{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 1000,
  "global_step": 12776,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00015654351909830932,
      "grad_norm": Infinity,
      "learning_rate": 0.0,
      "loss": 10.2049,
      "step": 1
    },
    {
      "epoch": 0.00031308703819661864,
      "grad_norm": Infinity,
      "learning_rate": 0.0,
      "loss": 10.1719,
      "step": 2
    },
    {
      "epoch": 0.000469630557294928,
      "grad_norm": 10.289196014404297,
      "learning_rate": 6e-07,
      "loss": 11.8275,
      "step": 3
    },
    {
      "epoch": 0.0006261740763932373,
      "grad_norm": 8.432656288146973,
      "learning_rate": 1.2e-06,
      "loss": 9.669,
      "step": 4
    },
    {
      "epoch": 0.0007827175954915466,
      "grad_norm": 10.236042022705078,
      "learning_rate": 1.8e-06,
      "loss": 11.8878,
      "step": 5
    },
    {
      "epoch": 0.000939261114589856,
      "grad_norm": 9.023536682128906,
      "learning_rate": 2.4e-06,
      "loss": 11.0993,
      "step": 6
    },
    {
      "epoch": 0.0010958046336881652,
      "grad_norm": 9.157316207885742,
      "learning_rate": 2.9999999999999997e-06,
      "loss": 11.0304,
      "step": 7
    },
    {
      "epoch": 0.0012523481527864746,
      "grad_norm": 15.224575996398926,
      "learning_rate": 3.6e-06,
      "loss": 18.5492,
      "step": 8
    },
    {
      "epoch": 0.001408891671884784,
      "grad_norm": 8.25251579284668,
      "learning_rate": 4.2e-06,
      "loss": 9.8113,
      "step": 9
    },
    {
      "epoch": 0.0015654351909830933,
      "grad_norm": 8.493621826171875,
      "learning_rate": 4.8e-06,
      "loss": 10.5777,
      "step": 10
    },
    {
      "epoch": 0.0017219787100814026,
      "grad_norm": 8.078389167785645,
      "learning_rate": 5.399999999999999e-06,
      "loss": 9.6918,
      "step": 11
    },
    {
      "epoch": 0.001878522229179712,
      "grad_norm": 12.197300910949707,
      "learning_rate": 5.999999999999999e-06,
      "loss": 15.5513,
      "step": 12
    },
    {
      "epoch": 0.002035065748278021,
      "grad_norm": 14.715069770812988,
      "learning_rate": 6.599999999999999e-06,
      "loss": 21.1878,
      "step": 13
    },
    {
      "epoch": 0.0021916092673763305,
      "grad_norm": 14.680959701538086,
      "learning_rate": 7.2e-06,
      "loss": 20.1223,
      "step": 14
    },
    {
      "epoch": 0.00234815278647464,
      "grad_norm": Infinity,
      "learning_rate": 7.2e-06,
      "loss": 25.9204,
      "step": 15
    },
    {
      "epoch": 0.002504696305572949,
      "grad_norm": 7.316311359405518,
      "learning_rate": 7.799999999999998e-06,
      "loss": 9.0609,
      "step": 16
    },
    {
      "epoch": 0.0026612398246712585,
      "grad_norm": 14.48494815826416,
      "learning_rate": 8.4e-06,
      "loss": 19.6195,
      "step": 17
    },
    {
      "epoch": 0.002817783343769568,
      "grad_norm": 24.607234954833984,
      "learning_rate": 8.999999999999999e-06,
      "loss": 24.8781,
      "step": 18
    },
    {
      "epoch": 0.002974326862867877,
      "grad_norm": 10.66619873046875,
      "learning_rate": 9.6e-06,
      "loss": 14.3883,
      "step": 19
    },
    {
      "epoch": 0.0031308703819661866,
      "grad_norm": 9.02160930633545,
      "learning_rate": 1.02e-05,
      "loss": 11.2116,
      "step": 20
    },
    {
      "epoch": 0.003287413901064496,
      "grad_norm": 8.273414611816406,
      "learning_rate": 1.0799999999999998e-05,
      "loss": 10.0936,
      "step": 21
    },
    {
      "epoch": 0.0034439574201628053,
      "grad_norm": 9.83197021484375,
      "learning_rate": 1.14e-05,
      "loss": 12.5294,
      "step": 22
    },
    {
      "epoch": 0.0036005009392611146,
      "grad_norm": 11.494529724121094,
      "learning_rate": 1.1999999999999999e-05,
      "loss": 15.1262,
      "step": 23
    },
    {
      "epoch": 0.003757044458359424,
      "grad_norm": 8.389466285705566,
      "learning_rate": 1.26e-05,
      "loss": 11.8818,
      "step": 24
    },
    {
      "epoch": 0.003913587977457733,
      "grad_norm": 11.484485626220703,
      "learning_rate": 1.3199999999999997e-05,
      "loss": 16.7747,
      "step": 25
    },
    {
      "epoch": 0.004070131496556042,
      "grad_norm": 11.260869026184082,
      "learning_rate": 1.3799999999999998e-05,
      "loss": 15.3859,
      "step": 26
    },
    {
      "epoch": 0.004226675015654352,
      "grad_norm": 17.069162368774414,
      "learning_rate": 1.44e-05,
      "loss": 19.4706,
      "step": 27
    },
    {
      "epoch": 0.004383218534752661,
      "grad_norm": 8.778854370117188,
      "learning_rate": 1.4999999999999999e-05,
      "loss": 13.0765,
      "step": 28
    },
    {
      "epoch": 0.004539762053850971,
      "grad_norm": 10.239326477050781,
      "learning_rate": 1.5599999999999996e-05,
      "loss": 14.7478,
      "step": 29
    },
    {
      "epoch": 0.00469630557294928,
      "grad_norm": 9.703652381896973,
      "learning_rate": 1.6199999999999997e-05,
      "loss": 13.5911,
      "step": 30
    },
    {
      "epoch": 0.004852849092047589,
      "grad_norm": 10.830334663391113,
      "learning_rate": 1.68e-05,
      "loss": 16.7388,
      "step": 31
    },
    {
      "epoch": 0.005009392611145898,
      "grad_norm": 14.102766990661621,
      "learning_rate": 1.74e-05,
      "loss": 19.3178,
      "step": 32
    },
    {
      "epoch": 0.005165936130244208,
      "grad_norm": 12.113805770874023,
      "learning_rate": 1.7999999999999997e-05,
      "loss": 17.4729,
      "step": 33
    },
    {
      "epoch": 0.005322479649342517,
      "grad_norm": 9.666101455688477,
      "learning_rate": 1.8599999999999998e-05,
      "loss": 14.8432,
      "step": 34
    },
    {
      "epoch": 0.005479023168440827,
      "grad_norm": 10.271428108215332,
      "learning_rate": 1.92e-05,
      "loss": 15.5869,
      "step": 35
    },
    {
      "epoch": 0.005635566687539136,
      "grad_norm": 27.508230209350586,
      "learning_rate": 1.98e-05,
      "loss": 13.2456,
      "step": 36
    },
    {
      "epoch": 0.0057921102066374455,
      "grad_norm": 10.40356159210205,
      "learning_rate": 2.04e-05,
      "loss": 15.4974,
      "step": 37
    },
    {
      "epoch": 0.005948653725735754,
      "grad_norm": 10.521185874938965,
      "learning_rate": 2.1e-05,
      "loss": 16.509,
      "step": 38
    },
    {
      "epoch": 0.006105197244834064,
      "grad_norm": 9.262508392333984,
      "learning_rate": 2.1599999999999996e-05,
      "loss": 12.5824,
      "step": 39
    },
    {
      "epoch": 0.006261740763932373,
      "grad_norm": 9.204741477966309,
      "learning_rate": 2.2199999999999998e-05,
      "loss": 13.576,
      "step": 40
    },
    {
      "epoch": 0.006418284283030683,
      "grad_norm": 6.91443395614624,
      "learning_rate": 2.28e-05,
      "loss": 11.8722,
      "step": 41
    },
    {
      "epoch": 0.006574827802128992,
      "grad_norm": 7.96762752532959,
      "learning_rate": 2.34e-05,
      "loss": 13.263,
      "step": 42
    },
    {
      "epoch": 0.006731371321227302,
      "grad_norm": 9.377372741699219,
      "learning_rate": 2.3999999999999997e-05,
      "loss": 12.1481,
      "step": 43
    },
    {
      "epoch": 0.0068879148403256105,
      "grad_norm": 6.440145015716553,
      "learning_rate": 2.4599999999999998e-05,
      "loss": 10.7579,
      "step": 44
    },
    {
      "epoch": 0.007044458359423919,
      "grad_norm": 5.9850077629089355,
      "learning_rate": 2.52e-05,
      "loss": 9.8462,
      "step": 45
    },
    {
      "epoch": 0.007201001878522229,
      "grad_norm": 10.260824203491211,
      "learning_rate": 2.5799999999999997e-05,
      "loss": 9.8883,
      "step": 46
    },
    {
      "epoch": 0.007357545397620538,
      "grad_norm": 10.321931838989258,
      "learning_rate": 2.6399999999999995e-05,
      "loss": 8.8922,
      "step": 47
    },
    {
      "epoch": 0.007514088916718848,
      "grad_norm": 5.632627964019775,
      "learning_rate": 2.6999999999999996e-05,
      "loss": 8.3128,
      "step": 48
    },
    {
      "epoch": 0.007670632435817157,
      "grad_norm": 4.798144817352295,
      "learning_rate": 2.7599999999999997e-05,
      "loss": 7.1669,
      "step": 49
    },
    {
      "epoch": 0.007827175954915467,
      "grad_norm": 4.642441272735596,
      "learning_rate": 2.8199999999999998e-05,
      "loss": 6.2127,
      "step": 50
    },
    {
      "epoch": 0.007983719474013776,
      "grad_norm": 12.44339370727539,
      "learning_rate": 2.88e-05,
      "loss": 11.9784,
      "step": 51
    },
    {
      "epoch": 0.008140262993112084,
      "grad_norm": 11.49549674987793,
      "learning_rate": 2.94e-05,
      "loss": 10.976,
      "step": 52
    },
    {
      "epoch": 0.008296806512210394,
      "grad_norm": 16.98020362854004,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 15.17,
      "step": 53
    },
    {
      "epoch": 0.008453350031308704,
      "grad_norm": 8.904106140136719,
      "learning_rate": 3.06e-05,
      "loss": 8.8607,
      "step": 54
    },
    {
      "epoch": 0.008609893550407014,
      "grad_norm": 9.727025985717773,
      "learning_rate": 3.119999999999999e-05,
      "loss": 9.6081,
      "step": 55
    },
    {
      "epoch": 0.008766437069505322,
      "grad_norm": 9.460471153259277,
      "learning_rate": 3.1799999999999994e-05,
      "loss": 9.9255,
      "step": 56
    },
    {
      "epoch": 0.008922980588603632,
      "grad_norm": 9.601983070373535,
      "learning_rate": 3.2399999999999995e-05,
      "loss": 9.9013,
      "step": 57
    },
    {
      "epoch": 0.009079524107701941,
      "grad_norm": 10.275938034057617,
      "learning_rate": 3.2999999999999996e-05,
      "loss": 11.9712,
      "step": 58
    },
    {
      "epoch": 0.009236067626800251,
      "grad_norm": 9.154601097106934,
      "learning_rate": 3.36e-05,
      "loss": 10.6476,
      "step": 59
    },
    {
      "epoch": 0.00939261114589856,
      "grad_norm": 8.198284149169922,
      "learning_rate": 3.42e-05,
      "loss": 9.7442,
      "step": 60
    },
    {
      "epoch": 0.009549154664996869,
      "grad_norm": 10.912142753601074,
      "learning_rate": 3.48e-05,
      "loss": 11.3455,
      "step": 61
    },
    {
      "epoch": 0.009705698184095179,
      "grad_norm": 6.028509616851807,
      "learning_rate": 3.539999999999999e-05,
      "loss": 7.7892,
      "step": 62
    },
    {
      "epoch": 0.009862241703193489,
      "grad_norm": 10.75712776184082,
      "learning_rate": 3.5999999999999994e-05,
      "loss": 10.4904,
      "step": 63
    },
    {
      "epoch": 0.010018785222291797,
      "grad_norm": 11.000633239746094,
      "learning_rate": 3.6599999999999995e-05,
      "loss": 9.4008,
      "step": 64
    },
    {
      "epoch": 0.010175328741390106,
      "grad_norm": 7.76064395904541,
      "learning_rate": 3.7199999999999996e-05,
      "loss": 7.2246,
      "step": 65
    },
    {
      "epoch": 0.010331872260488416,
      "grad_norm": 23.41922950744629,
      "learning_rate": 3.78e-05,
      "loss": 13.3124,
      "step": 66
    },
    {
      "epoch": 0.010488415779586726,
      "grad_norm": 11.076509475708008,
      "learning_rate": 3.84e-05,
      "loss": 7.637,
      "step": 67
    },
    {
      "epoch": 0.010644959298685034,
      "grad_norm": 29.545185089111328,
      "learning_rate": 3.9e-05,
      "loss": 14.3584,
      "step": 68
    },
    {
      "epoch": 0.010801502817783344,
      "grad_norm": 17.31962776184082,
      "learning_rate": 3.96e-05,
      "loss": 8.6004,
      "step": 69
    },
    {
      "epoch": 0.010958046336881654,
      "grad_norm": 11.55864143371582,
      "learning_rate": 4.02e-05,
      "loss": 6.3439,
      "step": 70
    },
    {
      "epoch": 0.011114589855979962,
      "grad_norm": 25.8043212890625,
      "learning_rate": 4.08e-05,
      "loss": 10.298,
      "step": 71
    },
    {
      "epoch": 0.011271133375078271,
      "grad_norm": 33.71727752685547,
      "learning_rate": 4.14e-05,
      "loss": 12.2877,
      "step": 72
    },
    {
      "epoch": 0.011427676894176581,
      "grad_norm": 20.209596633911133,
      "learning_rate": 4.2e-05,
      "loss": 7.7747,
      "step": 73
    },
    {
      "epoch": 0.011584220413274891,
      "grad_norm": 20.540761947631836,
      "learning_rate": 4.259999999999999e-05,
      "loss": 7.6235,
      "step": 74
    },
    {
      "epoch": 0.011740763932373199,
      "grad_norm": Infinity,
      "learning_rate": 4.259999999999999e-05,
      "loss": 12.327,
      "step": 75
    },
    {
      "epoch": 0.011897307451471509,
      "grad_norm": 19.35993766784668,
      "learning_rate": 4.319999999999999e-05,
      "loss": 7.1177,
      "step": 76
    },
    {
      "epoch": 0.012053850970569819,
      "grad_norm": 36.380619049072266,
      "learning_rate": 4.3799999999999994e-05,
      "loss": 10.6781,
      "step": 77
    },
    {
      "epoch": 0.012210394489668128,
      "grad_norm": 45.644840240478516,
      "learning_rate": 4.4399999999999995e-05,
      "loss": 12.4375,
      "step": 78
    },
    {
      "epoch": 0.012366938008766436,
      "grad_norm": 27.72957992553711,
      "learning_rate": 4.4999999999999996e-05,
      "loss": 8.1513,
      "step": 79
    },
    {
      "epoch": 0.012523481527864746,
      "grad_norm": 35.70543670654297,
      "learning_rate": 4.56e-05,
      "loss": 9.3938,
      "step": 80
    },
    {
      "epoch": 0.012680025046963056,
      "grad_norm": 51.08866500854492,
      "learning_rate": 4.62e-05,
      "loss": 12.0758,
      "step": 81
    },
    {
      "epoch": 0.012836568566061366,
      "grad_norm": 38.18144989013672,
      "learning_rate": 4.68e-05,
      "loss": 9.2489,
      "step": 82
    },
    {
      "epoch": 0.012993112085159674,
      "grad_norm": 25.754817962646484,
      "learning_rate": 4.7399999999999993e-05,
      "loss": 6.9829,
      "step": 83
    },
    {
      "epoch": 0.013149655604257984,
      "grad_norm": 33.58719253540039,
      "learning_rate": 4.7999999999999994e-05,
      "loss": 8.0805,
      "step": 84
    },
    {
      "epoch": 0.013306199123356293,
      "grad_norm": 19.882631301879883,
      "learning_rate": 4.8599999999999995e-05,
      "loss": 5.7876,
      "step": 85
    },
    {
      "epoch": 0.013462742642454603,
      "grad_norm": 28.052623748779297,
      "learning_rate": 4.9199999999999997e-05,
      "loss": 6.844,
      "step": 86
    },
    {
      "epoch": 0.013619286161552911,
      "grad_norm": 24.705547332763672,
      "learning_rate": 4.98e-05,
      "loss": 6.2694,
      "step": 87
    },
    {
      "epoch": 0.013775829680651221,
      "grad_norm": 27.596519470214844,
      "learning_rate": 5.04e-05,
      "loss": 6.6974,
      "step": 88
    },
    {
      "epoch": 0.01393237319974953,
      "grad_norm": 24.197368621826172,
      "learning_rate": 5.1e-05,
      "loss": 5.9251,
      "step": 89
    },
    {
      "epoch": 0.014088916718847839,
      "grad_norm": 33.817039489746094,
      "learning_rate": 5.1599999999999994e-05,
      "loss": 7.2591,
      "step": 90
    },
    {
      "epoch": 0.014245460237946149,
      "grad_norm": 28.664337158203125,
      "learning_rate": 5.2199999999999995e-05,
      "loss": 6.3747,
      "step": 91
    },
    {
      "epoch": 0.014402003757044458,
      "grad_norm": 19.04326057434082,
      "learning_rate": 5.279999999999999e-05,
      "loss": 5.2411,
      "step": 92
    },
    {
      "epoch": 0.014558547276142768,
      "grad_norm": 19.95840835571289,
      "learning_rate": 5.339999999999999e-05,
      "loss": 5.2001,
      "step": 93
    },
    {
      "epoch": 0.014715090795241076,
      "grad_norm": 17.750680923461914,
      "learning_rate": 5.399999999999999e-05,
      "loss": 4.9905,
      "step": 94
    },
    {
      "epoch": 0.014871634314339386,
      "grad_norm": 17.195247650146484,
      "learning_rate": 5.459999999999999e-05,
      "loss": 4.8285,
      "step": 95
    },
    {
      "epoch": 0.015028177833437696,
      "grad_norm": 12.27951431274414,
      "learning_rate": 5.519999999999999e-05,
      "loss": 4.311,
      "step": 96
    },
    {
      "epoch": 0.015184721352536006,
      "grad_norm": 10.93501091003418,
      "learning_rate": 5.5799999999999994e-05,
      "loss": 4.0848,
      "step": 97
    },
    {
      "epoch": 0.015341264871634314,
      "grad_norm": 8.672501564025879,
      "learning_rate": 5.6399999999999995e-05,
      "loss": 3.961,
      "step": 98
    },
    {
      "epoch": 0.015497808390732623,
      "grad_norm": 7.686964988708496,
      "learning_rate": 5.6999999999999996e-05,
      "loss": 3.8516,
      "step": 99
    },
    {
      "epoch": 0.015654351909830933,
      "grad_norm": 5.318123817443848,
      "learning_rate": 5.76e-05,
      "loss": 3.6987,
      "step": 100
    },
    {
      "epoch": 0.01581089542892924,
      "grad_norm": 13.304719924926758,
      "learning_rate": 5.82e-05,
      "loss": 4.1017,
      "step": 101
    },
    {
      "epoch": 0.015967438948027553,
      "grad_norm": 12.244338989257812,
      "learning_rate": 5.88e-05,
      "loss": 3.9793,
      "step": 102
    },
    {
      "epoch": 0.01612398246712586,
      "grad_norm": 11.684955596923828,
      "learning_rate": 5.94e-05,
      "loss": 3.9138,
      "step": 103
    },
    {
      "epoch": 0.01628052598622417,
      "grad_norm": 36.793663024902344,
      "learning_rate": 5.9999999999999995e-05,
      "loss": 6.8625,
      "step": 104
    },
    {
      "epoch": 0.01643706950532248,
      "grad_norm": 20.531827926635742,
      "learning_rate": 6.0599999999999996e-05,
      "loss": 4.8527,
      "step": 105
    },
    {
      "epoch": 0.01659361302442079,
      "grad_norm": 12.65087604522705,
      "learning_rate": 6.12e-05,
      "loss": 4.0059,
      "step": 106
    },
    {
      "epoch": 0.0167501565435191,
      "grad_norm": 9.573142051696777,
      "learning_rate": 6.18e-05,
      "loss": 3.7414,
      "step": 107
    },
    {
      "epoch": 0.016906700062617408,
      "grad_norm": 13.197820663452148,
      "learning_rate": 6.239999999999999e-05,
      "loss": 4.1523,
      "step": 108
    },
    {
      "epoch": 0.017063243581715716,
      "grad_norm": 10.532169342041016,
      "learning_rate": 6.299999999999999e-05,
      "loss": 3.8368,
      "step": 109
    },
    {
      "epoch": 0.017219787100814028,
      "grad_norm": 12.9858980178833,
      "learning_rate": 6.359999999999999e-05,
      "loss": 4.1369,
      "step": 110
    },
    {
      "epoch": 0.017376330619912336,
      "grad_norm": 29.88274574279785,
      "learning_rate": 6.419999999999999e-05,
      "loss": 5.8008,
      "step": 111
    },
    {
      "epoch": 0.017532874139010644,
      "grad_norm": 17.681074142456055,
      "learning_rate": 6.479999999999999e-05,
      "loss": 4.4349,
      "step": 112
    },
    {
      "epoch": 0.017689417658108955,
      "grad_norm": 14.819682121276855,
      "learning_rate": 6.539999999999999e-05,
      "loss": 4.3329,
      "step": 113
    },
    {
      "epoch": 0.017845961177207263,
      "grad_norm": 12.398885726928711,
      "learning_rate": 6.599999999999999e-05,
      "loss": 4.0054,
      "step": 114
    },
    {
      "epoch": 0.01800250469630557,
      "grad_norm": 11.930230140686035,
      "learning_rate": 6.659999999999999e-05,
      "loss": 3.9655,
      "step": 115
    },
    {
      "epoch": 0.018159048215403883,
      "grad_norm": 8.862527847290039,
      "learning_rate": 6.72e-05,
      "loss": 3.6875,
      "step": 116
    },
    {
      "epoch": 0.01831559173450219,
      "grad_norm": 21.191814422607422,
      "learning_rate": 6.78e-05,
      "loss": 4.9808,
      "step": 117
    },
    {
      "epoch": 0.018472135253600502,
      "grad_norm": 5.6609392166137695,
      "learning_rate": 6.84e-05,
      "loss": 3.446,
      "step": 118
    },
    {
      "epoch": 0.01862867877269881,
      "grad_norm": 15.799832344055176,
      "learning_rate": 6.9e-05,
      "loss": 4.3345,
      "step": 119
    },
    {
      "epoch": 0.01878522229179712,
      "grad_norm": 12.569608688354492,
      "learning_rate": 6.96e-05,
      "loss": 4.0145,
      "step": 120
    },
    {
      "epoch": 0.01894176581089543,
      "grad_norm": 6.250307559967041,
      "learning_rate": 7.02e-05,
      "loss": 3.531,
      "step": 121
    },
    {
      "epoch": 0.019098309329993738,
      "grad_norm": 9.823454856872559,
      "learning_rate": 7.079999999999999e-05,
      "loss": 3.8133,
      "step": 122
    },
    {
      "epoch": 0.019254852849092046,
      "grad_norm": 14.932891845703125,
      "learning_rate": 7.139999999999999e-05,
      "loss": 4.0484,
      "step": 123
    },
    {
      "epoch": 0.019411396368190358,
      "grad_norm": 8.835564613342285,
      "learning_rate": 7.199999999999999e-05,
      "loss": 3.775,
      "step": 124
    },
    {
      "epoch": 0.019567939887288666,
      "grad_norm": 2.77044939994812,
      "learning_rate": 7.259999999999999e-05,
      "loss": 3.385,
      "step": 125
    },
    {
      "epoch": 0.019724483406386977,
      "grad_norm": 19.207578659057617,
      "learning_rate": 7.319999999999999e-05,
      "loss": 4.4818,
      "step": 126
    },
    {
      "epoch": 0.019881026925485285,
      "grad_norm": 20.31056785583496,
      "learning_rate": 7.379999999999999e-05,
      "loss": 4.5493,
      "step": 127
    },
    {
      "epoch": 0.020037570444583593,
      "grad_norm": 9.711579322814941,
      "learning_rate": 7.439999999999999e-05,
      "loss": 3.7879,
      "step": 128
    },
    {
      "epoch": 0.020194113963681905,
      "grad_norm": 19.968996047973633,
      "learning_rate": 7.5e-05,
      "loss": 4.6096,
      "step": 129
    },
    {
      "epoch": 0.020350657482780213,
      "grad_norm": 13.640237808227539,
      "learning_rate": 7.56e-05,
      "loss": 4.1364,
      "step": 130
    },
    {
      "epoch": 0.02050720100187852,
      "grad_norm": 18.92345428466797,
      "learning_rate": 7.62e-05,
      "loss": 4.4219,
      "step": 131
    },
    {
      "epoch": 0.020663744520976832,
      "grad_norm": 5.510997295379639,
      "learning_rate": 7.68e-05,
      "loss": 3.615,
      "step": 132
    },
    {
      "epoch": 0.02082028804007514,
      "grad_norm": 11.286620140075684,
      "learning_rate": 7.74e-05,
      "loss": 4.0149,
      "step": 133
    },
    {
      "epoch": 0.020976831559173452,
      "grad_norm": 8.938575744628906,
      "learning_rate": 7.8e-05,
      "loss": 3.8435,
      "step": 134
    },
    {
      "epoch": 0.02113337507827176,
      "grad_norm": 8.202083587646484,
      "learning_rate": 7.86e-05,
      "loss": 3.8606,
      "step": 135
    },
    {
      "epoch": 0.021289918597370068,
      "grad_norm": 7.211292266845703,
      "learning_rate": 7.92e-05,
      "loss": 3.6637,
      "step": 136
    },
    {
      "epoch": 0.02144646211646838,
      "grad_norm": 6.565310001373291,
      "learning_rate": 7.98e-05,
      "loss": 3.602,
      "step": 137
    },
    {
      "epoch": 0.021603005635566688,
      "grad_norm": 5.222198009490967,
      "learning_rate": 8.04e-05,
      "loss": 3.6108,
      "step": 138
    },
    {
      "epoch": 0.021759549154664996,
      "grad_norm": 9.35428237915039,
      "learning_rate": 8.1e-05,
      "loss": 3.831,
      "step": 139
    },
    {
      "epoch": 0.021916092673763307,
      "grad_norm": 9.887560844421387,
      "learning_rate": 8.16e-05,
      "loss": 3.8316,
      "step": 140
    },
    {
      "epoch": 0.022072636192861615,
      "grad_norm": 8.195279121398926,
      "learning_rate": 8.22e-05,
      "loss": 3.642,
      "step": 141
    },
    {
      "epoch": 0.022229179711959923,
      "grad_norm": 3.218400478363037,
      "learning_rate": 8.28e-05,
      "loss": 3.44,
      "step": 142
    },
    {
      "epoch": 0.022385723231058235,
      "grad_norm": 2.472031354904175,
      "learning_rate": 8.34e-05,
      "loss": 3.4347,
      "step": 143
    },
    {
      "epoch": 0.022542266750156543,
      "grad_norm": 2.141615390777588,
      "learning_rate": 8.4e-05,
      "loss": 3.3126,
      "step": 144
    },
    {
      "epoch": 0.022698810269254854,
      "grad_norm": 3.026726007461548,
      "learning_rate": 8.459999999999998e-05,
      "loss": 3.3058,
      "step": 145
    },
    {
      "epoch": 0.022855353788353162,
      "grad_norm": 3.966043472290039,
      "learning_rate": 8.519999999999998e-05,
      "loss": 3.3199,
      "step": 146
    },
    {
      "epoch": 0.02301189730745147,
      "grad_norm": 3.906219720840454,
      "learning_rate": 8.579999999999998e-05,
      "loss": 3.2163,
      "step": 147
    },
    {
      "epoch": 0.023168440826549782,
      "grad_norm": 2.3450818061828613,
      "learning_rate": 8.639999999999999e-05,
      "loss": 3.1645,
      "step": 148
    },
    {
      "epoch": 0.02332498434564809,
      "grad_norm": 3.949995756149292,
      "learning_rate": 8.699999999999999e-05,
      "loss": 3.1981,
      "step": 149
    },
    {
      "epoch": 0.023481527864746398,
      "grad_norm": 2.897047996520996,
      "learning_rate": 8.759999999999999e-05,
      "loss": 3.0328,
      "step": 150
    },
    {
      "epoch": 0.02363807138384471,
      "grad_norm": 9.815628051757812,
      "learning_rate": 8.819999999999999e-05,
      "loss": 3.4079,
      "step": 151
    },
    {
      "epoch": 0.023794614902943018,
      "grad_norm": 8.956120491027832,
      "learning_rate": 8.879999999999999e-05,
      "loss": 3.4144,
      "step": 152
    },
    {
      "epoch": 0.02395115842204133,
      "grad_norm": 5.622773170471191,
      "learning_rate": 8.939999999999999e-05,
      "loss": 3.2383,
      "step": 153
    },
    {
      "epoch": 0.024107701941139637,
      "grad_norm": 10.2916259765625,
      "learning_rate": 8.999999999999999e-05,
      "loss": 3.6527,
      "step": 154
    },
    {
      "epoch": 0.024264245460237945,
      "grad_norm": 3.466508388519287,
      "learning_rate": 9.059999999999999e-05,
      "loss": 3.1562,
      "step": 155
    },
    {
      "epoch": 0.024420788979336257,
      "grad_norm": 3.6558587551116943,
      "learning_rate": 9.12e-05,
      "loss": 3.1937,
      "step": 156
    },
    {
      "epoch": 0.024577332498434565,
      "grad_norm": 1.539302945137024,
      "learning_rate": 9.18e-05,
      "loss": 3.0899,
      "step": 157
    },
    {
      "epoch": 0.024733876017532873,
      "grad_norm": 4.4443230628967285,
      "learning_rate": 9.24e-05,
      "loss": 3.3159,
      "step": 158
    },
    {
      "epoch": 0.024890419536631184,
      "grad_norm": 3.2545831203460693,
      "learning_rate": 9.3e-05,
      "loss": 3.1318,
      "step": 159
    },
    {
      "epoch": 0.025046963055729492,
      "grad_norm": 3.192456007003784,
      "learning_rate": 9.36e-05,
      "loss": 3.2075,
      "step": 160
    },
    {
      "epoch": 0.0252035065748278,
      "grad_norm": 1.9067258834838867,
      "learning_rate": 9.419999999999999e-05,
      "loss": 3.1137,
      "step": 161
    },
    {
      "epoch": 0.025360050093926112,
      "grad_norm": 1.6764792203903198,
      "learning_rate": 9.479999999999999e-05,
      "loss": 3.109,
      "step": 162
    },
    {
      "epoch": 0.02551659361302442,
      "grad_norm": 25.346240997314453,
      "learning_rate": 9.539999999999999e-05,
      "loss": 4.8797,
      "step": 163
    },
    {
      "epoch": 0.02567313713212273,
      "grad_norm": 1.1866326332092285,
      "learning_rate": 9.599999999999999e-05,
      "loss": 3.0522,
      "step": 164
    },
    {
      "epoch": 0.02582968065122104,
      "grad_norm": 1.4010124206542969,
      "learning_rate": 9.659999999999999e-05,
      "loss": 3.0789,
      "step": 165
    },
    {
      "epoch": 0.025986224170319348,
      "grad_norm": 8.419206619262695,
      "learning_rate": 9.719999999999999e-05,
      "loss": 3.3802,
      "step": 166
    },
    {
      "epoch": 0.02614276768941766,
      "grad_norm": 8.86681842803955,
      "learning_rate": 9.779999999999999e-05,
      "loss": 3.5289,
      "step": 167
    },
    {
      "epoch": 0.026299311208515967,
      "grad_norm": 23.243457794189453,
      "learning_rate": 9.839999999999999e-05,
      "loss": 4.5939,
      "step": 168
    },
    {
      "epoch": 0.026455854727614275,
      "grad_norm": 2.583829164505005,
      "learning_rate": 9.9e-05,
      "loss": 3.0461,
      "step": 169
    },
    {
      "epoch": 0.026612398246712587,
      "grad_norm": 3.8143973350524902,
      "learning_rate": 9.96e-05,
      "loss": 3.1916,
      "step": 170
    },
    {
      "epoch": 0.026768941765810895,
      "grad_norm": 3.9216201305389404,
      "learning_rate": 0.0001002,
      "loss": 3.2371,
      "step": 171
    },
    {
      "epoch": 0.026925485284909206,
      "grad_norm": 13.120063781738281,
      "learning_rate": 0.0001008,
      "loss": 3.7303,
      "step": 172
    },
    {
      "epoch": 0.027082028804007514,
      "grad_norm": 1.0374871492385864,
      "learning_rate": 0.0001014,
      "loss": 3.1096,
      "step": 173
    },
    {
      "epoch": 0.027238572323105822,
      "grad_norm": 2.4160315990448,
      "learning_rate": 0.000102,
      "loss": 3.2397,
      "step": 174
    },
    {
      "epoch": 0.027395115842204134,
      "grad_norm": 10.37614631652832,
      "learning_rate": 0.0001026,
      "loss": 3.6207,
      "step": 175
    },
    {
      "epoch": 0.027551659361302442,
      "grad_norm": 3.1003804206848145,
      "learning_rate": 0.00010319999999999999,
      "loss": 3.3122,
      "step": 176
    },
    {
      "epoch": 0.02770820288040075,
      "grad_norm": 1.6606565713882446,
      "learning_rate": 0.00010379999999999999,
      "loss": 3.0935,
      "step": 177
    },
    {
      "epoch": 0.02786474639949906,
      "grad_norm": 1.8800925016403198,
      "learning_rate": 0.00010439999999999999,
      "loss": 3.2649,
      "step": 178
    },
    {
      "epoch": 0.02802128991859737,
      "grad_norm": 6.265193462371826,
      "learning_rate": 0.00010499999999999999,
      "loss": 3.4573,
      "step": 179
    },
    {
      "epoch": 0.028177833437695678,
      "grad_norm": 8.842411994934082,
      "learning_rate": 0.00010559999999999998,
      "loss": 3.6856,
      "step": 180
    },
    {
      "epoch": 0.02833437695679399,
      "grad_norm": 1.783496379852295,
      "learning_rate": 0.00010619999999999998,
      "loss": 3.2177,
      "step": 181
    },
    {
      "epoch": 0.028490920475892297,
      "grad_norm": 3.2805280685424805,
      "learning_rate": 0.00010679999999999998,
      "loss": 3.2863,
      "step": 182
    },
    {
      "epoch": 0.02864746399499061,
      "grad_norm": 3.0294930934906006,
      "learning_rate": 0.00010739999999999998,
      "loss": 3.428,
      "step": 183
    },
    {
      "epoch": 0.028804007514088917,
      "grad_norm": 1.3118689060211182,
      "learning_rate": 0.00010799999999999998,
      "loss": 3.3734,
      "step": 184
    },
    {
      "epoch": 0.028960551033187225,
      "grad_norm": 4.640561103820801,
      "learning_rate": 0.00010859999999999998,
      "loss": 3.4229,
      "step": 185
    },
    {
      "epoch": 0.029117094552285536,
      "grad_norm": 2.2550694942474365,
      "learning_rate": 0.00010919999999999998,
      "loss": 3.1565,
      "step": 186
    },
    {
      "epoch": 0.029273638071383844,
      "grad_norm": 2.734482765197754,
      "learning_rate": 0.00010979999999999999,
      "loss": 3.1083,
      "step": 187
    },
    {
      "epoch": 0.029430181590482152,
      "grad_norm": 2.0161116123199463,
      "learning_rate": 0.00011039999999999999,
      "loss": 3.1897,
      "step": 188
    },
    {
      "epoch": 0.029586725109580464,
      "grad_norm": 1.5846065282821655,
      "learning_rate": 0.00011099999999999999,
      "loss": 3.1898,
      "step": 189
    },
    {
      "epoch": 0.029743268628678772,
      "grad_norm": 7.154541015625,
      "learning_rate": 0.00011159999999999999,
      "loss": 3.3738,
      "step": 190
    },
    {
      "epoch": 0.029899812147777084,
      "grad_norm": 2.3260960578918457,
      "learning_rate": 0.00011219999999999999,
      "loss": 3.2111,
      "step": 191
    },
    {
      "epoch": 0.03005635566687539,
      "grad_norm": 2.29046630859375,
      "learning_rate": 0.00011279999999999999,
      "loss": 3.2184,
      "step": 192
    },
    {
      "epoch": 0.0302128991859737,
      "grad_norm": 3.9989373683929443,
      "learning_rate": 0.00011339999999999999,
      "loss": 3.1476,
      "step": 193
    },
    {
      "epoch": 0.03036944270507201,
      "grad_norm": 1.7007783651351929,
      "learning_rate": 0.00011399999999999999,
      "loss": 3.098,
      "step": 194
    },
    {
      "epoch": 0.03052598622417032,
      "grad_norm": 1.5231850147247314,
      "learning_rate": 0.0001146,
      "loss": 3.1793,
      "step": 195
    },
    {
      "epoch": 0.030682529743268627,
      "grad_norm": 3.336979627609253,
      "learning_rate": 0.0001152,
      "loss": 3.0494,
      "step": 196
    },
    {
      "epoch": 0.03083907326236694,
      "grad_norm": 2.053584098815918,
      "learning_rate": 0.0001158,
      "loss": 2.9805,
      "step": 197
    },
    {
      "epoch": 0.030995616781465247,
      "grad_norm": 1.7036573886871338,
      "learning_rate": 0.0001164,
      "loss": 2.9734,
      "step": 198
    },
    {
      "epoch": 0.03115216030056356,
      "grad_norm": 3.5702524185180664,
      "learning_rate": 0.000117,
      "loss": 2.9142,
      "step": 199
    },
    {
      "epoch": 0.031308703819661866,
      "grad_norm": 3.3421378135681152,
      "learning_rate": 0.0001176,
      "loss": 2.954,
      "step": 200
    },
    {
      "epoch": 0.031465247338760174,
      "grad_norm": 59.050907135009766,
      "learning_rate": 0.0001182,
      "loss": 8.3104,
      "step": 201
    },
    {
      "epoch": 0.03162179085785848,
      "grad_norm": 17.812931060791016,
      "learning_rate": 0.0001188,
      "loss": 4.2312,
      "step": 202
    },
    {
      "epoch": 0.03177833437695679,
      "grad_norm": 3.1730802059173584,
      "learning_rate": 0.0001194,
      "loss": 3.0879,
      "step": 203
    },
    {
      "epoch": 0.031934877896055106,
      "grad_norm": 2.468261241912842,
      "learning_rate": 0.00011999999999999999,
      "loss": 3.0617,
      "step": 204
    },
    {
      "epoch": 0.032091421415153414,
      "grad_norm": 6.175726890563965,
      "learning_rate": 0.00012059999999999999,
      "loss": 3.1855,
      "step": 205
    },
    {
      "epoch": 0.03224796493425172,
      "grad_norm": 3.9613351821899414,
      "learning_rate": 0.00012119999999999999,
      "loss": 3.237,
      "step": 206
    },
    {
      "epoch": 0.03240450845335003,
      "grad_norm": 3.2016634941101074,
      "learning_rate": 0.00012179999999999999,
      "loss": 3.21,
      "step": 207
    },
    {
      "epoch": 0.03256105197244834,
      "grad_norm": 2.0282530784606934,
      "learning_rate": 0.0001224,
      "loss": 3.0034,
      "step": 208
    },
    {
      "epoch": 0.03271759549154665,
      "grad_norm": 1.9264556169509888,
      "learning_rate": 0.00012299999999999998,
      "loss": 3.0214,
      "step": 209
    },
    {
      "epoch": 0.03287413901064496,
      "grad_norm": 1.9033787250518799,
      "learning_rate": 0.0001236,
      "loss": 3.0177,
      "step": 210
    },
    {
      "epoch": 0.03303068252974327,
      "grad_norm": 1.997543215751648,
      "learning_rate": 0.00012419999999999998,
      "loss": 3.0139,
      "step": 211
    },
    {
      "epoch": 0.03318722604884158,
      "grad_norm": 3.2476229667663574,
      "learning_rate": 0.00012479999999999997,
      "loss": 3.069,
      "step": 212
    },
    {
      "epoch": 0.033343769567939885,
      "grad_norm": 2.2744576930999756,
      "learning_rate": 0.00012539999999999999,
      "loss": 2.9909,
      "step": 213
    },
    {
      "epoch": 0.0335003130870382,
      "grad_norm": 1.2331576347351074,
      "learning_rate": 0.00012599999999999997,
      "loss": 2.9783,
      "step": 214
    },
    {
      "epoch": 0.03365685660613651,
      "grad_norm": 2.7535603046417236,
      "learning_rate": 0.0001266,
      "loss": 2.9448,
      "step": 215
    },
    {
      "epoch": 0.033813400125234816,
      "grad_norm": 0.8149605989456177,
      "learning_rate": 0.00012719999999999997,
      "loss": 2.9163,
      "step": 216
    },
    {
      "epoch": 0.033969943644333124,
      "grad_norm": 12.775609970092773,
      "learning_rate": 0.0001278,
      "loss": 3.6752,
      "step": 217
    },
    {
      "epoch": 0.03412648716343143,
      "grad_norm": 3.7931580543518066,
      "learning_rate": 0.00012839999999999998,
      "loss": 3.0796,
      "step": 218
    },
    {
      "epoch": 0.03428303068252974,
      "grad_norm": 4.115076065063477,
      "learning_rate": 0.000129,
      "loss": 3.0605,
      "step": 219
    },
    {
      "epoch": 0.034439574201628055,
      "grad_norm": 0.8560850024223328,
      "learning_rate": 0.00012959999999999998,
      "loss": 2.994,
      "step": 220
    },
    {
      "epoch": 0.03459611772072636,
      "grad_norm": 4.085236072540283,
      "learning_rate": 0.0001302,
      "loss": 3.2076,
      "step": 221
    },
    {
      "epoch": 0.03475266123982467,
      "grad_norm": 1.9408749341964722,
      "learning_rate": 0.00013079999999999998,
      "loss": 3.0968,
      "step": 222
    },
    {
      "epoch": 0.03490920475892298,
      "grad_norm": 2.0618252754211426,
      "learning_rate": 0.0001314,
      "loss": 3.0112,
      "step": 223
    },
    {
      "epoch": 0.03506574827802129,
      "grad_norm": 2.527026414871216,
      "learning_rate": 0.00013199999999999998,
      "loss": 2.9357,
      "step": 224
    },
    {
      "epoch": 0.0352222917971196,
      "grad_norm": 8.99402904510498,
      "learning_rate": 0.0001326,
      "loss": 3.2706,
      "step": 225
    },
    {
      "epoch": 0.03537883531621791,
      "grad_norm": 7.271577835083008,
      "learning_rate": 0.00013319999999999999,
      "loss": 3.282,
      "step": 226
    },
    {
      "epoch": 0.03553537883531622,
      "grad_norm": 0.7550258040428162,
      "learning_rate": 0.0001338,
      "loss": 2.9374,
      "step": 227
    },
    {
      "epoch": 0.035691922354414526,
      "grad_norm": 2.4851882457733154,
      "learning_rate": 0.0001344,
      "loss": 3.0337,
      "step": 228
    },
    {
      "epoch": 0.035848465873512834,
      "grad_norm": 2.233167886734009,
      "learning_rate": 0.000135,
      "loss": 2.9723,
      "step": 229
    },
    {
      "epoch": 0.03600500939261114,
      "grad_norm": 1.3427929878234863,
      "learning_rate": 0.0001356,
      "loss": 3.0346,
      "step": 230
    },
    {
      "epoch": 0.03616155291170946,
      "grad_norm": 1.4141734838485718,
      "learning_rate": 0.0001362,
      "loss": 2.9946,
      "step": 231
    },
    {
      "epoch": 0.036318096430807766,
      "grad_norm": 0.9676278233528137,
      "learning_rate": 0.0001368,
      "loss": 3.0399,
      "step": 232
    },
    {
      "epoch": 0.036474639949906074,
      "grad_norm": 0.9752649664878845,
      "learning_rate": 0.0001374,
      "loss": 2.9977,
      "step": 233
    },
    {
      "epoch": 0.03663118346900438,
      "grad_norm": 1.0055956840515137,
      "learning_rate": 0.000138,
      "loss": 3.006,
      "step": 234
    },
    {
      "epoch": 0.03678772698810269,
      "grad_norm": 2.232508659362793,
      "learning_rate": 0.0001386,
      "loss": 2.9964,
      "step": 235
    },
    {
      "epoch": 0.036944270507201005,
      "grad_norm": 1.2427127361297607,
      "learning_rate": 0.0001392,
      "loss": 2.958,
      "step": 236
    },
    {
      "epoch": 0.03710081402629931,
      "grad_norm": 1.305584192276001,
      "learning_rate": 0.00013979999999999998,
      "loss": 3.0754,
      "step": 237
    },
    {
      "epoch": 0.03725735754539762,
      "grad_norm": 2.785289764404297,
      "learning_rate": 0.0001404,
      "loss": 3.0478,
      "step": 238
    },
    {
      "epoch": 0.03741390106449593,
      "grad_norm": 4.857529640197754,
      "learning_rate": 0.00014099999999999998,
      "loss": 3.2,
      "step": 239
    },
    {
      "epoch": 0.03757044458359424,
      "grad_norm": 2.906057119369507,
      "learning_rate": 0.00014159999999999997,
      "loss": 3.0873,
      "step": 240
    },
    {
      "epoch": 0.03772698810269255,
      "grad_norm": 1.3470922708511353,
      "learning_rate": 0.0001422,
      "loss": 3.0688,
      "step": 241
    },
    {
      "epoch": 0.03788353162179086,
      "grad_norm": 3.2114920616149902,
      "learning_rate": 0.00014279999999999997,
      "loss": 3.1128,
      "step": 242
    },
    {
      "epoch": 0.03804007514088917,
      "grad_norm": 2.603756904602051,
      "learning_rate": 0.0001434,
      "loss": 3.0977,
      "step": 243
    },
    {
      "epoch": 0.038196618659987476,
      "grad_norm": 4.524052143096924,
      "learning_rate": 0.00014399999999999998,
      "loss": 3.1806,
      "step": 244
    },
    {
      "epoch": 0.038353162179085784,
      "grad_norm": 2.9187328815460205,
      "learning_rate": 0.0001446,
      "loss": 3.0789,
      "step": 245
    },
    {
      "epoch": 0.03850970569818409,
      "grad_norm": 3.1184439659118652,
      "learning_rate": 0.00014519999999999998,
      "loss": 2.9883,
      "step": 246
    },
    {
      "epoch": 0.03866624921728241,
      "grad_norm": 3.0836615562438965,
      "learning_rate": 0.0001458,
      "loss": 2.9124,
      "step": 247
    },
    {
      "epoch": 0.038822792736380715,
      "grad_norm": 2.7621877193450928,
      "learning_rate": 0.00014639999999999998,
      "loss": 2.9268,
      "step": 248
    },
    {
      "epoch": 0.03897933625547902,
      "grad_norm": 1.8909045457839966,
      "learning_rate": 0.000147,
      "loss": 2.8235,
      "step": 249
    },
    {
      "epoch": 0.03913587977457733,
      "grad_norm": 3.872973680496216,
      "learning_rate": 0.00014759999999999998,
      "loss": 2.7703,
      "step": 250
    },
    {
      "epoch": 0.03929242329367564,
      "grad_norm": 4.677459716796875,
      "learning_rate": 0.0001482,
      "loss": 3.046,
      "step": 251
    },
    {
      "epoch": 0.039448966812773954,
      "grad_norm": 3.410970449447632,
      "learning_rate": 0.00014879999999999998,
      "loss": 3.038,
      "step": 252
    },
    {
      "epoch": 0.03960551033187226,
      "grad_norm": 6.155794143676758,
      "learning_rate": 0.0001494,
      "loss": 3.2031,
      "step": 253
    },
    {
      "epoch": 0.03976205385097057,
      "grad_norm": 2.345693588256836,
      "learning_rate": 0.00015,
      "loss": 2.9807,
      "step": 254
    },
    {
      "epoch": 0.03991859737006888,
      "grad_norm": 1.6279584169387817,
      "learning_rate": 0.00015059999999999997,
      "loss": 2.9993,
      "step": 255
    },
    {
      "epoch": 0.040075140889167186,
      "grad_norm": 1.0136637687683105,
      "learning_rate": 0.0001512,
      "loss": 2.9005,
      "step": 256
    },
    {
      "epoch": 0.040231684408265495,
      "grad_norm": 1.4269553422927856,
      "learning_rate": 0.00015179999999999998,
      "loss": 2.9279,
      "step": 257
    },
    {
      "epoch": 0.04038822792736381,
      "grad_norm": 1.676174283027649,
      "learning_rate": 0.0001524,
      "loss": 2.8742,
      "step": 258
    },
    {
      "epoch": 0.04054477144646212,
      "grad_norm": 9.41197395324707,
      "learning_rate": 0.00015299999999999998,
      "loss": 3.4308,
      "step": 259
    },
    {
      "epoch": 0.040701314965560426,
      "grad_norm": 3.7625269889831543,
      "learning_rate": 0.0001536,
      "loss": 2.9597,
      "step": 260
    },
    {
      "epoch": 0.040857858484658734,
      "grad_norm": 1.9438997507095337,
      "learning_rate": 0.00015419999999999998,
      "loss": 2.9289,
      "step": 261
    },
    {
      "epoch": 0.04101440200375704,
      "grad_norm": 1.4095935821533203,
      "learning_rate": 0.0001548,
      "loss": 2.8869,
      "step": 262
    },
    {
      "epoch": 0.04117094552285536,
      "grad_norm": 1.280892252922058,
      "learning_rate": 0.00015539999999999998,
      "loss": 2.9053,
      "step": 263
    },
    {
      "epoch": 0.041327489041953665,
      "grad_norm": 0.9870156049728394,
      "learning_rate": 0.000156,
      "loss": 2.8905,
      "step": 264
    },
    {
      "epoch": 0.04148403256105197,
      "grad_norm": 2.140601634979248,
      "learning_rate": 0.00015659999999999998,
      "loss": 2.9066,
      "step": 265
    },
    {
      "epoch": 0.04164057608015028,
      "grad_norm": 3.046438217163086,
      "learning_rate": 0.0001572,
      "loss": 2.9958,
      "step": 266
    },
    {
      "epoch": 0.04179711959924859,
      "grad_norm": 6.271746635437012,
      "learning_rate": 0.0001578,
      "loss": 3.2312,
      "step": 267
    },
    {
      "epoch": 0.041953663118346904,
      "grad_norm": 0.49755653738975525,
      "learning_rate": 0.0001584,
      "loss": 2.928,
      "step": 268
    },
    {
      "epoch": 0.04211020663744521,
      "grad_norm": 1.545334815979004,
      "learning_rate": 0.000159,
      "loss": 2.9341,
      "step": 269
    },
    {
      "epoch": 0.04226675015654352,
      "grad_norm": 0.7185292840003967,
      "learning_rate": 0.0001596,
      "loss": 2.9309,
      "step": 270
    },
    {
      "epoch": 0.04242329367564183,
      "grad_norm": 5.970613479614258,
      "learning_rate": 0.0001602,
      "loss": 3.1542,
      "step": 271
    },
    {
      "epoch": 0.042579837194740136,
      "grad_norm": 3.6000194549560547,
      "learning_rate": 0.0001608,
      "loss": 3.0439,
      "step": 272
    },
    {
      "epoch": 0.042736380713838444,
      "grad_norm": 3.2971136569976807,
      "learning_rate": 0.0001614,
      "loss": 2.982,
      "step": 273
    },
    {
      "epoch": 0.04289292423293676,
      "grad_norm": 1.305440068244934,
      "learning_rate": 0.000162,
      "loss": 3.0709,
      "step": 274
    },
    {
      "epoch": 0.04304946775203507,
      "grad_norm": 3.8561184406280518,
      "learning_rate": 0.0001626,
      "loss": 3.0722,
      "step": 275
    },
    {
      "epoch": 0.043206011271133375,
      "grad_norm": 4.944895267486572,
      "learning_rate": 0.0001632,
      "loss": 2.9751,
      "step": 276
    },
    {
      "epoch": 0.04336255479023168,
      "grad_norm": 5.420694351196289,
      "learning_rate": 0.0001638,
      "loss": 3.1554,
      "step": 277
    },
    {
      "epoch": 0.04351909830932999,
      "grad_norm": 1.4312067031860352,
      "learning_rate": 0.0001644,
      "loss": 3.1093,
      "step": 278
    },
    {
      "epoch": 0.043675641828428306,
      "grad_norm": 1.852827548980713,
      "learning_rate": 0.000165,
      "loss": 3.016,
      "step": 279
    },
    {
      "epoch": 0.043832185347526614,
      "grad_norm": 1.318656325340271,
      "learning_rate": 0.0001656,
      "loss": 2.9848,
      "step": 280
    },
    {
      "epoch": 0.04398872886662492,
      "grad_norm": 2.9079530239105225,
      "learning_rate": 0.0001662,
      "loss": 2.9815,
      "step": 281
    },
    {
      "epoch": 0.04414527238572323,
      "grad_norm": 2.3865318298339844,
      "learning_rate": 0.0001668,
      "loss": 2.9969,
      "step": 282
    },
    {
      "epoch": 0.04430181590482154,
      "grad_norm": 2.2731151580810547,
      "learning_rate": 0.0001674,
      "loss": 3.042,
      "step": 283
    },
    {
      "epoch": 0.044458359423919847,
      "grad_norm": 2.308046340942383,
      "learning_rate": 0.000168,
      "loss": 2.9983,
      "step": 284
    },
    {
      "epoch": 0.04461490294301816,
      "grad_norm": 2.4324429035186768,
      "learning_rate": 0.0001686,
      "loss": 3.0874,
      "step": 285
    },
    {
      "epoch": 0.04477144646211647,
      "grad_norm": 5.1173505783081055,
      "learning_rate": 0.00016919999999999997,
      "loss": 3.002,
      "step": 286
    },
    {
      "epoch": 0.04492798998121478,
      "grad_norm": 3.3832643032073975,
      "learning_rate": 0.00016979999999999998,
      "loss": 3.1237,
      "step": 287
    },
    {
      "epoch": 0.045084533500313086,
      "grad_norm": 1.6564899682998657,
      "learning_rate": 0.00017039999999999997,
      "loss": 3.1004,
      "step": 288
    },
    {
      "epoch": 0.045241077019411394,
      "grad_norm": 1.8179957866668701,
      "learning_rate": 0.00017099999999999998,
      "loss": 3.0704,
      "step": 289
    },
    {
      "epoch": 0.04539762053850971,
      "grad_norm": 1.3986287117004395,
      "learning_rate": 0.00017159999999999997,
      "loss": 3.088,
      "step": 290
    },
    {
      "epoch": 0.04555416405760802,
      "grad_norm": 1.6225603818893433,
      "learning_rate": 0.00017219999999999998,
      "loss": 2.9509,
      "step": 291
    },
    {
      "epoch": 0.045710707576706325,
      "grad_norm": 1.9594841003417969,
      "learning_rate": 0.00017279999999999997,
      "loss": 2.8125,
      "step": 292
    },
    {
      "epoch": 0.04586725109580463,
      "grad_norm": 2.098877191543579,
      "learning_rate": 0.00017339999999999996,
      "loss": 2.982,
      "step": 293
    },
    {
      "epoch": 0.04602379461490294,
      "grad_norm": 2.2931160926818848,
      "learning_rate": 0.00017399999999999997,
      "loss": 2.8949,
      "step": 294
    },
    {
      "epoch": 0.04618033813400125,
      "grad_norm": 1.8731071949005127,
      "learning_rate": 0.00017459999999999996,
      "loss": 3.0056,
      "step": 295
    },
    {
      "epoch": 0.046336881653099564,
      "grad_norm": 1.8290430307388306,
      "learning_rate": 0.00017519999999999998,
      "loss": 2.8271,
      "step": 296
    },
    {
      "epoch": 0.04649342517219787,
      "grad_norm": 2.112307548522949,
      "learning_rate": 0.00017579999999999996,
      "loss": 2.5751,
      "step": 297
    },
    {
      "epoch": 0.04664996869129618,
      "grad_norm": 1.703934907913208,
      "learning_rate": 0.00017639999999999998,
      "loss": 2.5481,
      "step": 298
    },
    {
      "epoch": 0.04680651221039449,
      "grad_norm": 2.3497684001922607,
      "learning_rate": 0.00017699999999999997,
      "loss": 2.3719,
      "step": 299
    },
    {
      "epoch": 0.046963055729492796,
      "grad_norm": 2.5405707359313965,
      "learning_rate": 0.00017759999999999998,
      "loss": 2.6147,
      "step": 300
    },
    {
      "epoch": 0.04711959924859111,
      "grad_norm": 41.07537841796875,
      "learning_rate": 0.00017819999999999997,
      "loss": 5.9212,
      "step": 301
    },
    {
      "epoch": 0.04727614276768942,
      "grad_norm": 3.55108904838562,
      "learning_rate": 0.00017879999999999998,
      "loss": 2.9271,
      "step": 302
    },
    {
      "epoch": 0.04743268628678773,
      "grad_norm": 16.451351165771484,
      "learning_rate": 0.00017939999999999997,
      "loss": 3.9858,
      "step": 303
    },
    {
      "epoch": 0.047589229805886035,
      "grad_norm": 1.0126781463623047,
      "learning_rate": 0.00017999999999999998,
      "loss": 2.9336,
      "step": 304
    },
    {
      "epoch": 0.04774577332498434,
      "grad_norm": 1.0213969945907593,
      "learning_rate": 0.00018059999999999997,
      "loss": 2.8967,
      "step": 305
    },
    {
      "epoch": 0.04790231684408266,
      "grad_norm": 2.1471173763275146,
      "learning_rate": 0.00018119999999999999,
      "loss": 2.8706,
      "step": 306
    },
    {
      "epoch": 0.048058860363180966,
      "grad_norm": 1.895684003829956,
      "learning_rate": 0.00018179999999999997,
      "loss": 2.9031,
      "step": 307
    },
    {
      "epoch": 0.048215403882279274,
      "grad_norm": 2.477410316467285,
      "learning_rate": 0.0001824,
      "loss": 2.9365,
      "step": 308
    },
    {
      "epoch": 0.04837194740137758,
      "grad_norm": 1.3221207857131958,
      "learning_rate": 0.00018299999999999998,
      "loss": 2.8607,
      "step": 309
    },
    {
      "epoch": 0.04852849092047589,
      "grad_norm": 1.6282726526260376,
      "learning_rate": 0.0001836,
      "loss": 2.9193,
      "step": 310
    },
    {
      "epoch": 0.0486850344395742,
      "grad_norm": 0.8280702829360962,
      "learning_rate": 0.00018419999999999998,
      "loss": 2.8905,
      "step": 311
    },
    {
      "epoch": 0.048841577958672514,
      "grad_norm": 1.868796706199646,
      "learning_rate": 0.0001848,
      "loss": 2.8843,
      "step": 312
    },
    {
      "epoch": 0.04899812147777082,
      "grad_norm": 2.179255723953247,
      "learning_rate": 0.00018539999999999998,
      "loss": 2.9584,
      "step": 313
    },
    {
      "epoch": 0.04915466499686913,
      "grad_norm": 1.0927103757858276,
      "learning_rate": 0.000186,
      "loss": 2.8541,
      "step": 314
    },
    {
      "epoch": 0.04931120851596744,
      "grad_norm": 1.2639646530151367,
      "learning_rate": 0.00018659999999999998,
      "loss": 2.8738,
      "step": 315
    },
    {
      "epoch": 0.049467752035065746,
      "grad_norm": 0.666633665561676,
      "learning_rate": 0.0001872,
      "loss": 2.9175,
      "step": 316
    },
    {
      "epoch": 0.04962429555416406,
      "grad_norm": 3.4493837356567383,
      "learning_rate": 0.00018779999999999998,
      "loss": 2.9048,
      "step": 317
    },
    {
      "epoch": 0.04978083907326237,
      "grad_norm": 0.9418443441390991,
      "learning_rate": 0.00018839999999999997,
      "loss": 2.8725,
      "step": 318
    },
    {
      "epoch": 0.04993738259236068,
      "grad_norm": 1.8465272188186646,
      "learning_rate": 0.00018899999999999999,
      "loss": 2.8598,
      "step": 319
    },
    {
      "epoch": 0.050093926111458985,
      "grad_norm": 1.2343814373016357,
      "learning_rate": 0.00018959999999999997,
      "loss": 2.9232,
      "step": 320
    },
    {
      "epoch": 0.05025046963055729,
      "grad_norm": 1.1628434658050537,
      "learning_rate": 0.0001902,
      "loss": 2.9062,
      "step": 321
    },
    {
      "epoch": 0.0504070131496556,
      "grad_norm": 3.26182222366333,
      "learning_rate": 0.00019079999999999998,
      "loss": 2.993,
      "step": 322
    },
    {
      "epoch": 0.050563556668753916,
      "grad_norm": 2.6465306282043457,
      "learning_rate": 0.0001914,
      "loss": 2.9229,
      "step": 323
    },
    {
      "epoch": 0.050720100187852224,
      "grad_norm": 1.6786047220230103,
      "learning_rate": 0.00019199999999999998,
      "loss": 2.9405,
      "step": 324
    },
    {
      "epoch": 0.05087664370695053,
      "grad_norm": 2.313570022583008,
      "learning_rate": 0.0001926,
      "loss": 2.8359,
      "step": 325
    },
    {
      "epoch": 0.05103318722604884,
      "grad_norm": 1.3497850894927979,
      "learning_rate": 0.00019319999999999998,
      "loss": 2.9089,
      "step": 326
    },
    {
      "epoch": 0.05118973074514715,
      "grad_norm": 0.9331809282302856,
      "learning_rate": 0.0001938,
      "loss": 2.8491,
      "step": 327
    },
    {
      "epoch": 0.05134627426424546,
      "grad_norm": 1.8141287565231323,
      "learning_rate": 0.00019439999999999998,
      "loss": 2.897,
      "step": 328
    },
    {
      "epoch": 0.05150281778334377,
      "grad_norm": 1.0781328678131104,
      "learning_rate": 0.000195,
      "loss": 2.8376,
      "step": 329
    },
    {
      "epoch": 0.05165936130244208,
      "grad_norm": 0.9681340456008911,
      "learning_rate": 0.00019559999999999998,
      "loss": 2.8983,
      "step": 330
    },
    {
      "epoch": 0.05181590482154039,
      "grad_norm": 4.107593059539795,
      "learning_rate": 0.0001962,
      "loss": 2.8991,
      "step": 331
    },
    {
      "epoch": 0.051972448340638695,
      "grad_norm": 2.0109505653381348,
      "learning_rate": 0.00019679999999999999,
      "loss": 2.8443,
      "step": 332
    },
    {
      "epoch": 0.05212899185973701,
      "grad_norm": 3.9468677043914795,
      "learning_rate": 0.0001974,
      "loss": 2.8434,
      "step": 333
    },
    {
      "epoch": 0.05228553537883532,
      "grad_norm": 1.277677059173584,
      "learning_rate": 0.000198,
      "loss": 2.7809,
      "step": 334
    },
    {
      "epoch": 0.052442078897933626,
      "grad_norm": 1.9392768144607544,
      "learning_rate": 0.0001986,
      "loss": 2.9161,
      "step": 335
    },
    {
      "epoch": 0.052598622417031934,
      "grad_norm": 3.3220739364624023,
      "learning_rate": 0.0001992,
      "loss": 2.7724,
      "step": 336
    },
    {
      "epoch": 0.05275516593613024,
      "grad_norm": 2.631103754043579,
      "learning_rate": 0.0001998,
      "loss": 2.8016,
      "step": 337
    },
    {
      "epoch": 0.05291170945522855,
      "grad_norm": 2.869077444076538,
      "learning_rate": 0.0002004,
      "loss": 2.8939,
      "step": 338
    },
    {
      "epoch": 0.053068252974326866,
      "grad_norm": 1.6763360500335693,
      "learning_rate": 0.000201,
      "loss": 2.7951,
      "step": 339
    },
    {
      "epoch": 0.053224796493425174,
      "grad_norm": 1.35333251953125,
      "learning_rate": 0.0002016,
      "loss": 2.6876,
      "step": 340
    },
    {
      "epoch": 0.05338134001252348,
      "grad_norm": 1.591059684753418,
      "learning_rate": 0.0002022,
      "loss": 2.7753,
      "step": 341
    },
    {
      "epoch": 0.05353788353162179,
      "grad_norm": 2.1885745525360107,
      "learning_rate": 0.0002028,
      "loss": 2.9564,
      "step": 342
    },
    {
      "epoch": 0.0536944270507201,
      "grad_norm": 6.893604755401611,
      "learning_rate": 0.00020339999999999998,
      "loss": 3.0515,
      "step": 343
    },
    {
      "epoch": 0.05385097056981841,
      "grad_norm": 4.528074264526367,
      "learning_rate": 0.000204,
      "loss": 2.6812,
      "step": 344
    },
    {
      "epoch": 0.05400751408891672,
      "grad_norm": 1.5735338926315308,
      "learning_rate": 0.00020459999999999999,
      "loss": 2.6635,
      "step": 345
    },
    {
      "epoch": 0.05416405760801503,
      "grad_norm": 2.046394109725952,
      "learning_rate": 0.0002052,
      "loss": 2.5343,
      "step": 346
    },
    {
      "epoch": 0.05432060112711334,
      "grad_norm": 1.8369107246398926,
      "learning_rate": 0.0002058,
      "loss": 2.4338,
      "step": 347
    },
    {
      "epoch": 0.054477144646211645,
      "grad_norm": 1.8167654275894165,
      "learning_rate": 0.00020639999999999998,
      "loss": 2.2824,
      "step": 348
    },
    {
      "epoch": 0.05463368816530995,
      "grad_norm": 3.679330825805664,
      "learning_rate": 0.00020699999999999996,
      "loss": 2.6324,
      "step": 349
    },
    {
      "epoch": 0.05479023168440827,
      "grad_norm": 3.2513327598571777,
      "learning_rate": 0.00020759999999999998,
      "loss": 2.2228,
      "step": 350
    },
    {
      "epoch": 0.054946775203506576,
      "grad_norm": 3.872000217437744,
      "learning_rate": 0.00020819999999999996,
      "loss": 2.9826,
      "step": 351
    },
    {
      "epoch": 0.055103318722604884,
      "grad_norm": 4.465301036834717,
      "learning_rate": 0.00020879999999999998,
      "loss": 3.3657,
      "step": 352
    },
    {
      "epoch": 0.05525986224170319,
      "grad_norm": 7.596004486083984,
      "learning_rate": 0.00020939999999999997,
      "loss": 3.4431,
      "step": 353
    },
    {
      "epoch": 0.0554164057608015,
      "grad_norm": 0.8127701282501221,
      "learning_rate": 0.00020999999999999998,
      "loss": 2.8607,
      "step": 354
    },
    {
      "epoch": 0.055572949279899815,
      "grad_norm": 3.6727442741394043,
      "learning_rate": 0.00021059999999999997,
      "loss": 2.9144,
      "step": 355
    },
    {
      "epoch": 0.05572949279899812,
      "grad_norm": 3.5418031215667725,
      "learning_rate": 0.00021119999999999996,
      "loss": 2.9496,
      "step": 356
    },
    {
      "epoch": 0.05588603631809643,
      "grad_norm": 0.981769323348999,
      "learning_rate": 0.00021179999999999997,
      "loss": 2.8433,
      "step": 357
    },
    {
      "epoch": 0.05604257983719474,
      "grad_norm": 0.742219090461731,
      "learning_rate": 0.00021239999999999996,
      "loss": 2.8561,
      "step": 358
    },
    {
      "epoch": 0.05619912335629305,
      "grad_norm": 14.127908706665039,
      "learning_rate": 0.00021299999999999997,
      "loss": 3.9582,
      "step": 359
    },
    {
      "epoch": 0.056355666875391355,
      "grad_norm": 1.440157175064087,
      "learning_rate": 0.00021359999999999996,
      "loss": 2.9154,
      "step": 360
    },
    {
      "epoch": 0.05651221039448967,
      "grad_norm": 12.539861679077148,
      "learning_rate": 0.00021419999999999998,
      "loss": 3.58,
      "step": 361
    },
    {
      "epoch": 0.05666875391358798,
      "grad_norm": 0.9830080270767212,
      "learning_rate": 0.00021479999999999996,
      "loss": 3.0251,
      "step": 362
    },
    {
      "epoch": 0.056825297432686286,
      "grad_norm": 0.5790128707885742,
      "learning_rate": 0.00021539999999999998,
      "loss": 2.8486,
      "step": 363
    },
    {
      "epoch": 0.056981840951784594,
      "grad_norm": 4.547713279724121,
      "learning_rate": 0.00021599999999999996,
      "loss": 3.0627,
      "step": 364
    },
    {
      "epoch": 0.0571383844708829,
      "grad_norm": 1.8868635892868042,
      "learning_rate": 0.00021659999999999998,
      "loss": 2.8365,
      "step": 365
    },
    {
      "epoch": 0.05729492798998122,
      "grad_norm": 1.4860725402832031,
      "learning_rate": 0.00021719999999999997,
      "loss": 2.8774,
      "step": 366
    },
    {
      "epoch": 0.057451471509079526,
      "grad_norm": 1.064705729484558,
      "learning_rate": 0.00021779999999999998,
      "loss": 2.9113,
      "step": 367
    },
    {
      "epoch": 0.057608015028177834,
      "grad_norm": 1.752041220664978,
      "learning_rate": 0.00021839999999999997,
      "loss": 2.9108,
      "step": 368
    },
    {
      "epoch": 0.05776455854727614,
      "grad_norm": 0.8162034153938293,
      "learning_rate": 0.00021899999999999998,
      "loss": 2.8598,
      "step": 369
    },
    {
      "epoch": 0.05792110206637445,
      "grad_norm": 0.7272463440895081,
      "learning_rate": 0.00021959999999999997,
      "loss": 2.856,
      "step": 370
    },
    {
      "epoch": 0.058077645585472765,
      "grad_norm": 0.9358817338943481,
      "learning_rate": 0.00022019999999999999,
      "loss": 2.8943,
      "step": 371
    },
    {
      "epoch": 0.05823418910457107,
      "grad_norm": 0.799842119216919,
      "learning_rate": 0.00022079999999999997,
      "loss": 2.822,
      "step": 372
    },
    {
      "epoch": 0.05839073262366938,
      "grad_norm": 2.1993589401245117,
      "learning_rate": 0.0002214,
      "loss": 2.9312,
      "step": 373
    },
    {
      "epoch": 0.05854727614276769,
      "grad_norm": 7.407040596008301,
      "learning_rate": 0.00022199999999999998,
      "loss": 3.2528,
      "step": 374
    },
    {
      "epoch": 0.058703819661866,
      "grad_norm": 0.9808345437049866,
      "learning_rate": 0.0002226,
      "loss": 2.8877,
      "step": 375
    },
    {
      "epoch": 0.058860363180964305,
      "grad_norm": 1.277418613433838,
      "learning_rate": 0.00022319999999999998,
      "loss": 2.779,
      "step": 376
    },
    {
      "epoch": 0.05901690670006262,
      "grad_norm": 1.093755841255188,
      "learning_rate": 0.0002238,
      "loss": 2.8769,
      "step": 377
    },
    {
      "epoch": 0.05917345021916093,
      "grad_norm": 2.1104671955108643,
      "learning_rate": 0.00022439999999999998,
      "loss": 2.8662,
      "step": 378
    },
    {
      "epoch": 0.059329993738259236,
      "grad_norm": 1.0569154024124146,
      "learning_rate": 0.000225,
      "loss": 2.8349,
      "step": 379
    },
    {
      "epoch": 0.059486537257357544,
      "grad_norm": 1.0086591243743896,
      "learning_rate": 0.00022559999999999998,
      "loss": 2.8008,
      "step": 380
    },
    {
      "epoch": 0.05964308077645585,
      "grad_norm": 2.0983245372772217,
      "learning_rate": 0.00022619999999999997,
      "loss": 2.7307,
      "step": 381
    },
    {
      "epoch": 0.05979962429555417,
      "grad_norm": 1.8184118270874023,
      "learning_rate": 0.00022679999999999998,
      "loss": 2.8152,
      "step": 382
    },
    {
      "epoch": 0.059956167814652475,
      "grad_norm": 1.5371543169021606,
      "learning_rate": 0.00022739999999999997,
      "loss": 2.6611,
      "step": 383
    },
    {
      "epoch": 0.06011271133375078,
      "grad_norm": 2.030109405517578,
      "learning_rate": 0.00022799999999999999,
      "loss": 2.9117,
      "step": 384
    },
    {
      "epoch": 0.06026925485284909,
      "grad_norm": 1.1472651958465576,
      "learning_rate": 0.00022859999999999997,
      "loss": 2.7593,
      "step": 385
    },
    {
      "epoch": 0.0604257983719474,
      "grad_norm": 2.547860622406006,
      "learning_rate": 0.0002292,
      "loss": 2.9494,
      "step": 386
    },
    {
      "epoch": 0.06058234189104571,
      "grad_norm": 4.287590980529785,
      "learning_rate": 0.00022979999999999997,
      "loss": 2.7367,
      "step": 387
    },
    {
      "epoch": 0.06073888541014402,
      "grad_norm": 2.5711324214935303,
      "learning_rate": 0.0002304,
      "loss": 2.8195,
      "step": 388
    },
    {
      "epoch": 0.06089542892924233,
      "grad_norm": 1.1869808435440063,
      "learning_rate": 0.00023099999999999998,
      "loss": 2.6613,
      "step": 389
    },
    {
      "epoch": 0.06105197244834064,
      "grad_norm": 2.388160467147827,
      "learning_rate": 0.0002316,
      "loss": 2.7794,
      "step": 390
    },
    {
      "epoch": 0.061208515967438946,
      "grad_norm": 2.665323257446289,
      "learning_rate": 0.00023219999999999998,
      "loss": 2.9327,
      "step": 391
    },
    {
      "epoch": 0.061365059486537255,
      "grad_norm": 1.2686117887496948,
      "learning_rate": 0.0002328,
      "loss": 2.7387,
      "step": 392
    },
    {
      "epoch": 0.06152160300563557,
      "grad_norm": 2.919185161590576,
      "learning_rate": 0.00023339999999999998,
      "loss": 2.6176,
      "step": 393
    },
    {
      "epoch": 0.06167814652473388,
      "grad_norm": 1.6040527820587158,
      "learning_rate": 0.000234,
      "loss": 2.4814,
      "step": 394
    },
    {
      "epoch": 0.061834690043832186,
      "grad_norm": 1.9547075033187866,
      "learning_rate": 0.00023459999999999998,
      "loss": 2.3291,
      "step": 395
    },
    {
      "epoch": 0.061991233562930494,
      "grad_norm": 2.361668109893799,
      "learning_rate": 0.0002352,
      "loss": 2.3533,
      "step": 396
    },
    {
      "epoch": 0.0621477770820288,
      "grad_norm": 3.003882646560669,
      "learning_rate": 0.00023579999999999999,
      "loss": 2.1218,
      "step": 397
    },
    {
      "epoch": 0.06230432060112712,
      "grad_norm": 2.0260565280914307,
      "learning_rate": 0.0002364,
      "loss": 2.3472,
      "step": 398
    },
    {
      "epoch": 0.062460864120225425,
      "grad_norm": 1.2082924842834473,
      "learning_rate": 0.000237,
      "loss": 2.2098,
      "step": 399
    },
    {
      "epoch": 0.06261740763932373,
      "grad_norm": 2.548179864883423,
      "learning_rate": 0.0002376,
      "loss": 2.006,
      "step": 400
    },
    {
      "epoch": 0.06277395115842205,
      "grad_norm": 25.11155128479004,
      "learning_rate": 0.0002382,
      "loss": 5.0331,
      "step": 401
    },
    {
      "epoch": 0.06293049467752035,
      "grad_norm": 7.020974159240723,
      "learning_rate": 0.0002388,
      "loss": 3.1208,
      "step": 402
    },
    {
      "epoch": 0.06308703819661866,
      "grad_norm": 5.667887210845947,
      "learning_rate": 0.0002394,
      "loss": 3.0382,
      "step": 403
    },
    {
      "epoch": 0.06324358171571696,
      "grad_norm": 1.9799600839614868,
      "learning_rate": 0.00023999999999999998,
      "loss": 2.8571,
      "step": 404
    },
    {
      "epoch": 0.06340012523481528,
      "grad_norm": 2.0452182292938232,
      "learning_rate": 0.0002406,
      "loss": 2.8546,
      "step": 405
    },
    {
      "epoch": 0.06355666875391358,
      "grad_norm": 3.450450897216797,
      "learning_rate": 0.00024119999999999998,
      "loss": 2.9338,
      "step": 406
    },
    {
      "epoch": 0.0637132122730119,
      "grad_norm": 3.7206132411956787,
      "learning_rate": 0.0002418,
      "loss": 2.8975,
      "step": 407
    },
    {
      "epoch": 0.06386975579211021,
      "grad_norm": 2.4928946495056152,
      "learning_rate": 0.00024239999999999998,
      "loss": 2.8855,
      "step": 408
    },
    {
      "epoch": 0.06402629931120851,
      "grad_norm": 1.6292774677276611,
      "learning_rate": 0.000243,
      "loss": 2.8278,
      "step": 409
    },
    {
      "epoch": 0.06418284283030683,
      "grad_norm": 1.6044974327087402,
      "learning_rate": 0.00024359999999999999,
      "loss": 2.8089,
      "step": 410
    },
    {
      "epoch": 0.06433938634940513,
      "grad_norm": 3.7000677585601807,
      "learning_rate": 0.00024419999999999997,
      "loss": 2.97,
      "step": 411
    },
    {
      "epoch": 0.06449592986850344,
      "grad_norm": 12.418329238891602,
      "learning_rate": 0.0002448,
      "loss": 3.5765,
      "step": 412
    },
    {
      "epoch": 0.06465247338760176,
      "grad_norm": 3.708932638168335,
      "learning_rate": 0.00024539999999999995,
      "loss": 2.889,
      "step": 413
    },
    {
      "epoch": 0.06480901690670006,
      "grad_norm": 2.5494682788848877,
      "learning_rate": 0.00024599999999999996,
      "loss": 2.8552,
      "step": 414
    },
    {
      "epoch": 0.06496556042579837,
      "grad_norm": 0.5013105273246765,
      "learning_rate": 0.0002466,
      "loss": 2.8204,
      "step": 415
    },
    {
      "epoch": 0.06512210394489668,
      "grad_norm": 0.6971216201782227,
      "learning_rate": 0.0002472,
      "loss": 2.7921,
      "step": 416
    },
    {
      "epoch": 0.06527864746399499,
      "grad_norm": 1.244098424911499,
      "learning_rate": 0.00024779999999999995,
      "loss": 2.8253,
      "step": 417
    },
    {
      "epoch": 0.0654351909830933,
      "grad_norm": 1.136309027671814,
      "learning_rate": 0.00024839999999999997,
      "loss": 2.7789,
      "step": 418
    },
    {
      "epoch": 0.0655917345021916,
      "grad_norm": 1.103389859199524,
      "learning_rate": 0.000249,
      "loss": 2.7981,
      "step": 419
    },
    {
      "epoch": 0.06574827802128992,
      "grad_norm": 0.6065533757209778,
      "learning_rate": 0.00024959999999999994,
      "loss": 2.777,
      "step": 420
    },
    {
      "epoch": 0.06590482154038822,
      "grad_norm": 1.2943936586380005,
      "learning_rate": 0.00025019999999999996,
      "loss": 2.779,
      "step": 421
    },
    {
      "epoch": 0.06606136505948654,
      "grad_norm": 1.5812445878982544,
      "learning_rate": 0.00025079999999999997,
      "loss": 2.7543,
      "step": 422
    },
    {
      "epoch": 0.06621790857858485,
      "grad_norm": 0.6058809757232666,
      "learning_rate": 0.0002514,
      "loss": 2.7617,
      "step": 423
    },
    {
      "epoch": 0.06637445209768315,
      "grad_norm": 0.6701329946517944,
      "learning_rate": 0.00025199999999999995,
      "loss": 2.776,
      "step": 424
    },
    {
      "epoch": 0.06653099561678147,
      "grad_norm": 1.4818625450134277,
      "learning_rate": 0.00025259999999999996,
      "loss": 2.8081,
      "step": 425
    },
    {
      "epoch": 0.06668753913587977,
      "grad_norm": 0.9696059226989746,
      "learning_rate": 0.0002532,
      "loss": 2.7473,
      "step": 426
    },
    {
      "epoch": 0.06684408265497808,
      "grad_norm": 0.6371885538101196,
      "learning_rate": 0.0002538,
      "loss": 2.7107,
      "step": 427
    },
    {
      "epoch": 0.0670006261740764,
      "grad_norm": 2.670564889907837,
      "learning_rate": 0.00025439999999999995,
      "loss": 2.7509,
      "step": 428
    },
    {
      "epoch": 0.0671571696931747,
      "grad_norm": 1.7934705018997192,
      "learning_rate": 0.00025499999999999996,
      "loss": 2.5508,
      "step": 429
    },
    {
      "epoch": 0.06731371321227302,
      "grad_norm": 1.4174633026123047,
      "learning_rate": 0.0002556,
      "loss": 2.7541,
      "step": 430
    },
    {
      "epoch": 0.06747025673137132,
      "grad_norm": 1.3677154779434204,
      "learning_rate": 0.0002562,
      "loss": 2.6337,
      "step": 431
    },
    {
      "epoch": 0.06762680025046963,
      "grad_norm": 3.6003217697143555,
      "learning_rate": 0.00025679999999999995,
      "loss": 2.6692,
      "step": 432
    },
    {
      "epoch": 0.06778334376956793,
      "grad_norm": 1.2599115371704102,
      "learning_rate": 0.00025739999999999997,
      "loss": 2.5543,
      "step": 433
    },
    {
      "epoch": 0.06793988728866625,
      "grad_norm": 0.8455353379249573,
      "learning_rate": 0.000258,
      "loss": 2.6359,
      "step": 434
    },
    {
      "epoch": 0.06809643080776456,
      "grad_norm": 0.9858604073524475,
      "learning_rate": 0.0002586,
      "loss": 2.7305,
      "step": 435
    },
    {
      "epoch": 0.06825297432686286,
      "grad_norm": 1.948320984840393,
      "learning_rate": 0.00025919999999999996,
      "loss": 2.7262,
      "step": 436
    },
    {
      "epoch": 0.06840951784596118,
      "grad_norm": 1.0108580589294434,
      "learning_rate": 0.00025979999999999997,
      "loss": 2.7447,
      "step": 437
    },
    {
      "epoch": 0.06856606136505948,
      "grad_norm": 1.686102032661438,
      "learning_rate": 0.0002604,
      "loss": 2.6928,
      "step": 438
    },
    {
      "epoch": 0.0687226048841578,
      "grad_norm": 1.4326525926589966,
      "learning_rate": 0.000261,
      "loss": 2.4659,
      "step": 439
    },
    {
      "epoch": 0.06887914840325611,
      "grad_norm": NaN,
      "learning_rate": 0.000261,
      "loss": 0.0,
      "step": 440
    },
    {
      "epoch": 0.06903569192235441,
      "grad_norm": 3.392507791519165,
      "learning_rate": 0.00026159999999999996,
      "loss": 2.502,
      "step": 441
    },
    {
      "epoch": 0.06919223544145273,
      "grad_norm": 1.325119137763977,
      "learning_rate": 0.0002622,
      "loss": 2.3608,
      "step": 442
    },
    {
      "epoch": 0.06934877896055103,
      "grad_norm": 2.0018181800842285,
      "learning_rate": 0.0002628,
      "loss": 2.7061,
      "step": 443
    },
    {
      "epoch": 0.06950532247964934,
      "grad_norm": 2.351816415786743,
      "learning_rate": 0.00026339999999999995,
      "loss": 2.0728,
      "step": 444
    },
    {
      "epoch": 0.06966186599874766,
      "grad_norm": 2.392810583114624,
      "learning_rate": 0.00026399999999999997,
      "loss": 2.4649,
      "step": 445
    },
    {
      "epoch": 0.06981840951784596,
      "grad_norm": 2.2967441082000732,
      "learning_rate": 0.0002646,
      "loss": 1.9439,
      "step": 446
    },
    {
      "epoch": 0.06997495303694427,
      "grad_norm": 1.640772819519043,
      "learning_rate": 0.0002652,
      "loss": 2.0041,
      "step": 447
    },
    {
      "epoch": 0.07013149655604257,
      "grad_norm": 2.1847622394561768,
      "learning_rate": 0.00026579999999999996,
      "loss": 1.9843,
      "step": 448
    },
    {
      "epoch": 0.07028804007514089,
      "grad_norm": 1.9607410430908203,
      "learning_rate": 0.00026639999999999997,
      "loss": 2.0595,
      "step": 449
    },
    {
      "epoch": 0.0704445835942392,
      "grad_norm": 1.476098656654358,
      "learning_rate": 0.000267,
      "loss": 1.9382,
      "step": 450
    },
    {
      "epoch": 0.0706011271133375,
      "grad_norm": 5.004290580749512,
      "learning_rate": 0.0002676,
      "loss": 3.0205,
      "step": 451
    },
    {
      "epoch": 0.07075767063243582,
      "grad_norm": 11.529528617858887,
      "learning_rate": 0.00026819999999999996,
      "loss": 3.5319,
      "step": 452
    },
    {
      "epoch": 0.07091421415153412,
      "grad_norm": 1.561353325843811,
      "learning_rate": 0.0002688,
      "loss": 2.7755,
      "step": 453
    },
    {
      "epoch": 0.07107075767063244,
      "grad_norm": 1.1581677198410034,
      "learning_rate": 0.0002694,
      "loss": 2.7657,
      "step": 454
    },
    {
      "epoch": 0.07122730118973075,
      "grad_norm": 0.837934136390686,
      "learning_rate": 0.00027,
      "loss": 2.6956,
      "step": 455
    },
    {
      "epoch": 0.07138384470882905,
      "grad_norm": 1.2540568113327026,
      "learning_rate": 0.00027059999999999996,
      "loss": 2.5976,
      "step": 456
    },
    {
      "epoch": 0.07154038822792737,
      "grad_norm": 1.265945315361023,
      "learning_rate": 0.0002712,
      "loss": 2.5529,
      "step": 457
    },
    {
      "epoch": 0.07169693174702567,
      "grad_norm": 0.9604879021644592,
      "learning_rate": 0.0002718,
      "loss": 2.5779,
      "step": 458
    },
    {
      "epoch": 0.07185347526612398,
      "grad_norm": 0.6527476906776428,
      "learning_rate": 0.0002724,
      "loss": 2.6235,
      "step": 459
    },
    {
      "epoch": 0.07201001878522229,
      "grad_norm": 0.6187277436256409,
      "learning_rate": 0.00027299999999999997,
      "loss": 2.4643,
      "step": 460
    },
    {
      "epoch": 0.0721665623043206,
      "grad_norm": 0.8495888710021973,
      "learning_rate": 0.0002736,
      "loss": 2.4603,
      "step": 461
    },
    {
      "epoch": 0.07232310582341892,
      "grad_norm": 0.812981128692627,
      "learning_rate": 0.0002742,
      "loss": 2.4654,
      "step": 462
    },
    {
      "epoch": 0.07247964934251722,
      "grad_norm": 3.680403709411621,
      "learning_rate": 0.0002748,
      "loss": 2.7148,
      "step": 463
    },
    {
      "epoch": 0.07263619286161553,
      "grad_norm": 0.837788462638855,
      "learning_rate": 0.00027539999999999997,
      "loss": 2.4459,
      "step": 464
    },
    {
      "epoch": 0.07279273638071383,
      "grad_norm": 0.9086034893989563,
      "learning_rate": 0.000276,
      "loss": 2.3291,
      "step": 465
    },
    {
      "epoch": 0.07294927989981215,
      "grad_norm": 1.12948739528656,
      "learning_rate": 0.0002766,
      "loss": 2.4085,
      "step": 466
    },
    {
      "epoch": 0.07310582341891046,
      "grad_norm": 0.9835497140884399,
      "learning_rate": 0.0002772,
      "loss": 2.2966,
      "step": 467
    },
    {
      "epoch": 0.07326236693800876,
      "grad_norm": 2.1482276916503906,
      "learning_rate": 0.0002778,
      "loss": 2.3619,
      "step": 468
    },
    {
      "epoch": 0.07341891045710708,
      "grad_norm": 3.3188507556915283,
      "learning_rate": 0.0002784,
      "loss": 2.3791,
      "step": 469
    },
    {
      "epoch": 0.07357545397620538,
      "grad_norm": 1.2607046365737915,
      "learning_rate": 0.000279,
      "loss": 2.3225,
      "step": 470
    },
    {
      "epoch": 0.0737319974953037,
      "grad_norm": 3.159202814102173,
      "learning_rate": 0.00027959999999999997,
      "loss": 2.4383,
      "step": 471
    },
    {
      "epoch": 0.07388854101440201,
      "grad_norm": 1.8510031700134277,
      "learning_rate": 0.0002802,
      "loss": 2.3881,
      "step": 472
    },
    {
      "epoch": 0.07404508453350031,
      "grad_norm": 1.1104274988174438,
      "learning_rate": 0.0002808,
      "loss": 2.2954,
      "step": 473
    },
    {
      "epoch": 0.07420162805259863,
      "grad_norm": 1.1513651609420776,
      "learning_rate": 0.00028139999999999996,
      "loss": 2.2092,
      "step": 474
    },
    {
      "epoch": 0.07435817157169693,
      "grad_norm": 1.2425847053527832,
      "learning_rate": 0.00028199999999999997,
      "loss": 2.2104,
      "step": 475
    },
    {
      "epoch": 0.07451471509079524,
      "grad_norm": 1.6782479286193848,
      "learning_rate": 0.0002826,
      "loss": 2.4311,
      "step": 476
    },
    {
      "epoch": 0.07467125860989356,
      "grad_norm": 3.411275625228882,
      "learning_rate": 0.00028319999999999994,
      "loss": 2.3067,
      "step": 477
    },
    {
      "epoch": 0.07482780212899186,
      "grad_norm": 1.4562594890594482,
      "learning_rate": 0.00028379999999999996,
      "loss": 2.2745,
      "step": 478
    },
    {
      "epoch": 0.07498434564809017,
      "grad_norm": 1.7229201793670654,
      "learning_rate": 0.0002844,
      "loss": 2.2636,
      "step": 479
    },
    {
      "epoch": 0.07514088916718847,
      "grad_norm": 1.804953932762146,
      "learning_rate": 0.000285,
      "loss": 2.2304,
      "step": 480
    },
    {
      "epoch": 0.07529743268628679,
      "grad_norm": 6.132996559143066,
      "learning_rate": 0.00028559999999999995,
      "loss": 2.9409,
      "step": 481
    },
    {
      "epoch": 0.0754539762053851,
      "grad_norm": 1.6565462350845337,
      "learning_rate": 0.00028619999999999996,
      "loss": 2.2299,
      "step": 482
    },
    {
      "epoch": 0.0756105197244834,
      "grad_norm": 1.50054132938385,
      "learning_rate": 0.0002868,
      "loss": 2.3357,
      "step": 483
    },
    {
      "epoch": 0.07576706324358172,
      "grad_norm": 2.0970003604888916,
      "learning_rate": 0.00028739999999999994,
      "loss": 2.4912,
      "step": 484
    },
    {
      "epoch": 0.07592360676268002,
      "grad_norm": 1.7060633897781372,
      "learning_rate": 0.00028799999999999995,
      "loss": 2.2346,
      "step": 485
    },
    {
      "epoch": 0.07608015028177834,
      "grad_norm": 3.9570698738098145,
      "learning_rate": 0.00028859999999999997,
      "loss": 2.2125,
      "step": 486
    },
    {
      "epoch": 0.07623669380087664,
      "grad_norm": 3.4521429538726807,
      "learning_rate": 0.0002892,
      "loss": 2.4241,
      "step": 487
    },
    {
      "epoch": 0.07639323731997495,
      "grad_norm": 2.9334945678710938,
      "learning_rate": 0.00028979999999999994,
      "loss": 2.4198,
      "step": 488
    },
    {
      "epoch": 0.07654978083907327,
      "grad_norm": 2.5098342895507812,
      "learning_rate": 0.00029039999999999996,
      "loss": 2.3674,
      "step": 489
    },
    {
      "epoch": 0.07670632435817157,
      "grad_norm": 2.8507261276245117,
      "learning_rate": 0.00029099999999999997,
      "loss": 2.1336,
      "step": 490
    },
    {
      "epoch": 0.07686286787726988,
      "grad_norm": 2.2666025161743164,
      "learning_rate": 0.0002916,
      "loss": 2.3893,
      "step": 491
    },
    {
      "epoch": 0.07701941139636818,
      "grad_norm": 2.3513152599334717,
      "learning_rate": 0.00029219999999999995,
      "loss": 1.7857,
      "step": 492
    },
    {
      "epoch": 0.0771759549154665,
      "grad_norm": 2.0074493885040283,
      "learning_rate": 0.00029279999999999996,
      "loss": 2.3626,
      "step": 493
    },
    {
      "epoch": 0.07733249843456481,
      "grad_norm": 1.9822458028793335,
      "learning_rate": 0.0002934,
      "loss": 2.0882,
      "step": 494
    },
    {
      "epoch": 0.07748904195366312,
      "grad_norm": 3.779900074005127,
      "learning_rate": 0.000294,
      "loss": 2.1448,
      "step": 495
    },
    {
      "epoch": 0.07764558547276143,
      "grad_norm": 2.198042869567871,
      "learning_rate": 0.00029459999999999995,
      "loss": 1.6344,
      "step": 496
    },
    {
      "epoch": 0.07780212899185973,
      "grad_norm": 1.7770249843597412,
      "learning_rate": 0.00029519999999999997,
      "loss": 1.8025,
      "step": 497
    },
    {
      "epoch": 0.07795867251095805,
      "grad_norm": 3.967764139175415,
      "learning_rate": 0.0002958,
      "loss": 1.6183,
      "step": 498
    },
    {
      "epoch": 0.07811521603005636,
      "grad_norm": 3.88845157623291,
      "learning_rate": 0.0002964,
      "loss": 1.6993,
      "step": 499
    },
    {
      "epoch": 0.07827175954915466,
      "grad_norm": 4.8480963706970215,
      "learning_rate": 0.00029699999999999996,
      "loss": 1.5443,
      "step": 500
    },
    {
      "epoch": 0.07842830306825298,
      "grad_norm": 1.4311107397079468,
      "learning_rate": 0.00029759999999999997,
      "loss": 2.4391,
      "step": 501
    },
    {
      "epoch": 0.07858484658735128,
      "grad_norm": 1.0577871799468994,
      "learning_rate": 0.0002982,
      "loss": 2.1257,
      "step": 502
    },
    {
      "epoch": 0.0787413901064496,
      "grad_norm": 0.7990356087684631,
      "learning_rate": 0.0002988,
      "loss": 2.0302,
      "step": 503
    },
    {
      "epoch": 0.07889793362554791,
      "grad_norm": 0.9843720197677612,
      "learning_rate": 0.00029939999999999996,
      "loss": 2.0466,
      "step": 504
    },
    {
      "epoch": 0.07905447714464621,
      "grad_norm": 0.8615551590919495,
      "learning_rate": 0.0003,
      "loss": 1.7693,
      "step": 505
    },
    {
      "epoch": 0.07921102066374452,
      "grad_norm": 1.1965469121932983,
      "learning_rate": 0.0002999755620723362,
      "loss": 1.7733,
      "step": 506
    },
    {
      "epoch": 0.07936756418284283,
      "grad_norm": 1.6512717008590698,
      "learning_rate": 0.0002999511241446725,
      "loss": 1.8785,
      "step": 507
    },
    {
      "epoch": 0.07952410770194114,
      "grad_norm": 1.4600224494934082,
      "learning_rate": 0.0002999266862170088,
      "loss": 1.807,
      "step": 508
    },
    {
      "epoch": 0.07968065122103946,
      "grad_norm": 0.969097375869751,
      "learning_rate": 0.00029990224828934503,
      "loss": 1.6916,
      "step": 509
    },
    {
      "epoch": 0.07983719474013776,
      "grad_norm": 10.695720672607422,
      "learning_rate": 0.0002998778103616813,
      "loss": 2.2747,
      "step": 510
    },
    {
      "epoch": 0.07999373825923607,
      "grad_norm": 1.4272046089172363,
      "learning_rate": 0.0002998533724340176,
      "loss": 1.7943,
      "step": 511
    },
    {
      "epoch": 0.08015028177833437,
      "grad_norm": 1.0465346574783325,
      "learning_rate": 0.00029982893450635384,
      "loss": 1.6995,
      "step": 512
    },
    {
      "epoch": 0.08030682529743269,
      "grad_norm": 1.2503383159637451,
      "learning_rate": 0.0002998044965786901,
      "loss": 1.6331,
      "step": 513
    },
    {
      "epoch": 0.08046336881653099,
      "grad_norm": 1.0055828094482422,
      "learning_rate": 0.00029978005865102634,
      "loss": 1.593,
      "step": 514
    },
    {
      "epoch": 0.0806199123356293,
      "grad_norm": 0.7897710800170898,
      "learning_rate": 0.00029975562072336265,
      "loss": 1.6474,
      "step": 515
    },
    {
      "epoch": 0.08077645585472762,
      "grad_norm": 2.2324774265289307,
      "learning_rate": 0.0002997311827956989,
      "loss": 1.8418,
      "step": 516
    },
    {
      "epoch": 0.08093299937382592,
      "grad_norm": 1.3233245611190796,
      "learning_rate": 0.00029970674486803515,
      "loss": 1.5869,
      "step": 517
    },
    {
      "epoch": 0.08108954289292424,
      "grad_norm": 1.2391108274459839,
      "learning_rate": 0.00029968230694037146,
      "loss": 1.604,
      "step": 518
    },
    {
      "epoch": 0.08124608641202254,
      "grad_norm": 1.746536135673523,
      "learning_rate": 0.0002996578690127077,
      "loss": 1.7185,
      "step": 519
    },
    {
      "epoch": 0.08140262993112085,
      "grad_norm": 2.1696672439575195,
      "learning_rate": 0.00029963343108504396,
      "loss": 1.7549,
      "step": 520
    },
    {
      "epoch": 0.08155917345021917,
      "grad_norm": 2.327106237411499,
      "learning_rate": 0.00029960899315738026,
      "loss": 1.7399,
      "step": 521
    },
    {
      "epoch": 0.08171571696931747,
      "grad_norm": 2.274477481842041,
      "learning_rate": 0.00029958455522971646,
      "loss": 1.6924,
      "step": 522
    },
    {
      "epoch": 0.08187226048841578,
      "grad_norm": 2.649080991744995,
      "learning_rate": 0.00029956011730205277,
      "loss": 1.7547,
      "step": 523
    },
    {
      "epoch": 0.08202880400751408,
      "grad_norm": 2.5556795597076416,
      "learning_rate": 0.000299535679374389,
      "loss": 1.7752,
      "step": 524
    },
    {
      "epoch": 0.0821853475266124,
      "grad_norm": 3.2737300395965576,
      "learning_rate": 0.00029951124144672527,
      "loss": 2.1346,
      "step": 525
    },
    {
      "epoch": 0.08234189104571071,
      "grad_norm": 1.695064663887024,
      "learning_rate": 0.0002994868035190616,
      "loss": 1.6453,
      "step": 526
    },
    {
      "epoch": 0.08249843456480901,
      "grad_norm": 9.204750061035156,
      "learning_rate": 0.0002994623655913978,
      "loss": 2.3186,
      "step": 527
    },
    {
      "epoch": 0.08265497808390733,
      "grad_norm": 3.981736660003662,
      "learning_rate": 0.0002994379276637341,
      "loss": 2.1661,
      "step": 528
    },
    {
      "epoch": 0.08281152160300563,
      "grad_norm": 4.276581764221191,
      "learning_rate": 0.00029941348973607033,
      "loss": 1.9903,
      "step": 529
    },
    {
      "epoch": 0.08296806512210395,
      "grad_norm": 3.1463425159454346,
      "learning_rate": 0.00029938905180840663,
      "loss": 1.9789,
      "step": 530
    },
    {
      "epoch": 0.08312460864120226,
      "grad_norm": 2.7212626934051514,
      "learning_rate": 0.0002993646138807429,
      "loss": 1.916,
      "step": 531
    },
    {
      "epoch": 0.08328115216030056,
      "grad_norm": 3.936119794845581,
      "learning_rate": 0.00029934017595307914,
      "loss": 2.077,
      "step": 532
    },
    {
      "epoch": 0.08343769567939888,
      "grad_norm": 3.289414882659912,
      "learning_rate": 0.00029931573802541544,
      "loss": 1.8905,
      "step": 533
    },
    {
      "epoch": 0.08359423919849718,
      "grad_norm": 4.605769157409668,
      "learning_rate": 0.0002992913000977517,
      "loss": 1.9716,
      "step": 534
    },
    {
      "epoch": 0.08375078271759549,
      "grad_norm": 1.6472567319869995,
      "learning_rate": 0.00029926686217008794,
      "loss": 1.9615,
      "step": 535
    },
    {
      "epoch": 0.08390732623669381,
      "grad_norm": 3.2355268001556396,
      "learning_rate": 0.00029924242424242425,
      "loss": 1.9301,
      "step": 536
    },
    {
      "epoch": 0.08406386975579211,
      "grad_norm": 1.935306429862976,
      "learning_rate": 0.00029921798631476045,
      "loss": 1.8207,
      "step": 537
    },
    {
      "epoch": 0.08422041327489042,
      "grad_norm": 2.387748956680298,
      "learning_rate": 0.00029919354838709675,
      "loss": 1.9752,
      "step": 538
    },
    {
      "epoch": 0.08437695679398872,
      "grad_norm": 3.236212730407715,
      "learning_rate": 0.000299169110459433,
      "loss": 2.2058,
      "step": 539
    },
    {
      "epoch": 0.08453350031308704,
      "grad_norm": 3.333184003829956,
      "learning_rate": 0.00029914467253176926,
      "loss": 2.0538,
      "step": 540
    },
    {
      "epoch": 0.08469004383218534,
      "grad_norm": 2.5119690895080566,
      "learning_rate": 0.00029912023460410556,
      "loss": 1.9744,
      "step": 541
    },
    {
      "epoch": 0.08484658735128366,
      "grad_norm": 3.832310438156128,
      "learning_rate": 0.0002990957966764418,
      "loss": 1.9949,
      "step": 542
    },
    {
      "epoch": 0.08500313087038197,
      "grad_norm": 3.179901599884033,
      "learning_rate": 0.00029907135874877806,
      "loss": 1.8598,
      "step": 543
    },
    {
      "epoch": 0.08515967438948027,
      "grad_norm": 4.294993877410889,
      "learning_rate": 0.00029904692082111437,
      "loss": 2.1715,
      "step": 544
    },
    {
      "epoch": 0.08531621790857859,
      "grad_norm": 5.152624607086182,
      "learning_rate": 0.0002990224828934506,
      "loss": 2.3794,
      "step": 545
    },
    {
      "epoch": 0.08547276142767689,
      "grad_norm": 3.927077054977417,
      "learning_rate": 0.00029899804496578687,
      "loss": 1.3454,
      "step": 546
    },
    {
      "epoch": 0.0856293049467752,
      "grad_norm": 2.089223623275757,
      "learning_rate": 0.0002989736070381231,
      "loss": 1.8446,
      "step": 547
    },
    {
      "epoch": 0.08578584846587352,
      "grad_norm": 3.640803813934326,
      "learning_rate": 0.00029894916911045943,
      "loss": 1.7771,
      "step": 548
    },
    {
      "epoch": 0.08594239198497182,
      "grad_norm": 2.4625141620635986,
      "learning_rate": 0.0002989247311827957,
      "loss": 1.6782,
      "step": 549
    },
    {
      "epoch": 0.08609893550407013,
      "grad_norm": 4.571941375732422,
      "learning_rate": 0.00029890029325513193,
      "loss": 2.3454,
      "step": 550
    },
    {
      "epoch": 0.08625547902316844,
      "grad_norm": 1.4779797792434692,
      "learning_rate": 0.00029887585532746824,
      "loss": 1.7439,
      "step": 551
    },
    {
      "epoch": 0.08641202254226675,
      "grad_norm": 1.2959303855895996,
      "learning_rate": 0.0002988514173998045,
      "loss": 1.6772,
      "step": 552
    },
    {
      "epoch": 0.08656856606136507,
      "grad_norm": 1.5447274446487427,
      "learning_rate": 0.00029882697947214074,
      "loss": 1.6696,
      "step": 553
    },
    {
      "epoch": 0.08672510958046337,
      "grad_norm": 1.5669810771942139,
      "learning_rate": 0.000298802541544477,
      "loss": 1.4124,
      "step": 554
    },
    {
      "epoch": 0.08688165309956168,
      "grad_norm": 1.109596848487854,
      "learning_rate": 0.00029877810361681324,
      "loss": 1.2168,
      "step": 555
    },
    {
      "epoch": 0.08703819661865998,
      "grad_norm": 1.1062586307525635,
      "learning_rate": 0.00029875366568914955,
      "loss": 1.35,
      "step": 556
    },
    {
      "epoch": 0.0871947401377583,
      "grad_norm": 1.2158498764038086,
      "learning_rate": 0.0002987292277614858,
      "loss": 1.2897,
      "step": 557
    },
    {
      "epoch": 0.08735128365685661,
      "grad_norm": 1.4337570667266846,
      "learning_rate": 0.00029870478983382205,
      "loss": 1.3757,
      "step": 558
    },
    {
      "epoch": 0.08750782717595491,
      "grad_norm": 1.0046571493148804,
      "learning_rate": 0.00029868035190615835,
      "loss": 1.5378,
      "step": 559
    },
    {
      "epoch": 0.08766437069505323,
      "grad_norm": 1.03334641456604,
      "learning_rate": 0.0002986559139784946,
      "loss": 1.2713,
      "step": 560
    },
    {
      "epoch": 0.08782091421415153,
      "grad_norm": 1.219976782798767,
      "learning_rate": 0.00029863147605083086,
      "loss": 1.3718,
      "step": 561
    },
    {
      "epoch": 0.08797745773324984,
      "grad_norm": 1.4024769067764282,
      "learning_rate": 0.0002986070381231671,
      "loss": 1.3139,
      "step": 562
    },
    {
      "epoch": 0.08813400125234815,
      "grad_norm": 1.8966326713562012,
      "learning_rate": 0.0002985826001955034,
      "loss": 1.4525,
      "step": 563
    },
    {
      "epoch": 0.08829054477144646,
      "grad_norm": 1.385165810585022,
      "learning_rate": 0.00029855816226783966,
      "loss": 1.2881,
      "step": 564
    },
    {
      "epoch": 0.08844708829054478,
      "grad_norm": 1.713485598564148,
      "learning_rate": 0.0002985337243401759,
      "loss": 1.1926,
      "step": 565
    },
    {
      "epoch": 0.08860363180964308,
      "grad_norm": 1.2496473789215088,
      "learning_rate": 0.0002985092864125122,
      "loss": 1.3264,
      "step": 566
    },
    {
      "epoch": 0.08876017532874139,
      "grad_norm": 2.36457896232605,
      "learning_rate": 0.00029848484848484847,
      "loss": 1.5377,
      "step": 567
    },
    {
      "epoch": 0.08891671884783969,
      "grad_norm": 1.7575597763061523,
      "learning_rate": 0.0002984604105571847,
      "loss": 1.562,
      "step": 568
    },
    {
      "epoch": 0.08907326236693801,
      "grad_norm": 1.9847209453582764,
      "learning_rate": 0.00029843597262952103,
      "loss": 1.2985,
      "step": 569
    },
    {
      "epoch": 0.08922980588603632,
      "grad_norm": 1.8547005653381348,
      "learning_rate": 0.0002984115347018572,
      "loss": 1.4982,
      "step": 570
    },
    {
      "epoch": 0.08938634940513462,
      "grad_norm": 1.179679036140442,
      "learning_rate": 0.00029838709677419353,
      "loss": 1.3506,
      "step": 571
    },
    {
      "epoch": 0.08954289292423294,
      "grad_norm": 2.9201154708862305,
      "learning_rate": 0.0002983626588465298,
      "loss": 1.6808,
      "step": 572
    },
    {
      "epoch": 0.08969943644333124,
      "grad_norm": 2.8319106101989746,
      "learning_rate": 0.00029833822091886603,
      "loss": 1.6371,
      "step": 573
    },
    {
      "epoch": 0.08985597996242956,
      "grad_norm": 3.5341687202453613,
      "learning_rate": 0.00029831378299120234,
      "loss": 2.353,
      "step": 574
    },
    {
      "epoch": 0.09001252348152787,
      "grad_norm": 1.8579005002975464,
      "learning_rate": 0.0002982893450635386,
      "loss": 1.3278,
      "step": 575
    },
    {
      "epoch": 0.09016906700062617,
      "grad_norm": 3.4177498817443848,
      "learning_rate": 0.00029826490713587484,
      "loss": 1.9979,
      "step": 576
    },
    {
      "epoch": 0.09032561051972449,
      "grad_norm": 3.2669591903686523,
      "learning_rate": 0.0002982404692082111,
      "loss": 1.9626,
      "step": 577
    },
    {
      "epoch": 0.09048215403882279,
      "grad_norm": 1.9549978971481323,
      "learning_rate": 0.0002982160312805474,
      "loss": 1.724,
      "step": 578
    },
    {
      "epoch": 0.0906386975579211,
      "grad_norm": 1.7460333108901978,
      "learning_rate": 0.00029819159335288365,
      "loss": 1.3989,
      "step": 579
    },
    {
      "epoch": 0.09079524107701942,
      "grad_norm": 3.614212989807129,
      "learning_rate": 0.0002981671554252199,
      "loss": 1.8983,
      "step": 580
    },
    {
      "epoch": 0.09095178459611772,
      "grad_norm": 3.4498374462127686,
      "learning_rate": 0.0002981427174975562,
      "loss": 1.606,
      "step": 581
    },
    {
      "epoch": 0.09110832811521603,
      "grad_norm": 4.136172771453857,
      "learning_rate": 0.00029811827956989246,
      "loss": 1.8912,
      "step": 582
    },
    {
      "epoch": 0.09126487163431433,
      "grad_norm": 3.7499725818634033,
      "learning_rate": 0.0002980938416422287,
      "loss": 1.941,
      "step": 583
    },
    {
      "epoch": 0.09142141515341265,
      "grad_norm": 4.676743507385254,
      "learning_rate": 0.000298069403714565,
      "loss": 1.6427,
      "step": 584
    },
    {
      "epoch": 0.09157795867251096,
      "grad_norm": 2.1480143070220947,
      "learning_rate": 0.0002980449657869012,
      "loss": 1.8666,
      "step": 585
    },
    {
      "epoch": 0.09173450219160927,
      "grad_norm": 2.891366481781006,
      "learning_rate": 0.0002980205278592375,
      "loss": 2.0098,
      "step": 586
    },
    {
      "epoch": 0.09189104571070758,
      "grad_norm": 2.557764768600464,
      "learning_rate": 0.00029799608993157377,
      "loss": 1.7633,
      "step": 587
    },
    {
      "epoch": 0.09204758922980588,
      "grad_norm": 2.8896427154541016,
      "learning_rate": 0.00029797165200391,
      "loss": 1.7614,
      "step": 588
    },
    {
      "epoch": 0.0922041327489042,
      "grad_norm": 2.5814003944396973,
      "learning_rate": 0.0002979472140762463,
      "loss": 1.5253,
      "step": 589
    },
    {
      "epoch": 0.0923606762680025,
      "grad_norm": 8.051748275756836,
      "learning_rate": 0.0002979227761485826,
      "loss": 2.0882,
      "step": 590
    },
    {
      "epoch": 0.09251721978710081,
      "grad_norm": 2.6856775283813477,
      "learning_rate": 0.00029789833822091883,
      "loss": 1.7916,
      "step": 591
    },
    {
      "epoch": 0.09267376330619913,
      "grad_norm": 2.2569031715393066,
      "learning_rate": 0.00029787390029325513,
      "loss": 1.8805,
      "step": 592
    },
    {
      "epoch": 0.09283030682529743,
      "grad_norm": 11.610511779785156,
      "learning_rate": 0.0002978494623655914,
      "loss": 1.635,
      "step": 593
    },
    {
      "epoch": 0.09298685034439574,
      "grad_norm": 2.5398268699645996,
      "learning_rate": 0.00029782502443792764,
      "loss": 1.8779,
      "step": 594
    },
    {
      "epoch": 0.09314339386349405,
      "grad_norm": 2.323976516723633,
      "learning_rate": 0.0002978005865102639,
      "loss": 1.9912,
      "step": 595
    },
    {
      "epoch": 0.09329993738259236,
      "grad_norm": 2.0534749031066895,
      "learning_rate": 0.0002977761485826002,
      "loss": 1.0067,
      "step": 596
    },
    {
      "epoch": 0.09345648090169068,
      "grad_norm": 3.8560521602630615,
      "learning_rate": 0.00029775171065493644,
      "loss": 1.5773,
      "step": 597
    },
    {
      "epoch": 0.09361302442078898,
      "grad_norm": 3.9110920429229736,
      "learning_rate": 0.0002977272727272727,
      "loss": 1.7498,
      "step": 598
    },
    {
      "epoch": 0.09376956793988729,
      "grad_norm": 3.5379350185394287,
      "learning_rate": 0.000297702834799609,
      "loss": 1.7439,
      "step": 599
    },
    {
      "epoch": 0.09392611145898559,
      "grad_norm": 3.4338223934173584,
      "learning_rate": 0.0002976783968719452,
      "loss": 1.3979,
      "step": 600
    },
    {
      "epoch": 0.09408265497808391,
      "grad_norm": 1.8097113370895386,
      "learning_rate": 0.0002976539589442815,
      "loss": 1.4324,
      "step": 601
    },
    {
      "epoch": 0.09423919849718222,
      "grad_norm": 1.1481614112854004,
      "learning_rate": 0.00029762952101661775,
      "loss": 1.3141,
      "step": 602
    },
    {
      "epoch": 0.09439574201628052,
      "grad_norm": 0.7774419784545898,
      "learning_rate": 0.000297605083088954,
      "loss": 1.1048,
      "step": 603
    },
    {
      "epoch": 0.09455228553537884,
      "grad_norm": 1.0336264371871948,
      "learning_rate": 0.0002975806451612903,
      "loss": 1.2183,
      "step": 604
    },
    {
      "epoch": 0.09470882905447714,
      "grad_norm": 0.7348364591598511,
      "learning_rate": 0.00029755620723362656,
      "loss": 1.0482,
      "step": 605
    },
    {
      "epoch": 0.09486537257357545,
      "grad_norm": 0.8298443555831909,
      "learning_rate": 0.0002975317693059628,
      "loss": 1.0892,
      "step": 606
    },
    {
      "epoch": 0.09502191609267377,
      "grad_norm": 0.7758945226669312,
      "learning_rate": 0.0002975073313782991,
      "loss": 0.9726,
      "step": 607
    },
    {
      "epoch": 0.09517845961177207,
      "grad_norm": 1.721008062362671,
      "learning_rate": 0.00029748289345063537,
      "loss": 1.0492,
      "step": 608
    },
    {
      "epoch": 0.09533500313087039,
      "grad_norm": 1.345603585243225,
      "learning_rate": 0.0002974584555229716,
      "loss": 1.0549,
      "step": 609
    },
    {
      "epoch": 0.09549154664996869,
      "grad_norm": 1.4059998989105225,
      "learning_rate": 0.00029743401759530787,
      "loss": 1.0506,
      "step": 610
    },
    {
      "epoch": 0.095648090169067,
      "grad_norm": 1.0252139568328857,
      "learning_rate": 0.0002974095796676442,
      "loss": 1.0309,
      "step": 611
    },
    {
      "epoch": 0.09580463368816532,
      "grad_norm": 1.4364956617355347,
      "learning_rate": 0.00029738514173998043,
      "loss": 1.0449,
      "step": 612
    },
    {
      "epoch": 0.09596117720726362,
      "grad_norm": 1.9643033742904663,
      "learning_rate": 0.0002973607038123167,
      "loss": 1.2821,
      "step": 613
    },
    {
      "epoch": 0.09611772072636193,
      "grad_norm": 1.3634884357452393,
      "learning_rate": 0.000297336265884653,
      "loss": 1.2619,
      "step": 614
    },
    {
      "epoch": 0.09627426424546023,
      "grad_norm": 2.063399314880371,
      "learning_rate": 0.00029731182795698924,
      "loss": 1.4142,
      "step": 615
    },
    {
      "epoch": 0.09643080776455855,
      "grad_norm": 1.287653923034668,
      "learning_rate": 0.0002972873900293255,
      "loss": 1.0253,
      "step": 616
    },
    {
      "epoch": 0.09658735128365685,
      "grad_norm": 1.6113243103027344,
      "learning_rate": 0.00029726295210166174,
      "loss": 1.2315,
      "step": 617
    },
    {
      "epoch": 0.09674389480275516,
      "grad_norm": 1.7345733642578125,
      "learning_rate": 0.000297238514173998,
      "loss": 1.1453,
      "step": 618
    },
    {
      "epoch": 0.09690043832185348,
      "grad_norm": 1.8146443367004395,
      "learning_rate": 0.0002972140762463343,
      "loss": 1.0814,
      "step": 619
    },
    {
      "epoch": 0.09705698184095178,
      "grad_norm": 1.6718642711639404,
      "learning_rate": 0.00029718963831867055,
      "loss": 1.4222,
      "step": 620
    },
    {
      "epoch": 0.0972135253600501,
      "grad_norm": 1.9962431192398071,
      "learning_rate": 0.0002971652003910068,
      "loss": 1.5162,
      "step": 621
    },
    {
      "epoch": 0.0973700688791484,
      "grad_norm": 1.7549268007278442,
      "learning_rate": 0.0002971407624633431,
      "loss": 1.5312,
      "step": 622
    },
    {
      "epoch": 0.09752661239824671,
      "grad_norm": 2.5582468509674072,
      "learning_rate": 0.00029711632453567936,
      "loss": 1.5691,
      "step": 623
    },
    {
      "epoch": 0.09768315591734503,
      "grad_norm": 2.2081313133239746,
      "learning_rate": 0.0002970918866080156,
      "loss": 1.1799,
      "step": 624
    },
    {
      "epoch": 0.09783969943644333,
      "grad_norm": 1.4021954536437988,
      "learning_rate": 0.00029706744868035186,
      "loss": 1.1547,
      "step": 625
    },
    {
      "epoch": 0.09799624295554164,
      "grad_norm": 1.895840048789978,
      "learning_rate": 0.00029704301075268816,
      "loss": 1.5232,
      "step": 626
    },
    {
      "epoch": 0.09815278647463994,
      "grad_norm": 2.2723934650421143,
      "learning_rate": 0.0002970185728250244,
      "loss": 1.6906,
      "step": 627
    },
    {
      "epoch": 0.09830932999373826,
      "grad_norm": 2.5103864669799805,
      "learning_rate": 0.00029699413489736067,
      "loss": 1.755,
      "step": 628
    },
    {
      "epoch": 0.09846587351283657,
      "grad_norm": 1.7477445602416992,
      "learning_rate": 0.00029696969696969697,
      "loss": 1.3139,
      "step": 629
    },
    {
      "epoch": 0.09862241703193488,
      "grad_norm": 2.9243004322052,
      "learning_rate": 0.0002969452590420332,
      "loss": 1.5794,
      "step": 630
    },
    {
      "epoch": 0.09877896055103319,
      "grad_norm": 3.4790265560150146,
      "learning_rate": 0.0002969208211143695,
      "loss": 1.5047,
      "step": 631
    },
    {
      "epoch": 0.09893550407013149,
      "grad_norm": 2.3917460441589355,
      "learning_rate": 0.0002968963831867058,
      "loss": 1.9341,
      "step": 632
    },
    {
      "epoch": 0.0990920475892298,
      "grad_norm": 3.003119707107544,
      "learning_rate": 0.000296871945259042,
      "loss": 1.6668,
      "step": 633
    },
    {
      "epoch": 0.09924859110832812,
      "grad_norm": 1.7967157363891602,
      "learning_rate": 0.0002968475073313783,
      "loss": 1.3301,
      "step": 634
    },
    {
      "epoch": 0.09940513462742642,
      "grad_norm": 9.354352951049805,
      "learning_rate": 0.00029682306940371453,
      "loss": 2.0308,
      "step": 635
    },
    {
      "epoch": 0.09956167814652474,
      "grad_norm": 2.55900502204895,
      "learning_rate": 0.0002967986314760508,
      "loss": 2.2328,
      "step": 636
    },
    {
      "epoch": 0.09971822166562304,
      "grad_norm": 3.0861406326293945,
      "learning_rate": 0.0002967741935483871,
      "loss": 1.5009,
      "step": 637
    },
    {
      "epoch": 0.09987476518472135,
      "grad_norm": 2.7047877311706543,
      "learning_rate": 0.00029674975562072334,
      "loss": 1.8303,
      "step": 638
    },
    {
      "epoch": 0.10003130870381967,
      "grad_norm": 1.7027578353881836,
      "learning_rate": 0.0002967253176930596,
      "loss": 1.4659,
      "step": 639
    },
    {
      "epoch": 0.10018785222291797,
      "grad_norm": 3.0926079750061035,
      "learning_rate": 0.0002967008797653959,
      "loss": 1.7596,
      "step": 640
    },
    {
      "epoch": 0.10034439574201628,
      "grad_norm": 3.1565775871276855,
      "learning_rate": 0.00029667644183773215,
      "loss": 2.2201,
      "step": 641
    },
    {
      "epoch": 0.10050093926111459,
      "grad_norm": 2.1882317066192627,
      "learning_rate": 0.0002966520039100684,
      "loss": 1.4988,
      "step": 642
    },
    {
      "epoch": 0.1006574827802129,
      "grad_norm": 3.5301172733306885,
      "learning_rate": 0.00029662756598240465,
      "loss": 1.6186,
      "step": 643
    },
    {
      "epoch": 0.1008140262993112,
      "grad_norm": 3.971935272216797,
      "learning_rate": 0.00029660312805474096,
      "loss": 2.141,
      "step": 644
    },
    {
      "epoch": 0.10097056981840952,
      "grad_norm": 4.350378513336182,
      "learning_rate": 0.0002965786901270772,
      "loss": 1.8005,
      "step": 645
    },
    {
      "epoch": 0.10112711333750783,
      "grad_norm": 4.860312461853027,
      "learning_rate": 0.00029655425219941346,
      "loss": 1.7147,
      "step": 646
    },
    {
      "epoch": 0.10128365685660613,
      "grad_norm": 4.879358768463135,
      "learning_rate": 0.00029652981427174977,
      "loss": 1.8563,
      "step": 647
    },
    {
      "epoch": 0.10144020037570445,
      "grad_norm": 2.9095821380615234,
      "learning_rate": 0.00029650537634408596,
      "loss": 1.569,
      "step": 648
    },
    {
      "epoch": 0.10159674389480275,
      "grad_norm": 2.8750619888305664,
      "learning_rate": 0.00029648093841642227,
      "loss": 1.724,
      "step": 649
    },
    {
      "epoch": 0.10175328741390106,
      "grad_norm": 4.468817234039307,
      "learning_rate": 0.0002964565004887585,
      "loss": 1.792,
      "step": 650
    },
    {
      "epoch": 0.10190983093299938,
      "grad_norm": 2.051830768585205,
      "learning_rate": 0.00029643206256109477,
      "loss": 1.322,
      "step": 651
    },
    {
      "epoch": 0.10206637445209768,
      "grad_norm": 1.3749841451644897,
      "learning_rate": 0.0002964076246334311,
      "loss": 1.0688,
      "step": 652
    },
    {
      "epoch": 0.102222917971196,
      "grad_norm": 1.0187915563583374,
      "learning_rate": 0.0002963831867057673,
      "loss": 1.0538,
      "step": 653
    },
    {
      "epoch": 0.1023794614902943,
      "grad_norm": 0.9093784689903259,
      "learning_rate": 0.0002963587487781036,
      "loss": 0.894,
      "step": 654
    },
    {
      "epoch": 0.10253600500939261,
      "grad_norm": 0.7980338931083679,
      "learning_rate": 0.0002963343108504399,
      "loss": 1.0852,
      "step": 655
    },
    {
      "epoch": 0.10269254852849093,
      "grad_norm": 1.0717920064926147,
      "learning_rate": 0.00029630987292277613,
      "loss": 0.9796,
      "step": 656
    },
    {
      "epoch": 0.10284909204758923,
      "grad_norm": 1.0452539920806885,
      "learning_rate": 0.0002962854349951124,
      "loss": 0.9234,
      "step": 657
    },
    {
      "epoch": 0.10300563556668754,
      "grad_norm": 1.031282901763916,
      "learning_rate": 0.00029626099706744864,
      "loss": 1.0287,
      "step": 658
    },
    {
      "epoch": 0.10316217908578584,
      "grad_norm": 1.1783666610717773,
      "learning_rate": 0.00029623655913978494,
      "loss": 0.8954,
      "step": 659
    },
    {
      "epoch": 0.10331872260488416,
      "grad_norm": 0.8439245223999023,
      "learning_rate": 0.0002962121212121212,
      "loss": 0.7738,
      "step": 660
    },
    {
      "epoch": 0.10347526612398247,
      "grad_norm": 0.914787769317627,
      "learning_rate": 0.00029618768328445745,
      "loss": 0.8843,
      "step": 661
    },
    {
      "epoch": 0.10363180964308077,
      "grad_norm": 1.0592864751815796,
      "learning_rate": 0.00029616324535679375,
      "loss": 1.0231,
      "step": 662
    },
    {
      "epoch": 0.10378835316217909,
      "grad_norm": 1.1694962978363037,
      "learning_rate": 0.00029613880742913,
      "loss": 1.096,
      "step": 663
    },
    {
      "epoch": 0.10394489668127739,
      "grad_norm": 2.1501312255859375,
      "learning_rate": 0.00029611436950146625,
      "loss": 0.9233,
      "step": 664
    },
    {
      "epoch": 0.1041014402003757,
      "grad_norm": 0.7547603249549866,
      "learning_rate": 0.0002960899315738025,
      "loss": 0.9509,
      "step": 665
    },
    {
      "epoch": 0.10425798371947402,
      "grad_norm": 1.070226788520813,
      "learning_rate": 0.00029606549364613876,
      "loss": 0.865,
      "step": 666
    },
    {
      "epoch": 0.10441452723857232,
      "grad_norm": 3.9022319316864014,
      "learning_rate": 0.00029604105571847506,
      "loss": 1.893,
      "step": 667
    },
    {
      "epoch": 0.10457107075767064,
      "grad_norm": 1.9020326137542725,
      "learning_rate": 0.0002960166177908113,
      "loss": 1.1844,
      "step": 668
    },
    {
      "epoch": 0.10472761427676894,
      "grad_norm": 1.318651795387268,
      "learning_rate": 0.00029599217986314756,
      "loss": 1.0922,
      "step": 669
    },
    {
      "epoch": 0.10488415779586725,
      "grad_norm": 1.523789405822754,
      "learning_rate": 0.00029596774193548387,
      "loss": 1.0611,
      "step": 670
    },
    {
      "epoch": 0.10504070131496555,
      "grad_norm": 1.3359650373458862,
      "learning_rate": 0.0002959433040078201,
      "loss": 1.0302,
      "step": 671
    },
    {
      "epoch": 0.10519724483406387,
      "grad_norm": 1.3721380233764648,
      "learning_rate": 0.00029591886608015637,
      "loss": 1.0434,
      "step": 672
    },
    {
      "epoch": 0.10535378835316218,
      "grad_norm": 1.5583863258361816,
      "learning_rate": 0.0002958944281524926,
      "loss": 1.4901,
      "step": 673
    },
    {
      "epoch": 0.10551033187226048,
      "grad_norm": 3.990788221359253,
      "learning_rate": 0.00029586999022482893,
      "loss": 1.4394,
      "step": 674
    },
    {
      "epoch": 0.1056668753913588,
      "grad_norm": 2.45947003364563,
      "learning_rate": 0.0002958455522971652,
      "loss": 1.4246,
      "step": 675
    },
    {
      "epoch": 0.1058234189104571,
      "grad_norm": 2.0882105827331543,
      "learning_rate": 0.00029582111436950143,
      "loss": 1.3557,
      "step": 676
    },
    {
      "epoch": 0.10597996242955542,
      "grad_norm": 1.8049339056015015,
      "learning_rate": 0.00029579667644183774,
      "loss": 0.9934,
      "step": 677
    },
    {
      "epoch": 0.10613650594865373,
      "grad_norm": 2.7573509216308594,
      "learning_rate": 0.000295772238514174,
      "loss": 1.2822,
      "step": 678
    },
    {
      "epoch": 0.10629304946775203,
      "grad_norm": 1.7646522521972656,
      "learning_rate": 0.00029574780058651024,
      "loss": 1.2641,
      "step": 679
    },
    {
      "epoch": 0.10644959298685035,
      "grad_norm": 3.6935081481933594,
      "learning_rate": 0.00029572336265884654,
      "loss": 1.6484,
      "step": 680
    },
    {
      "epoch": 0.10660613650594865,
      "grad_norm": 3.90690541267395,
      "learning_rate": 0.00029569892473118274,
      "loss": 1.738,
      "step": 681
    },
    {
      "epoch": 0.10676268002504696,
      "grad_norm": 3.629829168319702,
      "learning_rate": 0.00029567448680351905,
      "loss": 1.2565,
      "step": 682
    },
    {
      "epoch": 0.10691922354414528,
      "grad_norm": 2.447295904159546,
      "learning_rate": 0.0002956500488758553,
      "loss": 1.3839,
      "step": 683
    },
    {
      "epoch": 0.10707576706324358,
      "grad_norm": 3.1187667846679688,
      "learning_rate": 0.00029562561094819155,
      "loss": 1.2876,
      "step": 684
    },
    {
      "epoch": 0.1072323105823419,
      "grad_norm": 3.4593164920806885,
      "learning_rate": 0.00029560117302052785,
      "loss": 1.8006,
      "step": 685
    },
    {
      "epoch": 0.1073888541014402,
      "grad_norm": 2.576993465423584,
      "learning_rate": 0.0002955767350928641,
      "loss": 1.7238,
      "step": 686
    },
    {
      "epoch": 0.10754539762053851,
      "grad_norm": 4.6910319328308105,
      "learning_rate": 0.00029555229716520036,
      "loss": 1.8431,
      "step": 687
    },
    {
      "epoch": 0.10770194113963683,
      "grad_norm": 2.633889675140381,
      "learning_rate": 0.0002955278592375366,
      "loss": 1.9383,
      "step": 688
    },
    {
      "epoch": 0.10785848465873513,
      "grad_norm": 6.47677755355835,
      "learning_rate": 0.0002955034213098729,
      "loss": 2.1373,
      "step": 689
    },
    {
      "epoch": 0.10801502817783344,
      "grad_norm": 1.8168855905532837,
      "learning_rate": 0.00029547898338220917,
      "loss": 1.8526,
      "step": 690
    },
    {
      "epoch": 0.10817157169693174,
      "grad_norm": 2.5182387828826904,
      "learning_rate": 0.0002954545454545454,
      "loss": 1.632,
      "step": 691
    },
    {
      "epoch": 0.10832811521603006,
      "grad_norm": 2.240593671798706,
      "learning_rate": 0.0002954301075268817,
      "loss": 2.1292,
      "step": 692
    },
    {
      "epoch": 0.10848465873512837,
      "grad_norm": 2.15621018409729,
      "learning_rate": 0.000295405669599218,
      "loss": 1.9179,
      "step": 693
    },
    {
      "epoch": 0.10864120225422667,
      "grad_norm": 1.9122724533081055,
      "learning_rate": 0.0002953812316715542,
      "loss": 1.8236,
      "step": 694
    },
    {
      "epoch": 0.10879774577332499,
      "grad_norm": 2.539159059524536,
      "learning_rate": 0.00029535679374389053,
      "loss": 2.0443,
      "step": 695
    },
    {
      "epoch": 0.10895428929242329,
      "grad_norm": 1.3785215616226196,
      "learning_rate": 0.0002953323558162267,
      "loss": 1.0128,
      "step": 696
    },
    {
      "epoch": 0.1091108328115216,
      "grad_norm": 4.100296974182129,
      "learning_rate": 0.00029530791788856303,
      "loss": 1.38,
      "step": 697
    },
    {
      "epoch": 0.1092673763306199,
      "grad_norm": 3.4671149253845215,
      "learning_rate": 0.0002952834799608993,
      "loss": 1.4838,
      "step": 698
    },
    {
      "epoch": 0.10942391984971822,
      "grad_norm": 2.7299959659576416,
      "learning_rate": 0.00029525904203323553,
      "loss": 1.5185,
      "step": 699
    },
    {
      "epoch": 0.10958046336881654,
      "grad_norm": 3.1657235622406006,
      "learning_rate": 0.00029523460410557184,
      "loss": 1.625,
      "step": 700
    },
    {
      "epoch": 0.10973700688791484,
      "grad_norm": 1.4762587547302246,
      "learning_rate": 0.0002952101661779081,
      "loss": 1.2705,
      "step": 701
    },
    {
      "epoch": 0.10989355040701315,
      "grad_norm": 1.150429368019104,
      "learning_rate": 0.00029518572825024434,
      "loss": 1.068,
      "step": 702
    },
    {
      "epoch": 0.11005009392611145,
      "grad_norm": 1.3497153520584106,
      "learning_rate": 0.00029516129032258065,
      "loss": 1.1242,
      "step": 703
    },
    {
      "epoch": 0.11020663744520977,
      "grad_norm": 0.7703993320465088,
      "learning_rate": 0.0002951368523949169,
      "loss": 0.9952,
      "step": 704
    },
    {
      "epoch": 0.11036318096430808,
      "grad_norm": 1.7000598907470703,
      "learning_rate": 0.00029511241446725315,
      "loss": 0.9605,
      "step": 705
    },
    {
      "epoch": 0.11051972448340638,
      "grad_norm": 0.7951465845108032,
      "learning_rate": 0.0002950879765395894,
      "loss": 1.1336,
      "step": 706
    },
    {
      "epoch": 0.1106762680025047,
      "grad_norm": 0.787284791469574,
      "learning_rate": 0.0002950635386119257,
      "loss": 0.7303,
      "step": 707
    },
    {
      "epoch": 0.110832811521603,
      "grad_norm": 1.1658583879470825,
      "learning_rate": 0.00029503910068426196,
      "loss": 0.8081,
      "step": 708
    },
    {
      "epoch": 0.11098935504070132,
      "grad_norm": 1.4762721061706543,
      "learning_rate": 0.0002950146627565982,
      "loss": 1.0007,
      "step": 709
    },
    {
      "epoch": 0.11114589855979963,
      "grad_norm": 0.9633269309997559,
      "learning_rate": 0.0002949902248289345,
      "loss": 0.799,
      "step": 710
    },
    {
      "epoch": 0.11130244207889793,
      "grad_norm": 1.539566159248352,
      "learning_rate": 0.00029496578690127077,
      "loss": 1.012,
      "step": 711
    },
    {
      "epoch": 0.11145898559799625,
      "grad_norm": 1.1010069847106934,
      "learning_rate": 0.000294941348973607,
      "loss": 0.8838,
      "step": 712
    },
    {
      "epoch": 0.11161552911709455,
      "grad_norm": 1.7431788444519043,
      "learning_rate": 0.00029491691104594327,
      "loss": 1.1716,
      "step": 713
    },
    {
      "epoch": 0.11177207263619286,
      "grad_norm": 1.6751221418380737,
      "learning_rate": 0.0002948924731182795,
      "loss": 1.2777,
      "step": 714
    },
    {
      "epoch": 0.11192861615529118,
      "grad_norm": 1.2621047496795654,
      "learning_rate": 0.0002948680351906158,
      "loss": 0.887,
      "step": 715
    },
    {
      "epoch": 0.11208515967438948,
      "grad_norm": 2.204878091812134,
      "learning_rate": 0.0002948435972629521,
      "loss": 1.3467,
      "step": 716
    },
    {
      "epoch": 0.1122417031934878,
      "grad_norm": 1.4875999689102173,
      "learning_rate": 0.00029481915933528833,
      "loss": 1.2575,
      "step": 717
    },
    {
      "epoch": 0.1123982467125861,
      "grad_norm": 3.0924177169799805,
      "learning_rate": 0.00029479472140762463,
      "loss": 1.4872,
      "step": 718
    },
    {
      "epoch": 0.11255479023168441,
      "grad_norm": 2.1464388370513916,
      "learning_rate": 0.0002947702834799609,
      "loss": 1.063,
      "step": 719
    },
    {
      "epoch": 0.11271133375078271,
      "grad_norm": 2.1253466606140137,
      "learning_rate": 0.00029474584555229714,
      "loss": 1.0376,
      "step": 720
    },
    {
      "epoch": 0.11286787726988103,
      "grad_norm": 1.9266527891159058,
      "learning_rate": 0.0002947214076246334,
      "loss": 1.3224,
      "step": 721
    },
    {
      "epoch": 0.11302442078897934,
      "grad_norm": 1.8208385705947876,
      "learning_rate": 0.0002946969696969697,
      "loss": 1.186,
      "step": 722
    },
    {
      "epoch": 0.11318096430807764,
      "grad_norm": 1.7924067974090576,
      "learning_rate": 0.00029467253176930594,
      "loss": 1.2369,
      "step": 723
    },
    {
      "epoch": 0.11333750782717596,
      "grad_norm": 1.3854628801345825,
      "learning_rate": 0.0002946480938416422,
      "loss": 1.1493,
      "step": 724
    },
    {
      "epoch": 0.11349405134627426,
      "grad_norm": 2.6812336444854736,
      "learning_rate": 0.0002946236559139785,
      "loss": 1.3814,
      "step": 725
    },
    {
      "epoch": 0.11365059486537257,
      "grad_norm": 1.894299864768982,
      "learning_rate": 0.00029459921798631475,
      "loss": 1.117,
      "step": 726
    },
    {
      "epoch": 0.11380713838447089,
      "grad_norm": 4.2252726554870605,
      "learning_rate": 0.000294574780058651,
      "loss": 1.6541,
      "step": 727
    },
    {
      "epoch": 0.11396368190356919,
      "grad_norm": 1.660989761352539,
      "learning_rate": 0.0002945503421309873,
      "loss": 1.3064,
      "step": 728
    },
    {
      "epoch": 0.1141202254226675,
      "grad_norm": 1.3125436305999756,
      "learning_rate": 0.0002945259042033235,
      "loss": 1.2461,
      "step": 729
    },
    {
      "epoch": 0.1142767689417658,
      "grad_norm": 1.556378722190857,
      "learning_rate": 0.0002945014662756598,
      "loss": 1.4757,
      "step": 730
    },
    {
      "epoch": 0.11443331246086412,
      "grad_norm": 1.8206431865692139,
      "learning_rate": 0.00029447702834799606,
      "loss": 1.472,
      "step": 731
    },
    {
      "epoch": 0.11458985597996243,
      "grad_norm": 2.837712287902832,
      "learning_rate": 0.0002944525904203323,
      "loss": 1.6409,
      "step": 732
    },
    {
      "epoch": 0.11474639949906074,
      "grad_norm": 2.3481979370117188,
      "learning_rate": 0.0002944281524926686,
      "loss": 1.618,
      "step": 733
    },
    {
      "epoch": 0.11490294301815905,
      "grad_norm": 2.627960443496704,
      "learning_rate": 0.00029440371456500487,
      "loss": 1.1409,
      "step": 734
    },
    {
      "epoch": 0.11505948653725735,
      "grad_norm": 1.7479578256607056,
      "learning_rate": 0.0002943792766373411,
      "loss": 1.4743,
      "step": 735
    },
    {
      "epoch": 0.11521603005635567,
      "grad_norm": 2.662626028060913,
      "learning_rate": 0.0002943548387096774,
      "loss": 1.8053,
      "step": 736
    },
    {
      "epoch": 0.11537257357545398,
      "grad_norm": 3.780766487121582,
      "learning_rate": 0.0002943304007820137,
      "loss": 1.8668,
      "step": 737
    },
    {
      "epoch": 0.11552911709455228,
      "grad_norm": 3.494009017944336,
      "learning_rate": 0.00029430596285434993,
      "loss": 1.9487,
      "step": 738
    },
    {
      "epoch": 0.1156856606136506,
      "grad_norm": 2.3752057552337646,
      "learning_rate": 0.0002942815249266862,
      "loss": 1.8814,
      "step": 739
    },
    {
      "epoch": 0.1158422041327489,
      "grad_norm": 3.852254629135132,
      "learning_rate": 0.0002942570869990225,
      "loss": 1.9308,
      "step": 740
    },
    {
      "epoch": 0.11599874765184721,
      "grad_norm": 1.938531756401062,
      "learning_rate": 0.00029423264907135874,
      "loss": 1.4905,
      "step": 741
    },
    {
      "epoch": 0.11615529117094553,
      "grad_norm": 3.246478319168091,
      "learning_rate": 0.000294208211143695,
      "loss": 1.3597,
      "step": 742
    },
    {
      "epoch": 0.11631183469004383,
      "grad_norm": 2.4071877002716064,
      "learning_rate": 0.0002941837732160313,
      "loss": 1.1944,
      "step": 743
    },
    {
      "epoch": 0.11646837820914215,
      "grad_norm": 1.2850818634033203,
      "learning_rate": 0.0002941593352883675,
      "loss": 1.3349,
      "step": 744
    },
    {
      "epoch": 0.11662492172824045,
      "grad_norm": 2.346933126449585,
      "learning_rate": 0.0002941348973607038,
      "loss": 0.904,
      "step": 745
    },
    {
      "epoch": 0.11678146524733876,
      "grad_norm": 3.941329002380371,
      "learning_rate": 0.00029411045943304005,
      "loss": 2.2503,
      "step": 746
    },
    {
      "epoch": 0.11693800876643706,
      "grad_norm": 2.6362099647521973,
      "learning_rate": 0.0002940860215053763,
      "loss": 1.5694,
      "step": 747
    },
    {
      "epoch": 0.11709455228553538,
      "grad_norm": 4.092523097991943,
      "learning_rate": 0.0002940615835777126,
      "loss": 1.9898,
      "step": 748
    },
    {
      "epoch": 0.11725109580463369,
      "grad_norm": 3.4004411697387695,
      "learning_rate": 0.00029403714565004886,
      "loss": 1.589,
      "step": 749
    },
    {
      "epoch": 0.117407639323732,
      "grad_norm": 3.5993781089782715,
      "learning_rate": 0.0002940127077223851,
      "loss": 1.6423,
      "step": 750
    },
    {
      "epoch": 0.11756418284283031,
      "grad_norm": 0.811040461063385,
      "learning_rate": 0.0002939882697947214,
      "loss": 1.0251,
      "step": 751
    },
    {
      "epoch": 0.11772072636192861,
      "grad_norm": 0.8086991310119629,
      "learning_rate": 0.00029396383186705766,
      "loss": 0.9569,
      "step": 752
    },
    {
      "epoch": 0.11787726988102692,
      "grad_norm": 0.6266233325004578,
      "learning_rate": 0.0002939393939393939,
      "loss": 0.8385,
      "step": 753
    },
    {
      "epoch": 0.11803381340012524,
      "grad_norm": 2.2221486568450928,
      "learning_rate": 0.00029391495601173017,
      "loss": 1.0404,
      "step": 754
    },
    {
      "epoch": 0.11819035691922354,
      "grad_norm": 0.7100171446800232,
      "learning_rate": 0.00029389051808406647,
      "loss": 0.8026,
      "step": 755
    },
    {
      "epoch": 0.11834690043832186,
      "grad_norm": 1.3321337699890137,
      "learning_rate": 0.0002938660801564027,
      "loss": 0.8826,
      "step": 756
    },
    {
      "epoch": 0.11850344395742016,
      "grad_norm": 0.7835633754730225,
      "learning_rate": 0.000293841642228739,
      "loss": 0.7613,
      "step": 757
    },
    {
      "epoch": 0.11865998747651847,
      "grad_norm": 0.6961237788200378,
      "learning_rate": 0.0002938172043010753,
      "loss": 0.6241,
      "step": 758
    },
    {
      "epoch": 0.11881653099561679,
      "grad_norm": 1.128588080406189,
      "learning_rate": 0.0002937927663734115,
      "loss": 0.8578,
      "step": 759
    },
    {
      "epoch": 0.11897307451471509,
      "grad_norm": 0.9707612991333008,
      "learning_rate": 0.0002937683284457478,
      "loss": 0.9736,
      "step": 760
    },
    {
      "epoch": 0.1191296180338134,
      "grad_norm": 0.9797849655151367,
      "learning_rate": 0.00029374389051808403,
      "loss": 0.8492,
      "step": 761
    },
    {
      "epoch": 0.1192861615529117,
      "grad_norm": 1.5107133388519287,
      "learning_rate": 0.0002937194525904203,
      "loss": 0.7293,
      "step": 762
    },
    {
      "epoch": 0.11944270507201002,
      "grad_norm": 1.1908730268478394,
      "learning_rate": 0.0002936950146627566,
      "loss": 0.7209,
      "step": 763
    },
    {
      "epoch": 0.11959924859110833,
      "grad_norm": 1.0134694576263428,
      "learning_rate": 0.00029367057673509284,
      "loss": 0.8672,
      "step": 764
    },
    {
      "epoch": 0.11975579211020664,
      "grad_norm": 1.0121101140975952,
      "learning_rate": 0.0002936461388074291,
      "loss": 0.8697,
      "step": 765
    },
    {
      "epoch": 0.11991233562930495,
      "grad_norm": 0.948444128036499,
      "learning_rate": 0.0002936217008797654,
      "loss": 1.0135,
      "step": 766
    },
    {
      "epoch": 0.12006887914840325,
      "grad_norm": 1.0948662757873535,
      "learning_rate": 0.0002935972629521016,
      "loss": 0.7778,
      "step": 767
    },
    {
      "epoch": 0.12022542266750157,
      "grad_norm": 2.0492300987243652,
      "learning_rate": 0.0002935728250244379,
      "loss": 1.1036,
      "step": 768
    },
    {
      "epoch": 0.12038196618659988,
      "grad_norm": 1.7745238542556763,
      "learning_rate": 0.00029354838709677415,
      "loss": 0.9884,
      "step": 769
    },
    {
      "epoch": 0.12053850970569818,
      "grad_norm": 1.6761794090270996,
      "learning_rate": 0.0002935239491691104,
      "loss": 1.0637,
      "step": 770
    },
    {
      "epoch": 0.1206950532247965,
      "grad_norm": 1.1787947416305542,
      "learning_rate": 0.0002934995112414467,
      "loss": 0.8567,
      "step": 771
    },
    {
      "epoch": 0.1208515967438948,
      "grad_norm": 2.7158119678497314,
      "learning_rate": 0.00029347507331378296,
      "loss": 1.1482,
      "step": 772
    },
    {
      "epoch": 0.12100814026299311,
      "grad_norm": 1.5714401006698608,
      "learning_rate": 0.0002934506353861192,
      "loss": 1.0865,
      "step": 773
    },
    {
      "epoch": 0.12116468378209141,
      "grad_norm": 1.6347661018371582,
      "learning_rate": 0.0002934261974584555,
      "loss": 1.1607,
      "step": 774
    },
    {
      "epoch": 0.12132122730118973,
      "grad_norm": 2.3367502689361572,
      "learning_rate": 0.00029340175953079177,
      "loss": 1.1489,
      "step": 775
    },
    {
      "epoch": 0.12147777082028804,
      "grad_norm": 2.6965532302856445,
      "learning_rate": 0.000293377321603128,
      "loss": 1.0683,
      "step": 776
    },
    {
      "epoch": 0.12163431433938635,
      "grad_norm": 1.9645942449569702,
      "learning_rate": 0.00029335288367546427,
      "loss": 1.4028,
      "step": 777
    },
    {
      "epoch": 0.12179085785848466,
      "grad_norm": 2.6310694217681885,
      "learning_rate": 0.0002933284457478006,
      "loss": 1.2004,
      "step": 778
    },
    {
      "epoch": 0.12194740137758296,
      "grad_norm": 1.9831539392471313,
      "learning_rate": 0.00029330400782013683,
      "loss": 1.4612,
      "step": 779
    },
    {
      "epoch": 0.12210394489668128,
      "grad_norm": 2.4186670780181885,
      "learning_rate": 0.0002932795698924731,
      "loss": 1.4779,
      "step": 780
    },
    {
      "epoch": 0.12226048841577959,
      "grad_norm": 4.081188201904297,
      "learning_rate": 0.0002932551319648094,
      "loss": 1.713,
      "step": 781
    },
    {
      "epoch": 0.12241703193487789,
      "grad_norm": 2.4842689037323,
      "learning_rate": 0.00029323069403714564,
      "loss": 2.0171,
      "step": 782
    },
    {
      "epoch": 0.12257357545397621,
      "grad_norm": 2.6195974349975586,
      "learning_rate": 0.0002932062561094819,
      "loss": 1.212,
      "step": 783
    },
    {
      "epoch": 0.12273011897307451,
      "grad_norm": 2.697697162628174,
      "learning_rate": 0.00029318181818181814,
      "loss": 1.7433,
      "step": 784
    },
    {
      "epoch": 0.12288666249217282,
      "grad_norm": 2.484013319015503,
      "learning_rate": 0.0002931573802541544,
      "loss": 1.8923,
      "step": 785
    },
    {
      "epoch": 0.12304320601127114,
      "grad_norm": 4.01732873916626,
      "learning_rate": 0.0002931329423264907,
      "loss": 2.1281,
      "step": 786
    },
    {
      "epoch": 0.12319974953036944,
      "grad_norm": 2.3246097564697266,
      "learning_rate": 0.00029310850439882695,
      "loss": 2.1824,
      "step": 787
    },
    {
      "epoch": 0.12335629304946776,
      "grad_norm": 3.7635695934295654,
      "learning_rate": 0.0002930840664711632,
      "loss": 1.5211,
      "step": 788
    },
    {
      "epoch": 0.12351283656856606,
      "grad_norm": 2.600083827972412,
      "learning_rate": 0.0002930596285434995,
      "loss": 1.4717,
      "step": 789
    },
    {
      "epoch": 0.12366938008766437,
      "grad_norm": 4.064560890197754,
      "learning_rate": 0.00029303519061583575,
      "loss": 1.7544,
      "step": 790
    },
    {
      "epoch": 0.12382592360676269,
      "grad_norm": 4.859133720397949,
      "learning_rate": 0.000293010752688172,
      "loss": 1.9283,
      "step": 791
    },
    {
      "epoch": 0.12398246712586099,
      "grad_norm": 2.1018145084381104,
      "learning_rate": 0.00029298631476050826,
      "loss": 1.0601,
      "step": 792
    },
    {
      "epoch": 0.1241390106449593,
      "grad_norm": 2.4449431896209717,
      "learning_rate": 0.00029296187683284456,
      "loss": 1.8295,
      "step": 793
    },
    {
      "epoch": 0.1242955541640576,
      "grad_norm": 2.4376046657562256,
      "learning_rate": 0.0002929374389051808,
      "loss": 1.2816,
      "step": 794
    },
    {
      "epoch": 0.12445209768315592,
      "grad_norm": 3.415478229522705,
      "learning_rate": 0.00029291300097751706,
      "loss": 1.9171,
      "step": 795
    },
    {
      "epoch": 0.12460864120225423,
      "grad_norm": 3.3648712635040283,
      "learning_rate": 0.00029288856304985337,
      "loss": 1.2723,
      "step": 796
    },
    {
      "epoch": 0.12476518472135253,
      "grad_norm": 3.2523117065429688,
      "learning_rate": 0.0002928641251221896,
      "loss": 1.2449,
      "step": 797
    },
    {
      "epoch": 0.12492172824045085,
      "grad_norm": 2.2647125720977783,
      "learning_rate": 0.00029283968719452587,
      "loss": 1.2996,
      "step": 798
    },
    {
      "epoch": 0.12507827175954916,
      "grad_norm": 2.5591723918914795,
      "learning_rate": 0.0002928152492668622,
      "loss": 1.4211,
      "step": 799
    },
    {
      "epoch": 0.12523481527864747,
      "grad_norm": 1.7322767972946167,
      "learning_rate": 0.0002927908113391984,
      "loss": 1.3324,
      "step": 800
    },
    {
      "epoch": 0.12539135879774577,
      "grad_norm": 1.5625592470169067,
      "learning_rate": 0.0002927663734115347,
      "loss": 1.0354,
      "step": 801
    },
    {
      "epoch": 0.1255479023168441,
      "grad_norm": 1.2151803970336914,
      "learning_rate": 0.00029274193548387093,
      "loss": 0.7138,
      "step": 802
    },
    {
      "epoch": 0.1257044458359424,
      "grad_norm": 0.8814300894737244,
      "learning_rate": 0.0002927174975562072,
      "loss": 0.6718,
      "step": 803
    },
    {
      "epoch": 0.1258609893550407,
      "grad_norm": 0.789391279220581,
      "learning_rate": 0.0002926930596285435,
      "loss": 0.645,
      "step": 804
    },
    {
      "epoch": 0.126017532874139,
      "grad_norm": 0.6633252501487732,
      "learning_rate": 0.00029266862170087974,
      "loss": 0.6388,
      "step": 805
    },
    {
      "epoch": 0.12617407639323733,
      "grad_norm": 0.9132680892944336,
      "learning_rate": 0.000292644183773216,
      "loss": 0.8446,
      "step": 806
    },
    {
      "epoch": 0.12633061991233563,
      "grad_norm": 0.9453927278518677,
      "learning_rate": 0.00029261974584555224,
      "loss": 0.6193,
      "step": 807
    },
    {
      "epoch": 0.12648716343143393,
      "grad_norm": 1.8710074424743652,
      "learning_rate": 0.00029259530791788855,
      "loss": 0.8318,
      "step": 808
    },
    {
      "epoch": 0.12664370695053226,
      "grad_norm": 1.0109128952026367,
      "learning_rate": 0.0002925708699902248,
      "loss": 0.6593,
      "step": 809
    },
    {
      "epoch": 0.12680025046963056,
      "grad_norm": 1.1044448614120483,
      "learning_rate": 0.00029254643206256105,
      "loss": 0.7262,
      "step": 810
    },
    {
      "epoch": 0.12695679398872886,
      "grad_norm": 1.3992559909820557,
      "learning_rate": 0.00029252199413489736,
      "loss": 0.9756,
      "step": 811
    },
    {
      "epoch": 0.12711333750782716,
      "grad_norm": 1.5335423946380615,
      "learning_rate": 0.0002924975562072336,
      "loss": 0.7762,
      "step": 812
    },
    {
      "epoch": 0.1272698810269255,
      "grad_norm": 1.3663071393966675,
      "learning_rate": 0.00029247311827956986,
      "loss": 0.8961,
      "step": 813
    },
    {
      "epoch": 0.1274264245460238,
      "grad_norm": 1.9370208978652954,
      "learning_rate": 0.00029244868035190616,
      "loss": 0.8597,
      "step": 814
    },
    {
      "epoch": 0.1275829680651221,
      "grad_norm": 1.6557672023773193,
      "learning_rate": 0.00029242424242424236,
      "loss": 1.0157,
      "step": 815
    },
    {
      "epoch": 0.12773951158422042,
      "grad_norm": 1.4072177410125732,
      "learning_rate": 0.00029239980449657867,
      "loss": 0.9113,
      "step": 816
    },
    {
      "epoch": 0.12789605510331872,
      "grad_norm": 1.654747724533081,
      "learning_rate": 0.0002923753665689149,
      "loss": 0.8022,
      "step": 817
    },
    {
      "epoch": 0.12805259862241702,
      "grad_norm": 1.677295207977295,
      "learning_rate": 0.00029235092864125117,
      "loss": 0.7312,
      "step": 818
    },
    {
      "epoch": 0.12820914214151535,
      "grad_norm": 1.2632160186767578,
      "learning_rate": 0.0002923264907135875,
      "loss": 0.9197,
      "step": 819
    },
    {
      "epoch": 0.12836568566061365,
      "grad_norm": 1.6950252056121826,
      "learning_rate": 0.0002923020527859237,
      "loss": 1.268,
      "step": 820
    },
    {
      "epoch": 0.12852222917971196,
      "grad_norm": 1.9522795677185059,
      "learning_rate": 0.00029227761485826,
      "loss": 0.9591,
      "step": 821
    },
    {
      "epoch": 0.12867877269881026,
      "grad_norm": 2.316823959350586,
      "learning_rate": 0.0002922531769305963,
      "loss": 1.0732,
      "step": 822
    },
    {
      "epoch": 0.12883531621790859,
      "grad_norm": 2.60040283203125,
      "learning_rate": 0.00029222873900293253,
      "loss": 1.3269,
      "step": 823
    },
    {
      "epoch": 0.1289918597370069,
      "grad_norm": 2.1569325923919678,
      "learning_rate": 0.0002922043010752688,
      "loss": 1.3893,
      "step": 824
    },
    {
      "epoch": 0.1291484032561052,
      "grad_norm": 3.817329168319702,
      "learning_rate": 0.00029217986314760504,
      "loss": 1.3369,
      "step": 825
    },
    {
      "epoch": 0.12930494677520352,
      "grad_norm": 1.4851608276367188,
      "learning_rate": 0.00029215542521994134,
      "loss": 0.9969,
      "step": 826
    },
    {
      "epoch": 0.12946149029430182,
      "grad_norm": 2.9426980018615723,
      "learning_rate": 0.0002921309872922776,
      "loss": 1.6128,
      "step": 827
    },
    {
      "epoch": 0.12961803381340012,
      "grad_norm": 1.6474707126617432,
      "learning_rate": 0.00029210654936461384,
      "loss": 0.8915,
      "step": 828
    },
    {
      "epoch": 0.12977457733249845,
      "grad_norm": 2.1298272609710693,
      "learning_rate": 0.00029208211143695015,
      "loss": 1.0587,
      "step": 829
    },
    {
      "epoch": 0.12993112085159675,
      "grad_norm": 2.768442153930664,
      "learning_rate": 0.00029205767350928635,
      "loss": 1.544,
      "step": 830
    },
    {
      "epoch": 0.13008766437069505,
      "grad_norm": 2.516650676727295,
      "learning_rate": 0.00029203323558162265,
      "loss": 1.0509,
      "step": 831
    },
    {
      "epoch": 0.13024420788979335,
      "grad_norm": 2.4558215141296387,
      "learning_rate": 0.0002920087976539589,
      "loss": 1.4115,
      "step": 832
    },
    {
      "epoch": 0.13040075140889168,
      "grad_norm": 2.536071300506592,
      "learning_rate": 0.00029198435972629515,
      "loss": 1.4077,
      "step": 833
    },
    {
      "epoch": 0.13055729492798998,
      "grad_norm": 3.4631125926971436,
      "learning_rate": 0.00029195992179863146,
      "loss": 1.7084,
      "step": 834
    },
    {
      "epoch": 0.13071383844708828,
      "grad_norm": 2.3752031326293945,
      "learning_rate": 0.0002919354838709677,
      "loss": 1.1893,
      "step": 835
    },
    {
      "epoch": 0.1308703819661866,
      "grad_norm": 1.6237187385559082,
      "learning_rate": 0.00029191104594330396,
      "loss": 1.0723,
      "step": 836
    },
    {
      "epoch": 0.1310269254852849,
      "grad_norm": 2.3249778747558594,
      "learning_rate": 0.00029188660801564027,
      "loss": 1.6326,
      "step": 837
    },
    {
      "epoch": 0.1311834690043832,
      "grad_norm": 2.8457751274108887,
      "learning_rate": 0.0002918621700879765,
      "loss": 1.4018,
      "step": 838
    },
    {
      "epoch": 0.13134001252348151,
      "grad_norm": 3.0899088382720947,
      "learning_rate": 0.00029183773216031277,
      "loss": 2.1144,
      "step": 839
    },
    {
      "epoch": 0.13149655604257984,
      "grad_norm": 4.969150543212891,
      "learning_rate": 0.000291813294232649,
      "loss": 1.9739,
      "step": 840
    },
    {
      "epoch": 0.13165309956167814,
      "grad_norm": 2.7233245372772217,
      "learning_rate": 0.0002917888563049853,
      "loss": 1.7328,
      "step": 841
    },
    {
      "epoch": 0.13180964308077645,
      "grad_norm": 2.1063811779022217,
      "learning_rate": 0.0002917644183773216,
      "loss": 1.5687,
      "step": 842
    },
    {
      "epoch": 0.13196618659987477,
      "grad_norm": 1.7449232339859009,
      "learning_rate": 0.00029173998044965783,
      "loss": 1.8963,
      "step": 843
    },
    {
      "epoch": 0.13212273011897308,
      "grad_norm": 2.4782533645629883,
      "learning_rate": 0.00029171554252199413,
      "loss": 1.5947,
      "step": 844
    },
    {
      "epoch": 0.13227927363807138,
      "grad_norm": 2.6746630668640137,
      "learning_rate": 0.0002916911045943304,
      "loss": 2.6498,
      "step": 845
    },
    {
      "epoch": 0.1324358171571697,
      "grad_norm": 2.479112148284912,
      "learning_rate": 0.00029166666666666664,
      "loss": 1.5459,
      "step": 846
    },
    {
      "epoch": 0.132592360676268,
      "grad_norm": 1.7579686641693115,
      "learning_rate": 0.0002916422287390029,
      "loss": 0.6981,
      "step": 847
    },
    {
      "epoch": 0.1327489041953663,
      "grad_norm": 1.6483287811279297,
      "learning_rate": 0.00029161779081133914,
      "loss": 1.307,
      "step": 848
    },
    {
      "epoch": 0.1329054477144646,
      "grad_norm": 2.1981475353240967,
      "learning_rate": 0.00029159335288367544,
      "loss": 1.0251,
      "step": 849
    },
    {
      "epoch": 0.13306199123356294,
      "grad_norm": 2.7301342487335205,
      "learning_rate": 0.0002915689149560117,
      "loss": 1.0604,
      "step": 850
    },
    {
      "epoch": 0.13321853475266124,
      "grad_norm": 1.2275153398513794,
      "learning_rate": 0.00029154447702834795,
      "loss": 1.1503,
      "step": 851
    },
    {
      "epoch": 0.13337507827175954,
      "grad_norm": 0.9788094758987427,
      "learning_rate": 0.00029152003910068425,
      "loss": 0.8434,
      "step": 852
    },
    {
      "epoch": 0.13353162179085787,
      "grad_norm": 0.7427651286125183,
      "learning_rate": 0.0002914956011730205,
      "loss": 0.6953,
      "step": 853
    },
    {
      "epoch": 0.13368816530995617,
      "grad_norm": 0.8753380179405212,
      "learning_rate": 0.00029147116324535676,
      "loss": 0.6854,
      "step": 854
    },
    {
      "epoch": 0.13384470882905447,
      "grad_norm": 1.1522316932678223,
      "learning_rate": 0.000291446725317693,
      "loss": 0.5981,
      "step": 855
    },
    {
      "epoch": 0.1340012523481528,
      "grad_norm": 0.9101697206497192,
      "learning_rate": 0.0002914222873900293,
      "loss": 0.534,
      "step": 856
    },
    {
      "epoch": 0.1341577958672511,
      "grad_norm": 1.2821955680847168,
      "learning_rate": 0.00029139784946236556,
      "loss": 0.6715,
      "step": 857
    },
    {
      "epoch": 0.1343143393863494,
      "grad_norm": 1.4305757284164429,
      "learning_rate": 0.0002913734115347018,
      "loss": 0.8797,
      "step": 858
    },
    {
      "epoch": 0.1344708829054477,
      "grad_norm": 1.9941664934158325,
      "learning_rate": 0.0002913489736070381,
      "loss": 0.8174,
      "step": 859
    },
    {
      "epoch": 0.13462742642454603,
      "grad_norm": 1.482197880744934,
      "learning_rate": 0.00029132453567937437,
      "loss": 0.6398,
      "step": 860
    },
    {
      "epoch": 0.13478396994364433,
      "grad_norm": 1.1646844148635864,
      "learning_rate": 0.0002913000977517106,
      "loss": 0.6547,
      "step": 861
    },
    {
      "epoch": 0.13494051346274263,
      "grad_norm": 1.6788074970245361,
      "learning_rate": 0.00029127565982404693,
      "loss": 0.7803,
      "step": 862
    },
    {
      "epoch": 0.13509705698184096,
      "grad_norm": 2.2960433959960938,
      "learning_rate": 0.0002912512218963831,
      "loss": 1.1557,
      "step": 863
    },
    {
      "epoch": 0.13525360050093926,
      "grad_norm": 2.2098381519317627,
      "learning_rate": 0.00029122678396871943,
      "loss": 0.9451,
      "step": 864
    },
    {
      "epoch": 0.13541014402003757,
      "grad_norm": 1.51131272315979,
      "learning_rate": 0.0002912023460410557,
      "loss": 0.9505,
      "step": 865
    },
    {
      "epoch": 0.13556668753913587,
      "grad_norm": 0.9991070032119751,
      "learning_rate": 0.00029117790811339193,
      "loss": 0.7682,
      "step": 866
    },
    {
      "epoch": 0.1357232310582342,
      "grad_norm": 2.670018196105957,
      "learning_rate": 0.00029115347018572824,
      "loss": 1.3066,
      "step": 867
    },
    {
      "epoch": 0.1358797745773325,
      "grad_norm": 1.4845349788665771,
      "learning_rate": 0.0002911290322580645,
      "loss": 0.9315,
      "step": 868
    },
    {
      "epoch": 0.1360363180964308,
      "grad_norm": 1.2157962322235107,
      "learning_rate": 0.00029110459433040074,
      "loss": 0.6735,
      "step": 869
    },
    {
      "epoch": 0.13619286161552913,
      "grad_norm": 1.6585222482681274,
      "learning_rate": 0.00029108015640273705,
      "loss": 1.0187,
      "step": 870
    },
    {
      "epoch": 0.13634940513462743,
      "grad_norm": 1.76893949508667,
      "learning_rate": 0.0002910557184750733,
      "loss": 1.0663,
      "step": 871
    },
    {
      "epoch": 0.13650594865372573,
      "grad_norm": 0.9581948518753052,
      "learning_rate": 0.00029103128054740955,
      "loss": 0.7532,
      "step": 872
    },
    {
      "epoch": 0.13666249217282406,
      "grad_norm": 2.227740526199341,
      "learning_rate": 0.0002910068426197458,
      "loss": 0.8671,
      "step": 873
    },
    {
      "epoch": 0.13681903569192236,
      "grad_norm": 3.511131525039673,
      "learning_rate": 0.0002909824046920821,
      "loss": 1.1649,
      "step": 874
    },
    {
      "epoch": 0.13697557921102066,
      "grad_norm": 2.581437826156616,
      "learning_rate": 0.00029095796676441836,
      "loss": 1.1585,
      "step": 875
    },
    {
      "epoch": 0.13713212273011896,
      "grad_norm": 1.5962125062942505,
      "learning_rate": 0.0002909335288367546,
      "loss": 1.295,
      "step": 876
    },
    {
      "epoch": 0.1372886662492173,
      "grad_norm": 2.135364055633545,
      "learning_rate": 0.0002909090909090909,
      "loss": 1.5455,
      "step": 877
    },
    {
      "epoch": 0.1374452097683156,
      "grad_norm": 2.586113452911377,
      "learning_rate": 0.0002908846529814271,
      "loss": 1.6572,
      "step": 878
    },
    {
      "epoch": 0.1376017532874139,
      "grad_norm": 3.2886767387390137,
      "learning_rate": 0.0002908602150537634,
      "loss": 1.721,
      "step": 879
    },
    {
      "epoch": 0.13775829680651222,
      "grad_norm": 2.5917325019836426,
      "learning_rate": 0.00029083577712609967,
      "loss": 0.9838,
      "step": 880
    },
    {
      "epoch": 0.13791484032561052,
      "grad_norm": 2.5062735080718994,
      "learning_rate": 0.0002908113391984359,
      "loss": 1.3576,
      "step": 881
    },
    {
      "epoch": 0.13807138384470882,
      "grad_norm": 3.9728665351867676,
      "learning_rate": 0.0002907869012707722,
      "loss": 1.2969,
      "step": 882
    },
    {
      "epoch": 0.13822792736380715,
      "grad_norm": 2.223118305206299,
      "learning_rate": 0.0002907624633431085,
      "loss": 1.8066,
      "step": 883
    },
    {
      "epoch": 0.13838447088290545,
      "grad_norm": 3.254514694213867,
      "learning_rate": 0.0002907380254154447,
      "loss": 1.1451,
      "step": 884
    },
    {
      "epoch": 0.13854101440200375,
      "grad_norm": 7.31671667098999,
      "learning_rate": 0.00029071358748778103,
      "loss": 1.7801,
      "step": 885
    },
    {
      "epoch": 0.13869755792110205,
      "grad_norm": 3.2049777507781982,
      "learning_rate": 0.0002906891495601173,
      "loss": 2.0279,
      "step": 886
    },
    {
      "epoch": 0.13885410144020038,
      "grad_norm": 3.6071975231170654,
      "learning_rate": 0.00029066471163245353,
      "loss": 1.5702,
      "step": 887
    },
    {
      "epoch": 0.13901064495929868,
      "grad_norm": 2.798166275024414,
      "learning_rate": 0.0002906402737047898,
      "loss": 1.576,
      "step": 888
    },
    {
      "epoch": 0.13916718847839699,
      "grad_norm": 6.099799633026123,
      "learning_rate": 0.0002906158357771261,
      "loss": 1.8335,
      "step": 889
    },
    {
      "epoch": 0.13932373199749531,
      "grad_norm": 2.965550184249878,
      "learning_rate": 0.00029059139784946234,
      "loss": 1.548,
      "step": 890
    },
    {
      "epoch": 0.13948027551659362,
      "grad_norm": 2.5119330883026123,
      "learning_rate": 0.0002905669599217986,
      "loss": 1.9851,
      "step": 891
    },
    {
      "epoch": 0.13963681903569192,
      "grad_norm": 5.480480194091797,
      "learning_rate": 0.0002905425219941349,
      "loss": 1.4258,
      "step": 892
    },
    {
      "epoch": 0.13979336255479022,
      "grad_norm": 3.603321075439453,
      "learning_rate": 0.00029051808406647115,
      "loss": 1.541,
      "step": 893
    },
    {
      "epoch": 0.13994990607388855,
      "grad_norm": 4.580733776092529,
      "learning_rate": 0.0002904936461388074,
      "loss": 2.0883,
      "step": 894
    },
    {
      "epoch": 0.14010644959298685,
      "grad_norm": 2.0983352661132812,
      "learning_rate": 0.00029046920821114365,
      "loss": 1.5889,
      "step": 895
    },
    {
      "epoch": 0.14026299311208515,
      "grad_norm": 2.2917943000793457,
      "learning_rate": 0.0002904447702834799,
      "loss": 1.6588,
      "step": 896
    },
    {
      "epoch": 0.14041953663118348,
      "grad_norm": 3.498607873916626,
      "learning_rate": 0.0002904203323558162,
      "loss": 1.3828,
      "step": 897
    },
    {
      "epoch": 0.14057608015028178,
      "grad_norm": 2.6161184310913086,
      "learning_rate": 0.00029039589442815246,
      "loss": 1.1676,
      "step": 898
    },
    {
      "epoch": 0.14073262366938008,
      "grad_norm": 3.8199074268341064,
      "learning_rate": 0.0002903714565004887,
      "loss": 1.2942,
      "step": 899
    },
    {
      "epoch": 0.1408891671884784,
      "grad_norm": 1.9904534816741943,
      "learning_rate": 0.000290347018572825,
      "loss": 1.5531,
      "step": 900
    },
    {
      "epoch": 0.1410457107075767,
      "grad_norm": 0.8975843191146851,
      "learning_rate": 0.00029032258064516127,
      "loss": 0.8606,
      "step": 901
    },
    {
      "epoch": 0.141202254226675,
      "grad_norm": 0.7283688187599182,
      "learning_rate": 0.0002902981427174975,
      "loss": 0.7228,
      "step": 902
    },
    {
      "epoch": 0.1413587977457733,
      "grad_norm": 0.6623715162277222,
      "learning_rate": 0.00029027370478983377,
      "loss": 0.7472,
      "step": 903
    },
    {
      "epoch": 0.14151534126487164,
      "grad_norm": 0.7772536873817444,
      "learning_rate": 0.0002902492668621701,
      "loss": 0.5521,
      "step": 904
    },
    {
      "epoch": 0.14167188478396994,
      "grad_norm": 0.9219233989715576,
      "learning_rate": 0.00029022482893450633,
      "loss": 0.7454,
      "step": 905
    },
    {
      "epoch": 0.14182842830306824,
      "grad_norm": 0.8416048884391785,
      "learning_rate": 0.0002902003910068426,
      "loss": 0.7557,
      "step": 906
    },
    {
      "epoch": 0.14198497182216657,
      "grad_norm": 0.8040184378623962,
      "learning_rate": 0.0002901759530791789,
      "loss": 0.4868,
      "step": 907
    },
    {
      "epoch": 0.14214151534126487,
      "grad_norm": 0.9409207105636597,
      "learning_rate": 0.00029015151515151514,
      "loss": 0.6294,
      "step": 908
    },
    {
      "epoch": 0.14229805886036317,
      "grad_norm": 1.9797828197479248,
      "learning_rate": 0.0002901270772238514,
      "loss": 0.8302,
      "step": 909
    },
    {
      "epoch": 0.1424546023794615,
      "grad_norm": 1.6101166009902954,
      "learning_rate": 0.0002901026392961877,
      "loss": 1.0439,
      "step": 910
    },
    {
      "epoch": 0.1426111458985598,
      "grad_norm": 1.9140416383743286,
      "learning_rate": 0.0002900782013685239,
      "loss": 0.6344,
      "step": 911
    },
    {
      "epoch": 0.1427676894176581,
      "grad_norm": 0.9907447695732117,
      "learning_rate": 0.0002900537634408602,
      "loss": 0.6403,
      "step": 912
    },
    {
      "epoch": 0.1429242329367564,
      "grad_norm": 2.3162434101104736,
      "learning_rate": 0.00029002932551319645,
      "loss": 0.7338,
      "step": 913
    },
    {
      "epoch": 0.14308077645585474,
      "grad_norm": 1.7492724657058716,
      "learning_rate": 0.0002900048875855327,
      "loss": 0.6497,
      "step": 914
    },
    {
      "epoch": 0.14323731997495304,
      "grad_norm": 1.4608025550842285,
      "learning_rate": 0.000289980449657869,
      "loss": 0.7873,
      "step": 915
    },
    {
      "epoch": 0.14339386349405134,
      "grad_norm": 2.1269257068634033,
      "learning_rate": 0.00028995601173020525,
      "loss": 0.836,
      "step": 916
    },
    {
      "epoch": 0.14355040701314967,
      "grad_norm": 3.4742913246154785,
      "learning_rate": 0.0002899315738025415,
      "loss": 1.3075,
      "step": 917
    },
    {
      "epoch": 0.14370695053224797,
      "grad_norm": 2.038989305496216,
      "learning_rate": 0.00028990713587487776,
      "loss": 0.5859,
      "step": 918
    },
    {
      "epoch": 0.14386349405134627,
      "grad_norm": 2.4943230152130127,
      "learning_rate": 0.00028988269794721406,
      "loss": 0.925,
      "step": 919
    },
    {
      "epoch": 0.14402003757044457,
      "grad_norm": 1.818659782409668,
      "learning_rate": 0.0002898582600195503,
      "loss": 1.0819,
      "step": 920
    },
    {
      "epoch": 0.1441765810895429,
      "grad_norm": 2.3081328868865967,
      "learning_rate": 0.00028983382209188656,
      "loss": 0.7397,
      "step": 921
    },
    {
      "epoch": 0.1443331246086412,
      "grad_norm": 2.66398024559021,
      "learning_rate": 0.00028980938416422287,
      "loss": 0.7526,
      "step": 922
    },
    {
      "epoch": 0.1444896681277395,
      "grad_norm": 1.6944574117660522,
      "learning_rate": 0.0002897849462365591,
      "loss": 1.3631,
      "step": 923
    },
    {
      "epoch": 0.14464621164683783,
      "grad_norm": 3.4251976013183594,
      "learning_rate": 0.00028976050830889537,
      "loss": 1.302,
      "step": 924
    },
    {
      "epoch": 0.14480275516593613,
      "grad_norm": 6.1071062088012695,
      "learning_rate": 0.0002897360703812317,
      "loss": 0.9843,
      "step": 925
    },
    {
      "epoch": 0.14495929868503443,
      "grad_norm": 4.849330902099609,
      "learning_rate": 0.0002897116324535679,
      "loss": 1.802,
      "step": 926
    },
    {
      "epoch": 0.14511584220413276,
      "grad_norm": 3.4698123931884766,
      "learning_rate": 0.0002896871945259042,
      "loss": 1.2331,
      "step": 927
    },
    {
      "epoch": 0.14527238572323106,
      "grad_norm": 2.4112536907196045,
      "learning_rate": 0.00028966275659824043,
      "loss": 1.5486,
      "step": 928
    },
    {
      "epoch": 0.14542892924232936,
      "grad_norm": 3.808506965637207,
      "learning_rate": 0.0002896383186705767,
      "loss": 1.6323,
      "step": 929
    },
    {
      "epoch": 0.14558547276142766,
      "grad_norm": 2.8173084259033203,
      "learning_rate": 0.000289613880742913,
      "loss": 1.4004,
      "step": 930
    },
    {
      "epoch": 0.145742016280526,
      "grad_norm": 2.575488805770874,
      "learning_rate": 0.00028958944281524924,
      "loss": 1.2148,
      "step": 931
    },
    {
      "epoch": 0.1458985597996243,
      "grad_norm": 2.422727108001709,
      "learning_rate": 0.0002895650048875855,
      "loss": 1.6757,
      "step": 932
    },
    {
      "epoch": 0.1460551033187226,
      "grad_norm": 1.202463984489441,
      "learning_rate": 0.0002895405669599218,
      "loss": 0.7642,
      "step": 933
    },
    {
      "epoch": 0.14621164683782092,
      "grad_norm": 2.1168575286865234,
      "learning_rate": 0.00028951612903225805,
      "loss": 1.628,
      "step": 934
    },
    {
      "epoch": 0.14636819035691923,
      "grad_norm": 5.3358917236328125,
      "learning_rate": 0.0002894916911045943,
      "loss": 1.873,
      "step": 935
    },
    {
      "epoch": 0.14652473387601753,
      "grad_norm": 4.962363243103027,
      "learning_rate": 0.00028946725317693055,
      "loss": 1.5203,
      "step": 936
    },
    {
      "epoch": 0.14668127739511586,
      "grad_norm": 2.7353806495666504,
      "learning_rate": 0.00028944281524926686,
      "loss": 1.3402,
      "step": 937
    },
    {
      "epoch": 0.14683782091421416,
      "grad_norm": 2.9888954162597656,
      "learning_rate": 0.0002894183773216031,
      "loss": 1.7889,
      "step": 938
    },
    {
      "epoch": 0.14699436443331246,
      "grad_norm": 2.7625157833099365,
      "learning_rate": 0.00028939393939393936,
      "loss": 1.7595,
      "step": 939
    },
    {
      "epoch": 0.14715090795241076,
      "grad_norm": 1.8934247493743896,
      "learning_rate": 0.00028936950146627566,
      "loss": 1.1726,
      "step": 940
    },
    {
      "epoch": 0.1473074514715091,
      "grad_norm": 2.089092969894409,
      "learning_rate": 0.0002893450635386119,
      "loss": 1.0545,
      "step": 941
    },
    {
      "epoch": 0.1474639949906074,
      "grad_norm": 2.7847442626953125,
      "learning_rate": 0.00028932062561094817,
      "loss": 1.4845,
      "step": 942
    },
    {
      "epoch": 0.1476205385097057,
      "grad_norm": 5.1051740646362305,
      "learning_rate": 0.0002892961876832844,
      "loss": 1.5035,
      "step": 943
    },
    {
      "epoch": 0.14777708202880402,
      "grad_norm": 4.7735161781311035,
      "learning_rate": 0.00028927174975562067,
      "loss": 1.6538,
      "step": 944
    },
    {
      "epoch": 0.14793362554790232,
      "grad_norm": 2.4726431369781494,
      "learning_rate": 0.000289247311827957,
      "loss": 1.2094,
      "step": 945
    },
    {
      "epoch": 0.14809016906700062,
      "grad_norm": 3.6386454105377197,
      "learning_rate": 0.0002892228739002932,
      "loss": 1.1057,
      "step": 946
    },
    {
      "epoch": 0.14824671258609892,
      "grad_norm": 2.5107581615448,
      "learning_rate": 0.0002891984359726295,
      "loss": 0.9869,
      "step": 947
    },
    {
      "epoch": 0.14840325610519725,
      "grad_norm": 2.7597451210021973,
      "learning_rate": 0.0002891739980449658,
      "loss": 1.413,
      "step": 948
    },
    {
      "epoch": 0.14855979962429555,
      "grad_norm": 5.333527565002441,
      "learning_rate": 0.00028914956011730203,
      "loss": 1.6703,
      "step": 949
    },
    {
      "epoch": 0.14871634314339385,
      "grad_norm": 2.9326138496398926,
      "learning_rate": 0.0002891251221896383,
      "loss": 1.8701,
      "step": 950
    },
    {
      "epoch": 0.14887288666249218,
      "grad_norm": 1.1659024953842163,
      "learning_rate": 0.00028910068426197454,
      "loss": 0.9318,
      "step": 951
    },
    {
      "epoch": 0.14902943018159048,
      "grad_norm": 0.989399790763855,
      "learning_rate": 0.00028907624633431084,
      "loss": 0.824,
      "step": 952
    },
    {
      "epoch": 0.14918597370068878,
      "grad_norm": 0.8183377385139465,
      "learning_rate": 0.0002890518084066471,
      "loss": 0.9731,
      "step": 953
    },
    {
      "epoch": 0.1493425172197871,
      "grad_norm": 0.9016216993331909,
      "learning_rate": 0.00028902737047898334,
      "loss": 0.6404,
      "step": 954
    },
    {
      "epoch": 0.14949906073888541,
      "grad_norm": 0.7788622975349426,
      "learning_rate": 0.00028900293255131965,
      "loss": 0.7309,
      "step": 955
    },
    {
      "epoch": 0.14965560425798372,
      "grad_norm": 0.9964921474456787,
      "learning_rate": 0.0002889784946236559,
      "loss": 0.8296,
      "step": 956
    },
    {
      "epoch": 0.14981214777708202,
      "grad_norm": 1.0400006771087646,
      "learning_rate": 0.00028895405669599215,
      "loss": 0.7825,
      "step": 957
    },
    {
      "epoch": 0.14996869129618035,
      "grad_norm": 1.093723177909851,
      "learning_rate": 0.00028892961876832846,
      "loss": 0.6284,
      "step": 958
    },
    {
      "epoch": 0.15012523481527865,
      "grad_norm": 0.8602253198623657,
      "learning_rate": 0.00028890518084066465,
      "loss": 0.606,
      "step": 959
    },
    {
      "epoch": 0.15028177833437695,
      "grad_norm": 1.0207772254943848,
      "learning_rate": 0.00028888074291300096,
      "loss": 0.6316,
      "step": 960
    },
    {
      "epoch": 0.15043832185347528,
      "grad_norm": 1.8470277786254883,
      "learning_rate": 0.0002888563049853372,
      "loss": 0.9523,
      "step": 961
    },
    {
      "epoch": 0.15059486537257358,
      "grad_norm": 1.5962973833084106,
      "learning_rate": 0.00028883186705767346,
      "loss": 0.7024,
      "step": 962
    },
    {
      "epoch": 0.15075140889167188,
      "grad_norm": 1.2844789028167725,
      "learning_rate": 0.00028880742913000977,
      "loss": 1.0965,
      "step": 963
    },
    {
      "epoch": 0.1509079524107702,
      "grad_norm": 1.1468544006347656,
      "learning_rate": 0.000288782991202346,
      "loss": 0.8194,
      "step": 964
    },
    {
      "epoch": 0.1510644959298685,
      "grad_norm": 1.3954484462738037,
      "learning_rate": 0.00028875855327468227,
      "loss": 0.5367,
      "step": 965
    },
    {
      "epoch": 0.1512210394489668,
      "grad_norm": 1.3916863203048706,
      "learning_rate": 0.0002887341153470185,
      "loss": 0.7351,
      "step": 966
    },
    {
      "epoch": 0.1513775829680651,
      "grad_norm": 1.7426979541778564,
      "learning_rate": 0.0002887096774193548,
      "loss": 1.0045,
      "step": 967
    },
    {
      "epoch": 0.15153412648716344,
      "grad_norm": 1.407542109489441,
      "learning_rate": 0.0002886852394916911,
      "loss": 0.889,
      "step": 968
    },
    {
      "epoch": 0.15169067000626174,
      "grad_norm": 1.1983486413955688,
      "learning_rate": 0.00028866080156402733,
      "loss": 0.9749,
      "step": 969
    },
    {
      "epoch": 0.15184721352536004,
      "grad_norm": 2.5818674564361572,
      "learning_rate": 0.00028863636363636363,
      "loss": 0.8094,
      "step": 970
    },
    {
      "epoch": 0.15200375704445837,
      "grad_norm": 1.2199925184249878,
      "learning_rate": 0.0002886119257086999,
      "loss": 1.3122,
      "step": 971
    },
    {
      "epoch": 0.15216030056355667,
      "grad_norm": 1.924811840057373,
      "learning_rate": 0.00028858748778103614,
      "loss": 0.9118,
      "step": 972
    },
    {
      "epoch": 0.15231684408265497,
      "grad_norm": 1.6180088520050049,
      "learning_rate": 0.00028856304985337244,
      "loss": 0.9576,
      "step": 973
    },
    {
      "epoch": 0.15247338760175327,
      "grad_norm": 2.2538881301879883,
      "learning_rate": 0.00028853861192570864,
      "loss": 1.1137,
      "step": 974
    },
    {
      "epoch": 0.1526299311208516,
      "grad_norm": 3.2477221488952637,
      "learning_rate": 0.00028851417399804494,
      "loss": 1.6183,
      "step": 975
    },
    {
      "epoch": 0.1527864746399499,
      "grad_norm": 2.37235426902771,
      "learning_rate": 0.0002884897360703812,
      "loss": 0.978,
      "step": 976
    },
    {
      "epoch": 0.1529430181590482,
      "grad_norm": 2.178849697113037,
      "learning_rate": 0.00028846529814271745,
      "loss": 1.0761,
      "step": 977
    },
    {
      "epoch": 0.15309956167814653,
      "grad_norm": 2.8794000148773193,
      "learning_rate": 0.00028844086021505375,
      "loss": 1.2112,
      "step": 978
    },
    {
      "epoch": 0.15325610519724484,
      "grad_norm": 2.8864026069641113,
      "learning_rate": 0.00028841642228739,
      "loss": 1.4875,
      "step": 979
    },
    {
      "epoch": 0.15341264871634314,
      "grad_norm": 2.0675783157348633,
      "learning_rate": 0.00028839198435972626,
      "loss": 1.2043,
      "step": 980
    },
    {
      "epoch": 0.15356919223544147,
      "grad_norm": 1.339497685432434,
      "learning_rate": 0.00028836754643206256,
      "loss": 1.2546,
      "step": 981
    },
    {
      "epoch": 0.15372573575453977,
      "grad_norm": 2.3659565448760986,
      "learning_rate": 0.0002883431085043988,
      "loss": 1.3646,
      "step": 982
    },
    {
      "epoch": 0.15388227927363807,
      "grad_norm": 2.711576461791992,
      "learning_rate": 0.00028831867057673506,
      "loss": 1.3132,
      "step": 983
    },
    {
      "epoch": 0.15403882279273637,
      "grad_norm": 2.318260431289673,
      "learning_rate": 0.0002882942326490713,
      "loss": 1.5859,
      "step": 984
    },
    {
      "epoch": 0.1541953663118347,
      "grad_norm": 1.689608097076416,
      "learning_rate": 0.0002882697947214076,
      "loss": 1.1586,
      "step": 985
    },
    {
      "epoch": 0.154351909830933,
      "grad_norm": 2.5589444637298584,
      "learning_rate": 0.00028824535679374387,
      "loss": 1.1056,
      "step": 986
    },
    {
      "epoch": 0.1545084533500313,
      "grad_norm": 2.314589262008667,
      "learning_rate": 0.0002882209188660801,
      "loss": 1.7521,
      "step": 987
    },
    {
      "epoch": 0.15466499686912963,
      "grad_norm": 3.8083949089050293,
      "learning_rate": 0.00028819648093841643,
      "loss": 1.7522,
      "step": 988
    },
    {
      "epoch": 0.15482154038822793,
      "grad_norm": 2.452573537826538,
      "learning_rate": 0.0002881720430107526,
      "loss": 1.1393,
      "step": 989
    },
    {
      "epoch": 0.15497808390732623,
      "grad_norm": 2.592909336090088,
      "learning_rate": 0.00028814760508308893,
      "loss": 2.0492,
      "step": 990
    },
    {
      "epoch": 0.15513462742642456,
      "grad_norm": 2.1113009452819824,
      "learning_rate": 0.0002881231671554252,
      "loss": 1.3064,
      "step": 991
    },
    {
      "epoch": 0.15529117094552286,
      "grad_norm": 2.0039310455322266,
      "learning_rate": 0.00028809872922776143,
      "loss": 1.1432,
      "step": 992
    },
    {
      "epoch": 0.15544771446462116,
      "grad_norm": 3.2623660564422607,
      "learning_rate": 0.00028807429130009774,
      "loss": 1.5518,
      "step": 993
    },
    {
      "epoch": 0.15560425798371946,
      "grad_norm": 1.7344200611114502,
      "learning_rate": 0.000288049853372434,
      "loss": 1.2908,
      "step": 994
    },
    {
      "epoch": 0.1557608015028178,
      "grad_norm": 3.0457141399383545,
      "learning_rate": 0.00028802541544477024,
      "loss": 1.8197,
      "step": 995
    },
    {
      "epoch": 0.1559173450219161,
      "grad_norm": 3.5465786457061768,
      "learning_rate": 0.00028800097751710655,
      "loss": 1.2153,
      "step": 996
    },
    {
      "epoch": 0.1560738885410144,
      "grad_norm": 2.4094431400299072,
      "learning_rate": 0.0002879765395894428,
      "loss": 1.3038,
      "step": 997
    },
    {
      "epoch": 0.15623043206011272,
      "grad_norm": 4.36583948135376,
      "learning_rate": 0.00028795210166177905,
      "loss": 1.4192,
      "step": 998
    },
    {
      "epoch": 0.15638697557921102,
      "grad_norm": 2.3406410217285156,
      "learning_rate": 0.0002879276637341153,
      "loss": 1.0493,
      "step": 999
    },
    {
      "epoch": 0.15654351909830932,
      "grad_norm": 1.9417458772659302,
      "learning_rate": 0.0002879032258064516,
      "loss": 1.0919,
      "step": 1000
    },
    {
      "epoch": 0.15654351909830932,
      "eval_loss": 1.01686429977417,
      "eval_runtime": 205.1557,
      "eval_samples_per_second": 60.359,
      "eval_steps_per_second": 3.773,
      "eval_wer": 0.7063561212997004,
      "step": 1000
    },
    {
      "epoch": 0.15670006261740763,
      "grad_norm": 1.0563576221466064,
      "learning_rate": 0.00028787878787878786,
      "loss": 0.8153,
      "step": 1001
    },
    {
      "epoch": 0.15685660613650595,
      "grad_norm": 0.817237913608551,
      "learning_rate": 0.0002878543499511241,
      "loss": 0.6999,
      "step": 1002
    },
    {
      "epoch": 0.15701314965560426,
      "grad_norm": 0.6916754841804504,
      "learning_rate": 0.0002878299120234604,
      "loss": 0.5458,
      "step": 1003
    },
    {
      "epoch": 0.15716969317470256,
      "grad_norm": 0.7108484506607056,
      "learning_rate": 0.00028780547409579666,
      "loss": 0.677,
      "step": 1004
    },
    {
      "epoch": 0.15732623669380089,
      "grad_norm": 1.1377077102661133,
      "learning_rate": 0.0002877810361681329,
      "loss": 0.7012,
      "step": 1005
    },
    {
      "epoch": 0.1574827802128992,
      "grad_norm": 1.456892967224121,
      "learning_rate": 0.00028775659824046917,
      "loss": 0.6656,
      "step": 1006
    },
    {
      "epoch": 0.1576393237319975,
      "grad_norm": 1.1236454248428345,
      "learning_rate": 0.0002877321603128054,
      "loss": 0.6829,
      "step": 1007
    },
    {
      "epoch": 0.15779586725109582,
      "grad_norm": 2.5703699588775635,
      "learning_rate": 0.0002877077223851417,
      "loss": 1.0528,
      "step": 1008
    },
    {
      "epoch": 0.15795241077019412,
      "grad_norm": 1.1044304370880127,
      "learning_rate": 0.000287683284457478,
      "loss": 0.6504,
      "step": 1009
    },
    {
      "epoch": 0.15810895428929242,
      "grad_norm": 1.558677315711975,
      "learning_rate": 0.0002876588465298142,
      "loss": 0.654,
      "step": 1010
    },
    {
      "epoch": 0.15826549780839072,
      "grad_norm": 1.6468641757965088,
      "learning_rate": 0.00028763440860215053,
      "loss": 0.7489,
      "step": 1011
    },
    {
      "epoch": 0.15842204132748905,
      "grad_norm": 1.1569092273712158,
      "learning_rate": 0.0002876099706744868,
      "loss": 0.8098,
      "step": 1012
    },
    {
      "epoch": 0.15857858484658735,
      "grad_norm": 1.2596278190612793,
      "learning_rate": 0.00028758553274682303,
      "loss": 0.8457,
      "step": 1013
    },
    {
      "epoch": 0.15873512836568565,
      "grad_norm": 0.9945486187934875,
      "learning_rate": 0.0002875610948191593,
      "loss": 0.6101,
      "step": 1014
    },
    {
      "epoch": 0.15889167188478398,
      "grad_norm": 1.4740206003189087,
      "learning_rate": 0.0002875366568914956,
      "loss": 0.7989,
      "step": 1015
    },
    {
      "epoch": 0.15904821540388228,
      "grad_norm": 4.350058555603027,
      "learning_rate": 0.00028751221896383184,
      "loss": 0.9334,
      "step": 1016
    },
    {
      "epoch": 0.15920475892298058,
      "grad_norm": 1.1375359296798706,
      "learning_rate": 0.0002874877810361681,
      "loss": 0.6877,
      "step": 1017
    },
    {
      "epoch": 0.1593613024420789,
      "grad_norm": 2.1333682537078857,
      "learning_rate": 0.0002874633431085044,
      "loss": 0.8379,
      "step": 1018
    },
    {
      "epoch": 0.1595178459611772,
      "grad_norm": 1.3724780082702637,
      "learning_rate": 0.00028743890518084065,
      "loss": 1.0991,
      "step": 1019
    },
    {
      "epoch": 0.1596743894802755,
      "grad_norm": 4.463011741638184,
      "learning_rate": 0.0002874144672531769,
      "loss": 1.4866,
      "step": 1020
    },
    {
      "epoch": 0.15983093299937381,
      "grad_norm": 2.178107738494873,
      "learning_rate": 0.0002873900293255132,
      "loss": 0.7986,
      "step": 1021
    },
    {
      "epoch": 0.15998747651847214,
      "grad_norm": 1.3757675886154175,
      "learning_rate": 0.0002873655913978494,
      "loss": 0.8131,
      "step": 1022
    },
    {
      "epoch": 0.16014402003757044,
      "grad_norm": 1.6082704067230225,
      "learning_rate": 0.0002873411534701857,
      "loss": 1.0131,
      "step": 1023
    },
    {
      "epoch": 0.16030056355666875,
      "grad_norm": 1.9035452604293823,
      "learning_rate": 0.00028731671554252196,
      "loss": 0.9901,
      "step": 1024
    },
    {
      "epoch": 0.16045710707576707,
      "grad_norm": 2.227452039718628,
      "learning_rate": 0.0002872922776148582,
      "loss": 1.0808,
      "step": 1025
    },
    {
      "epoch": 0.16061365059486538,
      "grad_norm": 2.5133438110351562,
      "learning_rate": 0.0002872678396871945,
      "loss": 1.4922,
      "step": 1026
    },
    {
      "epoch": 0.16077019411396368,
      "grad_norm": 3.3356120586395264,
      "learning_rate": 0.00028724340175953077,
      "loss": 1.2555,
      "step": 1027
    },
    {
      "epoch": 0.16092673763306198,
      "grad_norm": 2.868877649307251,
      "learning_rate": 0.000287218963831867,
      "loss": 1.3604,
      "step": 1028
    },
    {
      "epoch": 0.1610832811521603,
      "grad_norm": 2.3197638988494873,
      "learning_rate": 0.0002871945259042033,
      "loss": 1.4672,
      "step": 1029
    },
    {
      "epoch": 0.1612398246712586,
      "grad_norm": 1.5867116451263428,
      "learning_rate": 0.0002871700879765396,
      "loss": 0.7615,
      "step": 1030
    },
    {
      "epoch": 0.1613963681903569,
      "grad_norm": 1.823372721672058,
      "learning_rate": 0.00028714565004887583,
      "loss": 1.4227,
      "step": 1031
    },
    {
      "epoch": 0.16155291170945524,
      "grad_norm": 3.756685972213745,
      "learning_rate": 0.0002871212121212121,
      "loss": 1.9678,
      "step": 1032
    },
    {
      "epoch": 0.16170945522855354,
      "grad_norm": 2.313270330429077,
      "learning_rate": 0.0002870967741935484,
      "loss": 0.9197,
      "step": 1033
    },
    {
      "epoch": 0.16186599874765184,
      "grad_norm": 1.9607908725738525,
      "learning_rate": 0.00028707233626588464,
      "loss": 1.4155,
      "step": 1034
    },
    {
      "epoch": 0.16202254226675017,
      "grad_norm": 2.0244407653808594,
      "learning_rate": 0.0002870478983382209,
      "loss": 1.2159,
      "step": 1035
    },
    {
      "epoch": 0.16217908578584847,
      "grad_norm": 2.4551143646240234,
      "learning_rate": 0.0002870234604105572,
      "loss": 1.8048,
      "step": 1036
    },
    {
      "epoch": 0.16233562930494677,
      "grad_norm": 2.5346062183380127,
      "learning_rate": 0.0002869990224828934,
      "loss": 1.2969,
      "step": 1037
    },
    {
      "epoch": 0.16249217282404507,
      "grad_norm": 3.634019374847412,
      "learning_rate": 0.0002869745845552297,
      "loss": 2.0716,
      "step": 1038
    },
    {
      "epoch": 0.1626487163431434,
      "grad_norm": 2.7607200145721436,
      "learning_rate": 0.00028695014662756595,
      "loss": 0.8884,
      "step": 1039
    },
    {
      "epoch": 0.1628052598622417,
      "grad_norm": 2.8792600631713867,
      "learning_rate": 0.0002869257086999022,
      "loss": 1.5306,
      "step": 1040
    },
    {
      "epoch": 0.16296180338134,
      "grad_norm": 5.755788803100586,
      "learning_rate": 0.0002869012707722385,
      "loss": 1.4916,
      "step": 1041
    },
    {
      "epoch": 0.16311834690043833,
      "grad_norm": 2.4009017944335938,
      "learning_rate": 0.00028687683284457475,
      "loss": 1.3055,
      "step": 1042
    },
    {
      "epoch": 0.16327489041953663,
      "grad_norm": 2.3000261783599854,
      "learning_rate": 0.000286852394916911,
      "loss": 1.3189,
      "step": 1043
    },
    {
      "epoch": 0.16343143393863493,
      "grad_norm": 2.4022369384765625,
      "learning_rate": 0.0002868279569892473,
      "loss": 1.5442,
      "step": 1044
    },
    {
      "epoch": 0.16358797745773326,
      "grad_norm": 2.216566801071167,
      "learning_rate": 0.00028680351906158356,
      "loss": 1.7154,
      "step": 1045
    },
    {
      "epoch": 0.16374452097683156,
      "grad_norm": 2.9098117351531982,
      "learning_rate": 0.0002867790811339198,
      "loss": 1.1951,
      "step": 1046
    },
    {
      "epoch": 0.16390106449592987,
      "grad_norm": 4.8952178955078125,
      "learning_rate": 0.00028675464320625606,
      "loss": 1.6244,
      "step": 1047
    },
    {
      "epoch": 0.16405760801502817,
      "grad_norm": 4.096086502075195,
      "learning_rate": 0.00028673020527859237,
      "loss": 1.3874,
      "step": 1048
    },
    {
      "epoch": 0.1642141515341265,
      "grad_norm": 3.0238194465637207,
      "learning_rate": 0.0002867057673509286,
      "loss": 1.345,
      "step": 1049
    },
    {
      "epoch": 0.1643706950532248,
      "grad_norm": 2.8986921310424805,
      "learning_rate": 0.00028668132942326487,
      "loss": 1.2903,
      "step": 1050
    },
    {
      "epoch": 0.1645272385723231,
      "grad_norm": 0.9072627425193787,
      "learning_rate": 0.0002866568914956012,
      "loss": 0.6808,
      "step": 1051
    },
    {
      "epoch": 0.16468378209142143,
      "grad_norm": 0.8426551222801208,
      "learning_rate": 0.00028663245356793743,
      "loss": 0.6256,
      "step": 1052
    },
    {
      "epoch": 0.16484032561051973,
      "grad_norm": 0.7460830807685852,
      "learning_rate": 0.0002866080156402737,
      "loss": 0.5074,
      "step": 1053
    },
    {
      "epoch": 0.16499686912961803,
      "grad_norm": 0.606028139591217,
      "learning_rate": 0.00028658357771260993,
      "loss": 0.6061,
      "step": 1054
    },
    {
      "epoch": 0.16515341264871633,
      "grad_norm": 0.833991527557373,
      "learning_rate": 0.0002865591397849462,
      "loss": 0.5883,
      "step": 1055
    },
    {
      "epoch": 0.16530995616781466,
      "grad_norm": 1.9847928285598755,
      "learning_rate": 0.0002865347018572825,
      "loss": 0.7446,
      "step": 1056
    },
    {
      "epoch": 0.16546649968691296,
      "grad_norm": 0.9032483100891113,
      "learning_rate": 0.00028651026392961874,
      "loss": 0.5827,
      "step": 1057
    },
    {
      "epoch": 0.16562304320601126,
      "grad_norm": 0.7685295343399048,
      "learning_rate": 0.000286485826001955,
      "loss": 0.6579,
      "step": 1058
    },
    {
      "epoch": 0.1657795867251096,
      "grad_norm": 0.6518795490264893,
      "learning_rate": 0.0002864613880742913,
      "loss": 0.5019,
      "step": 1059
    },
    {
      "epoch": 0.1659361302442079,
      "grad_norm": 1.02455735206604,
      "learning_rate": 0.00028643695014662755,
      "loss": 0.4796,
      "step": 1060
    },
    {
      "epoch": 0.1660926737633062,
      "grad_norm": 1.0418224334716797,
      "learning_rate": 0.0002864125122189638,
      "loss": 0.5102,
      "step": 1061
    },
    {
      "epoch": 0.16624921728240452,
      "grad_norm": 2.0449554920196533,
      "learning_rate": 0.00028638807429130005,
      "loss": 0.7181,
      "step": 1062
    },
    {
      "epoch": 0.16640576080150282,
      "grad_norm": 1.0994571447372437,
      "learning_rate": 0.00028636363636363636,
      "loss": 0.7923,
      "step": 1063
    },
    {
      "epoch": 0.16656230432060112,
      "grad_norm": 0.9208387136459351,
      "learning_rate": 0.0002863391984359726,
      "loss": 0.5434,
      "step": 1064
    },
    {
      "epoch": 0.16671884783969942,
      "grad_norm": 1.2651475667953491,
      "learning_rate": 0.00028631476050830886,
      "loss": 1.0223,
      "step": 1065
    },
    {
      "epoch": 0.16687539135879775,
      "grad_norm": 1.9309346675872803,
      "learning_rate": 0.00028629032258064516,
      "loss": 0.8762,
      "step": 1066
    },
    {
      "epoch": 0.16703193487789605,
      "grad_norm": 2.4084434509277344,
      "learning_rate": 0.0002862658846529814,
      "loss": 0.9494,
      "step": 1067
    },
    {
      "epoch": 0.16718847839699436,
      "grad_norm": 1.4075591564178467,
      "learning_rate": 0.00028624144672531767,
      "loss": 0.7541,
      "step": 1068
    },
    {
      "epoch": 0.16734502191609268,
      "grad_norm": 2.102005958557129,
      "learning_rate": 0.00028621700879765397,
      "loss": 0.8695,
      "step": 1069
    },
    {
      "epoch": 0.16750156543519099,
      "grad_norm": 2.49064040184021,
      "learning_rate": 0.00028619257086999017,
      "loss": 0.7304,
      "step": 1070
    },
    {
      "epoch": 0.1676581089542893,
      "grad_norm": 1.221191167831421,
      "learning_rate": 0.0002861681329423265,
      "loss": 0.6588,
      "step": 1071
    },
    {
      "epoch": 0.16781465247338762,
      "grad_norm": 1.5001965761184692,
      "learning_rate": 0.0002861436950146627,
      "loss": 1.1985,
      "step": 1072
    },
    {
      "epoch": 0.16797119599248592,
      "grad_norm": 1.4838846921920776,
      "learning_rate": 0.000286119257086999,
      "loss": 0.9565,
      "step": 1073
    },
    {
      "epoch": 0.16812773951158422,
      "grad_norm": 1.5427438020706177,
      "learning_rate": 0.0002860948191593353,
      "loss": 0.8253,
      "step": 1074
    },
    {
      "epoch": 0.16828428303068252,
      "grad_norm": 1.4111217260360718,
      "learning_rate": 0.00028607038123167153,
      "loss": 0.6618,
      "step": 1075
    },
    {
      "epoch": 0.16844082654978085,
      "grad_norm": 1.6827608346939087,
      "learning_rate": 0.0002860459433040078,
      "loss": 0.9508,
      "step": 1076
    },
    {
      "epoch": 0.16859737006887915,
      "grad_norm": 1.9233096837997437,
      "learning_rate": 0.00028602150537634404,
      "loss": 1.5639,
      "step": 1077
    },
    {
      "epoch": 0.16875391358797745,
      "grad_norm": 1.4510252475738525,
      "learning_rate": 0.00028599706744868034,
      "loss": 0.9704,
      "step": 1078
    },
    {
      "epoch": 0.16891045710707578,
      "grad_norm": 3.087716817855835,
      "learning_rate": 0.0002859726295210166,
      "loss": 0.979,
      "step": 1079
    },
    {
      "epoch": 0.16906700062617408,
      "grad_norm": 3.0797057151794434,
      "learning_rate": 0.00028594819159335284,
      "loss": 0.6662,
      "step": 1080
    },
    {
      "epoch": 0.16922354414527238,
      "grad_norm": 2.5428311824798584,
      "learning_rate": 0.00028592375366568915,
      "loss": 1.4854,
      "step": 1081
    },
    {
      "epoch": 0.16938008766437068,
      "grad_norm": 1.8000260591506958,
      "learning_rate": 0.0002858993157380254,
      "loss": 1.3866,
      "step": 1082
    },
    {
      "epoch": 0.169536631183469,
      "grad_norm": 3.2034716606140137,
      "learning_rate": 0.00028587487781036165,
      "loss": 1.1556,
      "step": 1083
    },
    {
      "epoch": 0.1696931747025673,
      "grad_norm": 1.544206142425537,
      "learning_rate": 0.00028585043988269796,
      "loss": 1.0316,
      "step": 1084
    },
    {
      "epoch": 0.1698497182216656,
      "grad_norm": 2.489361047744751,
      "learning_rate": 0.00028582600195503415,
      "loss": 1.1934,
      "step": 1085
    },
    {
      "epoch": 0.17000626174076394,
      "grad_norm": 2.156745672225952,
      "learning_rate": 0.00028580156402737046,
      "loss": 1.4824,
      "step": 1086
    },
    {
      "epoch": 0.17016280525986224,
      "grad_norm": 2.317676305770874,
      "learning_rate": 0.0002857771260997067,
      "loss": 1.266,
      "step": 1087
    },
    {
      "epoch": 0.17031934877896054,
      "grad_norm": 1.6004157066345215,
      "learning_rate": 0.00028575268817204296,
      "loss": 1.0838,
      "step": 1088
    },
    {
      "epoch": 0.17047589229805887,
      "grad_norm": 1.9848181009292603,
      "learning_rate": 0.00028572825024437927,
      "loss": 1.274,
      "step": 1089
    },
    {
      "epoch": 0.17063243581715717,
      "grad_norm": 2.925771474838257,
      "learning_rate": 0.0002857038123167155,
      "loss": 1.2818,
      "step": 1090
    },
    {
      "epoch": 0.17078897933625548,
      "grad_norm": 1.8362958431243896,
      "learning_rate": 0.00028567937438905177,
      "loss": 1.0246,
      "step": 1091
    },
    {
      "epoch": 0.17094552285535378,
      "grad_norm": 4.179997444152832,
      "learning_rate": 0.0002856549364613881,
      "loss": 2.1207,
      "step": 1092
    },
    {
      "epoch": 0.1711020663744521,
      "grad_norm": 4.3508429527282715,
      "learning_rate": 0.0002856304985337243,
      "loss": 2.0854,
      "step": 1093
    },
    {
      "epoch": 0.1712586098935504,
      "grad_norm": 6.442362308502197,
      "learning_rate": 0.0002856060606060606,
      "loss": 1.654,
      "step": 1094
    },
    {
      "epoch": 0.1714151534126487,
      "grad_norm": 1.9675512313842773,
      "learning_rate": 0.00028558162267839683,
      "loss": 1.2959,
      "step": 1095
    },
    {
      "epoch": 0.17157169693174704,
      "grad_norm": 2.1139070987701416,
      "learning_rate": 0.00028555718475073313,
      "loss": 1.4101,
      "step": 1096
    },
    {
      "epoch": 0.17172824045084534,
      "grad_norm": 1.998644232749939,
      "learning_rate": 0.0002855327468230694,
      "loss": 0.6606,
      "step": 1097
    },
    {
      "epoch": 0.17188478396994364,
      "grad_norm": 3.3093795776367188,
      "learning_rate": 0.00028550830889540564,
      "loss": 0.8759,
      "step": 1098
    },
    {
      "epoch": 0.17204132748904197,
      "grad_norm": 3.0865354537963867,
      "learning_rate": 0.00028548387096774194,
      "loss": 0.5971,
      "step": 1099
    },
    {
      "epoch": 0.17219787100814027,
      "grad_norm": 1.6972962617874146,
      "learning_rate": 0.0002854594330400782,
      "loss": 0.8417,
      "step": 1100
    },
    {
      "epoch": 0.17235441452723857,
      "grad_norm": 0.6339183449745178,
      "learning_rate": 0.00028543499511241445,
      "loss": 0.6272,
      "step": 1101
    },
    {
      "epoch": 0.17251095804633687,
      "grad_norm": 0.7550747990608215,
      "learning_rate": 0.0002854105571847507,
      "loss": 0.6742,
      "step": 1102
    },
    {
      "epoch": 0.1726675015654352,
      "grad_norm": 0.7940789461135864,
      "learning_rate": 0.00028538611925708695,
      "loss": 0.4962,
      "step": 1103
    },
    {
      "epoch": 0.1728240450845335,
      "grad_norm": 0.7967172265052795,
      "learning_rate": 0.00028536168132942325,
      "loss": 0.591,
      "step": 1104
    },
    {
      "epoch": 0.1729805886036318,
      "grad_norm": 0.9077056050300598,
      "learning_rate": 0.0002853372434017595,
      "loss": 0.5865,
      "step": 1105
    },
    {
      "epoch": 0.17313713212273013,
      "grad_norm": 0.7266988158226013,
      "learning_rate": 0.00028531280547409576,
      "loss": 0.5431,
      "step": 1106
    },
    {
      "epoch": 0.17329367564182843,
      "grad_norm": 1.1024484634399414,
      "learning_rate": 0.00028528836754643206,
      "loss": 0.68,
      "step": 1107
    },
    {
      "epoch": 0.17345021916092673,
      "grad_norm": 1.058496117591858,
      "learning_rate": 0.0002852639296187683,
      "loss": 0.6728,
      "step": 1108
    },
    {
      "epoch": 0.17360676268002503,
      "grad_norm": 1.3563542366027832,
      "learning_rate": 0.00028523949169110456,
      "loss": 0.6104,
      "step": 1109
    },
    {
      "epoch": 0.17376330619912336,
      "grad_norm": 0.9580490589141846,
      "learning_rate": 0.0002852150537634408,
      "loss": 0.4989,
      "step": 1110
    },
    {
      "epoch": 0.17391984971822166,
      "grad_norm": 1.4669703245162964,
      "learning_rate": 0.0002851906158357771,
      "loss": 0.7394,
      "step": 1111
    },
    {
      "epoch": 0.17407639323731997,
      "grad_norm": 1.063644289970398,
      "learning_rate": 0.00028516617790811337,
      "loss": 0.5999,
      "step": 1112
    },
    {
      "epoch": 0.1742329367564183,
      "grad_norm": 1.5729597806930542,
      "learning_rate": 0.0002851417399804496,
      "loss": 0.8043,
      "step": 1113
    },
    {
      "epoch": 0.1743894802755166,
      "grad_norm": 0.8515031933784485,
      "learning_rate": 0.00028511730205278593,
      "loss": 0.5842,
      "step": 1114
    },
    {
      "epoch": 0.1745460237946149,
      "grad_norm": 2.940446138381958,
      "learning_rate": 0.0002850928641251222,
      "loss": 1.1475,
      "step": 1115
    },
    {
      "epoch": 0.17470256731371323,
      "grad_norm": 1.5992013216018677,
      "learning_rate": 0.00028506842619745843,
      "loss": 1.125,
      "step": 1116
    },
    {
      "epoch": 0.17485911083281153,
      "grad_norm": 1.8026546239852905,
      "learning_rate": 0.00028504398826979474,
      "loss": 0.8755,
      "step": 1117
    },
    {
      "epoch": 0.17501565435190983,
      "grad_norm": 0.9795113801956177,
      "learning_rate": 0.00028501955034213093,
      "loss": 0.5423,
      "step": 1118
    },
    {
      "epoch": 0.17517219787100813,
      "grad_norm": 2.1793277263641357,
      "learning_rate": 0.00028499511241446724,
      "loss": 0.907,
      "step": 1119
    },
    {
      "epoch": 0.17532874139010646,
      "grad_norm": 6.160722255706787,
      "learning_rate": 0.0002849706744868035,
      "loss": 1.1782,
      "step": 1120
    },
    {
      "epoch": 0.17548528490920476,
      "grad_norm": 3.347403049468994,
      "learning_rate": 0.00028494623655913974,
      "loss": 1.0413,
      "step": 1121
    },
    {
      "epoch": 0.17564182842830306,
      "grad_norm": 2.783139705657959,
      "learning_rate": 0.00028492179863147605,
      "loss": 0.8199,
      "step": 1122
    },
    {
      "epoch": 0.1757983719474014,
      "grad_norm": 1.6603683233261108,
      "learning_rate": 0.0002848973607038123,
      "loss": 0.7946,
      "step": 1123
    },
    {
      "epoch": 0.1759549154664997,
      "grad_norm": 2.4885950088500977,
      "learning_rate": 0.00028487292277614855,
      "loss": 1.1803,
      "step": 1124
    },
    {
      "epoch": 0.176111458985598,
      "grad_norm": 2.89567494392395,
      "learning_rate": 0.0002848484848484848,
      "loss": 1.2259,
      "step": 1125
    },
    {
      "epoch": 0.1762680025046963,
      "grad_norm": 1.2964775562286377,
      "learning_rate": 0.0002848240469208211,
      "loss": 1.0174,
      "step": 1126
    },
    {
      "epoch": 0.17642454602379462,
      "grad_norm": 1.6365723609924316,
      "learning_rate": 0.00028479960899315736,
      "loss": 0.9492,
      "step": 1127
    },
    {
      "epoch": 0.17658108954289292,
      "grad_norm": 1.2834248542785645,
      "learning_rate": 0.0002847751710654936,
      "loss": 0.7886,
      "step": 1128
    },
    {
      "epoch": 0.17673763306199122,
      "grad_norm": 1.8275222778320312,
      "learning_rate": 0.0002847507331378299,
      "loss": 1.1145,
      "step": 1129
    },
    {
      "epoch": 0.17689417658108955,
      "grad_norm": 2.4772050380706787,
      "learning_rate": 0.00028472629521016617,
      "loss": 0.9815,
      "step": 1130
    },
    {
      "epoch": 0.17705072010018785,
      "grad_norm": 5.258149147033691,
      "learning_rate": 0.0002847018572825024,
      "loss": 1.1874,
      "step": 1131
    },
    {
      "epoch": 0.17720726361928615,
      "grad_norm": 2.3318662643432617,
      "learning_rate": 0.0002846774193548387,
      "loss": 1.308,
      "step": 1132
    },
    {
      "epoch": 0.17736380713838448,
      "grad_norm": 3.756080389022827,
      "learning_rate": 0.0002846529814271749,
      "loss": 1.0855,
      "step": 1133
    },
    {
      "epoch": 0.17752035065748278,
      "grad_norm": 2.8273825645446777,
      "learning_rate": 0.0002846285434995112,
      "loss": 1.395,
      "step": 1134
    },
    {
      "epoch": 0.17767689417658108,
      "grad_norm": 1.8789290189743042,
      "learning_rate": 0.0002846041055718475,
      "loss": 1.2854,
      "step": 1135
    },
    {
      "epoch": 0.17783343769567939,
      "grad_norm": 5.230047225952148,
      "learning_rate": 0.0002845796676441837,
      "loss": 1.3411,
      "step": 1136
    },
    {
      "epoch": 0.17798998121477771,
      "grad_norm": 3.2810747623443604,
      "learning_rate": 0.00028455522971652003,
      "loss": 1.972,
      "step": 1137
    },
    {
      "epoch": 0.17814652473387602,
      "grad_norm": 3.59177565574646,
      "learning_rate": 0.0002845307917888563,
      "loss": 1.4195,
      "step": 1138
    },
    {
      "epoch": 0.17830306825297432,
      "grad_norm": 3.564577341079712,
      "learning_rate": 0.00028450635386119253,
      "loss": 1.0904,
      "step": 1139
    },
    {
      "epoch": 0.17845961177207265,
      "grad_norm": 2.349719762802124,
      "learning_rate": 0.00028448191593352884,
      "loss": 1.4299,
      "step": 1140
    },
    {
      "epoch": 0.17861615529117095,
      "grad_norm": 2.8419268131256104,
      "learning_rate": 0.0002844574780058651,
      "loss": 1.9473,
      "step": 1141
    },
    {
      "epoch": 0.17877269881026925,
      "grad_norm": 5.5093159675598145,
      "learning_rate": 0.00028443304007820134,
      "loss": 1.7942,
      "step": 1142
    },
    {
      "epoch": 0.17892924232936758,
      "grad_norm": 2.4165680408477783,
      "learning_rate": 0.0002844086021505376,
      "loss": 1.3642,
      "step": 1143
    },
    {
      "epoch": 0.17908578584846588,
      "grad_norm": 3.020798444747925,
      "learning_rate": 0.0002843841642228739,
      "loss": 1.649,
      "step": 1144
    },
    {
      "epoch": 0.17924232936756418,
      "grad_norm": 2.0584144592285156,
      "learning_rate": 0.00028435972629521015,
      "loss": 1.4339,
      "step": 1145
    },
    {
      "epoch": 0.17939887288666248,
      "grad_norm": 3.8811330795288086,
      "learning_rate": 0.0002843352883675464,
      "loss": 1.143,
      "step": 1146
    },
    {
      "epoch": 0.1795554164057608,
      "grad_norm": 3.469479560852051,
      "learning_rate": 0.0002843108504398827,
      "loss": 1.3439,
      "step": 1147
    },
    {
      "epoch": 0.1797119599248591,
      "grad_norm": 2.772313117980957,
      "learning_rate": 0.0002842864125122189,
      "loss": 1.253,
      "step": 1148
    },
    {
      "epoch": 0.1798685034439574,
      "grad_norm": 2.618361473083496,
      "learning_rate": 0.0002842619745845552,
      "loss": 1.6485,
      "step": 1149
    },
    {
      "epoch": 0.18002504696305574,
      "grad_norm": 1.9248026609420776,
      "learning_rate": 0.00028423753665689146,
      "loss": 1.4382,
      "step": 1150
    },
    {
      "epoch": 0.18018159048215404,
      "grad_norm": 1.16201913356781,
      "learning_rate": 0.0002842130987292277,
      "loss": 0.6461,
      "step": 1151
    },
    {
      "epoch": 0.18033813400125234,
      "grad_norm": 0.9263600707054138,
      "learning_rate": 0.000284188660801564,
      "loss": 0.6146,
      "step": 1152
    },
    {
      "epoch": 0.18049467752035064,
      "grad_norm": 0.6873730421066284,
      "learning_rate": 0.00028416422287390027,
      "loss": 0.5212,
      "step": 1153
    },
    {
      "epoch": 0.18065122103944897,
      "grad_norm": 0.7068181037902832,
      "learning_rate": 0.0002841397849462365,
      "loss": 0.478,
      "step": 1154
    },
    {
      "epoch": 0.18080776455854727,
      "grad_norm": 0.824680745601654,
      "learning_rate": 0.0002841153470185728,
      "loss": 0.7398,
      "step": 1155
    },
    {
      "epoch": 0.18096430807764557,
      "grad_norm": 0.7223101854324341,
      "learning_rate": 0.0002840909090909091,
      "loss": 0.4756,
      "step": 1156
    },
    {
      "epoch": 0.1811208515967439,
      "grad_norm": 1.0076377391815186,
      "learning_rate": 0.00028406647116324533,
      "loss": 0.518,
      "step": 1157
    },
    {
      "epoch": 0.1812773951158422,
      "grad_norm": 1.0383641719818115,
      "learning_rate": 0.0002840420332355816,
      "loss": 0.604,
      "step": 1158
    },
    {
      "epoch": 0.1814339386349405,
      "grad_norm": 1.6516761779785156,
      "learning_rate": 0.0002840175953079179,
      "loss": 0.6473,
      "step": 1159
    },
    {
      "epoch": 0.18159048215403883,
      "grad_norm": 1.3219435214996338,
      "learning_rate": 0.00028399315738025414,
      "loss": 0.8136,
      "step": 1160
    },
    {
      "epoch": 0.18174702567313714,
      "grad_norm": 1.0588476657867432,
      "learning_rate": 0.0002839687194525904,
      "loss": 0.5458,
      "step": 1161
    },
    {
      "epoch": 0.18190356919223544,
      "grad_norm": 1.2911880016326904,
      "learning_rate": 0.0002839442815249267,
      "loss": 0.6775,
      "step": 1162
    },
    {
      "epoch": 0.18206011271133374,
      "grad_norm": 1.0801639556884766,
      "learning_rate": 0.00028391984359726294,
      "loss": 0.7164,
      "step": 1163
    },
    {
      "epoch": 0.18221665623043207,
      "grad_norm": 1.2428066730499268,
      "learning_rate": 0.0002838954056695992,
      "loss": 0.7186,
      "step": 1164
    },
    {
      "epoch": 0.18237319974953037,
      "grad_norm": 2.018033504486084,
      "learning_rate": 0.00028387096774193545,
      "loss": 1.0134,
      "step": 1165
    },
    {
      "epoch": 0.18252974326862867,
      "grad_norm": 1.2658600807189941,
      "learning_rate": 0.0002838465298142717,
      "loss": 0.6296,
      "step": 1166
    },
    {
      "epoch": 0.182686286787727,
      "grad_norm": 1.356482744216919,
      "learning_rate": 0.000283822091886608,
      "loss": 0.8566,
      "step": 1167
    },
    {
      "epoch": 0.1828428303068253,
      "grad_norm": 1.4970476627349854,
      "learning_rate": 0.00028379765395894425,
      "loss": 0.9361,
      "step": 1168
    },
    {
      "epoch": 0.1829993738259236,
      "grad_norm": 2.531439781188965,
      "learning_rate": 0.0002837732160312805,
      "loss": 1.3125,
      "step": 1169
    },
    {
      "epoch": 0.18315591734502193,
      "grad_norm": 1.9374449253082275,
      "learning_rate": 0.0002837487781036168,
      "loss": 1.1296,
      "step": 1170
    },
    {
      "epoch": 0.18331246086412023,
      "grad_norm": 1.2566038370132446,
      "learning_rate": 0.00028372434017595306,
      "loss": 0.9607,
      "step": 1171
    },
    {
      "epoch": 0.18346900438321853,
      "grad_norm": 1.5955857038497925,
      "learning_rate": 0.0002836999022482893,
      "loss": 1.2914,
      "step": 1172
    },
    {
      "epoch": 0.18362554790231683,
      "grad_norm": 1.207210898399353,
      "learning_rate": 0.00028367546432062557,
      "loss": 0.8891,
      "step": 1173
    },
    {
      "epoch": 0.18378209142141516,
      "grad_norm": 1.754010796546936,
      "learning_rate": 0.00028365102639296187,
      "loss": 0.8335,
      "step": 1174
    },
    {
      "epoch": 0.18393863494051346,
      "grad_norm": 1.5274769067764282,
      "learning_rate": 0.0002836265884652981,
      "loss": 1.0153,
      "step": 1175
    },
    {
      "epoch": 0.18409517845961176,
      "grad_norm": 1.2438735961914062,
      "learning_rate": 0.0002836021505376344,
      "loss": 0.7795,
      "step": 1176
    },
    {
      "epoch": 0.1842517219787101,
      "grad_norm": 1.854723572731018,
      "learning_rate": 0.0002835777126099707,
      "loss": 1.1812,
      "step": 1177
    },
    {
      "epoch": 0.1844082654978084,
      "grad_norm": 1.8179014921188354,
      "learning_rate": 0.00028355327468230693,
      "loss": 1.0775,
      "step": 1178
    },
    {
      "epoch": 0.1845648090169067,
      "grad_norm": 1.427675724029541,
      "learning_rate": 0.0002835288367546432,
      "loss": 0.7761,
      "step": 1179
    },
    {
      "epoch": 0.184721352536005,
      "grad_norm": 2.1459126472473145,
      "learning_rate": 0.0002835043988269795,
      "loss": 1.3916,
      "step": 1180
    },
    {
      "epoch": 0.18487789605510332,
      "grad_norm": 1.9748454093933105,
      "learning_rate": 0.0002834799608993157,
      "loss": 1.0506,
      "step": 1181
    },
    {
      "epoch": 0.18503443957420163,
      "grad_norm": 2.326005220413208,
      "learning_rate": 0.000283455522971652,
      "loss": 1.7368,
      "step": 1182
    },
    {
      "epoch": 0.18519098309329993,
      "grad_norm": 3.0131938457489014,
      "learning_rate": 0.00028343108504398824,
      "loss": 1.5369,
      "step": 1183
    },
    {
      "epoch": 0.18534752661239826,
      "grad_norm": 2.223701000213623,
      "learning_rate": 0.0002834066471163245,
      "loss": 1.1345,
      "step": 1184
    },
    {
      "epoch": 0.18550407013149656,
      "grad_norm": 3.37308669090271,
      "learning_rate": 0.0002833822091886608,
      "loss": 1.7771,
      "step": 1185
    },
    {
      "epoch": 0.18566061365059486,
      "grad_norm": 2.0770747661590576,
      "learning_rate": 0.00028335777126099705,
      "loss": 1.2859,
      "step": 1186
    },
    {
      "epoch": 0.1858171571696932,
      "grad_norm": 2.7626192569732666,
      "learning_rate": 0.0002833333333333333,
      "loss": 1.3574,
      "step": 1187
    },
    {
      "epoch": 0.1859737006887915,
      "grad_norm": 2.9744863510131836,
      "learning_rate": 0.0002833088954056696,
      "loss": 1.9453,
      "step": 1188
    },
    {
      "epoch": 0.1861302442078898,
      "grad_norm": 2.143836736679077,
      "learning_rate": 0.00028328445747800586,
      "loss": 1.4961,
      "step": 1189
    },
    {
      "epoch": 0.1862867877269881,
      "grad_norm": 1.8865716457366943,
      "learning_rate": 0.0002832600195503421,
      "loss": 1.7137,
      "step": 1190
    },
    {
      "epoch": 0.18644333124608642,
      "grad_norm": 2.325896978378296,
      "learning_rate": 0.00028323558162267836,
      "loss": 1.4174,
      "step": 1191
    },
    {
      "epoch": 0.18659987476518472,
      "grad_norm": 2.0641846656799316,
      "learning_rate": 0.00028321114369501466,
      "loss": 1.7222,
      "step": 1192
    },
    {
      "epoch": 0.18675641828428302,
      "grad_norm": 1.5752524137496948,
      "learning_rate": 0.0002831867057673509,
      "loss": 1.4698,
      "step": 1193
    },
    {
      "epoch": 0.18691296180338135,
      "grad_norm": 1.7678941488265991,
      "learning_rate": 0.00028316226783968717,
      "loss": 1.1898,
      "step": 1194
    },
    {
      "epoch": 0.18706950532247965,
      "grad_norm": 2.59869384765625,
      "learning_rate": 0.00028313782991202347,
      "loss": 1.2828,
      "step": 1195
    },
    {
      "epoch": 0.18722604884157795,
      "grad_norm": 1.4962000846862793,
      "learning_rate": 0.00028311339198435967,
      "loss": 0.9471,
      "step": 1196
    },
    {
      "epoch": 0.18738259236067628,
      "grad_norm": 2.0714030265808105,
      "learning_rate": 0.000283088954056696,
      "loss": 1.1181,
      "step": 1197
    },
    {
      "epoch": 0.18753913587977458,
      "grad_norm": 1.3907320499420166,
      "learning_rate": 0.0002830645161290322,
      "loss": 1.0763,
      "step": 1198
    },
    {
      "epoch": 0.18769567939887288,
      "grad_norm": 2.7507219314575195,
      "learning_rate": 0.0002830400782013685,
      "loss": 1.3243,
      "step": 1199
    },
    {
      "epoch": 0.18785222291797118,
      "grad_norm": 2.2302870750427246,
      "learning_rate": 0.0002830156402737048,
      "loss": 1.1937,
      "step": 1200
    },
    {
      "epoch": 0.1880087664370695,
      "grad_norm": 0.9416796565055847,
      "learning_rate": 0.00028299120234604103,
      "loss": 0.6451,
      "step": 1201
    },
    {
      "epoch": 0.18816530995616781,
      "grad_norm": 0.6642339825630188,
      "learning_rate": 0.0002829667644183773,
      "loss": 0.483,
      "step": 1202
    },
    {
      "epoch": 0.18832185347526612,
      "grad_norm": 0.9438838958740234,
      "learning_rate": 0.0002829423264907136,
      "loss": 0.9268,
      "step": 1203
    },
    {
      "epoch": 0.18847839699436444,
      "grad_norm": 0.6609278917312622,
      "learning_rate": 0.00028291788856304984,
      "loss": 0.6353,
      "step": 1204
    },
    {
      "epoch": 0.18863494051346275,
      "grad_norm": 0.7672062516212463,
      "learning_rate": 0.0002828934506353861,
      "loss": 0.6421,
      "step": 1205
    },
    {
      "epoch": 0.18879148403256105,
      "grad_norm": 0.7625839114189148,
      "learning_rate": 0.00028286901270772234,
      "loss": 0.579,
      "step": 1206
    },
    {
      "epoch": 0.18894802755165935,
      "grad_norm": 1.4474471807479858,
      "learning_rate": 0.00028284457478005865,
      "loss": 0.7153,
      "step": 1207
    },
    {
      "epoch": 0.18910457107075768,
      "grad_norm": 1.2392566204071045,
      "learning_rate": 0.0002828201368523949,
      "loss": 0.5436,
      "step": 1208
    },
    {
      "epoch": 0.18926111458985598,
      "grad_norm": 1.2932655811309814,
      "learning_rate": 0.00028279569892473115,
      "loss": 0.5473,
      "step": 1209
    },
    {
      "epoch": 0.18941765810895428,
      "grad_norm": 1.3752779960632324,
      "learning_rate": 0.00028277126099706746,
      "loss": 0.5555,
      "step": 1210
    },
    {
      "epoch": 0.1895742016280526,
      "grad_norm": 0.6900332570075989,
      "learning_rate": 0.0002827468230694037,
      "loss": 0.4844,
      "step": 1211
    },
    {
      "epoch": 0.1897307451471509,
      "grad_norm": 1.8527002334594727,
      "learning_rate": 0.00028272238514173996,
      "loss": 1.2104,
      "step": 1212
    },
    {
      "epoch": 0.1898872886662492,
      "grad_norm": 1.6955235004425049,
      "learning_rate": 0.0002826979472140762,
      "loss": 0.7639,
      "step": 1213
    },
    {
      "epoch": 0.19004383218534754,
      "grad_norm": 1.2351977825164795,
      "learning_rate": 0.00028267350928641246,
      "loss": 0.6054,
      "step": 1214
    },
    {
      "epoch": 0.19020037570444584,
      "grad_norm": 1.1425282955169678,
      "learning_rate": 0.00028264907135874877,
      "loss": 0.6811,
      "step": 1215
    },
    {
      "epoch": 0.19035691922354414,
      "grad_norm": 1.043776273727417,
      "learning_rate": 0.000282624633431085,
      "loss": 0.7065,
      "step": 1216
    },
    {
      "epoch": 0.19051346274264244,
      "grad_norm": 1.4897061586380005,
      "learning_rate": 0.00028260019550342127,
      "loss": 0.8076,
      "step": 1217
    },
    {
      "epoch": 0.19067000626174077,
      "grad_norm": 1.984785556793213,
      "learning_rate": 0.0002825757575757576,
      "loss": 1.0173,
      "step": 1218
    },
    {
      "epoch": 0.19082654978083907,
      "grad_norm": 1.847044587135315,
      "learning_rate": 0.00028255131964809383,
      "loss": 1.0036,
      "step": 1219
    },
    {
      "epoch": 0.19098309329993737,
      "grad_norm": 1.3059883117675781,
      "learning_rate": 0.0002825268817204301,
      "loss": 0.7437,
      "step": 1220
    },
    {
      "epoch": 0.1911396368190357,
      "grad_norm": 1.8383262157440186,
      "learning_rate": 0.00028250244379276633,
      "loss": 0.785,
      "step": 1221
    },
    {
      "epoch": 0.191296180338134,
      "grad_norm": 1.516766905784607,
      "learning_rate": 0.00028247800586510264,
      "loss": 1.0545,
      "step": 1222
    },
    {
      "epoch": 0.1914527238572323,
      "grad_norm": 1.1955336332321167,
      "learning_rate": 0.0002824535679374389,
      "loss": 0.7535,
      "step": 1223
    },
    {
      "epoch": 0.19160926737633063,
      "grad_norm": 1.8770792484283447,
      "learning_rate": 0.00028242913000977514,
      "loss": 0.7899,
      "step": 1224
    },
    {
      "epoch": 0.19176581089542893,
      "grad_norm": 2.0006120204925537,
      "learning_rate": 0.00028240469208211144,
      "loss": 0.903,
      "step": 1225
    },
    {
      "epoch": 0.19192235441452724,
      "grad_norm": 1.9902926683425903,
      "learning_rate": 0.0002823802541544477,
      "loss": 1.5116,
      "step": 1226
    },
    {
      "epoch": 0.19207889793362554,
      "grad_norm": 1.5198619365692139,
      "learning_rate": 0.00028235581622678395,
      "loss": 0.917,
      "step": 1227
    },
    {
      "epoch": 0.19223544145272387,
      "grad_norm": 1.8800374269485474,
      "learning_rate": 0.00028233137829912025,
      "loss": 1.1258,
      "step": 1228
    },
    {
      "epoch": 0.19239198497182217,
      "grad_norm": 1.7788026332855225,
      "learning_rate": 0.00028230694037145645,
      "loss": 1.0894,
      "step": 1229
    },
    {
      "epoch": 0.19254852849092047,
      "grad_norm": 1.4421257972717285,
      "learning_rate": 0.00028228250244379275,
      "loss": 0.812,
      "step": 1230
    },
    {
      "epoch": 0.1927050720100188,
      "grad_norm": 2.770679235458374,
      "learning_rate": 0.000282258064516129,
      "loss": 1.2619,
      "step": 1231
    },
    {
      "epoch": 0.1928616155291171,
      "grad_norm": 1.2351292371749878,
      "learning_rate": 0.00028223362658846526,
      "loss": 0.9904,
      "step": 1232
    },
    {
      "epoch": 0.1930181590482154,
      "grad_norm": 1.4025189876556396,
      "learning_rate": 0.00028220918866080156,
      "loss": 1.2112,
      "step": 1233
    },
    {
      "epoch": 0.1931747025673137,
      "grad_norm": 1.9613655805587769,
      "learning_rate": 0.0002821847507331378,
      "loss": 1.3794,
      "step": 1234
    },
    {
      "epoch": 0.19333124608641203,
      "grad_norm": 1.3618839979171753,
      "learning_rate": 0.00028216031280547406,
      "loss": 1.0788,
      "step": 1235
    },
    {
      "epoch": 0.19348778960551033,
      "grad_norm": 2.786076068878174,
      "learning_rate": 0.0002821358748778103,
      "loss": 1.62,
      "step": 1236
    },
    {
      "epoch": 0.19364433312460863,
      "grad_norm": 2.0850629806518555,
      "learning_rate": 0.0002821114369501466,
      "loss": 1.5648,
      "step": 1237
    },
    {
      "epoch": 0.19380087664370696,
      "grad_norm": 2.4790842533111572,
      "learning_rate": 0.00028208699902248287,
      "loss": 1.2221,
      "step": 1238
    },
    {
      "epoch": 0.19395742016280526,
      "grad_norm": 2.154008388519287,
      "learning_rate": 0.0002820625610948191,
      "loss": 1.0456,
      "step": 1239
    },
    {
      "epoch": 0.19411396368190356,
      "grad_norm": 4.071086883544922,
      "learning_rate": 0.00028203812316715543,
      "loss": 1.4998,
      "step": 1240
    },
    {
      "epoch": 0.1942705072010019,
      "grad_norm": 2.761547803878784,
      "learning_rate": 0.0002820136852394917,
      "loss": 2.0477,
      "step": 1241
    },
    {
      "epoch": 0.1944270507201002,
      "grad_norm": 3.3864057064056396,
      "learning_rate": 0.00028198924731182793,
      "loss": 1.1553,
      "step": 1242
    },
    {
      "epoch": 0.1945835942391985,
      "grad_norm": 2.1232552528381348,
      "learning_rate": 0.00028196480938416424,
      "loss": 1.2943,
      "step": 1243
    },
    {
      "epoch": 0.1947401377582968,
      "grad_norm": 4.488090991973877,
      "learning_rate": 0.00028194037145650043,
      "loss": 1.0515,
      "step": 1244
    },
    {
      "epoch": 0.19489668127739512,
      "grad_norm": 2.625746726989746,
      "learning_rate": 0.00028191593352883674,
      "loss": 0.3637,
      "step": 1245
    },
    {
      "epoch": 0.19505322479649342,
      "grad_norm": 3.204906940460205,
      "learning_rate": 0.000281891495601173,
      "loss": 0.7695,
      "step": 1246
    },
    {
      "epoch": 0.19520976831559173,
      "grad_norm": 2.906982660293579,
      "learning_rate": 0.00028186705767350924,
      "loss": 0.871,
      "step": 1247
    },
    {
      "epoch": 0.19536631183469005,
      "grad_norm": 3.8746962547302246,
      "learning_rate": 0.00028184261974584555,
      "loss": 1.6993,
      "step": 1248
    },
    {
      "epoch": 0.19552285535378836,
      "grad_norm": 5.038626194000244,
      "learning_rate": 0.0002818181818181818,
      "loss": 1.3819,
      "step": 1249
    },
    {
      "epoch": 0.19567939887288666,
      "grad_norm": 2.480462074279785,
      "learning_rate": 0.00028179374389051805,
      "loss": 0.9957,
      "step": 1250
    },
    {
      "epoch": 0.19583594239198499,
      "grad_norm": 0.802400529384613,
      "learning_rate": 0.00028176930596285436,
      "loss": 0.5353,
      "step": 1251
    },
    {
      "epoch": 0.1959924859110833,
      "grad_norm": 0.9732524752616882,
      "learning_rate": 0.0002817448680351906,
      "loss": 0.7606,
      "step": 1252
    },
    {
      "epoch": 0.1961490294301816,
      "grad_norm": 1.2732630968093872,
      "learning_rate": 0.00028172043010752686,
      "loss": 0.7711,
      "step": 1253
    },
    {
      "epoch": 0.1963055729492799,
      "grad_norm": 0.6224183440208435,
      "learning_rate": 0.0002816959921798631,
      "loss": 0.5173,
      "step": 1254
    },
    {
      "epoch": 0.19646211646837822,
      "grad_norm": 0.9690751433372498,
      "learning_rate": 0.0002816715542521994,
      "loss": 0.5248,
      "step": 1255
    },
    {
      "epoch": 0.19661865998747652,
      "grad_norm": 1.045830249786377,
      "learning_rate": 0.00028164711632453567,
      "loss": 0.4887,
      "step": 1256
    },
    {
      "epoch": 0.19677520350657482,
      "grad_norm": 0.9513697624206543,
      "learning_rate": 0.0002816226783968719,
      "loss": 0.5496,
      "step": 1257
    },
    {
      "epoch": 0.19693174702567315,
      "grad_norm": 0.6548385620117188,
      "learning_rate": 0.0002815982404692082,
      "loss": 0.4298,
      "step": 1258
    },
    {
      "epoch": 0.19708829054477145,
      "grad_norm": 0.9877871870994568,
      "learning_rate": 0.0002815738025415445,
      "loss": 0.603,
      "step": 1259
    },
    {
      "epoch": 0.19724483406386975,
      "grad_norm": 0.9719914197921753,
      "learning_rate": 0.0002815493646138807,
      "loss": 0.5035,
      "step": 1260
    },
    {
      "epoch": 0.19740137758296805,
      "grad_norm": 0.5327731966972351,
      "learning_rate": 0.000281524926686217,
      "loss": 0.3911,
      "step": 1261
    },
    {
      "epoch": 0.19755792110206638,
      "grad_norm": 0.874661386013031,
      "learning_rate": 0.00028150048875855323,
      "loss": 0.7128,
      "step": 1262
    },
    {
      "epoch": 0.19771446462116468,
      "grad_norm": 1.2676417827606201,
      "learning_rate": 0.00028147605083088953,
      "loss": 0.6342,
      "step": 1263
    },
    {
      "epoch": 0.19787100814026298,
      "grad_norm": 1.1647354364395142,
      "learning_rate": 0.0002814516129032258,
      "loss": 0.7849,
      "step": 1264
    },
    {
      "epoch": 0.1980275516593613,
      "grad_norm": 2.0960235595703125,
      "learning_rate": 0.00028142717497556204,
      "loss": 0.4386,
      "step": 1265
    },
    {
      "epoch": 0.1981840951784596,
      "grad_norm": 0.8338558673858643,
      "learning_rate": 0.00028140273704789834,
      "loss": 0.5409,
      "step": 1266
    },
    {
      "epoch": 0.1983406386975579,
      "grad_norm": 1.7632123231887817,
      "learning_rate": 0.0002813782991202346,
      "loss": 0.7232,
      "step": 1267
    },
    {
      "epoch": 0.19849718221665624,
      "grad_norm": 2.810450792312622,
      "learning_rate": 0.00028135386119257084,
      "loss": 0.6558,
      "step": 1268
    },
    {
      "epoch": 0.19865372573575454,
      "grad_norm": 1.2503775358200073,
      "learning_rate": 0.0002813294232649071,
      "loss": 0.84,
      "step": 1269
    },
    {
      "epoch": 0.19881026925485284,
      "grad_norm": 1.8836698532104492,
      "learning_rate": 0.0002813049853372434,
      "loss": 0.8859,
      "step": 1270
    },
    {
      "epoch": 0.19896681277395115,
      "grad_norm": 1.3264356851577759,
      "learning_rate": 0.00028128054740957965,
      "loss": 0.7314,
      "step": 1271
    },
    {
      "epoch": 0.19912335629304947,
      "grad_norm": 4.775301933288574,
      "learning_rate": 0.0002812561094819159,
      "loss": 1.4003,
      "step": 1272
    },
    {
      "epoch": 0.19927989981214778,
      "grad_norm": 0.8449872136116028,
      "learning_rate": 0.00028123167155425215,
      "loss": 0.7736,
      "step": 1273
    },
    {
      "epoch": 0.19943644333124608,
      "grad_norm": 1.8365225791931152,
      "learning_rate": 0.00028120723362658846,
      "loss": 1.187,
      "step": 1274
    },
    {
      "epoch": 0.1995929868503444,
      "grad_norm": 1.0617682933807373,
      "learning_rate": 0.0002811827956989247,
      "loss": 0.9421,
      "step": 1275
    },
    {
      "epoch": 0.1997495303694427,
      "grad_norm": 1.617153525352478,
      "learning_rate": 0.00028115835777126096,
      "loss": 1.0415,
      "step": 1276
    },
    {
      "epoch": 0.199906073888541,
      "grad_norm": 1.6880416870117188,
      "learning_rate": 0.0002811339198435972,
      "loss": 0.984,
      "step": 1277
    },
    {
      "epoch": 0.20006261740763934,
      "grad_norm": 1.4345347881317139,
      "learning_rate": 0.0002811094819159335,
      "loss": 1.0823,
      "step": 1278
    },
    {
      "epoch": 0.20021916092673764,
      "grad_norm": 1.5855309963226318,
      "learning_rate": 0.00028108504398826977,
      "loss": 1.1963,
      "step": 1279
    },
    {
      "epoch": 0.20037570444583594,
      "grad_norm": 1.754294991493225,
      "learning_rate": 0.000281060606060606,
      "loss": 1.0804,
      "step": 1280
    },
    {
      "epoch": 0.20053224796493424,
      "grad_norm": 2.036597967147827,
      "learning_rate": 0.0002810361681329423,
      "loss": 0.9677,
      "step": 1281
    },
    {
      "epoch": 0.20068879148403257,
      "grad_norm": 3.59908390045166,
      "learning_rate": 0.0002810117302052786,
      "loss": 1.2978,
      "step": 1282
    },
    {
      "epoch": 0.20084533500313087,
      "grad_norm": 3.524487257003784,
      "learning_rate": 0.00028098729227761483,
      "loss": 1.0612,
      "step": 1283
    },
    {
      "epoch": 0.20100187852222917,
      "grad_norm": 2.481109857559204,
      "learning_rate": 0.0002809628543499511,
      "loss": 1.1174,
      "step": 1284
    },
    {
      "epoch": 0.2011584220413275,
      "grad_norm": 1.8362330198287964,
      "learning_rate": 0.00028093841642228733,
      "loss": 1.2175,
      "step": 1285
    },
    {
      "epoch": 0.2013149655604258,
      "grad_norm": 2.3617823123931885,
      "learning_rate": 0.00028091397849462364,
      "loss": 1.3735,
      "step": 1286
    },
    {
      "epoch": 0.2014715090795241,
      "grad_norm": 2.753002405166626,
      "learning_rate": 0.0002808895405669599,
      "loss": 1.0858,
      "step": 1287
    },
    {
      "epoch": 0.2016280525986224,
      "grad_norm": 2.5607731342315674,
      "learning_rate": 0.00028086510263929614,
      "loss": 0.93,
      "step": 1288
    },
    {
      "epoch": 0.20178459611772073,
      "grad_norm": 2.9476540088653564,
      "learning_rate": 0.00028084066471163244,
      "loss": 1.6911,
      "step": 1289
    },
    {
      "epoch": 0.20194113963681903,
      "grad_norm": 3.2327563762664795,
      "learning_rate": 0.0002808162267839687,
      "loss": 1.402,
      "step": 1290
    },
    {
      "epoch": 0.20209768315591733,
      "grad_norm": 3.618028163909912,
      "learning_rate": 0.00028079178885630495,
      "loss": 1.945,
      "step": 1291
    },
    {
      "epoch": 0.20225422667501566,
      "grad_norm": 2.141831159591675,
      "learning_rate": 0.0002807673509286412,
      "loss": 1.8992,
      "step": 1292
    },
    {
      "epoch": 0.20241077019411396,
      "grad_norm": 2.144073963165283,
      "learning_rate": 0.0002807429130009775,
      "loss": 1.0292,
      "step": 1293
    },
    {
      "epoch": 0.20256731371321227,
      "grad_norm": 2.449118137359619,
      "learning_rate": 0.00028071847507331376,
      "loss": 1.18,
      "step": 1294
    },
    {
      "epoch": 0.2027238572323106,
      "grad_norm": 2.5604381561279297,
      "learning_rate": 0.00028069403714565,
      "loss": 1.7014,
      "step": 1295
    },
    {
      "epoch": 0.2028804007514089,
      "grad_norm": 1.5793492794036865,
      "learning_rate": 0.0002806695992179863,
      "loss": 0.9777,
      "step": 1296
    },
    {
      "epoch": 0.2030369442705072,
      "grad_norm": 2.9324707984924316,
      "learning_rate": 0.00028064516129032256,
      "loss": 1.0947,
      "step": 1297
    },
    {
      "epoch": 0.2031934877896055,
      "grad_norm": 1.8581868410110474,
      "learning_rate": 0.0002806207233626588,
      "loss": 0.8496,
      "step": 1298
    },
    {
      "epoch": 0.20335003130870383,
      "grad_norm": 2.117748975753784,
      "learning_rate": 0.0002805962854349951,
      "loss": 1.2021,
      "step": 1299
    },
    {
      "epoch": 0.20350657482780213,
      "grad_norm": 3.1243643760681152,
      "learning_rate": 0.0002805718475073313,
      "loss": 1.5356,
      "step": 1300
    },
    {
      "epoch": 0.20366311834690043,
      "grad_norm": 1.2260499000549316,
      "learning_rate": 0.0002805474095796676,
      "loss": 0.745,
      "step": 1301
    },
    {
      "epoch": 0.20381966186599876,
      "grad_norm": 0.8840999007225037,
      "learning_rate": 0.0002805229716520039,
      "loss": 0.5955,
      "step": 1302
    },
    {
      "epoch": 0.20397620538509706,
      "grad_norm": 0.9450700879096985,
      "learning_rate": 0.0002804985337243401,
      "loss": 0.8156,
      "step": 1303
    },
    {
      "epoch": 0.20413274890419536,
      "grad_norm": 0.9631698727607727,
      "learning_rate": 0.00028047409579667643,
      "loss": 0.6363,
      "step": 1304
    },
    {
      "epoch": 0.2042892924232937,
      "grad_norm": 0.7808576822280884,
      "learning_rate": 0.0002804496578690127,
      "loss": 0.6415,
      "step": 1305
    },
    {
      "epoch": 0.204445835942392,
      "grad_norm": 0.8394602537155151,
      "learning_rate": 0.00028042521994134893,
      "loss": 0.5683,
      "step": 1306
    },
    {
      "epoch": 0.2046023794614903,
      "grad_norm": 0.9493279457092285,
      "learning_rate": 0.0002804007820136852,
      "loss": 0.6006,
      "step": 1307
    },
    {
      "epoch": 0.2047589229805886,
      "grad_norm": 1.2991026639938354,
      "learning_rate": 0.0002803763440860215,
      "loss": 0.5773,
      "step": 1308
    },
    {
      "epoch": 0.20491546649968692,
      "grad_norm": 0.9779866337776184,
      "learning_rate": 0.00028035190615835774,
      "loss": 0.6043,
      "step": 1309
    },
    {
      "epoch": 0.20507201001878522,
      "grad_norm": 0.8985393047332764,
      "learning_rate": 0.000280327468230694,
      "loss": 0.5852,
      "step": 1310
    },
    {
      "epoch": 0.20522855353788352,
      "grad_norm": 6.3509039878845215,
      "learning_rate": 0.0002803030303030303,
      "loss": 1.5883,
      "step": 1311
    },
    {
      "epoch": 0.20538509705698185,
      "grad_norm": 0.9153957366943359,
      "learning_rate": 0.00028027859237536655,
      "loss": 0.4863,
      "step": 1312
    },
    {
      "epoch": 0.20554164057608015,
      "grad_norm": 1.6397264003753662,
      "learning_rate": 0.0002802541544477028,
      "loss": 0.7647,
      "step": 1313
    },
    {
      "epoch": 0.20569818409517845,
      "grad_norm": 1.2118104696273804,
      "learning_rate": 0.0002802297165200391,
      "loss": 0.5911,
      "step": 1314
    },
    {
      "epoch": 0.20585472761427676,
      "grad_norm": 1.76604425907135,
      "learning_rate": 0.0002802052785923753,
      "loss": 0.9774,
      "step": 1315
    },
    {
      "epoch": 0.20601127113337508,
      "grad_norm": 1.2922953367233276,
      "learning_rate": 0.0002801808406647116,
      "loss": 0.6841,
      "step": 1316
    },
    {
      "epoch": 0.20616781465247339,
      "grad_norm": 1.8913264274597168,
      "learning_rate": 0.00028015640273704786,
      "loss": 0.5421,
      "step": 1317
    },
    {
      "epoch": 0.2063243581715717,
      "grad_norm": 4.230543613433838,
      "learning_rate": 0.0002801319648093841,
      "loss": 0.8927,
      "step": 1318
    },
    {
      "epoch": 0.20648090169067002,
      "grad_norm": 1.4822667837142944,
      "learning_rate": 0.0002801075268817204,
      "loss": 0.8219,
      "step": 1319
    },
    {
      "epoch": 0.20663744520976832,
      "grad_norm": 1.4265705347061157,
      "learning_rate": 0.00028008308895405667,
      "loss": 0.7132,
      "step": 1320
    },
    {
      "epoch": 0.20679398872886662,
      "grad_norm": 1.4394387006759644,
      "learning_rate": 0.0002800586510263929,
      "loss": 0.6372,
      "step": 1321
    },
    {
      "epoch": 0.20695053224796495,
      "grad_norm": 1.241612434387207,
      "learning_rate": 0.0002800342130987292,
      "loss": 0.7006,
      "step": 1322
    },
    {
      "epoch": 0.20710707576706325,
      "grad_norm": 1.5274121761322021,
      "learning_rate": 0.0002800097751710655,
      "loss": 1.122,
      "step": 1323
    },
    {
      "epoch": 0.20726361928616155,
      "grad_norm": 1.7099493741989136,
      "learning_rate": 0.0002799853372434017,
      "loss": 0.7326,
      "step": 1324
    },
    {
      "epoch": 0.20742016280525985,
      "grad_norm": 2.0768227577209473,
      "learning_rate": 0.000279960899315738,
      "loss": 0.7534,
      "step": 1325
    },
    {
      "epoch": 0.20757670632435818,
      "grad_norm": 0.9548947811126709,
      "learning_rate": 0.0002799364613880743,
      "loss": 0.6914,
      "step": 1326
    },
    {
      "epoch": 0.20773324984345648,
      "grad_norm": 2.6987247467041016,
      "learning_rate": 0.00027991202346041053,
      "loss": 1.5067,
      "step": 1327
    },
    {
      "epoch": 0.20788979336255478,
      "grad_norm": 1.8522530794143677,
      "learning_rate": 0.0002798875855327468,
      "loss": 0.8489,
      "step": 1328
    },
    {
      "epoch": 0.2080463368816531,
      "grad_norm": 2.1194634437561035,
      "learning_rate": 0.0002798631476050831,
      "loss": 0.7161,
      "step": 1329
    },
    {
      "epoch": 0.2082028804007514,
      "grad_norm": 1.979345440864563,
      "learning_rate": 0.0002798387096774193,
      "loss": 1.1511,
      "step": 1330
    },
    {
      "epoch": 0.2083594239198497,
      "grad_norm": 2.7786378860473633,
      "learning_rate": 0.0002798142717497556,
      "loss": 1.0264,
      "step": 1331
    },
    {
      "epoch": 0.20851596743894804,
      "grad_norm": 1.8285077810287476,
      "learning_rate": 0.00027978983382209184,
      "loss": 0.9509,
      "step": 1332
    },
    {
      "epoch": 0.20867251095804634,
      "grad_norm": 1.5117707252502441,
      "learning_rate": 0.0002797653958944281,
      "loss": 1.1794,
      "step": 1333
    },
    {
      "epoch": 0.20882905447714464,
      "grad_norm": 1.830258846282959,
      "learning_rate": 0.0002797409579667644,
      "loss": 1.2072,
      "step": 1334
    },
    {
      "epoch": 0.20898559799624294,
      "grad_norm": 2.282845973968506,
      "learning_rate": 0.00027971652003910065,
      "loss": 1.6122,
      "step": 1335
    },
    {
      "epoch": 0.20914214151534127,
      "grad_norm": 2.070908308029175,
      "learning_rate": 0.0002796920821114369,
      "loss": 1.3019,
      "step": 1336
    },
    {
      "epoch": 0.20929868503443957,
      "grad_norm": 1.86088228225708,
      "learning_rate": 0.0002796676441837732,
      "loss": 1.2159,
      "step": 1337
    },
    {
      "epoch": 0.20945522855353788,
      "grad_norm": 1.9150466918945312,
      "learning_rate": 0.00027964320625610946,
      "loss": 1.5197,
      "step": 1338
    },
    {
      "epoch": 0.2096117720726362,
      "grad_norm": 2.406750202178955,
      "learning_rate": 0.0002796187683284457,
      "loss": 1.5689,
      "step": 1339
    },
    {
      "epoch": 0.2097683155917345,
      "grad_norm": 3.363661289215088,
      "learning_rate": 0.00027959433040078196,
      "loss": 1.4533,
      "step": 1340
    },
    {
      "epoch": 0.2099248591108328,
      "grad_norm": 2.635338306427002,
      "learning_rate": 0.00027956989247311827,
      "loss": 1.4408,
      "step": 1341
    },
    {
      "epoch": 0.2100814026299311,
      "grad_norm": 2.8413336277008057,
      "learning_rate": 0.0002795454545454545,
      "loss": 1.3341,
      "step": 1342
    },
    {
      "epoch": 0.21023794614902944,
      "grad_norm": 1.84250009059906,
      "learning_rate": 0.00027952101661779077,
      "loss": 1.2058,
      "step": 1343
    },
    {
      "epoch": 0.21039448966812774,
      "grad_norm": 2.388916492462158,
      "learning_rate": 0.0002794965786901271,
      "loss": 1.3711,
      "step": 1344
    },
    {
      "epoch": 0.21055103318722604,
      "grad_norm": 2.6313962936401367,
      "learning_rate": 0.00027947214076246333,
      "loss": 1.0988,
      "step": 1345
    },
    {
      "epoch": 0.21070757670632437,
      "grad_norm": 1.7488915920257568,
      "learning_rate": 0.0002794477028347996,
      "loss": 1.0163,
      "step": 1346
    },
    {
      "epoch": 0.21086412022542267,
      "grad_norm": 2.03892183303833,
      "learning_rate": 0.0002794232649071359,
      "loss": 0.9676,
      "step": 1347
    },
    {
      "epoch": 0.21102066374452097,
      "grad_norm": 1.5612317323684692,
      "learning_rate": 0.0002793988269794721,
      "loss": 0.8602,
      "step": 1348
    },
    {
      "epoch": 0.2111772072636193,
      "grad_norm": 2.5052318572998047,
      "learning_rate": 0.0002793743890518084,
      "loss": 0.6831,
      "step": 1349
    },
    {
      "epoch": 0.2113337507827176,
      "grad_norm": 2.1519994735717773,
      "learning_rate": 0.00027934995112414464,
      "loss": 1.084,
      "step": 1350
    },
    {
      "epoch": 0.2114902943018159,
      "grad_norm": 0.670945405960083,
      "learning_rate": 0.0002793255131964809,
      "loss": 0.4423,
      "step": 1351
    },
    {
      "epoch": 0.2116468378209142,
      "grad_norm": 1.1006921529769897,
      "learning_rate": 0.0002793010752688172,
      "loss": 0.6916,
      "step": 1352
    },
    {
      "epoch": 0.21180338134001253,
      "grad_norm": 0.8006137609481812,
      "learning_rate": 0.00027927663734115345,
      "loss": 0.4832,
      "step": 1353
    },
    {
      "epoch": 0.21195992485911083,
      "grad_norm": 0.7051506042480469,
      "learning_rate": 0.0002792521994134897,
      "loss": 0.5289,
      "step": 1354
    },
    {
      "epoch": 0.21211646837820913,
      "grad_norm": 0.6033449769020081,
      "learning_rate": 0.00027922776148582595,
      "loss": 0.423,
      "step": 1355
    },
    {
      "epoch": 0.21227301189730746,
      "grad_norm": 1.0505086183547974,
      "learning_rate": 0.00027920332355816225,
      "loss": 0.5403,
      "step": 1356
    },
    {
      "epoch": 0.21242955541640576,
      "grad_norm": 1.063795804977417,
      "learning_rate": 0.0002791788856304985,
      "loss": 0.6797,
      "step": 1357
    },
    {
      "epoch": 0.21258609893550406,
      "grad_norm": 0.77169269323349,
      "learning_rate": 0.00027915444770283476,
      "loss": 0.7163,
      "step": 1358
    },
    {
      "epoch": 0.2127426424546024,
      "grad_norm": 1.3137171268463135,
      "learning_rate": 0.00027913000977517106,
      "loss": 0.6437,
      "step": 1359
    },
    {
      "epoch": 0.2128991859737007,
      "grad_norm": 1.344262719154358,
      "learning_rate": 0.0002791055718475073,
      "loss": 0.5225,
      "step": 1360
    },
    {
      "epoch": 0.213055729492799,
      "grad_norm": 0.8262065649032593,
      "learning_rate": 0.00027908113391984356,
      "loss": 0.6403,
      "step": 1361
    },
    {
      "epoch": 0.2132122730118973,
      "grad_norm": 1.4010947942733765,
      "learning_rate": 0.00027905669599217987,
      "loss": 0.8737,
      "step": 1362
    },
    {
      "epoch": 0.21336881653099563,
      "grad_norm": 1.177219033241272,
      "learning_rate": 0.00027903225806451607,
      "loss": 0.7244,
      "step": 1363
    },
    {
      "epoch": 0.21352536005009393,
      "grad_norm": 1.3933367729187012,
      "learning_rate": 0.00027900782013685237,
      "loss": 0.4727,
      "step": 1364
    },
    {
      "epoch": 0.21368190356919223,
      "grad_norm": 1.8369626998901367,
      "learning_rate": 0.0002789833822091886,
      "loss": 0.6863,
      "step": 1365
    },
    {
      "epoch": 0.21383844708829056,
      "grad_norm": 1.523794174194336,
      "learning_rate": 0.0002789589442815249,
      "loss": 0.667,
      "step": 1366
    },
    {
      "epoch": 0.21399499060738886,
      "grad_norm": 1.0985807180404663,
      "learning_rate": 0.0002789345063538612,
      "loss": 1.0209,
      "step": 1367
    },
    {
      "epoch": 0.21415153412648716,
      "grad_norm": 1.0308488607406616,
      "learning_rate": 0.00027891006842619743,
      "loss": 0.6309,
      "step": 1368
    },
    {
      "epoch": 0.21430807764558546,
      "grad_norm": 2.2205519676208496,
      "learning_rate": 0.0002788856304985337,
      "loss": 1.1807,
      "step": 1369
    },
    {
      "epoch": 0.2144646211646838,
      "grad_norm": 1.8049101829528809,
      "learning_rate": 0.00027886119257087,
      "loss": 0.9443,
      "step": 1370
    },
    {
      "epoch": 0.2146211646837821,
      "grad_norm": 1.7300370931625366,
      "learning_rate": 0.00027883675464320624,
      "loss": 0.7854,
      "step": 1371
    },
    {
      "epoch": 0.2147777082028804,
      "grad_norm": 1.587482213973999,
      "learning_rate": 0.0002788123167155425,
      "loss": 0.5499,
      "step": 1372
    },
    {
      "epoch": 0.21493425172197872,
      "grad_norm": 1.539663553237915,
      "learning_rate": 0.00027878787878787874,
      "loss": 0.9599,
      "step": 1373
    },
    {
      "epoch": 0.21509079524107702,
      "grad_norm": 3.231137990951538,
      "learning_rate": 0.00027876344086021505,
      "loss": 0.9099,
      "step": 1374
    },
    {
      "epoch": 0.21524733876017532,
      "grad_norm": 1.5350894927978516,
      "learning_rate": 0.0002787390029325513,
      "loss": 1.0147,
      "step": 1375
    },
    {
      "epoch": 0.21540388227927365,
      "grad_norm": 2.6407129764556885,
      "learning_rate": 0.00027871456500488755,
      "loss": 0.7973,
      "step": 1376
    },
    {
      "epoch": 0.21556042579837195,
      "grad_norm": 2.9032108783721924,
      "learning_rate": 0.00027869012707722386,
      "loss": 1.1,
      "step": 1377
    },
    {
      "epoch": 0.21571696931747025,
      "grad_norm": 2.3868675231933594,
      "learning_rate": 0.00027866568914956005,
      "loss": 0.625,
      "step": 1378
    },
    {
      "epoch": 0.21587351283656855,
      "grad_norm": 1.203199863433838,
      "learning_rate": 0.00027864125122189636,
      "loss": 0.7701,
      "step": 1379
    },
    {
      "epoch": 0.21603005635566688,
      "grad_norm": 5.6415205001831055,
      "learning_rate": 0.0002786168132942326,
      "loss": 1.1647,
      "step": 1380
    },
    {
      "epoch": 0.21618659987476518,
      "grad_norm": 1.5552973747253418,
      "learning_rate": 0.00027859237536656886,
      "loss": 1.0549,
      "step": 1381
    },
    {
      "epoch": 0.21634314339386349,
      "grad_norm": 4.893740653991699,
      "learning_rate": 0.00027856793743890517,
      "loss": 1.311,
      "step": 1382
    },
    {
      "epoch": 0.21649968691296181,
      "grad_norm": 5.882252216339111,
      "learning_rate": 0.0002785434995112414,
      "loss": 1.5808,
      "step": 1383
    },
    {
      "epoch": 0.21665623043206012,
      "grad_norm": 2.043102502822876,
      "learning_rate": 0.00027851906158357767,
      "loss": 1.5347,
      "step": 1384
    },
    {
      "epoch": 0.21681277395115842,
      "grad_norm": 2.2881131172180176,
      "learning_rate": 0.000278494623655914,
      "loss": 1.37,
      "step": 1385
    },
    {
      "epoch": 0.21696931747025675,
      "grad_norm": 2.5173704624176025,
      "learning_rate": 0.0002784701857282502,
      "loss": 1.5788,
      "step": 1386
    },
    {
      "epoch": 0.21712586098935505,
      "grad_norm": 2.1015539169311523,
      "learning_rate": 0.0002784457478005865,
      "loss": 1.5984,
      "step": 1387
    },
    {
      "epoch": 0.21728240450845335,
      "grad_norm": 2.4034738540649414,
      "learning_rate": 0.00027842130987292273,
      "loss": 1.49,
      "step": 1388
    },
    {
      "epoch": 0.21743894802755165,
      "grad_norm": 2.5903968811035156,
      "learning_rate": 0.00027839687194525903,
      "loss": 1.4753,
      "step": 1389
    },
    {
      "epoch": 0.21759549154664998,
      "grad_norm": 3.2042367458343506,
      "learning_rate": 0.0002783724340175953,
      "loss": 1.6426,
      "step": 1390
    },
    {
      "epoch": 0.21775203506574828,
      "grad_norm": 4.184605598449707,
      "learning_rate": 0.00027834799608993154,
      "loss": 1.9565,
      "step": 1391
    },
    {
      "epoch": 0.21790857858484658,
      "grad_norm": 2.502986431121826,
      "learning_rate": 0.00027832355816226784,
      "loss": 1.7027,
      "step": 1392
    },
    {
      "epoch": 0.2180651221039449,
      "grad_norm": 2.451610565185547,
      "learning_rate": 0.0002782991202346041,
      "loss": 1.2844,
      "step": 1393
    },
    {
      "epoch": 0.2182216656230432,
      "grad_norm": 1.821632742881775,
      "learning_rate": 0.00027827468230694034,
      "loss": 0.9655,
      "step": 1394
    },
    {
      "epoch": 0.2183782091421415,
      "grad_norm": 4.602065563201904,
      "learning_rate": 0.0002782502443792766,
      "loss": 2.3643,
      "step": 1395
    },
    {
      "epoch": 0.2185347526612398,
      "grad_norm": 3.144266366958618,
      "learning_rate": 0.00027822580645161285,
      "loss": 1.0854,
      "step": 1396
    },
    {
      "epoch": 0.21869129618033814,
      "grad_norm": 2.5353446006774902,
      "learning_rate": 0.00027820136852394915,
      "loss": 1.2546,
      "step": 1397
    },
    {
      "epoch": 0.21884783969943644,
      "grad_norm": 2.2268052101135254,
      "learning_rate": 0.0002781769305962854,
      "loss": 1.4616,
      "step": 1398
    },
    {
      "epoch": 0.21900438321853474,
      "grad_norm": 3.656038284301758,
      "learning_rate": 0.00027815249266862165,
      "loss": 1.7655,
      "step": 1399
    },
    {
      "epoch": 0.21916092673763307,
      "grad_norm": 2.7248010635375977,
      "learning_rate": 0.00027812805474095796,
      "loss": 1.8593,
      "step": 1400
    },
    {
      "epoch": 0.21931747025673137,
      "grad_norm": 0.8559685945510864,
      "learning_rate": 0.0002781036168132942,
      "loss": 0.5366,
      "step": 1401
    },
    {
      "epoch": 0.21947401377582967,
      "grad_norm": 0.8044238090515137,
      "learning_rate": 0.00027807917888563046,
      "loss": 0.5512,
      "step": 1402
    },
    {
      "epoch": 0.219630557294928,
      "grad_norm": 0.5511070489883423,
      "learning_rate": 0.0002780547409579667,
      "loss": 0.4996,
      "step": 1403
    },
    {
      "epoch": 0.2197871008140263,
      "grad_norm": 0.9777888655662537,
      "learning_rate": 0.000278030303030303,
      "loss": 0.6064,
      "step": 1404
    },
    {
      "epoch": 0.2199436443331246,
      "grad_norm": 0.7745730876922607,
      "learning_rate": 0.00027800586510263927,
      "loss": 0.4713,
      "step": 1405
    },
    {
      "epoch": 0.2201001878522229,
      "grad_norm": 1.0580544471740723,
      "learning_rate": 0.0002779814271749755,
      "loss": 0.6403,
      "step": 1406
    },
    {
      "epoch": 0.22025673137132123,
      "grad_norm": 0.7956675291061401,
      "learning_rate": 0.0002779569892473118,
      "loss": 0.4613,
      "step": 1407
    },
    {
      "epoch": 0.22041327489041954,
      "grad_norm": 0.678325891494751,
      "learning_rate": 0.0002779325513196481,
      "loss": 0.3812,
      "step": 1408
    },
    {
      "epoch": 0.22056981840951784,
      "grad_norm": 0.7760977149009705,
      "learning_rate": 0.00027790811339198433,
      "loss": 0.3709,
      "step": 1409
    },
    {
      "epoch": 0.22072636192861617,
      "grad_norm": 1.0391250848770142,
      "learning_rate": 0.00027788367546432063,
      "loss": 0.4915,
      "step": 1410
    },
    {
      "epoch": 0.22088290544771447,
      "grad_norm": 0.9946457743644714,
      "learning_rate": 0.00027785923753665683,
      "loss": 0.5706,
      "step": 1411
    },
    {
      "epoch": 0.22103944896681277,
      "grad_norm": 1.5022060871124268,
      "learning_rate": 0.00027783479960899314,
      "loss": 0.6768,
      "step": 1412
    },
    {
      "epoch": 0.2211959924859111,
      "grad_norm": 1.4307317733764648,
      "learning_rate": 0.0002778103616813294,
      "loss": 0.6929,
      "step": 1413
    },
    {
      "epoch": 0.2213525360050094,
      "grad_norm": 0.9843283891677856,
      "learning_rate": 0.00027778592375366564,
      "loss": 0.4171,
      "step": 1414
    },
    {
      "epoch": 0.2215090795241077,
      "grad_norm": 1.1952210664749146,
      "learning_rate": 0.00027776148582600195,
      "loss": 0.5666,
      "step": 1415
    },
    {
      "epoch": 0.221665623043206,
      "grad_norm": 1.7626519203186035,
      "learning_rate": 0.0002777370478983382,
      "loss": 0.8112,
      "step": 1416
    },
    {
      "epoch": 0.22182216656230433,
      "grad_norm": 3.326972246170044,
      "learning_rate": 0.00027771260997067445,
      "loss": 1.1599,
      "step": 1417
    },
    {
      "epoch": 0.22197871008140263,
      "grad_norm": 0.9253565669059753,
      "learning_rate": 0.00027768817204301075,
      "loss": 0.6991,
      "step": 1418
    },
    {
      "epoch": 0.22213525360050093,
      "grad_norm": 1.0200566053390503,
      "learning_rate": 0.000277663734115347,
      "loss": 0.5801,
      "step": 1419
    },
    {
      "epoch": 0.22229179711959926,
      "grad_norm": 1.5075781345367432,
      "learning_rate": 0.00027763929618768326,
      "loss": 0.4883,
      "step": 1420
    },
    {
      "epoch": 0.22244834063869756,
      "grad_norm": 1.3740901947021484,
      "learning_rate": 0.0002776148582600195,
      "loss": 0.7034,
      "step": 1421
    },
    {
      "epoch": 0.22260488415779586,
      "grad_norm": 2.125110149383545,
      "learning_rate": 0.0002775904203323558,
      "loss": 0.7774,
      "step": 1422
    },
    {
      "epoch": 0.22276142767689416,
      "grad_norm": 2.414090633392334,
      "learning_rate": 0.00027756598240469206,
      "loss": 0.7952,
      "step": 1423
    },
    {
      "epoch": 0.2229179711959925,
      "grad_norm": 4.026956081390381,
      "learning_rate": 0.0002775415444770283,
      "loss": 0.9317,
      "step": 1424
    },
    {
      "epoch": 0.2230745147150908,
      "grad_norm": 1.9889763593673706,
      "learning_rate": 0.0002775171065493646,
      "loss": 1.0914,
      "step": 1425
    },
    {
      "epoch": 0.2232310582341891,
      "grad_norm": 1.7664798498153687,
      "learning_rate": 0.0002774926686217008,
      "loss": 0.7606,
      "step": 1426
    },
    {
      "epoch": 0.22338760175328742,
      "grad_norm": 1.880422830581665,
      "learning_rate": 0.0002774682306940371,
      "loss": 1.0761,
      "step": 1427
    },
    {
      "epoch": 0.22354414527238572,
      "grad_norm": 2.686260223388672,
      "learning_rate": 0.0002774437927663734,
      "loss": 0.9571,
      "step": 1428
    },
    {
      "epoch": 0.22370068879148403,
      "grad_norm": 1.939051628112793,
      "learning_rate": 0.0002774193548387096,
      "loss": 1.0255,
      "step": 1429
    },
    {
      "epoch": 0.22385723231058235,
      "grad_norm": 2.358743190765381,
      "learning_rate": 0.00027739491691104593,
      "loss": 1.0628,
      "step": 1430
    },
    {
      "epoch": 0.22401377582968066,
      "grad_norm": 2.4641621112823486,
      "learning_rate": 0.0002773704789833822,
      "loss": 1.2966,
      "step": 1431
    },
    {
      "epoch": 0.22417031934877896,
      "grad_norm": 1.8473012447357178,
      "learning_rate": 0.00027734604105571843,
      "loss": 1.1513,
      "step": 1432
    },
    {
      "epoch": 0.22432686286787726,
      "grad_norm": 1.4949876070022583,
      "learning_rate": 0.00027732160312805474,
      "loss": 1.306,
      "step": 1433
    },
    {
      "epoch": 0.2244834063869756,
      "grad_norm": 1.902312159538269,
      "learning_rate": 0.000277297165200391,
      "loss": 1.6001,
      "step": 1434
    },
    {
      "epoch": 0.2246399499060739,
      "grad_norm": 1.7652263641357422,
      "learning_rate": 0.00027727272727272724,
      "loss": 1.3238,
      "step": 1435
    },
    {
      "epoch": 0.2247964934251722,
      "grad_norm": 2.4492361545562744,
      "learning_rate": 0.0002772482893450635,
      "loss": 1.3513,
      "step": 1436
    },
    {
      "epoch": 0.22495303694427052,
      "grad_norm": 2.270918369293213,
      "learning_rate": 0.0002772238514173998,
      "loss": 1.626,
      "step": 1437
    },
    {
      "epoch": 0.22510958046336882,
      "grad_norm": 2.2354772090911865,
      "learning_rate": 0.00027719941348973605,
      "loss": 1.3422,
      "step": 1438
    },
    {
      "epoch": 0.22526612398246712,
      "grad_norm": 1.9211229085922241,
      "learning_rate": 0.0002771749755620723,
      "loss": 1.5828,
      "step": 1439
    },
    {
      "epoch": 0.22542266750156542,
      "grad_norm": 1.6848493814468384,
      "learning_rate": 0.0002771505376344086,
      "loss": 1.2256,
      "step": 1440
    },
    {
      "epoch": 0.22557921102066375,
      "grad_norm": 2.3783888816833496,
      "learning_rate": 0.00027712609970674486,
      "loss": 1.1445,
      "step": 1441
    },
    {
      "epoch": 0.22573575453976205,
      "grad_norm": 3.050572156906128,
      "learning_rate": 0.0002771016617790811,
      "loss": 1.2094,
      "step": 1442
    },
    {
      "epoch": 0.22589229805886035,
      "grad_norm": 2.4172866344451904,
      "learning_rate": 0.00027707722385141736,
      "loss": 1.415,
      "step": 1443
    },
    {
      "epoch": 0.22604884157795868,
      "grad_norm": 2.319094181060791,
      "learning_rate": 0.0002770527859237536,
      "loss": 1.7239,
      "step": 1444
    },
    {
      "epoch": 0.22620538509705698,
      "grad_norm": 2.561666965484619,
      "learning_rate": 0.0002770283479960899,
      "loss": 1.2432,
      "step": 1445
    },
    {
      "epoch": 0.22636192861615528,
      "grad_norm": 2.667259454727173,
      "learning_rate": 0.00027700391006842617,
      "loss": 0.9331,
      "step": 1446
    },
    {
      "epoch": 0.2265184721352536,
      "grad_norm": 2.7518651485443115,
      "learning_rate": 0.0002769794721407624,
      "loss": 1.0571,
      "step": 1447
    },
    {
      "epoch": 0.2266750156543519,
      "grad_norm": 2.4335968494415283,
      "learning_rate": 0.0002769550342130987,
      "loss": 1.1156,
      "step": 1448
    },
    {
      "epoch": 0.22683155917345021,
      "grad_norm": 5.546756744384766,
      "learning_rate": 0.000276930596285435,
      "loss": 0.6606,
      "step": 1449
    },
    {
      "epoch": 0.22698810269254852,
      "grad_norm": 2.5599935054779053,
      "learning_rate": 0.0002769061583577712,
      "loss": 1.3594,
      "step": 1450
    },
    {
      "epoch": 0.22714464621164684,
      "grad_norm": 0.9069414734840393,
      "learning_rate": 0.0002768817204301075,
      "loss": 0.533,
      "step": 1451
    },
    {
      "epoch": 0.22730118973074515,
      "grad_norm": 0.9586019515991211,
      "learning_rate": 0.0002768572825024438,
      "loss": 0.5733,
      "step": 1452
    },
    {
      "epoch": 0.22745773324984345,
      "grad_norm": 1.2516900300979614,
      "learning_rate": 0.00027683284457478003,
      "loss": 0.6321,
      "step": 1453
    },
    {
      "epoch": 0.22761427676894178,
      "grad_norm": 0.7019465565681458,
      "learning_rate": 0.0002768084066471163,
      "loss": 0.534,
      "step": 1454
    },
    {
      "epoch": 0.22777082028804008,
      "grad_norm": 0.9529553651809692,
      "learning_rate": 0.0002767839687194526,
      "loss": 0.4528,
      "step": 1455
    },
    {
      "epoch": 0.22792736380713838,
      "grad_norm": 2.888197660446167,
      "learning_rate": 0.00027675953079178884,
      "loss": 0.5788,
      "step": 1456
    },
    {
      "epoch": 0.2280839073262367,
      "grad_norm": 0.9398375153541565,
      "learning_rate": 0.0002767350928641251,
      "loss": 0.7626,
      "step": 1457
    },
    {
      "epoch": 0.228240450845335,
      "grad_norm": 1.111086368560791,
      "learning_rate": 0.0002767106549364614,
      "loss": 0.4823,
      "step": 1458
    },
    {
      "epoch": 0.2283969943644333,
      "grad_norm": 0.8638045191764832,
      "learning_rate": 0.0002766862170087976,
      "loss": 0.5664,
      "step": 1459
    },
    {
      "epoch": 0.2285535378835316,
      "grad_norm": 1.1331050395965576,
      "learning_rate": 0.0002766617790811339,
      "loss": 0.5838,
      "step": 1460
    },
    {
      "epoch": 0.22871008140262994,
      "grad_norm": 1.5367311239242554,
      "learning_rate": 0.00027663734115347015,
      "loss": 0.5199,
      "step": 1461
    },
    {
      "epoch": 0.22886662492172824,
      "grad_norm": 1.0833733081817627,
      "learning_rate": 0.0002766129032258064,
      "loss": 0.6322,
      "step": 1462
    },
    {
      "epoch": 0.22902316844082654,
      "grad_norm": 1.2636278867721558,
      "learning_rate": 0.0002765884652981427,
      "loss": 0.8835,
      "step": 1463
    },
    {
      "epoch": 0.22917971195992487,
      "grad_norm": 0.8570249080657959,
      "learning_rate": 0.00027656402737047896,
      "loss": 0.5407,
      "step": 1464
    },
    {
      "epoch": 0.22933625547902317,
      "grad_norm": 1.7874391078948975,
      "learning_rate": 0.0002765395894428152,
      "loss": 0.6971,
      "step": 1465
    },
    {
      "epoch": 0.22949279899812147,
      "grad_norm": 1.138791799545288,
      "learning_rate": 0.00027651515151515146,
      "loss": 0.6228,
      "step": 1466
    },
    {
      "epoch": 0.22964934251721977,
      "grad_norm": 1.4818170070648193,
      "learning_rate": 0.00027649071358748777,
      "loss": 0.6939,
      "step": 1467
    },
    {
      "epoch": 0.2298058860363181,
      "grad_norm": 1.402114987373352,
      "learning_rate": 0.000276466275659824,
      "loss": 0.6931,
      "step": 1468
    },
    {
      "epoch": 0.2299624295554164,
      "grad_norm": 1.5063778162002563,
      "learning_rate": 0.00027644183773216027,
      "loss": 0.7895,
      "step": 1469
    },
    {
      "epoch": 0.2301189730745147,
      "grad_norm": 2.0992255210876465,
      "learning_rate": 0.0002764173998044966,
      "loss": 0.6326,
      "step": 1470
    },
    {
      "epoch": 0.23027551659361303,
      "grad_norm": 1.7399237155914307,
      "learning_rate": 0.00027639296187683283,
      "loss": 0.5391,
      "step": 1471
    },
    {
      "epoch": 0.23043206011271133,
      "grad_norm": 2.003880023956299,
      "learning_rate": 0.0002763685239491691,
      "loss": 0.7821,
      "step": 1472
    },
    {
      "epoch": 0.23058860363180964,
      "grad_norm": 1.794185996055603,
      "learning_rate": 0.0002763440860215054,
      "loss": 0.7965,
      "step": 1473
    },
    {
      "epoch": 0.23074514715090796,
      "grad_norm": 1.893991470336914,
      "learning_rate": 0.0002763196480938416,
      "loss": 0.6783,
      "step": 1474
    },
    {
      "epoch": 0.23090169067000627,
      "grad_norm": 2.7751851081848145,
      "learning_rate": 0.0002762952101661779,
      "loss": 0.9839,
      "step": 1475
    },
    {
      "epoch": 0.23105823418910457,
      "grad_norm": 1.9507943391799927,
      "learning_rate": 0.00027627077223851414,
      "loss": 0.5881,
      "step": 1476
    },
    {
      "epoch": 0.23121477770820287,
      "grad_norm": 2.9333903789520264,
      "learning_rate": 0.0002762463343108504,
      "loss": 0.6818,
      "step": 1477
    },
    {
      "epoch": 0.2313713212273012,
      "grad_norm": 2.3272409439086914,
      "learning_rate": 0.0002762218963831867,
      "loss": 0.9943,
      "step": 1478
    },
    {
      "epoch": 0.2315278647463995,
      "grad_norm": 3.2229695320129395,
      "learning_rate": 0.00027619745845552295,
      "loss": 1.225,
      "step": 1479
    },
    {
      "epoch": 0.2316844082654978,
      "grad_norm": 2.4953155517578125,
      "learning_rate": 0.0002761730205278592,
      "loss": 1.034,
      "step": 1480
    },
    {
      "epoch": 0.23184095178459613,
      "grad_norm": 2.438466787338257,
      "learning_rate": 0.0002761485826001955,
      "loss": 1.2024,
      "step": 1481
    },
    {
      "epoch": 0.23199749530369443,
      "grad_norm": 4.884873390197754,
      "learning_rate": 0.00027612414467253175,
      "loss": 1.0249,
      "step": 1482
    },
    {
      "epoch": 0.23215403882279273,
      "grad_norm": 2.4104747772216797,
      "learning_rate": 0.000276099706744868,
      "loss": 1.2849,
      "step": 1483
    },
    {
      "epoch": 0.23231058234189106,
      "grad_norm": 3.2531588077545166,
      "learning_rate": 0.00027607526881720426,
      "loss": 0.8408,
      "step": 1484
    },
    {
      "epoch": 0.23246712586098936,
      "grad_norm": 2.4069766998291016,
      "learning_rate": 0.00027605083088954056,
      "loss": 1.4731,
      "step": 1485
    },
    {
      "epoch": 0.23262366938008766,
      "grad_norm": 2.5518875122070312,
      "learning_rate": 0.0002760263929618768,
      "loss": 1.1409,
      "step": 1486
    },
    {
      "epoch": 0.23278021289918596,
      "grad_norm": 2.899477958679199,
      "learning_rate": 0.00027600195503421306,
      "loss": 1.2555,
      "step": 1487
    },
    {
      "epoch": 0.2329367564182843,
      "grad_norm": 1.8513123989105225,
      "learning_rate": 0.00027597751710654937,
      "loss": 0.8074,
      "step": 1488
    },
    {
      "epoch": 0.2330932999373826,
      "grad_norm": 2.3798718452453613,
      "learning_rate": 0.00027595307917888557,
      "loss": 1.7678,
      "step": 1489
    },
    {
      "epoch": 0.2332498434564809,
      "grad_norm": 2.9262709617614746,
      "learning_rate": 0.00027592864125122187,
      "loss": 1.5967,
      "step": 1490
    },
    {
      "epoch": 0.23340638697557922,
      "grad_norm": 2.1684608459472656,
      "learning_rate": 0.0002759042033235581,
      "loss": 1.9874,
      "step": 1491
    },
    {
      "epoch": 0.23356293049467752,
      "grad_norm": 2.6140377521514893,
      "learning_rate": 0.0002758797653958944,
      "loss": 0.7234,
      "step": 1492
    },
    {
      "epoch": 0.23371947401377582,
      "grad_norm": 2.1321074962615967,
      "learning_rate": 0.0002758553274682307,
      "loss": 1.0605,
      "step": 1493
    },
    {
      "epoch": 0.23387601753287413,
      "grad_norm": 2.2342660427093506,
      "learning_rate": 0.00027583088954056693,
      "loss": 1.8914,
      "step": 1494
    },
    {
      "epoch": 0.23403256105197245,
      "grad_norm": 3.242147445678711,
      "learning_rate": 0.0002758064516129032,
      "loss": 1.8859,
      "step": 1495
    },
    {
      "epoch": 0.23418910457107076,
      "grad_norm": 4.4763946533203125,
      "learning_rate": 0.0002757820136852395,
      "loss": 1.3996,
      "step": 1496
    },
    {
      "epoch": 0.23434564809016906,
      "grad_norm": 2.6398067474365234,
      "learning_rate": 0.00027575757575757574,
      "loss": 0.9282,
      "step": 1497
    },
    {
      "epoch": 0.23450219160926739,
      "grad_norm": 1.8949030637741089,
      "learning_rate": 0.000275733137829912,
      "loss": 0.9894,
      "step": 1498
    },
    {
      "epoch": 0.2346587351283657,
      "grad_norm": 1.6546046733856201,
      "learning_rate": 0.00027570869990224824,
      "loss": 0.7834,
      "step": 1499
    },
    {
      "epoch": 0.234815278647464,
      "grad_norm": 1.8807889223098755,
      "learning_rate": 0.00027568426197458455,
      "loss": 1.1045,
      "step": 1500
    },
    {
      "epoch": 0.23497182216656232,
      "grad_norm": 0.7086061835289001,
      "learning_rate": 0.0002756598240469208,
      "loss": 0.5635,
      "step": 1501
    },
    {
      "epoch": 0.23512836568566062,
      "grad_norm": 1.064099669456482,
      "learning_rate": 0.00027563538611925705,
      "loss": 0.6135,
      "step": 1502
    },
    {
      "epoch": 0.23528490920475892,
      "grad_norm": 0.6339705586433411,
      "learning_rate": 0.00027561094819159336,
      "loss": 0.4156,
      "step": 1503
    },
    {
      "epoch": 0.23544145272385722,
      "grad_norm": 0.6692638397216797,
      "learning_rate": 0.0002755865102639296,
      "loss": 0.5255,
      "step": 1504
    },
    {
      "epoch": 0.23559799624295555,
      "grad_norm": 0.6374533772468567,
      "learning_rate": 0.00027556207233626586,
      "loss": 0.4684,
      "step": 1505
    },
    {
      "epoch": 0.23575453976205385,
      "grad_norm": 0.621849000453949,
      "learning_rate": 0.00027553763440860216,
      "loss": 0.4455,
      "step": 1506
    },
    {
      "epoch": 0.23591108328115215,
      "grad_norm": 1.2488616704940796,
      "learning_rate": 0.00027551319648093836,
      "loss": 0.5984,
      "step": 1507
    },
    {
      "epoch": 0.23606762680025048,
      "grad_norm": 0.7494067549705505,
      "learning_rate": 0.00027548875855327467,
      "loss": 0.5569,
      "step": 1508
    },
    {
      "epoch": 0.23622417031934878,
      "grad_norm": 5.239253520965576,
      "learning_rate": 0.0002754643206256109,
      "loss": 1.5669,
      "step": 1509
    },
    {
      "epoch": 0.23638071383844708,
      "grad_norm": 1.0418435335159302,
      "learning_rate": 0.00027543988269794717,
      "loss": 0.4972,
      "step": 1510
    },
    {
      "epoch": 0.2365372573575454,
      "grad_norm": 1.3810733556747437,
      "learning_rate": 0.0002754154447702835,
      "loss": 0.7008,
      "step": 1511
    },
    {
      "epoch": 0.2366938008766437,
      "grad_norm": 1.964684247970581,
      "learning_rate": 0.0002753910068426197,
      "loss": 0.598,
      "step": 1512
    },
    {
      "epoch": 0.236850344395742,
      "grad_norm": 1.2020512819290161,
      "learning_rate": 0.000275366568914956,
      "loss": 0.7735,
      "step": 1513
    },
    {
      "epoch": 0.23700688791484031,
      "grad_norm": 1.2619682550430298,
      "learning_rate": 0.00027534213098729223,
      "loss": 0.6185,
      "step": 1514
    },
    {
      "epoch": 0.23716343143393864,
      "grad_norm": 1.121608018875122,
      "learning_rate": 0.00027531769305962853,
      "loss": 0.6125,
      "step": 1515
    },
    {
      "epoch": 0.23731997495303694,
      "grad_norm": 1.6488595008850098,
      "learning_rate": 0.0002752932551319648,
      "loss": 0.5002,
      "step": 1516
    },
    {
      "epoch": 0.23747651847213525,
      "grad_norm": 2.0960159301757812,
      "learning_rate": 0.00027526881720430104,
      "loss": 0.9657,
      "step": 1517
    },
    {
      "epoch": 0.23763306199123357,
      "grad_norm": 1.6600353717803955,
      "learning_rate": 0.00027524437927663734,
      "loss": 0.6478,
      "step": 1518
    },
    {
      "epoch": 0.23778960551033188,
      "grad_norm": 1.770453691482544,
      "learning_rate": 0.0002752199413489736,
      "loss": 0.8878,
      "step": 1519
    },
    {
      "epoch": 0.23794614902943018,
      "grad_norm": 1.5966682434082031,
      "learning_rate": 0.00027519550342130984,
      "loss": 1.122,
      "step": 1520
    },
    {
      "epoch": 0.23810269254852848,
      "grad_norm": 1.4263012409210205,
      "learning_rate": 0.00027517106549364615,
      "loss": 0.6308,
      "step": 1521
    },
    {
      "epoch": 0.2382592360676268,
      "grad_norm": 3.0444369316101074,
      "learning_rate": 0.00027514662756598235,
      "loss": 0.8983,
      "step": 1522
    },
    {
      "epoch": 0.2384157795867251,
      "grad_norm": 1.5377203226089478,
      "learning_rate": 0.00027512218963831865,
      "loss": 0.6754,
      "step": 1523
    },
    {
      "epoch": 0.2385723231058234,
      "grad_norm": 1.37147855758667,
      "learning_rate": 0.0002750977517106549,
      "loss": 0.8112,
      "step": 1524
    },
    {
      "epoch": 0.23872886662492174,
      "grad_norm": 1.6660315990447998,
      "learning_rate": 0.00027507331378299115,
      "loss": 1.0561,
      "step": 1525
    },
    {
      "epoch": 0.23888541014402004,
      "grad_norm": 1.588474988937378,
      "learning_rate": 0.00027504887585532746,
      "loss": 1.1573,
      "step": 1526
    },
    {
      "epoch": 0.23904195366311834,
      "grad_norm": 2.331662178039551,
      "learning_rate": 0.0002750244379276637,
      "loss": 1.0771,
      "step": 1527
    },
    {
      "epoch": 0.23919849718221667,
      "grad_norm": 3.051088571548462,
      "learning_rate": 0.00027499999999999996,
      "loss": 0.8527,
      "step": 1528
    },
    {
      "epoch": 0.23935504070131497,
      "grad_norm": 2.258211374282837,
      "learning_rate": 0.00027497556207233627,
      "loss": 1.2319,
      "step": 1529
    },
    {
      "epoch": 0.23951158422041327,
      "grad_norm": 2.485002279281616,
      "learning_rate": 0.0002749511241446725,
      "loss": 1.1426,
      "step": 1530
    },
    {
      "epoch": 0.23966812773951157,
      "grad_norm": 2.4680216312408447,
      "learning_rate": 0.00027492668621700877,
      "loss": 1.1472,
      "step": 1531
    },
    {
      "epoch": 0.2398246712586099,
      "grad_norm": 1.421920895576477,
      "learning_rate": 0.000274902248289345,
      "loss": 1.0561,
      "step": 1532
    },
    {
      "epoch": 0.2399812147777082,
      "grad_norm": 2.4300615787506104,
      "learning_rate": 0.0002748778103616813,
      "loss": 1.5544,
      "step": 1533
    },
    {
      "epoch": 0.2401377582968065,
      "grad_norm": 3.0856986045837402,
      "learning_rate": 0.0002748533724340176,
      "loss": 0.8778,
      "step": 1534
    },
    {
      "epoch": 0.24029430181590483,
      "grad_norm": 2.003622055053711,
      "learning_rate": 0.00027482893450635383,
      "loss": 1.1826,
      "step": 1535
    },
    {
      "epoch": 0.24045084533500313,
      "grad_norm": 2.1644046306610107,
      "learning_rate": 0.00027480449657869013,
      "loss": 1.3446,
      "step": 1536
    },
    {
      "epoch": 0.24060738885410143,
      "grad_norm": 2.3185083866119385,
      "learning_rate": 0.00027478005865102633,
      "loss": 1.2308,
      "step": 1537
    },
    {
      "epoch": 0.24076393237319976,
      "grad_norm": 2.0810067653656006,
      "learning_rate": 0.00027475562072336264,
      "loss": 0.9243,
      "step": 1538
    },
    {
      "epoch": 0.24092047589229806,
      "grad_norm": 2.293769359588623,
      "learning_rate": 0.0002747311827956989,
      "loss": 1.7686,
      "step": 1539
    },
    {
      "epoch": 0.24107701941139636,
      "grad_norm": 2.4113895893096924,
      "learning_rate": 0.00027470674486803514,
      "loss": 1.3467,
      "step": 1540
    },
    {
      "epoch": 0.24123356293049467,
      "grad_norm": 2.769763708114624,
      "learning_rate": 0.00027468230694037145,
      "loss": 1.8965,
      "step": 1541
    },
    {
      "epoch": 0.241390106449593,
      "grad_norm": 2.0843498706817627,
      "learning_rate": 0.0002746578690127077,
      "loss": 1.2388,
      "step": 1542
    },
    {
      "epoch": 0.2415466499686913,
      "grad_norm": 3.0376205444335938,
      "learning_rate": 0.00027463343108504395,
      "loss": 1.3435,
      "step": 1543
    },
    {
      "epoch": 0.2417031934877896,
      "grad_norm": 4.603314399719238,
      "learning_rate": 0.00027460899315738025,
      "loss": 1.5941,
      "step": 1544
    },
    {
      "epoch": 0.24185973700688793,
      "grad_norm": 3.8465962409973145,
      "learning_rate": 0.0002745845552297165,
      "loss": 1.5791,
      "step": 1545
    },
    {
      "epoch": 0.24201628052598623,
      "grad_norm": 1.3982967138290405,
      "learning_rate": 0.00027456011730205276,
      "loss": 1.0725,
      "step": 1546
    },
    {
      "epoch": 0.24217282404508453,
      "grad_norm": 2.9826741218566895,
      "learning_rate": 0.000274535679374389,
      "loss": 1.2196,
      "step": 1547
    },
    {
      "epoch": 0.24232936756418283,
      "grad_norm": 4.3426055908203125,
      "learning_rate": 0.0002745112414467253,
      "loss": 1.1231,
      "step": 1548
    },
    {
      "epoch": 0.24248591108328116,
      "grad_norm": 2.9004628658294678,
      "learning_rate": 0.00027448680351906156,
      "loss": 1.4186,
      "step": 1549
    },
    {
      "epoch": 0.24264245460237946,
      "grad_norm": 1.7319072484970093,
      "learning_rate": 0.0002744623655913978,
      "loss": 0.8112,
      "step": 1550
    },
    {
      "epoch": 0.24279899812147776,
      "grad_norm": 0.6421094536781311,
      "learning_rate": 0.0002744379276637341,
      "loss": 0.463,
      "step": 1551
    },
    {
      "epoch": 0.2429555416405761,
      "grad_norm": 0.7612909078598022,
      "learning_rate": 0.00027441348973607037,
      "loss": 0.4454,
      "step": 1552
    },
    {
      "epoch": 0.2431120851596744,
      "grad_norm": 0.5956478118896484,
      "learning_rate": 0.0002743890518084066,
      "loss": 0.5241,
      "step": 1553
    },
    {
      "epoch": 0.2432686286787727,
      "grad_norm": 0.6601691246032715,
      "learning_rate": 0.0002743646138807429,
      "loss": 0.481,
      "step": 1554
    },
    {
      "epoch": 0.24342517219787102,
      "grad_norm": 0.757583498954773,
      "learning_rate": 0.0002743401759530791,
      "loss": 0.565,
      "step": 1555
    },
    {
      "epoch": 0.24358171571696932,
      "grad_norm": 1.0948114395141602,
      "learning_rate": 0.00027431573802541543,
      "loss": 0.555,
      "step": 1556
    },
    {
      "epoch": 0.24373825923606762,
      "grad_norm": 0.9005800485610962,
      "learning_rate": 0.0002742913000977517,
      "loss": 0.4812,
      "step": 1557
    },
    {
      "epoch": 0.24389480275516592,
      "grad_norm": 1.3342481851577759,
      "learning_rate": 0.00027426686217008793,
      "loss": 0.5122,
      "step": 1558
    },
    {
      "epoch": 0.24405134627426425,
      "grad_norm": 0.7070523500442505,
      "learning_rate": 0.00027424242424242424,
      "loss": 0.3781,
      "step": 1559
    },
    {
      "epoch": 0.24420788979336255,
      "grad_norm": 0.6875860095024109,
      "learning_rate": 0.0002742179863147605,
      "loss": 0.5231,
      "step": 1560
    },
    {
      "epoch": 0.24436443331246085,
      "grad_norm": 1.033418893814087,
      "learning_rate": 0.00027419354838709674,
      "loss": 0.5012,
      "step": 1561
    },
    {
      "epoch": 0.24452097683155918,
      "grad_norm": 2.32181978225708,
      "learning_rate": 0.000274169110459433,
      "loss": 0.6205,
      "step": 1562
    },
    {
      "epoch": 0.24467752035065748,
      "grad_norm": 1.1538715362548828,
      "learning_rate": 0.0002741446725317693,
      "loss": 0.5589,
      "step": 1563
    },
    {
      "epoch": 0.24483406386975579,
      "grad_norm": 1.1103618144989014,
      "learning_rate": 0.00027412023460410555,
      "loss": 0.5474,
      "step": 1564
    },
    {
      "epoch": 0.24499060738885411,
      "grad_norm": 2.351726531982422,
      "learning_rate": 0.0002740957966764418,
      "loss": 0.5257,
      "step": 1565
    },
    {
      "epoch": 0.24514715090795242,
      "grad_norm": 1.7007243633270264,
      "learning_rate": 0.0002740713587487781,
      "loss": 0.6285,
      "step": 1566
    },
    {
      "epoch": 0.24530369442705072,
      "grad_norm": 1.454067349433899,
      "learning_rate": 0.00027404692082111436,
      "loss": 0.9199,
      "step": 1567
    },
    {
      "epoch": 0.24546023794614902,
      "grad_norm": 0.8819766640663147,
      "learning_rate": 0.0002740224828934506,
      "loss": 0.7621,
      "step": 1568
    },
    {
      "epoch": 0.24561678146524735,
      "grad_norm": 1.5095868110656738,
      "learning_rate": 0.0002739980449657869,
      "loss": 1.0348,
      "step": 1569
    },
    {
      "epoch": 0.24577332498434565,
      "grad_norm": 0.8652186393737793,
      "learning_rate": 0.0002739736070381231,
      "loss": 0.8568,
      "step": 1570
    },
    {
      "epoch": 0.24592986850344395,
      "grad_norm": 1.0354799032211304,
      "learning_rate": 0.0002739491691104594,
      "loss": 0.5486,
      "step": 1571
    },
    {
      "epoch": 0.24608641202254228,
      "grad_norm": 1.751391887664795,
      "learning_rate": 0.00027392473118279567,
      "loss": 0.9729,
      "step": 1572
    },
    {
      "epoch": 0.24624295554164058,
      "grad_norm": 1.665823221206665,
      "learning_rate": 0.0002739002932551319,
      "loss": 1.1064,
      "step": 1573
    },
    {
      "epoch": 0.24639949906073888,
      "grad_norm": 1.3109946250915527,
      "learning_rate": 0.0002738758553274682,
      "loss": 0.7137,
      "step": 1574
    },
    {
      "epoch": 0.24655604257983718,
      "grad_norm": 3.6172003746032715,
      "learning_rate": 0.0002738514173998045,
      "loss": 1.0663,
      "step": 1575
    },
    {
      "epoch": 0.2467125860989355,
      "grad_norm": 1.8106416463851929,
      "learning_rate": 0.0002738269794721407,
      "loss": 0.8352,
      "step": 1576
    },
    {
      "epoch": 0.2468691296180338,
      "grad_norm": 1.7313213348388672,
      "learning_rate": 0.00027380254154447703,
      "loss": 1.143,
      "step": 1577
    },
    {
      "epoch": 0.2470256731371321,
      "grad_norm": 2.3070340156555176,
      "learning_rate": 0.0002737781036168133,
      "loss": 0.8652,
      "step": 1578
    },
    {
      "epoch": 0.24718221665623044,
      "grad_norm": 1.9005930423736572,
      "learning_rate": 0.00027375366568914953,
      "loss": 1.108,
      "step": 1579
    },
    {
      "epoch": 0.24733876017532874,
      "grad_norm": 1.746368408203125,
      "learning_rate": 0.0002737292277614858,
      "loss": 0.9733,
      "step": 1580
    },
    {
      "epoch": 0.24749530369442704,
      "grad_norm": 2.9574499130249023,
      "learning_rate": 0.0002737047898338221,
      "loss": 0.8982,
      "step": 1581
    },
    {
      "epoch": 0.24765184721352537,
      "grad_norm": 2.634615659713745,
      "learning_rate": 0.00027368035190615834,
      "loss": 1.5592,
      "step": 1582
    },
    {
      "epoch": 0.24780839073262367,
      "grad_norm": 2.2195236682891846,
      "learning_rate": 0.0002736559139784946,
      "loss": 1.8192,
      "step": 1583
    },
    {
      "epoch": 0.24796493425172197,
      "grad_norm": 3.19441294670105,
      "learning_rate": 0.0002736314760508309,
      "loss": 1.2147,
      "step": 1584
    },
    {
      "epoch": 0.24812147777082028,
      "grad_norm": 1.6516034603118896,
      "learning_rate": 0.0002736070381231671,
      "loss": 1.0827,
      "step": 1585
    },
    {
      "epoch": 0.2482780212899186,
      "grad_norm": 3.0163512229919434,
      "learning_rate": 0.0002735826001955034,
      "loss": 1.8071,
      "step": 1586
    },
    {
      "epoch": 0.2484345648090169,
      "grad_norm": 3.311854362487793,
      "learning_rate": 0.00027355816226783965,
      "loss": 1.1725,
      "step": 1587
    },
    {
      "epoch": 0.2485911083281152,
      "grad_norm": 2.583314895629883,
      "learning_rate": 0.0002735337243401759,
      "loss": 1.9088,
      "step": 1588
    },
    {
      "epoch": 0.24874765184721354,
      "grad_norm": 3.644285202026367,
      "learning_rate": 0.0002735092864125122,
      "loss": 2.057,
      "step": 1589
    },
    {
      "epoch": 0.24890419536631184,
      "grad_norm": 2.473665475845337,
      "learning_rate": 0.00027348484848484846,
      "loss": 1.0609,
      "step": 1590
    },
    {
      "epoch": 0.24906073888541014,
      "grad_norm": 2.75217342376709,
      "learning_rate": 0.0002734604105571847,
      "loss": 1.9401,
      "step": 1591
    },
    {
      "epoch": 0.24921728240450847,
      "grad_norm": 4.07191801071167,
      "learning_rate": 0.000273435972629521,
      "loss": 1.9969,
      "step": 1592
    },
    {
      "epoch": 0.24937382592360677,
      "grad_norm": 2.538170337677002,
      "learning_rate": 0.00027341153470185727,
      "loss": 1.451,
      "step": 1593
    },
    {
      "epoch": 0.24953036944270507,
      "grad_norm": 1.8366115093231201,
      "learning_rate": 0.0002733870967741935,
      "loss": 1.7211,
      "step": 1594
    },
    {
      "epoch": 0.24968691296180337,
      "grad_norm": 2.4829647541046143,
      "learning_rate": 0.00027336265884652977,
      "loss": 1.3786,
      "step": 1595
    },
    {
      "epoch": 0.2498434564809017,
      "grad_norm": 2.2429263591766357,
      "learning_rate": 0.0002733382209188661,
      "loss": 1.4635,
      "step": 1596
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1900274753570557,
      "learning_rate": 0.00027331378299120233,
      "loss": 0.9664,
      "step": 1597
    },
    {
      "epoch": 0.25015654351909833,
      "grad_norm": 1.736470341682434,
      "learning_rate": 0.0002732893450635386,
      "loss": 1.2012,
      "step": 1598
    },
    {
      "epoch": 0.2503130870381966,
      "grad_norm": 2.750772714614868,
      "learning_rate": 0.0002732649071358749,
      "loss": 0.5847,
      "step": 1599
    },
    {
      "epoch": 0.25046963055729493,
      "grad_norm": 3.358440637588501,
      "learning_rate": 0.00027324046920821114,
      "loss": 1.3006,
      "step": 1600
    },
    {
      "epoch": 0.25062617407639326,
      "grad_norm": 0.7446082830429077,
      "learning_rate": 0.0002732160312805474,
      "loss": 0.474,
      "step": 1601
    },
    {
      "epoch": 0.25078271759549153,
      "grad_norm": 0.5928311347961426,
      "learning_rate": 0.00027319159335288364,
      "loss": 0.4473,
      "step": 1602
    },
    {
      "epoch": 0.25093926111458986,
      "grad_norm": 0.8611072897911072,
      "learning_rate": 0.0002731671554252199,
      "loss": 0.4695,
      "step": 1603
    },
    {
      "epoch": 0.2510958046336882,
      "grad_norm": 0.6917060613632202,
      "learning_rate": 0.0002731427174975562,
      "loss": 0.5228,
      "step": 1604
    },
    {
      "epoch": 0.25125234815278646,
      "grad_norm": 1.070335865020752,
      "learning_rate": 0.00027311827956989245,
      "loss": 1.0444,
      "step": 1605
    },
    {
      "epoch": 0.2514088916718848,
      "grad_norm": 1.0247135162353516,
      "learning_rate": 0.0002730938416422287,
      "loss": 0.7059,
      "step": 1606
    },
    {
      "epoch": 0.25156543519098307,
      "grad_norm": 1.059765100479126,
      "learning_rate": 0.000273069403714565,
      "loss": 0.5032,
      "step": 1607
    },
    {
      "epoch": 0.2517219787100814,
      "grad_norm": 0.8535358905792236,
      "learning_rate": 0.00027304496578690125,
      "loss": 0.4397,
      "step": 1608
    },
    {
      "epoch": 0.2518785222291797,
      "grad_norm": 1.4279924631118774,
      "learning_rate": 0.0002730205278592375,
      "loss": 0.6977,
      "step": 1609
    },
    {
      "epoch": 0.252035065748278,
      "grad_norm": 0.8265482187271118,
      "learning_rate": 0.00027299608993157376,
      "loss": 0.4416,
      "step": 1610
    },
    {
      "epoch": 0.2521916092673763,
      "grad_norm": 1.0426527261734009,
      "learning_rate": 0.00027297165200391006,
      "loss": 0.4451,
      "step": 1611
    },
    {
      "epoch": 0.25234815278647466,
      "grad_norm": 1.2460665702819824,
      "learning_rate": 0.0002729472140762463,
      "loss": 0.7127,
      "step": 1612
    },
    {
      "epoch": 0.25250469630557293,
      "grad_norm": 1.153203010559082,
      "learning_rate": 0.00027292277614858257,
      "loss": 0.5075,
      "step": 1613
    },
    {
      "epoch": 0.25266123982467126,
      "grad_norm": 1.497168779373169,
      "learning_rate": 0.00027289833822091887,
      "loss": 0.6462,
      "step": 1614
    },
    {
      "epoch": 0.2528177833437696,
      "grad_norm": 0.8022387027740479,
      "learning_rate": 0.0002728739002932551,
      "loss": 0.5667,
      "step": 1615
    },
    {
      "epoch": 0.25297432686286786,
      "grad_norm": 1.270932912826538,
      "learning_rate": 0.0002728494623655914,
      "loss": 0.4302,
      "step": 1616
    },
    {
      "epoch": 0.2531308703819662,
      "grad_norm": 2.111393928527832,
      "learning_rate": 0.0002728250244379277,
      "loss": 0.7754,
      "step": 1617
    },
    {
      "epoch": 0.2532874139010645,
      "grad_norm": 0.9829224348068237,
      "learning_rate": 0.0002728005865102639,
      "loss": 0.5654,
      "step": 1618
    },
    {
      "epoch": 0.2534439574201628,
      "grad_norm": 1.0931804180145264,
      "learning_rate": 0.0002727761485826002,
      "loss": 0.4941,
      "step": 1619
    },
    {
      "epoch": 0.2536005009392611,
      "grad_norm": 1.211100459098816,
      "learning_rate": 0.00027275171065493643,
      "loss": 0.8934,
      "step": 1620
    },
    {
      "epoch": 0.25375704445835945,
      "grad_norm": 1.7143733501434326,
      "learning_rate": 0.0002727272727272727,
      "loss": 1.0193,
      "step": 1621
    },
    {
      "epoch": 0.2539135879774577,
      "grad_norm": 1.8827818632125854,
      "learning_rate": 0.000272702834799609,
      "loss": 1.0734,
      "step": 1622
    },
    {
      "epoch": 0.25407013149655605,
      "grad_norm": 1.6230251789093018,
      "learning_rate": 0.00027267839687194524,
      "loss": 1.0196,
      "step": 1623
    },
    {
      "epoch": 0.2542266750156543,
      "grad_norm": 1.9909404516220093,
      "learning_rate": 0.0002726539589442815,
      "loss": 0.8388,
      "step": 1624
    },
    {
      "epoch": 0.25438321853475265,
      "grad_norm": 1.6396865844726562,
      "learning_rate": 0.00027262952101661774,
      "loss": 0.803,
      "step": 1625
    },
    {
      "epoch": 0.254539762053851,
      "grad_norm": 1.7983092069625854,
      "learning_rate": 0.00027260508308895405,
      "loss": 0.8712,
      "step": 1626
    },
    {
      "epoch": 0.25469630557294926,
      "grad_norm": 2.2103474140167236,
      "learning_rate": 0.0002725806451612903,
      "loss": 0.9533,
      "step": 1627
    },
    {
      "epoch": 0.2548528490920476,
      "grad_norm": 1.7631006240844727,
      "learning_rate": 0.00027255620723362655,
      "loss": 0.9976,
      "step": 1628
    },
    {
      "epoch": 0.2550093926111459,
      "grad_norm": 3.220822811126709,
      "learning_rate": 0.00027253176930596286,
      "loss": 1.206,
      "step": 1629
    },
    {
      "epoch": 0.2551659361302442,
      "grad_norm": 1.8891594409942627,
      "learning_rate": 0.0002725073313782991,
      "loss": 1.0133,
      "step": 1630
    },
    {
      "epoch": 0.2553224796493425,
      "grad_norm": 1.985506534576416,
      "learning_rate": 0.00027248289345063536,
      "loss": 0.8946,
      "step": 1631
    },
    {
      "epoch": 0.25547902316844084,
      "grad_norm": 3.2439825534820557,
      "learning_rate": 0.00027245845552297166,
      "loss": 1.3356,
      "step": 1632
    },
    {
      "epoch": 0.2556355666875391,
      "grad_norm": 1.7113157510757446,
      "learning_rate": 0.00027243401759530786,
      "loss": 0.8516,
      "step": 1633
    },
    {
      "epoch": 0.25579211020663745,
      "grad_norm": 2.8625941276550293,
      "learning_rate": 0.00027240957966764417,
      "loss": 1.5037,
      "step": 1634
    },
    {
      "epoch": 0.2559486537257358,
      "grad_norm": 1.7289516925811768,
      "learning_rate": 0.0002723851417399804,
      "loss": 1.167,
      "step": 1635
    },
    {
      "epoch": 0.25610519724483405,
      "grad_norm": 1.9103654623031616,
      "learning_rate": 0.00027236070381231667,
      "loss": 1.2579,
      "step": 1636
    },
    {
      "epoch": 0.2562617407639324,
      "grad_norm": 2.2144134044647217,
      "learning_rate": 0.000272336265884653,
      "loss": 1.6795,
      "step": 1637
    },
    {
      "epoch": 0.2564182842830307,
      "grad_norm": 2.3531837463378906,
      "learning_rate": 0.0002723118279569892,
      "loss": 1.2621,
      "step": 1638
    },
    {
      "epoch": 0.256574827802129,
      "grad_norm": 3.0877432823181152,
      "learning_rate": 0.0002722873900293255,
      "loss": 1.9715,
      "step": 1639
    },
    {
      "epoch": 0.2567313713212273,
      "grad_norm": 1.3705271482467651,
      "learning_rate": 0.0002722629521016618,
      "loss": 1.0049,
      "step": 1640
    },
    {
      "epoch": 0.25688791484032564,
      "grad_norm": 4.535170078277588,
      "learning_rate": 0.00027223851417399803,
      "loss": 1.4665,
      "step": 1641
    },
    {
      "epoch": 0.2570444583594239,
      "grad_norm": 2.5787107944488525,
      "learning_rate": 0.0002722140762463343,
      "loss": 1.2625,
      "step": 1642
    },
    {
      "epoch": 0.25720100187852224,
      "grad_norm": 2.6251347064971924,
      "learning_rate": 0.00027218963831867054,
      "loss": 1.1023,
      "step": 1643
    },
    {
      "epoch": 0.2573575453976205,
      "grad_norm": 1.5550967454910278,
      "learning_rate": 0.00027216520039100684,
      "loss": 0.7943,
      "step": 1644
    },
    {
      "epoch": 0.25751408891671884,
      "grad_norm": 3.5643835067749023,
      "learning_rate": 0.0002721407624633431,
      "loss": 1.731,
      "step": 1645
    },
    {
      "epoch": 0.25767063243581717,
      "grad_norm": 1.838036298751831,
      "learning_rate": 0.00027211632453567934,
      "loss": 1.2029,
      "step": 1646
    },
    {
      "epoch": 0.25782717595491544,
      "grad_norm": 1.7118128538131714,
      "learning_rate": 0.00027209188660801565,
      "loss": 0.5492,
      "step": 1647
    },
    {
      "epoch": 0.2579837194740138,
      "grad_norm": 1.6779016256332397,
      "learning_rate": 0.00027206744868035185,
      "loss": 0.6806,
      "step": 1648
    },
    {
      "epoch": 0.2581402629931121,
      "grad_norm": 1.5262585878372192,
      "learning_rate": 0.00027204301075268815,
      "loss": 0.4437,
      "step": 1649
    },
    {
      "epoch": 0.2582968065122104,
      "grad_norm": 4.473031520843506,
      "learning_rate": 0.0002720185728250244,
      "loss": 1.9419,
      "step": 1650
    },
    {
      "epoch": 0.2584533500313087,
      "grad_norm": 0.5745874643325806,
      "learning_rate": 0.00027199413489736065,
      "loss": 0.4091,
      "step": 1651
    },
    {
      "epoch": 0.25860989355040703,
      "grad_norm": 0.6224820017814636,
      "learning_rate": 0.00027196969696969696,
      "loss": 0.3966,
      "step": 1652
    },
    {
      "epoch": 0.2587664370695053,
      "grad_norm": 1.2538436651229858,
      "learning_rate": 0.0002719452590420332,
      "loss": 0.6685,
      "step": 1653
    },
    {
      "epoch": 0.25892298058860364,
      "grad_norm": 0.558447003364563,
      "learning_rate": 0.00027192082111436946,
      "loss": 0.3777,
      "step": 1654
    },
    {
      "epoch": 0.25907952410770196,
      "grad_norm": 0.6431753635406494,
      "learning_rate": 0.00027189638318670577,
      "loss": 0.3588,
      "step": 1655
    },
    {
      "epoch": 0.25923606762680024,
      "grad_norm": 0.5876404047012329,
      "learning_rate": 0.000271871945259042,
      "loss": 0.3271,
      "step": 1656
    },
    {
      "epoch": 0.25939261114589857,
      "grad_norm": 1.1488398313522339,
      "learning_rate": 0.00027184750733137827,
      "loss": 0.5317,
      "step": 1657
    },
    {
      "epoch": 0.2595491546649969,
      "grad_norm": 1.7112611532211304,
      "learning_rate": 0.0002718230694037145,
      "loss": 0.7552,
      "step": 1658
    },
    {
      "epoch": 0.25970569818409517,
      "grad_norm": 0.7966535687446594,
      "learning_rate": 0.00027179863147605083,
      "loss": 0.5836,
      "step": 1659
    },
    {
      "epoch": 0.2598622417031935,
      "grad_norm": 0.847993016242981,
      "learning_rate": 0.0002717741935483871,
      "loss": 0.3743,
      "step": 1660
    },
    {
      "epoch": 0.26001878522229177,
      "grad_norm": 1.1151633262634277,
      "learning_rate": 0.00027174975562072333,
      "loss": 0.5694,
      "step": 1661
    },
    {
      "epoch": 0.2601753287413901,
      "grad_norm": 1.010614275932312,
      "learning_rate": 0.00027172531769305964,
      "loss": 0.5105,
      "step": 1662
    },
    {
      "epoch": 0.26033187226048843,
      "grad_norm": 1.286232590675354,
      "learning_rate": 0.0002717008797653959,
      "loss": 0.4341,
      "step": 1663
    },
    {
      "epoch": 0.2604884157795867,
      "grad_norm": 1.0436598062515259,
      "learning_rate": 0.00027167644183773214,
      "loss": 0.5435,
      "step": 1664
    },
    {
      "epoch": 0.26064495929868503,
      "grad_norm": 1.534713625907898,
      "learning_rate": 0.00027165200391006844,
      "loss": 0.703,
      "step": 1665
    },
    {
      "epoch": 0.26080150281778336,
      "grad_norm": 1.9755645990371704,
      "learning_rate": 0.00027162756598240464,
      "loss": 0.8479,
      "step": 1666
    },
    {
      "epoch": 0.26095804633688163,
      "grad_norm": 2.3660671710968018,
      "learning_rate": 0.00027160312805474095,
      "loss": 0.9621,
      "step": 1667
    },
    {
      "epoch": 0.26111458985597996,
      "grad_norm": 0.8814460039138794,
      "learning_rate": 0.0002715786901270772,
      "loss": 0.6693,
      "step": 1668
    },
    {
      "epoch": 0.2612711333750783,
      "grad_norm": 1.214966893196106,
      "learning_rate": 0.00027155425219941345,
      "loss": 0.5287,
      "step": 1669
    },
    {
      "epoch": 0.26142767689417656,
      "grad_norm": 1.7603254318237305,
      "learning_rate": 0.00027152981427174975,
      "loss": 0.9414,
      "step": 1670
    },
    {
      "epoch": 0.2615842204132749,
      "grad_norm": 1.4818147420883179,
      "learning_rate": 0.000271505376344086,
      "loss": 0.6992,
      "step": 1671
    },
    {
      "epoch": 0.2617407639323732,
      "grad_norm": 1.955994725227356,
      "learning_rate": 0.00027148093841642226,
      "loss": 0.6655,
      "step": 1672
    },
    {
      "epoch": 0.2618973074514715,
      "grad_norm": 1.3508237600326538,
      "learning_rate": 0.0002714565004887585,
      "loss": 0.9702,
      "step": 1673
    },
    {
      "epoch": 0.2620538509705698,
      "grad_norm": 2.0264861583709717,
      "learning_rate": 0.0002714320625610948,
      "loss": 1.1312,
      "step": 1674
    },
    {
      "epoch": 0.26221039448966815,
      "grad_norm": 1.562551736831665,
      "learning_rate": 0.00027140762463343106,
      "loss": 0.8464,
      "step": 1675
    },
    {
      "epoch": 0.2623669380087664,
      "grad_norm": 1.3331489562988281,
      "learning_rate": 0.0002713831867057673,
      "loss": 0.9695,
      "step": 1676
    },
    {
      "epoch": 0.26252348152786475,
      "grad_norm": 1.8584641218185425,
      "learning_rate": 0.0002713587487781036,
      "loss": 0.8289,
      "step": 1677
    },
    {
      "epoch": 0.26268002504696303,
      "grad_norm": 2.1813502311706543,
      "learning_rate": 0.00027133431085043987,
      "loss": 0.8063,
      "step": 1678
    },
    {
      "epoch": 0.26283656856606136,
      "grad_norm": 1.9859607219696045,
      "learning_rate": 0.0002713098729227761,
      "loss": 0.9312,
      "step": 1679
    },
    {
      "epoch": 0.2629931120851597,
      "grad_norm": 4.345942974090576,
      "learning_rate": 0.00027128543499511243,
      "loss": 1.1479,
      "step": 1680
    },
    {
      "epoch": 0.26314965560425796,
      "grad_norm": 2.3821566104888916,
      "learning_rate": 0.0002712609970674486,
      "loss": 1.2104,
      "step": 1681
    },
    {
      "epoch": 0.2633061991233563,
      "grad_norm": 1.971257209777832,
      "learning_rate": 0.00027123655913978493,
      "loss": 1.3765,
      "step": 1682
    },
    {
      "epoch": 0.2634627426424546,
      "grad_norm": 2.1938669681549072,
      "learning_rate": 0.0002712121212121212,
      "loss": 1.3679,
      "step": 1683
    },
    {
      "epoch": 0.2636192861615529,
      "grad_norm": 2.430959701538086,
      "learning_rate": 0.00027118768328445743,
      "loss": 0.8468,
      "step": 1684
    },
    {
      "epoch": 0.2637758296806512,
      "grad_norm": 3.02817964553833,
      "learning_rate": 0.00027116324535679374,
      "loss": 1.5774,
      "step": 1685
    },
    {
      "epoch": 0.26393237319974955,
      "grad_norm": 2.2461435794830322,
      "learning_rate": 0.00027113880742913,
      "loss": 0.9381,
      "step": 1686
    },
    {
      "epoch": 0.2640889167188478,
      "grad_norm": 1.5934573411941528,
      "learning_rate": 0.00027111436950146624,
      "loss": 0.7952,
      "step": 1687
    },
    {
      "epoch": 0.26424546023794615,
      "grad_norm": 2.398531436920166,
      "learning_rate": 0.00027108993157380255,
      "loss": 1.4084,
      "step": 1688
    },
    {
      "epoch": 0.2644020037570445,
      "grad_norm": 2.056870698928833,
      "learning_rate": 0.0002710654936461388,
      "loss": 1.2535,
      "step": 1689
    },
    {
      "epoch": 0.26455854727614275,
      "grad_norm": 2.2823963165283203,
      "learning_rate": 0.00027104105571847505,
      "loss": 1.2529,
      "step": 1690
    },
    {
      "epoch": 0.2647150907952411,
      "grad_norm": 1.608046293258667,
      "learning_rate": 0.0002710166177908113,
      "loss": 1.3346,
      "step": 1691
    },
    {
      "epoch": 0.2648716343143394,
      "grad_norm": 3.5349557399749756,
      "learning_rate": 0.0002709921798631476,
      "loss": 1.4861,
      "step": 1692
    },
    {
      "epoch": 0.2650281778334377,
      "grad_norm": 2.209035634994507,
      "learning_rate": 0.00027096774193548386,
      "loss": 1.4499,
      "step": 1693
    },
    {
      "epoch": 0.265184721352536,
      "grad_norm": 2.169724464416504,
      "learning_rate": 0.0002709433040078201,
      "loss": 0.9006,
      "step": 1694
    },
    {
      "epoch": 0.26534126487163434,
      "grad_norm": 2.5454139709472656,
      "learning_rate": 0.0002709188660801564,
      "loss": 1.2534,
      "step": 1695
    },
    {
      "epoch": 0.2654978083907326,
      "grad_norm": 1.16903555393219,
      "learning_rate": 0.0002708944281524926,
      "loss": 0.8634,
      "step": 1696
    },
    {
      "epoch": 0.26565435190983094,
      "grad_norm": 2.4895148277282715,
      "learning_rate": 0.0002708699902248289,
      "loss": 1.026,
      "step": 1697
    },
    {
      "epoch": 0.2658108954289292,
      "grad_norm": 1.859477162361145,
      "learning_rate": 0.00027084555229716517,
      "loss": 0.9984,
      "step": 1698
    },
    {
      "epoch": 0.26596743894802755,
      "grad_norm": 3.284623146057129,
      "learning_rate": 0.0002708211143695014,
      "loss": 1.1007,
      "step": 1699
    },
    {
      "epoch": 0.2661239824671259,
      "grad_norm": 3.7506582736968994,
      "learning_rate": 0.0002707966764418377,
      "loss": 1.955,
      "step": 1700
    },
    {
      "epoch": 0.26628052598622415,
      "grad_norm": 0.6009975671768188,
      "learning_rate": 0.000270772238514174,
      "loss": 0.4298,
      "step": 1701
    },
    {
      "epoch": 0.2664370695053225,
      "grad_norm": 1.3551870584487915,
      "learning_rate": 0.00027074780058651023,
      "loss": 0.622,
      "step": 1702
    },
    {
      "epoch": 0.2665936130244208,
      "grad_norm": 1.262035608291626,
      "learning_rate": 0.00027072336265884653,
      "loss": 0.6164,
      "step": 1703
    },
    {
      "epoch": 0.2667501565435191,
      "grad_norm": 0.7926015257835388,
      "learning_rate": 0.0002706989247311828,
      "loss": 0.4694,
      "step": 1704
    },
    {
      "epoch": 0.2669067000626174,
      "grad_norm": 1.2867344617843628,
      "learning_rate": 0.00027067448680351904,
      "loss": 0.66,
      "step": 1705
    },
    {
      "epoch": 0.26706324358171574,
      "grad_norm": 1.134917140007019,
      "learning_rate": 0.0002706500488758553,
      "loss": 0.9503,
      "step": 1706
    },
    {
      "epoch": 0.267219787100814,
      "grad_norm": 3.0994365215301514,
      "learning_rate": 0.0002706256109481916,
      "loss": 0.9773,
      "step": 1707
    },
    {
      "epoch": 0.26737633061991234,
      "grad_norm": 0.9256912469863892,
      "learning_rate": 0.00027060117302052784,
      "loss": 0.4688,
      "step": 1708
    },
    {
      "epoch": 0.26753287413901067,
      "grad_norm": 1.101236343383789,
      "learning_rate": 0.0002705767350928641,
      "loss": 0.6223,
      "step": 1709
    },
    {
      "epoch": 0.26768941765810894,
      "grad_norm": 0.7278368473052979,
      "learning_rate": 0.0002705522971652004,
      "loss": 0.4361,
      "step": 1710
    },
    {
      "epoch": 0.26784596117720727,
      "grad_norm": 2.448662757873535,
      "learning_rate": 0.00027052785923753665,
      "loss": 0.8,
      "step": 1711
    },
    {
      "epoch": 0.2680025046963056,
      "grad_norm": 1.1800535917282104,
      "learning_rate": 0.0002705034213098729,
      "loss": 0.5331,
      "step": 1712
    },
    {
      "epoch": 0.2681590482154039,
      "grad_norm": 2.716262102127075,
      "learning_rate": 0.00027047898338220915,
      "loss": 1.0572,
      "step": 1713
    },
    {
      "epoch": 0.2683155917345022,
      "grad_norm": 0.9216450452804565,
      "learning_rate": 0.0002704545454545454,
      "loss": 0.6161,
      "step": 1714
    },
    {
      "epoch": 0.2684721352536005,
      "grad_norm": 1.1275360584259033,
      "learning_rate": 0.0002704301075268817,
      "loss": 0.6809,
      "step": 1715
    },
    {
      "epoch": 0.2686286787726988,
      "grad_norm": 1.6321802139282227,
      "learning_rate": 0.00027040566959921796,
      "loss": 0.8089,
      "step": 1716
    },
    {
      "epoch": 0.26878522229179713,
      "grad_norm": 1.1545872688293457,
      "learning_rate": 0.0002703812316715542,
      "loss": 0.9202,
      "step": 1717
    },
    {
      "epoch": 0.2689417658108954,
      "grad_norm": 2.506645679473877,
      "learning_rate": 0.0002703567937438905,
      "loss": 0.7897,
      "step": 1718
    },
    {
      "epoch": 0.26909830932999373,
      "grad_norm": 1.1123408079147339,
      "learning_rate": 0.00027033235581622677,
      "loss": 0.5747,
      "step": 1719
    },
    {
      "epoch": 0.26925485284909206,
      "grad_norm": 0.9035351872444153,
      "learning_rate": 0.000270307917888563,
      "loss": 0.5402,
      "step": 1720
    },
    {
      "epoch": 0.26941139636819034,
      "grad_norm": 1.5112823247909546,
      "learning_rate": 0.00027028347996089927,
      "loss": 0.5699,
      "step": 1721
    },
    {
      "epoch": 0.26956793988728867,
      "grad_norm": 1.4688304662704468,
      "learning_rate": 0.0002702590420332356,
      "loss": 0.5282,
      "step": 1722
    },
    {
      "epoch": 0.269724483406387,
      "grad_norm": 1.3641126155853271,
      "learning_rate": 0.00027023460410557183,
      "loss": 0.6053,
      "step": 1723
    },
    {
      "epoch": 0.26988102692548527,
      "grad_norm": 1.88762366771698,
      "learning_rate": 0.0002702101661779081,
      "loss": 0.8346,
      "step": 1724
    },
    {
      "epoch": 0.2700375704445836,
      "grad_norm": 1.9120090007781982,
      "learning_rate": 0.0002701857282502444,
      "loss": 0.9424,
      "step": 1725
    },
    {
      "epoch": 0.2701941139636819,
      "grad_norm": 1.8165515661239624,
      "learning_rate": 0.00027016129032258064,
      "loss": 0.9418,
      "step": 1726
    },
    {
      "epoch": 0.2703506574827802,
      "grad_norm": 2.26983904838562,
      "learning_rate": 0.0002701368523949169,
      "loss": 0.8333,
      "step": 1727
    },
    {
      "epoch": 0.27050720100187853,
      "grad_norm": 1.2658028602600098,
      "learning_rate": 0.0002701124144672532,
      "loss": 0.8788,
      "step": 1728
    },
    {
      "epoch": 0.27066374452097686,
      "grad_norm": 2.336582660675049,
      "learning_rate": 0.0002700879765395894,
      "loss": 0.8146,
      "step": 1729
    },
    {
      "epoch": 0.27082028804007513,
      "grad_norm": 1.6825093030929565,
      "learning_rate": 0.0002700635386119257,
      "loss": 0.8878,
      "step": 1730
    },
    {
      "epoch": 0.27097683155917346,
      "grad_norm": 2.3204588890075684,
      "learning_rate": 0.00027003910068426195,
      "loss": 1.2894,
      "step": 1731
    },
    {
      "epoch": 0.27113337507827173,
      "grad_norm": 2.438342809677124,
      "learning_rate": 0.0002700146627565982,
      "loss": 1.2633,
      "step": 1732
    },
    {
      "epoch": 0.27128991859737006,
      "grad_norm": 2.8188915252685547,
      "learning_rate": 0.0002699902248289345,
      "loss": 1.0154,
      "step": 1733
    },
    {
      "epoch": 0.2714464621164684,
      "grad_norm": 2.1101574897766113,
      "learning_rate": 0.00026996578690127076,
      "loss": 1.3788,
      "step": 1734
    },
    {
      "epoch": 0.27160300563556666,
      "grad_norm": 2.3884902000427246,
      "learning_rate": 0.000269941348973607,
      "loss": 1.0987,
      "step": 1735
    },
    {
      "epoch": 0.271759549154665,
      "grad_norm": 3.389667510986328,
      "learning_rate": 0.0002699169110459433,
      "loss": 1.1042,
      "step": 1736
    },
    {
      "epoch": 0.2719160926737633,
      "grad_norm": 2.0286519527435303,
      "learning_rate": 0.00026989247311827956,
      "loss": 1.6512,
      "step": 1737
    },
    {
      "epoch": 0.2720726361928616,
      "grad_norm": 2.0738837718963623,
      "learning_rate": 0.0002698680351906158,
      "loss": 1.1985,
      "step": 1738
    },
    {
      "epoch": 0.2722291797119599,
      "grad_norm": 2.440089464187622,
      "learning_rate": 0.00026984359726295207,
      "loss": 1.225,
      "step": 1739
    },
    {
      "epoch": 0.27238572323105825,
      "grad_norm": 2.8969979286193848,
      "learning_rate": 0.00026981915933528837,
      "loss": 2.0094,
      "step": 1740
    },
    {
      "epoch": 0.2725422667501565,
      "grad_norm": 2.208665609359741,
      "learning_rate": 0.0002697947214076246,
      "loss": 1.367,
      "step": 1741
    },
    {
      "epoch": 0.27269881026925485,
      "grad_norm": 2.2335593700408936,
      "learning_rate": 0.0002697702834799609,
      "loss": 1.2281,
      "step": 1742
    },
    {
      "epoch": 0.2728553537883532,
      "grad_norm": 1.435701608657837,
      "learning_rate": 0.0002697458455522972,
      "loss": 0.98,
      "step": 1743
    },
    {
      "epoch": 0.27301189730745146,
      "grad_norm": 2.6410810947418213,
      "learning_rate": 0.0002697214076246334,
      "loss": 1.5297,
      "step": 1744
    },
    {
      "epoch": 0.2731684408265498,
      "grad_norm": 3.8984954357147217,
      "learning_rate": 0.0002696969696969697,
      "loss": 1.528,
      "step": 1745
    },
    {
      "epoch": 0.2733249843456481,
      "grad_norm": 1.9594566822052002,
      "learning_rate": 0.00026967253176930593,
      "loss": 1.1372,
      "step": 1746
    },
    {
      "epoch": 0.2734815278647464,
      "grad_norm": 1.6567978858947754,
      "learning_rate": 0.0002696480938416422,
      "loss": 0.8802,
      "step": 1747
    },
    {
      "epoch": 0.2736380713838447,
      "grad_norm": 1.2935997247695923,
      "learning_rate": 0.0002696236559139785,
      "loss": 0.7041,
      "step": 1748
    },
    {
      "epoch": 0.27379461490294305,
      "grad_norm": 3.2540674209594727,
      "learning_rate": 0.00026959921798631474,
      "loss": 1.4213,
      "step": 1749
    },
    {
      "epoch": 0.2739511584220413,
      "grad_norm": 2.091407060623169,
      "learning_rate": 0.000269574780058651,
      "loss": 1.246,
      "step": 1750
    },
    {
      "epoch": 0.27410770194113965,
      "grad_norm": 0.7957701683044434,
      "learning_rate": 0.0002695503421309873,
      "loss": 0.4907,
      "step": 1751
    },
    {
      "epoch": 0.2742642454602379,
      "grad_norm": 0.6549016833305359,
      "learning_rate": 0.00026952590420332355,
      "loss": 0.3429,
      "step": 1752
    },
    {
      "epoch": 0.27442078897933625,
      "grad_norm": 0.8715705275535583,
      "learning_rate": 0.0002695014662756598,
      "loss": 0.4866,
      "step": 1753
    },
    {
      "epoch": 0.2745773324984346,
      "grad_norm": 0.8090335726737976,
      "learning_rate": 0.00026947702834799605,
      "loss": 0.4019,
      "step": 1754
    },
    {
      "epoch": 0.27473387601753285,
      "grad_norm": 0.5464131236076355,
      "learning_rate": 0.00026945259042033236,
      "loss": 0.4006,
      "step": 1755
    },
    {
      "epoch": 0.2748904195366312,
      "grad_norm": 1.1286975145339966,
      "learning_rate": 0.0002694281524926686,
      "loss": 0.6231,
      "step": 1756
    },
    {
      "epoch": 0.2750469630557295,
      "grad_norm": 1.059890627861023,
      "learning_rate": 0.00026940371456500486,
      "loss": 0.4803,
      "step": 1757
    },
    {
      "epoch": 0.2752035065748278,
      "grad_norm": 1.019981861114502,
      "learning_rate": 0.00026937927663734116,
      "loss": 0.4935,
      "step": 1758
    },
    {
      "epoch": 0.2753600500939261,
      "grad_norm": 0.9815790057182312,
      "learning_rate": 0.0002693548387096774,
      "loss": 0.5164,
      "step": 1759
    },
    {
      "epoch": 0.27551659361302444,
      "grad_norm": 1.1376479864120483,
      "learning_rate": 0.00026933040078201367,
      "loss": 0.5096,
      "step": 1760
    },
    {
      "epoch": 0.2756731371321227,
      "grad_norm": 0.9377894997596741,
      "learning_rate": 0.0002693059628543499,
      "loss": 0.4121,
      "step": 1761
    },
    {
      "epoch": 0.27582968065122104,
      "grad_norm": 0.6440132856369019,
      "learning_rate": 0.00026928152492668617,
      "loss": 0.4009,
      "step": 1762
    },
    {
      "epoch": 0.27598622417031937,
      "grad_norm": 1.1063646078109741,
      "learning_rate": 0.0002692570869990225,
      "loss": 0.5908,
      "step": 1763
    },
    {
      "epoch": 0.27614276768941765,
      "grad_norm": 1.2366384267807007,
      "learning_rate": 0.0002692326490713587,
      "loss": 0.6754,
      "step": 1764
    },
    {
      "epoch": 0.276299311208516,
      "grad_norm": 1.511243224143982,
      "learning_rate": 0.000269208211143695,
      "loss": 0.6837,
      "step": 1765
    },
    {
      "epoch": 0.2764558547276143,
      "grad_norm": 2.002196788787842,
      "learning_rate": 0.0002691837732160313,
      "loss": 0.9414,
      "step": 1766
    },
    {
      "epoch": 0.2766123982467126,
      "grad_norm": 0.955176591873169,
      "learning_rate": 0.00026915933528836753,
      "loss": 0.6387,
      "step": 1767
    },
    {
      "epoch": 0.2767689417658109,
      "grad_norm": 1.9328628778457642,
      "learning_rate": 0.0002691348973607038,
      "loss": 0.7687,
      "step": 1768
    },
    {
      "epoch": 0.2769254852849092,
      "grad_norm": 1.1169955730438232,
      "learning_rate": 0.00026911045943304004,
      "loss": 0.5347,
      "step": 1769
    },
    {
      "epoch": 0.2770820288040075,
      "grad_norm": 1.7990258932113647,
      "learning_rate": 0.00026908602150537634,
      "loss": 0.4861,
      "step": 1770
    },
    {
      "epoch": 0.27723857232310584,
      "grad_norm": 1.4510148763656616,
      "learning_rate": 0.0002690615835777126,
      "loss": 0.7257,
      "step": 1771
    },
    {
      "epoch": 0.2773951158422041,
      "grad_norm": 2.4672670364379883,
      "learning_rate": 0.00026903714565004884,
      "loss": 0.7709,
      "step": 1772
    },
    {
      "epoch": 0.27755165936130244,
      "grad_norm": 1.4069247245788574,
      "learning_rate": 0.00026901270772238515,
      "loss": 0.7594,
      "step": 1773
    },
    {
      "epoch": 0.27770820288040077,
      "grad_norm": 1.4989938735961914,
      "learning_rate": 0.0002689882697947214,
      "loss": 0.7431,
      "step": 1774
    },
    {
      "epoch": 0.27786474639949904,
      "grad_norm": 1.9093029499053955,
      "learning_rate": 0.00026896383186705765,
      "loss": 1.0465,
      "step": 1775
    },
    {
      "epoch": 0.27802128991859737,
      "grad_norm": 1.3027005195617676,
      "learning_rate": 0.00026893939393939396,
      "loss": 1.0814,
      "step": 1776
    },
    {
      "epoch": 0.2781778334376957,
      "grad_norm": 2.1441051959991455,
      "learning_rate": 0.00026891495601173016,
      "loss": 0.8204,
      "step": 1777
    },
    {
      "epoch": 0.27833437695679397,
      "grad_norm": 2.118739128112793,
      "learning_rate": 0.00026889051808406646,
      "loss": 1.1269,
      "step": 1778
    },
    {
      "epoch": 0.2784909204758923,
      "grad_norm": 1.8077781200408936,
      "learning_rate": 0.0002688660801564027,
      "loss": 0.8659,
      "step": 1779
    },
    {
      "epoch": 0.27864746399499063,
      "grad_norm": 2.058222770690918,
      "learning_rate": 0.00026884164222873896,
      "loss": 0.9428,
      "step": 1780
    },
    {
      "epoch": 0.2788040075140889,
      "grad_norm": 2.2564380168914795,
      "learning_rate": 0.00026881720430107527,
      "loss": 0.8841,
      "step": 1781
    },
    {
      "epoch": 0.27896055103318723,
      "grad_norm": 3.851144313812256,
      "learning_rate": 0.0002687927663734115,
      "loss": 0.7061,
      "step": 1782
    },
    {
      "epoch": 0.27911709455228556,
      "grad_norm": 1.6981579065322876,
      "learning_rate": 0.00026876832844574777,
      "loss": 1.2807,
      "step": 1783
    },
    {
      "epoch": 0.27927363807138383,
      "grad_norm": 3.6595795154571533,
      "learning_rate": 0.000268743890518084,
      "loss": 1.2373,
      "step": 1784
    },
    {
      "epoch": 0.27943018159048216,
      "grad_norm": 2.8983232975006104,
      "learning_rate": 0.0002687194525904203,
      "loss": 1.029,
      "step": 1785
    },
    {
      "epoch": 0.27958672510958044,
      "grad_norm": 1.8303897380828857,
      "learning_rate": 0.0002686950146627566,
      "loss": 1.1833,
      "step": 1786
    },
    {
      "epoch": 0.27974326862867877,
      "grad_norm": 2.681189775466919,
      "learning_rate": 0.00026867057673509283,
      "loss": 1.5727,
      "step": 1787
    },
    {
      "epoch": 0.2798998121477771,
      "grad_norm": 2.851393699645996,
      "learning_rate": 0.0002686461388074291,
      "loss": 1.4353,
      "step": 1788
    },
    {
      "epoch": 0.28005635566687537,
      "grad_norm": 2.83847713470459,
      "learning_rate": 0.0002686217008797654,
      "loss": 1.3998,
      "step": 1789
    },
    {
      "epoch": 0.2802128991859737,
      "grad_norm": 2.401233673095703,
      "learning_rate": 0.00026859726295210164,
      "loss": 1.2927,
      "step": 1790
    },
    {
      "epoch": 0.280369442705072,
      "grad_norm": 2.81550931930542,
      "learning_rate": 0.0002685728250244379,
      "loss": 1.3452,
      "step": 1791
    },
    {
      "epoch": 0.2805259862241703,
      "grad_norm": 2.0500288009643555,
      "learning_rate": 0.00026854838709677414,
      "loss": 1.2581,
      "step": 1792
    },
    {
      "epoch": 0.2806825297432686,
      "grad_norm": 2.6344480514526367,
      "learning_rate": 0.00026852394916911045,
      "loss": 0.9827,
      "step": 1793
    },
    {
      "epoch": 0.28083907326236696,
      "grad_norm": 1.7978938817977905,
      "learning_rate": 0.0002684995112414467,
      "loss": 0.8994,
      "step": 1794
    },
    {
      "epoch": 0.28099561678146523,
      "grad_norm": 2.9591362476348877,
      "learning_rate": 0.00026847507331378295,
      "loss": 1.9596,
      "step": 1795
    },
    {
      "epoch": 0.28115216030056356,
      "grad_norm": 2.2683701515197754,
      "learning_rate": 0.00026845063538611925,
      "loss": 1.1068,
      "step": 1796
    },
    {
      "epoch": 0.2813087038196619,
      "grad_norm": 1.6625263690948486,
      "learning_rate": 0.0002684261974584555,
      "loss": 0.515,
      "step": 1797
    },
    {
      "epoch": 0.28146524733876016,
      "grad_norm": 1.8417325019836426,
      "learning_rate": 0.00026840175953079176,
      "loss": 0.6351,
      "step": 1798
    },
    {
      "epoch": 0.2816217908578585,
      "grad_norm": 2.7256836891174316,
      "learning_rate": 0.00026837732160312806,
      "loss": 1.6654,
      "step": 1799
    },
    {
      "epoch": 0.2817783343769568,
      "grad_norm": 2.1390318870544434,
      "learning_rate": 0.00026835288367546426,
      "loss": 1.1826,
      "step": 1800
    },
    {
      "epoch": 0.2819348778960551,
      "grad_norm": 0.6134200096130371,
      "learning_rate": 0.00026832844574780056,
      "loss": 0.4343,
      "step": 1801
    },
    {
      "epoch": 0.2820914214151534,
      "grad_norm": 0.9784292578697205,
      "learning_rate": 0.0002683040078201368,
      "loss": 0.3972,
      "step": 1802
    },
    {
      "epoch": 0.28224796493425175,
      "grad_norm": 0.6980729699134827,
      "learning_rate": 0.00026827956989247307,
      "loss": 0.4933,
      "step": 1803
    },
    {
      "epoch": 0.28240450845335,
      "grad_norm": 0.6361396312713623,
      "learning_rate": 0.00026825513196480937,
      "loss": 0.3944,
      "step": 1804
    },
    {
      "epoch": 0.28256105197244835,
      "grad_norm": 0.7284366488456726,
      "learning_rate": 0.0002682306940371456,
      "loss": 0.443,
      "step": 1805
    },
    {
      "epoch": 0.2827175954915466,
      "grad_norm": 0.8181057572364807,
      "learning_rate": 0.0002682062561094819,
      "loss": 0.5435,
      "step": 1806
    },
    {
      "epoch": 0.28287413901064495,
      "grad_norm": 0.7504498362541199,
      "learning_rate": 0.0002681818181818181,
      "loss": 0.4323,
      "step": 1807
    },
    {
      "epoch": 0.2830306825297433,
      "grad_norm": 1.1806275844573975,
      "learning_rate": 0.00026815738025415443,
      "loss": 0.5852,
      "step": 1808
    },
    {
      "epoch": 0.28318722604884156,
      "grad_norm": 1.2418806552886963,
      "learning_rate": 0.0002681329423264907,
      "loss": 0.4522,
      "step": 1809
    },
    {
      "epoch": 0.2833437695679399,
      "grad_norm": 1.2037335634231567,
      "learning_rate": 0.00026810850439882693,
      "loss": 0.6192,
      "step": 1810
    },
    {
      "epoch": 0.2835003130870382,
      "grad_norm": 1.9327640533447266,
      "learning_rate": 0.00026808406647116324,
      "loss": 0.4402,
      "step": 1811
    },
    {
      "epoch": 0.2836568566061365,
      "grad_norm": 1.134433388710022,
      "learning_rate": 0.0002680596285434995,
      "loss": 0.4974,
      "step": 1812
    },
    {
      "epoch": 0.2838134001252348,
      "grad_norm": 0.9470615386962891,
      "learning_rate": 0.00026803519061583574,
      "loss": 0.4516,
      "step": 1813
    },
    {
      "epoch": 0.28396994364433314,
      "grad_norm": 1.848379135131836,
      "learning_rate": 0.00026801075268817205,
      "loss": 0.7034,
      "step": 1814
    },
    {
      "epoch": 0.2841264871634314,
      "grad_norm": 1.2914631366729736,
      "learning_rate": 0.00026798631476050824,
      "loss": 0.6472,
      "step": 1815
    },
    {
      "epoch": 0.28428303068252975,
      "grad_norm": 1.4809837341308594,
      "learning_rate": 0.00026796187683284455,
      "loss": 0.5035,
      "step": 1816
    },
    {
      "epoch": 0.2844395742016281,
      "grad_norm": 1.478471279144287,
      "learning_rate": 0.0002679374389051808,
      "loss": 0.7554,
      "step": 1817
    },
    {
      "epoch": 0.28459611772072635,
      "grad_norm": 1.0676370859146118,
      "learning_rate": 0.00026791300097751705,
      "loss": 0.5768,
      "step": 1818
    },
    {
      "epoch": 0.2847526612398247,
      "grad_norm": 1.5760576725006104,
      "learning_rate": 0.00026788856304985336,
      "loss": 0.6962,
      "step": 1819
    },
    {
      "epoch": 0.284909204758923,
      "grad_norm": 0.8975147604942322,
      "learning_rate": 0.0002678641251221896,
      "loss": 0.4515,
      "step": 1820
    },
    {
      "epoch": 0.2850657482780213,
      "grad_norm": 2.1367173194885254,
      "learning_rate": 0.00026783968719452586,
      "loss": 0.6175,
      "step": 1821
    },
    {
      "epoch": 0.2852222917971196,
      "grad_norm": 2.5454366207122803,
      "learning_rate": 0.00026781524926686217,
      "loss": 0.7304,
      "step": 1822
    },
    {
      "epoch": 0.2853788353162179,
      "grad_norm": 1.3795878887176514,
      "learning_rate": 0.0002677908113391984,
      "loss": 0.7314,
      "step": 1823
    },
    {
      "epoch": 0.2855353788353162,
      "grad_norm": 1.7124171257019043,
      "learning_rate": 0.00026776637341153467,
      "loss": 0.9597,
      "step": 1824
    },
    {
      "epoch": 0.28569192235441454,
      "grad_norm": 2.3599631786346436,
      "learning_rate": 0.0002677419354838709,
      "loss": 1.0985,
      "step": 1825
    },
    {
      "epoch": 0.2858484658735128,
      "grad_norm": 2.4604930877685547,
      "learning_rate": 0.0002677174975562072,
      "loss": 0.7528,
      "step": 1826
    },
    {
      "epoch": 0.28600500939261114,
      "grad_norm": 2.542212724685669,
      "learning_rate": 0.0002676930596285435,
      "loss": 1.0192,
      "step": 1827
    },
    {
      "epoch": 0.28616155291170947,
      "grad_norm": 2.8099851608276367,
      "learning_rate": 0.00026766862170087973,
      "loss": 1.0454,
      "step": 1828
    },
    {
      "epoch": 0.28631809643080774,
      "grad_norm": 1.8754301071166992,
      "learning_rate": 0.00026764418377321603,
      "loss": 1.0623,
      "step": 1829
    },
    {
      "epoch": 0.2864746399499061,
      "grad_norm": 1.9678064584732056,
      "learning_rate": 0.0002676197458455523,
      "loss": 0.9486,
      "step": 1830
    },
    {
      "epoch": 0.2866311834690044,
      "grad_norm": 1.242255687713623,
      "learning_rate": 0.00026759530791788854,
      "loss": 0.6753,
      "step": 1831
    },
    {
      "epoch": 0.2867877269881027,
      "grad_norm": 2.036165475845337,
      "learning_rate": 0.0002675708699902248,
      "loss": 1.168,
      "step": 1832
    },
    {
      "epoch": 0.286944270507201,
      "grad_norm": 2.8120241165161133,
      "learning_rate": 0.00026754643206256104,
      "loss": 0.8798,
      "step": 1833
    },
    {
      "epoch": 0.28710081402629933,
      "grad_norm": 1.734350562095642,
      "learning_rate": 0.00026752199413489734,
      "loss": 1.0636,
      "step": 1834
    },
    {
      "epoch": 0.2872573575453976,
      "grad_norm": 2.014326572418213,
      "learning_rate": 0.0002674975562072336,
      "loss": 1.011,
      "step": 1835
    },
    {
      "epoch": 0.28741390106449594,
      "grad_norm": 2.908611536026001,
      "learning_rate": 0.00026747311827956985,
      "loss": 1.0242,
      "step": 1836
    },
    {
      "epoch": 0.28757044458359426,
      "grad_norm": 2.8601036071777344,
      "learning_rate": 0.00026744868035190615,
      "loss": 1.1949,
      "step": 1837
    },
    {
      "epoch": 0.28772698810269254,
      "grad_norm": 2.5931968688964844,
      "learning_rate": 0.0002674242424242424,
      "loss": 0.8634,
      "step": 1838
    },
    {
      "epoch": 0.28788353162179087,
      "grad_norm": 2.2925524711608887,
      "learning_rate": 0.00026739980449657865,
      "loss": 1.0344,
      "step": 1839
    },
    {
      "epoch": 0.28804007514088914,
      "grad_norm": 1.680653691291809,
      "learning_rate": 0.0002673753665689149,
      "loss": 1.3297,
      "step": 1840
    },
    {
      "epoch": 0.28819661865998747,
      "grad_norm": 1.445891261100769,
      "learning_rate": 0.0002673509286412512,
      "loss": 1.2022,
      "step": 1841
    },
    {
      "epoch": 0.2883531621790858,
      "grad_norm": 3.1469690799713135,
      "learning_rate": 0.00026732649071358746,
      "loss": 1.3316,
      "step": 1842
    },
    {
      "epoch": 0.28850970569818407,
      "grad_norm": 1.903037190437317,
      "learning_rate": 0.0002673020527859237,
      "loss": 1.4019,
      "step": 1843
    },
    {
      "epoch": 0.2886662492172824,
      "grad_norm": 5.236410617828369,
      "learning_rate": 0.00026727761485826,
      "loss": 1.8396,
      "step": 1844
    },
    {
      "epoch": 0.28882279273638073,
      "grad_norm": 4.735574722290039,
      "learning_rate": 0.00026725317693059627,
      "loss": 1.8712,
      "step": 1845
    },
    {
      "epoch": 0.288979336255479,
      "grad_norm": 1.4214756488800049,
      "learning_rate": 0.0002672287390029325,
      "loss": 1.0403,
      "step": 1846
    },
    {
      "epoch": 0.28913587977457733,
      "grad_norm": 2.3396573066711426,
      "learning_rate": 0.0002672043010752688,
      "loss": 1.6399,
      "step": 1847
    },
    {
      "epoch": 0.28929242329367566,
      "grad_norm": 3.088986873626709,
      "learning_rate": 0.000267179863147605,
      "loss": 1.184,
      "step": 1848
    },
    {
      "epoch": 0.28944896681277393,
      "grad_norm": 2.6393306255340576,
      "learning_rate": 0.00026715542521994133,
      "loss": 1.3135,
      "step": 1849
    },
    {
      "epoch": 0.28960551033187226,
      "grad_norm": 1.8152228593826294,
      "learning_rate": 0.0002671309872922776,
      "loss": 1.4623,
      "step": 1850
    },
    {
      "epoch": 0.2897620538509706,
      "grad_norm": 0.7032530903816223,
      "learning_rate": 0.00026710654936461383,
      "loss": 0.5949,
      "step": 1851
    },
    {
      "epoch": 0.28991859737006886,
      "grad_norm": 0.7973015904426575,
      "learning_rate": 0.00026708211143695014,
      "loss": 0.3533,
      "step": 1852
    },
    {
      "epoch": 0.2900751408891672,
      "grad_norm": 1.0158485174179077,
      "learning_rate": 0.0002670576735092864,
      "loss": 0.5212,
      "step": 1853
    },
    {
      "epoch": 0.2902316844082655,
      "grad_norm": 0.6405438184738159,
      "learning_rate": 0.00026703323558162264,
      "loss": 0.5227,
      "step": 1854
    },
    {
      "epoch": 0.2903882279273638,
      "grad_norm": 0.8163948059082031,
      "learning_rate": 0.0002670087976539589,
      "loss": 0.5069,
      "step": 1855
    },
    {
      "epoch": 0.2905447714464621,
      "grad_norm": 2.606863021850586,
      "learning_rate": 0.0002669843597262952,
      "loss": 0.9599,
      "step": 1856
    },
    {
      "epoch": 0.29070131496556045,
      "grad_norm": 0.8287584781646729,
      "learning_rate": 0.00026695992179863145,
      "loss": 0.5773,
      "step": 1857
    },
    {
      "epoch": 0.2908578584846587,
      "grad_norm": 0.5821240544319153,
      "learning_rate": 0.0002669354838709677,
      "loss": 0.2348,
      "step": 1858
    },
    {
      "epoch": 0.29101440200375706,
      "grad_norm": 0.9026615023612976,
      "learning_rate": 0.000266911045943304,
      "loss": 0.3831,
      "step": 1859
    },
    {
      "epoch": 0.29117094552285533,
      "grad_norm": 1.199336290359497,
      "learning_rate": 0.00026688660801564026,
      "loss": 0.4268,
      "step": 1860
    },
    {
      "epoch": 0.29132748904195366,
      "grad_norm": 1.1749603748321533,
      "learning_rate": 0.0002668621700879765,
      "loss": 0.5279,
      "step": 1861
    },
    {
      "epoch": 0.291484032561052,
      "grad_norm": 0.8679521083831787,
      "learning_rate": 0.0002668377321603128,
      "loss": 0.3649,
      "step": 1862
    },
    {
      "epoch": 0.29164057608015026,
      "grad_norm": 1.6801387071609497,
      "learning_rate": 0.000266813294232649,
      "loss": 0.6133,
      "step": 1863
    },
    {
      "epoch": 0.2917971195992486,
      "grad_norm": 1.3388230800628662,
      "learning_rate": 0.0002667888563049853,
      "loss": 0.4792,
      "step": 1864
    },
    {
      "epoch": 0.2919536631183469,
      "grad_norm": 1.3929903507232666,
      "learning_rate": 0.00026676441837732157,
      "loss": 0.6117,
      "step": 1865
    },
    {
      "epoch": 0.2921102066374452,
      "grad_norm": 1.5032627582550049,
      "learning_rate": 0.0002667399804496578,
      "loss": 0.9218,
      "step": 1866
    },
    {
      "epoch": 0.2922667501565435,
      "grad_norm": 1.732055902481079,
      "learning_rate": 0.0002667155425219941,
      "loss": 0.7628,
      "step": 1867
    },
    {
      "epoch": 0.29242329367564185,
      "grad_norm": 1.4332983493804932,
      "learning_rate": 0.0002666911045943304,
      "loss": 0.652,
      "step": 1868
    },
    {
      "epoch": 0.2925798371947401,
      "grad_norm": 2.0694615840911865,
      "learning_rate": 0.0002666666666666666,
      "loss": 0.7529,
      "step": 1869
    },
    {
      "epoch": 0.29273638071383845,
      "grad_norm": 1.9924960136413574,
      "learning_rate": 0.00026664222873900293,
      "loss": 0.7831,
      "step": 1870
    },
    {
      "epoch": 0.2928929242329368,
      "grad_norm": 1.686233401298523,
      "learning_rate": 0.0002666177908113392,
      "loss": 1.0141,
      "step": 1871
    },
    {
      "epoch": 0.29304946775203505,
      "grad_norm": 1.5367885828018188,
      "learning_rate": 0.00026659335288367543,
      "loss": 0.6643,
      "step": 1872
    },
    {
      "epoch": 0.2932060112711334,
      "grad_norm": 1.6426396369934082,
      "learning_rate": 0.0002665689149560117,
      "loss": 0.8695,
      "step": 1873
    },
    {
      "epoch": 0.2933625547902317,
      "grad_norm": 2.6876118183135986,
      "learning_rate": 0.000266544477028348,
      "loss": 0.7657,
      "step": 1874
    },
    {
      "epoch": 0.29351909830933,
      "grad_norm": 2.0909829139709473,
      "learning_rate": 0.00026652003910068424,
      "loss": 1.0113,
      "step": 1875
    },
    {
      "epoch": 0.2936756418284283,
      "grad_norm": 1.8119468688964844,
      "learning_rate": 0.0002664956011730205,
      "loss": 1.0539,
      "step": 1876
    },
    {
      "epoch": 0.2938321853475266,
      "grad_norm": 2.4083352088928223,
      "learning_rate": 0.0002664711632453568,
      "loss": 0.8561,
      "step": 1877
    },
    {
      "epoch": 0.2939887288666249,
      "grad_norm": 1.916272521018982,
      "learning_rate": 0.000266446725317693,
      "loss": 0.6055,
      "step": 1878
    },
    {
      "epoch": 0.29414527238572324,
      "grad_norm": 1.9073026180267334,
      "learning_rate": 0.0002664222873900293,
      "loss": 1.3,
      "step": 1879
    },
    {
      "epoch": 0.2943018159048215,
      "grad_norm": 2.932504415512085,
      "learning_rate": 0.00026639784946236555,
      "loss": 1.0639,
      "step": 1880
    },
    {
      "epoch": 0.29445835942391985,
      "grad_norm": 2.1795926094055176,
      "learning_rate": 0.0002663734115347018,
      "loss": 1.1687,
      "step": 1881
    },
    {
      "epoch": 0.2946149029430182,
      "grad_norm": 2.62552809715271,
      "learning_rate": 0.0002663489736070381,
      "loss": 1.2471,
      "step": 1882
    },
    {
      "epoch": 0.29477144646211645,
      "grad_norm": 2.430758237838745,
      "learning_rate": 0.00026632453567937436,
      "loss": 1.0549,
      "step": 1883
    },
    {
      "epoch": 0.2949279899812148,
      "grad_norm": 1.8063433170318604,
      "learning_rate": 0.0002663000977517106,
      "loss": 1.0626,
      "step": 1884
    },
    {
      "epoch": 0.2950845335003131,
      "grad_norm": 7.484827995300293,
      "learning_rate": 0.0002662756598240469,
      "loss": 1.1665,
      "step": 1885
    },
    {
      "epoch": 0.2952410770194114,
      "grad_norm": 3.3804683685302734,
      "learning_rate": 0.00026625122189638317,
      "loss": 1.9209,
      "step": 1886
    },
    {
      "epoch": 0.2953976205385097,
      "grad_norm": 1.9869343042373657,
      "learning_rate": 0.0002662267839687194,
      "loss": 1.1098,
      "step": 1887
    },
    {
      "epoch": 0.29555416405760804,
      "grad_norm": 3.7342841625213623,
      "learning_rate": 0.00026620234604105567,
      "loss": 1.3265,
      "step": 1888
    },
    {
      "epoch": 0.2957107075767063,
      "grad_norm": 1.7713956832885742,
      "learning_rate": 0.000266177908113392,
      "loss": 1.5005,
      "step": 1889
    },
    {
      "epoch": 0.29586725109580464,
      "grad_norm": 2.1043894290924072,
      "learning_rate": 0.0002661534701857282,
      "loss": 0.8232,
      "step": 1890
    },
    {
      "epoch": 0.29602379461490297,
      "grad_norm": 2.7855064868927,
      "learning_rate": 0.0002661290322580645,
      "loss": 1.1112,
      "step": 1891
    },
    {
      "epoch": 0.29618033813400124,
      "grad_norm": 2.4430272579193115,
      "learning_rate": 0.0002661045943304008,
      "loss": 1.4161,
      "step": 1892
    },
    {
      "epoch": 0.29633688165309957,
      "grad_norm": 2.2559332847595215,
      "learning_rate": 0.00026608015640273703,
      "loss": 1.8431,
      "step": 1893
    },
    {
      "epoch": 0.29649342517219784,
      "grad_norm": 6.513355255126953,
      "learning_rate": 0.0002660557184750733,
      "loss": 1.9822,
      "step": 1894
    },
    {
      "epoch": 0.2966499686912962,
      "grad_norm": 2.599079132080078,
      "learning_rate": 0.0002660312805474096,
      "loss": 1.2078,
      "step": 1895
    },
    {
      "epoch": 0.2968065122103945,
      "grad_norm": NaN,
      "learning_rate": 0.0002660312805474096,
      "loss": 0.0,
      "step": 1896
    },
    {
      "epoch": 0.2969630557294928,
      "grad_norm": 5.364233016967773,
      "learning_rate": 0.0002660068426197458,
      "loss": 1.1353,
      "step": 1897
    },
    {
      "epoch": 0.2971195992485911,
      "grad_norm": 3.103199005126953,
      "learning_rate": 0.0002659824046920821,
      "loss": 0.9987,
      "step": 1898
    },
    {
      "epoch": 0.29727614276768943,
      "grad_norm": 1.9960848093032837,
      "learning_rate": 0.00026595796676441835,
      "loss": 0.7689,
      "step": 1899
    },
    {
      "epoch": 0.2974326862867877,
      "grad_norm": 1.9379894733428955,
      "learning_rate": 0.0002659335288367546,
      "loss": 1.1175,
      "step": 1900
    },
    {
      "epoch": 0.29758922980588604,
      "grad_norm": 0.6607356667518616,
      "learning_rate": 0.0002659090909090909,
      "loss": 0.3971,
      "step": 1901
    },
    {
      "epoch": 0.29774577332498436,
      "grad_norm": 0.8567434549331665,
      "learning_rate": 0.00026588465298142715,
      "loss": 0.4854,
      "step": 1902
    },
    {
      "epoch": 0.29790231684408264,
      "grad_norm": 0.6764739751815796,
      "learning_rate": 0.0002658602150537634,
      "loss": 0.3991,
      "step": 1903
    },
    {
      "epoch": 0.29805886036318097,
      "grad_norm": 0.630438506603241,
      "learning_rate": 0.00026583577712609966,
      "loss": 0.3442,
      "step": 1904
    },
    {
      "epoch": 0.2982154038822793,
      "grad_norm": 0.5661547780036926,
      "learning_rate": 0.00026581133919843596,
      "loss": 0.3289,
      "step": 1905
    },
    {
      "epoch": 0.29837194740137757,
      "grad_norm": 0.7079960107803345,
      "learning_rate": 0.0002657869012707722,
      "loss": 0.4095,
      "step": 1906
    },
    {
      "epoch": 0.2985284909204759,
      "grad_norm": 0.7579357624053955,
      "learning_rate": 0.00026576246334310846,
      "loss": 0.4561,
      "step": 1907
    },
    {
      "epoch": 0.2986850344395742,
      "grad_norm": 0.753103494644165,
      "learning_rate": 0.00026573802541544477,
      "loss": 0.4748,
      "step": 1908
    },
    {
      "epoch": 0.2988415779586725,
      "grad_norm": 0.5085161328315735,
      "learning_rate": 0.000265713587487781,
      "loss": 0.3129,
      "step": 1909
    },
    {
      "epoch": 0.29899812147777083,
      "grad_norm": 1.400939702987671,
      "learning_rate": 0.00026568914956011727,
      "loss": 0.8525,
      "step": 1910
    },
    {
      "epoch": 0.29915466499686916,
      "grad_norm": 1.0124390125274658,
      "learning_rate": 0.0002656647116324536,
      "loss": 0.5729,
      "step": 1911
    },
    {
      "epoch": 0.29931120851596743,
      "grad_norm": 1.1342532634735107,
      "learning_rate": 0.0002656402737047898,
      "loss": 0.6094,
      "step": 1912
    },
    {
      "epoch": 0.29946775203506576,
      "grad_norm": 0.8373773694038391,
      "learning_rate": 0.0002656158357771261,
      "loss": 0.5798,
      "step": 1913
    },
    {
      "epoch": 0.29962429555416403,
      "grad_norm": 1.389040470123291,
      "learning_rate": 0.00026559139784946233,
      "loss": 0.6012,
      "step": 1914
    },
    {
      "epoch": 0.29978083907326236,
      "grad_norm": 0.9821066856384277,
      "learning_rate": 0.0002655669599217986,
      "loss": 0.7038,
      "step": 1915
    },
    {
      "epoch": 0.2999373825923607,
      "grad_norm": 0.9827203750610352,
      "learning_rate": 0.0002655425219941349,
      "loss": 0.5822,
      "step": 1916
    },
    {
      "epoch": 0.30009392611145896,
      "grad_norm": 1.2951587438583374,
      "learning_rate": 0.00026551808406647114,
      "loss": 0.4766,
      "step": 1917
    },
    {
      "epoch": 0.3002504696305573,
      "grad_norm": 1.0430768728256226,
      "learning_rate": 0.0002654936461388074,
      "loss": 0.5216,
      "step": 1918
    },
    {
      "epoch": 0.3004070131496556,
      "grad_norm": 0.9382654428482056,
      "learning_rate": 0.0002654692082111437,
      "loss": 0.4903,
      "step": 1919
    },
    {
      "epoch": 0.3005635566687539,
      "grad_norm": 4.105330944061279,
      "learning_rate": 0.00026544477028347995,
      "loss": 1.2844,
      "step": 1920
    },
    {
      "epoch": 0.3007201001878522,
      "grad_norm": 1.663966178894043,
      "learning_rate": 0.0002654203323558162,
      "loss": 0.9027,
      "step": 1921
    },
    {
      "epoch": 0.30087664370695055,
      "grad_norm": 2.5586514472961426,
      "learning_rate": 0.00026539589442815245,
      "loss": 1.0099,
      "step": 1922
    },
    {
      "epoch": 0.3010331872260488,
      "grad_norm": 1.8075790405273438,
      "learning_rate": 0.00026537145650048875,
      "loss": 0.6638,
      "step": 1923
    },
    {
      "epoch": 0.30118973074514716,
      "grad_norm": 1.4178240299224854,
      "learning_rate": 0.000265347018572825,
      "loss": 0.5234,
      "step": 1924
    },
    {
      "epoch": 0.3013462742642455,
      "grad_norm": 8.215217590332031,
      "learning_rate": 0.00026532258064516126,
      "loss": 1.948,
      "step": 1925
    },
    {
      "epoch": 0.30150281778334376,
      "grad_norm": 2.2477364540100098,
      "learning_rate": 0.00026529814271749756,
      "loss": 0.7896,
      "step": 1926
    },
    {
      "epoch": 0.3016593613024421,
      "grad_norm": 2.111276626586914,
      "learning_rate": 0.00026527370478983376,
      "loss": 0.9689,
      "step": 1927
    },
    {
      "epoch": 0.3018159048215404,
      "grad_norm": 5.124137878417969,
      "learning_rate": 0.00026524926686217006,
      "loss": 1.6356,
      "step": 1928
    },
    {
      "epoch": 0.3019724483406387,
      "grad_norm": 1.8129172325134277,
      "learning_rate": 0.0002652248289345063,
      "loss": 0.7388,
      "step": 1929
    },
    {
      "epoch": 0.302128991859737,
      "grad_norm": 2.3607561588287354,
      "learning_rate": 0.00026520039100684257,
      "loss": 0.8196,
      "step": 1930
    },
    {
      "epoch": 0.3022855353788353,
      "grad_norm": 2.1512415409088135,
      "learning_rate": 0.0002651759530791789,
      "loss": 0.8933,
      "step": 1931
    },
    {
      "epoch": 0.3024420788979336,
      "grad_norm": 1.5390371084213257,
      "learning_rate": 0.0002651515151515151,
      "loss": 0.944,
      "step": 1932
    },
    {
      "epoch": 0.30259862241703195,
      "grad_norm": 1.9705681800842285,
      "learning_rate": 0.0002651270772238514,
      "loss": 1.0435,
      "step": 1933
    },
    {
      "epoch": 0.3027551659361302,
      "grad_norm": 1.8132861852645874,
      "learning_rate": 0.0002651026392961877,
      "loss": 0.7962,
      "step": 1934
    },
    {
      "epoch": 0.30291170945522855,
      "grad_norm": 3.0275063514709473,
      "learning_rate": 0.00026507820136852393,
      "loss": 1.2715,
      "step": 1935
    },
    {
      "epoch": 0.3030682529743269,
      "grad_norm": 1.90733003616333,
      "learning_rate": 0.0002650537634408602,
      "loss": 1.5763,
      "step": 1936
    },
    {
      "epoch": 0.30322479649342515,
      "grad_norm": 2.937795400619507,
      "learning_rate": 0.00026502932551319643,
      "loss": 1.7163,
      "step": 1937
    },
    {
      "epoch": 0.3033813400125235,
      "grad_norm": 3.4657444953918457,
      "learning_rate": 0.00026500488758553274,
      "loss": 1.1086,
      "step": 1938
    },
    {
      "epoch": 0.3035378835316218,
      "grad_norm": 2.1386613845825195,
      "learning_rate": 0.000264980449657869,
      "loss": 1.3036,
      "step": 1939
    },
    {
      "epoch": 0.3036944270507201,
      "grad_norm": 3.0543034076690674,
      "learning_rate": 0.00026495601173020524,
      "loss": 1.1613,
      "step": 1940
    },
    {
      "epoch": 0.3038509705698184,
      "grad_norm": 4.59813117980957,
      "learning_rate": 0.00026493157380254155,
      "loss": 1.7809,
      "step": 1941
    },
    {
      "epoch": 0.30400751408891674,
      "grad_norm": 1.5989995002746582,
      "learning_rate": 0.0002649071358748778,
      "loss": 1.3399,
      "step": 1942
    },
    {
      "epoch": 0.304164057608015,
      "grad_norm": 3.1781742572784424,
      "learning_rate": 0.00026488269794721405,
      "loss": 1.2275,
      "step": 1943
    },
    {
      "epoch": 0.30432060112711334,
      "grad_norm": 1.8627430200576782,
      "learning_rate": 0.0002648582600195503,
      "loss": 1.4865,
      "step": 1944
    },
    {
      "epoch": 0.3044771446462117,
      "grad_norm": 2.872692823410034,
      "learning_rate": 0.00026483382209188655,
      "loss": 1.359,
      "step": 1945
    },
    {
      "epoch": 0.30463368816530995,
      "grad_norm": 2.231991767883301,
      "learning_rate": 0.00026480938416422286,
      "loss": 1.3721,
      "step": 1946
    },
    {
      "epoch": 0.3047902316844083,
      "grad_norm": 4.821183681488037,
      "learning_rate": 0.0002647849462365591,
      "loss": 1.0801,
      "step": 1947
    },
    {
      "epoch": 0.30494677520350655,
      "grad_norm": 3.1386799812316895,
      "learning_rate": 0.00026476050830889536,
      "loss": 0.9191,
      "step": 1948
    },
    {
      "epoch": 0.3051033187226049,
      "grad_norm": 1.6399694681167603,
      "learning_rate": 0.00026473607038123167,
      "loss": 1.0087,
      "step": 1949
    },
    {
      "epoch": 0.3052598622417032,
      "grad_norm": 1.6671782732009888,
      "learning_rate": 0.0002647116324535679,
      "loss": 1.5208,
      "step": 1950
    },
    {
      "epoch": 0.3054164057608015,
      "grad_norm": 0.6199254393577576,
      "learning_rate": 0.00026468719452590417,
      "loss": 0.4585,
      "step": 1951
    },
    {
      "epoch": 0.3055729492798998,
      "grad_norm": 0.783209502696991,
      "learning_rate": 0.0002646627565982404,
      "loss": 0.4495,
      "step": 1952
    },
    {
      "epoch": 0.30572949279899814,
      "grad_norm": 0.6731662154197693,
      "learning_rate": 0.0002646383186705767,
      "loss": 0.452,
      "step": 1953
    },
    {
      "epoch": 0.3058860363180964,
      "grad_norm": 0.6478911638259888,
      "learning_rate": 0.000264613880742913,
      "loss": 0.4503,
      "step": 1954
    },
    {
      "epoch": 0.30604257983719474,
      "grad_norm": 1.0483900308609009,
      "learning_rate": 0.00026458944281524923,
      "loss": 0.4677,
      "step": 1955
    },
    {
      "epoch": 0.30619912335629307,
      "grad_norm": 0.8272550106048584,
      "learning_rate": 0.00026456500488758553,
      "loss": 0.4934,
      "step": 1956
    },
    {
      "epoch": 0.30635566687539134,
      "grad_norm": 0.5366849899291992,
      "learning_rate": 0.0002645405669599218,
      "loss": 0.3189,
      "step": 1957
    },
    {
      "epoch": 0.30651221039448967,
      "grad_norm": 2.845299243927002,
      "learning_rate": 0.00026451612903225804,
      "loss": 0.4301,
      "step": 1958
    },
    {
      "epoch": 0.306668753913588,
      "grad_norm": 1.312286615371704,
      "learning_rate": 0.00026449169110459434,
      "loss": 0.5793,
      "step": 1959
    },
    {
      "epoch": 0.3068252974326863,
      "grad_norm": 1.1284027099609375,
      "learning_rate": 0.00026446725317693054,
      "loss": 0.4355,
      "step": 1960
    },
    {
      "epoch": 0.3069818409517846,
      "grad_norm": 1.250265121459961,
      "learning_rate": 0.00026444281524926684,
      "loss": 0.6103,
      "step": 1961
    },
    {
      "epoch": 0.30713838447088293,
      "grad_norm": 1.059794545173645,
      "learning_rate": 0.0002644183773216031,
      "loss": 0.4448,
      "step": 1962
    },
    {
      "epoch": 0.3072949279899812,
      "grad_norm": 1.0672193765640259,
      "learning_rate": 0.00026439393939393935,
      "loss": 0.4652,
      "step": 1963
    },
    {
      "epoch": 0.30745147150907953,
      "grad_norm": 1.6198948621749878,
      "learning_rate": 0.00026436950146627565,
      "loss": 0.855,
      "step": 1964
    },
    {
      "epoch": 0.3076080150281778,
      "grad_norm": 1.385394811630249,
      "learning_rate": 0.0002643450635386119,
      "loss": 0.6643,
      "step": 1965
    },
    {
      "epoch": 0.30776455854727613,
      "grad_norm": 1.2622168064117432,
      "learning_rate": 0.00026432062561094815,
      "loss": 0.7861,
      "step": 1966
    },
    {
      "epoch": 0.30792110206637446,
      "grad_norm": 1.0766664743423462,
      "learning_rate": 0.0002642961876832844,
      "loss": 0.5213,
      "step": 1967
    },
    {
      "epoch": 0.30807764558547274,
      "grad_norm": 1.1218162775039673,
      "learning_rate": 0.0002642717497556207,
      "loss": 0.6966,
      "step": 1968
    },
    {
      "epoch": 0.30823418910457107,
      "grad_norm": 1.4492287635803223,
      "learning_rate": 0.00026424731182795696,
      "loss": 0.538,
      "step": 1969
    },
    {
      "epoch": 0.3083907326236694,
      "grad_norm": 1.70439875125885,
      "learning_rate": 0.0002642228739002932,
      "loss": 0.8476,
      "step": 1970
    },
    {
      "epoch": 0.30854727614276767,
      "grad_norm": 1.5610077381134033,
      "learning_rate": 0.0002641984359726295,
      "loss": 0.641,
      "step": 1971
    },
    {
      "epoch": 0.308703819661866,
      "grad_norm": 1.6430991888046265,
      "learning_rate": 0.00026417399804496577,
      "loss": 0.9228,
      "step": 1972
    },
    {
      "epoch": 0.3088603631809643,
      "grad_norm": 3.3181350231170654,
      "learning_rate": 0.000264149560117302,
      "loss": 1.2539,
      "step": 1973
    },
    {
      "epoch": 0.3090169067000626,
      "grad_norm": 1.5349977016448975,
      "learning_rate": 0.00026412512218963833,
      "loss": 0.8461,
      "step": 1974
    },
    {
      "epoch": 0.30917345021916093,
      "grad_norm": 1.3889473676681519,
      "learning_rate": 0.0002641006842619745,
      "loss": 0.8088,
      "step": 1975
    },
    {
      "epoch": 0.30932999373825926,
      "grad_norm": 3.5759482383728027,
      "learning_rate": 0.00026407624633431083,
      "loss": 1.0715,
      "step": 1976
    },
    {
      "epoch": 0.30948653725735753,
      "grad_norm": 5.998546123504639,
      "learning_rate": 0.0002640518084066471,
      "loss": 1.1101,
      "step": 1977
    },
    {
      "epoch": 0.30964308077645586,
      "grad_norm": 1.6314876079559326,
      "learning_rate": 0.00026402737047898333,
      "loss": 0.9514,
      "step": 1978
    },
    {
      "epoch": 0.3097996242955542,
      "grad_norm": 1.3417260646820068,
      "learning_rate": 0.00026400293255131964,
      "loss": 0.9667,
      "step": 1979
    },
    {
      "epoch": 0.30995616781465246,
      "grad_norm": 1.557165265083313,
      "learning_rate": 0.0002639784946236559,
      "loss": 0.9157,
      "step": 1980
    },
    {
      "epoch": 0.3101127113337508,
      "grad_norm": 1.574750542640686,
      "learning_rate": 0.00026395405669599214,
      "loss": 0.7029,
      "step": 1981
    },
    {
      "epoch": 0.3102692548528491,
      "grad_norm": 2.596886157989502,
      "learning_rate": 0.00026392961876832845,
      "loss": 1.3352,
      "step": 1982
    },
    {
      "epoch": 0.3104257983719474,
      "grad_norm": 1.6578302383422852,
      "learning_rate": 0.0002639051808406647,
      "loss": 0.6754,
      "step": 1983
    },
    {
      "epoch": 0.3105823418910457,
      "grad_norm": 1.7295254468917847,
      "learning_rate": 0.00026388074291300095,
      "loss": 1.0541,
      "step": 1984
    },
    {
      "epoch": 0.310738885410144,
      "grad_norm": 1.5874801874160767,
      "learning_rate": 0.0002638563049853372,
      "loss": 1.1757,
      "step": 1985
    },
    {
      "epoch": 0.3108954289292423,
      "grad_norm": 2.339158535003662,
      "learning_rate": 0.0002638318670576735,
      "loss": 1.4245,
      "step": 1986
    },
    {
      "epoch": 0.31105197244834065,
      "grad_norm": 1.606540560722351,
      "learning_rate": 0.00026380742913000976,
      "loss": 1.1071,
      "step": 1987
    },
    {
      "epoch": 0.3112085159674389,
      "grad_norm": 3.946507215499878,
      "learning_rate": 0.000263782991202346,
      "loss": 1.4799,
      "step": 1988
    },
    {
      "epoch": 0.31136505948653725,
      "grad_norm": 3.6631851196289062,
      "learning_rate": 0.0002637585532746823,
      "loss": 1.7003,
      "step": 1989
    },
    {
      "epoch": 0.3115216030056356,
      "grad_norm": 2.883183240890503,
      "learning_rate": 0.00026373411534701856,
      "loss": 1.3078,
      "step": 1990
    },
    {
      "epoch": 0.31167814652473386,
      "grad_norm": 1.4559465646743774,
      "learning_rate": 0.0002637096774193548,
      "loss": 1.4201,
      "step": 1991
    },
    {
      "epoch": 0.3118346900438322,
      "grad_norm": 1.9634108543395996,
      "learning_rate": 0.00026368523949169107,
      "loss": 1.3366,
      "step": 1992
    },
    {
      "epoch": 0.3119912335629305,
      "grad_norm": 2.3829219341278076,
      "learning_rate": 0.0002636608015640273,
      "loss": 1.6637,
      "step": 1993
    },
    {
      "epoch": 0.3121477770820288,
      "grad_norm": 2.1768481731414795,
      "learning_rate": 0.0002636363636363636,
      "loss": 1.1262,
      "step": 1994
    },
    {
      "epoch": 0.3123043206011271,
      "grad_norm": 2.132509231567383,
      "learning_rate": 0.0002636119257086999,
      "loss": 1.3264,
      "step": 1995
    },
    {
      "epoch": 0.31246086412022545,
      "grad_norm": 3.3182246685028076,
      "learning_rate": 0.0002635874877810361,
      "loss": 0.7347,
      "step": 1996
    },
    {
      "epoch": 0.3126174076393237,
      "grad_norm": 1.8568942546844482,
      "learning_rate": 0.00026356304985337243,
      "loss": 0.9086,
      "step": 1997
    },
    {
      "epoch": 0.31277395115842205,
      "grad_norm": 2.9322288036346436,
      "learning_rate": 0.0002635386119257087,
      "loss": 1.2884,
      "step": 1998
    },
    {
      "epoch": 0.3129304946775204,
      "grad_norm": 5.352544784545898,
      "learning_rate": 0.00026351417399804493,
      "loss": 1.1595,
      "step": 1999
    },
    {
      "epoch": 0.31308703819661865,
      "grad_norm": 2.5578665733337402,
      "learning_rate": 0.0002634897360703812,
      "loss": 1.4768,
      "step": 2000
    },
    {
      "epoch": 0.31308703819661865,
      "eval_loss": 0.7156243920326233,
      "eval_runtime": 203.0672,
      "eval_samples_per_second": 60.98,
      "eval_steps_per_second": 3.812,
      "eval_wer": 0.4355993794759547,
      "step": 2000
    },
    {
      "epoch": 0.313243581715717,
      "grad_norm": 0.7377122044563293,
      "learning_rate": 0.0002634652981427175,
      "loss": 0.5763,
      "step": 2001
    },
    {
      "epoch": 0.31340012523481525,
      "grad_norm": 0.6151639819145203,
      "learning_rate": 0.00026344086021505374,
      "loss": 0.4244,
      "step": 2002
    },
    {
      "epoch": 0.3135566687539136,
      "grad_norm": 0.7877262830734253,
      "learning_rate": 0.00026341642228739,
      "loss": 0.4682,
      "step": 2003
    },
    {
      "epoch": 0.3137132122730119,
      "grad_norm": 0.6403424143791199,
      "learning_rate": 0.0002633919843597263,
      "loss": 0.4186,
      "step": 2004
    },
    {
      "epoch": 0.3138697557921102,
      "grad_norm": 0.7786056995391846,
      "learning_rate": 0.00026336754643206255,
      "loss": 0.5447,
      "step": 2005
    },
    {
      "epoch": 0.3140262993112085,
      "grad_norm": 1.1472690105438232,
      "learning_rate": 0.0002633431085043988,
      "loss": 0.4301,
      "step": 2006
    },
    {
      "epoch": 0.31418284283030684,
      "grad_norm": 0.6084722280502319,
      "learning_rate": 0.0002633186705767351,
      "loss": 0.3747,
      "step": 2007
    },
    {
      "epoch": 0.3143393863494051,
      "grad_norm": 0.780753493309021,
      "learning_rate": 0.0002632942326490713,
      "loss": 0.6128,
      "step": 2008
    },
    {
      "epoch": 0.31449592986850344,
      "grad_norm": 0.6669083833694458,
      "learning_rate": 0.0002632697947214076,
      "loss": 0.5328,
      "step": 2009
    },
    {
      "epoch": 0.31465247338760177,
      "grad_norm": 1.1285808086395264,
      "learning_rate": 0.00026324535679374386,
      "loss": 0.457,
      "step": 2010
    },
    {
      "epoch": 0.31480901690670005,
      "grad_norm": 0.7840350270271301,
      "learning_rate": 0.0002632209188660801,
      "loss": 0.5403,
      "step": 2011
    },
    {
      "epoch": 0.3149655604257984,
      "grad_norm": 0.7977207899093628,
      "learning_rate": 0.0002631964809384164,
      "loss": 0.5125,
      "step": 2012
    },
    {
      "epoch": 0.3151221039448967,
      "grad_norm": 3.0473263263702393,
      "learning_rate": 0.00026317204301075267,
      "loss": 0.5996,
      "step": 2013
    },
    {
      "epoch": 0.315278647463995,
      "grad_norm": 1.2086435556411743,
      "learning_rate": 0.0002631476050830889,
      "loss": 0.5371,
      "step": 2014
    },
    {
      "epoch": 0.3154351909830933,
      "grad_norm": 1.242156744003296,
      "learning_rate": 0.00026312316715542517,
      "loss": 0.5623,
      "step": 2015
    },
    {
      "epoch": 0.31559173450219163,
      "grad_norm": 1.3142002820968628,
      "learning_rate": 0.0002630987292277615,
      "loss": 0.7507,
      "step": 2016
    },
    {
      "epoch": 0.3157482780212899,
      "grad_norm": 1.0907604694366455,
      "learning_rate": 0.0002630742913000977,
      "loss": 0.5674,
      "step": 2017
    },
    {
      "epoch": 0.31590482154038824,
      "grad_norm": 2.107612133026123,
      "learning_rate": 0.000263049853372434,
      "loss": 0.9088,
      "step": 2018
    },
    {
      "epoch": 0.3160613650594865,
      "grad_norm": 2.506220817565918,
      "learning_rate": 0.0002630254154447703,
      "loss": 0.9719,
      "step": 2019
    },
    {
      "epoch": 0.31621790857858484,
      "grad_norm": 1.3462073802947998,
      "learning_rate": 0.00026300097751710654,
      "loss": 0.7515,
      "step": 2020
    },
    {
      "epoch": 0.31637445209768317,
      "grad_norm": 2.0875937938690186,
      "learning_rate": 0.0002629765395894428,
      "loss": 0.7944,
      "step": 2021
    },
    {
      "epoch": 0.31653099561678144,
      "grad_norm": 2.3353264331817627,
      "learning_rate": 0.0002629521016617791,
      "loss": 1.2749,
      "step": 2022
    },
    {
      "epoch": 0.31668753913587977,
      "grad_norm": 2.3827357292175293,
      "learning_rate": 0.0002629276637341153,
      "loss": 1.3145,
      "step": 2023
    },
    {
      "epoch": 0.3168440826549781,
      "grad_norm": 1.2165812253952026,
      "learning_rate": 0.0002629032258064516,
      "loss": 0.6288,
      "step": 2024
    },
    {
      "epoch": 0.31700062617407637,
      "grad_norm": 2.692824363708496,
      "learning_rate": 0.00026287878787878785,
      "loss": 0.9532,
      "step": 2025
    },
    {
      "epoch": 0.3171571696931747,
      "grad_norm": 1.6349315643310547,
      "learning_rate": 0.0002628543499511241,
      "loss": 0.6626,
      "step": 2026
    },
    {
      "epoch": 0.31731371321227303,
      "grad_norm": 1.819525957107544,
      "learning_rate": 0.0002628299120234604,
      "loss": 0.6427,
      "step": 2027
    },
    {
      "epoch": 0.3174702567313713,
      "grad_norm": 1.595823049545288,
      "learning_rate": 0.00026280547409579665,
      "loss": 0.7039,
      "step": 2028
    },
    {
      "epoch": 0.31762680025046963,
      "grad_norm": 1.4594745635986328,
      "learning_rate": 0.0002627810361681329,
      "loss": 1.2991,
      "step": 2029
    },
    {
      "epoch": 0.31778334376956796,
      "grad_norm": 2.08130145072937,
      "learning_rate": 0.0002627565982404692,
      "loss": 0.9092,
      "step": 2030
    },
    {
      "epoch": 0.31793988728866623,
      "grad_norm": 1.4197502136230469,
      "learning_rate": 0.00026273216031280546,
      "loss": 0.8779,
      "step": 2031
    },
    {
      "epoch": 0.31809643080776456,
      "grad_norm": 1.6842551231384277,
      "learning_rate": 0.0002627077223851417,
      "loss": 0.8164,
      "step": 2032
    },
    {
      "epoch": 0.3182529743268629,
      "grad_norm": 2.0086166858673096,
      "learning_rate": 0.00026268328445747796,
      "loss": 1.2897,
      "step": 2033
    },
    {
      "epoch": 0.31840951784596117,
      "grad_norm": 1.6585288047790527,
      "learning_rate": 0.00026265884652981427,
      "loss": 0.9483,
      "step": 2034
    },
    {
      "epoch": 0.3185660613650595,
      "grad_norm": 2.2773189544677734,
      "learning_rate": 0.0002626344086021505,
      "loss": 0.9985,
      "step": 2035
    },
    {
      "epoch": 0.3187226048841578,
      "grad_norm": 1.6106942892074585,
      "learning_rate": 0.00026260997067448677,
      "loss": 0.8284,
      "step": 2036
    },
    {
      "epoch": 0.3188791484032561,
      "grad_norm": 4.228276252746582,
      "learning_rate": 0.0002625855327468231,
      "loss": 1.1095,
      "step": 2037
    },
    {
      "epoch": 0.3190356919223544,
      "grad_norm": 2.243229866027832,
      "learning_rate": 0.0002625610948191593,
      "loss": 1.3409,
      "step": 2038
    },
    {
      "epoch": 0.3191922354414527,
      "grad_norm": 1.684822678565979,
      "learning_rate": 0.0002625366568914956,
      "loss": 1.0103,
      "step": 2039
    },
    {
      "epoch": 0.319348778960551,
      "grad_norm": 2.6656551361083984,
      "learning_rate": 0.00026251221896383183,
      "loss": 1.5314,
      "step": 2040
    },
    {
      "epoch": 0.31950532247964936,
      "grad_norm": 2.3108103275299072,
      "learning_rate": 0.0002624877810361681,
      "loss": 1.4516,
      "step": 2041
    },
    {
      "epoch": 0.31966186599874763,
      "grad_norm": 3.1501247882843018,
      "learning_rate": 0.0002624633431085044,
      "loss": 1.5163,
      "step": 2042
    },
    {
      "epoch": 0.31981840951784596,
      "grad_norm": 3.26369571685791,
      "learning_rate": 0.00026243890518084064,
      "loss": 1.4617,
      "step": 2043
    },
    {
      "epoch": 0.3199749530369443,
      "grad_norm": 2.4706554412841797,
      "learning_rate": 0.0002624144672531769,
      "loss": 1.5977,
      "step": 2044
    },
    {
      "epoch": 0.32013149655604256,
      "grad_norm": 2.1074883937835693,
      "learning_rate": 0.0002623900293255132,
      "loss": 1.0742,
      "step": 2045
    },
    {
      "epoch": 0.3202880400751409,
      "grad_norm": 1.818274974822998,
      "learning_rate": 0.00026236559139784945,
      "loss": 0.9828,
      "step": 2046
    },
    {
      "epoch": 0.3204445835942392,
      "grad_norm": 4.391941070556641,
      "learning_rate": 0.0002623411534701857,
      "loss": 1.3465,
      "step": 2047
    },
    {
      "epoch": 0.3206011271133375,
      "grad_norm": 2.8843443393707275,
      "learning_rate": 0.00026231671554252195,
      "loss": 1.4826,
      "step": 2048
    },
    {
      "epoch": 0.3207576706324358,
      "grad_norm": 1.6411322355270386,
      "learning_rate": 0.00026229227761485825,
      "loss": 0.8778,
      "step": 2049
    },
    {
      "epoch": 0.32091421415153415,
      "grad_norm": 1.7798197269439697,
      "learning_rate": 0.0002622678396871945,
      "loss": 1.4236,
      "step": 2050
    },
    {
      "epoch": 0.3210707576706324,
      "grad_norm": 0.6216705441474915,
      "learning_rate": 0.00026224340175953076,
      "loss": 0.3424,
      "step": 2051
    },
    {
      "epoch": 0.32122730118973075,
      "grad_norm": 0.7121636271476746,
      "learning_rate": 0.00026221896383186706,
      "loss": 0.3667,
      "step": 2052
    },
    {
      "epoch": 0.3213838447088291,
      "grad_norm": 1.5780848264694214,
      "learning_rate": 0.0002621945259042033,
      "loss": 0.5881,
      "step": 2053
    },
    {
      "epoch": 0.32154038822792735,
      "grad_norm": 1.3239952325820923,
      "learning_rate": 0.00026217008797653957,
      "loss": 0.6034,
      "step": 2054
    },
    {
      "epoch": 0.3216969317470257,
      "grad_norm": 0.8389119505882263,
      "learning_rate": 0.00026214565004887587,
      "loss": 0.4193,
      "step": 2055
    },
    {
      "epoch": 0.32185347526612396,
      "grad_norm": 1.1425360441207886,
      "learning_rate": 0.00026212121212121207,
      "loss": 0.4454,
      "step": 2056
    },
    {
      "epoch": 0.3220100187852223,
      "grad_norm": 0.9988243579864502,
      "learning_rate": 0.0002620967741935484,
      "loss": 0.4994,
      "step": 2057
    },
    {
      "epoch": 0.3221665623043206,
      "grad_norm": 0.9738520383834839,
      "learning_rate": 0.0002620723362658846,
      "loss": 0.4687,
      "step": 2058
    },
    {
      "epoch": 0.3223231058234189,
      "grad_norm": 0.7718535661697388,
      "learning_rate": 0.0002620478983382209,
      "loss": 0.5803,
      "step": 2059
    },
    {
      "epoch": 0.3224796493425172,
      "grad_norm": 1.0445855855941772,
      "learning_rate": 0.0002620234604105572,
      "loss": 0.5174,
      "step": 2060
    },
    {
      "epoch": 0.32263619286161555,
      "grad_norm": 1.0972551107406616,
      "learning_rate": 0.00026199902248289343,
      "loss": 0.5589,
      "step": 2061
    },
    {
      "epoch": 0.3227927363807138,
      "grad_norm": 0.9006467461585999,
      "learning_rate": 0.0002619745845552297,
      "loss": 0.6339,
      "step": 2062
    },
    {
      "epoch": 0.32294927989981215,
      "grad_norm": 1.221784234046936,
      "learning_rate": 0.00026195014662756594,
      "loss": 0.6673,
      "step": 2063
    },
    {
      "epoch": 0.3231058234189105,
      "grad_norm": 0.9309597611427307,
      "learning_rate": 0.00026192570869990224,
      "loss": 0.5886,
      "step": 2064
    },
    {
      "epoch": 0.32326236693800875,
      "grad_norm": 1.0560715198516846,
      "learning_rate": 0.0002619012707722385,
      "loss": 0.8406,
      "step": 2065
    },
    {
      "epoch": 0.3234189104571071,
      "grad_norm": 1.26978600025177,
      "learning_rate": 0.00026187683284457474,
      "loss": 0.4253,
      "step": 2066
    },
    {
      "epoch": 0.3235754539762054,
      "grad_norm": 1.005332350730896,
      "learning_rate": 0.00026185239491691105,
      "loss": 0.6387,
      "step": 2067
    },
    {
      "epoch": 0.3237319974953037,
      "grad_norm": 2.363250970840454,
      "learning_rate": 0.0002618279569892473,
      "loss": 0.7234,
      "step": 2068
    },
    {
      "epoch": 0.323888541014402,
      "grad_norm": 1.3625165224075317,
      "learning_rate": 0.00026180351906158355,
      "loss": 0.6617,
      "step": 2069
    },
    {
      "epoch": 0.32404508453350034,
      "grad_norm": 1.591560959815979,
      "learning_rate": 0.00026177908113391986,
      "loss": 0.7068,
      "step": 2070
    },
    {
      "epoch": 0.3242016280525986,
      "grad_norm": 1.429697036743164,
      "learning_rate": 0.00026175464320625605,
      "loss": 0.7811,
      "step": 2071
    },
    {
      "epoch": 0.32435817157169694,
      "grad_norm": 2.3727924823760986,
      "learning_rate": 0.00026173020527859236,
      "loss": 0.885,
      "step": 2072
    },
    {
      "epoch": 0.3245147150907952,
      "grad_norm": 1.7005720138549805,
      "learning_rate": 0.0002617057673509286,
      "loss": 0.8557,
      "step": 2073
    },
    {
      "epoch": 0.32467125860989354,
      "grad_norm": 1.7432682514190674,
      "learning_rate": 0.00026168132942326486,
      "loss": 0.8502,
      "step": 2074
    },
    {
      "epoch": 0.32482780212899187,
      "grad_norm": 2.942030906677246,
      "learning_rate": 0.00026165689149560117,
      "loss": 1.4696,
      "step": 2075
    },
    {
      "epoch": 0.32498434564809014,
      "grad_norm": 1.5261107683181763,
      "learning_rate": 0.0002616324535679374,
      "loss": 0.5626,
      "step": 2076
    },
    {
      "epoch": 0.3251408891671885,
      "grad_norm": 1.8354300260543823,
      "learning_rate": 0.00026160801564027367,
      "loss": 0.7561,
      "step": 2077
    },
    {
      "epoch": 0.3252974326862868,
      "grad_norm": 1.2063688039779663,
      "learning_rate": 0.00026158357771261,
      "loss": 0.4853,
      "step": 2078
    },
    {
      "epoch": 0.3254539762053851,
      "grad_norm": 2.0545923709869385,
      "learning_rate": 0.0002615591397849462,
      "loss": 0.8312,
      "step": 2079
    },
    {
      "epoch": 0.3256105197244834,
      "grad_norm": 8.21367073059082,
      "learning_rate": 0.0002615347018572825,
      "loss": 1.2077,
      "step": 2080
    },
    {
      "epoch": 0.32576706324358173,
      "grad_norm": 1.7242724895477295,
      "learning_rate": 0.00026151026392961873,
      "loss": 0.5682,
      "step": 2081
    },
    {
      "epoch": 0.32592360676268,
      "grad_norm": 3.253326177597046,
      "learning_rate": 0.00026148582600195503,
      "loss": 1.3158,
      "step": 2082
    },
    {
      "epoch": 0.32608015028177834,
      "grad_norm": 2.2287137508392334,
      "learning_rate": 0.0002614613880742913,
      "loss": 1.1925,
      "step": 2083
    },
    {
      "epoch": 0.32623669380087666,
      "grad_norm": 2.752636194229126,
      "learning_rate": 0.00026143695014662754,
      "loss": 1.5877,
      "step": 2084
    },
    {
      "epoch": 0.32639323731997494,
      "grad_norm": 1.1275129318237305,
      "learning_rate": 0.00026141251221896384,
      "loss": 0.4494,
      "step": 2085
    },
    {
      "epoch": 0.32654978083907327,
      "grad_norm": 2.6209516525268555,
      "learning_rate": 0.00026138807429130004,
      "loss": 1.5737,
      "step": 2086
    },
    {
      "epoch": 0.3267063243581716,
      "grad_norm": 2.409445285797119,
      "learning_rate": 0.00026136363636363634,
      "loss": 0.9766,
      "step": 2087
    },
    {
      "epoch": 0.32686286787726987,
      "grad_norm": 3.750723123550415,
      "learning_rate": 0.0002613391984359726,
      "loss": 1.4515,
      "step": 2088
    },
    {
      "epoch": 0.3270194113963682,
      "grad_norm": 2.0334136486053467,
      "learning_rate": 0.00026131476050830885,
      "loss": 0.7738,
      "step": 2089
    },
    {
      "epoch": 0.3271759549154665,
      "grad_norm": 3.3774590492248535,
      "learning_rate": 0.00026129032258064515,
      "loss": 1.5601,
      "step": 2090
    },
    {
      "epoch": 0.3273324984345648,
      "grad_norm": 1.3902050256729126,
      "learning_rate": 0.0002612658846529814,
      "loss": 0.8408,
      "step": 2091
    },
    {
      "epoch": 0.32748904195366313,
      "grad_norm": 1.9110989570617676,
      "learning_rate": 0.00026124144672531765,
      "loss": 0.886,
      "step": 2092
    },
    {
      "epoch": 0.3276455854727614,
      "grad_norm": 1.9537615776062012,
      "learning_rate": 0.00026121700879765396,
      "loss": 1.8936,
      "step": 2093
    },
    {
      "epoch": 0.32780212899185973,
      "grad_norm": 3.947450876235962,
      "learning_rate": 0.0002611925708699902,
      "loss": 1.3683,
      "step": 2094
    },
    {
      "epoch": 0.32795867251095806,
      "grad_norm": 1.9001954793930054,
      "learning_rate": 0.00026116813294232646,
      "loss": 1.0555,
      "step": 2095
    },
    {
      "epoch": 0.32811521603005633,
      "grad_norm": 1.3852555751800537,
      "learning_rate": 0.0002611436950146627,
      "loss": 0.8577,
      "step": 2096
    },
    {
      "epoch": 0.32827175954915466,
      "grad_norm": 1.6378505229949951,
      "learning_rate": 0.000261119257086999,
      "loss": 0.5502,
      "step": 2097
    },
    {
      "epoch": 0.328428303068253,
      "grad_norm": 1.4456698894500732,
      "learning_rate": 0.00026109481915933527,
      "loss": 1.164,
      "step": 2098
    },
    {
      "epoch": 0.32858484658735126,
      "grad_norm": 2.254760265350342,
      "learning_rate": 0.0002610703812316715,
      "loss": 1.174,
      "step": 2099
    },
    {
      "epoch": 0.3287413901064496,
      "grad_norm": 2.9925920963287354,
      "learning_rate": 0.00026104594330400783,
      "loss": 1.4058,
      "step": 2100
    },
    {
      "epoch": 0.3288979336255479,
      "grad_norm": 0.5673936605453491,
      "learning_rate": 0.0002610215053763441,
      "loss": 0.4183,
      "step": 2101
    },
    {
      "epoch": 0.3290544771446462,
      "grad_norm": 0.6673199534416199,
      "learning_rate": 0.00026099706744868033,
      "loss": 0.3369,
      "step": 2102
    },
    {
      "epoch": 0.3292110206637445,
      "grad_norm": 0.8384460806846619,
      "learning_rate": 0.0002609726295210166,
      "loss": 0.4785,
      "step": 2103
    },
    {
      "epoch": 0.32936756418284285,
      "grad_norm": 0.8750125765800476,
      "learning_rate": 0.00026094819159335283,
      "loss": 0.5129,
      "step": 2104
    },
    {
      "epoch": 0.3295241077019411,
      "grad_norm": 0.729326069355011,
      "learning_rate": 0.00026092375366568914,
      "loss": 0.5408,
      "step": 2105
    },
    {
      "epoch": 0.32968065122103946,
      "grad_norm": 0.8103402256965637,
      "learning_rate": 0.0002608993157380254,
      "loss": 0.5732,
      "step": 2106
    },
    {
      "epoch": 0.3298371947401378,
      "grad_norm": 1.0650441646575928,
      "learning_rate": 0.00026087487781036164,
      "loss": 0.5434,
      "step": 2107
    },
    {
      "epoch": 0.32999373825923606,
      "grad_norm": 0.8528415560722351,
      "learning_rate": 0.00026085043988269795,
      "loss": 0.4231,
      "step": 2108
    },
    {
      "epoch": 0.3301502817783344,
      "grad_norm": 1.0440559387207031,
      "learning_rate": 0.0002608260019550342,
      "loss": 0.533,
      "step": 2109
    },
    {
      "epoch": 0.33030682529743266,
      "grad_norm": 0.9997760653495789,
      "learning_rate": 0.00026080156402737045,
      "loss": 0.6891,
      "step": 2110
    },
    {
      "epoch": 0.330463368816531,
      "grad_norm": 1.1891223192214966,
      "learning_rate": 0.0002607771260997067,
      "loss": 0.4688,
      "step": 2111
    },
    {
      "epoch": 0.3306199123356293,
      "grad_norm": 1.3714631795883179,
      "learning_rate": 0.000260752688172043,
      "loss": 0.5096,
      "step": 2112
    },
    {
      "epoch": 0.3307764558547276,
      "grad_norm": 1.0900437831878662,
      "learning_rate": 0.00026072825024437926,
      "loss": 0.6755,
      "step": 2113
    },
    {
      "epoch": 0.3309329993738259,
      "grad_norm": 2.1338839530944824,
      "learning_rate": 0.0002607038123167155,
      "loss": 0.8436,
      "step": 2114
    },
    {
      "epoch": 0.33108954289292425,
      "grad_norm": 1.6171543598175049,
      "learning_rate": 0.0002606793743890518,
      "loss": 0.7621,
      "step": 2115
    },
    {
      "epoch": 0.3312460864120225,
      "grad_norm": 0.8001659512519836,
      "learning_rate": 0.00026065493646138806,
      "loss": 0.491,
      "step": 2116
    },
    {
      "epoch": 0.33140262993112085,
      "grad_norm": 1.0575730800628662,
      "learning_rate": 0.0002606304985337243,
      "loss": 0.6333,
      "step": 2117
    },
    {
      "epoch": 0.3315591734502192,
      "grad_norm": 1.6386489868164062,
      "learning_rate": 0.0002606060606060606,
      "loss": 0.73,
      "step": 2118
    },
    {
      "epoch": 0.33171571696931745,
      "grad_norm": 1.1848156452178955,
      "learning_rate": 0.0002605816226783968,
      "loss": 0.5608,
      "step": 2119
    },
    {
      "epoch": 0.3318722604884158,
      "grad_norm": 1.1664621829986572,
      "learning_rate": 0.0002605571847507331,
      "loss": 0.5895,
      "step": 2120
    },
    {
      "epoch": 0.3320288040075141,
      "grad_norm": 1.0166929960250854,
      "learning_rate": 0.0002605327468230694,
      "loss": 0.5018,
      "step": 2121
    },
    {
      "epoch": 0.3321853475266124,
      "grad_norm": 3.1172375679016113,
      "learning_rate": 0.0002605083088954056,
      "loss": 0.7414,
      "step": 2122
    },
    {
      "epoch": 0.3323418910457107,
      "grad_norm": 1.3032346963882446,
      "learning_rate": 0.00026048387096774193,
      "loss": 0.5268,
      "step": 2123
    },
    {
      "epoch": 0.33249843456480904,
      "grad_norm": 2.1473283767700195,
      "learning_rate": 0.0002604594330400782,
      "loss": 0.8175,
      "step": 2124
    },
    {
      "epoch": 0.3326549780839073,
      "grad_norm": 2.331975221633911,
      "learning_rate": 0.00026043499511241443,
      "loss": 1.0549,
      "step": 2125
    },
    {
      "epoch": 0.33281152160300564,
      "grad_norm": 2.3260395526885986,
      "learning_rate": 0.0002604105571847507,
      "loss": 1.1403,
      "step": 2126
    },
    {
      "epoch": 0.3329680651221039,
      "grad_norm": 3.693405866622925,
      "learning_rate": 0.000260386119257087,
      "loss": 0.8268,
      "step": 2127
    },
    {
      "epoch": 0.33312460864120225,
      "grad_norm": 1.5741246938705444,
      "learning_rate": 0.00026036168132942324,
      "loss": 1.0704,
      "step": 2128
    },
    {
      "epoch": 0.3332811521603006,
      "grad_norm": 1.5389891862869263,
      "learning_rate": 0.0002603372434017595,
      "loss": 0.8378,
      "step": 2129
    },
    {
      "epoch": 0.33343769567939885,
      "grad_norm": 1.7030190229415894,
      "learning_rate": 0.0002603128054740958,
      "loss": 1.0205,
      "step": 2130
    },
    {
      "epoch": 0.3335942391984972,
      "grad_norm": 2.1614034175872803,
      "learning_rate": 0.00026028836754643205,
      "loss": 1.128,
      "step": 2131
    },
    {
      "epoch": 0.3337507827175955,
      "grad_norm": 1.5320353507995605,
      "learning_rate": 0.0002602639296187683,
      "loss": 0.7876,
      "step": 2132
    },
    {
      "epoch": 0.3339073262366938,
      "grad_norm": 1.6267305612564087,
      "learning_rate": 0.0002602394916911046,
      "loss": 0.8237,
      "step": 2133
    },
    {
      "epoch": 0.3340638697557921,
      "grad_norm": 3.2950265407562256,
      "learning_rate": 0.0002602150537634408,
      "loss": 1.1185,
      "step": 2134
    },
    {
      "epoch": 0.33422041327489044,
      "grad_norm": 2.2613959312438965,
      "learning_rate": 0.0002601906158357771,
      "loss": 0.91,
      "step": 2135
    },
    {
      "epoch": 0.3343769567939887,
      "grad_norm": 1.922223687171936,
      "learning_rate": 0.00026016617790811336,
      "loss": 1.006,
      "step": 2136
    },
    {
      "epoch": 0.33453350031308704,
      "grad_norm": 3.277189016342163,
      "learning_rate": 0.0002601417399804496,
      "loss": 1.4087,
      "step": 2137
    },
    {
      "epoch": 0.33469004383218537,
      "grad_norm": 2.7033369541168213,
      "learning_rate": 0.0002601173020527859,
      "loss": 1.3359,
      "step": 2138
    },
    {
      "epoch": 0.33484658735128364,
      "grad_norm": 1.8570423126220703,
      "learning_rate": 0.00026009286412512217,
      "loss": 0.9543,
      "step": 2139
    },
    {
      "epoch": 0.33500313087038197,
      "grad_norm": 4.569610595703125,
      "learning_rate": 0.0002600684261974584,
      "loss": 1.0798,
      "step": 2140
    },
    {
      "epoch": 0.3351596743894803,
      "grad_norm": 2.035679578781128,
      "learning_rate": 0.0002600439882697947,
      "loss": 2.1632,
      "step": 2141
    },
    {
      "epoch": 0.3353162179085786,
      "grad_norm": 1.7678215503692627,
      "learning_rate": 0.000260019550342131,
      "loss": 1.4521,
      "step": 2142
    },
    {
      "epoch": 0.3354727614276769,
      "grad_norm": 2.673259735107422,
      "learning_rate": 0.00025999511241446723,
      "loss": 1.5297,
      "step": 2143
    },
    {
      "epoch": 0.33562930494677523,
      "grad_norm": 2.435270309448242,
      "learning_rate": 0.0002599706744868035,
      "loss": 1.1949,
      "step": 2144
    },
    {
      "epoch": 0.3357858484658735,
      "grad_norm": 2.4045560359954834,
      "learning_rate": 0.0002599462365591398,
      "loss": 1.356,
      "step": 2145
    },
    {
      "epoch": 0.33594239198497183,
      "grad_norm": 2.4822990894317627,
      "learning_rate": 0.00025992179863147604,
      "loss": 0.9344,
      "step": 2146
    },
    {
      "epoch": 0.3360989355040701,
      "grad_norm": 1.8672354221343994,
      "learning_rate": 0.0002598973607038123,
      "loss": 0.7547,
      "step": 2147
    },
    {
      "epoch": 0.33625547902316844,
      "grad_norm": 0.980976939201355,
      "learning_rate": 0.0002598729227761486,
      "loss": 0.8016,
      "step": 2148
    },
    {
      "epoch": 0.33641202254226676,
      "grad_norm": 1.5991452932357788,
      "learning_rate": 0.00025984848484848484,
      "loss": 1.0073,
      "step": 2149
    },
    {
      "epoch": 0.33656856606136504,
      "grad_norm": 1.6782243251800537,
      "learning_rate": 0.0002598240469208211,
      "loss": 1.0327,
      "step": 2150
    },
    {
      "epoch": 0.33672510958046337,
      "grad_norm": 0.7231286764144897,
      "learning_rate": 0.00025979960899315735,
      "loss": 0.4323,
      "step": 2151
    },
    {
      "epoch": 0.3368816530995617,
      "grad_norm": 0.8727384805679321,
      "learning_rate": 0.0002597751710654936,
      "loss": 0.5182,
      "step": 2152
    },
    {
      "epoch": 0.33703819661865997,
      "grad_norm": 0.49551793932914734,
      "learning_rate": 0.0002597507331378299,
      "loss": 0.374,
      "step": 2153
    },
    {
      "epoch": 0.3371947401377583,
      "grad_norm": 1.4629757404327393,
      "learning_rate": 0.00025972629521016615,
      "loss": 0.4585,
      "step": 2154
    },
    {
      "epoch": 0.3373512836568566,
      "grad_norm": 0.6694740653038025,
      "learning_rate": 0.0002597018572825024,
      "loss": 0.4198,
      "step": 2155
    },
    {
      "epoch": 0.3375078271759549,
      "grad_norm": 0.9819236993789673,
      "learning_rate": 0.0002596774193548387,
      "loss": 0.4105,
      "step": 2156
    },
    {
      "epoch": 0.33766437069505323,
      "grad_norm": 1.0089393854141235,
      "learning_rate": 0.00025965298142717496,
      "loss": 0.5237,
      "step": 2157
    },
    {
      "epoch": 0.33782091421415156,
      "grad_norm": 0.5175392627716064,
      "learning_rate": 0.0002596285434995112,
      "loss": 0.4374,
      "step": 2158
    },
    {
      "epoch": 0.33797745773324983,
      "grad_norm": 0.8120709657669067,
      "learning_rate": 0.00025960410557184746,
      "loss": 0.5646,
      "step": 2159
    },
    {
      "epoch": 0.33813400125234816,
      "grad_norm": 0.9096924662590027,
      "learning_rate": 0.00025957966764418377,
      "loss": 0.5908,
      "step": 2160
    },
    {
      "epoch": 0.3382905447714465,
      "grad_norm": 0.8007375001907349,
      "learning_rate": 0.00025955522971652,
      "loss": 0.4136,
      "step": 2161
    },
    {
      "epoch": 0.33844708829054476,
      "grad_norm": 1.0222002267837524,
      "learning_rate": 0.00025953079178885627,
      "loss": 0.7746,
      "step": 2162
    },
    {
      "epoch": 0.3386036318096431,
      "grad_norm": 0.897656261920929,
      "learning_rate": 0.0002595063538611926,
      "loss": 0.4193,
      "step": 2163
    },
    {
      "epoch": 0.33876017532874136,
      "grad_norm": 1.0817437171936035,
      "learning_rate": 0.00025948191593352883,
      "loss": 0.3561,
      "step": 2164
    },
    {
      "epoch": 0.3389167188478397,
      "grad_norm": 1.0257635116577148,
      "learning_rate": 0.0002594574780058651,
      "loss": 0.4638,
      "step": 2165
    },
    {
      "epoch": 0.339073262366938,
      "grad_norm": 1.651721715927124,
      "learning_rate": 0.0002594330400782014,
      "loss": 0.8187,
      "step": 2166
    },
    {
      "epoch": 0.3392298058860363,
      "grad_norm": 1.1372004747390747,
      "learning_rate": 0.0002594086021505376,
      "loss": 0.5079,
      "step": 2167
    },
    {
      "epoch": 0.3393863494051346,
      "grad_norm": 1.8931221961975098,
      "learning_rate": 0.0002593841642228739,
      "loss": 0.7663,
      "step": 2168
    },
    {
      "epoch": 0.33954289292423295,
      "grad_norm": 1.1629186868667603,
      "learning_rate": 0.00025935972629521014,
      "loss": 0.6915,
      "step": 2169
    },
    {
      "epoch": 0.3396994364433312,
      "grad_norm": 1.5948940515518188,
      "learning_rate": 0.0002593352883675464,
      "loss": 0.9156,
      "step": 2170
    },
    {
      "epoch": 0.33985597996242956,
      "grad_norm": 1.6868611574172974,
      "learning_rate": 0.0002593108504398827,
      "loss": 0.6997,
      "step": 2171
    },
    {
      "epoch": 0.3400125234815279,
      "grad_norm": 2.6906371116638184,
      "learning_rate": 0.00025928641251221895,
      "loss": 0.744,
      "step": 2172
    },
    {
      "epoch": 0.34016906700062616,
      "grad_norm": 1.4386720657348633,
      "learning_rate": 0.0002592619745845552,
      "loss": 0.5701,
      "step": 2173
    },
    {
      "epoch": 0.3403256105197245,
      "grad_norm": 1.7945785522460938,
      "learning_rate": 0.00025923753665689145,
      "loss": 0.9001,
      "step": 2174
    },
    {
      "epoch": 0.3404821540388228,
      "grad_norm": 1.8731091022491455,
      "learning_rate": 0.00025921309872922776,
      "loss": 0.7199,
      "step": 2175
    },
    {
      "epoch": 0.3406386975579211,
      "grad_norm": 2.1651723384857178,
      "learning_rate": 0.000259188660801564,
      "loss": 1.005,
      "step": 2176
    },
    {
      "epoch": 0.3407952410770194,
      "grad_norm": 2.3972442150115967,
      "learning_rate": 0.00025916422287390026,
      "loss": 1.1601,
      "step": 2177
    },
    {
      "epoch": 0.34095178459611775,
      "grad_norm": 1.9120584726333618,
      "learning_rate": 0.00025913978494623656,
      "loss": 0.8559,
      "step": 2178
    },
    {
      "epoch": 0.341108328115216,
      "grad_norm": 2.741961717605591,
      "learning_rate": 0.0002591153470185728,
      "loss": 0.9065,
      "step": 2179
    },
    {
      "epoch": 0.34126487163431435,
      "grad_norm": 2.225860834121704,
      "learning_rate": 0.00025909090909090907,
      "loss": 1.1946,
      "step": 2180
    },
    {
      "epoch": 0.3414214151534126,
      "grad_norm": 2.113170623779297,
      "learning_rate": 0.00025906647116324537,
      "loss": 1.0144,
      "step": 2181
    },
    {
      "epoch": 0.34157795867251095,
      "grad_norm": 3.7552993297576904,
      "learning_rate": 0.00025904203323558157,
      "loss": 1.0589,
      "step": 2182
    },
    {
      "epoch": 0.3417345021916093,
      "grad_norm": 2.8972368240356445,
      "learning_rate": 0.0002590175953079179,
      "loss": 1.034,
      "step": 2183
    },
    {
      "epoch": 0.34189104571070755,
      "grad_norm": 9.414438247680664,
      "learning_rate": 0.0002589931573802541,
      "loss": 1.3211,
      "step": 2184
    },
    {
      "epoch": 0.3420475892298059,
      "grad_norm": 2.1980931758880615,
      "learning_rate": 0.0002589687194525904,
      "loss": 1.1937,
      "step": 2185
    },
    {
      "epoch": 0.3422041327489042,
      "grad_norm": 1.7976131439208984,
      "learning_rate": 0.0002589442815249267,
      "loss": 1.2868,
      "step": 2186
    },
    {
      "epoch": 0.3423606762680025,
      "grad_norm": 3.20818829536438,
      "learning_rate": 0.00025891984359726293,
      "loss": 1.3121,
      "step": 2187
    },
    {
      "epoch": 0.3425172197871008,
      "grad_norm": 2.546349287033081,
      "learning_rate": 0.0002588954056695992,
      "loss": 1.4822,
      "step": 2188
    },
    {
      "epoch": 0.34267376330619914,
      "grad_norm": 2.756331205368042,
      "learning_rate": 0.0002588709677419355,
      "loss": 1.3909,
      "step": 2189
    },
    {
      "epoch": 0.3428303068252974,
      "grad_norm": 2.7843143939971924,
      "learning_rate": 0.00025884652981427174,
      "loss": 1.4207,
      "step": 2190
    },
    {
      "epoch": 0.34298685034439574,
      "grad_norm": 2.695613145828247,
      "learning_rate": 0.000258822091886608,
      "loss": 1.0539,
      "step": 2191
    },
    {
      "epoch": 0.3431433938634941,
      "grad_norm": 2.704921007156372,
      "learning_rate": 0.00025879765395894424,
      "loss": 1.2176,
      "step": 2192
    },
    {
      "epoch": 0.34329993738259235,
      "grad_norm": 2.946237802505493,
      "learning_rate": 0.00025877321603128055,
      "loss": 1.6507,
      "step": 2193
    },
    {
      "epoch": 0.3434564809016907,
      "grad_norm": 2.0598270893096924,
      "learning_rate": 0.0002587487781036168,
      "loss": 1.5005,
      "step": 2194
    },
    {
      "epoch": 0.343613024420789,
      "grad_norm": 2.0431604385375977,
      "learning_rate": 0.00025872434017595305,
      "loss": 0.7967,
      "step": 2195
    },
    {
      "epoch": 0.3437695679398873,
      "grad_norm": 3.8533642292022705,
      "learning_rate": 0.00025869990224828936,
      "loss": 1.6454,
      "step": 2196
    },
    {
      "epoch": 0.3439261114589856,
      "grad_norm": 1.8645732402801514,
      "learning_rate": 0.00025867546432062555,
      "loss": 1.0377,
      "step": 2197
    },
    {
      "epoch": 0.34408265497808393,
      "grad_norm": 2.5659990310668945,
      "learning_rate": 0.00025865102639296186,
      "loss": 0.9068,
      "step": 2198
    },
    {
      "epoch": 0.3442391984971822,
      "grad_norm": 2.7070820331573486,
      "learning_rate": 0.0002586265884652981,
      "loss": 0.9772,
      "step": 2199
    },
    {
      "epoch": 0.34439574201628054,
      "grad_norm": 1.378251075744629,
      "learning_rate": 0.00025860215053763436,
      "loss": 1.4371,
      "step": 2200
    },
    {
      "epoch": 0.3445522855353788,
      "grad_norm": 0.5317636132240295,
      "learning_rate": 0.00025857771260997067,
      "loss": 0.4837,
      "step": 2201
    },
    {
      "epoch": 0.34470882905447714,
      "grad_norm": 0.7095066905021667,
      "learning_rate": 0.0002585532746823069,
      "loss": 0.579,
      "step": 2202
    },
    {
      "epoch": 0.34486537257357547,
      "grad_norm": 0.9275551438331604,
      "learning_rate": 0.00025852883675464317,
      "loss": 0.6504,
      "step": 2203
    },
    {
      "epoch": 0.34502191609267374,
      "grad_norm": 1.716207504272461,
      "learning_rate": 0.0002585043988269795,
      "loss": 0.4966,
      "step": 2204
    },
    {
      "epoch": 0.34517845961177207,
      "grad_norm": 0.6072659492492676,
      "learning_rate": 0.0002584799608993157,
      "loss": 0.5413,
      "step": 2205
    },
    {
      "epoch": 0.3453350031308704,
      "grad_norm": 1.4741132259368896,
      "learning_rate": 0.000258455522971652,
      "loss": 0.4245,
      "step": 2206
    },
    {
      "epoch": 0.3454915466499687,
      "grad_norm": 0.9101395010948181,
      "learning_rate": 0.00025843108504398823,
      "loss": 0.5105,
      "step": 2207
    },
    {
      "epoch": 0.345648090169067,
      "grad_norm": 0.8929093480110168,
      "learning_rate": 0.00025840664711632453,
      "loss": 0.5917,
      "step": 2208
    },
    {
      "epoch": 0.34580463368816533,
      "grad_norm": 0.9677044153213501,
      "learning_rate": 0.0002583822091886608,
      "loss": 0.618,
      "step": 2209
    },
    {
      "epoch": 0.3459611772072636,
      "grad_norm": 0.9117663502693176,
      "learning_rate": 0.00025835777126099704,
      "loss": 0.5954,
      "step": 2210
    },
    {
      "epoch": 0.34611772072636193,
      "grad_norm": 0.8379285335540771,
      "learning_rate": 0.00025833333333333334,
      "loss": 0.4431,
      "step": 2211
    },
    {
      "epoch": 0.34627426424546026,
      "grad_norm": 0.8588935136795044,
      "learning_rate": 0.0002583088954056696,
      "loss": 0.4436,
      "step": 2212
    },
    {
      "epoch": 0.34643080776455853,
      "grad_norm": 1.0925568342208862,
      "learning_rate": 0.00025828445747800584,
      "loss": 0.3352,
      "step": 2213
    },
    {
      "epoch": 0.34658735128365686,
      "grad_norm": 1.6619585752487183,
      "learning_rate": 0.00025826001955034215,
      "loss": 0.6069,
      "step": 2214
    },
    {
      "epoch": 0.3467438948027552,
      "grad_norm": 2.354327440261841,
      "learning_rate": 0.00025823558162267835,
      "loss": 0.7189,
      "step": 2215
    },
    {
      "epoch": 0.34690043832185347,
      "grad_norm": 1.1300790309906006,
      "learning_rate": 0.00025821114369501465,
      "loss": 0.4776,
      "step": 2216
    },
    {
      "epoch": 0.3470569818409518,
      "grad_norm": 1.8510233163833618,
      "learning_rate": 0.0002581867057673509,
      "loss": 0.8701,
      "step": 2217
    },
    {
      "epoch": 0.34721352536005007,
      "grad_norm": 1.0941369533538818,
      "learning_rate": 0.00025816226783968716,
      "loss": 0.621,
      "step": 2218
    },
    {
      "epoch": 0.3473700688791484,
      "grad_norm": 1.9270963668823242,
      "learning_rate": 0.00025813782991202346,
      "loss": 0.4966,
      "step": 2219
    },
    {
      "epoch": 0.3475266123982467,
      "grad_norm": 2.781996726989746,
      "learning_rate": 0.0002581133919843597,
      "loss": 0.7897,
      "step": 2220
    },
    {
      "epoch": 0.347683155917345,
      "grad_norm": 1.3982248306274414,
      "learning_rate": 0.00025808895405669596,
      "loss": 0.7361,
      "step": 2221
    },
    {
      "epoch": 0.34783969943644333,
      "grad_norm": 1.4898085594177246,
      "learning_rate": 0.0002580645161290322,
      "loss": 0.6928,
      "step": 2222
    },
    {
      "epoch": 0.34799624295554166,
      "grad_norm": 1.5134382247924805,
      "learning_rate": 0.0002580400782013685,
      "loss": 0.6501,
      "step": 2223
    },
    {
      "epoch": 0.34815278647463993,
      "grad_norm": 2.149883985519409,
      "learning_rate": 0.00025801564027370477,
      "loss": 0.8948,
      "step": 2224
    },
    {
      "epoch": 0.34830932999373826,
      "grad_norm": 2.79316782951355,
      "learning_rate": 0.000257991202346041,
      "loss": 0.9539,
      "step": 2225
    },
    {
      "epoch": 0.3484658735128366,
      "grad_norm": 1.0646382570266724,
      "learning_rate": 0.00025796676441837733,
      "loss": 0.6291,
      "step": 2226
    },
    {
      "epoch": 0.34862241703193486,
      "grad_norm": 1.6639641523361206,
      "learning_rate": 0.0002579423264907136,
      "loss": 0.8177,
      "step": 2227
    },
    {
      "epoch": 0.3487789605510332,
      "grad_norm": 1.5588454008102417,
      "learning_rate": 0.00025791788856304983,
      "loss": 0.6531,
      "step": 2228
    },
    {
      "epoch": 0.3489355040701315,
      "grad_norm": 2.3921139240264893,
      "learning_rate": 0.00025789345063538614,
      "loss": 0.976,
      "step": 2229
    },
    {
      "epoch": 0.3490920475892298,
      "grad_norm": 3.0724098682403564,
      "learning_rate": 0.00025786901270772233,
      "loss": 1.2253,
      "step": 2230
    },
    {
      "epoch": 0.3492485911083281,
      "grad_norm": 1.900876760482788,
      "learning_rate": 0.00025784457478005864,
      "loss": 1.3439,
      "step": 2231
    },
    {
      "epoch": 0.34940513462742645,
      "grad_norm": 1.975421667098999,
      "learning_rate": 0.0002578201368523949,
      "loss": 1.0304,
      "step": 2232
    },
    {
      "epoch": 0.3495616781465247,
      "grad_norm": 1.7853161096572876,
      "learning_rate": 0.00025779569892473114,
      "loss": 0.8912,
      "step": 2233
    },
    {
      "epoch": 0.34971822166562305,
      "grad_norm": 2.0983622074127197,
      "learning_rate": 0.00025777126099706745,
      "loss": 0.7942,
      "step": 2234
    },
    {
      "epoch": 0.3498747651847213,
      "grad_norm": 2.663015127182007,
      "learning_rate": 0.0002577468230694037,
      "loss": 1.3245,
      "step": 2235
    },
    {
      "epoch": 0.35003130870381965,
      "grad_norm": 3.564725399017334,
      "learning_rate": 0.00025772238514173995,
      "loss": 1.5237,
      "step": 2236
    },
    {
      "epoch": 0.350187852222918,
      "grad_norm": 3.92618727684021,
      "learning_rate": 0.00025769794721407625,
      "loss": 0.9078,
      "step": 2237
    },
    {
      "epoch": 0.35034439574201626,
      "grad_norm": 2.12117862701416,
      "learning_rate": 0.0002576735092864125,
      "loss": 1.2595,
      "step": 2238
    },
    {
      "epoch": 0.3505009392611146,
      "grad_norm": 2.0806806087493896,
      "learning_rate": 0.00025764907135874876,
      "loss": 1.0926,
      "step": 2239
    },
    {
      "epoch": 0.3506574827802129,
      "grad_norm": 2.7820568084716797,
      "learning_rate": 0.000257624633431085,
      "loss": 1.1132,
      "step": 2240
    },
    {
      "epoch": 0.3508140262993112,
      "grad_norm": 2.921229362487793,
      "learning_rate": 0.0002576001955034213,
      "loss": 1.3909,
      "step": 2241
    },
    {
      "epoch": 0.3509705698184095,
      "grad_norm": 2.3714778423309326,
      "learning_rate": 0.00025757575757575756,
      "loss": 1.3431,
      "step": 2242
    },
    {
      "epoch": 0.35112711333750785,
      "grad_norm": 1.7340811491012573,
      "learning_rate": 0.0002575513196480938,
      "loss": 1.0429,
      "step": 2243
    },
    {
      "epoch": 0.3512836568566061,
      "grad_norm": 2.6049928665161133,
      "learning_rate": 0.0002575268817204301,
      "loss": 1.5378,
      "step": 2244
    },
    {
      "epoch": 0.35144020037570445,
      "grad_norm": 2.498192310333252,
      "learning_rate": 0.0002575024437927663,
      "loss": 0.9605,
      "step": 2245
    },
    {
      "epoch": 0.3515967438948028,
      "grad_norm": 2.289841890335083,
      "learning_rate": 0.0002574780058651026,
      "loss": 1.601,
      "step": 2246
    },
    {
      "epoch": 0.35175328741390105,
      "grad_norm": 3.376864194869995,
      "learning_rate": 0.0002574535679374389,
      "loss": 0.9655,
      "step": 2247
    },
    {
      "epoch": 0.3519098309329994,
      "grad_norm": 1.3867840766906738,
      "learning_rate": 0.0002574291300097751,
      "loss": 0.8568,
      "step": 2248
    },
    {
      "epoch": 0.3520663744520977,
      "grad_norm": 2.4591383934020996,
      "learning_rate": 0.00025740469208211143,
      "loss": 0.9717,
      "step": 2249
    },
    {
      "epoch": 0.352222917971196,
      "grad_norm": 2.244575023651123,
      "learning_rate": 0.0002573802541544477,
      "loss": 1.1796,
      "step": 2250
    },
    {
      "epoch": 0.3523794614902943,
      "grad_norm": 0.7717861533164978,
      "learning_rate": 0.00025735581622678393,
      "loss": 0.3882,
      "step": 2251
    },
    {
      "epoch": 0.3525360050093926,
      "grad_norm": 0.5575485825538635,
      "learning_rate": 0.00025733137829912024,
      "loss": 0.2957,
      "step": 2252
    },
    {
      "epoch": 0.3526925485284909,
      "grad_norm": 0.7117191553115845,
      "learning_rate": 0.0002573069403714565,
      "loss": 0.5051,
      "step": 2253
    },
    {
      "epoch": 0.35284909204758924,
      "grad_norm": 0.5327755808830261,
      "learning_rate": 0.00025728250244379274,
      "loss": 0.3964,
      "step": 2254
    },
    {
      "epoch": 0.3530056355666875,
      "grad_norm": 0.7493363618850708,
      "learning_rate": 0.000257258064516129,
      "loss": 0.4411,
      "step": 2255
    },
    {
      "epoch": 0.35316217908578584,
      "grad_norm": 0.5741692185401917,
      "learning_rate": 0.0002572336265884653,
      "loss": 0.3451,
      "step": 2256
    },
    {
      "epoch": 0.3533187226048842,
      "grad_norm": 0.6874625086784363,
      "learning_rate": 0.00025720918866080155,
      "loss": 0.3136,
      "step": 2257
    },
    {
      "epoch": 0.35347526612398245,
      "grad_norm": 1.1320338249206543,
      "learning_rate": 0.0002571847507331378,
      "loss": 0.5177,
      "step": 2258
    },
    {
      "epoch": 0.3536318096430808,
      "grad_norm": 1.1008598804473877,
      "learning_rate": 0.0002571603128054741,
      "loss": 0.3906,
      "step": 2259
    },
    {
      "epoch": 0.3537883531621791,
      "grad_norm": 1.950392484664917,
      "learning_rate": 0.00025713587487781036,
      "loss": 0.4661,
      "step": 2260
    },
    {
      "epoch": 0.3539448966812774,
      "grad_norm": 0.9202426075935364,
      "learning_rate": 0.0002571114369501466,
      "loss": 0.3923,
      "step": 2261
    },
    {
      "epoch": 0.3541014402003757,
      "grad_norm": 1.0292601585388184,
      "learning_rate": 0.00025708699902248286,
      "loss": 0.5517,
      "step": 2262
    },
    {
      "epoch": 0.35425798371947403,
      "grad_norm": 0.9722676277160645,
      "learning_rate": 0.0002570625610948191,
      "loss": 0.432,
      "step": 2263
    },
    {
      "epoch": 0.3544145272385723,
      "grad_norm": 2.0393378734588623,
      "learning_rate": 0.0002570381231671554,
      "loss": 0.4121,
      "step": 2264
    },
    {
      "epoch": 0.35457107075767064,
      "grad_norm": 1.5482399463653564,
      "learning_rate": 0.00025701368523949167,
      "loss": 0.571,
      "step": 2265
    },
    {
      "epoch": 0.35472761427676897,
      "grad_norm": 1.6801763772964478,
      "learning_rate": 0.0002569892473118279,
      "loss": 0.6699,
      "step": 2266
    },
    {
      "epoch": 0.35488415779586724,
      "grad_norm": 1.4256728887557983,
      "learning_rate": 0.0002569648093841642,
      "loss": 0.9005,
      "step": 2267
    },
    {
      "epoch": 0.35504070131496557,
      "grad_norm": 1.3114697933197021,
      "learning_rate": 0.0002569403714565005,
      "loss": 0.5094,
      "step": 2268
    },
    {
      "epoch": 0.3551972448340639,
      "grad_norm": 1.3349591493606567,
      "learning_rate": 0.00025691593352883673,
      "loss": 0.7312,
      "step": 2269
    },
    {
      "epoch": 0.35535378835316217,
      "grad_norm": 1.4924392700195312,
      "learning_rate": 0.000256891495601173,
      "loss": 0.6074,
      "step": 2270
    },
    {
      "epoch": 0.3555103318722605,
      "grad_norm": 1.610140323638916,
      "learning_rate": 0.0002568670576735093,
      "loss": 0.9784,
      "step": 2271
    },
    {
      "epoch": 0.35566687539135877,
      "grad_norm": 3.816253662109375,
      "learning_rate": 0.00025684261974584554,
      "loss": 0.738,
      "step": 2272
    },
    {
      "epoch": 0.3558234189104571,
      "grad_norm": 2.6387505531311035,
      "learning_rate": 0.0002568181818181818,
      "loss": 0.6391,
      "step": 2273
    },
    {
      "epoch": 0.35597996242955543,
      "grad_norm": 2.228098154067993,
      "learning_rate": 0.0002567937438905181,
      "loss": 1.1959,
      "step": 2274
    },
    {
      "epoch": 0.3561365059486537,
      "grad_norm": 2.4509754180908203,
      "learning_rate": 0.00025676930596285434,
      "loss": 0.9127,
      "step": 2275
    },
    {
      "epoch": 0.35629304946775203,
      "grad_norm": 2.0629775524139404,
      "learning_rate": 0.0002567448680351906,
      "loss": 1.3263,
      "step": 2276
    },
    {
      "epoch": 0.35644959298685036,
      "grad_norm": 1.7958539724349976,
      "learning_rate": 0.0002567204301075269,
      "loss": 1.176,
      "step": 2277
    },
    {
      "epoch": 0.35660613650594863,
      "grad_norm": 1.8170077800750732,
      "learning_rate": 0.0002566959921798631,
      "loss": 0.7305,
      "step": 2278
    },
    {
      "epoch": 0.35676268002504696,
      "grad_norm": 2.358222007751465,
      "learning_rate": 0.0002566715542521994,
      "loss": 1.1738,
      "step": 2279
    },
    {
      "epoch": 0.3569192235441453,
      "grad_norm": 1.7362068891525269,
      "learning_rate": 0.00025664711632453565,
      "loss": 1.1321,
      "step": 2280
    },
    {
      "epoch": 0.35707576706324357,
      "grad_norm": 1.7529265880584717,
      "learning_rate": 0.0002566226783968719,
      "loss": 1.4302,
      "step": 2281
    },
    {
      "epoch": 0.3572323105823419,
      "grad_norm": 2.1040377616882324,
      "learning_rate": 0.0002565982404692082,
      "loss": 0.9679,
      "step": 2282
    },
    {
      "epoch": 0.3573888541014402,
      "grad_norm": 3.2066938877105713,
      "learning_rate": 0.00025657380254154446,
      "loss": 0.93,
      "step": 2283
    },
    {
      "epoch": 0.3575453976205385,
      "grad_norm": 1.855434536933899,
      "learning_rate": 0.0002565493646138807,
      "loss": 1.1485,
      "step": 2284
    },
    {
      "epoch": 0.3577019411396368,
      "grad_norm": 1.5777641534805298,
      "learning_rate": 0.00025652492668621696,
      "loss": 1.2011,
      "step": 2285
    },
    {
      "epoch": 0.35785848465873515,
      "grad_norm": 2.2551398277282715,
      "learning_rate": 0.00025650048875855327,
      "loss": 1.5703,
      "step": 2286
    },
    {
      "epoch": 0.3580150281778334,
      "grad_norm": 2.300269365310669,
      "learning_rate": 0.0002564760508308895,
      "loss": 1.4385,
      "step": 2287
    },
    {
      "epoch": 0.35817157169693176,
      "grad_norm": 1.81515634059906,
      "learning_rate": 0.00025645161290322577,
      "loss": 1.1124,
      "step": 2288
    },
    {
      "epoch": 0.35832811521603003,
      "grad_norm": 1.9418299198150635,
      "learning_rate": 0.0002564271749755621,
      "loss": 1.3713,
      "step": 2289
    },
    {
      "epoch": 0.35848465873512836,
      "grad_norm": 2.397949695587158,
      "learning_rate": 0.00025640273704789833,
      "loss": 1.1791,
      "step": 2290
    },
    {
      "epoch": 0.3586412022542267,
      "grad_norm": 2.4161155223846436,
      "learning_rate": 0.0002563782991202346,
      "loss": 1.7395,
      "step": 2291
    },
    {
      "epoch": 0.35879774577332496,
      "grad_norm": 2.1113173961639404,
      "learning_rate": 0.0002563538611925709,
      "loss": 0.9496,
      "step": 2292
    },
    {
      "epoch": 0.3589542892924233,
      "grad_norm": 1.8949174880981445,
      "learning_rate": 0.0002563294232649071,
      "loss": 1.4005,
      "step": 2293
    },
    {
      "epoch": 0.3591108328115216,
      "grad_norm": 3.1635854244232178,
      "learning_rate": 0.0002563049853372434,
      "loss": 1.3552,
      "step": 2294
    },
    {
      "epoch": 0.3592673763306199,
      "grad_norm": 1.8511496782302856,
      "learning_rate": 0.00025628054740957964,
      "loss": 1.6418,
      "step": 2295
    },
    {
      "epoch": 0.3594239198497182,
      "grad_norm": 2.0520694255828857,
      "learning_rate": 0.0002562561094819159,
      "loss": 0.8151,
      "step": 2296
    },
    {
      "epoch": 0.35958046336881655,
      "grad_norm": 2.5839831829071045,
      "learning_rate": 0.0002562316715542522,
      "loss": 0.7499,
      "step": 2297
    },
    {
      "epoch": 0.3597370068879148,
      "grad_norm": 1.368067979812622,
      "learning_rate": 0.00025620723362658845,
      "loss": 0.8066,
      "step": 2298
    },
    {
      "epoch": 0.35989355040701315,
      "grad_norm": 2.371605634689331,
      "learning_rate": 0.0002561827956989247,
      "loss": 1.376,
      "step": 2299
    },
    {
      "epoch": 0.3600500939261115,
      "grad_norm": 2.2831456661224365,
      "learning_rate": 0.000256158357771261,
      "loss": 1.2232,
      "step": 2300
    },
    {
      "epoch": 0.36020663744520975,
      "grad_norm": 0.44485318660736084,
      "learning_rate": 0.0002561339198435972,
      "loss": 0.3244,
      "step": 2301
    },
    {
      "epoch": 0.3603631809643081,
      "grad_norm": 0.7267107367515564,
      "learning_rate": 0.0002561094819159335,
      "loss": 0.5188,
      "step": 2302
    },
    {
      "epoch": 0.3605197244834064,
      "grad_norm": 0.7034205794334412,
      "learning_rate": 0.00025608504398826976,
      "loss": 0.404,
      "step": 2303
    },
    {
      "epoch": 0.3606762680025047,
      "grad_norm": 0.8780544996261597,
      "learning_rate": 0.000256060606060606,
      "loss": 0.4213,
      "step": 2304
    },
    {
      "epoch": 0.360832811521603,
      "grad_norm": 0.5896530747413635,
      "learning_rate": 0.0002560361681329423,
      "loss": 0.387,
      "step": 2305
    },
    {
      "epoch": 0.3609893550407013,
      "grad_norm": 0.7313452959060669,
      "learning_rate": 0.00025601173020527857,
      "loss": 0.3646,
      "step": 2306
    },
    {
      "epoch": 0.3611458985597996,
      "grad_norm": 0.7097526788711548,
      "learning_rate": 0.0002559872922776148,
      "loss": 0.4382,
      "step": 2307
    },
    {
      "epoch": 0.36130244207889795,
      "grad_norm": 0.685795247554779,
      "learning_rate": 0.0002559628543499511,
      "loss": 0.4068,
      "step": 2308
    },
    {
      "epoch": 0.3614589855979962,
      "grad_norm": 0.6047770380973816,
      "learning_rate": 0.0002559384164222874,
      "loss": 0.4213,
      "step": 2309
    },
    {
      "epoch": 0.36161552911709455,
      "grad_norm": 13.313465118408203,
      "learning_rate": 0.0002559139784946236,
      "loss": 3.0132,
      "step": 2310
    },
    {
      "epoch": 0.3617720726361929,
      "grad_norm": 1.098598599433899,
      "learning_rate": 0.0002558895405669599,
      "loss": 0.396,
      "step": 2311
    },
    {
      "epoch": 0.36192861615529115,
      "grad_norm": 0.8045850396156311,
      "learning_rate": 0.0002558651026392962,
      "loss": 0.4162,
      "step": 2312
    },
    {
      "epoch": 0.3620851596743895,
      "grad_norm": 1.3596242666244507,
      "learning_rate": 0.00025584066471163243,
      "loss": 0.7236,
      "step": 2313
    },
    {
      "epoch": 0.3622417031934878,
      "grad_norm": 1.0989056825637817,
      "learning_rate": 0.0002558162267839687,
      "loss": 0.4784,
      "step": 2314
    },
    {
      "epoch": 0.3623982467125861,
      "grad_norm": 5.293839931488037,
      "learning_rate": 0.000255791788856305,
      "loss": 0.9534,
      "step": 2315
    },
    {
      "epoch": 0.3625547902316844,
      "grad_norm": 0.8783222436904907,
      "learning_rate": 0.0002557673509286412,
      "loss": 0.7846,
      "step": 2316
    },
    {
      "epoch": 0.36271133375078274,
      "grad_norm": 1.2049531936645508,
      "learning_rate": 0.0002557429130009775,
      "loss": 0.5091,
      "step": 2317
    },
    {
      "epoch": 0.362867877269881,
      "grad_norm": 1.0293569564819336,
      "learning_rate": 0.00025571847507331374,
      "loss": 0.5985,
      "step": 2318
    },
    {
      "epoch": 0.36302442078897934,
      "grad_norm": 1.8985306024551392,
      "learning_rate": 0.00025569403714565,
      "loss": 0.7279,
      "step": 2319
    },
    {
      "epoch": 0.36318096430807767,
      "grad_norm": 1.2464323043823242,
      "learning_rate": 0.0002556695992179863,
      "loss": 0.6954,
      "step": 2320
    },
    {
      "epoch": 0.36333750782717594,
      "grad_norm": 0.8314011096954346,
      "learning_rate": 0.00025564516129032255,
      "loss": 0.461,
      "step": 2321
    },
    {
      "epoch": 0.36349405134627427,
      "grad_norm": 1.2355091571807861,
      "learning_rate": 0.0002556207233626588,
      "loss": 0.719,
      "step": 2322
    },
    {
      "epoch": 0.3636505948653726,
      "grad_norm": 1.9257937669754028,
      "learning_rate": 0.0002555962854349951,
      "loss": 0.5286,
      "step": 2323
    },
    {
      "epoch": 0.3638071383844709,
      "grad_norm": 1.4777895212173462,
      "learning_rate": 0.00025557184750733136,
      "loss": 0.709,
      "step": 2324
    },
    {
      "epoch": 0.3639636819035692,
      "grad_norm": 2.033358573913574,
      "learning_rate": 0.0002555474095796676,
      "loss": 0.936,
      "step": 2325
    },
    {
      "epoch": 0.3641202254226675,
      "grad_norm": 2.625703811645508,
      "learning_rate": 0.00025552297165200386,
      "loss": 1.2012,
      "step": 2326
    },
    {
      "epoch": 0.3642767689417658,
      "grad_norm": 2.080801486968994,
      "learning_rate": 0.00025549853372434017,
      "loss": 0.7433,
      "step": 2327
    },
    {
      "epoch": 0.36443331246086413,
      "grad_norm": 1.9644887447357178,
      "learning_rate": 0.0002554740957966764,
      "loss": 0.8065,
      "step": 2328
    },
    {
      "epoch": 0.3645898559799624,
      "grad_norm": 3.320422410964966,
      "learning_rate": 0.00025544965786901267,
      "loss": 1.1196,
      "step": 2329
    },
    {
      "epoch": 0.36474639949906074,
      "grad_norm": 1.20960533618927,
      "learning_rate": 0.000255425219941349,
      "loss": 0.8819,
      "step": 2330
    },
    {
      "epoch": 0.36490294301815906,
      "grad_norm": 1.6307368278503418,
      "learning_rate": 0.0002554007820136852,
      "loss": 0.5572,
      "step": 2331
    },
    {
      "epoch": 0.36505948653725734,
      "grad_norm": 1.9175331592559814,
      "learning_rate": 0.0002553763440860215,
      "loss": 1.4483,
      "step": 2332
    },
    {
      "epoch": 0.36521603005635567,
      "grad_norm": 3.3411760330200195,
      "learning_rate": 0.00025535190615835773,
      "loss": 1.2187,
      "step": 2333
    },
    {
      "epoch": 0.365372573575454,
      "grad_norm": 2.1116483211517334,
      "learning_rate": 0.000255327468230694,
      "loss": 1.3473,
      "step": 2334
    },
    {
      "epoch": 0.36552911709455227,
      "grad_norm": 3.1497905254364014,
      "learning_rate": 0.0002553030303030303,
      "loss": 0.8682,
      "step": 2335
    },
    {
      "epoch": 0.3656856606136506,
      "grad_norm": 2.2446794509887695,
      "learning_rate": 0.00025527859237536654,
      "loss": 1.1501,
      "step": 2336
    },
    {
      "epoch": 0.3658422041327489,
      "grad_norm": 1.977690577507019,
      "learning_rate": 0.0002552541544477028,
      "loss": 1.2884,
      "step": 2337
    },
    {
      "epoch": 0.3659987476518472,
      "grad_norm": 2.598881721496582,
      "learning_rate": 0.0002552297165200391,
      "loss": 1.3231,
      "step": 2338
    },
    {
      "epoch": 0.36615529117094553,
      "grad_norm": 2.4923884868621826,
      "learning_rate": 0.00025520527859237535,
      "loss": 0.86,
      "step": 2339
    },
    {
      "epoch": 0.36631183469004386,
      "grad_norm": 2.453099250793457,
      "learning_rate": 0.0002551808406647116,
      "loss": 1.0986,
      "step": 2340
    },
    {
      "epoch": 0.36646837820914213,
      "grad_norm": 3.555971384048462,
      "learning_rate": 0.00025515640273704785,
      "loss": 1.8547,
      "step": 2341
    },
    {
      "epoch": 0.36662492172824046,
      "grad_norm": 1.9464694261550903,
      "learning_rate": 0.00025513196480938415,
      "loss": 1.1017,
      "step": 2342
    },
    {
      "epoch": 0.36678146524733873,
      "grad_norm": 3.1742851734161377,
      "learning_rate": 0.0002551075268817204,
      "loss": 1.7284,
      "step": 2343
    },
    {
      "epoch": 0.36693800876643706,
      "grad_norm": 1.5996363162994385,
      "learning_rate": 0.00025508308895405666,
      "loss": 1.1764,
      "step": 2344
    },
    {
      "epoch": 0.3670945522855354,
      "grad_norm": 3.3336408138275146,
      "learning_rate": 0.00025505865102639296,
      "loss": 1.5985,
      "step": 2345
    },
    {
      "epoch": 0.36725109580463366,
      "grad_norm": 1.3342739343643188,
      "learning_rate": 0.0002550342130987292,
      "loss": 1.2361,
      "step": 2346
    },
    {
      "epoch": 0.367407639323732,
      "grad_norm": 2.0389742851257324,
      "learning_rate": 0.00025500977517106546,
      "loss": 0.6936,
      "step": 2347
    },
    {
      "epoch": 0.3675641828428303,
      "grad_norm": 1.7741936445236206,
      "learning_rate": 0.00025498533724340177,
      "loss": 0.6738,
      "step": 2348
    },
    {
      "epoch": 0.3677207263619286,
      "grad_norm": 2.689279079437256,
      "learning_rate": 0.00025496089931573797,
      "loss": 1.4255,
      "step": 2349
    },
    {
      "epoch": 0.3678772698810269,
      "grad_norm": 1.4633771181106567,
      "learning_rate": 0.00025493646138807427,
      "loss": 0.8265,
      "step": 2350
    },
    {
      "epoch": 0.36803381340012525,
      "grad_norm": 0.6005833148956299,
      "learning_rate": 0.0002549120234604105,
      "loss": 0.3965,
      "step": 2351
    },
    {
      "epoch": 0.3681903569192235,
      "grad_norm": 0.5253506898880005,
      "learning_rate": 0.0002548875855327468,
      "loss": 0.4102,
      "step": 2352
    },
    {
      "epoch": 0.36834690043832186,
      "grad_norm": 0.582460880279541,
      "learning_rate": 0.0002548631476050831,
      "loss": 0.2965,
      "step": 2353
    },
    {
      "epoch": 0.3685034439574202,
      "grad_norm": 0.8253412246704102,
      "learning_rate": 0.00025483870967741933,
      "loss": 0.3247,
      "step": 2354
    },
    {
      "epoch": 0.36865998747651846,
      "grad_norm": 0.6425923705101013,
      "learning_rate": 0.0002548142717497556,
      "loss": 0.4133,
      "step": 2355
    },
    {
      "epoch": 0.3688165309956168,
      "grad_norm": 1.085408091545105,
      "learning_rate": 0.00025478983382209183,
      "loss": 0.4947,
      "step": 2356
    },
    {
      "epoch": 0.3689730745147151,
      "grad_norm": 1.145398497581482,
      "learning_rate": 0.00025476539589442814,
      "loss": 0.6325,
      "step": 2357
    },
    {
      "epoch": 0.3691296180338134,
      "grad_norm": 0.7742732167243958,
      "learning_rate": 0.0002547409579667644,
      "loss": 0.4053,
      "step": 2358
    },
    {
      "epoch": 0.3692861615529117,
      "grad_norm": 1.4050893783569336,
      "learning_rate": 0.00025471652003910064,
      "loss": 0.7099,
      "step": 2359
    },
    {
      "epoch": 0.36944270507201,
      "grad_norm": 0.5724520683288574,
      "learning_rate": 0.00025469208211143695,
      "loss": 0.4195,
      "step": 2360
    },
    {
      "epoch": 0.3695992485911083,
      "grad_norm": 1.2148267030715942,
      "learning_rate": 0.0002546676441837732,
      "loss": 0.6613,
      "step": 2361
    },
    {
      "epoch": 0.36975579211020665,
      "grad_norm": 1.6140097379684448,
      "learning_rate": 0.00025464320625610945,
      "loss": 0.5106,
      "step": 2362
    },
    {
      "epoch": 0.3699123356293049,
      "grad_norm": 1.0580832958221436,
      "learning_rate": 0.00025461876832844575,
      "loss": 0.6509,
      "step": 2363
    },
    {
      "epoch": 0.37006887914840325,
      "grad_norm": 1.0968374013900757,
      "learning_rate": 0.00025459433040078195,
      "loss": 0.4407,
      "step": 2364
    },
    {
      "epoch": 0.3702254226675016,
      "grad_norm": 1.8869554996490479,
      "learning_rate": 0.00025456989247311826,
      "loss": 0.8746,
      "step": 2365
    },
    {
      "epoch": 0.37038196618659985,
      "grad_norm": 1.1201070547103882,
      "learning_rate": 0.0002545454545454545,
      "loss": 0.4115,
      "step": 2366
    },
    {
      "epoch": 0.3705385097056982,
      "grad_norm": 1.391060709953308,
      "learning_rate": 0.00025452101661779076,
      "loss": 0.8105,
      "step": 2367
    },
    {
      "epoch": 0.3706950532247965,
      "grad_norm": 1.707555890083313,
      "learning_rate": 0.00025449657869012707,
      "loss": 0.5989,
      "step": 2368
    },
    {
      "epoch": 0.3708515967438948,
      "grad_norm": 1.1574084758758545,
      "learning_rate": 0.0002544721407624633,
      "loss": 0.7763,
      "step": 2369
    },
    {
      "epoch": 0.3710081402629931,
      "grad_norm": 4.23545503616333,
      "learning_rate": 0.00025444770283479957,
      "loss": 1.369,
      "step": 2370
    },
    {
      "epoch": 0.37116468378209144,
      "grad_norm": 1.9482706785202026,
      "learning_rate": 0.0002544232649071359,
      "loss": 0.5819,
      "step": 2371
    },
    {
      "epoch": 0.3713212273011897,
      "grad_norm": 1.2901068925857544,
      "learning_rate": 0.0002543988269794721,
      "loss": 1.0075,
      "step": 2372
    },
    {
      "epoch": 0.37147777082028804,
      "grad_norm": 1.395151138305664,
      "learning_rate": 0.0002543743890518084,
      "loss": 0.5711,
      "step": 2373
    },
    {
      "epoch": 0.3716343143393864,
      "grad_norm": 1.3401544094085693,
      "learning_rate": 0.0002543499511241446,
      "loss": 0.7202,
      "step": 2374
    },
    {
      "epoch": 0.37179085785848465,
      "grad_norm": 0.9693850874900818,
      "learning_rate": 0.00025432551319648093,
      "loss": 0.7058,
      "step": 2375
    },
    {
      "epoch": 0.371947401377583,
      "grad_norm": 1.058971881866455,
      "learning_rate": 0.0002543010752688172,
      "loss": 0.6258,
      "step": 2376
    },
    {
      "epoch": 0.3721039448966813,
      "grad_norm": 2.7098138332366943,
      "learning_rate": 0.00025427663734115343,
      "loss": 0.741,
      "step": 2377
    },
    {
      "epoch": 0.3722604884157796,
      "grad_norm": 1.293344259262085,
      "learning_rate": 0.00025425219941348974,
      "loss": 0.7096,
      "step": 2378
    },
    {
      "epoch": 0.3724170319348779,
      "grad_norm": 6.409755706787109,
      "learning_rate": 0.000254227761485826,
      "loss": 0.9657,
      "step": 2379
    },
    {
      "epoch": 0.3725735754539762,
      "grad_norm": 3.3196206092834473,
      "learning_rate": 0.00025420332355816224,
      "loss": 0.8149,
      "step": 2380
    },
    {
      "epoch": 0.3727301189730745,
      "grad_norm": 1.7499516010284424,
      "learning_rate": 0.0002541788856304985,
      "loss": 1.0022,
      "step": 2381
    },
    {
      "epoch": 0.37288666249217284,
      "grad_norm": 2.6857728958129883,
      "learning_rate": 0.00025415444770283475,
      "loss": 0.8736,
      "step": 2382
    },
    {
      "epoch": 0.3730432060112711,
      "grad_norm": 2.264000415802002,
      "learning_rate": 0.00025413000977517105,
      "loss": 0.6507,
      "step": 2383
    },
    {
      "epoch": 0.37319974953036944,
      "grad_norm": 2.5130653381347656,
      "learning_rate": 0.0002541055718475073,
      "loss": 0.8562,
      "step": 2384
    },
    {
      "epoch": 0.37335629304946777,
      "grad_norm": 6.680568218231201,
      "learning_rate": 0.00025408113391984355,
      "loss": 1.6511,
      "step": 2385
    },
    {
      "epoch": 0.37351283656856604,
      "grad_norm": 3.2456836700439453,
      "learning_rate": 0.00025405669599217986,
      "loss": 1.1887,
      "step": 2386
    },
    {
      "epoch": 0.37366938008766437,
      "grad_norm": 2.358682632446289,
      "learning_rate": 0.0002540322580645161,
      "loss": 1.1949,
      "step": 2387
    },
    {
      "epoch": 0.3738259236067627,
      "grad_norm": 1.589746117591858,
      "learning_rate": 0.00025400782013685236,
      "loss": 1.248,
      "step": 2388
    },
    {
      "epoch": 0.373982467125861,
      "grad_norm": 3.969412326812744,
      "learning_rate": 0.0002539833822091886,
      "loss": 1.4367,
      "step": 2389
    },
    {
      "epoch": 0.3741390106449593,
      "grad_norm": 1.8342620134353638,
      "learning_rate": 0.0002539589442815249,
      "loss": 0.6383,
      "step": 2390
    },
    {
      "epoch": 0.37429555416405763,
      "grad_norm": 5.95599365234375,
      "learning_rate": 0.00025393450635386117,
      "loss": 1.3961,
      "step": 2391
    },
    {
      "epoch": 0.3744520976831559,
      "grad_norm": 2.8930656909942627,
      "learning_rate": 0.0002539100684261974,
      "loss": 1.5589,
      "step": 2392
    },
    {
      "epoch": 0.37460864120225423,
      "grad_norm": 3.1002862453460693,
      "learning_rate": 0.0002538856304985337,
      "loss": 1.4962,
      "step": 2393
    },
    {
      "epoch": 0.37476518472135256,
      "grad_norm": 4.003942489624023,
      "learning_rate": 0.00025386119257087,
      "loss": 1.2726,
      "step": 2394
    },
    {
      "epoch": 0.37492172824045084,
      "grad_norm": 3.102221965789795,
      "learning_rate": 0.00025383675464320623,
      "loss": 1.2946,
      "step": 2395
    },
    {
      "epoch": 0.37507827175954916,
      "grad_norm": 2.444598913192749,
      "learning_rate": 0.00025381231671554253,
      "loss": 1.1978,
      "step": 2396
    },
    {
      "epoch": 0.37523481527864744,
      "grad_norm": 2.2664988040924072,
      "learning_rate": 0.00025378787878787873,
      "loss": 0.7367,
      "step": 2397
    },
    {
      "epoch": 0.37539135879774577,
      "grad_norm": 1.9962775707244873,
      "learning_rate": 0.00025376344086021504,
      "loss": 0.4199,
      "step": 2398
    },
    {
      "epoch": 0.3755479023168441,
      "grad_norm": 2.8011834621429443,
      "learning_rate": 0.0002537390029325513,
      "loss": 1.2034,
      "step": 2399
    },
    {
      "epoch": 0.37570444583594237,
      "grad_norm": 1.9701213836669922,
      "learning_rate": 0.00025371456500488754,
      "loss": 1.0624,
      "step": 2400
    },
    {
      "epoch": 0.3758609893550407,
      "grad_norm": 0.7180734276771545,
      "learning_rate": 0.00025369012707722384,
      "loss": 0.3932,
      "step": 2401
    },
    {
      "epoch": 0.376017532874139,
      "grad_norm": 0.5675762295722961,
      "learning_rate": 0.0002536656891495601,
      "loss": 0.3901,
      "step": 2402
    },
    {
      "epoch": 0.3761740763932373,
      "grad_norm": 0.6413562893867493,
      "learning_rate": 0.00025364125122189635,
      "loss": 0.4672,
      "step": 2403
    },
    {
      "epoch": 0.37633061991233563,
      "grad_norm": 0.9517688155174255,
      "learning_rate": 0.0002536168132942326,
      "loss": 0.6919,
      "step": 2404
    },
    {
      "epoch": 0.37648716343143396,
      "grad_norm": 0.6282806396484375,
      "learning_rate": 0.0002535923753665689,
      "loss": 0.451,
      "step": 2405
    },
    {
      "epoch": 0.37664370695053223,
      "grad_norm": 0.590168833732605,
      "learning_rate": 0.00025356793743890515,
      "loss": 0.4078,
      "step": 2406
    },
    {
      "epoch": 0.37680025046963056,
      "grad_norm": 0.7684517502784729,
      "learning_rate": 0.0002535434995112414,
      "loss": 0.4757,
      "step": 2407
    },
    {
      "epoch": 0.3769567939887289,
      "grad_norm": 0.9231551289558411,
      "learning_rate": 0.0002535190615835777,
      "loss": 0.6356,
      "step": 2408
    },
    {
      "epoch": 0.37711333750782716,
      "grad_norm": 0.8362495303153992,
      "learning_rate": 0.00025349462365591396,
      "loss": 0.4089,
      "step": 2409
    },
    {
      "epoch": 0.3772698810269255,
      "grad_norm": 0.8577970862388611,
      "learning_rate": 0.0002534701857282502,
      "loss": 0.4651,
      "step": 2410
    },
    {
      "epoch": 0.3774264245460238,
      "grad_norm": 1.4198194742202759,
      "learning_rate": 0.0002534457478005865,
      "loss": 0.6993,
      "step": 2411
    },
    {
      "epoch": 0.3775829680651221,
      "grad_norm": 1.1189966201782227,
      "learning_rate": 0.0002534213098729227,
      "loss": 0.7051,
      "step": 2412
    },
    {
      "epoch": 0.3777395115842204,
      "grad_norm": 0.8378294706344604,
      "learning_rate": 0.000253396871945259,
      "loss": 0.4614,
      "step": 2413
    },
    {
      "epoch": 0.3778960551033187,
      "grad_norm": 0.8094697594642639,
      "learning_rate": 0.0002533724340175953,
      "loss": 0.4466,
      "step": 2414
    },
    {
      "epoch": 0.378052598622417,
      "grad_norm": 1.419098138809204,
      "learning_rate": 0.0002533479960899315,
      "loss": 0.7824,
      "step": 2415
    },
    {
      "epoch": 0.37820914214151535,
      "grad_norm": 2.8481099605560303,
      "learning_rate": 0.00025332355816226783,
      "loss": 0.643,
      "step": 2416
    },
    {
      "epoch": 0.3783656856606136,
      "grad_norm": 1.6152693033218384,
      "learning_rate": 0.0002532991202346041,
      "loss": 0.6802,
      "step": 2417
    },
    {
      "epoch": 0.37852222917971196,
      "grad_norm": 1.3891299962997437,
      "learning_rate": 0.00025327468230694033,
      "loss": 0.9134,
      "step": 2418
    },
    {
      "epoch": 0.3786787726988103,
      "grad_norm": 0.8565380573272705,
      "learning_rate": 0.00025325024437927664,
      "loss": 0.5214,
      "step": 2419
    },
    {
      "epoch": 0.37883531621790856,
      "grad_norm": 1.4971650838851929,
      "learning_rate": 0.0002532258064516129,
      "loss": 0.6881,
      "step": 2420
    },
    {
      "epoch": 0.3789918597370069,
      "grad_norm": 2.303706169128418,
      "learning_rate": 0.00025320136852394914,
      "loss": 0.9213,
      "step": 2421
    },
    {
      "epoch": 0.3791484032561052,
      "grad_norm": 1.7516024112701416,
      "learning_rate": 0.0002531769305962854,
      "loss": 0.7242,
      "step": 2422
    },
    {
      "epoch": 0.3793049467752035,
      "grad_norm": 1.990359902381897,
      "learning_rate": 0.0002531524926686217,
      "loss": 0.829,
      "step": 2423
    },
    {
      "epoch": 0.3794614902943018,
      "grad_norm": 3.0393333435058594,
      "learning_rate": 0.00025312805474095795,
      "loss": 1.1958,
      "step": 2424
    },
    {
      "epoch": 0.37961803381340015,
      "grad_norm": 1.408226728439331,
      "learning_rate": 0.0002531036168132942,
      "loss": 0.7958,
      "step": 2425
    },
    {
      "epoch": 0.3797745773324984,
      "grad_norm": 4.30161190032959,
      "learning_rate": 0.0002530791788856305,
      "loss": 1.0334,
      "step": 2426
    },
    {
      "epoch": 0.37993112085159675,
      "grad_norm": 1.420586109161377,
      "learning_rate": 0.0002530547409579667,
      "loss": 0.8091,
      "step": 2427
    },
    {
      "epoch": 0.3800876643706951,
      "grad_norm": 1.1459065675735474,
      "learning_rate": 0.000253030303030303,
      "loss": 0.5026,
      "step": 2428
    },
    {
      "epoch": 0.38024420788979335,
      "grad_norm": 1.717860460281372,
      "learning_rate": 0.00025300586510263926,
      "loss": 0.8863,
      "step": 2429
    },
    {
      "epoch": 0.3804007514088917,
      "grad_norm": 1.8267467021942139,
      "learning_rate": 0.0002529814271749755,
      "loss": 0.926,
      "step": 2430
    },
    {
      "epoch": 0.38055729492799,
      "grad_norm": 1.825463056564331,
      "learning_rate": 0.0002529569892473118,
      "loss": 1.062,
      "step": 2431
    },
    {
      "epoch": 0.3807138384470883,
      "grad_norm": 4.6027140617370605,
      "learning_rate": 0.00025293255131964807,
      "loss": 1.0772,
      "step": 2432
    },
    {
      "epoch": 0.3808703819661866,
      "grad_norm": 1.6209831237792969,
      "learning_rate": 0.0002529081133919843,
      "loss": 1.2083,
      "step": 2433
    },
    {
      "epoch": 0.3810269254852849,
      "grad_norm": 1.9430205821990967,
      "learning_rate": 0.0002528836754643206,
      "loss": 0.8559,
      "step": 2434
    },
    {
      "epoch": 0.3811834690043832,
      "grad_norm": 5.019834041595459,
      "learning_rate": 0.0002528592375366569,
      "loss": 0.9144,
      "step": 2435
    },
    {
      "epoch": 0.38134001252348154,
      "grad_norm": 1.813771367073059,
      "learning_rate": 0.0002528347996089931,
      "loss": 0.5865,
      "step": 2436
    },
    {
      "epoch": 0.3814965560425798,
      "grad_norm": 6.178098201751709,
      "learning_rate": 0.0002528103616813294,
      "loss": 1.0331,
      "step": 2437
    },
    {
      "epoch": 0.38165309956167814,
      "grad_norm": 2.6770286560058594,
      "learning_rate": 0.0002527859237536657,
      "loss": 1.4776,
      "step": 2438
    },
    {
      "epoch": 0.3818096430807765,
      "grad_norm": 2.0814406871795654,
      "learning_rate": 0.00025276148582600193,
      "loss": 1.2574,
      "step": 2439
    },
    {
      "epoch": 0.38196618659987475,
      "grad_norm": 2.1399121284484863,
      "learning_rate": 0.0002527370478983382,
      "loss": 1.2914,
      "step": 2440
    },
    {
      "epoch": 0.3821227301189731,
      "grad_norm": 2.3518238067626953,
      "learning_rate": 0.0002527126099706745,
      "loss": 1.0346,
      "step": 2441
    },
    {
      "epoch": 0.3822792736380714,
      "grad_norm": 2.9069864749908447,
      "learning_rate": 0.00025268817204301074,
      "loss": 1.0897,
      "step": 2442
    },
    {
      "epoch": 0.3824358171571697,
      "grad_norm": 10.248844146728516,
      "learning_rate": 0.000252663734115347,
      "loss": 1.375,
      "step": 2443
    },
    {
      "epoch": 0.382592360676268,
      "grad_norm": 2.556163787841797,
      "learning_rate": 0.00025263929618768324,
      "loss": 1.5689,
      "step": 2444
    },
    {
      "epoch": 0.38274890419536634,
      "grad_norm": 6.150379657745361,
      "learning_rate": 0.0002526148582600195,
      "loss": 1.5573,
      "step": 2445
    },
    {
      "epoch": 0.3829054477144646,
      "grad_norm": 2.1215577125549316,
      "learning_rate": 0.0002525904203323558,
      "loss": 0.7524,
      "step": 2446
    },
    {
      "epoch": 0.38306199123356294,
      "grad_norm": 1.778437852859497,
      "learning_rate": 0.00025256598240469205,
      "loss": 1.0496,
      "step": 2447
    },
    {
      "epoch": 0.38321853475266127,
      "grad_norm": 1.631048560142517,
      "learning_rate": 0.0002525415444770283,
      "loss": 0.676,
      "step": 2448
    },
    {
      "epoch": 0.38337507827175954,
      "grad_norm": 2.2763094902038574,
      "learning_rate": 0.0002525171065493646,
      "loss": 1.1262,
      "step": 2449
    },
    {
      "epoch": 0.38353162179085787,
      "grad_norm": 1.9400389194488525,
      "learning_rate": 0.00025249266862170086,
      "loss": 1.5447,
      "step": 2450
    },
    {
      "epoch": 0.38368816530995614,
      "grad_norm": 0.6042339205741882,
      "learning_rate": 0.0002524682306940371,
      "loss": 0.4302,
      "step": 2451
    },
    {
      "epoch": 0.38384470882905447,
      "grad_norm": 0.6273007392883301,
      "learning_rate": 0.00025244379276637336,
      "loss": 0.4748,
      "step": 2452
    },
    {
      "epoch": 0.3840012523481528,
      "grad_norm": 0.5941194891929626,
      "learning_rate": 0.00025241935483870967,
      "loss": 0.3699,
      "step": 2453
    },
    {
      "epoch": 0.3841577958672511,
      "grad_norm": 0.6583676338195801,
      "learning_rate": 0.0002523949169110459,
      "loss": 0.5617,
      "step": 2454
    },
    {
      "epoch": 0.3843143393863494,
      "grad_norm": 0.9693568348884583,
      "learning_rate": 0.00025237047898338217,
      "loss": 0.4727,
      "step": 2455
    },
    {
      "epoch": 0.38447088290544773,
      "grad_norm": 0.7565197944641113,
      "learning_rate": 0.0002523460410557185,
      "loss": 0.4105,
      "step": 2456
    },
    {
      "epoch": 0.384627426424546,
      "grad_norm": 0.9457018375396729,
      "learning_rate": 0.00025232160312805473,
      "loss": 0.6106,
      "step": 2457
    },
    {
      "epoch": 0.38478396994364433,
      "grad_norm": 0.8594042062759399,
      "learning_rate": 0.000252297165200391,
      "loss": 0.4825,
      "step": 2458
    },
    {
      "epoch": 0.38494051346274266,
      "grad_norm": 0.9214059114456177,
      "learning_rate": 0.0002522727272727273,
      "loss": 0.4283,
      "step": 2459
    },
    {
      "epoch": 0.38509705698184094,
      "grad_norm": 1.0324896574020386,
      "learning_rate": 0.0002522482893450635,
      "loss": 0.4448,
      "step": 2460
    },
    {
      "epoch": 0.38525360050093926,
      "grad_norm": 0.8373879194259644,
      "learning_rate": 0.0002522238514173998,
      "loss": 0.4815,
      "step": 2461
    },
    {
      "epoch": 0.3854101440200376,
      "grad_norm": 0.5907357931137085,
      "learning_rate": 0.00025219941348973604,
      "loss": 0.3629,
      "step": 2462
    },
    {
      "epoch": 0.38556668753913587,
      "grad_norm": 0.8201125860214233,
      "learning_rate": 0.0002521749755620723,
      "loss": 0.4437,
      "step": 2463
    },
    {
      "epoch": 0.3857232310582342,
      "grad_norm": 1.6409950256347656,
      "learning_rate": 0.0002521505376344086,
      "loss": 0.6879,
      "step": 2464
    },
    {
      "epoch": 0.3858797745773325,
      "grad_norm": 1.9288908243179321,
      "learning_rate": 0.00025212609970674485,
      "loss": 0.8157,
      "step": 2465
    },
    {
      "epoch": 0.3860363180964308,
      "grad_norm": 1.1502376794815063,
      "learning_rate": 0.0002521016617790811,
      "loss": 0.746,
      "step": 2466
    },
    {
      "epoch": 0.3861928616155291,
      "grad_norm": 1.607825756072998,
      "learning_rate": 0.0002520772238514174,
      "loss": 0.4716,
      "step": 2467
    },
    {
      "epoch": 0.3863494051346274,
      "grad_norm": 1.3579707145690918,
      "learning_rate": 0.00025205278592375365,
      "loss": 0.486,
      "step": 2468
    },
    {
      "epoch": 0.38650594865372573,
      "grad_norm": 1.369374394416809,
      "learning_rate": 0.0002520283479960899,
      "loss": 0.6688,
      "step": 2469
    },
    {
      "epoch": 0.38666249217282406,
      "grad_norm": 0.9472814798355103,
      "learning_rate": 0.00025200391006842616,
      "loss": 0.4558,
      "step": 2470
    },
    {
      "epoch": 0.38681903569192233,
      "grad_norm": 1.0959384441375732,
      "learning_rate": 0.00025197947214076246,
      "loss": 0.7678,
      "step": 2471
    },
    {
      "epoch": 0.38697557921102066,
      "grad_norm": 1.783223032951355,
      "learning_rate": 0.0002519550342130987,
      "loss": 0.8826,
      "step": 2472
    },
    {
      "epoch": 0.387132122730119,
      "grad_norm": 0.7266517281532288,
      "learning_rate": 0.00025193059628543496,
      "loss": 0.3746,
      "step": 2473
    },
    {
      "epoch": 0.38728866624921726,
      "grad_norm": 1.6912201642990112,
      "learning_rate": 0.00025190615835777127,
      "loss": 0.7786,
      "step": 2474
    },
    {
      "epoch": 0.3874452097683156,
      "grad_norm": 2.1097683906555176,
      "learning_rate": 0.00025188172043010747,
      "loss": 1.0013,
      "step": 2475
    },
    {
      "epoch": 0.3876017532874139,
      "grad_norm": 2.697650671005249,
      "learning_rate": 0.00025185728250244377,
      "loss": 0.9238,
      "step": 2476
    },
    {
      "epoch": 0.3877582968065122,
      "grad_norm": 1.2909862995147705,
      "learning_rate": 0.00025183284457478,
      "loss": 0.613,
      "step": 2477
    },
    {
      "epoch": 0.3879148403256105,
      "grad_norm": 2.043966293334961,
      "learning_rate": 0.0002518084066471163,
      "loss": 0.698,
      "step": 2478
    },
    {
      "epoch": 0.38807138384470885,
      "grad_norm": 1.7088145017623901,
      "learning_rate": 0.0002517839687194526,
      "loss": 0.8166,
      "step": 2479
    },
    {
      "epoch": 0.3882279273638071,
      "grad_norm": 1.015830397605896,
      "learning_rate": 0.00025175953079178883,
      "loss": 0.5839,
      "step": 2480
    },
    {
      "epoch": 0.38838447088290545,
      "grad_norm": 1.558098316192627,
      "learning_rate": 0.0002517350928641251,
      "loss": 0.7847,
      "step": 2481
    },
    {
      "epoch": 0.3885410144020038,
      "grad_norm": 1.4762747287750244,
      "learning_rate": 0.0002517106549364614,
      "loss": 0.6532,
      "step": 2482
    },
    {
      "epoch": 0.38869755792110205,
      "grad_norm": 2.2666919231414795,
      "learning_rate": 0.00025168621700879764,
      "loss": 0.9531,
      "step": 2483
    },
    {
      "epoch": 0.3888541014402004,
      "grad_norm": 3.644498586654663,
      "learning_rate": 0.0002516617790811339,
      "loss": 1.3798,
      "step": 2484
    },
    {
      "epoch": 0.3890106449592987,
      "grad_norm": 1.4934632778167725,
      "learning_rate": 0.00025163734115347014,
      "loss": 0.6577,
      "step": 2485
    },
    {
      "epoch": 0.389167188478397,
      "grad_norm": 2.9513094425201416,
      "learning_rate": 0.00025161290322580645,
      "loss": 1.0388,
      "step": 2486
    },
    {
      "epoch": 0.3893237319974953,
      "grad_norm": 2.517242670059204,
      "learning_rate": 0.0002515884652981427,
      "loss": 1.1773,
      "step": 2487
    },
    {
      "epoch": 0.3894802755165936,
      "grad_norm": 2.8088479042053223,
      "learning_rate": 0.00025156402737047895,
      "loss": 1.2082,
      "step": 2488
    },
    {
      "epoch": 0.3896368190356919,
      "grad_norm": 2.6290640830993652,
      "learning_rate": 0.00025153958944281526,
      "loss": 1.6824,
      "step": 2489
    },
    {
      "epoch": 0.38979336255479025,
      "grad_norm": 1.979951024055481,
      "learning_rate": 0.0002515151515151515,
      "loss": 1.1155,
      "step": 2490
    },
    {
      "epoch": 0.3899499060738885,
      "grad_norm": 3.4212875366210938,
      "learning_rate": 0.00025149071358748776,
      "loss": 1.1699,
      "step": 2491
    },
    {
      "epoch": 0.39010644959298685,
      "grad_norm": 2.455173969268799,
      "learning_rate": 0.000251466275659824,
      "loss": 1.516,
      "step": 2492
    },
    {
      "epoch": 0.3902629931120852,
      "grad_norm": 3.4120118618011475,
      "learning_rate": 0.00025144183773216026,
      "loss": 1.3543,
      "step": 2493
    },
    {
      "epoch": 0.39041953663118345,
      "grad_norm": 1.8049226999282837,
      "learning_rate": 0.00025141739980449657,
      "loss": 1.1935,
      "step": 2494
    },
    {
      "epoch": 0.3905760801502818,
      "grad_norm": 2.091107130050659,
      "learning_rate": 0.0002513929618768328,
      "loss": 1.9788,
      "step": 2495
    },
    {
      "epoch": 0.3907326236693801,
      "grad_norm": 1.4167218208312988,
      "learning_rate": 0.00025136852394916907,
      "loss": 0.7783,
      "step": 2496
    },
    {
      "epoch": 0.3908891671884784,
      "grad_norm": 2.366852045059204,
      "learning_rate": 0.0002513440860215054,
      "loss": 0.9097,
      "step": 2497
    },
    {
      "epoch": 0.3910457107075767,
      "grad_norm": 1.7226179838180542,
      "learning_rate": 0.0002513196480938416,
      "loss": 0.9851,
      "step": 2498
    },
    {
      "epoch": 0.39120225422667504,
      "grad_norm": 1.691483974456787,
      "learning_rate": 0.0002512952101661779,
      "loss": 1.281,
      "step": 2499
    },
    {
      "epoch": 0.3913587977457733,
      "grad_norm": 2.1025145053863525,
      "learning_rate": 0.00025127077223851413,
      "loss": 1.495,
      "step": 2500
    },
    {
      "epoch": 0.39151534126487164,
      "grad_norm": 0.49316316843032837,
      "learning_rate": 0.00025124633431085043,
      "loss": 0.4291,
      "step": 2501
    },
    {
      "epoch": 0.39167188478396997,
      "grad_norm": 0.47687578201293945,
      "learning_rate": 0.0002512218963831867,
      "loss": 0.3186,
      "step": 2502
    },
    {
      "epoch": 0.39182842830306824,
      "grad_norm": 0.6680770516395569,
      "learning_rate": 0.00025119745845552294,
      "loss": 0.4185,
      "step": 2503
    },
    {
      "epoch": 0.3919849718221666,
      "grad_norm": 0.6595609188079834,
      "learning_rate": 0.00025117302052785924,
      "loss": 0.3147,
      "step": 2504
    },
    {
      "epoch": 0.39214151534126485,
      "grad_norm": 0.8479176759719849,
      "learning_rate": 0.0002511485826001955,
      "loss": 0.3578,
      "step": 2505
    },
    {
      "epoch": 0.3922980588603632,
      "grad_norm": 0.6460444331169128,
      "learning_rate": 0.00025112414467253174,
      "loss": 0.3305,
      "step": 2506
    },
    {
      "epoch": 0.3924546023794615,
      "grad_norm": 0.7635279297828674,
      "learning_rate": 0.00025109970674486805,
      "loss": 0.3328,
      "step": 2507
    },
    {
      "epoch": 0.3926111458985598,
      "grad_norm": 0.6837889552116394,
      "learning_rate": 0.00025107526881720425,
      "loss": 0.3137,
      "step": 2508
    },
    {
      "epoch": 0.3927676894176581,
      "grad_norm": 3.4760055541992188,
      "learning_rate": 0.00025105083088954055,
      "loss": 1.145,
      "step": 2509
    },
    {
      "epoch": 0.39292423293675643,
      "grad_norm": 0.936640739440918,
      "learning_rate": 0.0002510263929618768,
      "loss": 0.4839,
      "step": 2510
    },
    {
      "epoch": 0.3930807764558547,
      "grad_norm": 0.6478950381278992,
      "learning_rate": 0.00025100195503421305,
      "loss": 0.239,
      "step": 2511
    },
    {
      "epoch": 0.39323731997495304,
      "grad_norm": 1.125580906867981,
      "learning_rate": 0.00025097751710654936,
      "loss": 0.5681,
      "step": 2512
    },
    {
      "epoch": 0.39339386349405137,
      "grad_norm": 1.4228874444961548,
      "learning_rate": 0.0002509530791788856,
      "loss": 0.7442,
      "step": 2513
    },
    {
      "epoch": 0.39355040701314964,
      "grad_norm": 1.260831356048584,
      "learning_rate": 0.00025092864125122186,
      "loss": 0.6783,
      "step": 2514
    },
    {
      "epoch": 0.39370695053224797,
      "grad_norm": 1.2701672315597534,
      "learning_rate": 0.0002509042033235581,
      "loss": 0.426,
      "step": 2515
    },
    {
      "epoch": 0.3938634940513463,
      "grad_norm": 0.7969043850898743,
      "learning_rate": 0.0002508797653958944,
      "loss": 0.4279,
      "step": 2516
    },
    {
      "epoch": 0.39402003757044457,
      "grad_norm": 2.375305414199829,
      "learning_rate": 0.00025085532746823067,
      "loss": 0.8547,
      "step": 2517
    },
    {
      "epoch": 0.3941765810895429,
      "grad_norm": 1.0514167547225952,
      "learning_rate": 0.0002508308895405669,
      "loss": 0.5921,
      "step": 2518
    },
    {
      "epoch": 0.39433312460864123,
      "grad_norm": 1.659162998199463,
      "learning_rate": 0.0002508064516129032,
      "loss": 0.792,
      "step": 2519
    },
    {
      "epoch": 0.3944896681277395,
      "grad_norm": 1.6312371492385864,
      "learning_rate": 0.0002507820136852395,
      "loss": 0.8341,
      "step": 2520
    },
    {
      "epoch": 0.39464621164683783,
      "grad_norm": 1.0578227043151855,
      "learning_rate": 0.00025075757575757573,
      "loss": 0.6725,
      "step": 2521
    },
    {
      "epoch": 0.3948027551659361,
      "grad_norm": 1.3463261127471924,
      "learning_rate": 0.00025073313782991203,
      "loss": 0.753,
      "step": 2522
    },
    {
      "epoch": 0.39495929868503443,
      "grad_norm": 1.1711524724960327,
      "learning_rate": 0.00025070869990224823,
      "loss": 0.6932,
      "step": 2523
    },
    {
      "epoch": 0.39511584220413276,
      "grad_norm": 1.7872284650802612,
      "learning_rate": 0.00025068426197458454,
      "loss": 0.8229,
      "step": 2524
    },
    {
      "epoch": 0.39527238572323103,
      "grad_norm": 1.4881881475448608,
      "learning_rate": 0.0002506598240469208,
      "loss": 1.1885,
      "step": 2525
    },
    {
      "epoch": 0.39542892924232936,
      "grad_norm": 2.591526508331299,
      "learning_rate": 0.00025063538611925704,
      "loss": 0.9122,
      "step": 2526
    },
    {
      "epoch": 0.3955854727614277,
      "grad_norm": 2.517484188079834,
      "learning_rate": 0.00025061094819159334,
      "loss": 1.1493,
      "step": 2527
    },
    {
      "epoch": 0.39574201628052597,
      "grad_norm": 1.369888424873352,
      "learning_rate": 0.0002505865102639296,
      "loss": 0.7883,
      "step": 2528
    },
    {
      "epoch": 0.3958985597996243,
      "grad_norm": 1.5327993631362915,
      "learning_rate": 0.00025056207233626585,
      "loss": 0.5566,
      "step": 2529
    },
    {
      "epoch": 0.3960551033187226,
      "grad_norm": 2.494425058364868,
      "learning_rate": 0.00025053763440860215,
      "loss": 0.8489,
      "step": 2530
    },
    {
      "epoch": 0.3962116468378209,
      "grad_norm": 1.5433497428894043,
      "learning_rate": 0.0002505131964809384,
      "loss": 0.8051,
      "step": 2531
    },
    {
      "epoch": 0.3963681903569192,
      "grad_norm": 3.3626677989959717,
      "learning_rate": 0.00025048875855327465,
      "loss": 1.0817,
      "step": 2532
    },
    {
      "epoch": 0.39652473387601755,
      "grad_norm": 2.291754722595215,
      "learning_rate": 0.0002504643206256109,
      "loss": 0.9022,
      "step": 2533
    },
    {
      "epoch": 0.3966812773951158,
      "grad_norm": 2.296135425567627,
      "learning_rate": 0.0002504398826979472,
      "loss": 1.4655,
      "step": 2534
    },
    {
      "epoch": 0.39683782091421416,
      "grad_norm": 3.5448482036590576,
      "learning_rate": 0.00025041544477028346,
      "loss": 0.9305,
      "step": 2535
    },
    {
      "epoch": 0.3969943644333125,
      "grad_norm": 2.3300135135650635,
      "learning_rate": 0.0002503910068426197,
      "loss": 1.3307,
      "step": 2536
    },
    {
      "epoch": 0.39715090795241076,
      "grad_norm": 1.741479516029358,
      "learning_rate": 0.000250366568914956,
      "loss": 0.7833,
      "step": 2537
    },
    {
      "epoch": 0.3973074514715091,
      "grad_norm": 4.329046249389648,
      "learning_rate": 0.00025034213098729227,
      "loss": 1.0903,
      "step": 2538
    },
    {
      "epoch": 0.3974639949906074,
      "grad_norm": 5.144536972045898,
      "learning_rate": 0.0002503176930596285,
      "loss": 1.4296,
      "step": 2539
    },
    {
      "epoch": 0.3976205385097057,
      "grad_norm": 2.033552408218384,
      "learning_rate": 0.0002502932551319648,
      "loss": 1.2994,
      "step": 2540
    },
    {
      "epoch": 0.397777082028804,
      "grad_norm": 2.1732325553894043,
      "learning_rate": 0.000250268817204301,
      "loss": 1.3316,
      "step": 2541
    },
    {
      "epoch": 0.3979336255479023,
      "grad_norm": 4.526739597320557,
      "learning_rate": 0.00025024437927663733,
      "loss": 1.187,
      "step": 2542
    },
    {
      "epoch": 0.3980901690670006,
      "grad_norm": 1.4033749103546143,
      "learning_rate": 0.0002502199413489736,
      "loss": 0.8175,
      "step": 2543
    },
    {
      "epoch": 0.39824671258609895,
      "grad_norm": 3.880048990249634,
      "learning_rate": 0.00025019550342130983,
      "loss": 1.6048,
      "step": 2544
    },
    {
      "epoch": 0.3984032561051972,
      "grad_norm": 2.1508264541625977,
      "learning_rate": 0.00025017106549364614,
      "loss": 1.3564,
      "step": 2545
    },
    {
      "epoch": 0.39855979962429555,
      "grad_norm": 1.6752632856369019,
      "learning_rate": 0.0002501466275659824,
      "loss": 0.6449,
      "step": 2546
    },
    {
      "epoch": 0.3987163431433939,
      "grad_norm": 3.97983980178833,
      "learning_rate": 0.00025012218963831864,
      "loss": 1.7772,
      "step": 2547
    },
    {
      "epoch": 0.39887288666249215,
      "grad_norm": 2.7829971313476562,
      "learning_rate": 0.0002500977517106549,
      "loss": 1.109,
      "step": 2548
    },
    {
      "epoch": 0.3990294301815905,
      "grad_norm": 1.2870094776153564,
      "learning_rate": 0.0002500733137829912,
      "loss": 0.4101,
      "step": 2549
    },
    {
      "epoch": 0.3991859737006888,
      "grad_norm": 2.9641854763031006,
      "learning_rate": 0.00025004887585532745,
      "loss": 1.1036,
      "step": 2550
    },
    {
      "epoch": 0.3993425172197871,
      "grad_norm": 0.6355754137039185,
      "learning_rate": 0.0002500244379276637,
      "loss": 0.3884,
      "step": 2551
    },
    {
      "epoch": 0.3994990607388854,
      "grad_norm": 0.7654311656951904,
      "learning_rate": 0.00025,
      "loss": 0.4459,
      "step": 2552
    },
    {
      "epoch": 0.39965560425798374,
      "grad_norm": 0.6441348791122437,
      "learning_rate": 0.00024997556207233626,
      "loss": 0.3962,
      "step": 2553
    },
    {
      "epoch": 0.399812147777082,
      "grad_norm": 0.8454661965370178,
      "learning_rate": 0.0002499511241446725,
      "loss": 0.3226,
      "step": 2554
    },
    {
      "epoch": 0.39996869129618035,
      "grad_norm": 0.7234505414962769,
      "learning_rate": 0.0002499266862170088,
      "loss": 0.3463,
      "step": 2555
    },
    {
      "epoch": 0.4001252348152787,
      "grad_norm": 0.7027657628059387,
      "learning_rate": 0.000249902248289345,
      "loss": 0.2914,
      "step": 2556
    },
    {
      "epoch": 0.40028177833437695,
      "grad_norm": 0.9003955721855164,
      "learning_rate": 0.0002498778103616813,
      "loss": 0.3581,
      "step": 2557
    },
    {
      "epoch": 0.4004383218534753,
      "grad_norm": 0.8857301473617554,
      "learning_rate": 0.00024985337243401757,
      "loss": 0.6009,
      "step": 2558
    },
    {
      "epoch": 0.40059486537257355,
      "grad_norm": 0.8024937510490417,
      "learning_rate": 0.0002498289345063538,
      "loss": 0.407,
      "step": 2559
    },
    {
      "epoch": 0.4007514088916719,
      "grad_norm": 1.1914548873901367,
      "learning_rate": 0.0002498044965786901,
      "loss": 0.5759,
      "step": 2560
    },
    {
      "epoch": 0.4009079524107702,
      "grad_norm": 1.0876412391662598,
      "learning_rate": 0.0002497800586510264,
      "loss": 0.4644,
      "step": 2561
    },
    {
      "epoch": 0.4010644959298685,
      "grad_norm": 0.8005790710449219,
      "learning_rate": 0.0002497556207233626,
      "loss": 0.327,
      "step": 2562
    },
    {
      "epoch": 0.4012210394489668,
      "grad_norm": 0.8703690767288208,
      "learning_rate": 0.0002497311827956989,
      "loss": 0.6214,
      "step": 2563
    },
    {
      "epoch": 0.40137758296806514,
      "grad_norm": 1.3815338611602783,
      "learning_rate": 0.0002497067448680352,
      "loss": 0.4908,
      "step": 2564
    },
    {
      "epoch": 0.4015341264871634,
      "grad_norm": 3.5727379322052,
      "learning_rate": 0.00024968230694037143,
      "loss": 0.4925,
      "step": 2565
    },
    {
      "epoch": 0.40169067000626174,
      "grad_norm": 1.0240099430084229,
      "learning_rate": 0.0002496578690127077,
      "loss": 0.661,
      "step": 2566
    },
    {
      "epoch": 0.40184721352536007,
      "grad_norm": 1.9735991954803467,
      "learning_rate": 0.000249633431085044,
      "loss": 0.5323,
      "step": 2567
    },
    {
      "epoch": 0.40200375704445834,
      "grad_norm": 1.2893562316894531,
      "learning_rate": 0.00024960899315738024,
      "loss": 0.5383,
      "step": 2568
    },
    {
      "epoch": 0.40216030056355667,
      "grad_norm": 0.8876537680625916,
      "learning_rate": 0.0002495845552297165,
      "loss": 0.6309,
      "step": 2569
    },
    {
      "epoch": 0.402316844082655,
      "grad_norm": 1.0564181804656982,
      "learning_rate": 0.0002495601173020528,
      "loss": 0.5627,
      "step": 2570
    },
    {
      "epoch": 0.4024733876017533,
      "grad_norm": 1.0052175521850586,
      "learning_rate": 0.000249535679374389,
      "loss": 0.6418,
      "step": 2571
    },
    {
      "epoch": 0.4026299311208516,
      "grad_norm": 0.7646985054016113,
      "learning_rate": 0.0002495112414467253,
      "loss": 0.4353,
      "step": 2572
    },
    {
      "epoch": 0.40278647463994993,
      "grad_norm": 1.8414807319641113,
      "learning_rate": 0.00024948680351906155,
      "loss": 0.5096,
      "step": 2573
    },
    {
      "epoch": 0.4029430181590482,
      "grad_norm": 1.9422529935836792,
      "learning_rate": 0.0002494623655913978,
      "loss": 0.6357,
      "step": 2574
    },
    {
      "epoch": 0.40309956167814653,
      "grad_norm": 2.9409773349761963,
      "learning_rate": 0.0002494379276637341,
      "loss": 0.5801,
      "step": 2575
    },
    {
      "epoch": 0.4032561051972448,
      "grad_norm": 1.64481782913208,
      "learning_rate": 0.00024941348973607036,
      "loss": 1.0987,
      "step": 2576
    },
    {
      "epoch": 0.40341264871634314,
      "grad_norm": 2.1421914100646973,
      "learning_rate": 0.0002493890518084066,
      "loss": 0.7097,
      "step": 2577
    },
    {
      "epoch": 0.40356919223544147,
      "grad_norm": 1.6423346996307373,
      "learning_rate": 0.0002493646138807429,
      "loss": 0.8432,
      "step": 2578
    },
    {
      "epoch": 0.40372573575453974,
      "grad_norm": 4.2599005699157715,
      "learning_rate": 0.00024934017595307917,
      "loss": 1.2652,
      "step": 2579
    },
    {
      "epoch": 0.40388227927363807,
      "grad_norm": 2.666485071182251,
      "learning_rate": 0.0002493157380254154,
      "loss": 1.0066,
      "step": 2580
    },
    {
      "epoch": 0.4040388227927364,
      "grad_norm": 3.4429917335510254,
      "learning_rate": 0.00024929130009775167,
      "loss": 1.0087,
      "step": 2581
    },
    {
      "epoch": 0.40419536631183467,
      "grad_norm": 3.632044553756714,
      "learning_rate": 0.000249266862170088,
      "loss": 1.2341,
      "step": 2582
    },
    {
      "epoch": 0.404351909830933,
      "grad_norm": 1.675054907798767,
      "learning_rate": 0.00024924242424242423,
      "loss": 0.939,
      "step": 2583
    },
    {
      "epoch": 0.4045084533500313,
      "grad_norm": 2.2389726638793945,
      "learning_rate": 0.0002492179863147605,
      "loss": 1.3942,
      "step": 2584
    },
    {
      "epoch": 0.4046649968691296,
      "grad_norm": 1.801397442817688,
      "learning_rate": 0.0002491935483870968,
      "loss": 0.9123,
      "step": 2585
    },
    {
      "epoch": 0.40482154038822793,
      "grad_norm": 3.3526382446289062,
      "learning_rate": 0.000249169110459433,
      "loss": 1.5424,
      "step": 2586
    },
    {
      "epoch": 0.40497808390732626,
      "grad_norm": 1.4892598390579224,
      "learning_rate": 0.0002491446725317693,
      "loss": 0.7216,
      "step": 2587
    },
    {
      "epoch": 0.40513462742642453,
      "grad_norm": 2.2398314476013184,
      "learning_rate": 0.00024912023460410554,
      "loss": 1.1292,
      "step": 2588
    },
    {
      "epoch": 0.40529117094552286,
      "grad_norm": 7.038582801818848,
      "learning_rate": 0.0002490957966764418,
      "loss": 1.1889,
      "step": 2589
    },
    {
      "epoch": 0.4054477144646212,
      "grad_norm": 2.2800958156585693,
      "learning_rate": 0.0002490713587487781,
      "loss": 1.0995,
      "step": 2590
    },
    {
      "epoch": 0.40560425798371946,
      "grad_norm": 2.6573739051818848,
      "learning_rate": 0.00024904692082111435,
      "loss": 1.5452,
      "step": 2591
    },
    {
      "epoch": 0.4057608015028178,
      "grad_norm": 2.7160732746124268,
      "learning_rate": 0.0002490224828934506,
      "loss": 1.134,
      "step": 2592
    },
    {
      "epoch": 0.40591734502191607,
      "grad_norm": 3.265036106109619,
      "learning_rate": 0.0002489980449657869,
      "loss": 1.2917,
      "step": 2593
    },
    {
      "epoch": 0.4060738885410144,
      "grad_norm": 3.3533453941345215,
      "learning_rate": 0.00024897360703812315,
      "loss": 1.5418,
      "step": 2594
    },
    {
      "epoch": 0.4062304320601127,
      "grad_norm": 1.740634560585022,
      "learning_rate": 0.0002489491691104594,
      "loss": 1.3963,
      "step": 2595
    },
    {
      "epoch": 0.406386975579211,
      "grad_norm": 5.385989665985107,
      "learning_rate": 0.00024892473118279566,
      "loss": 0.9535,
      "step": 2596
    },
    {
      "epoch": 0.4065435190983093,
      "grad_norm": 2.453171730041504,
      "learning_rate": 0.00024890029325513196,
      "loss": 1.1681,
      "step": 2597
    },
    {
      "epoch": 0.40670006261740765,
      "grad_norm": 2.1263811588287354,
      "learning_rate": 0.0002488758553274682,
      "loss": 0.795,
      "step": 2598
    },
    {
      "epoch": 0.4068566061365059,
      "grad_norm": 2.388746738433838,
      "learning_rate": 0.00024885141739980446,
      "loss": 0.9264,
      "step": 2599
    },
    {
      "epoch": 0.40701314965560426,
      "grad_norm": 1.1103012561798096,
      "learning_rate": 0.00024882697947214077,
      "loss": 0.8049,
      "step": 2600
    },
    {
      "epoch": 0.4071696931747026,
      "grad_norm": 0.601675271987915,
      "learning_rate": 0.000248802541544477,
      "loss": 0.365,
      "step": 2601
    },
    {
      "epoch": 0.40732623669380086,
      "grad_norm": 0.5282478332519531,
      "learning_rate": 0.00024877810361681327,
      "loss": 0.3728,
      "step": 2602
    },
    {
      "epoch": 0.4074827802128992,
      "grad_norm": 0.5197291374206543,
      "learning_rate": 0.0002487536656891495,
      "loss": 0.3763,
      "step": 2603
    },
    {
      "epoch": 0.4076393237319975,
      "grad_norm": 1.0491033792495728,
      "learning_rate": 0.0002487292277614858,
      "loss": 0.4424,
      "step": 2604
    },
    {
      "epoch": 0.4077958672510958,
      "grad_norm": 1.314231038093567,
      "learning_rate": 0.0002487047898338221,
      "loss": 0.5056,
      "step": 2605
    },
    {
      "epoch": 0.4079524107701941,
      "grad_norm": 0.6822581887245178,
      "learning_rate": 0.00024868035190615833,
      "loss": 0.3068,
      "step": 2606
    },
    {
      "epoch": 0.40810895428929245,
      "grad_norm": 1.021838665008545,
      "learning_rate": 0.0002486559139784946,
      "loss": 0.4735,
      "step": 2607
    },
    {
      "epoch": 0.4082654978083907,
      "grad_norm": 0.770111083984375,
      "learning_rate": 0.0002486314760508309,
      "loss": 0.4265,
      "step": 2608
    },
    {
      "epoch": 0.40842204132748905,
      "grad_norm": 0.6351111531257629,
      "learning_rate": 0.00024860703812316714,
      "loss": 0.4441,
      "step": 2609
    },
    {
      "epoch": 0.4085785848465874,
      "grad_norm": 1.667920470237732,
      "learning_rate": 0.0002485826001955034,
      "loss": 0.8832,
      "step": 2610
    },
    {
      "epoch": 0.40873512836568565,
      "grad_norm": 0.9965152740478516,
      "learning_rate": 0.00024855816226783964,
      "loss": 0.4595,
      "step": 2611
    },
    {
      "epoch": 0.408891671884784,
      "grad_norm": 1.3201706409454346,
      "learning_rate": 0.00024853372434017595,
      "loss": 0.7611,
      "step": 2612
    },
    {
      "epoch": 0.40904821540388225,
      "grad_norm": 0.8612977266311646,
      "learning_rate": 0.0002485092864125122,
      "loss": 0.4856,
      "step": 2613
    },
    {
      "epoch": 0.4092047589229806,
      "grad_norm": 2.6356048583984375,
      "learning_rate": 0.00024848484848484845,
      "loss": 0.9506,
      "step": 2614
    },
    {
      "epoch": 0.4093613024420789,
      "grad_norm": 1.1005220413208008,
      "learning_rate": 0.00024846041055718476,
      "loss": 0.4625,
      "step": 2615
    },
    {
      "epoch": 0.4095178459611772,
      "grad_norm": 1.277478814125061,
      "learning_rate": 0.000248435972629521,
      "loss": 0.747,
      "step": 2616
    },
    {
      "epoch": 0.4096743894802755,
      "grad_norm": 1.216950535774231,
      "learning_rate": 0.00024841153470185726,
      "loss": 0.5756,
      "step": 2617
    },
    {
      "epoch": 0.40983093299937384,
      "grad_norm": 1.0903007984161377,
      "learning_rate": 0.00024838709677419356,
      "loss": 0.7352,
      "step": 2618
    },
    {
      "epoch": 0.4099874765184721,
      "grad_norm": 1.3664308786392212,
      "learning_rate": 0.00024836265884652976,
      "loss": 0.6177,
      "step": 2619
    },
    {
      "epoch": 0.41014402003757044,
      "grad_norm": 1.4325757026672363,
      "learning_rate": 0.00024833822091886607,
      "loss": 0.7601,
      "step": 2620
    },
    {
      "epoch": 0.4103005635566688,
      "grad_norm": 1.858127236366272,
      "learning_rate": 0.0002483137829912023,
      "loss": 0.7936,
      "step": 2621
    },
    {
      "epoch": 0.41045710707576705,
      "grad_norm": 3.4538111686706543,
      "learning_rate": 0.00024828934506353857,
      "loss": 0.9523,
      "step": 2622
    },
    {
      "epoch": 0.4106136505948654,
      "grad_norm": 1.925356388092041,
      "learning_rate": 0.0002482649071358749,
      "loss": 0.8096,
      "step": 2623
    },
    {
      "epoch": 0.4107701941139637,
      "grad_norm": 1.6007717847824097,
      "learning_rate": 0.0002482404692082111,
      "loss": 0.9204,
      "step": 2624
    },
    {
      "epoch": 0.410926737633062,
      "grad_norm": 2.12320613861084,
      "learning_rate": 0.0002482160312805474,
      "loss": 0.9734,
      "step": 2625
    },
    {
      "epoch": 0.4110832811521603,
      "grad_norm": 1.3324270248413086,
      "learning_rate": 0.0002481915933528837,
      "loss": 1.0131,
      "step": 2626
    },
    {
      "epoch": 0.41123982467125864,
      "grad_norm": 1.9677183628082275,
      "learning_rate": 0.00024816715542521993,
      "loss": 0.9945,
      "step": 2627
    },
    {
      "epoch": 0.4113963681903569,
      "grad_norm": 1.487841010093689,
      "learning_rate": 0.0002481427174975562,
      "loss": 0.813,
      "step": 2628
    },
    {
      "epoch": 0.41155291170945524,
      "grad_norm": 2.215928554534912,
      "learning_rate": 0.00024811827956989244,
      "loss": 0.8284,
      "step": 2629
    },
    {
      "epoch": 0.4117094552285535,
      "grad_norm": 2.1060638427734375,
      "learning_rate": 0.00024809384164222874,
      "loss": 0.9693,
      "step": 2630
    },
    {
      "epoch": 0.41186599874765184,
      "grad_norm": 1.2802146673202515,
      "learning_rate": 0.000248069403714565,
      "loss": 0.5589,
      "step": 2631
    },
    {
      "epoch": 0.41202254226675017,
      "grad_norm": 2.4846386909484863,
      "learning_rate": 0.00024804496578690124,
      "loss": 0.9417,
      "step": 2632
    },
    {
      "epoch": 0.41217908578584844,
      "grad_norm": 1.465616226196289,
      "learning_rate": 0.00024802052785923755,
      "loss": 0.523,
      "step": 2633
    },
    {
      "epoch": 0.41233562930494677,
      "grad_norm": 1.689579963684082,
      "learning_rate": 0.00024799608993157375,
      "loss": 0.7951,
      "step": 2634
    },
    {
      "epoch": 0.4124921728240451,
      "grad_norm": 2.5636866092681885,
      "learning_rate": 0.00024797165200391005,
      "loss": 1.4816,
      "step": 2635
    },
    {
      "epoch": 0.4126487163431434,
      "grad_norm": 2.3543753623962402,
      "learning_rate": 0.0002479472140762463,
      "loss": 1.4351,
      "step": 2636
    },
    {
      "epoch": 0.4128052598622417,
      "grad_norm": 2.550464630126953,
      "learning_rate": 0.00024792277614858255,
      "loss": 1.5062,
      "step": 2637
    },
    {
      "epoch": 0.41296180338134003,
      "grad_norm": 2.0879738330841064,
      "learning_rate": 0.00024789833822091886,
      "loss": 1.5573,
      "step": 2638
    },
    {
      "epoch": 0.4131183469004383,
      "grad_norm": 2.426447629928589,
      "learning_rate": 0.0002478739002932551,
      "loss": 1.5449,
      "step": 2639
    },
    {
      "epoch": 0.41327489041953663,
      "grad_norm": 2.501875877380371,
      "learning_rate": 0.00024784946236559136,
      "loss": 1.1526,
      "step": 2640
    },
    {
      "epoch": 0.41343143393863496,
      "grad_norm": 3.55033540725708,
      "learning_rate": 0.00024782502443792767,
      "loss": 1.4765,
      "step": 2641
    },
    {
      "epoch": 0.41358797745773324,
      "grad_norm": 4.644171237945557,
      "learning_rate": 0.0002478005865102639,
      "loss": 1.6887,
      "step": 2642
    },
    {
      "epoch": 0.41374452097683156,
      "grad_norm": 2.9000136852264404,
      "learning_rate": 0.00024777614858260017,
      "loss": 1.3512,
      "step": 2643
    },
    {
      "epoch": 0.4139010644959299,
      "grad_norm": 2.809741735458374,
      "learning_rate": 0.0002477517106549364,
      "loss": 1.1566,
      "step": 2644
    },
    {
      "epoch": 0.41405760801502817,
      "grad_norm": 2.349257230758667,
      "learning_rate": 0.0002477272727272727,
      "loss": 1.4035,
      "step": 2645
    },
    {
      "epoch": 0.4142141515341265,
      "grad_norm": 3.1145989894866943,
      "learning_rate": 0.000247702834799609,
      "loss": 1.3006,
      "step": 2646
    },
    {
      "epoch": 0.41437069505322477,
      "grad_norm": 1.8506964445114136,
      "learning_rate": 0.00024767839687194523,
      "loss": 1.3462,
      "step": 2647
    },
    {
      "epoch": 0.4145272385723231,
      "grad_norm": 1.8519948720932007,
      "learning_rate": 0.00024765395894428153,
      "loss": 0.7869,
      "step": 2648
    },
    {
      "epoch": 0.4146837820914214,
      "grad_norm": 1.9720544815063477,
      "learning_rate": 0.0002476295210166178,
      "loss": 1.4319,
      "step": 2649
    },
    {
      "epoch": 0.4148403256105197,
      "grad_norm": 3.180431365966797,
      "learning_rate": 0.00024760508308895404,
      "loss": 2.3146,
      "step": 2650
    },
    {
      "epoch": 0.41499686912961803,
      "grad_norm": 0.8477596640586853,
      "learning_rate": 0.0002475806451612903,
      "loss": 0.4092,
      "step": 2651
    },
    {
      "epoch": 0.41515341264871636,
      "grad_norm": 0.5798821449279785,
      "learning_rate": 0.00024755620723362654,
      "loss": 0.3916,
      "step": 2652
    },
    {
      "epoch": 0.41530995616781463,
      "grad_norm": 0.638412356376648,
      "learning_rate": 0.00024753176930596284,
      "loss": 0.3177,
      "step": 2653
    },
    {
      "epoch": 0.41546649968691296,
      "grad_norm": 0.5856349468231201,
      "learning_rate": 0.0002475073313782991,
      "loss": 0.3968,
      "step": 2654
    },
    {
      "epoch": 0.4156230432060113,
      "grad_norm": 0.852562665939331,
      "learning_rate": 0.00024748289345063535,
      "loss": 0.3886,
      "step": 2655
    },
    {
      "epoch": 0.41577958672510956,
      "grad_norm": 0.7616521716117859,
      "learning_rate": 0.00024745845552297165,
      "loss": 0.4342,
      "step": 2656
    },
    {
      "epoch": 0.4159361302442079,
      "grad_norm": 0.7631804943084717,
      "learning_rate": 0.0002474340175953079,
      "loss": 0.4359,
      "step": 2657
    },
    {
      "epoch": 0.4160926737633062,
      "grad_norm": 1.0951449871063232,
      "learning_rate": 0.00024740957966764416,
      "loss": 0.3401,
      "step": 2658
    },
    {
      "epoch": 0.4162492172824045,
      "grad_norm": 0.5410890579223633,
      "learning_rate": 0.0002473851417399804,
      "loss": 0.3564,
      "step": 2659
    },
    {
      "epoch": 0.4164057608015028,
      "grad_norm": 0.9513525366783142,
      "learning_rate": 0.0002473607038123167,
      "loss": 0.5835,
      "step": 2660
    },
    {
      "epoch": 0.41656230432060115,
      "grad_norm": 1.1458948850631714,
      "learning_rate": 0.00024733626588465296,
      "loss": 0.5584,
      "step": 2661
    },
    {
      "epoch": 0.4167188478396994,
      "grad_norm": 1.0255465507507324,
      "learning_rate": 0.0002473118279569892,
      "loss": 0.4138,
      "step": 2662
    },
    {
      "epoch": 0.41687539135879775,
      "grad_norm": 2.739941120147705,
      "learning_rate": 0.0002472873900293255,
      "loss": 0.7455,
      "step": 2663
    },
    {
      "epoch": 0.4170319348778961,
      "grad_norm": 1.1839171648025513,
      "learning_rate": 0.00024726295210166177,
      "loss": 0.6217,
      "step": 2664
    },
    {
      "epoch": 0.41718847839699436,
      "grad_norm": 1.1585807800292969,
      "learning_rate": 0.000247238514173998,
      "loss": 0.5179,
      "step": 2665
    },
    {
      "epoch": 0.4173450219160927,
      "grad_norm": 1.2621674537658691,
      "learning_rate": 0.00024721407624633433,
      "loss": 0.5774,
      "step": 2666
    },
    {
      "epoch": 0.41750156543519096,
      "grad_norm": 1.1701115369796753,
      "learning_rate": 0.0002471896383186705,
      "loss": 0.6508,
      "step": 2667
    },
    {
      "epoch": 0.4176581089542893,
      "grad_norm": 1.1155378818511963,
      "learning_rate": 0.00024716520039100683,
      "loss": 0.809,
      "step": 2668
    },
    {
      "epoch": 0.4178146524733876,
      "grad_norm": 1.2293123006820679,
      "learning_rate": 0.0002471407624633431,
      "loss": 0.5816,
      "step": 2669
    },
    {
      "epoch": 0.4179711959924859,
      "grad_norm": 1.4187663793563843,
      "learning_rate": 0.00024711632453567933,
      "loss": 0.6369,
      "step": 2670
    },
    {
      "epoch": 0.4181277395115842,
      "grad_norm": 1.2454630136489868,
      "learning_rate": 0.00024709188660801564,
      "loss": 0.715,
      "step": 2671
    },
    {
      "epoch": 0.41828428303068255,
      "grad_norm": 1.0575640201568604,
      "learning_rate": 0.0002470674486803519,
      "loss": 0.5052,
      "step": 2672
    },
    {
      "epoch": 0.4184408265497808,
      "grad_norm": 1.1785027980804443,
      "learning_rate": 0.00024704301075268814,
      "loss": 0.8882,
      "step": 2673
    },
    {
      "epoch": 0.41859737006887915,
      "grad_norm": 2.9896340370178223,
      "learning_rate": 0.0002470185728250244,
      "loss": 1.3041,
      "step": 2674
    },
    {
      "epoch": 0.4187539135879775,
      "grad_norm": 2.0894863605499268,
      "learning_rate": 0.0002469941348973607,
      "loss": 0.7306,
      "step": 2675
    },
    {
      "epoch": 0.41891045710707575,
      "grad_norm": 1.6486024856567383,
      "learning_rate": 0.00024696969696969695,
      "loss": 0.846,
      "step": 2676
    },
    {
      "epoch": 0.4190670006261741,
      "grad_norm": 2.0100576877593994,
      "learning_rate": 0.0002469452590420332,
      "loss": 0.8831,
      "step": 2677
    },
    {
      "epoch": 0.4192235441452724,
      "grad_norm": 2.7158892154693604,
      "learning_rate": 0.0002469208211143695,
      "loss": 0.8862,
      "step": 2678
    },
    {
      "epoch": 0.4193800876643707,
      "grad_norm": 2.027427911758423,
      "learning_rate": 0.00024689638318670576,
      "loss": 1.0893,
      "step": 2679
    },
    {
      "epoch": 0.419536631183469,
      "grad_norm": 1.634826421737671,
      "learning_rate": 0.000246871945259042,
      "loss": 0.9038,
      "step": 2680
    },
    {
      "epoch": 0.41969317470256734,
      "grad_norm": 1.4327385425567627,
      "learning_rate": 0.0002468475073313783,
      "loss": 0.6251,
      "step": 2681
    },
    {
      "epoch": 0.4198497182216656,
      "grad_norm": 2.376344919204712,
      "learning_rate": 0.0002468230694037145,
      "loss": 0.9759,
      "step": 2682
    },
    {
      "epoch": 0.42000626174076394,
      "grad_norm": 2.1796486377716064,
      "learning_rate": 0.0002467986314760508,
      "loss": 1.1159,
      "step": 2683
    },
    {
      "epoch": 0.4201628052598622,
      "grad_norm": 2.264559507369995,
      "learning_rate": 0.00024677419354838707,
      "loss": 1.0303,
      "step": 2684
    },
    {
      "epoch": 0.42031934877896054,
      "grad_norm": 3.5693933963775635,
      "learning_rate": 0.0002467497556207233,
      "loss": 1.2559,
      "step": 2685
    },
    {
      "epoch": 0.4204758922980589,
      "grad_norm": 1.9500762224197388,
      "learning_rate": 0.0002467253176930596,
      "loss": 1.3884,
      "step": 2686
    },
    {
      "epoch": 0.42063243581715715,
      "grad_norm": 1.6748422384262085,
      "learning_rate": 0.0002467008797653959,
      "loss": 1.1449,
      "step": 2687
    },
    {
      "epoch": 0.4207889793362555,
      "grad_norm": 1.6067205667495728,
      "learning_rate": 0.0002466764418377321,
      "loss": 0.8418,
      "step": 2688
    },
    {
      "epoch": 0.4209455228553538,
      "grad_norm": 1.628675937652588,
      "learning_rate": 0.00024665200391006843,
      "loss": 0.8212,
      "step": 2689
    },
    {
      "epoch": 0.4211020663744521,
      "grad_norm": 2.2137458324432373,
      "learning_rate": 0.0002466275659824047,
      "loss": 1.5843,
      "step": 2690
    },
    {
      "epoch": 0.4212586098935504,
      "grad_norm": 2.58758282661438,
      "learning_rate": 0.00024660312805474093,
      "loss": 1.4545,
      "step": 2691
    },
    {
      "epoch": 0.42141515341264874,
      "grad_norm": 2.075272560119629,
      "learning_rate": 0.0002465786901270772,
      "loss": 1.8048,
      "step": 2692
    },
    {
      "epoch": 0.421571696931747,
      "grad_norm": 2.119429588317871,
      "learning_rate": 0.0002465542521994135,
      "loss": 1.3897,
      "step": 2693
    },
    {
      "epoch": 0.42172824045084534,
      "grad_norm": 1.7578437328338623,
      "learning_rate": 0.00024652981427174974,
      "loss": 1.8498,
      "step": 2694
    },
    {
      "epoch": 0.42188478396994367,
      "grad_norm": 1.8442625999450684,
      "learning_rate": 0.000246505376344086,
      "loss": 1.3389,
      "step": 2695
    },
    {
      "epoch": 0.42204132748904194,
      "grad_norm": 1.9508384466171265,
      "learning_rate": 0.0002464809384164223,
      "loss": 0.8474,
      "step": 2696
    },
    {
      "epoch": 0.42219787100814027,
      "grad_norm": 1.3713496923446655,
      "learning_rate": 0.00024645650048875855,
      "loss": 0.8622,
      "step": 2697
    },
    {
      "epoch": 0.4223544145272386,
      "grad_norm": 2.2516250610351562,
      "learning_rate": 0.0002464320625610948,
      "loss": 0.6771,
      "step": 2698
    },
    {
      "epoch": 0.42251095804633687,
      "grad_norm": 2.1349215507507324,
      "learning_rate": 0.00024640762463343105,
      "loss": 1.3575,
      "step": 2699
    },
    {
      "epoch": 0.4226675015654352,
      "grad_norm": 3.8896472454071045,
      "learning_rate": 0.0002463831867057673,
      "loss": 1.4243,
      "step": 2700
    },
    {
      "epoch": 0.4228240450845335,
      "grad_norm": 0.8927620053291321,
      "learning_rate": 0.0002463587487781036,
      "loss": 0.3883,
      "step": 2701
    },
    {
      "epoch": 0.4229805886036318,
      "grad_norm": 0.9689701199531555,
      "learning_rate": 0.00024633431085043986,
      "loss": 0.5873,
      "step": 2702
    },
    {
      "epoch": 0.42313713212273013,
      "grad_norm": 0.4493149220943451,
      "learning_rate": 0.0002463098729227761,
      "loss": 0.276,
      "step": 2703
    },
    {
      "epoch": 0.4232936756418284,
      "grad_norm": 0.5155251026153564,
      "learning_rate": 0.0002462854349951124,
      "loss": 0.3299,
      "step": 2704
    },
    {
      "epoch": 0.42345021916092673,
      "grad_norm": 0.7345922589302063,
      "learning_rate": 0.00024626099706744867,
      "loss": 0.3684,
      "step": 2705
    },
    {
      "epoch": 0.42360676268002506,
      "grad_norm": 0.633811354637146,
      "learning_rate": 0.0002462365591397849,
      "loss": 0.4542,
      "step": 2706
    },
    {
      "epoch": 0.42376330619912334,
      "grad_norm": 0.9088723063468933,
      "learning_rate": 0.00024621212121212117,
      "loss": 0.4833,
      "step": 2707
    },
    {
      "epoch": 0.42391984971822166,
      "grad_norm": 1.057075023651123,
      "learning_rate": 0.0002461876832844575,
      "loss": 0.461,
      "step": 2708
    },
    {
      "epoch": 0.42407639323732,
      "grad_norm": 0.996385395526886,
      "learning_rate": 0.00024616324535679373,
      "loss": 0.3916,
      "step": 2709
    },
    {
      "epoch": 0.42423293675641827,
      "grad_norm": 1.1317174434661865,
      "learning_rate": 0.00024613880742913,
      "loss": 0.4329,
      "step": 2710
    },
    {
      "epoch": 0.4243894802755166,
      "grad_norm": 0.8854171633720398,
      "learning_rate": 0.0002461143695014663,
      "loss": 0.3573,
      "step": 2711
    },
    {
      "epoch": 0.4245460237946149,
      "grad_norm": 1.3034350872039795,
      "learning_rate": 0.00024608993157380254,
      "loss": 0.6152,
      "step": 2712
    },
    {
      "epoch": 0.4247025673137132,
      "grad_norm": 0.8485596776008606,
      "learning_rate": 0.0002460654936461388,
      "loss": 0.5314,
      "step": 2713
    },
    {
      "epoch": 0.4248591108328115,
      "grad_norm": 1.0428731441497803,
      "learning_rate": 0.0002460410557184751,
      "loss": 0.4594,
      "step": 2714
    },
    {
      "epoch": 0.42501565435190986,
      "grad_norm": 1.12111496925354,
      "learning_rate": 0.0002460166177908113,
      "loss": 0.4709,
      "step": 2715
    },
    {
      "epoch": 0.42517219787100813,
      "grad_norm": 1.1140966415405273,
      "learning_rate": 0.0002459921798631476,
      "loss": 0.5229,
      "step": 2716
    },
    {
      "epoch": 0.42532874139010646,
      "grad_norm": 0.8085500597953796,
      "learning_rate": 0.00024596774193548385,
      "loss": 0.5124,
      "step": 2717
    },
    {
      "epoch": 0.4254852849092048,
      "grad_norm": 1.5303175449371338,
      "learning_rate": 0.0002459433040078201,
      "loss": 0.6329,
      "step": 2718
    },
    {
      "epoch": 0.42564182842830306,
      "grad_norm": 1.0877023935317993,
      "learning_rate": 0.0002459188660801564,
      "loss": 0.6225,
      "step": 2719
    },
    {
      "epoch": 0.4257983719474014,
      "grad_norm": 1.3418067693710327,
      "learning_rate": 0.00024589442815249265,
      "loss": 0.5107,
      "step": 2720
    },
    {
      "epoch": 0.42595491546649966,
      "grad_norm": 1.9135111570358276,
      "learning_rate": 0.0002458699902248289,
      "loss": 0.6342,
      "step": 2721
    },
    {
      "epoch": 0.426111458985598,
      "grad_norm": 1.0036910772323608,
      "learning_rate": 0.00024584555229716516,
      "loss": 0.5549,
      "step": 2722
    },
    {
      "epoch": 0.4262680025046963,
      "grad_norm": 1.8668835163116455,
      "learning_rate": 0.00024582111436950146,
      "loss": 0.6197,
      "step": 2723
    },
    {
      "epoch": 0.4264245460237946,
      "grad_norm": 2.176187515258789,
      "learning_rate": 0.0002457966764418377,
      "loss": 0.6798,
      "step": 2724
    },
    {
      "epoch": 0.4265810895428929,
      "grad_norm": 1.4894177913665771,
      "learning_rate": 0.00024577223851417396,
      "loss": 1.0341,
      "step": 2725
    },
    {
      "epoch": 0.42673763306199125,
      "grad_norm": 2.1725120544433594,
      "learning_rate": 0.00024574780058651027,
      "loss": 0.8838,
      "step": 2726
    },
    {
      "epoch": 0.4268941765810895,
      "grad_norm": 1.8074668645858765,
      "learning_rate": 0.0002457233626588465,
      "loss": 0.5326,
      "step": 2727
    },
    {
      "epoch": 0.42705072010018785,
      "grad_norm": 2.6459243297576904,
      "learning_rate": 0.00024569892473118277,
      "loss": 0.6351,
      "step": 2728
    },
    {
      "epoch": 0.4272072636192862,
      "grad_norm": 1.6028372049331665,
      "learning_rate": 0.0002456744868035191,
      "loss": 0.9502,
      "step": 2729
    },
    {
      "epoch": 0.42736380713838445,
      "grad_norm": 2.833115816116333,
      "learning_rate": 0.0002456500488758553,
      "loss": 1.3844,
      "step": 2730
    },
    {
      "epoch": 0.4275203506574828,
      "grad_norm": 1.2911146879196167,
      "learning_rate": 0.0002456256109481916,
      "loss": 1.0888,
      "step": 2731
    },
    {
      "epoch": 0.4276768941765811,
      "grad_norm": 1.904571771621704,
      "learning_rate": 0.00024560117302052783,
      "loss": 1.0575,
      "step": 2732
    },
    {
      "epoch": 0.4278334376956794,
      "grad_norm": 2.163661241531372,
      "learning_rate": 0.0002455767350928641,
      "loss": 1.5075,
      "step": 2733
    },
    {
      "epoch": 0.4279899812147777,
      "grad_norm": 5.065518379211426,
      "learning_rate": 0.0002455522971652004,
      "loss": 1.4266,
      "step": 2734
    },
    {
      "epoch": 0.42814652473387604,
      "grad_norm": 1.9618192911148071,
      "learning_rate": 0.00024552785923753664,
      "loss": 1.0812,
      "step": 2735
    },
    {
      "epoch": 0.4283030682529743,
      "grad_norm": 2.5747668743133545,
      "learning_rate": 0.0002455034213098729,
      "loss": 1.0698,
      "step": 2736
    },
    {
      "epoch": 0.42845961177207265,
      "grad_norm": 2.1782901287078857,
      "learning_rate": 0.0002454789833822092,
      "loss": 0.883,
      "step": 2737
    },
    {
      "epoch": 0.4286161552911709,
      "grad_norm": 5.35093879699707,
      "learning_rate": 0.00024545454545454545,
      "loss": 1.4407,
      "step": 2738
    },
    {
      "epoch": 0.42877269881026925,
      "grad_norm": 2.183814287185669,
      "learning_rate": 0.0002454301075268817,
      "loss": 1.0695,
      "step": 2739
    },
    {
      "epoch": 0.4289292423293676,
      "grad_norm": 2.1607635021209717,
      "learning_rate": 0.00024540566959921795,
      "loss": 1.0842,
      "step": 2740
    },
    {
      "epoch": 0.42908578584846585,
      "grad_norm": 2.969998598098755,
      "learning_rate": 0.00024538123167155426,
      "loss": 1.4593,
      "step": 2741
    },
    {
      "epoch": 0.4292423293675642,
      "grad_norm": 1.807499647140503,
      "learning_rate": 0.0002453567937438905,
      "loss": 0.8874,
      "step": 2742
    },
    {
      "epoch": 0.4293988728866625,
      "grad_norm": 1.7791450023651123,
      "learning_rate": 0.00024533235581622676,
      "loss": 1.5912,
      "step": 2743
    },
    {
      "epoch": 0.4295554164057608,
      "grad_norm": 1.3194993734359741,
      "learning_rate": 0.00024530791788856306,
      "loss": 1.0056,
      "step": 2744
    },
    {
      "epoch": 0.4297119599248591,
      "grad_norm": 2.4051010608673096,
      "learning_rate": 0.00024528347996089926,
      "loss": 1.4255,
      "step": 2745
    },
    {
      "epoch": 0.42986850344395744,
      "grad_norm": 3.256500482559204,
      "learning_rate": 0.00024525904203323557,
      "loss": 1.2331,
      "step": 2746
    },
    {
      "epoch": 0.4300250469630557,
      "grad_norm": 2.3121085166931152,
      "learning_rate": 0.0002452346041055718,
      "loss": 0.845,
      "step": 2747
    },
    {
      "epoch": 0.43018159048215404,
      "grad_norm": 2.94036865234375,
      "learning_rate": 0.00024521016617790807,
      "loss": 0.851,
      "step": 2748
    },
    {
      "epoch": 0.43033813400125237,
      "grad_norm": 2.072420597076416,
      "learning_rate": 0.0002451857282502444,
      "loss": 0.5205,
      "step": 2749
    },
    {
      "epoch": 0.43049467752035064,
      "grad_norm": 2.3945791721343994,
      "learning_rate": 0.0002451612903225806,
      "loss": 0.9723,
      "step": 2750
    },
    {
      "epoch": 0.430651221039449,
      "grad_norm": 0.6630196571350098,
      "learning_rate": 0.0002451368523949169,
      "loss": 0.3922,
      "step": 2751
    },
    {
      "epoch": 0.4308077645585473,
      "grad_norm": 0.6978053450584412,
      "learning_rate": 0.0002451124144672532,
      "loss": 0.4675,
      "step": 2752
    },
    {
      "epoch": 0.4309643080776456,
      "grad_norm": 0.570740818977356,
      "learning_rate": 0.00024508797653958943,
      "loss": 0.3968,
      "step": 2753
    },
    {
      "epoch": 0.4311208515967439,
      "grad_norm": 0.5810028314590454,
      "learning_rate": 0.0002450635386119257,
      "loss": 0.3249,
      "step": 2754
    },
    {
      "epoch": 0.4312773951158422,
      "grad_norm": 1.180005431175232,
      "learning_rate": 0.00024503910068426194,
      "loss": 0.4897,
      "step": 2755
    },
    {
      "epoch": 0.4314339386349405,
      "grad_norm": 0.7174784541130066,
      "learning_rate": 0.00024501466275659824,
      "loss": 0.3243,
      "step": 2756
    },
    {
      "epoch": 0.43159048215403883,
      "grad_norm": 0.7873983979225159,
      "learning_rate": 0.0002449902248289345,
      "loss": 0.5946,
      "step": 2757
    },
    {
      "epoch": 0.4317470256731371,
      "grad_norm": 0.7748199105262756,
      "learning_rate": 0.00024496578690127074,
      "loss": 0.4663,
      "step": 2758
    },
    {
      "epoch": 0.43190356919223544,
      "grad_norm": 0.571399450302124,
      "learning_rate": 0.00024494134897360705,
      "loss": 0.3024,
      "step": 2759
    },
    {
      "epoch": 0.43206011271133377,
      "grad_norm": 0.838038980960846,
      "learning_rate": 0.0002449169110459433,
      "loss": 0.5797,
      "step": 2760
    },
    {
      "epoch": 0.43221665623043204,
      "grad_norm": 0.9804670214653015,
      "learning_rate": 0.00024489247311827955,
      "loss": 0.4022,
      "step": 2761
    },
    {
      "epoch": 0.43237319974953037,
      "grad_norm": 0.7213824391365051,
      "learning_rate": 0.0002448680351906158,
      "loss": 0.3703,
      "step": 2762
    },
    {
      "epoch": 0.4325297432686287,
      "grad_norm": 1.013265609741211,
      "learning_rate": 0.00024484359726295205,
      "loss": 0.3412,
      "step": 2763
    },
    {
      "epoch": 0.43268628678772697,
      "grad_norm": 1.2366046905517578,
      "learning_rate": 0.00024481915933528836,
      "loss": 0.8917,
      "step": 2764
    },
    {
      "epoch": 0.4328428303068253,
      "grad_norm": 0.8313443660736084,
      "learning_rate": 0.0002447947214076246,
      "loss": 0.3694,
      "step": 2765
    },
    {
      "epoch": 0.43299937382592363,
      "grad_norm": 1.9965428113937378,
      "learning_rate": 0.00024477028347996086,
      "loss": 0.7908,
      "step": 2766
    },
    {
      "epoch": 0.4331559173450219,
      "grad_norm": 1.3897510766983032,
      "learning_rate": 0.00024474584555229717,
      "loss": 0.5598,
      "step": 2767
    },
    {
      "epoch": 0.43331246086412023,
      "grad_norm": 1.1880437135696411,
      "learning_rate": 0.0002447214076246334,
      "loss": 0.5628,
      "step": 2768
    },
    {
      "epoch": 0.43346900438321856,
      "grad_norm": 3.2662549018859863,
      "learning_rate": 0.00024469696969696967,
      "loss": 0.9813,
      "step": 2769
    },
    {
      "epoch": 0.43362554790231683,
      "grad_norm": 1.4377611875534058,
      "learning_rate": 0.0002446725317693059,
      "loss": 0.5714,
      "step": 2770
    },
    {
      "epoch": 0.43378209142141516,
      "grad_norm": 2.769331693649292,
      "learning_rate": 0.0002446480938416422,
      "loss": 1.2337,
      "step": 2771
    },
    {
      "epoch": 0.4339386349405135,
      "grad_norm": 1.2153619527816772,
      "learning_rate": 0.0002446236559139785,
      "loss": 0.5278,
      "step": 2772
    },
    {
      "epoch": 0.43409517845961176,
      "grad_norm": 2.0345957279205322,
      "learning_rate": 0.00024459921798631473,
      "loss": 0.7495,
      "step": 2773
    },
    {
      "epoch": 0.4342517219787101,
      "grad_norm": 1.450657844543457,
      "learning_rate": 0.00024457478005865103,
      "loss": 0.6209,
      "step": 2774
    },
    {
      "epoch": 0.43440826549780837,
      "grad_norm": 1.648074746131897,
      "learning_rate": 0.0002445503421309873,
      "loss": 0.6324,
      "step": 2775
    },
    {
      "epoch": 0.4345648090169067,
      "grad_norm": 1.3603814840316772,
      "learning_rate": 0.00024452590420332354,
      "loss": 0.754,
      "step": 2776
    },
    {
      "epoch": 0.434721352536005,
      "grad_norm": 3.350707769393921,
      "learning_rate": 0.00024450146627565984,
      "loss": 0.73,
      "step": 2777
    },
    {
      "epoch": 0.4348778960551033,
      "grad_norm": 2.1769869327545166,
      "learning_rate": 0.00024447702834799604,
      "loss": 0.9358,
      "step": 2778
    },
    {
      "epoch": 0.4350344395742016,
      "grad_norm": 1.703329086303711,
      "learning_rate": 0.00024445259042033235,
      "loss": 0.6444,
      "step": 2779
    },
    {
      "epoch": 0.43519098309329995,
      "grad_norm": 2.654348611831665,
      "learning_rate": 0.0002444281524926686,
      "loss": 0.7654,
      "step": 2780
    },
    {
      "epoch": 0.43534752661239823,
      "grad_norm": 1.5147393941879272,
      "learning_rate": 0.00024440371456500485,
      "loss": 0.9084,
      "step": 2781
    },
    {
      "epoch": 0.43550407013149656,
      "grad_norm": 2.269134759902954,
      "learning_rate": 0.00024437927663734115,
      "loss": 1.033,
      "step": 2782
    },
    {
      "epoch": 0.4356606136505949,
      "grad_norm": 1.4680894613265991,
      "learning_rate": 0.0002443548387096774,
      "loss": 1.0404,
      "step": 2783
    },
    {
      "epoch": 0.43581715716969316,
      "grad_norm": 1.8765166997909546,
      "learning_rate": 0.00024433040078201366,
      "loss": 1.4057,
      "step": 2784
    },
    {
      "epoch": 0.4359737006887915,
      "grad_norm": 1.2388074398040771,
      "learning_rate": 0.00024430596285434996,
      "loss": 0.7339,
      "step": 2785
    },
    {
      "epoch": 0.4361302442078898,
      "grad_norm": 1.540697693824768,
      "learning_rate": 0.0002442815249266862,
      "loss": 0.8482,
      "step": 2786
    },
    {
      "epoch": 0.4362867877269881,
      "grad_norm": 2.0315816402435303,
      "learning_rate": 0.00024425708699902246,
      "loss": 0.9921,
      "step": 2787
    },
    {
      "epoch": 0.4364433312460864,
      "grad_norm": 2.7498373985290527,
      "learning_rate": 0.0002442326490713587,
      "loss": 1.6373,
      "step": 2788
    },
    {
      "epoch": 0.43659987476518475,
      "grad_norm": 2.1657485961914062,
      "learning_rate": 0.000244208211143695,
      "loss": 1.6076,
      "step": 2789
    },
    {
      "epoch": 0.436756418284283,
      "grad_norm": 3.260566234588623,
      "learning_rate": 0.00024418377321603127,
      "loss": 1.1333,
      "step": 2790
    },
    {
      "epoch": 0.43691296180338135,
      "grad_norm": 2.2284598350524902,
      "learning_rate": 0.0002441593352883675,
      "loss": 0.9482,
      "step": 2791
    },
    {
      "epoch": 0.4370695053224796,
      "grad_norm": 2.1188924312591553,
      "learning_rate": 0.0002441348973607038,
      "loss": 1.0199,
      "step": 2792
    },
    {
      "epoch": 0.43722604884157795,
      "grad_norm": 5.20643949508667,
      "learning_rate": 0.00024411045943304005,
      "loss": 1.8085,
      "step": 2793
    },
    {
      "epoch": 0.4373825923606763,
      "grad_norm": 4.37985372543335,
      "learning_rate": 0.00024408602150537633,
      "loss": 1.2478,
      "step": 2794
    },
    {
      "epoch": 0.43753913587977455,
      "grad_norm": 1.3027615547180176,
      "learning_rate": 0.0002440615835777126,
      "loss": 1.2622,
      "step": 2795
    },
    {
      "epoch": 0.4376956793988729,
      "grad_norm": 2.000991106033325,
      "learning_rate": 0.00024403714565004886,
      "loss": 0.8131,
      "step": 2796
    },
    {
      "epoch": 0.4378522229179712,
      "grad_norm": 2.1807901859283447,
      "learning_rate": 0.00024401270772238514,
      "loss": 0.7618,
      "step": 2797
    },
    {
      "epoch": 0.4380087664370695,
      "grad_norm": 2.6867918968200684,
      "learning_rate": 0.0002439882697947214,
      "loss": 0.6396,
      "step": 2798
    },
    {
      "epoch": 0.4381653099561678,
      "grad_norm": 1.687910795211792,
      "learning_rate": 0.00024396383186705764,
      "loss": 1.2688,
      "step": 2799
    },
    {
      "epoch": 0.43832185347526614,
      "grad_norm": 2.035445213317871,
      "learning_rate": 0.00024393939393939392,
      "loss": 1.1152,
      "step": 2800
    },
    {
      "epoch": 0.4384783969943644,
      "grad_norm": 1.4942790269851685,
      "learning_rate": 0.0002439149560117302,
      "loss": 0.6478,
      "step": 2801
    },
    {
      "epoch": 0.43863494051346275,
      "grad_norm": 0.5058338642120361,
      "learning_rate": 0.00024389051808406645,
      "loss": 0.257,
      "step": 2802
    },
    {
      "epoch": 0.4387914840325611,
      "grad_norm": 0.5301083922386169,
      "learning_rate": 0.00024386608015640273,
      "loss": 0.3152,
      "step": 2803
    },
    {
      "epoch": 0.43894802755165935,
      "grad_norm": 3.6920485496520996,
      "learning_rate": 0.000243841642228739,
      "loss": 1.2555,
      "step": 2804
    },
    {
      "epoch": 0.4391045710707577,
      "grad_norm": 0.8687499761581421,
      "learning_rate": 0.00024381720430107523,
      "loss": 0.4101,
      "step": 2805
    },
    {
      "epoch": 0.439261114589856,
      "grad_norm": 0.5601373314857483,
      "learning_rate": 0.0002437927663734115,
      "loss": 0.4841,
      "step": 2806
    },
    {
      "epoch": 0.4394176581089543,
      "grad_norm": 0.5990231037139893,
      "learning_rate": 0.0002437683284457478,
      "loss": 0.3959,
      "step": 2807
    },
    {
      "epoch": 0.4395742016280526,
      "grad_norm": 0.8136112093925476,
      "learning_rate": 0.00024374389051808404,
      "loss": 0.5858,
      "step": 2808
    },
    {
      "epoch": 0.4397307451471509,
      "grad_norm": 0.8082829713821411,
      "learning_rate": 0.00024371945259042032,
      "loss": 0.4747,
      "step": 2809
    },
    {
      "epoch": 0.4398872886662492,
      "grad_norm": 0.751895010471344,
      "learning_rate": 0.00024369501466275657,
      "loss": 0.4398,
      "step": 2810
    },
    {
      "epoch": 0.44004383218534754,
      "grad_norm": 0.7629719972610474,
      "learning_rate": 0.00024367057673509285,
      "loss": 0.3338,
      "step": 2811
    },
    {
      "epoch": 0.4402003757044458,
      "grad_norm": 0.9700342416763306,
      "learning_rate": 0.00024364613880742912,
      "loss": 0.5572,
      "step": 2812
    },
    {
      "epoch": 0.44035691922354414,
      "grad_norm": 1.136712670326233,
      "learning_rate": 0.00024362170087976535,
      "loss": 0.5201,
      "step": 2813
    },
    {
      "epoch": 0.44051346274264247,
      "grad_norm": 0.9968575239181519,
      "learning_rate": 0.00024359726295210163,
      "loss": 0.5493,
      "step": 2814
    },
    {
      "epoch": 0.44067000626174074,
      "grad_norm": 1.318066120147705,
      "learning_rate": 0.0002435728250244379,
      "loss": 0.5117,
      "step": 2815
    },
    {
      "epoch": 0.44082654978083907,
      "grad_norm": 1.940822720527649,
      "learning_rate": 0.00024354838709677416,
      "loss": 0.6577,
      "step": 2816
    },
    {
      "epoch": 0.4409830932999374,
      "grad_norm": 1.8296147584915161,
      "learning_rate": 0.00024352394916911043,
      "loss": 0.7105,
      "step": 2817
    },
    {
      "epoch": 0.4411396368190357,
      "grad_norm": 2.1246185302734375,
      "learning_rate": 0.0002434995112414467,
      "loss": 0.8139,
      "step": 2818
    },
    {
      "epoch": 0.441296180338134,
      "grad_norm": 1.4536235332489014,
      "learning_rate": 0.00024347507331378296,
      "loss": 0.6666,
      "step": 2819
    },
    {
      "epoch": 0.44145272385723233,
      "grad_norm": 1.9893560409545898,
      "learning_rate": 0.00024345063538611924,
      "loss": 0.7701,
      "step": 2820
    },
    {
      "epoch": 0.4416092673763306,
      "grad_norm": 2.051508903503418,
      "learning_rate": 0.0002434261974584555,
      "loss": 0.4967,
      "step": 2821
    },
    {
      "epoch": 0.44176581089542893,
      "grad_norm": 3.648225784301758,
      "learning_rate": 0.00024340175953079175,
      "loss": 0.7787,
      "step": 2822
    },
    {
      "epoch": 0.44192235441452726,
      "grad_norm": 1.885310411453247,
      "learning_rate": 0.00024337732160312802,
      "loss": 0.5775,
      "step": 2823
    },
    {
      "epoch": 0.44207889793362554,
      "grad_norm": 2.8299691677093506,
      "learning_rate": 0.0002433528836754643,
      "loss": 0.9215,
      "step": 2824
    },
    {
      "epoch": 0.44223544145272387,
      "grad_norm": 1.727311134338379,
      "learning_rate": 0.00024332844574780055,
      "loss": 0.8729,
      "step": 2825
    },
    {
      "epoch": 0.4423919849718222,
      "grad_norm": 3.334357738494873,
      "learning_rate": 0.00024330400782013683,
      "loss": 0.7647,
      "step": 2826
    },
    {
      "epoch": 0.44254852849092047,
      "grad_norm": 2.5358595848083496,
      "learning_rate": 0.0002432795698924731,
      "loss": 0.808,
      "step": 2827
    },
    {
      "epoch": 0.4427050720100188,
      "grad_norm": 1.919858694076538,
      "learning_rate": 0.00024325513196480933,
      "loss": 0.8237,
      "step": 2828
    },
    {
      "epoch": 0.44286161552911707,
      "grad_norm": 2.2071852684020996,
      "learning_rate": 0.0002432306940371456,
      "loss": 0.7848,
      "step": 2829
    },
    {
      "epoch": 0.4430181590482154,
      "grad_norm": 2.1200320720672607,
      "learning_rate": 0.0002432062561094819,
      "loss": 0.8957,
      "step": 2830
    },
    {
      "epoch": 0.4431747025673137,
      "grad_norm": 2.553398609161377,
      "learning_rate": 0.00024318181818181814,
      "loss": 1.4184,
      "step": 2831
    },
    {
      "epoch": 0.443331246086412,
      "grad_norm": 1.989119529724121,
      "learning_rate": 0.00024315738025415442,
      "loss": 1.1927,
      "step": 2832
    },
    {
      "epoch": 0.44348778960551033,
      "grad_norm": 1.7680295705795288,
      "learning_rate": 0.0002431329423264907,
      "loss": 0.7261,
      "step": 2833
    },
    {
      "epoch": 0.44364433312460866,
      "grad_norm": 3.1489715576171875,
      "learning_rate": 0.00024310850439882695,
      "loss": 1.2965,
      "step": 2834
    },
    {
      "epoch": 0.44380087664370693,
      "grad_norm": 2.7406511306762695,
      "learning_rate": 0.00024308406647116323,
      "loss": 0.9612,
      "step": 2835
    },
    {
      "epoch": 0.44395742016280526,
      "grad_norm": 2.736985921859741,
      "learning_rate": 0.0002430596285434995,
      "loss": 1.1864,
      "step": 2836
    },
    {
      "epoch": 0.4441139636819036,
      "grad_norm": 2.205655097961426,
      "learning_rate": 0.00024303519061583573,
      "loss": 1.6187,
      "step": 2837
    },
    {
      "epoch": 0.44427050720100186,
      "grad_norm": 1.7284249067306519,
      "learning_rate": 0.000243010752688172,
      "loss": 0.9562,
      "step": 2838
    },
    {
      "epoch": 0.4444270507201002,
      "grad_norm": 2.0134189128875732,
      "learning_rate": 0.0002429863147605083,
      "loss": 1.3006,
      "step": 2839
    },
    {
      "epoch": 0.4445835942391985,
      "grad_norm": 2.37020206451416,
      "learning_rate": 0.00024296187683284454,
      "loss": 1.5716,
      "step": 2840
    },
    {
      "epoch": 0.4447401377582968,
      "grad_norm": 2.1553592681884766,
      "learning_rate": 0.00024293743890518082,
      "loss": 1.1951,
      "step": 2841
    },
    {
      "epoch": 0.4448966812773951,
      "grad_norm": 2.4819650650024414,
      "learning_rate": 0.0002429130009775171,
      "loss": 0.8878,
      "step": 2842
    },
    {
      "epoch": 0.44505322479649345,
      "grad_norm": 2.196474552154541,
      "learning_rate": 0.00024288856304985335,
      "loss": 1.4834,
      "step": 2843
    },
    {
      "epoch": 0.4452097683155917,
      "grad_norm": 3.71829891204834,
      "learning_rate": 0.00024286412512218963,
      "loss": 1.1742,
      "step": 2844
    },
    {
      "epoch": 0.44536631183469005,
      "grad_norm": 2.282411575317383,
      "learning_rate": 0.00024283968719452588,
      "loss": 1.3603,
      "step": 2845
    },
    {
      "epoch": 0.4455228553537883,
      "grad_norm": 1.549609899520874,
      "learning_rate": 0.00024281524926686213,
      "loss": 0.7373,
      "step": 2846
    },
    {
      "epoch": 0.44567939887288666,
      "grad_norm": 0.8783185482025146,
      "learning_rate": 0.0002427908113391984,
      "loss": 0.435,
      "step": 2847
    },
    {
      "epoch": 0.445835942391985,
      "grad_norm": 4.473077774047852,
      "learning_rate": 0.00024276637341153468,
      "loss": 1.096,
      "step": 2848
    },
    {
      "epoch": 0.44599248591108326,
      "grad_norm": 1.7083899974822998,
      "learning_rate": 0.00024274193548387094,
      "loss": 0.9034,
      "step": 2849
    },
    {
      "epoch": 0.4461490294301816,
      "grad_norm": 1.6015340089797974,
      "learning_rate": 0.00024271749755620721,
      "loss": 0.8184,
      "step": 2850
    },
    {
      "epoch": 0.4463055729492799,
      "grad_norm": 0.5668478608131409,
      "learning_rate": 0.0002426930596285435,
      "loss": 0.3711,
      "step": 2851
    },
    {
      "epoch": 0.4464621164683782,
      "grad_norm": 0.603583812713623,
      "learning_rate": 0.00024266862170087972,
      "loss": 0.284,
      "step": 2852
    },
    {
      "epoch": 0.4466186599874765,
      "grad_norm": 0.8786858916282654,
      "learning_rate": 0.000242644183773216,
      "loss": 0.2688,
      "step": 2853
    },
    {
      "epoch": 0.44677520350657485,
      "grad_norm": 0.5734667778015137,
      "learning_rate": 0.00024261974584555227,
      "loss": 0.3223,
      "step": 2854
    },
    {
      "epoch": 0.4469317470256731,
      "grad_norm": 0.926365077495575,
      "learning_rate": 0.00024259530791788852,
      "loss": 0.44,
      "step": 2855
    },
    {
      "epoch": 0.44708829054477145,
      "grad_norm": 0.661888062953949,
      "learning_rate": 0.0002425708699902248,
      "loss": 0.3443,
      "step": 2856
    },
    {
      "epoch": 0.4472448340638698,
      "grad_norm": 0.6842008233070374,
      "learning_rate": 0.00024254643206256108,
      "loss": 0.4118,
      "step": 2857
    },
    {
      "epoch": 0.44740137758296805,
      "grad_norm": 0.7839653491973877,
      "learning_rate": 0.00024252199413489733,
      "loss": 0.3634,
      "step": 2858
    },
    {
      "epoch": 0.4475579211020664,
      "grad_norm": 1.1856542825698853,
      "learning_rate": 0.0002424975562072336,
      "loss": 0.6146,
      "step": 2859
    },
    {
      "epoch": 0.4477144646211647,
      "grad_norm": 1.2870744466781616,
      "learning_rate": 0.0002424731182795699,
      "loss": 0.4388,
      "step": 2860
    },
    {
      "epoch": 0.447871008140263,
      "grad_norm": 0.8525822162628174,
      "learning_rate": 0.0002424486803519061,
      "loss": 0.5828,
      "step": 2861
    },
    {
      "epoch": 0.4480275516593613,
      "grad_norm": 0.8361726403236389,
      "learning_rate": 0.0002424242424242424,
      "loss": 0.4072,
      "step": 2862
    },
    {
      "epoch": 0.4481840951784596,
      "grad_norm": 3.588017225265503,
      "learning_rate": 0.00024239980449657867,
      "loss": 0.7043,
      "step": 2863
    },
    {
      "epoch": 0.4483406386975579,
      "grad_norm": 0.9456626176834106,
      "learning_rate": 0.00024237536656891492,
      "loss": 0.6288,
      "step": 2864
    },
    {
      "epoch": 0.44849718221665624,
      "grad_norm": 0.8043878078460693,
      "learning_rate": 0.0002423509286412512,
      "loss": 0.3946,
      "step": 2865
    },
    {
      "epoch": 0.4486537257357545,
      "grad_norm": 1.2611615657806396,
      "learning_rate": 0.00024232649071358748,
      "loss": 0.5556,
      "step": 2866
    },
    {
      "epoch": 0.44881026925485284,
      "grad_norm": 1.1336512565612793,
      "learning_rate": 0.00024230205278592373,
      "loss": 0.6662,
      "step": 2867
    },
    {
      "epoch": 0.4489668127739512,
      "grad_norm": 1.0719926357269287,
      "learning_rate": 0.00024227761485826,
      "loss": 0.4993,
      "step": 2868
    },
    {
      "epoch": 0.44912335629304945,
      "grad_norm": 1.3766525983810425,
      "learning_rate": 0.00024225317693059626,
      "loss": 0.5825,
      "step": 2869
    },
    {
      "epoch": 0.4492798998121478,
      "grad_norm": 1.647732138633728,
      "learning_rate": 0.0002422287390029325,
      "loss": 0.5897,
      "step": 2870
    },
    {
      "epoch": 0.4494364433312461,
      "grad_norm": 2.6894371509552,
      "learning_rate": 0.0002422043010752688,
      "loss": 0.8135,
      "step": 2871
    },
    {
      "epoch": 0.4495929868503444,
      "grad_norm": 2.326774835586548,
      "learning_rate": 0.00024217986314760507,
      "loss": 0.6992,
      "step": 2872
    },
    {
      "epoch": 0.4497495303694427,
      "grad_norm": 0.996810257434845,
      "learning_rate": 0.00024215542521994132,
      "loss": 0.4537,
      "step": 2873
    },
    {
      "epoch": 0.44990607388854104,
      "grad_norm": 1.5761237144470215,
      "learning_rate": 0.0002421309872922776,
      "loss": 0.6694,
      "step": 2874
    },
    {
      "epoch": 0.4500626174076393,
      "grad_norm": 3.633239507675171,
      "learning_rate": 0.00024210654936461387,
      "loss": 0.9042,
      "step": 2875
    },
    {
      "epoch": 0.45021916092673764,
      "grad_norm": 2.0842907428741455,
      "learning_rate": 0.0002420821114369501,
      "loss": 1.0301,
      "step": 2876
    },
    {
      "epoch": 0.45037570444583597,
      "grad_norm": 1.3994956016540527,
      "learning_rate": 0.00024205767350928638,
      "loss": 0.6357,
      "step": 2877
    },
    {
      "epoch": 0.45053224796493424,
      "grad_norm": 1.4533970355987549,
      "learning_rate": 0.00024203323558162266,
      "loss": 0.6302,
      "step": 2878
    },
    {
      "epoch": 0.45068879148403257,
      "grad_norm": 1.189319133758545,
      "learning_rate": 0.0002420087976539589,
      "loss": 0.5899,
      "step": 2879
    },
    {
      "epoch": 0.45084533500313084,
      "grad_norm": 3.1489529609680176,
      "learning_rate": 0.00024198435972629519,
      "loss": 0.931,
      "step": 2880
    },
    {
      "epoch": 0.45100187852222917,
      "grad_norm": 1.8433654308319092,
      "learning_rate": 0.00024195992179863146,
      "loss": 0.4861,
      "step": 2881
    },
    {
      "epoch": 0.4511584220413275,
      "grad_norm": 3.465445041656494,
      "learning_rate": 0.00024193548387096771,
      "loss": 0.6557,
      "step": 2882
    },
    {
      "epoch": 0.4513149655604258,
      "grad_norm": 2.037400484085083,
      "learning_rate": 0.000241911045943304,
      "loss": 1.3114,
      "step": 2883
    },
    {
      "epoch": 0.4514715090795241,
      "grad_norm": 1.4593843221664429,
      "learning_rate": 0.00024188660801564027,
      "loss": 0.81,
      "step": 2884
    },
    {
      "epoch": 0.45162805259862243,
      "grad_norm": 2.352095365524292,
      "learning_rate": 0.0002418621700879765,
      "loss": 0.6537,
      "step": 2885
    },
    {
      "epoch": 0.4517845961177207,
      "grad_norm": 2.0871784687042236,
      "learning_rate": 0.00024183773216031277,
      "loss": 1.0747,
      "step": 2886
    },
    {
      "epoch": 0.45194113963681903,
      "grad_norm": 3.217430591583252,
      "learning_rate": 0.00024181329423264905,
      "loss": 0.8365,
      "step": 2887
    },
    {
      "epoch": 0.45209768315591736,
      "grad_norm": 3.391012668609619,
      "learning_rate": 0.0002417888563049853,
      "loss": 0.7699,
      "step": 2888
    },
    {
      "epoch": 0.45225422667501564,
      "grad_norm": 4.206727981567383,
      "learning_rate": 0.00024176441837732158,
      "loss": 1.3047,
      "step": 2889
    },
    {
      "epoch": 0.45241077019411396,
      "grad_norm": 2.0663089752197266,
      "learning_rate": 0.00024173998044965786,
      "loss": 0.8944,
      "step": 2890
    },
    {
      "epoch": 0.4525673137132123,
      "grad_norm": 3.9211676120758057,
      "learning_rate": 0.0002417155425219941,
      "loss": 2.0434,
      "step": 2891
    },
    {
      "epoch": 0.45272385723231057,
      "grad_norm": 1.6745312213897705,
      "learning_rate": 0.00024169110459433036,
      "loss": 1.4715,
      "step": 2892
    },
    {
      "epoch": 0.4528804007514089,
      "grad_norm": 2.4523682594299316,
      "learning_rate": 0.00024166666666666664,
      "loss": 0.9836,
      "step": 2893
    },
    {
      "epoch": 0.4530369442705072,
      "grad_norm": 1.776292085647583,
      "learning_rate": 0.0002416422287390029,
      "loss": 1.4415,
      "step": 2894
    },
    {
      "epoch": 0.4531934877896055,
      "grad_norm": 2.345961570739746,
      "learning_rate": 0.00024161779081133917,
      "loss": 1.3364,
      "step": 2895
    },
    {
      "epoch": 0.4533500313087038,
      "grad_norm": 2.3358938694000244,
      "learning_rate": 0.00024159335288367545,
      "loss": 1.2469,
      "step": 2896
    },
    {
      "epoch": 0.45350657482780216,
      "grad_norm": 2.8481311798095703,
      "learning_rate": 0.0002415689149560117,
      "loss": 0.6862,
      "step": 2897
    },
    {
      "epoch": 0.45366311834690043,
      "grad_norm": 3.2002439498901367,
      "learning_rate": 0.00024154447702834798,
      "loss": 0.8734,
      "step": 2898
    },
    {
      "epoch": 0.45381966186599876,
      "grad_norm": 3.30496883392334,
      "learning_rate": 0.00024152003910068426,
      "loss": 1.1847,
      "step": 2899
    },
    {
      "epoch": 0.45397620538509703,
      "grad_norm": 2.5524821281433105,
      "learning_rate": 0.00024149560117302048,
      "loss": 1.1897,
      "step": 2900
    },
    {
      "epoch": 0.45413274890419536,
      "grad_norm": 0.7851200103759766,
      "learning_rate": 0.00024147116324535676,
      "loss": 0.5016,
      "step": 2901
    },
    {
      "epoch": 0.4542892924232937,
      "grad_norm": 0.8391879796981812,
      "learning_rate": 0.00024144672531769304,
      "loss": 0.4045,
      "step": 2902
    },
    {
      "epoch": 0.45444583594239196,
      "grad_norm": 1.0512319803237915,
      "learning_rate": 0.0002414222873900293,
      "loss": 0.4671,
      "step": 2903
    },
    {
      "epoch": 0.4546023794614903,
      "grad_norm": 0.7638587951660156,
      "learning_rate": 0.00024139784946236557,
      "loss": 0.4057,
      "step": 2904
    },
    {
      "epoch": 0.4547589229805886,
      "grad_norm": 0.5966224670410156,
      "learning_rate": 0.00024137341153470185,
      "loss": 0.3474,
      "step": 2905
    },
    {
      "epoch": 0.4549154664996869,
      "grad_norm": 0.6843605041503906,
      "learning_rate": 0.0002413489736070381,
      "loss": 0.259,
      "step": 2906
    },
    {
      "epoch": 0.4550720100187852,
      "grad_norm": 0.8733233213424683,
      "learning_rate": 0.00024132453567937438,
      "loss": 0.5368,
      "step": 2907
    },
    {
      "epoch": 0.45522855353788355,
      "grad_norm": 0.8052546977996826,
      "learning_rate": 0.00024130009775171065,
      "loss": 0.4997,
      "step": 2908
    },
    {
      "epoch": 0.4553850970569818,
      "grad_norm": 0.7011566162109375,
      "learning_rate": 0.00024127565982404688,
      "loss": 0.4456,
      "step": 2909
    },
    {
      "epoch": 0.45554164057608015,
      "grad_norm": 2.446593999862671,
      "learning_rate": 0.00024125122189638316,
      "loss": 0.4917,
      "step": 2910
    },
    {
      "epoch": 0.4556981840951785,
      "grad_norm": 0.5837518572807312,
      "learning_rate": 0.00024122678396871943,
      "loss": 0.4428,
      "step": 2911
    },
    {
      "epoch": 0.45585472761427676,
      "grad_norm": 0.7179192900657654,
      "learning_rate": 0.00024120234604105569,
      "loss": 0.4092,
      "step": 2912
    },
    {
      "epoch": 0.4560112711333751,
      "grad_norm": 0.9555642008781433,
      "learning_rate": 0.00024117790811339196,
      "loss": 0.4673,
      "step": 2913
    },
    {
      "epoch": 0.4561678146524734,
      "grad_norm": 0.891258955001831,
      "learning_rate": 0.00024115347018572824,
      "loss": 0.3904,
      "step": 2914
    },
    {
      "epoch": 0.4563243581715717,
      "grad_norm": 0.9778416752815247,
      "learning_rate": 0.0002411290322580645,
      "loss": 0.4525,
      "step": 2915
    },
    {
      "epoch": 0.45648090169067,
      "grad_norm": 1.230553150177002,
      "learning_rate": 0.00024110459433040074,
      "loss": 0.3054,
      "step": 2916
    },
    {
      "epoch": 0.4566374452097683,
      "grad_norm": 1.3313859701156616,
      "learning_rate": 0.00024108015640273702,
      "loss": 0.3882,
      "step": 2917
    },
    {
      "epoch": 0.4567939887288666,
      "grad_norm": 1.3717501163482666,
      "learning_rate": 0.00024105571847507327,
      "loss": 0.6825,
      "step": 2918
    },
    {
      "epoch": 0.45695053224796495,
      "grad_norm": 2.073382616043091,
      "learning_rate": 0.00024103128054740955,
      "loss": 0.5989,
      "step": 2919
    },
    {
      "epoch": 0.4571070757670632,
      "grad_norm": 1.1343777179718018,
      "learning_rate": 0.00024100684261974583,
      "loss": 0.5252,
      "step": 2920
    },
    {
      "epoch": 0.45726361928616155,
      "grad_norm": 1.8335095643997192,
      "learning_rate": 0.00024098240469208208,
      "loss": 0.4679,
      "step": 2921
    },
    {
      "epoch": 0.4574201628052599,
      "grad_norm": 2.010408639907837,
      "learning_rate": 0.00024095796676441836,
      "loss": 0.6568,
      "step": 2922
    },
    {
      "epoch": 0.45757670632435815,
      "grad_norm": 1.3354651927947998,
      "learning_rate": 0.00024093352883675464,
      "loss": 0.545,
      "step": 2923
    },
    {
      "epoch": 0.4577332498434565,
      "grad_norm": 2.0071985721588135,
      "learning_rate": 0.00024090909090909086,
      "loss": 0.9399,
      "step": 2924
    },
    {
      "epoch": 0.4578897933625548,
      "grad_norm": 3.2481436729431152,
      "learning_rate": 0.00024088465298142714,
      "loss": 1.0217,
      "step": 2925
    },
    {
      "epoch": 0.4580463368816531,
      "grad_norm": 2.576655387878418,
      "learning_rate": 0.00024086021505376342,
      "loss": 0.5155,
      "step": 2926
    },
    {
      "epoch": 0.4582028804007514,
      "grad_norm": 1.2466546297073364,
      "learning_rate": 0.00024083577712609967,
      "loss": 0.6947,
      "step": 2927
    },
    {
      "epoch": 0.45835942391984974,
      "grad_norm": 1.3399920463562012,
      "learning_rate": 0.00024081133919843595,
      "loss": 0.6762,
      "step": 2928
    },
    {
      "epoch": 0.458515967438948,
      "grad_norm": 1.9821996688842773,
      "learning_rate": 0.00024078690127077223,
      "loss": 0.8415,
      "step": 2929
    },
    {
      "epoch": 0.45867251095804634,
      "grad_norm": 2.0268774032592773,
      "learning_rate": 0.00024076246334310848,
      "loss": 0.7196,
      "step": 2930
    },
    {
      "epoch": 0.45882905447714467,
      "grad_norm": 1.9509254693984985,
      "learning_rate": 0.00024073802541544476,
      "loss": 0.7176,
      "step": 2931
    },
    {
      "epoch": 0.45898559799624294,
      "grad_norm": 1.5287137031555176,
      "learning_rate": 0.00024071358748778104,
      "loss": 0.9125,
      "step": 2932
    },
    {
      "epoch": 0.4591421415153413,
      "grad_norm": 2.1340198516845703,
      "learning_rate": 0.00024068914956011726,
      "loss": 0.7953,
      "step": 2933
    },
    {
      "epoch": 0.45929868503443955,
      "grad_norm": 1.94748055934906,
      "learning_rate": 0.00024066471163245354,
      "loss": 0.9377,
      "step": 2934
    },
    {
      "epoch": 0.4594552285535379,
      "grad_norm": 1.8639601469039917,
      "learning_rate": 0.00024064027370478982,
      "loss": 0.9098,
      "step": 2935
    },
    {
      "epoch": 0.4596117720726362,
      "grad_norm": 1.7481262683868408,
      "learning_rate": 0.00024061583577712607,
      "loss": 1.1671,
      "step": 2936
    },
    {
      "epoch": 0.4597683155917345,
      "grad_norm": 2.4911282062530518,
      "learning_rate": 0.00024059139784946235,
      "loss": 1.5963,
      "step": 2937
    },
    {
      "epoch": 0.4599248591108328,
      "grad_norm": 3.1671178340911865,
      "learning_rate": 0.00024056695992179862,
      "loss": 0.8903,
      "step": 2938
    },
    {
      "epoch": 0.46008140262993114,
      "grad_norm": 2.2914581298828125,
      "learning_rate": 0.00024054252199413488,
      "loss": 1.2117,
      "step": 2939
    },
    {
      "epoch": 0.4602379461490294,
      "grad_norm": 2.8251593112945557,
      "learning_rate": 0.00024051808406647113,
      "loss": 1.4033,
      "step": 2940
    },
    {
      "epoch": 0.46039448966812774,
      "grad_norm": 2.4043309688568115,
      "learning_rate": 0.0002404936461388074,
      "loss": 0.9579,
      "step": 2941
    },
    {
      "epoch": 0.46055103318722607,
      "grad_norm": 2.894033908843994,
      "learning_rate": 0.00024046920821114366,
      "loss": 0.833,
      "step": 2942
    },
    {
      "epoch": 0.46070757670632434,
      "grad_norm": 2.439354658126831,
      "learning_rate": 0.00024044477028347994,
      "loss": 1.7688,
      "step": 2943
    },
    {
      "epoch": 0.46086412022542267,
      "grad_norm": 3.430408000946045,
      "learning_rate": 0.00024042033235581621,
      "loss": 1.5622,
      "step": 2944
    },
    {
      "epoch": 0.461020663744521,
      "grad_norm": 1.8975231647491455,
      "learning_rate": 0.00024039589442815246,
      "loss": 1.3692,
      "step": 2945
    },
    {
      "epoch": 0.46117720726361927,
      "grad_norm": 1.8430036306381226,
      "learning_rate": 0.00024037145650048874,
      "loss": 1.1524,
      "step": 2946
    },
    {
      "epoch": 0.4613337507827176,
      "grad_norm": 3.267629861831665,
      "learning_rate": 0.00024034701857282502,
      "loss": 0.8915,
      "step": 2947
    },
    {
      "epoch": 0.46149029430181593,
      "grad_norm": 2.6302919387817383,
      "learning_rate": 0.00024032258064516125,
      "loss": 1.2047,
      "step": 2948
    },
    {
      "epoch": 0.4616468378209142,
      "grad_norm": 2.9128334522247314,
      "learning_rate": 0.00024029814271749752,
      "loss": 1.3952,
      "step": 2949
    },
    {
      "epoch": 0.46180338134001253,
      "grad_norm": 1.4925674200057983,
      "learning_rate": 0.0002402737047898338,
      "loss": 1.4002,
      "step": 2950
    },
    {
      "epoch": 0.46195992485911086,
      "grad_norm": 0.5663087964057922,
      "learning_rate": 0.00024024926686217005,
      "loss": 0.3536,
      "step": 2951
    },
    {
      "epoch": 0.46211646837820913,
      "grad_norm": 0.5711171627044678,
      "learning_rate": 0.00024022482893450633,
      "loss": 0.4303,
      "step": 2952
    },
    {
      "epoch": 0.46227301189730746,
      "grad_norm": 0.7608370184898376,
      "learning_rate": 0.0002402003910068426,
      "loss": 0.4472,
      "step": 2953
    },
    {
      "epoch": 0.46242955541640574,
      "grad_norm": 0.5618414282798767,
      "learning_rate": 0.00024017595307917886,
      "loss": 0.4728,
      "step": 2954
    },
    {
      "epoch": 0.46258609893550406,
      "grad_norm": 0.5996711254119873,
      "learning_rate": 0.00024015151515151514,
      "loss": 0.4194,
      "step": 2955
    },
    {
      "epoch": 0.4627426424546024,
      "grad_norm": 0.6966267228126526,
      "learning_rate": 0.00024012707722385142,
      "loss": 0.4361,
      "step": 2956
    },
    {
      "epoch": 0.46289918597370067,
      "grad_norm": 0.6041519045829773,
      "learning_rate": 0.00024010263929618764,
      "loss": 0.3586,
      "step": 2957
    },
    {
      "epoch": 0.463055729492799,
      "grad_norm": 0.6143047213554382,
      "learning_rate": 0.00024007820136852392,
      "loss": 0.3958,
      "step": 2958
    },
    {
      "epoch": 0.4632122730118973,
      "grad_norm": 0.727200448513031,
      "learning_rate": 0.0002400537634408602,
      "loss": 0.4222,
      "step": 2959
    },
    {
      "epoch": 0.4633688165309956,
      "grad_norm": 0.7016879320144653,
      "learning_rate": 0.00024002932551319645,
      "loss": 0.3269,
      "step": 2960
    },
    {
      "epoch": 0.4635253600500939,
      "grad_norm": 1.3324402570724487,
      "learning_rate": 0.00024000488758553273,
      "loss": 0.4065,
      "step": 2961
    },
    {
      "epoch": 0.46368190356919226,
      "grad_norm": 0.8815485835075378,
      "learning_rate": 0.000239980449657869,
      "loss": 0.4625,
      "step": 2962
    },
    {
      "epoch": 0.46383844708829053,
      "grad_norm": 1.0616453886032104,
      "learning_rate": 0.00023995601173020523,
      "loss": 0.5971,
      "step": 2963
    },
    {
      "epoch": 0.46399499060738886,
      "grad_norm": 0.8224049806594849,
      "learning_rate": 0.0002399315738025415,
      "loss": 0.4368,
      "step": 2964
    },
    {
      "epoch": 0.4641515341264872,
      "grad_norm": 1.420790195465088,
      "learning_rate": 0.0002399071358748778,
      "loss": 0.5855,
      "step": 2965
    },
    {
      "epoch": 0.46430807764558546,
      "grad_norm": 1.2988914251327515,
      "learning_rate": 0.00023988269794721404,
      "loss": 0.6298,
      "step": 2966
    },
    {
      "epoch": 0.4644646211646838,
      "grad_norm": 2.2684569358825684,
      "learning_rate": 0.00023985826001955032,
      "loss": 0.6326,
      "step": 2967
    },
    {
      "epoch": 0.4646211646837821,
      "grad_norm": 2.2783150672912598,
      "learning_rate": 0.0002398338220918866,
      "loss": 0.7841,
      "step": 2968
    },
    {
      "epoch": 0.4647777082028804,
      "grad_norm": 1.206459879875183,
      "learning_rate": 0.00023980938416422285,
      "loss": 0.7821,
      "step": 2969
    },
    {
      "epoch": 0.4649342517219787,
      "grad_norm": 3.7730114459991455,
      "learning_rate": 0.00023978494623655913,
      "loss": 1.0646,
      "step": 2970
    },
    {
      "epoch": 0.465090795241077,
      "grad_norm": 1.3835830688476562,
      "learning_rate": 0.0002397605083088954,
      "loss": 0.452,
      "step": 2971
    },
    {
      "epoch": 0.4652473387601753,
      "grad_norm": 2.1477081775665283,
      "learning_rate": 0.00023973607038123163,
      "loss": 0.9363,
      "step": 2972
    },
    {
      "epoch": 0.46540388227927365,
      "grad_norm": 1.1559685468673706,
      "learning_rate": 0.0002397116324535679,
      "loss": 0.5786,
      "step": 2973
    },
    {
      "epoch": 0.4655604257983719,
      "grad_norm": 2.111488103866577,
      "learning_rate": 0.00023968719452590418,
      "loss": 0.5976,
      "step": 2974
    },
    {
      "epoch": 0.46571696931747025,
      "grad_norm": 1.2530162334442139,
      "learning_rate": 0.00023966275659824044,
      "loss": 0.5426,
      "step": 2975
    },
    {
      "epoch": 0.4658735128365686,
      "grad_norm": 4.318511962890625,
      "learning_rate": 0.00023963831867057671,
      "loss": 1.0584,
      "step": 2976
    },
    {
      "epoch": 0.46603005635566686,
      "grad_norm": 1.5593619346618652,
      "learning_rate": 0.000239613880742913,
      "loss": 0.5268,
      "step": 2977
    },
    {
      "epoch": 0.4661865998747652,
      "grad_norm": 1.6333059072494507,
      "learning_rate": 0.00023958944281524924,
      "loss": 0.9127,
      "step": 2978
    },
    {
      "epoch": 0.4663431433938635,
      "grad_norm": 1.942104697227478,
      "learning_rate": 0.00023956500488758552,
      "loss": 1.2134,
      "step": 2979
    },
    {
      "epoch": 0.4664996869129618,
      "grad_norm": 3.2156219482421875,
      "learning_rate": 0.00023954056695992177,
      "loss": 1.0458,
      "step": 2980
    },
    {
      "epoch": 0.4666562304320601,
      "grad_norm": 1.9920734167099,
      "learning_rate": 0.00023951612903225802,
      "loss": 0.7571,
      "step": 2981
    },
    {
      "epoch": 0.46681277395115844,
      "grad_norm": 4.246744632720947,
      "learning_rate": 0.0002394916911045943,
      "loss": 0.7142,
      "step": 2982
    },
    {
      "epoch": 0.4669693174702567,
      "grad_norm": 1.2510920763015747,
      "learning_rate": 0.00023946725317693058,
      "loss": 0.8802,
      "step": 2983
    },
    {
      "epoch": 0.46712586098935505,
      "grad_norm": 1.446738839149475,
      "learning_rate": 0.00023944281524926683,
      "loss": 0.9898,
      "step": 2984
    },
    {
      "epoch": 0.4672824045084534,
      "grad_norm": 2.1608774662017822,
      "learning_rate": 0.0002394183773216031,
      "loss": 1.2076,
      "step": 2985
    },
    {
      "epoch": 0.46743894802755165,
      "grad_norm": 2.4812753200531006,
      "learning_rate": 0.0002393939393939394,
      "loss": 0.8113,
      "step": 2986
    },
    {
      "epoch": 0.46759549154665,
      "grad_norm": 2.0435714721679688,
      "learning_rate": 0.0002393695014662756,
      "loss": 1.0752,
      "step": 2987
    },
    {
      "epoch": 0.46775203506574825,
      "grad_norm": 3.043316602706909,
      "learning_rate": 0.0002393450635386119,
      "loss": 1.6354,
      "step": 2988
    },
    {
      "epoch": 0.4679085785848466,
      "grad_norm": 3.5616087913513184,
      "learning_rate": 0.00023932062561094817,
      "loss": 1.0435,
      "step": 2989
    },
    {
      "epoch": 0.4680651221039449,
      "grad_norm": 3.942680835723877,
      "learning_rate": 0.00023929618768328442,
      "loss": 1.6489,
      "step": 2990
    },
    {
      "epoch": 0.4682216656230432,
      "grad_norm": 1.8026114702224731,
      "learning_rate": 0.0002392717497556207,
      "loss": 1.0482,
      "step": 2991
    },
    {
      "epoch": 0.4683782091421415,
      "grad_norm": 3.0630009174346924,
      "learning_rate": 0.00023924731182795698,
      "loss": 1.5673,
      "step": 2992
    },
    {
      "epoch": 0.46853475266123984,
      "grad_norm": 2.128530740737915,
      "learning_rate": 0.00023922287390029323,
      "loss": 1.3448,
      "step": 2993
    },
    {
      "epoch": 0.4686912961803381,
      "grad_norm": 2.782782554626465,
      "learning_rate": 0.0002391984359726295,
      "loss": 1.4261,
      "step": 2994
    },
    {
      "epoch": 0.46884783969943644,
      "grad_norm": 2.1580779552459717,
      "learning_rate": 0.00023917399804496579,
      "loss": 1.6046,
      "step": 2995
    },
    {
      "epoch": 0.46900438321853477,
      "grad_norm": 1.6903536319732666,
      "learning_rate": 0.000239149560117302,
      "loss": 0.9642,
      "step": 2996
    },
    {
      "epoch": 0.46916092673763304,
      "grad_norm": 3.0075113773345947,
      "learning_rate": 0.0002391251221896383,
      "loss": 0.7686,
      "step": 2997
    },
    {
      "epoch": 0.4693174702567314,
      "grad_norm": 1.6930934190750122,
      "learning_rate": 0.00023910068426197457,
      "loss": 1.2032,
      "step": 2998
    },
    {
      "epoch": 0.4694740137758297,
      "grad_norm": 2.7833595275878906,
      "learning_rate": 0.00023907624633431082,
      "loss": 1.0076,
      "step": 2999
    },
    {
      "epoch": 0.469630557294928,
      "grad_norm": 1.994335651397705,
      "learning_rate": 0.0002390518084066471,
      "loss": 0.9728,
      "step": 3000
    },
    {
      "epoch": 0.469630557294928,
      "eval_loss": 0.6461995244026184,
      "eval_runtime": 205.814,
      "eval_samples_per_second": 60.166,
      "eval_steps_per_second": 3.761,
      "eval_wer": 0.4029793654504112,
      "step": 3000
    },
    {
      "epoch": 0.4697871008140263,
      "grad_norm": 0.5218053460121155,
      "learning_rate": 0.00023902737047898337,
      "loss": 0.2968,
      "step": 3001
    },
    {
      "epoch": 0.46994364433312463,
      "grad_norm": 0.8626418113708496,
      "learning_rate": 0.00023900293255131963,
      "loss": 0.363,
      "step": 3002
    },
    {
      "epoch": 0.4701001878522229,
      "grad_norm": 0.6451046466827393,
      "learning_rate": 0.0002389784946236559,
      "loss": 0.3505,
      "step": 3003
    },
    {
      "epoch": 0.47025673137132123,
      "grad_norm": 1.0618706941604614,
      "learning_rate": 0.00023895405669599216,
      "loss": 0.3542,
      "step": 3004
    },
    {
      "epoch": 0.47041327489041956,
      "grad_norm": 0.6961786150932312,
      "learning_rate": 0.0002389296187683284,
      "loss": 0.4272,
      "step": 3005
    },
    {
      "epoch": 0.47056981840951784,
      "grad_norm": 0.6767577528953552,
      "learning_rate": 0.00023890518084066469,
      "loss": 0.3825,
      "step": 3006
    },
    {
      "epoch": 0.47072636192861617,
      "grad_norm": 0.8151198029518127,
      "learning_rate": 0.00023888074291300096,
      "loss": 0.3995,
      "step": 3007
    },
    {
      "epoch": 0.47088290544771444,
      "grad_norm": 1.8509833812713623,
      "learning_rate": 0.00023885630498533722,
      "loss": 0.5605,
      "step": 3008
    },
    {
      "epoch": 0.47103944896681277,
      "grad_norm": 0.6328085660934448,
      "learning_rate": 0.0002388318670576735,
      "loss": 0.4549,
      "step": 3009
    },
    {
      "epoch": 0.4711959924859111,
      "grad_norm": 1.1745184659957886,
      "learning_rate": 0.00023880742913000977,
      "loss": 0.5468,
      "step": 3010
    },
    {
      "epoch": 0.47135253600500937,
      "grad_norm": 1.2673914432525635,
      "learning_rate": 0.000238782991202346,
      "loss": 0.6493,
      "step": 3011
    },
    {
      "epoch": 0.4715090795241077,
      "grad_norm": 0.6951119899749756,
      "learning_rate": 0.00023875855327468227,
      "loss": 0.3245,
      "step": 3012
    },
    {
      "epoch": 0.47166562304320603,
      "grad_norm": 1.6163688898086548,
      "learning_rate": 0.00023873411534701855,
      "loss": 0.4878,
      "step": 3013
    },
    {
      "epoch": 0.4718221665623043,
      "grad_norm": 1.3467936515808105,
      "learning_rate": 0.0002387096774193548,
      "loss": 0.4346,
      "step": 3014
    },
    {
      "epoch": 0.47197871008140263,
      "grad_norm": 0.9435334205627441,
      "learning_rate": 0.00023868523949169108,
      "loss": 0.6272,
      "step": 3015
    },
    {
      "epoch": 0.47213525360050096,
      "grad_norm": 1.2836298942565918,
      "learning_rate": 0.00023866080156402736,
      "loss": 0.6137,
      "step": 3016
    },
    {
      "epoch": 0.47229179711959923,
      "grad_norm": 1.1199934482574463,
      "learning_rate": 0.0002386363636363636,
      "loss": 0.5952,
      "step": 3017
    },
    {
      "epoch": 0.47244834063869756,
      "grad_norm": 1.2082144021987915,
      "learning_rate": 0.0002386119257086999,
      "loss": 0.6759,
      "step": 3018
    },
    {
      "epoch": 0.4726048841577959,
      "grad_norm": 0.8209431171417236,
      "learning_rate": 0.00023858748778103617,
      "loss": 0.4364,
      "step": 3019
    },
    {
      "epoch": 0.47276142767689416,
      "grad_norm": 1.712845802307129,
      "learning_rate": 0.0002385630498533724,
      "loss": 0.7077,
      "step": 3020
    },
    {
      "epoch": 0.4729179711959925,
      "grad_norm": 2.234483242034912,
      "learning_rate": 0.00023853861192570867,
      "loss": 0.7401,
      "step": 3021
    },
    {
      "epoch": 0.4730745147150908,
      "grad_norm": 2.4135541915893555,
      "learning_rate": 0.00023851417399804495,
      "loss": 1.2215,
      "step": 3022
    },
    {
      "epoch": 0.4732310582341891,
      "grad_norm": 1.4404560327529907,
      "learning_rate": 0.0002384897360703812,
      "loss": 0.8349,
      "step": 3023
    },
    {
      "epoch": 0.4733876017532874,
      "grad_norm": 2.8168258666992188,
      "learning_rate": 0.00023846529814271748,
      "loss": 1.2226,
      "step": 3024
    },
    {
      "epoch": 0.4735441452723857,
      "grad_norm": 2.413266181945801,
      "learning_rate": 0.00023844086021505376,
      "loss": 0.7827,
      "step": 3025
    },
    {
      "epoch": 0.473700688791484,
      "grad_norm": 1.8607373237609863,
      "learning_rate": 0.00023841642228739,
      "loss": 0.9485,
      "step": 3026
    },
    {
      "epoch": 0.47385723231058235,
      "grad_norm": 2.2613914012908936,
      "learning_rate": 0.0002383919843597263,
      "loss": 0.6483,
      "step": 3027
    },
    {
      "epoch": 0.47401377582968063,
      "grad_norm": 1.8476412296295166,
      "learning_rate": 0.00023836754643206254,
      "loss": 0.9358,
      "step": 3028
    },
    {
      "epoch": 0.47417031934877896,
      "grad_norm": 5.6817851066589355,
      "learning_rate": 0.0002383431085043988,
      "loss": 1.1602,
      "step": 3029
    },
    {
      "epoch": 0.4743268628678773,
      "grad_norm": 1.4337266683578491,
      "learning_rate": 0.00023831867057673507,
      "loss": 0.5406,
      "step": 3030
    },
    {
      "epoch": 0.47448340638697556,
      "grad_norm": 1.3688822984695435,
      "learning_rate": 0.00023829423264907135,
      "loss": 0.6749,
      "step": 3031
    },
    {
      "epoch": 0.4746399499060739,
      "grad_norm": 1.8527987003326416,
      "learning_rate": 0.0002382697947214076,
      "loss": 0.6256,
      "step": 3032
    },
    {
      "epoch": 0.4747964934251722,
      "grad_norm": 2.1295547485351562,
      "learning_rate": 0.00023824535679374388,
      "loss": 0.8839,
      "step": 3033
    },
    {
      "epoch": 0.4749530369442705,
      "grad_norm": 7.561639308929443,
      "learning_rate": 0.00023822091886608015,
      "loss": 1.5188,
      "step": 3034
    },
    {
      "epoch": 0.4751095804633688,
      "grad_norm": 2.218928575515747,
      "learning_rate": 0.00023819648093841638,
      "loss": 1.0713,
      "step": 3035
    },
    {
      "epoch": 0.47526612398246715,
      "grad_norm": 1.8199081420898438,
      "learning_rate": 0.00023817204301075266,
      "loss": 0.9386,
      "step": 3036
    },
    {
      "epoch": 0.4754226675015654,
      "grad_norm": 2.755200147628784,
      "learning_rate": 0.00023814760508308893,
      "loss": 1.3609,
      "step": 3037
    },
    {
      "epoch": 0.47557921102066375,
      "grad_norm": 2.402803897857666,
      "learning_rate": 0.00023812316715542519,
      "loss": 0.9989,
      "step": 3038
    },
    {
      "epoch": 0.4757357545397621,
      "grad_norm": 2.486337184906006,
      "learning_rate": 0.00023809872922776146,
      "loss": 1.2102,
      "step": 3039
    },
    {
      "epoch": 0.47589229805886035,
      "grad_norm": 2.5728185176849365,
      "learning_rate": 0.00023807429130009774,
      "loss": 1.1949,
      "step": 3040
    },
    {
      "epoch": 0.4760488415779587,
      "grad_norm": 2.1376590728759766,
      "learning_rate": 0.000238049853372434,
      "loss": 1.2828,
      "step": 3041
    },
    {
      "epoch": 0.47620538509705695,
      "grad_norm": 1.944007158279419,
      "learning_rate": 0.00023802541544477027,
      "loss": 2.2113,
      "step": 3042
    },
    {
      "epoch": 0.4763619286161553,
      "grad_norm": 3.550218343734741,
      "learning_rate": 0.00023800097751710655,
      "loss": 0.9221,
      "step": 3043
    },
    {
      "epoch": 0.4765184721352536,
      "grad_norm": 3.1462228298187256,
      "learning_rate": 0.00023797653958944277,
      "loss": 1.6563,
      "step": 3044
    },
    {
      "epoch": 0.4766750156543519,
      "grad_norm": 4.181352615356445,
      "learning_rate": 0.00023795210166177905,
      "loss": 1.7042,
      "step": 3045
    },
    {
      "epoch": 0.4768315591734502,
      "grad_norm": 2.818664312362671,
      "learning_rate": 0.00023792766373411533,
      "loss": 1.0479,
      "step": 3046
    },
    {
      "epoch": 0.47698810269254854,
      "grad_norm": 1.8017176389694214,
      "learning_rate": 0.00023790322580645158,
      "loss": 0.4576,
      "step": 3047
    },
    {
      "epoch": 0.4771446462116468,
      "grad_norm": 3.6153724193573,
      "learning_rate": 0.00023787878787878786,
      "loss": 1.3568,
      "step": 3048
    },
    {
      "epoch": 0.47730118973074515,
      "grad_norm": 2.630364418029785,
      "learning_rate": 0.00023785434995112414,
      "loss": 1.0744,
      "step": 3049
    },
    {
      "epoch": 0.4774577332498435,
      "grad_norm": 2.6814486980438232,
      "learning_rate": 0.0002378299120234604,
      "loss": 1.015,
      "step": 3050
    },
    {
      "epoch": 0.47761427676894175,
      "grad_norm": 0.47881409525871277,
      "learning_rate": 0.00023780547409579664,
      "loss": 0.4242,
      "step": 3051
    },
    {
      "epoch": 0.4777708202880401,
      "grad_norm": 0.47194862365722656,
      "learning_rate": 0.00023778103616813292,
      "loss": 0.3254,
      "step": 3052
    },
    {
      "epoch": 0.4779273638071384,
      "grad_norm": 1.0471508502960205,
      "learning_rate": 0.00023775659824046917,
      "loss": 0.5085,
      "step": 3053
    },
    {
      "epoch": 0.4780839073262367,
      "grad_norm": 0.921794593334198,
      "learning_rate": 0.00023773216031280545,
      "loss": 0.5577,
      "step": 3054
    },
    {
      "epoch": 0.478240450845335,
      "grad_norm": 0.5158277750015259,
      "learning_rate": 0.00023770772238514173,
      "loss": 0.3387,
      "step": 3055
    },
    {
      "epoch": 0.47839699436443334,
      "grad_norm": 0.7019078135490417,
      "learning_rate": 0.00023768328445747798,
      "loss": 0.3237,
      "step": 3056
    },
    {
      "epoch": 0.4785535378835316,
      "grad_norm": 0.46547719836235046,
      "learning_rate": 0.00023765884652981426,
      "loss": 0.2525,
      "step": 3057
    },
    {
      "epoch": 0.47871008140262994,
      "grad_norm": 1.2258836030960083,
      "learning_rate": 0.00023763440860215054,
      "loss": 0.5739,
      "step": 3058
    },
    {
      "epoch": 0.47886662492172827,
      "grad_norm": 0.8825581073760986,
      "learning_rate": 0.00023760997067448676,
      "loss": 0.3721,
      "step": 3059
    },
    {
      "epoch": 0.47902316844082654,
      "grad_norm": 0.7724276185035706,
      "learning_rate": 0.00023758553274682304,
      "loss": 0.3145,
      "step": 3060
    },
    {
      "epoch": 0.47917971195992487,
      "grad_norm": 6.942146301269531,
      "learning_rate": 0.00023756109481915932,
      "loss": 0.716,
      "step": 3061
    },
    {
      "epoch": 0.47933625547902314,
      "grad_norm": 0.8972859978675842,
      "learning_rate": 0.00023753665689149557,
      "loss": 0.4538,
      "step": 3062
    },
    {
      "epoch": 0.47949279899812147,
      "grad_norm": 1.213957667350769,
      "learning_rate": 0.00023751221896383185,
      "loss": 0.6098,
      "step": 3063
    },
    {
      "epoch": 0.4796493425172198,
      "grad_norm": 1.26592218875885,
      "learning_rate": 0.00023748778103616813,
      "loss": 0.5618,
      "step": 3064
    },
    {
      "epoch": 0.4798058860363181,
      "grad_norm": 1.0946046113967896,
      "learning_rate": 0.00023746334310850438,
      "loss": 0.5241,
      "step": 3065
    },
    {
      "epoch": 0.4799624295554164,
      "grad_norm": 1.3569597005844116,
      "learning_rate": 0.00023743890518084065,
      "loss": 0.5417,
      "step": 3066
    },
    {
      "epoch": 0.48011897307451473,
      "grad_norm": 1.2722363471984863,
      "learning_rate": 0.00023741446725317693,
      "loss": 0.7591,
      "step": 3067
    },
    {
      "epoch": 0.480275516593613,
      "grad_norm": 2.4486443996429443,
      "learning_rate": 0.00023739002932551316,
      "loss": 0.7028,
      "step": 3068
    },
    {
      "epoch": 0.48043206011271133,
      "grad_norm": 1.7454280853271484,
      "learning_rate": 0.00023736559139784944,
      "loss": 0.7086,
      "step": 3069
    },
    {
      "epoch": 0.48058860363180966,
      "grad_norm": 1.4875236749649048,
      "learning_rate": 0.00023734115347018571,
      "loss": 0.7306,
      "step": 3070
    },
    {
      "epoch": 0.48074514715090794,
      "grad_norm": 1.795547366142273,
      "learning_rate": 0.00023731671554252197,
      "loss": 0.6128,
      "step": 3071
    },
    {
      "epoch": 0.48090169067000627,
      "grad_norm": 1.589440107345581,
      "learning_rate": 0.00023729227761485824,
      "loss": 0.5872,
      "step": 3072
    },
    {
      "epoch": 0.4810582341891046,
      "grad_norm": 2.098665714263916,
      "learning_rate": 0.00023726783968719452,
      "loss": 0.7743,
      "step": 3073
    },
    {
      "epoch": 0.48121477770820287,
      "grad_norm": 1.5738600492477417,
      "learning_rate": 0.00023724340175953077,
      "loss": 0.8935,
      "step": 3074
    },
    {
      "epoch": 0.4813713212273012,
      "grad_norm": 1.6582045555114746,
      "learning_rate": 0.00023721896383186702,
      "loss": 0.9405,
      "step": 3075
    },
    {
      "epoch": 0.4815278647463995,
      "grad_norm": 1.581145167350769,
      "learning_rate": 0.0002371945259042033,
      "loss": 0.7447,
      "step": 3076
    },
    {
      "epoch": 0.4816844082654978,
      "grad_norm": 1.4934779405593872,
      "learning_rate": 0.00023717008797653955,
      "loss": 0.5348,
      "step": 3077
    },
    {
      "epoch": 0.4818409517845961,
      "grad_norm": 1.6471208333969116,
      "learning_rate": 0.00023714565004887583,
      "loss": 0.9434,
      "step": 3078
    },
    {
      "epoch": 0.4819974953036944,
      "grad_norm": 2.7180702686309814,
      "learning_rate": 0.0002371212121212121,
      "loss": 0.8578,
      "step": 3079
    },
    {
      "epoch": 0.48215403882279273,
      "grad_norm": 1.2886486053466797,
      "learning_rate": 0.00023709677419354836,
      "loss": 0.7375,
      "step": 3080
    },
    {
      "epoch": 0.48231058234189106,
      "grad_norm": 2.2205405235290527,
      "learning_rate": 0.00023707233626588464,
      "loss": 1.0982,
      "step": 3081
    },
    {
      "epoch": 0.48246712586098933,
      "grad_norm": 3.079237461090088,
      "learning_rate": 0.00023704789833822092,
      "loss": 0.9257,
      "step": 3082
    },
    {
      "epoch": 0.48262366938008766,
      "grad_norm": 2.3623974323272705,
      "learning_rate": 0.00023702346041055714,
      "loss": 1.1405,
      "step": 3083
    },
    {
      "epoch": 0.482780212899186,
      "grad_norm": 1.2514792680740356,
      "learning_rate": 0.00023699902248289342,
      "loss": 0.6099,
      "step": 3084
    },
    {
      "epoch": 0.48293675641828426,
      "grad_norm": 1.4864928722381592,
      "learning_rate": 0.0002369745845552297,
      "loss": 0.82,
      "step": 3085
    },
    {
      "epoch": 0.4830932999373826,
      "grad_norm": 1.9229499101638794,
      "learning_rate": 0.00023695014662756595,
      "loss": 1.0889,
      "step": 3086
    },
    {
      "epoch": 0.4832498434564809,
      "grad_norm": 2.564469575881958,
      "learning_rate": 0.00023692570869990223,
      "loss": 0.9904,
      "step": 3087
    },
    {
      "epoch": 0.4834063869755792,
      "grad_norm": 3.783491611480713,
      "learning_rate": 0.0002369012707722385,
      "loss": 1.594,
      "step": 3088
    },
    {
      "epoch": 0.4835629304946775,
      "grad_norm": 2.4869840145111084,
      "learning_rate": 0.00023687683284457476,
      "loss": 0.8637,
      "step": 3089
    },
    {
      "epoch": 0.48371947401377585,
      "grad_norm": 2.5603604316711426,
      "learning_rate": 0.00023685239491691104,
      "loss": 1.1697,
      "step": 3090
    },
    {
      "epoch": 0.4838760175328741,
      "grad_norm": 3.291196823120117,
      "learning_rate": 0.00023682795698924732,
      "loss": 2.2811,
      "step": 3091
    },
    {
      "epoch": 0.48403256105197245,
      "grad_norm": 1.8274985551834106,
      "learning_rate": 0.00023680351906158354,
      "loss": 1.0696,
      "step": 3092
    },
    {
      "epoch": 0.4841891045710708,
      "grad_norm": 1.265817403793335,
      "learning_rate": 0.00023677908113391982,
      "loss": 0.9953,
      "step": 3093
    },
    {
      "epoch": 0.48434564809016906,
      "grad_norm": NaN,
      "learning_rate": 0.00023677908113391982,
      "loss": 0.0,
      "step": 3094
    },
    {
      "epoch": 0.4845021916092674,
      "grad_norm": 3.3963782787323,
      "learning_rate": 0.0002367546432062561,
      "loss": 1.2792,
      "step": 3095
    },
    {
      "epoch": 0.48465873512836566,
      "grad_norm": 1.3875449895858765,
      "learning_rate": 0.00023673020527859235,
      "loss": 0.7411,
      "step": 3096
    },
    {
      "epoch": 0.484815278647464,
      "grad_norm": 3.187683582305908,
      "learning_rate": 0.00023670576735092863,
      "loss": 0.9312,
      "step": 3097
    },
    {
      "epoch": 0.4849718221665623,
      "grad_norm": 3.2056405544281006,
      "learning_rate": 0.0002366813294232649,
      "loss": 1.432,
      "step": 3098
    },
    {
      "epoch": 0.4851283656856606,
      "grad_norm": 4.896152019500732,
      "learning_rate": 0.00023665689149560116,
      "loss": 0.9695,
      "step": 3099
    },
    {
      "epoch": 0.4852849092047589,
      "grad_norm": 2.358058452606201,
      "learning_rate": 0.0002366324535679374,
      "loss": 1.3928,
      "step": 3100
    },
    {
      "epoch": 0.48544145272385725,
      "grad_norm": 0.6357040405273438,
      "learning_rate": 0.00023660801564027369,
      "loss": 0.3465,
      "step": 3101
    },
    {
      "epoch": 0.4855979962429555,
      "grad_norm": 0.48957034945487976,
      "learning_rate": 0.00023658357771260994,
      "loss": 0.4029,
      "step": 3102
    },
    {
      "epoch": 0.48575453976205385,
      "grad_norm": 0.7128604054450989,
      "learning_rate": 0.00023655913978494621,
      "loss": 0.3123,
      "step": 3103
    },
    {
      "epoch": 0.4859110832811522,
      "grad_norm": 0.5227282643318176,
      "learning_rate": 0.0002365347018572825,
      "loss": 0.3385,
      "step": 3104
    },
    {
      "epoch": 0.48606762680025045,
      "grad_norm": 0.5651166439056396,
      "learning_rate": 0.00023651026392961874,
      "loss": 0.3359,
      "step": 3105
    },
    {
      "epoch": 0.4862241703193488,
      "grad_norm": 0.8126301169395447,
      "learning_rate": 0.00023648582600195502,
      "loss": 0.4309,
      "step": 3106
    },
    {
      "epoch": 0.4863807138384471,
      "grad_norm": 1.4748018980026245,
      "learning_rate": 0.0002364613880742913,
      "loss": 0.4009,
      "step": 3107
    },
    {
      "epoch": 0.4865372573575454,
      "grad_norm": 1.2960294485092163,
      "learning_rate": 0.00023643695014662753,
      "loss": 0.4647,
      "step": 3108
    },
    {
      "epoch": 0.4866938008766437,
      "grad_norm": 1.0774697065353394,
      "learning_rate": 0.0002364125122189638,
      "loss": 0.4079,
      "step": 3109
    },
    {
      "epoch": 0.48685034439574204,
      "grad_norm": 1.326656460762024,
      "learning_rate": 0.00023638807429130008,
      "loss": 0.3332,
      "step": 3110
    },
    {
      "epoch": 0.4870068879148403,
      "grad_norm": 0.907184362411499,
      "learning_rate": 0.00023636363636363633,
      "loss": 0.4756,
      "step": 3111
    },
    {
      "epoch": 0.48716343143393864,
      "grad_norm": 0.7261903882026672,
      "learning_rate": 0.0002363391984359726,
      "loss": 0.3881,
      "step": 3112
    },
    {
      "epoch": 0.48731997495303697,
      "grad_norm": 1.3236558437347412,
      "learning_rate": 0.0002363147605083089,
      "loss": 0.4808,
      "step": 3113
    },
    {
      "epoch": 0.48747651847213525,
      "grad_norm": 1.082720160484314,
      "learning_rate": 0.00023629032258064514,
      "loss": 0.4554,
      "step": 3114
    },
    {
      "epoch": 0.4876330619912336,
      "grad_norm": 1.2888387441635132,
      "learning_rate": 0.00023626588465298142,
      "loss": 0.7315,
      "step": 3115
    },
    {
      "epoch": 0.48778960551033185,
      "grad_norm": 1.2054266929626465,
      "learning_rate": 0.0002362414467253177,
      "loss": 0.8247,
      "step": 3116
    },
    {
      "epoch": 0.4879461490294302,
      "grad_norm": 0.8966947197914124,
      "learning_rate": 0.00023621700879765392,
      "loss": 0.4858,
      "step": 3117
    },
    {
      "epoch": 0.4881026925485285,
      "grad_norm": 2.668046712875366,
      "learning_rate": 0.0002361925708699902,
      "loss": 0.7636,
      "step": 3118
    },
    {
      "epoch": 0.4882592360676268,
      "grad_norm": 2.526010513305664,
      "learning_rate": 0.00023616813294232648,
      "loss": 0.7442,
      "step": 3119
    },
    {
      "epoch": 0.4884157795867251,
      "grad_norm": 1.6158323287963867,
      "learning_rate": 0.00023614369501466273,
      "loss": 0.6907,
      "step": 3120
    },
    {
      "epoch": 0.48857232310582344,
      "grad_norm": 1.133370041847229,
      "learning_rate": 0.000236119257086999,
      "loss": 0.773,
      "step": 3121
    },
    {
      "epoch": 0.4887288666249217,
      "grad_norm": 1.4854271411895752,
      "learning_rate": 0.0002360948191593353,
      "loss": 0.9368,
      "step": 3122
    },
    {
      "epoch": 0.48888541014402004,
      "grad_norm": 2.469548463821411,
      "learning_rate": 0.0002360703812316715,
      "loss": 0.9644,
      "step": 3123
    },
    {
      "epoch": 0.48904195366311837,
      "grad_norm": 1.3509788513183594,
      "learning_rate": 0.0002360459433040078,
      "loss": 0.8894,
      "step": 3124
    },
    {
      "epoch": 0.48919849718221664,
      "grad_norm": 1.1007205247879028,
      "learning_rate": 0.00023602150537634407,
      "loss": 0.6317,
      "step": 3125
    },
    {
      "epoch": 0.48935504070131497,
      "grad_norm": 1.8395272493362427,
      "learning_rate": 0.00023599706744868032,
      "loss": 0.8419,
      "step": 3126
    },
    {
      "epoch": 0.4895115842204133,
      "grad_norm": 2.4748945236206055,
      "learning_rate": 0.0002359726295210166,
      "loss": 1.069,
      "step": 3127
    },
    {
      "epoch": 0.48966812773951157,
      "grad_norm": 2.535205602645874,
      "learning_rate": 0.00023594819159335288,
      "loss": 0.9937,
      "step": 3128
    },
    {
      "epoch": 0.4898246712586099,
      "grad_norm": 1.4972389936447144,
      "learning_rate": 0.00023592375366568913,
      "loss": 0.6853,
      "step": 3129
    },
    {
      "epoch": 0.48998121477770823,
      "grad_norm": 1.4133657217025757,
      "learning_rate": 0.0002358993157380254,
      "loss": 0.6998,
      "step": 3130
    },
    {
      "epoch": 0.4901377582968065,
      "grad_norm": 2.480767250061035,
      "learning_rate": 0.00023587487781036168,
      "loss": 1.6381,
      "step": 3131
    },
    {
      "epoch": 0.49029430181590483,
      "grad_norm": 2.512920618057251,
      "learning_rate": 0.0002358504398826979,
      "loss": 0.7757,
      "step": 3132
    },
    {
      "epoch": 0.4904508453350031,
      "grad_norm": 2.0442888736724854,
      "learning_rate": 0.00023582600195503419,
      "loss": 1.1028,
      "step": 3133
    },
    {
      "epoch": 0.49060738885410143,
      "grad_norm": 2.166085720062256,
      "learning_rate": 0.00023580156402737046,
      "loss": 1.0315,
      "step": 3134
    },
    {
      "epoch": 0.49076393237319976,
      "grad_norm": 2.705939292907715,
      "learning_rate": 0.00023577712609970672,
      "loss": 0.9559,
      "step": 3135
    },
    {
      "epoch": 0.49092047589229804,
      "grad_norm": 1.7874341011047363,
      "learning_rate": 0.000235752688172043,
      "loss": 0.719,
      "step": 3136
    },
    {
      "epoch": 0.49107701941139636,
      "grad_norm": 2.1287143230438232,
      "learning_rate": 0.00023572825024437927,
      "loss": 0.8894,
      "step": 3137
    },
    {
      "epoch": 0.4912335629304947,
      "grad_norm": 2.1680641174316406,
      "learning_rate": 0.00023570381231671552,
      "loss": 1.2279,
      "step": 3138
    },
    {
      "epoch": 0.49139010644959297,
      "grad_norm": 3.1743721961975098,
      "learning_rate": 0.0002356793743890518,
      "loss": 1.1262,
      "step": 3139
    },
    {
      "epoch": 0.4915466499686913,
      "grad_norm": 2.6501054763793945,
      "learning_rate": 0.00023565493646138805,
      "loss": 1.4381,
      "step": 3140
    },
    {
      "epoch": 0.4917031934877896,
      "grad_norm": 2.002523422241211,
      "learning_rate": 0.0002356304985337243,
      "loss": 0.9156,
      "step": 3141
    },
    {
      "epoch": 0.4918597370068879,
      "grad_norm": 3.3505101203918457,
      "learning_rate": 0.00023560606060606058,
      "loss": 1.7951,
      "step": 3142
    },
    {
      "epoch": 0.4920162805259862,
      "grad_norm": 2.2706339359283447,
      "learning_rate": 0.00023558162267839686,
      "loss": 0.957,
      "step": 3143
    },
    {
      "epoch": 0.49217282404508456,
      "grad_norm": 1.9583147764205933,
      "learning_rate": 0.0002355571847507331,
      "loss": 1.1053,
      "step": 3144
    },
    {
      "epoch": 0.49232936756418283,
      "grad_norm": 1.6880264282226562,
      "learning_rate": 0.0002355327468230694,
      "loss": 1.3319,
      "step": 3145
    },
    {
      "epoch": 0.49248591108328116,
      "grad_norm": 2.1848926544189453,
      "learning_rate": 0.00023550830889540567,
      "loss": 1.1314,
      "step": 3146
    },
    {
      "epoch": 0.4926424546023795,
      "grad_norm": 2.028449773788452,
      "learning_rate": 0.0002354838709677419,
      "loss": 1.2468,
      "step": 3147
    },
    {
      "epoch": 0.49279899812147776,
      "grad_norm": 1.6452484130859375,
      "learning_rate": 0.00023545943304007817,
      "loss": 0.7077,
      "step": 3148
    },
    {
      "epoch": 0.4929555416405761,
      "grad_norm": 1.4108058214187622,
      "learning_rate": 0.00023543499511241445,
      "loss": 0.7043,
      "step": 3149
    },
    {
      "epoch": 0.49311208515967436,
      "grad_norm": 3.064290761947632,
      "learning_rate": 0.0002354105571847507,
      "loss": 0.8504,
      "step": 3150
    },
    {
      "epoch": 0.4932686286787727,
      "grad_norm": 0.6606544256210327,
      "learning_rate": 0.00023538611925708698,
      "loss": 0.3642,
      "step": 3151
    },
    {
      "epoch": 0.493425172197871,
      "grad_norm": 0.7985259890556335,
      "learning_rate": 0.00023536168132942326,
      "loss": 0.3771,
      "step": 3152
    },
    {
      "epoch": 0.4935817157169693,
      "grad_norm": 0.8636269569396973,
      "learning_rate": 0.0002353372434017595,
      "loss": 0.2973,
      "step": 3153
    },
    {
      "epoch": 0.4937382592360676,
      "grad_norm": 0.8866095542907715,
      "learning_rate": 0.0002353128054740958,
      "loss": 0.3708,
      "step": 3154
    },
    {
      "epoch": 0.49389480275516595,
      "grad_norm": 0.6658779978752136,
      "learning_rate": 0.00023528836754643207,
      "loss": 0.3621,
      "step": 3155
    },
    {
      "epoch": 0.4940513462742642,
      "grad_norm": 0.5546128749847412,
      "learning_rate": 0.0002352639296187683,
      "loss": 0.2645,
      "step": 3156
    },
    {
      "epoch": 0.49420788979336255,
      "grad_norm": 0.9286956191062927,
      "learning_rate": 0.00023523949169110457,
      "loss": 0.4566,
      "step": 3157
    },
    {
      "epoch": 0.4943644333124609,
      "grad_norm": 1.118327021598816,
      "learning_rate": 0.00023521505376344085,
      "loss": 0.3843,
      "step": 3158
    },
    {
      "epoch": 0.49452097683155916,
      "grad_norm": 1.691359281539917,
      "learning_rate": 0.0002351906158357771,
      "loss": 0.306,
      "step": 3159
    },
    {
      "epoch": 0.4946775203506575,
      "grad_norm": 1.1028999090194702,
      "learning_rate": 0.00023516617790811338,
      "loss": 0.5754,
      "step": 3160
    },
    {
      "epoch": 0.4948340638697558,
      "grad_norm": 0.7714839577674866,
      "learning_rate": 0.00023514173998044965,
      "loss": 0.4509,
      "step": 3161
    },
    {
      "epoch": 0.4949906073888541,
      "grad_norm": 0.7560392022132874,
      "learning_rate": 0.0002351173020527859,
      "loss": 0.3924,
      "step": 3162
    },
    {
      "epoch": 0.4951471509079524,
      "grad_norm": 0.8828386664390564,
      "learning_rate": 0.00023509286412512218,
      "loss": 0.6857,
      "step": 3163
    },
    {
      "epoch": 0.49530369442705074,
      "grad_norm": 2.5070831775665283,
      "learning_rate": 0.00023506842619745844,
      "loss": 0.671,
      "step": 3164
    },
    {
      "epoch": 0.495460237946149,
      "grad_norm": 1.2520339488983154,
      "learning_rate": 0.0002350439882697947,
      "loss": 0.5964,
      "step": 3165
    },
    {
      "epoch": 0.49561678146524735,
      "grad_norm": 0.9470450282096863,
      "learning_rate": 0.00023501955034213096,
      "loss": 0.5922,
      "step": 3166
    },
    {
      "epoch": 0.4957733249843457,
      "grad_norm": 0.9970950484275818,
      "learning_rate": 0.00023499511241446724,
      "loss": 0.6786,
      "step": 3167
    },
    {
      "epoch": 0.49592986850344395,
      "grad_norm": 1.0567865371704102,
      "learning_rate": 0.0002349706744868035,
      "loss": 0.6266,
      "step": 3168
    },
    {
      "epoch": 0.4960864120225423,
      "grad_norm": 0.6511049866676331,
      "learning_rate": 0.00023494623655913977,
      "loss": 0.3637,
      "step": 3169
    },
    {
      "epoch": 0.49624295554164055,
      "grad_norm": 1.0755113363265991,
      "learning_rate": 0.00023492179863147605,
      "loss": 0.538,
      "step": 3170
    },
    {
      "epoch": 0.4963994990607389,
      "grad_norm": 1.8359781503677368,
      "learning_rate": 0.00023489736070381228,
      "loss": 0.8472,
      "step": 3171
    },
    {
      "epoch": 0.4965560425798372,
      "grad_norm": 0.943196713924408,
      "learning_rate": 0.00023487292277614855,
      "loss": 0.5028,
      "step": 3172
    },
    {
      "epoch": 0.4967125860989355,
      "grad_norm": 2.1813058853149414,
      "learning_rate": 0.00023484848484848483,
      "loss": 0.6239,
      "step": 3173
    },
    {
      "epoch": 0.4968691296180338,
      "grad_norm": 1.8853085041046143,
      "learning_rate": 0.00023482404692082108,
      "loss": 0.544,
      "step": 3174
    },
    {
      "epoch": 0.49702567313713214,
      "grad_norm": 1.5820690393447876,
      "learning_rate": 0.00023479960899315736,
      "loss": 0.7816,
      "step": 3175
    },
    {
      "epoch": 0.4971822166562304,
      "grad_norm": 1.5415321588516235,
      "learning_rate": 0.00023477517106549364,
      "loss": 0.5894,
      "step": 3176
    },
    {
      "epoch": 0.49733876017532874,
      "grad_norm": 1.8562960624694824,
      "learning_rate": 0.0002347507331378299,
      "loss": 0.8592,
      "step": 3177
    },
    {
      "epoch": 0.49749530369442707,
      "grad_norm": 2.251617908477783,
      "learning_rate": 0.00023472629521016617,
      "loss": 1.0993,
      "step": 3178
    },
    {
      "epoch": 0.49765184721352534,
      "grad_norm": 3.1134870052337646,
      "learning_rate": 0.00023470185728250245,
      "loss": 0.9617,
      "step": 3179
    },
    {
      "epoch": 0.4978083907326237,
      "grad_norm": 2.8295364379882812,
      "learning_rate": 0.00023467741935483867,
      "loss": 0.7751,
      "step": 3180
    },
    {
      "epoch": 0.497964934251722,
      "grad_norm": 2.4342901706695557,
      "learning_rate": 0.00023465298142717495,
      "loss": 0.638,
      "step": 3181
    },
    {
      "epoch": 0.4981214777708203,
      "grad_norm": 2.19533371925354,
      "learning_rate": 0.00023462854349951123,
      "loss": 1.0993,
      "step": 3182
    },
    {
      "epoch": 0.4982780212899186,
      "grad_norm": 2.2838051319122314,
      "learning_rate": 0.00023460410557184748,
      "loss": 1.1031,
      "step": 3183
    },
    {
      "epoch": 0.49843456480901693,
      "grad_norm": 2.4727444648742676,
      "learning_rate": 0.00023457966764418376,
      "loss": 1.0324,
      "step": 3184
    },
    {
      "epoch": 0.4985911083281152,
      "grad_norm": 2.0869252681732178,
      "learning_rate": 0.00023455522971652004,
      "loss": 1.0483,
      "step": 3185
    },
    {
      "epoch": 0.49874765184721354,
      "grad_norm": 2.266732931137085,
      "learning_rate": 0.0002345307917888563,
      "loss": 0.8777,
      "step": 3186
    },
    {
      "epoch": 0.4989041953663118,
      "grad_norm": 2.169137954711914,
      "learning_rate": 0.00023450635386119257,
      "loss": 1.1718,
      "step": 3187
    },
    {
      "epoch": 0.49906073888541014,
      "grad_norm": 2.2153878211975098,
      "learning_rate": 0.00023448191593352882,
      "loss": 1.1595,
      "step": 3188
    },
    {
      "epoch": 0.49921728240450847,
      "grad_norm": 2.4565927982330322,
      "learning_rate": 0.00023445747800586507,
      "loss": 1.2587,
      "step": 3189
    },
    {
      "epoch": 0.49937382592360674,
      "grad_norm": 1.9531141519546509,
      "learning_rate": 0.00023443304007820135,
      "loss": 0.9619,
      "step": 3190
    },
    {
      "epoch": 0.49953036944270507,
      "grad_norm": 3.107656478881836,
      "learning_rate": 0.00023440860215053763,
      "loss": 1.2016,
      "step": 3191
    },
    {
      "epoch": 0.4996869129618034,
      "grad_norm": 2.456040143966675,
      "learning_rate": 0.00023438416422287388,
      "loss": 1.7411,
      "step": 3192
    },
    {
      "epoch": 0.49984345648090167,
      "grad_norm": 2.5410706996917725,
      "learning_rate": 0.00023435972629521016,
      "loss": 1.5113,
      "step": 3193
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8548656702041626,
      "learning_rate": 0.00023433528836754643,
      "loss": 1.2314,
      "step": 3194
    },
    {
      "epoch": 0.5001565435190983,
      "grad_norm": 2.335625648498535,
      "learning_rate": 0.00023431085043988266,
      "loss": 0.9746,
      "step": 3195
    },
    {
      "epoch": 0.5003130870381967,
      "grad_norm": 1.950925350189209,
      "learning_rate": 0.00023428641251221894,
      "loss": 0.6818,
      "step": 3196
    },
    {
      "epoch": 0.5004696305572949,
      "grad_norm": 2.8675320148468018,
      "learning_rate": 0.00023426197458455521,
      "loss": 1.3223,
      "step": 3197
    },
    {
      "epoch": 0.5006261740763932,
      "grad_norm": 7.516252040863037,
      "learning_rate": 0.00023423753665689147,
      "loss": 0.8275,
      "step": 3198
    },
    {
      "epoch": 0.5007827175954915,
      "grad_norm": 3.5480682849884033,
      "learning_rate": 0.00023421309872922774,
      "loss": 0.8734,
      "step": 3199
    },
    {
      "epoch": 0.5009392611145899,
      "grad_norm": 1.8720403909683228,
      "learning_rate": 0.00023418866080156402,
      "loss": 0.9164,
      "step": 3200
    },
    {
      "epoch": 0.5010958046336882,
      "grad_norm": 0.5714183449745178,
      "learning_rate": 0.00023416422287390027,
      "loss": 0.3279,
      "step": 3201
    },
    {
      "epoch": 0.5012523481527865,
      "grad_norm": 0.5906309485435486,
      "learning_rate": 0.00023413978494623655,
      "loss": 0.4663,
      "step": 3202
    },
    {
      "epoch": 0.5014088916718847,
      "grad_norm": 0.49747294187545776,
      "learning_rate": 0.00023411534701857283,
      "loss": 0.3761,
      "step": 3203
    },
    {
      "epoch": 0.5015654351909831,
      "grad_norm": 0.6094551086425781,
      "learning_rate": 0.00023409090909090905,
      "loss": 0.3538,
      "step": 3204
    },
    {
      "epoch": 0.5017219787100814,
      "grad_norm": 0.549359142780304,
      "learning_rate": 0.00023406647116324533,
      "loss": 0.259,
      "step": 3205
    },
    {
      "epoch": 0.5018785222291797,
      "grad_norm": 0.6701558232307434,
      "learning_rate": 0.0002340420332355816,
      "loss": 0.3294,
      "step": 3206
    },
    {
      "epoch": 0.502035065748278,
      "grad_norm": 0.6899073719978333,
      "learning_rate": 0.00023401759530791786,
      "loss": 0.3143,
      "step": 3207
    },
    {
      "epoch": 0.5021916092673764,
      "grad_norm": 0.5624547004699707,
      "learning_rate": 0.00023399315738025414,
      "loss": 0.4073,
      "step": 3208
    },
    {
      "epoch": 0.5023481527864746,
      "grad_norm": 1.0639121532440186,
      "learning_rate": 0.00023396871945259042,
      "loss": 0.5554,
      "step": 3209
    },
    {
      "epoch": 0.5025046963055729,
      "grad_norm": 1.0245072841644287,
      "learning_rate": 0.00023394428152492667,
      "loss": 1.0342,
      "step": 3210
    },
    {
      "epoch": 0.5026612398246713,
      "grad_norm": 0.8923421502113342,
      "learning_rate": 0.00023391984359726292,
      "loss": 0.3688,
      "step": 3211
    },
    {
      "epoch": 0.5028177833437696,
      "grad_norm": 1.4009363651275635,
      "learning_rate": 0.0002338954056695992,
      "loss": 0.6074,
      "step": 3212
    },
    {
      "epoch": 0.5029743268628679,
      "grad_norm": 1.919011116027832,
      "learning_rate": 0.00023387096774193545,
      "loss": 0.5968,
      "step": 3213
    },
    {
      "epoch": 0.5031308703819661,
      "grad_norm": 1.0158270597457886,
      "learning_rate": 0.00023384652981427173,
      "loss": 0.5708,
      "step": 3214
    },
    {
      "epoch": 0.5032874139010645,
      "grad_norm": 1.5985990762710571,
      "learning_rate": 0.000233822091886608,
      "loss": 0.7346,
      "step": 3215
    },
    {
      "epoch": 0.5034439574201628,
      "grad_norm": 1.1101726293563843,
      "learning_rate": 0.00023379765395894426,
      "loss": 0.5624,
      "step": 3216
    },
    {
      "epoch": 0.5036005009392611,
      "grad_norm": 1.8938688039779663,
      "learning_rate": 0.00023377321603128054,
      "loss": 0.4229,
      "step": 3217
    },
    {
      "epoch": 0.5037570444583594,
      "grad_norm": 0.9973450303077698,
      "learning_rate": 0.00023374877810361682,
      "loss": 0.6207,
      "step": 3218
    },
    {
      "epoch": 0.5039135879774578,
      "grad_norm": 1.006881833076477,
      "learning_rate": 0.00023372434017595304,
      "loss": 0.4491,
      "step": 3219
    },
    {
      "epoch": 0.504070131496556,
      "grad_norm": 1.4391586780548096,
      "learning_rate": 0.00023369990224828932,
      "loss": 0.6227,
      "step": 3220
    },
    {
      "epoch": 0.5042266750156543,
      "grad_norm": 1.5591343641281128,
      "learning_rate": 0.0002336754643206256,
      "loss": 0.5966,
      "step": 3221
    },
    {
      "epoch": 0.5043832185347527,
      "grad_norm": 1.0018322467803955,
      "learning_rate": 0.00023365102639296185,
      "loss": 0.5168,
      "step": 3222
    },
    {
      "epoch": 0.504539762053851,
      "grad_norm": 1.1433907747268677,
      "learning_rate": 0.00023362658846529813,
      "loss": 0.6349,
      "step": 3223
    },
    {
      "epoch": 0.5046963055729493,
      "grad_norm": 1.683487057685852,
      "learning_rate": 0.0002336021505376344,
      "loss": 0.9885,
      "step": 3224
    },
    {
      "epoch": 0.5048528490920476,
      "grad_norm": 1.9962869882583618,
      "learning_rate": 0.00023357771260997066,
      "loss": 0.9242,
      "step": 3225
    },
    {
      "epoch": 0.5050093926111459,
      "grad_norm": 2.4794111251831055,
      "learning_rate": 0.00023355327468230693,
      "loss": 1.2199,
      "step": 3226
    },
    {
      "epoch": 0.5051659361302442,
      "grad_norm": 1.2889044284820557,
      "learning_rate": 0.0002335288367546432,
      "loss": 0.4802,
      "step": 3227
    },
    {
      "epoch": 0.5053224796493425,
      "grad_norm": 1.9017808437347412,
      "learning_rate": 0.00023350439882697944,
      "loss": 0.9971,
      "step": 3228
    },
    {
      "epoch": 0.5054790231684408,
      "grad_norm": 2.0930917263031006,
      "learning_rate": 0.00023347996089931572,
      "loss": 0.6098,
      "step": 3229
    },
    {
      "epoch": 0.5056355666875392,
      "grad_norm": 2.39898943901062,
      "learning_rate": 0.000233455522971652,
      "loss": 1.0619,
      "step": 3230
    },
    {
      "epoch": 0.5057921102066374,
      "grad_norm": 2.192613124847412,
      "learning_rate": 0.00023343108504398824,
      "loss": 0.9109,
      "step": 3231
    },
    {
      "epoch": 0.5059486537257357,
      "grad_norm": 2.8761684894561768,
      "learning_rate": 0.00023340664711632452,
      "loss": 0.8734,
      "step": 3232
    },
    {
      "epoch": 0.506105197244834,
      "grad_norm": 3.7933554649353027,
      "learning_rate": 0.0002333822091886608,
      "loss": 1.1158,
      "step": 3233
    },
    {
      "epoch": 0.5062617407639324,
      "grad_norm": 3.8589084148406982,
      "learning_rate": 0.00023335777126099705,
      "loss": 0.9443,
      "step": 3234
    },
    {
      "epoch": 0.5064182842830307,
      "grad_norm": 2.620635986328125,
      "learning_rate": 0.0002333333333333333,
      "loss": 1.1503,
      "step": 3235
    },
    {
      "epoch": 0.506574827802129,
      "grad_norm": 2.9023118019104004,
      "learning_rate": 0.00023330889540566958,
      "loss": 1.1161,
      "step": 3236
    },
    {
      "epoch": 0.5067313713212273,
      "grad_norm": 2.676131248474121,
      "learning_rate": 0.00023328445747800583,
      "loss": 1.3424,
      "step": 3237
    },
    {
      "epoch": 0.5068879148403256,
      "grad_norm": 5.856595516204834,
      "learning_rate": 0.0002332600195503421,
      "loss": 1.2241,
      "step": 3238
    },
    {
      "epoch": 0.5070444583594239,
      "grad_norm": 2.3947947025299072,
      "learning_rate": 0.0002332355816226784,
      "loss": 1.5444,
      "step": 3239
    },
    {
      "epoch": 0.5072010018785222,
      "grad_norm": 2.4484972953796387,
      "learning_rate": 0.00023321114369501464,
      "loss": 0.983,
      "step": 3240
    },
    {
      "epoch": 0.5073575453976206,
      "grad_norm": 1.774258017539978,
      "learning_rate": 0.00023318670576735092,
      "loss": 1.0836,
      "step": 3241
    },
    {
      "epoch": 0.5075140889167189,
      "grad_norm": 3.7993736267089844,
      "learning_rate": 0.0002331622678396872,
      "loss": 1.9356,
      "step": 3242
    },
    {
      "epoch": 0.5076706324358171,
      "grad_norm": 2.240050792694092,
      "learning_rate": 0.00023313782991202342,
      "loss": 1.4901,
      "step": 3243
    },
    {
      "epoch": 0.5078271759549154,
      "grad_norm": 2.437676429748535,
      "learning_rate": 0.0002331133919843597,
      "loss": 1.0782,
      "step": 3244
    },
    {
      "epoch": 0.5079837194740138,
      "grad_norm": 2.2974166870117188,
      "learning_rate": 0.00023308895405669598,
      "loss": 1.5938,
      "step": 3245
    },
    {
      "epoch": 0.5081402629931121,
      "grad_norm": 1.5529897212982178,
      "learning_rate": 0.00023306451612903223,
      "loss": 0.4601,
      "step": 3246
    },
    {
      "epoch": 0.5082968065122104,
      "grad_norm": 1.3996665477752686,
      "learning_rate": 0.0002330400782013685,
      "loss": 0.6735,
      "step": 3247
    },
    {
      "epoch": 0.5084533500313086,
      "grad_norm": 1.607757329940796,
      "learning_rate": 0.0002330156402737048,
      "loss": 0.9503,
      "step": 3248
    },
    {
      "epoch": 0.508609893550407,
      "grad_norm": 1.984406590461731,
      "learning_rate": 0.00023299120234604104,
      "loss": 0.844,
      "step": 3249
    },
    {
      "epoch": 0.5087664370695053,
      "grad_norm": 2.105788230895996,
      "learning_rate": 0.00023296676441837732,
      "loss": 1.3768,
      "step": 3250
    },
    {
      "epoch": 0.5089229805886036,
      "grad_norm": 0.5248891115188599,
      "learning_rate": 0.0002329423264907136,
      "loss": 0.3916,
      "step": 3251
    },
    {
      "epoch": 0.509079524107702,
      "grad_norm": 0.6104065179824829,
      "learning_rate": 0.00023291788856304982,
      "loss": 0.3574,
      "step": 3252
    },
    {
      "epoch": 0.5092360676268003,
      "grad_norm": 0.3868466913700104,
      "learning_rate": 0.0002328934506353861,
      "loss": 0.2812,
      "step": 3253
    },
    {
      "epoch": 0.5093926111458985,
      "grad_norm": 0.5593306422233582,
      "learning_rate": 0.00023286901270772238,
      "loss": 0.3395,
      "step": 3254
    },
    {
      "epoch": 0.5095491546649968,
      "grad_norm": 0.5168365836143494,
      "learning_rate": 0.00023284457478005863,
      "loss": 0.2767,
      "step": 3255
    },
    {
      "epoch": 0.5097056981840952,
      "grad_norm": 0.5223838686943054,
      "learning_rate": 0.0002328201368523949,
      "loss": 0.4546,
      "step": 3256
    },
    {
      "epoch": 0.5098622417031935,
      "grad_norm": 0.9906821846961975,
      "learning_rate": 0.00023279569892473118,
      "loss": 0.4196,
      "step": 3257
    },
    {
      "epoch": 0.5100187852222918,
      "grad_norm": 1.1748069524765015,
      "learning_rate": 0.00023277126099706743,
      "loss": 0.6491,
      "step": 3258
    },
    {
      "epoch": 0.5101753287413902,
      "grad_norm": 0.8128048777580261,
      "learning_rate": 0.00023274682306940369,
      "loss": 0.3803,
      "step": 3259
    },
    {
      "epoch": 0.5103318722604884,
      "grad_norm": 1.0079333782196045,
      "learning_rate": 0.00023272238514173996,
      "loss": 0.4582,
      "step": 3260
    },
    {
      "epoch": 0.5104884157795867,
      "grad_norm": 1.1329032182693481,
      "learning_rate": 0.00023269794721407622,
      "loss": 0.5789,
      "step": 3261
    },
    {
      "epoch": 0.510644959298685,
      "grad_norm": 0.8500834703445435,
      "learning_rate": 0.0002326735092864125,
      "loss": 0.3144,
      "step": 3262
    },
    {
      "epoch": 0.5108015028177834,
      "grad_norm": 0.8295961022377014,
      "learning_rate": 0.00023264907135874877,
      "loss": 0.4592,
      "step": 3263
    },
    {
      "epoch": 0.5109580463368817,
      "grad_norm": 1.165689468383789,
      "learning_rate": 0.00023262463343108502,
      "loss": 0.4923,
      "step": 3264
    },
    {
      "epoch": 0.51111458985598,
      "grad_norm": 1.2372647523880005,
      "learning_rate": 0.0002326001955034213,
      "loss": 0.4771,
      "step": 3265
    },
    {
      "epoch": 0.5112711333750782,
      "grad_norm": 1.0081822872161865,
      "learning_rate": 0.00023257575757575758,
      "loss": 0.4397,
      "step": 3266
    },
    {
      "epoch": 0.5114276768941766,
      "grad_norm": 1.4410061836242676,
      "learning_rate": 0.0002325513196480938,
      "loss": 0.5526,
      "step": 3267
    },
    {
      "epoch": 0.5115842204132749,
      "grad_norm": 1.1677322387695312,
      "learning_rate": 0.00023252688172043008,
      "loss": 0.4355,
      "step": 3268
    },
    {
      "epoch": 0.5117407639323732,
      "grad_norm": 1.1747709512710571,
      "learning_rate": 0.00023250244379276636,
      "loss": 0.5189,
      "step": 3269
    },
    {
      "epoch": 0.5118973074514716,
      "grad_norm": 1.338614821434021,
      "learning_rate": 0.0002324780058651026,
      "loss": 0.7244,
      "step": 3270
    },
    {
      "epoch": 0.5120538509705698,
      "grad_norm": 1.5333219766616821,
      "learning_rate": 0.0002324535679374389,
      "loss": 0.727,
      "step": 3271
    },
    {
      "epoch": 0.5122103944896681,
      "grad_norm": 1.5182764530181885,
      "learning_rate": 0.00023242913000977517,
      "loss": 0.737,
      "step": 3272
    },
    {
      "epoch": 0.5123669380087664,
      "grad_norm": 1.514541745185852,
      "learning_rate": 0.00023240469208211142,
      "loss": 0.672,
      "step": 3273
    },
    {
      "epoch": 0.5125234815278648,
      "grad_norm": 1.7871018648147583,
      "learning_rate": 0.0002323802541544477,
      "loss": 0.8974,
      "step": 3274
    },
    {
      "epoch": 0.5126800250469631,
      "grad_norm": 3.3677568435668945,
      "learning_rate": 0.00023235581622678398,
      "loss": 0.7339,
      "step": 3275
    },
    {
      "epoch": 0.5128365685660614,
      "grad_norm": 1.6806715726852417,
      "learning_rate": 0.0002323313782991202,
      "loss": 0.6502,
      "step": 3276
    },
    {
      "epoch": 0.5129931120851596,
      "grad_norm": 3.0515427589416504,
      "learning_rate": 0.00023230694037145648,
      "loss": 0.7692,
      "step": 3277
    },
    {
      "epoch": 0.513149655604258,
      "grad_norm": 1.5666495561599731,
      "learning_rate": 0.00023228250244379276,
      "loss": 1.0262,
      "step": 3278
    },
    {
      "epoch": 0.5133061991233563,
      "grad_norm": 2.228578805923462,
      "learning_rate": 0.000232258064516129,
      "loss": 0.5212,
      "step": 3279
    },
    {
      "epoch": 0.5134627426424546,
      "grad_norm": 1.372183084487915,
      "learning_rate": 0.0002322336265884653,
      "loss": 0.5467,
      "step": 3280
    },
    {
      "epoch": 0.513619286161553,
      "grad_norm": 3.1377692222595215,
      "learning_rate": 0.00023220918866080157,
      "loss": 1.0146,
      "step": 3281
    },
    {
      "epoch": 0.5137758296806513,
      "grad_norm": 2.914731979370117,
      "learning_rate": 0.0002321847507331378,
      "loss": 0.6966,
      "step": 3282
    },
    {
      "epoch": 0.5139323731997495,
      "grad_norm": 1.5598711967468262,
      "learning_rate": 0.00023216031280547407,
      "loss": 0.9771,
      "step": 3283
    },
    {
      "epoch": 0.5140889167188478,
      "grad_norm": 2.859375238418579,
      "learning_rate": 0.00023213587487781035,
      "loss": 1.0935,
      "step": 3284
    },
    {
      "epoch": 0.5142454602379462,
      "grad_norm": NaN,
      "learning_rate": 0.00023213587487781035,
      "loss": 0.0,
      "step": 3285
    },
    {
      "epoch": 0.5144020037570445,
      "grad_norm": 3.383396625518799,
      "learning_rate": 0.0002321114369501466,
      "loss": 1.3124,
      "step": 3286
    },
    {
      "epoch": 0.5145585472761428,
      "grad_norm": 4.065854549407959,
      "learning_rate": 0.00023208699902248288,
      "loss": 0.9329,
      "step": 3287
    },
    {
      "epoch": 0.514715090795241,
      "grad_norm": 2.2295284271240234,
      "learning_rate": 0.00023206256109481915,
      "loss": 1.4921,
      "step": 3288
    },
    {
      "epoch": 0.5148716343143394,
      "grad_norm": 1.8836601972579956,
      "learning_rate": 0.0002320381231671554,
      "loss": 1.0573,
      "step": 3289
    },
    {
      "epoch": 0.5150281778334377,
      "grad_norm": 3.414989709854126,
      "learning_rate": 0.00023201368523949168,
      "loss": 1.5235,
      "step": 3290
    },
    {
      "epoch": 0.515184721352536,
      "grad_norm": 2.405909776687622,
      "learning_rate": 0.00023198924731182796,
      "loss": 1.2622,
      "step": 3291
    },
    {
      "epoch": 0.5153412648716343,
      "grad_norm": 2.359362840652466,
      "learning_rate": 0.0002319648093841642,
      "loss": 1.4015,
      "step": 3292
    },
    {
      "epoch": 0.5154978083907327,
      "grad_norm": 1.3080377578735352,
      "learning_rate": 0.00023194037145650047,
      "loss": 1.2348,
      "step": 3293
    },
    {
      "epoch": 0.5156543519098309,
      "grad_norm": 3.4239389896392822,
      "learning_rate": 0.00023191593352883674,
      "loss": 1.8937,
      "step": 3294
    },
    {
      "epoch": 0.5158108954289292,
      "grad_norm": 2.4649817943573,
      "learning_rate": 0.000231891495601173,
      "loss": 1.7055,
      "step": 3295
    },
    {
      "epoch": 0.5159674389480275,
      "grad_norm": 2.2212533950805664,
      "learning_rate": 0.00023186705767350927,
      "loss": 0.9789,
      "step": 3296
    },
    {
      "epoch": 0.5161239824671259,
      "grad_norm": 1.765533685684204,
      "learning_rate": 0.00023184261974584555,
      "loss": 1.141,
      "step": 3297
    },
    {
      "epoch": 0.5162805259862242,
      "grad_norm": 2.404294013977051,
      "learning_rate": 0.0002318181818181818,
      "loss": 0.4344,
      "step": 3298
    },
    {
      "epoch": 0.5164370695053225,
      "grad_norm": 2.090888500213623,
      "learning_rate": 0.00023179374389051808,
      "loss": 0.9269,
      "step": 3299
    },
    {
      "epoch": 0.5165936130244208,
      "grad_norm": 1.3735568523406982,
      "learning_rate": 0.00023176930596285433,
      "loss": 1.101,
      "step": 3300
    },
    {
      "epoch": 0.5167501565435191,
      "grad_norm": 0.6953723430633545,
      "learning_rate": 0.00023174486803519058,
      "loss": 0.3792,
      "step": 3301
    },
    {
      "epoch": 0.5169067000626174,
      "grad_norm": 1.5250662565231323,
      "learning_rate": 0.00023172043010752686,
      "loss": 0.4269,
      "step": 3302
    },
    {
      "epoch": 0.5170632435817157,
      "grad_norm": 0.602741539478302,
      "learning_rate": 0.00023169599217986314,
      "loss": 0.3612,
      "step": 3303
    },
    {
      "epoch": 0.5172197871008141,
      "grad_norm": 2.0099756717681885,
      "learning_rate": 0.0002316715542521994,
      "loss": 0.5513,
      "step": 3304
    },
    {
      "epoch": 0.5173763306199123,
      "grad_norm": 0.6258875131607056,
      "learning_rate": 0.00023164711632453567,
      "loss": 0.3664,
      "step": 3305
    },
    {
      "epoch": 0.5175328741390106,
      "grad_norm": 0.8901511430740356,
      "learning_rate": 0.00023162267839687195,
      "loss": 0.5725,
      "step": 3306
    },
    {
      "epoch": 0.5176894176581089,
      "grad_norm": 0.6851283311843872,
      "learning_rate": 0.00023159824046920817,
      "loss": 0.6214,
      "step": 3307
    },
    {
      "epoch": 0.5178459611772073,
      "grad_norm": 0.9573870897293091,
      "learning_rate": 0.00023157380254154445,
      "loss": 0.4796,
      "step": 3308
    },
    {
      "epoch": 0.5180025046963056,
      "grad_norm": 1.0000059604644775,
      "learning_rate": 0.00023154936461388073,
      "loss": 0.3986,
      "step": 3309
    },
    {
      "epoch": 0.5181590482154039,
      "grad_norm": 2.725654125213623,
      "learning_rate": 0.00023152492668621698,
      "loss": 0.8138,
      "step": 3310
    },
    {
      "epoch": 0.5183155917345021,
      "grad_norm": 0.7232942581176758,
      "learning_rate": 0.00023150048875855326,
      "loss": 0.4892,
      "step": 3311
    },
    {
      "epoch": 0.5184721352536005,
      "grad_norm": 0.7372367978096008,
      "learning_rate": 0.00023147605083088954,
      "loss": 0.359,
      "step": 3312
    },
    {
      "epoch": 0.5186286787726988,
      "grad_norm": 0.6439573168754578,
      "learning_rate": 0.0002314516129032258,
      "loss": 0.4568,
      "step": 3313
    },
    {
      "epoch": 0.5187852222917971,
      "grad_norm": 0.9462945461273193,
      "learning_rate": 0.00023142717497556207,
      "loss": 0.4037,
      "step": 3314
    },
    {
      "epoch": 0.5189417658108955,
      "grad_norm": 1.0442487001419067,
      "learning_rate": 0.00023140273704789835,
      "loss": 0.5252,
      "step": 3315
    },
    {
      "epoch": 0.5190983093299938,
      "grad_norm": 1.7141872644424438,
      "learning_rate": 0.00023137829912023457,
      "loss": 0.6814,
      "step": 3316
    },
    {
      "epoch": 0.519254852849092,
      "grad_norm": 1.502058982849121,
      "learning_rate": 0.00023135386119257085,
      "loss": 0.7735,
      "step": 3317
    },
    {
      "epoch": 0.5194113963681903,
      "grad_norm": 1.0354453325271606,
      "learning_rate": 0.00023132942326490713,
      "loss": 0.4362,
      "step": 3318
    },
    {
      "epoch": 0.5195679398872887,
      "grad_norm": 0.9755420088768005,
      "learning_rate": 0.00023130498533724338,
      "loss": 0.5186,
      "step": 3319
    },
    {
      "epoch": 0.519724483406387,
      "grad_norm": 2.601315975189209,
      "learning_rate": 0.00023128054740957966,
      "loss": 0.5573,
      "step": 3320
    },
    {
      "epoch": 0.5198810269254853,
      "grad_norm": 0.9432253241539001,
      "learning_rate": 0.00023125610948191593,
      "loss": 0.5652,
      "step": 3321
    },
    {
      "epoch": 0.5200375704445835,
      "grad_norm": 1.0160396099090576,
      "learning_rate": 0.00023123167155425219,
      "loss": 0.63,
      "step": 3322
    },
    {
      "epoch": 0.5201941139636819,
      "grad_norm": 1.788796067237854,
      "learning_rate": 0.00023120723362658846,
      "loss": 0.6753,
      "step": 3323
    },
    {
      "epoch": 0.5203506574827802,
      "grad_norm": 2.158668041229248,
      "learning_rate": 0.0002311827956989247,
      "loss": 0.5324,
      "step": 3324
    },
    {
      "epoch": 0.5205072010018785,
      "grad_norm": 2.5072875022888184,
      "learning_rate": 0.00023115835777126097,
      "loss": 0.8801,
      "step": 3325
    },
    {
      "epoch": 0.5206637445209769,
      "grad_norm": 2.151085138320923,
      "learning_rate": 0.00023113391984359724,
      "loss": 0.6469,
      "step": 3326
    },
    {
      "epoch": 0.5208202880400752,
      "grad_norm": 1.2307536602020264,
      "learning_rate": 0.0002311094819159335,
      "loss": 0.5898,
      "step": 3327
    },
    {
      "epoch": 0.5209768315591734,
      "grad_norm": 1.785718321800232,
      "learning_rate": 0.00023108504398826977,
      "loss": 0.498,
      "step": 3328
    },
    {
      "epoch": 0.5211333750782717,
      "grad_norm": 2.295013666152954,
      "learning_rate": 0.00023106060606060605,
      "loss": 0.9652,
      "step": 3329
    },
    {
      "epoch": 0.5212899185973701,
      "grad_norm": 2.417727470397949,
      "learning_rate": 0.00023103616813294228,
      "loss": 0.7332,
      "step": 3330
    },
    {
      "epoch": 0.5214464621164684,
      "grad_norm": 2.488974094390869,
      "learning_rate": 0.00023101173020527855,
      "loss": 0.6885,
      "step": 3331
    },
    {
      "epoch": 0.5216030056355667,
      "grad_norm": 1.3490437269210815,
      "learning_rate": 0.00023098729227761483,
      "loss": 0.5777,
      "step": 3332
    },
    {
      "epoch": 0.521759549154665,
      "grad_norm": 2.2661380767822266,
      "learning_rate": 0.00023096285434995108,
      "loss": 0.9452,
      "step": 3333
    },
    {
      "epoch": 0.5219160926737633,
      "grad_norm": 3.1678786277770996,
      "learning_rate": 0.00023093841642228736,
      "loss": 0.7481,
      "step": 3334
    },
    {
      "epoch": 0.5220726361928616,
      "grad_norm": 3.216134548187256,
      "learning_rate": 0.00023091397849462364,
      "loss": 1.22,
      "step": 3335
    },
    {
      "epoch": 0.5222291797119599,
      "grad_norm": 1.301865816116333,
      "learning_rate": 0.0002308895405669599,
      "loss": 0.9253,
      "step": 3336
    },
    {
      "epoch": 0.5223857232310583,
      "grad_norm": 2.037886142730713,
      "learning_rate": 0.00023086510263929617,
      "loss": 0.5437,
      "step": 3337
    },
    {
      "epoch": 0.5225422667501566,
      "grad_norm": 2.8060178756713867,
      "learning_rate": 0.00023084066471163245,
      "loss": 1.2142,
      "step": 3338
    },
    {
      "epoch": 0.5226988102692548,
      "grad_norm": 2.1127607822418213,
      "learning_rate": 0.00023081622678396867,
      "loss": 1.3183,
      "step": 3339
    },
    {
      "epoch": 0.5228553537883531,
      "grad_norm": 2.004124879837036,
      "learning_rate": 0.00023079178885630495,
      "loss": 1.2346,
      "step": 3340
    },
    {
      "epoch": 0.5230118973074515,
      "grad_norm": 3.113992691040039,
      "learning_rate": 0.00023076735092864123,
      "loss": 1.6289,
      "step": 3341
    },
    {
      "epoch": 0.5231684408265498,
      "grad_norm": 1.6076124906539917,
      "learning_rate": 0.00023074291300097748,
      "loss": 1.2212,
      "step": 3342
    },
    {
      "epoch": 0.5233249843456481,
      "grad_norm": 1.8852514028549194,
      "learning_rate": 0.00023071847507331376,
      "loss": 1.0095,
      "step": 3343
    },
    {
      "epoch": 0.5234815278647464,
      "grad_norm": 4.059223651885986,
      "learning_rate": 0.00023069403714565004,
      "loss": 1.042,
      "step": 3344
    },
    {
      "epoch": 0.5236380713838447,
      "grad_norm": 2.242157459259033,
      "learning_rate": 0.0002306695992179863,
      "loss": 1.6616,
      "step": 3345
    },
    {
      "epoch": 0.523794614902943,
      "grad_norm": 2.6829006671905518,
      "learning_rate": 0.00023064516129032257,
      "loss": 1.2571,
      "step": 3346
    },
    {
      "epoch": 0.5239511584220413,
      "grad_norm": 1.4247349500656128,
      "learning_rate": 0.00023062072336265885,
      "loss": 0.7588,
      "step": 3347
    },
    {
      "epoch": 0.5241077019411396,
      "grad_norm": 4.124013423919678,
      "learning_rate": 0.00023059628543499507,
      "loss": 0.9346,
      "step": 3348
    },
    {
      "epoch": 0.524264245460238,
      "grad_norm": 1.38236665725708,
      "learning_rate": 0.00023057184750733135,
      "loss": 0.7006,
      "step": 3349
    },
    {
      "epoch": 0.5244207889793363,
      "grad_norm": 1.1349596977233887,
      "learning_rate": 0.00023054740957966763,
      "loss": 0.7983,
      "step": 3350
    },
    {
      "epoch": 0.5245773324984345,
      "grad_norm": 0.6386610269546509,
      "learning_rate": 0.00023052297165200388,
      "loss": 0.4338,
      "step": 3351
    },
    {
      "epoch": 0.5247338760175329,
      "grad_norm": 0.4819841980934143,
      "learning_rate": 0.00023049853372434016,
      "loss": 0.2964,
      "step": 3352
    },
    {
      "epoch": 0.5248904195366312,
      "grad_norm": 0.8476313352584839,
      "learning_rate": 0.00023047409579667643,
      "loss": 0.4256,
      "step": 3353
    },
    {
      "epoch": 0.5250469630557295,
      "grad_norm": 0.8270666003227234,
      "learning_rate": 0.00023044965786901266,
      "loss": 0.3657,
      "step": 3354
    },
    {
      "epoch": 0.5252035065748278,
      "grad_norm": 0.8374744653701782,
      "learning_rate": 0.00023042521994134894,
      "loss": 0.4129,
      "step": 3355
    },
    {
      "epoch": 0.5253600500939261,
      "grad_norm": 0.6448666453361511,
      "learning_rate": 0.00023040078201368522,
      "loss": 0.3006,
      "step": 3356
    },
    {
      "epoch": 0.5255165936130244,
      "grad_norm": 1.3359466791152954,
      "learning_rate": 0.00023037634408602147,
      "loss": 0.5692,
      "step": 3357
    },
    {
      "epoch": 0.5256731371321227,
      "grad_norm": 0.5898628830909729,
      "learning_rate": 0.00023035190615835775,
      "loss": 0.3944,
      "step": 3358
    },
    {
      "epoch": 0.525829680651221,
      "grad_norm": 1.0883667469024658,
      "learning_rate": 0.00023032746823069402,
      "loss": 0.5376,
      "step": 3359
    },
    {
      "epoch": 0.5259862241703194,
      "grad_norm": 0.5729881525039673,
      "learning_rate": 0.00023030303030303027,
      "loss": 0.4935,
      "step": 3360
    },
    {
      "epoch": 0.5261427676894177,
      "grad_norm": 1.0635184049606323,
      "learning_rate": 0.00023027859237536655,
      "loss": 0.4492,
      "step": 3361
    },
    {
      "epoch": 0.5262993112085159,
      "grad_norm": 15.57529067993164,
      "learning_rate": 0.00023025415444770283,
      "loss": 3.5422,
      "step": 3362
    },
    {
      "epoch": 0.5264558547276142,
      "grad_norm": 0.9564080238342285,
      "learning_rate": 0.00023022971652003906,
      "loss": 0.3876,
      "step": 3363
    },
    {
      "epoch": 0.5266123982467126,
      "grad_norm": 1.117563247680664,
      "learning_rate": 0.00023020527859237533,
      "loss": 0.493,
      "step": 3364
    },
    {
      "epoch": 0.5267689417658109,
      "grad_norm": 1.619956135749817,
      "learning_rate": 0.0002301808406647116,
      "loss": 0.7056,
      "step": 3365
    },
    {
      "epoch": 0.5269254852849092,
      "grad_norm": 1.5382646322250366,
      "learning_rate": 0.00023015640273704786,
      "loss": 0.5172,
      "step": 3366
    },
    {
      "epoch": 0.5270820288040076,
      "grad_norm": 2.17067813873291,
      "learning_rate": 0.00023013196480938414,
      "loss": 0.8012,
      "step": 3367
    },
    {
      "epoch": 0.5272385723231058,
      "grad_norm": 1.4361052513122559,
      "learning_rate": 0.00023010752688172042,
      "loss": 0.6448,
      "step": 3368
    },
    {
      "epoch": 0.5273951158422041,
      "grad_norm": 1.2153189182281494,
      "learning_rate": 0.00023008308895405667,
      "loss": 0.5594,
      "step": 3369
    },
    {
      "epoch": 0.5275516593613024,
      "grad_norm": 1.1564387083053589,
      "learning_rate": 0.00023005865102639295,
      "loss": 0.6764,
      "step": 3370
    },
    {
      "epoch": 0.5277082028804008,
      "grad_norm": 1.91134512424469,
      "learning_rate": 0.0002300342130987292,
      "loss": 0.3817,
      "step": 3371
    },
    {
      "epoch": 0.5278647463994991,
      "grad_norm": 1.3591880798339844,
      "learning_rate": 0.00023000977517106545,
      "loss": 0.8981,
      "step": 3372
    },
    {
      "epoch": 0.5280212899185974,
      "grad_norm": 0.9462252259254456,
      "learning_rate": 0.00022998533724340173,
      "loss": 0.777,
      "step": 3373
    },
    {
      "epoch": 0.5281778334376956,
      "grad_norm": 1.3694474697113037,
      "learning_rate": 0.000229960899315738,
      "loss": 0.5385,
      "step": 3374
    },
    {
      "epoch": 0.528334376956794,
      "grad_norm": 1.7175570726394653,
      "learning_rate": 0.00022993646138807426,
      "loss": 0.7427,
      "step": 3375
    },
    {
      "epoch": 0.5284909204758923,
      "grad_norm": 1.3409452438354492,
      "learning_rate": 0.00022991202346041054,
      "loss": 0.8763,
      "step": 3376
    },
    {
      "epoch": 0.5286474639949906,
      "grad_norm": 2.1568386554718018,
      "learning_rate": 0.00022988758553274682,
      "loss": 0.8583,
      "step": 3377
    },
    {
      "epoch": 0.528804007514089,
      "grad_norm": 1.5670572519302368,
      "learning_rate": 0.00022986314760508304,
      "loss": 0.7728,
      "step": 3378
    },
    {
      "epoch": 0.5289605510331872,
      "grad_norm": 3.02644419670105,
      "learning_rate": 0.00022983870967741932,
      "loss": 1.4348,
      "step": 3379
    },
    {
      "epoch": 0.5291170945522855,
      "grad_norm": 1.4505369663238525,
      "learning_rate": 0.0002298142717497556,
      "loss": 0.7838,
      "step": 3380
    },
    {
      "epoch": 0.5292736380713838,
      "grad_norm": 1.8097779750823975,
      "learning_rate": 0.00022978983382209185,
      "loss": 0.6987,
      "step": 3381
    },
    {
      "epoch": 0.5294301815904822,
      "grad_norm": 2.318065643310547,
      "learning_rate": 0.00022976539589442813,
      "loss": 0.9155,
      "step": 3382
    },
    {
      "epoch": 0.5295867251095805,
      "grad_norm": 1.7021375894546509,
      "learning_rate": 0.0002297409579667644,
      "loss": 0.4227,
      "step": 3383
    },
    {
      "epoch": 0.5297432686286788,
      "grad_norm": 1.7113087177276611,
      "learning_rate": 0.00022971652003910066,
      "loss": 0.9867,
      "step": 3384
    },
    {
      "epoch": 0.529899812147777,
      "grad_norm": 1.9517741203308105,
      "learning_rate": 0.00022969208211143694,
      "loss": 0.745,
      "step": 3385
    },
    {
      "epoch": 0.5300563556668754,
      "grad_norm": 2.208434820175171,
      "learning_rate": 0.00022966764418377321,
      "loss": 1.193,
      "step": 3386
    },
    {
      "epoch": 0.5302128991859737,
      "grad_norm": 1.7686792612075806,
      "learning_rate": 0.00022964320625610944,
      "loss": 1.2062,
      "step": 3387
    },
    {
      "epoch": 0.530369442705072,
      "grad_norm": 2.9696571826934814,
      "learning_rate": 0.00022961876832844572,
      "loss": 1.2893,
      "step": 3388
    },
    {
      "epoch": 0.5305259862241704,
      "grad_norm": 1.779247522354126,
      "learning_rate": 0.000229594330400782,
      "loss": 0.9809,
      "step": 3389
    },
    {
      "epoch": 0.5306825297432687,
      "grad_norm": 3.7535805702209473,
      "learning_rate": 0.00022956989247311825,
      "loss": 0.998,
      "step": 3390
    },
    {
      "epoch": 0.5308390732623669,
      "grad_norm": 4.040312767028809,
      "learning_rate": 0.00022954545454545452,
      "loss": 1.7128,
      "step": 3391
    },
    {
      "epoch": 0.5309956167814652,
      "grad_norm": 3.6977481842041016,
      "learning_rate": 0.0002295210166177908,
      "loss": 1.7015,
      "step": 3392
    },
    {
      "epoch": 0.5311521603005636,
      "grad_norm": 3.1999452114105225,
      "learning_rate": 0.00022949657869012705,
      "loss": 1.397,
      "step": 3393
    },
    {
      "epoch": 0.5313087038196619,
      "grad_norm": 1.5966978073120117,
      "learning_rate": 0.00022947214076246333,
      "loss": 0.6898,
      "step": 3394
    },
    {
      "epoch": 0.5314652473387602,
      "grad_norm": 3.1094870567321777,
      "learning_rate": 0.00022944770283479958,
      "loss": 1.1193,
      "step": 3395
    },
    {
      "epoch": 0.5316217908578584,
      "grad_norm": 4.34362268447876,
      "learning_rate": 0.00022942326490713583,
      "loss": 1.4771,
      "step": 3396
    },
    {
      "epoch": 0.5317783343769568,
      "grad_norm": 2.2344250679016113,
      "learning_rate": 0.0002293988269794721,
      "loss": 0.9172,
      "step": 3397
    },
    {
      "epoch": 0.5319348778960551,
      "grad_norm": 1.4617431163787842,
      "learning_rate": 0.0002293743890518084,
      "loss": 0.6987,
      "step": 3398
    },
    {
      "epoch": 0.5320914214151534,
      "grad_norm": 2.481635093688965,
      "learning_rate": 0.00022934995112414464,
      "loss": 1.653,
      "step": 3399
    },
    {
      "epoch": 0.5322479649342517,
      "grad_norm": 6.106346607208252,
      "learning_rate": 0.00022932551319648092,
      "loss": 1.5556,
      "step": 3400
    },
    {
      "epoch": 0.5324045084533501,
      "grad_norm": 0.7310137748718262,
      "learning_rate": 0.0002293010752688172,
      "loss": 0.4391,
      "step": 3401
    },
    {
      "epoch": 0.5325610519724483,
      "grad_norm": 0.5438380241394043,
      "learning_rate": 0.00022927663734115342,
      "loss": 0.3034,
      "step": 3402
    },
    {
      "epoch": 0.5327175954915466,
      "grad_norm": 0.5217984318733215,
      "learning_rate": 0.0002292521994134897,
      "loss": 0.3638,
      "step": 3403
    },
    {
      "epoch": 0.532874139010645,
      "grad_norm": 0.9906480312347412,
      "learning_rate": 0.00022922776148582598,
      "loss": 0.5653,
      "step": 3404
    },
    {
      "epoch": 0.5330306825297433,
      "grad_norm": 0.5368504524230957,
      "learning_rate": 0.00022920332355816223,
      "loss": 0.2615,
      "step": 3405
    },
    {
      "epoch": 0.5331872260488416,
      "grad_norm": 0.5763323903083801,
      "learning_rate": 0.0002291788856304985,
      "loss": 0.3562,
      "step": 3406
    },
    {
      "epoch": 0.5333437695679399,
      "grad_norm": 0.5000870227813721,
      "learning_rate": 0.0002291544477028348,
      "loss": 0.3222,
      "step": 3407
    },
    {
      "epoch": 0.5335003130870382,
      "grad_norm": 1.1395305395126343,
      "learning_rate": 0.00022913000977517104,
      "loss": 0.5258,
      "step": 3408
    },
    {
      "epoch": 0.5336568566061365,
      "grad_norm": 0.6404957175254822,
      "learning_rate": 0.00022910557184750732,
      "loss": 0.394,
      "step": 3409
    },
    {
      "epoch": 0.5338134001252348,
      "grad_norm": 0.6041001677513123,
      "learning_rate": 0.0002290811339198436,
      "loss": 0.4798,
      "step": 3410
    },
    {
      "epoch": 0.5339699436443331,
      "grad_norm": 0.9585782289505005,
      "learning_rate": 0.00022905669599217982,
      "loss": 0.4944,
      "step": 3411
    },
    {
      "epoch": 0.5341264871634315,
      "grad_norm": 1.1818557977676392,
      "learning_rate": 0.0002290322580645161,
      "loss": 0.4911,
      "step": 3412
    },
    {
      "epoch": 0.5342830306825297,
      "grad_norm": 1.2576113939285278,
      "learning_rate": 0.00022900782013685238,
      "loss": 0.4152,
      "step": 3413
    },
    {
      "epoch": 0.534439574201628,
      "grad_norm": 1.7009164094924927,
      "learning_rate": 0.00022898338220918863,
      "loss": 0.5462,
      "step": 3414
    },
    {
      "epoch": 0.5345961177207263,
      "grad_norm": 2.0021321773529053,
      "learning_rate": 0.0002289589442815249,
      "loss": 0.5423,
      "step": 3415
    },
    {
      "epoch": 0.5347526612398247,
      "grad_norm": 1.420309066772461,
      "learning_rate": 0.00022893450635386118,
      "loss": 0.4347,
      "step": 3416
    },
    {
      "epoch": 0.534909204758923,
      "grad_norm": 1.0252100229263306,
      "learning_rate": 0.00022891006842619744,
      "loss": 0.5028,
      "step": 3417
    },
    {
      "epoch": 0.5350657482780213,
      "grad_norm": 1.3138428926467896,
      "learning_rate": 0.00022888563049853371,
      "loss": 0.7375,
      "step": 3418
    },
    {
      "epoch": 0.5352222917971196,
      "grad_norm": 0.9500388503074646,
      "learning_rate": 0.00022886119257086997,
      "loss": 0.3795,
      "step": 3419
    },
    {
      "epoch": 0.5353788353162179,
      "grad_norm": 0.8862390518188477,
      "learning_rate": 0.00022883675464320622,
      "loss": 0.3207,
      "step": 3420
    },
    {
      "epoch": 0.5355353788353162,
      "grad_norm": 1.376956582069397,
      "learning_rate": 0.0002288123167155425,
      "loss": 0.6341,
      "step": 3421
    },
    {
      "epoch": 0.5356919223544145,
      "grad_norm": 3.1740610599517822,
      "learning_rate": 0.00022878787878787877,
      "loss": 0.7873,
      "step": 3422
    },
    {
      "epoch": 0.5358484658735129,
      "grad_norm": 1.5563631057739258,
      "learning_rate": 0.00022876344086021502,
      "loss": 0.6119,
      "step": 3423
    },
    {
      "epoch": 0.5360050093926112,
      "grad_norm": 2.3107411861419678,
      "learning_rate": 0.0002287390029325513,
      "loss": 0.9051,
      "step": 3424
    },
    {
      "epoch": 0.5361615529117094,
      "grad_norm": 1.220873236656189,
      "learning_rate": 0.00022871456500488758,
      "loss": 0.5476,
      "step": 3425
    },
    {
      "epoch": 0.5363180964308077,
      "grad_norm": 3.4480738639831543,
      "learning_rate": 0.0002286901270772238,
      "loss": 1.1999,
      "step": 3426
    },
    {
      "epoch": 0.5364746399499061,
      "grad_norm": 2.0392181873321533,
      "learning_rate": 0.00022866568914956008,
      "loss": 0.8199,
      "step": 3427
    },
    {
      "epoch": 0.5366311834690044,
      "grad_norm": 1.5251580476760864,
      "learning_rate": 0.00022864125122189636,
      "loss": 0.9086,
      "step": 3428
    },
    {
      "epoch": 0.5367877269881027,
      "grad_norm": 1.5712864398956299,
      "learning_rate": 0.00022861681329423261,
      "loss": 1.0547,
      "step": 3429
    },
    {
      "epoch": 0.536944270507201,
      "grad_norm": 2.3451268672943115,
      "learning_rate": 0.0002285923753665689,
      "loss": 0.6112,
      "step": 3430
    },
    {
      "epoch": 0.5371008140262993,
      "grad_norm": 2.3855226039886475,
      "learning_rate": 0.00022856793743890517,
      "loss": 1.3906,
      "step": 3431
    },
    {
      "epoch": 0.5372573575453976,
      "grad_norm": 2.014420747756958,
      "learning_rate": 0.00022854349951124142,
      "loss": 0.9376,
      "step": 3432
    },
    {
      "epoch": 0.5374139010644959,
      "grad_norm": 3.3102900981903076,
      "learning_rate": 0.0002285190615835777,
      "loss": 0.8464,
      "step": 3433
    },
    {
      "epoch": 0.5375704445835943,
      "grad_norm": 2.707771062850952,
      "learning_rate": 0.00022849462365591398,
      "loss": 0.8186,
      "step": 3434
    },
    {
      "epoch": 0.5377269881026926,
      "grad_norm": 1.4002277851104736,
      "learning_rate": 0.0002284701857282502,
      "loss": 1.0236,
      "step": 3435
    },
    {
      "epoch": 0.5378835316217908,
      "grad_norm": 2.523838996887207,
      "learning_rate": 0.00022844574780058648,
      "loss": 1.027,
      "step": 3436
    },
    {
      "epoch": 0.5380400751408891,
      "grad_norm": 2.5714972019195557,
      "learning_rate": 0.00022842130987292276,
      "loss": 1.0235,
      "step": 3437
    },
    {
      "epoch": 0.5381966186599875,
      "grad_norm": 3.162550926208496,
      "learning_rate": 0.000228396871945259,
      "loss": 1.1276,
      "step": 3438
    },
    {
      "epoch": 0.5383531621790858,
      "grad_norm": 1.9457319974899292,
      "learning_rate": 0.0002283724340175953,
      "loss": 1.0137,
      "step": 3439
    },
    {
      "epoch": 0.5385097056981841,
      "grad_norm": 1.903373122215271,
      "learning_rate": 0.00022834799608993157,
      "loss": 1.1942,
      "step": 3440
    },
    {
      "epoch": 0.5386662492172825,
      "grad_norm": 2.4530985355377197,
      "learning_rate": 0.00022832355816226782,
      "loss": 0.9727,
      "step": 3441
    },
    {
      "epoch": 0.5388227927363807,
      "grad_norm": 3.6654627323150635,
      "learning_rate": 0.00022829912023460407,
      "loss": 1.8357,
      "step": 3442
    },
    {
      "epoch": 0.538979336255479,
      "grad_norm": 2.052656650543213,
      "learning_rate": 0.00022827468230694035,
      "loss": 1.2149,
      "step": 3443
    },
    {
      "epoch": 0.5391358797745773,
      "grad_norm": 2.6127777099609375,
      "learning_rate": 0.0002282502443792766,
      "loss": 1.7517,
      "step": 3444
    },
    {
      "epoch": 0.5392924232936757,
      "grad_norm": 1.7864444255828857,
      "learning_rate": 0.00022822580645161288,
      "loss": 2.0915,
      "step": 3445
    },
    {
      "epoch": 0.539448966812774,
      "grad_norm": 1.7597533464431763,
      "learning_rate": 0.00022820136852394916,
      "loss": 0.6787,
      "step": 3446
    },
    {
      "epoch": 0.5396055103318722,
      "grad_norm": 1.367319941520691,
      "learning_rate": 0.0002281769305962854,
      "loss": 0.7049,
      "step": 3447
    },
    {
      "epoch": 0.5397620538509705,
      "grad_norm": 3.0287587642669678,
      "learning_rate": 0.00022815249266862169,
      "loss": 1.3499,
      "step": 3448
    },
    {
      "epoch": 0.5399185973700689,
      "grad_norm": 1.6455621719360352,
      "learning_rate": 0.00022812805474095796,
      "loss": 0.6301,
      "step": 3449
    },
    {
      "epoch": 0.5400751408891672,
      "grad_norm": 1.5209710597991943,
      "learning_rate": 0.0002281036168132942,
      "loss": 1.35,
      "step": 3450
    },
    {
      "epoch": 0.5402316844082655,
      "grad_norm": 0.5051299333572388,
      "learning_rate": 0.00022807917888563047,
      "loss": 0.3548,
      "step": 3451
    },
    {
      "epoch": 0.5403882279273639,
      "grad_norm": 0.9242395162582397,
      "learning_rate": 0.00022805474095796674,
      "loss": 0.518,
      "step": 3452
    },
    {
      "epoch": 0.5405447714464621,
      "grad_norm": 0.676452100276947,
      "learning_rate": 0.000228030303030303,
      "loss": 0.2494,
      "step": 3453
    },
    {
      "epoch": 0.5407013149655604,
      "grad_norm": 0.5453509092330933,
      "learning_rate": 0.00022800586510263927,
      "loss": 0.3723,
      "step": 3454
    },
    {
      "epoch": 0.5408578584846587,
      "grad_norm": 0.7466689348220825,
      "learning_rate": 0.00022798142717497555,
      "loss": 0.3795,
      "step": 3455
    },
    {
      "epoch": 0.5410144020037571,
      "grad_norm": 2.0270116329193115,
      "learning_rate": 0.0002279569892473118,
      "loss": 0.5283,
      "step": 3456
    },
    {
      "epoch": 0.5411709455228554,
      "grad_norm": 0.8889615535736084,
      "learning_rate": 0.00022793255131964808,
      "loss": 0.4212,
      "step": 3457
    },
    {
      "epoch": 0.5413274890419537,
      "grad_norm": 0.9550113677978516,
      "learning_rate": 0.00022790811339198436,
      "loss": 0.4152,
      "step": 3458
    },
    {
      "epoch": 0.5414840325610519,
      "grad_norm": 0.8343170881271362,
      "learning_rate": 0.00022788367546432058,
      "loss": 0.298,
      "step": 3459
    },
    {
      "epoch": 0.5416405760801503,
      "grad_norm": 0.9619960784912109,
      "learning_rate": 0.00022785923753665686,
      "loss": 0.4314,
      "step": 3460
    },
    {
      "epoch": 0.5417971195992486,
      "grad_norm": 0.5965786576271057,
      "learning_rate": 0.00022783479960899314,
      "loss": 0.3688,
      "step": 3461
    },
    {
      "epoch": 0.5419536631183469,
      "grad_norm": 0.9296935200691223,
      "learning_rate": 0.0002278103616813294,
      "loss": 0.4797,
      "step": 3462
    },
    {
      "epoch": 0.5421102066374452,
      "grad_norm": 1.0688129663467407,
      "learning_rate": 0.00022778592375366567,
      "loss": 0.513,
      "step": 3463
    },
    {
      "epoch": 0.5422667501565435,
      "grad_norm": 1.701122522354126,
      "learning_rate": 0.00022776148582600195,
      "loss": 0.6811,
      "step": 3464
    },
    {
      "epoch": 0.5424232936756418,
      "grad_norm": 1.191488265991211,
      "learning_rate": 0.0002277370478983382,
      "loss": 0.5554,
      "step": 3465
    },
    {
      "epoch": 0.5425798371947401,
      "grad_norm": 0.9095986485481262,
      "learning_rate": 0.00022771260997067445,
      "loss": 0.5331,
      "step": 3466
    },
    {
      "epoch": 0.5427363807138385,
      "grad_norm": 1.1927762031555176,
      "learning_rate": 0.00022768817204301073,
      "loss": 0.4866,
      "step": 3467
    },
    {
      "epoch": 0.5428929242329368,
      "grad_norm": 1.4253158569335938,
      "learning_rate": 0.00022766373411534698,
      "loss": 0.9596,
      "step": 3468
    },
    {
      "epoch": 0.5430494677520351,
      "grad_norm": 1.2457616329193115,
      "learning_rate": 0.00022763929618768326,
      "loss": 0.7098,
      "step": 3469
    },
    {
      "epoch": 0.5432060112711333,
      "grad_norm": 1.5363194942474365,
      "learning_rate": 0.00022761485826001954,
      "loss": 0.9893,
      "step": 3470
    },
    {
      "epoch": 0.5433625547902317,
      "grad_norm": 1.4052916765213013,
      "learning_rate": 0.0002275904203323558,
      "loss": 0.671,
      "step": 3471
    },
    {
      "epoch": 0.54351909830933,
      "grad_norm": 1.5679124593734741,
      "learning_rate": 0.00022756598240469207,
      "loss": 0.9099,
      "step": 3472
    },
    {
      "epoch": 0.5436756418284283,
      "grad_norm": 1.642220139503479,
      "learning_rate": 0.00022754154447702835,
      "loss": 0.6937,
      "step": 3473
    },
    {
      "epoch": 0.5438321853475266,
      "grad_norm": 2.2005414962768555,
      "learning_rate": 0.00022751710654936457,
      "loss": 0.6677,
      "step": 3474
    },
    {
      "epoch": 0.543988728866625,
      "grad_norm": 1.1939611434936523,
      "learning_rate": 0.00022749266862170085,
      "loss": 0.6974,
      "step": 3475
    },
    {
      "epoch": 0.5441452723857232,
      "grad_norm": 1.181011438369751,
      "learning_rate": 0.00022746823069403713,
      "loss": 0.7005,
      "step": 3476
    },
    {
      "epoch": 0.5443018159048215,
      "grad_norm": 2.0277912616729736,
      "learning_rate": 0.00022744379276637338,
      "loss": 1.0437,
      "step": 3477
    },
    {
      "epoch": 0.5444583594239198,
      "grad_norm": 2.596615791320801,
      "learning_rate": 0.00022741935483870966,
      "loss": 1.2138,
      "step": 3478
    },
    {
      "epoch": 0.5446149029430182,
      "grad_norm": 1.2545539140701294,
      "learning_rate": 0.00022739491691104593,
      "loss": 0.6488,
      "step": 3479
    },
    {
      "epoch": 0.5447714464621165,
      "grad_norm": 1.7148044109344482,
      "learning_rate": 0.00022737047898338219,
      "loss": 0.8829,
      "step": 3480
    },
    {
      "epoch": 0.5449279899812148,
      "grad_norm": 3.2928473949432373,
      "learning_rate": 0.00022734604105571846,
      "loss": 1.189,
      "step": 3481
    },
    {
      "epoch": 0.545084533500313,
      "grad_norm": 2.6867687702178955,
      "learning_rate": 0.00022732160312805474,
      "loss": 1.0388,
      "step": 3482
    },
    {
      "epoch": 0.5452410770194114,
      "grad_norm": 2.231804132461548,
      "learning_rate": 0.00022729716520039097,
      "loss": 0.8676,
      "step": 3483
    },
    {
      "epoch": 0.5453976205385097,
      "grad_norm": 1.7448841333389282,
      "learning_rate": 0.00022727272727272725,
      "loss": 0.8428,
      "step": 3484
    },
    {
      "epoch": 0.545554164057608,
      "grad_norm": 1.843652606010437,
      "learning_rate": 0.00022724828934506352,
      "loss": 0.788,
      "step": 3485
    },
    {
      "epoch": 0.5457107075767064,
      "grad_norm": 2.4735724925994873,
      "learning_rate": 0.00022722385141739978,
      "loss": 1.2652,
      "step": 3486
    },
    {
      "epoch": 0.5458672510958046,
      "grad_norm": 1.6568492650985718,
      "learning_rate": 0.00022719941348973605,
      "loss": 1.159,
      "step": 3487
    },
    {
      "epoch": 0.5460237946149029,
      "grad_norm": 2.2103464603424072,
      "learning_rate": 0.00022717497556207233,
      "loss": 1.0109,
      "step": 3488
    },
    {
      "epoch": 0.5461803381340012,
      "grad_norm": 2.2817304134368896,
      "learning_rate": 0.00022715053763440856,
      "loss": 0.7408,
      "step": 3489
    },
    {
      "epoch": 0.5463368816530996,
      "grad_norm": 1.5361602306365967,
      "learning_rate": 0.00022712609970674483,
      "loss": 0.9899,
      "step": 3490
    },
    {
      "epoch": 0.5464934251721979,
      "grad_norm": 2.22003436088562,
      "learning_rate": 0.0002271016617790811,
      "loss": 1.565,
      "step": 3491
    },
    {
      "epoch": 0.5466499686912962,
      "grad_norm": 1.5563693046569824,
      "learning_rate": 0.00022707722385141736,
      "loss": 1.1689,
      "step": 3492
    },
    {
      "epoch": 0.5468065122103944,
      "grad_norm": 1.7934682369232178,
      "learning_rate": 0.00022705278592375364,
      "loss": 1.2611,
      "step": 3493
    },
    {
      "epoch": 0.5469630557294928,
      "grad_norm": 1.765387773513794,
      "learning_rate": 0.00022702834799608992,
      "loss": 1.3886,
      "step": 3494
    },
    {
      "epoch": 0.5471195992485911,
      "grad_norm": 4.801526069641113,
      "learning_rate": 0.00022700391006842617,
      "loss": 1.7589,
      "step": 3495
    },
    {
      "epoch": 0.5472761427676894,
      "grad_norm": NaN,
      "learning_rate": 0.00022700391006842617,
      "loss": 0.0,
      "step": 3496
    },
    {
      "epoch": 0.5474326862867878,
      "grad_norm": 2.799443244934082,
      "learning_rate": 0.00022697947214076245,
      "loss": 1.0594,
      "step": 3497
    },
    {
      "epoch": 0.5475892298058861,
      "grad_norm": 1.0865267515182495,
      "learning_rate": 0.00022695503421309873,
      "loss": 0.6357,
      "step": 3498
    },
    {
      "epoch": 0.5477457733249843,
      "grad_norm": 1.8002020120620728,
      "learning_rate": 0.00022693059628543495,
      "loss": 0.5283,
      "step": 3499
    },
    {
      "epoch": 0.5479023168440826,
      "grad_norm": 2.474811315536499,
      "learning_rate": 0.00022690615835777123,
      "loss": 1.3573,
      "step": 3500
    },
    {
      "epoch": 0.548058860363181,
      "grad_norm": 0.7308578491210938,
      "learning_rate": 0.0002268817204301075,
      "loss": 0.255,
      "step": 3501
    },
    {
      "epoch": 0.5482154038822793,
      "grad_norm": 0.9229421615600586,
      "learning_rate": 0.00022685728250244376,
      "loss": 0.5089,
      "step": 3502
    },
    {
      "epoch": 0.5483719474013776,
      "grad_norm": 0.6674167513847351,
      "learning_rate": 0.00022683284457478004,
      "loss": 0.3171,
      "step": 3503
    },
    {
      "epoch": 0.5485284909204758,
      "grad_norm": 0.6612511873245239,
      "learning_rate": 0.00022680840664711632,
      "loss": 0.3403,
      "step": 3504
    },
    {
      "epoch": 0.5486850344395742,
      "grad_norm": 0.6101376414299011,
      "learning_rate": 0.00022678396871945257,
      "loss": 0.3597,
      "step": 3505
    },
    {
      "epoch": 0.5488415779586725,
      "grad_norm": 0.6959841251373291,
      "learning_rate": 0.00022675953079178885,
      "loss": 0.4338,
      "step": 3506
    },
    {
      "epoch": 0.5489981214777708,
      "grad_norm": 0.6419839262962341,
      "learning_rate": 0.00022673509286412513,
      "loss": 0.3483,
      "step": 3507
    },
    {
      "epoch": 0.5491546649968692,
      "grad_norm": 0.5177487134933472,
      "learning_rate": 0.00022671065493646135,
      "loss": 0.2916,
      "step": 3508
    },
    {
      "epoch": 0.5493112085159675,
      "grad_norm": 0.5536925196647644,
      "learning_rate": 0.00022668621700879763,
      "loss": 0.2922,
      "step": 3509
    },
    {
      "epoch": 0.5494677520350657,
      "grad_norm": 0.7985596656799316,
      "learning_rate": 0.0002266617790811339,
      "loss": 0.4654,
      "step": 3510
    },
    {
      "epoch": 0.549624295554164,
      "grad_norm": 0.6217026710510254,
      "learning_rate": 0.00022663734115347016,
      "loss": 0.423,
      "step": 3511
    },
    {
      "epoch": 0.5497808390732624,
      "grad_norm": 0.8252955675125122,
      "learning_rate": 0.00022661290322580644,
      "loss": 0.3649,
      "step": 3512
    },
    {
      "epoch": 0.5499373825923607,
      "grad_norm": 0.7620217204093933,
      "learning_rate": 0.00022658846529814271,
      "loss": 0.2662,
      "step": 3513
    },
    {
      "epoch": 0.550093926111459,
      "grad_norm": 1.2428134679794312,
      "learning_rate": 0.00022656402737047894,
      "loss": 0.8678,
      "step": 3514
    },
    {
      "epoch": 0.5502504696305573,
      "grad_norm": 1.2475255727767944,
      "learning_rate": 0.00022653958944281522,
      "loss": 0.6157,
      "step": 3515
    },
    {
      "epoch": 0.5504070131496556,
      "grad_norm": 0.6180065870285034,
      "learning_rate": 0.0002265151515151515,
      "loss": 0.4179,
      "step": 3516
    },
    {
      "epoch": 0.5505635566687539,
      "grad_norm": 1.0421558618545532,
      "learning_rate": 0.00022649071358748775,
      "loss": 0.5168,
      "step": 3517
    },
    {
      "epoch": 0.5507201001878522,
      "grad_norm": 1.2510666847229004,
      "learning_rate": 0.00022646627565982402,
      "loss": 0.4618,
      "step": 3518
    },
    {
      "epoch": 0.5508766437069506,
      "grad_norm": 0.9947577118873596,
      "learning_rate": 0.0002264418377321603,
      "loss": 0.4542,
      "step": 3519
    },
    {
      "epoch": 0.5510331872260489,
      "grad_norm": 1.770212173461914,
      "learning_rate": 0.00022641739980449655,
      "loss": 0.5551,
      "step": 3520
    },
    {
      "epoch": 0.5511897307451471,
      "grad_norm": 1.2235960960388184,
      "learning_rate": 0.00022639296187683283,
      "loss": 0.5831,
      "step": 3521
    },
    {
      "epoch": 0.5513462742642454,
      "grad_norm": 3.5953123569488525,
      "learning_rate": 0.0002263685239491691,
      "loss": 0.8161,
      "step": 3522
    },
    {
      "epoch": 0.5515028177833438,
      "grad_norm": 2.296663761138916,
      "learning_rate": 0.00022634408602150533,
      "loss": 0.8925,
      "step": 3523
    },
    {
      "epoch": 0.5516593613024421,
      "grad_norm": 1.9417630434036255,
      "learning_rate": 0.0002263196480938416,
      "loss": 1.1062,
      "step": 3524
    },
    {
      "epoch": 0.5518159048215404,
      "grad_norm": 1.476485013961792,
      "learning_rate": 0.0002262952101661779,
      "loss": 0.7668,
      "step": 3525
    },
    {
      "epoch": 0.5519724483406387,
      "grad_norm": 2.0487520694732666,
      "learning_rate": 0.00022627077223851414,
      "loss": 0.9898,
      "step": 3526
    },
    {
      "epoch": 0.552128991859737,
      "grad_norm": 4.198084354400635,
      "learning_rate": 0.00022624633431085042,
      "loss": 1.3848,
      "step": 3527
    },
    {
      "epoch": 0.5522855353788353,
      "grad_norm": 2.556408166885376,
      "learning_rate": 0.0002262218963831867,
      "loss": 0.8235,
      "step": 3528
    },
    {
      "epoch": 0.5524420788979336,
      "grad_norm": 2.0827739238739014,
      "learning_rate": 0.00022619745845552295,
      "loss": 0.8973,
      "step": 3529
    },
    {
      "epoch": 0.552598622417032,
      "grad_norm": 3.2092297077178955,
      "learning_rate": 0.00022617302052785923,
      "loss": 1.2371,
      "step": 3530
    },
    {
      "epoch": 0.5527551659361303,
      "grad_norm": 2.5607619285583496,
      "learning_rate": 0.00022614858260019548,
      "loss": 0.9781,
      "step": 3531
    },
    {
      "epoch": 0.5529117094552286,
      "grad_norm": 1.599326729774475,
      "learning_rate": 0.00022612414467253173,
      "loss": 1.0107,
      "step": 3532
    },
    {
      "epoch": 0.5530682529743268,
      "grad_norm": 2.7297661304473877,
      "learning_rate": 0.000226099706744868,
      "loss": 0.766,
      "step": 3533
    },
    {
      "epoch": 0.5532247964934252,
      "grad_norm": 2.919910430908203,
      "learning_rate": 0.0002260752688172043,
      "loss": 0.9601,
      "step": 3534
    },
    {
      "epoch": 0.5533813400125235,
      "grad_norm": 2.319716691970825,
      "learning_rate": 0.00022605083088954054,
      "loss": 0.9165,
      "step": 3535
    },
    {
      "epoch": 0.5535378835316218,
      "grad_norm": 2.5673739910125732,
      "learning_rate": 0.00022602639296187682,
      "loss": 0.954,
      "step": 3536
    },
    {
      "epoch": 0.5536944270507201,
      "grad_norm": 3.195618152618408,
      "learning_rate": 0.0002260019550342131,
      "loss": 0.9416,
      "step": 3537
    },
    {
      "epoch": 0.5538509705698184,
      "grad_norm": 3.3584465980529785,
      "learning_rate": 0.00022597751710654932,
      "loss": 1.2491,
      "step": 3538
    },
    {
      "epoch": 0.5540075140889167,
      "grad_norm": 2.711599349975586,
      "learning_rate": 0.0002259530791788856,
      "loss": 0.9559,
      "step": 3539
    },
    {
      "epoch": 0.554164057608015,
      "grad_norm": 3.927182912826538,
      "learning_rate": 0.00022592864125122188,
      "loss": 1.0383,
      "step": 3540
    },
    {
      "epoch": 0.5543206011271133,
      "grad_norm": 2.409851312637329,
      "learning_rate": 0.00022590420332355813,
      "loss": 1.2182,
      "step": 3541
    },
    {
      "epoch": 0.5544771446462117,
      "grad_norm": 2.19413423538208,
      "learning_rate": 0.0002258797653958944,
      "loss": 1.6216,
      "step": 3542
    },
    {
      "epoch": 0.55463368816531,
      "grad_norm": 2.6822988986968994,
      "learning_rate": 0.00022585532746823069,
      "loss": 1.5274,
      "step": 3543
    },
    {
      "epoch": 0.5547902316844082,
      "grad_norm": 1.415163516998291,
      "learning_rate": 0.00022583088954056694,
      "loss": 1.097,
      "step": 3544
    },
    {
      "epoch": 0.5549467752035065,
      "grad_norm": 2.0902304649353027,
      "learning_rate": 0.00022580645161290321,
      "loss": 0.7808,
      "step": 3545
    },
    {
      "epoch": 0.5551033187226049,
      "grad_norm": 2.3548595905303955,
      "learning_rate": 0.0002257820136852395,
      "loss": 0.4876,
      "step": 3546
    },
    {
      "epoch": 0.5552598622417032,
      "grad_norm": 1.715476632118225,
      "learning_rate": 0.00022575757575757572,
      "loss": 0.8252,
      "step": 3547
    },
    {
      "epoch": 0.5554164057608015,
      "grad_norm": 2.3043689727783203,
      "learning_rate": 0.000225733137829912,
      "loss": 0.8234,
      "step": 3548
    },
    {
      "epoch": 0.5555729492798999,
      "grad_norm": 1.6674572229385376,
      "learning_rate": 0.00022570869990224827,
      "loss": 1.091,
      "step": 3549
    },
    {
      "epoch": 0.5557294927989981,
      "grad_norm": 4.341742038726807,
      "learning_rate": 0.00022568426197458453,
      "loss": 1.4916,
      "step": 3550
    },
    {
      "epoch": 0.5558860363180964,
      "grad_norm": 0.5292091369628906,
      "learning_rate": 0.0002256598240469208,
      "loss": 0.3373,
      "step": 3551
    },
    {
      "epoch": 0.5560425798371947,
      "grad_norm": 0.5695773959159851,
      "learning_rate": 0.00022563538611925708,
      "loss": 0.316,
      "step": 3552
    },
    {
      "epoch": 0.5561991233562931,
      "grad_norm": 0.6100854277610779,
      "learning_rate": 0.00022561094819159333,
      "loss": 0.3825,
      "step": 3553
    },
    {
      "epoch": 0.5563556668753914,
      "grad_norm": 0.7489566802978516,
      "learning_rate": 0.0002255865102639296,
      "loss": 0.4018,
      "step": 3554
    },
    {
      "epoch": 0.5565122103944896,
      "grad_norm": 0.8366956114768982,
      "learning_rate": 0.00022556207233626586,
      "loss": 0.4573,
      "step": 3555
    },
    {
      "epoch": 0.5566687539135879,
      "grad_norm": 0.6405544877052307,
      "learning_rate": 0.00022553763440860211,
      "loss": 0.4246,
      "step": 3556
    },
    {
      "epoch": 0.5568252974326863,
      "grad_norm": 0.7330952882766724,
      "learning_rate": 0.0002255131964809384,
      "loss": 0.2276,
      "step": 3557
    },
    {
      "epoch": 0.5569818409517846,
      "grad_norm": 0.7685224413871765,
      "learning_rate": 0.00022548875855327467,
      "loss": 0.3543,
      "step": 3558
    },
    {
      "epoch": 0.5571383844708829,
      "grad_norm": 0.679145872592926,
      "learning_rate": 0.00022546432062561092,
      "loss": 0.506,
      "step": 3559
    },
    {
      "epoch": 0.5572949279899813,
      "grad_norm": 0.8586009740829468,
      "learning_rate": 0.0002254398826979472,
      "loss": 0.4348,
      "step": 3560
    },
    {
      "epoch": 0.5574514715090795,
      "grad_norm": 0.8567745685577393,
      "learning_rate": 0.00022541544477028348,
      "loss": 0.3119,
      "step": 3561
    },
    {
      "epoch": 0.5576080150281778,
      "grad_norm": 0.9251272082328796,
      "learning_rate": 0.0002253910068426197,
      "loss": 0.4098,
      "step": 3562
    },
    {
      "epoch": 0.5577645585472761,
      "grad_norm": 0.8240041136741638,
      "learning_rate": 0.00022536656891495598,
      "loss": 0.3483,
      "step": 3563
    },
    {
      "epoch": 0.5579211020663745,
      "grad_norm": 0.5420299172401428,
      "learning_rate": 0.00022534213098729226,
      "loss": 0.3981,
      "step": 3564
    },
    {
      "epoch": 0.5580776455854728,
      "grad_norm": NaN,
      "learning_rate": 0.00022534213098729226,
      "loss": 0.0,
      "step": 3565
    },
    {
      "epoch": 0.5582341891045711,
      "grad_norm": 5.230978488922119,
      "learning_rate": 0.0002253176930596285,
      "loss": 0.8704,
      "step": 3566
    },
    {
      "epoch": 0.5583907326236693,
      "grad_norm": 1.3879233598709106,
      "learning_rate": 0.0002252932551319648,
      "loss": 0.5912,
      "step": 3567
    },
    {
      "epoch": 0.5585472761427677,
      "grad_norm": 1.5150583982467651,
      "learning_rate": 0.00022526881720430107,
      "loss": 0.6871,
      "step": 3568
    },
    {
      "epoch": 0.558703819661866,
      "grad_norm": 1.0342755317687988,
      "learning_rate": 0.00022524437927663732,
      "loss": 0.6863,
      "step": 3569
    },
    {
      "epoch": 0.5588603631809643,
      "grad_norm": 1.3024189472198486,
      "learning_rate": 0.0002252199413489736,
      "loss": 0.7084,
      "step": 3570
    },
    {
      "epoch": 0.5590169067000627,
      "grad_norm": 1.985766887664795,
      "learning_rate": 0.00022519550342130988,
      "loss": 0.9109,
      "step": 3571
    },
    {
      "epoch": 0.5591734502191609,
      "grad_norm": 1.3010162115097046,
      "learning_rate": 0.0002251710654936461,
      "loss": 0.5658,
      "step": 3572
    },
    {
      "epoch": 0.5593299937382592,
      "grad_norm": 1.4666105508804321,
      "learning_rate": 0.00022514662756598238,
      "loss": 1.1083,
      "step": 3573
    },
    {
      "epoch": 0.5594865372573575,
      "grad_norm": 1.7869631052017212,
      "learning_rate": 0.00022512218963831866,
      "loss": 0.655,
      "step": 3574
    },
    {
      "epoch": 0.5596430807764559,
      "grad_norm": 1.4627180099487305,
      "learning_rate": 0.0002250977517106549,
      "loss": 0.62,
      "step": 3575
    },
    {
      "epoch": 0.5597996242955542,
      "grad_norm": 2.270118236541748,
      "learning_rate": 0.00022507331378299119,
      "loss": 1.0856,
      "step": 3576
    },
    {
      "epoch": 0.5599561678146525,
      "grad_norm": 1.2834758758544922,
      "learning_rate": 0.00022504887585532746,
      "loss": 0.3643,
      "step": 3577
    },
    {
      "epoch": 0.5601127113337507,
      "grad_norm": 2.207064628601074,
      "learning_rate": 0.00022502443792766372,
      "loss": 0.6055,
      "step": 3578
    },
    {
      "epoch": 0.5602692548528491,
      "grad_norm": 4.434436798095703,
      "learning_rate": 0.000225,
      "loss": 1.3798,
      "step": 3579
    },
    {
      "epoch": 0.5604257983719474,
      "grad_norm": 2.0381810665130615,
      "learning_rate": 0.00022497556207233625,
      "loss": 1.0831,
      "step": 3580
    },
    {
      "epoch": 0.5605823418910457,
      "grad_norm": 2.15852689743042,
      "learning_rate": 0.0002249511241446725,
      "loss": 0.7524,
      "step": 3581
    },
    {
      "epoch": 0.560738885410144,
      "grad_norm": 3.1960198879241943,
      "learning_rate": 0.00022492668621700877,
      "loss": 0.8807,
      "step": 3582
    },
    {
      "epoch": 0.5608954289292424,
      "grad_norm": 1.4990772008895874,
      "learning_rate": 0.00022490224828934505,
      "loss": 0.4864,
      "step": 3583
    },
    {
      "epoch": 0.5610519724483406,
      "grad_norm": 2.38069486618042,
      "learning_rate": 0.0002248778103616813,
      "loss": 0.9871,
      "step": 3584
    },
    {
      "epoch": 0.5612085159674389,
      "grad_norm": 3.0171408653259277,
      "learning_rate": 0.00022485337243401758,
      "loss": 1.1582,
      "step": 3585
    },
    {
      "epoch": 0.5613650594865373,
      "grad_norm": 2.1977977752685547,
      "learning_rate": 0.00022482893450635386,
      "loss": 0.6874,
      "step": 3586
    },
    {
      "epoch": 0.5615216030056356,
      "grad_norm": 2.109252452850342,
      "learning_rate": 0.00022480449657869009,
      "loss": 0.8322,
      "step": 3587
    },
    {
      "epoch": 0.5616781465247339,
      "grad_norm": 2.189072370529175,
      "learning_rate": 0.00022478005865102636,
      "loss": 1.0376,
      "step": 3588
    },
    {
      "epoch": 0.5618346900438321,
      "grad_norm": 3.6662800312042236,
      "learning_rate": 0.00022475562072336264,
      "loss": 1.7067,
      "step": 3589
    },
    {
      "epoch": 0.5619912335629305,
      "grad_norm": 2.0984466075897217,
      "learning_rate": 0.0002247311827956989,
      "loss": 0.9702,
      "step": 3590
    },
    {
      "epoch": 0.5621477770820288,
      "grad_norm": 1.8233792781829834,
      "learning_rate": 0.00022470674486803517,
      "loss": 1.1355,
      "step": 3591
    },
    {
      "epoch": 0.5623043206011271,
      "grad_norm": 1.7763501405715942,
      "learning_rate": 0.00022468230694037145,
      "loss": 1.3035,
      "step": 3592
    },
    {
      "epoch": 0.5624608641202254,
      "grad_norm": 3.240084171295166,
      "learning_rate": 0.0002246578690127077,
      "loss": 1.4573,
      "step": 3593
    },
    {
      "epoch": 0.5626174076393238,
      "grad_norm": 2.5310521125793457,
      "learning_rate": 0.00022463343108504398,
      "loss": 1.5207,
      "step": 3594
    },
    {
      "epoch": 0.562773951158422,
      "grad_norm": 1.6432092189788818,
      "learning_rate": 0.00022460899315738026,
      "loss": 1.0834,
      "step": 3595
    },
    {
      "epoch": 0.5629304946775203,
      "grad_norm": 0.8675611019134521,
      "learning_rate": 0.00022458455522971648,
      "loss": 0.5104,
      "step": 3596
    },
    {
      "epoch": 0.5630870381966186,
      "grad_norm": 1.3985061645507812,
      "learning_rate": 0.00022456011730205276,
      "loss": 0.7592,
      "step": 3597
    },
    {
      "epoch": 0.563243581715717,
      "grad_norm": 2.843593120574951,
      "learning_rate": 0.00022453567937438904,
      "loss": 0.8605,
      "step": 3598
    },
    {
      "epoch": 0.5634001252348153,
      "grad_norm": 2.444558620452881,
      "learning_rate": 0.0002245112414467253,
      "loss": 0.9727,
      "step": 3599
    },
    {
      "epoch": 0.5635566687539136,
      "grad_norm": 2.5446102619171143,
      "learning_rate": 0.00022448680351906157,
      "loss": 1.3655,
      "step": 3600
    },
    {
      "epoch": 0.5637132122730119,
      "grad_norm": 0.5711503624916077,
      "learning_rate": 0.00022446236559139785,
      "loss": 0.35,
      "step": 3601
    },
    {
      "epoch": 0.5638697557921102,
      "grad_norm": 0.39799460768699646,
      "learning_rate": 0.0002244379276637341,
      "loss": 0.238,
      "step": 3602
    },
    {
      "epoch": 0.5640262993112085,
      "grad_norm": 0.5093404054641724,
      "learning_rate": 0.00022441348973607035,
      "loss": 0.4052,
      "step": 3603
    },
    {
      "epoch": 0.5641828428303068,
      "grad_norm": 0.7445962429046631,
      "learning_rate": 0.00022438905180840663,
      "loss": 0.6028,
      "step": 3604
    },
    {
      "epoch": 0.5643393863494052,
      "grad_norm": 0.8046715259552002,
      "learning_rate": 0.00022436461388074288,
      "loss": 0.3982,
      "step": 3605
    },
    {
      "epoch": 0.5644959298685035,
      "grad_norm": 0.7545210123062134,
      "learning_rate": 0.00022434017595307916,
      "loss": 0.3914,
      "step": 3606
    },
    {
      "epoch": 0.5646524733876017,
      "grad_norm": 0.8643534183502197,
      "learning_rate": 0.00022431573802541544,
      "loss": 0.4382,
      "step": 3607
    },
    {
      "epoch": 0.5648090169067,
      "grad_norm": 0.7606629133224487,
      "learning_rate": 0.0002242913000977517,
      "loss": 0.3623,
      "step": 3608
    },
    {
      "epoch": 0.5649655604257984,
      "grad_norm": 1.1187993288040161,
      "learning_rate": 0.00022426686217008796,
      "loss": 0.5698,
      "step": 3609
    },
    {
      "epoch": 0.5651221039448967,
      "grad_norm": 0.6099269390106201,
      "learning_rate": 0.00022424242424242424,
      "loss": 0.319,
      "step": 3610
    },
    {
      "epoch": 0.565278647463995,
      "grad_norm": 1.3326719999313354,
      "learning_rate": 0.00022421798631476047,
      "loss": 0.6025,
      "step": 3611
    },
    {
      "epoch": 0.5654351909830932,
      "grad_norm": 2.37567400932312,
      "learning_rate": 0.00022419354838709675,
      "loss": 1.0935,
      "step": 3612
    },
    {
      "epoch": 0.5655917345021916,
      "grad_norm": 0.8539568185806274,
      "learning_rate": 0.00022416911045943302,
      "loss": 0.3043,
      "step": 3613
    },
    {
      "epoch": 0.5657482780212899,
      "grad_norm": 1.237491488456726,
      "learning_rate": 0.00022414467253176928,
      "loss": 0.4338,
      "step": 3614
    },
    {
      "epoch": 0.5659048215403882,
      "grad_norm": 0.9956717491149902,
      "learning_rate": 0.00022412023460410555,
      "loss": 0.4656,
      "step": 3615
    },
    {
      "epoch": 0.5660613650594866,
      "grad_norm": 1.1546032428741455,
      "learning_rate": 0.00022409579667644183,
      "loss": 0.7444,
      "step": 3616
    },
    {
      "epoch": 0.5662179085785849,
      "grad_norm": 1.377625823020935,
      "learning_rate": 0.00022407135874877808,
      "loss": 0.8117,
      "step": 3617
    },
    {
      "epoch": 0.5663744520976831,
      "grad_norm": 1.8123804330825806,
      "learning_rate": 0.00022404692082111436,
      "loss": 0.566,
      "step": 3618
    },
    {
      "epoch": 0.5665309956167814,
      "grad_norm": 1.1724187135696411,
      "learning_rate": 0.00022402248289345064,
      "loss": 0.5954,
      "step": 3619
    },
    {
      "epoch": 0.5666875391358798,
      "grad_norm": 1.4252662658691406,
      "learning_rate": 0.00022399804496578686,
      "loss": 0.6772,
      "step": 3620
    },
    {
      "epoch": 0.5668440826549781,
      "grad_norm": 2.668489456176758,
      "learning_rate": 0.00022397360703812314,
      "loss": 0.897,
      "step": 3621
    },
    {
      "epoch": 0.5670006261740764,
      "grad_norm": 1.0190789699554443,
      "learning_rate": 0.00022394916911045942,
      "loss": 0.6887,
      "step": 3622
    },
    {
      "epoch": 0.5671571696931748,
      "grad_norm": 1.3881101608276367,
      "learning_rate": 0.00022392473118279567,
      "loss": 0.7768,
      "step": 3623
    },
    {
      "epoch": 0.567313713212273,
      "grad_norm": 1.1338660717010498,
      "learning_rate": 0.00022390029325513195,
      "loss": 0.6853,
      "step": 3624
    },
    {
      "epoch": 0.5674702567313713,
      "grad_norm": 1.3985800743103027,
      "learning_rate": 0.00022387585532746823,
      "loss": 0.7319,
      "step": 3625
    },
    {
      "epoch": 0.5676268002504696,
      "grad_norm": 3.043156147003174,
      "learning_rate": 0.00022385141739980448,
      "loss": 1.2041,
      "step": 3626
    },
    {
      "epoch": 0.567783343769568,
      "grad_norm": 1.2043315172195435,
      "learning_rate": 0.00022382697947214073,
      "loss": 0.5941,
      "step": 3627
    },
    {
      "epoch": 0.5679398872886663,
      "grad_norm": 1.9357560873031616,
      "learning_rate": 0.000223802541544477,
      "loss": 1.1688,
      "step": 3628
    },
    {
      "epoch": 0.5680964308077645,
      "grad_norm": 2.4211204051971436,
      "learning_rate": 0.00022377810361681326,
      "loss": 0.9962,
      "step": 3629
    },
    {
      "epoch": 0.5682529743268628,
      "grad_norm": 2.997868299484253,
      "learning_rate": 0.00022375366568914954,
      "loss": 0.7847,
      "step": 3630
    },
    {
      "epoch": 0.5684095178459612,
      "grad_norm": 2.5542614459991455,
      "learning_rate": 0.00022372922776148582,
      "loss": 1.0314,
      "step": 3631
    },
    {
      "epoch": 0.5685660613650595,
      "grad_norm": 2.0041849613189697,
      "learning_rate": 0.00022370478983382207,
      "loss": 1.4012,
      "step": 3632
    },
    {
      "epoch": 0.5687226048841578,
      "grad_norm": 2.1820318698883057,
      "learning_rate": 0.00022368035190615835,
      "loss": 1.2137,
      "step": 3633
    },
    {
      "epoch": 0.5688791484032562,
      "grad_norm": 2.45853853225708,
      "learning_rate": 0.00022365591397849463,
      "loss": 1.0993,
      "step": 3634
    },
    {
      "epoch": 0.5690356919223544,
      "grad_norm": 2.2817904949188232,
      "learning_rate": 0.00022363147605083085,
      "loss": 1.1108,
      "step": 3635
    },
    {
      "epoch": 0.5691922354414527,
      "grad_norm": 1.8414242267608643,
      "learning_rate": 0.00022360703812316713,
      "loss": 0.9754,
      "step": 3636
    },
    {
      "epoch": 0.569348778960551,
      "grad_norm": 1.4161016941070557,
      "learning_rate": 0.0002235826001955034,
      "loss": 0.6568,
      "step": 3637
    },
    {
      "epoch": 0.5695053224796494,
      "grad_norm": 1.995675802230835,
      "learning_rate": 0.00022355816226783966,
      "loss": 1.1306,
      "step": 3638
    },
    {
      "epoch": 0.5696618659987477,
      "grad_norm": 2.4192090034484863,
      "learning_rate": 0.00022353372434017594,
      "loss": 1.55,
      "step": 3639
    },
    {
      "epoch": 0.569818409517846,
      "grad_norm": 3.0786073207855225,
      "learning_rate": 0.00022350928641251221,
      "loss": 1.8578,
      "step": 3640
    },
    {
      "epoch": 0.5699749530369442,
      "grad_norm": 2.814448118209839,
      "learning_rate": 0.00022348484848484847,
      "loss": 1.5919,
      "step": 3641
    },
    {
      "epoch": 0.5701314965560426,
      "grad_norm": 1.6886667013168335,
      "learning_rate": 0.00022346041055718474,
      "loss": 0.9254,
      "step": 3642
    },
    {
      "epoch": 0.5702880400751409,
      "grad_norm": 3.294468879699707,
      "learning_rate": 0.00022343597262952102,
      "loss": 1.1636,
      "step": 3643
    },
    {
      "epoch": 0.5704445835942392,
      "grad_norm": 2.3327720165252686,
      "learning_rate": 0.00022341153470185725,
      "loss": 1.2624,
      "step": 3644
    },
    {
      "epoch": 0.5706011271133375,
      "grad_norm": 1.5095564126968384,
      "learning_rate": 0.00022338709677419352,
      "loss": 0.4536,
      "step": 3645
    },
    {
      "epoch": 0.5707576706324358,
      "grad_norm": 3.851837635040283,
      "learning_rate": 0.0002233626588465298,
      "loss": 0.9293,
      "step": 3646
    },
    {
      "epoch": 0.5709142141515341,
      "grad_norm": 2.2999038696289062,
      "learning_rate": 0.00022333822091886605,
      "loss": 0.8068,
      "step": 3647
    },
    {
      "epoch": 0.5710707576706324,
      "grad_norm": 1.686329960823059,
      "learning_rate": 0.00022331378299120233,
      "loss": 0.3304,
      "step": 3648
    },
    {
      "epoch": 0.5712273011897308,
      "grad_norm": 2.073951244354248,
      "learning_rate": 0.0002232893450635386,
      "loss": 0.7134,
      "step": 3649
    },
    {
      "epoch": 0.5713838447088291,
      "grad_norm": 1.7170746326446533,
      "learning_rate": 0.00022326490713587484,
      "loss": 1.1497,
      "step": 3650
    },
    {
      "epoch": 0.5715403882279274,
      "grad_norm": 0.47634878754615784,
      "learning_rate": 0.00022324046920821111,
      "loss": 0.3378,
      "step": 3651
    },
    {
      "epoch": 0.5716969317470256,
      "grad_norm": 0.629563570022583,
      "learning_rate": 0.0002232160312805474,
      "loss": 0.4027,
      "step": 3652
    },
    {
      "epoch": 0.571853475266124,
      "grad_norm": 0.8603024482727051,
      "learning_rate": 0.00022319159335288364,
      "loss": 0.3611,
      "step": 3653
    },
    {
      "epoch": 0.5720100187852223,
      "grad_norm": 0.5923214554786682,
      "learning_rate": 0.00022316715542521992,
      "loss": 0.3809,
      "step": 3654
    },
    {
      "epoch": 0.5721665623043206,
      "grad_norm": 0.8817975521087646,
      "learning_rate": 0.0002231427174975562,
      "loss": 0.4392,
      "step": 3655
    },
    {
      "epoch": 0.5723231058234189,
      "grad_norm": 0.8190473914146423,
      "learning_rate": 0.00022311827956989245,
      "loss": 0.4117,
      "step": 3656
    },
    {
      "epoch": 0.5724796493425173,
      "grad_norm": 0.6798378229141235,
      "learning_rate": 0.00022309384164222873,
      "loss": 0.2774,
      "step": 3657
    },
    {
      "epoch": 0.5726361928616155,
      "grad_norm": 0.8386220932006836,
      "learning_rate": 0.000223069403714565,
      "loss": 0.4713,
      "step": 3658
    },
    {
      "epoch": 0.5727927363807138,
      "grad_norm": 0.8392332196235657,
      "learning_rate": 0.00022304496578690123,
      "loss": 0.5223,
      "step": 3659
    },
    {
      "epoch": 0.5729492798998121,
      "grad_norm": 1.1230872869491577,
      "learning_rate": 0.0002230205278592375,
      "loss": 0.4972,
      "step": 3660
    },
    {
      "epoch": 0.5731058234189105,
      "grad_norm": 1.0100561380386353,
      "learning_rate": 0.0002229960899315738,
      "loss": 0.3689,
      "step": 3661
    },
    {
      "epoch": 0.5732623669380088,
      "grad_norm": 0.7954285144805908,
      "learning_rate": 0.00022297165200391004,
      "loss": 0.3825,
      "step": 3662
    },
    {
      "epoch": 0.573418910457107,
      "grad_norm": 1.0235164165496826,
      "learning_rate": 0.00022294721407624632,
      "loss": 0.5164,
      "step": 3663
    },
    {
      "epoch": 0.5735754539762054,
      "grad_norm": 1.4720885753631592,
      "learning_rate": 0.0002229227761485826,
      "loss": 0.5941,
      "step": 3664
    },
    {
      "epoch": 0.5737319974953037,
      "grad_norm": 0.6552546620368958,
      "learning_rate": 0.00022289833822091885,
      "loss": 0.4673,
      "step": 3665
    },
    {
      "epoch": 0.573888541014402,
      "grad_norm": 0.9891802072525024,
      "learning_rate": 0.00022287390029325513,
      "loss": 0.5022,
      "step": 3666
    },
    {
      "epoch": 0.5740450845335003,
      "grad_norm": 1.503885269165039,
      "learning_rate": 0.0002228494623655914,
      "loss": 0.5776,
      "step": 3667
    },
    {
      "epoch": 0.5742016280525987,
      "grad_norm": 1.1407116651535034,
      "learning_rate": 0.00022282502443792763,
      "loss": 0.5049,
      "step": 3668
    },
    {
      "epoch": 0.5743581715716969,
      "grad_norm": 1.1594735383987427,
      "learning_rate": 0.0002228005865102639,
      "loss": 0.5241,
      "step": 3669
    },
    {
      "epoch": 0.5745147150907952,
      "grad_norm": 0.8916786313056946,
      "learning_rate": 0.00022277614858260019,
      "loss": 0.3553,
      "step": 3670
    },
    {
      "epoch": 0.5746712586098935,
      "grad_norm": 1.9808990955352783,
      "learning_rate": 0.00022275171065493644,
      "loss": 0.7108,
      "step": 3671
    },
    {
      "epoch": 0.5748278021289919,
      "grad_norm": 1.4254517555236816,
      "learning_rate": 0.00022272727272727272,
      "loss": 0.5996,
      "step": 3672
    },
    {
      "epoch": 0.5749843456480902,
      "grad_norm": 1.409542441368103,
      "learning_rate": 0.000222702834799609,
      "loss": 0.691,
      "step": 3673
    },
    {
      "epoch": 0.5751408891671885,
      "grad_norm": 1.5165317058563232,
      "learning_rate": 0.00022267839687194522,
      "loss": 0.5769,
      "step": 3674
    },
    {
      "epoch": 0.5752974326862867,
      "grad_norm": 2.6416120529174805,
      "learning_rate": 0.0002226539589442815,
      "loss": 0.7262,
      "step": 3675
    },
    {
      "epoch": 0.5754539762053851,
      "grad_norm": 2.1240227222442627,
      "learning_rate": 0.00022262952101661777,
      "loss": 0.6493,
      "step": 3676
    },
    {
      "epoch": 0.5756105197244834,
      "grad_norm": 1.722774624824524,
      "learning_rate": 0.00022260508308895403,
      "loss": 0.8552,
      "step": 3677
    },
    {
      "epoch": 0.5757670632435817,
      "grad_norm": 1.5990309715270996,
      "learning_rate": 0.0002225806451612903,
      "loss": 1.0542,
      "step": 3678
    },
    {
      "epoch": 0.5759236067626801,
      "grad_norm": 1.7411298751831055,
      "learning_rate": 0.00022255620723362658,
      "loss": 0.752,
      "step": 3679
    },
    {
      "epoch": 0.5760801502817783,
      "grad_norm": 2.393709182739258,
      "learning_rate": 0.00022253176930596283,
      "loss": 1.3362,
      "step": 3680
    },
    {
      "epoch": 0.5762366938008766,
      "grad_norm": 1.4573614597320557,
      "learning_rate": 0.0002225073313782991,
      "loss": 0.77,
      "step": 3681
    },
    {
      "epoch": 0.5763932373199749,
      "grad_norm": 1.8054333925247192,
      "learning_rate": 0.0002224828934506354,
      "loss": 0.9428,
      "step": 3682
    },
    {
      "epoch": 0.5765497808390733,
      "grad_norm": 2.158707618713379,
      "learning_rate": 0.00022245845552297161,
      "loss": 0.8786,
      "step": 3683
    },
    {
      "epoch": 0.5767063243581716,
      "grad_norm": 1.175864577293396,
      "learning_rate": 0.0002224340175953079,
      "loss": 0.7271,
      "step": 3684
    },
    {
      "epoch": 0.5768628678772699,
      "grad_norm": 3.2046947479248047,
      "learning_rate": 0.00022240957966764417,
      "loss": 1.1071,
      "step": 3685
    },
    {
      "epoch": 0.5770194113963681,
      "grad_norm": 2.6925206184387207,
      "learning_rate": 0.00022238514173998042,
      "loss": 1.247,
      "step": 3686
    },
    {
      "epoch": 0.5771759549154665,
      "grad_norm": 2.9383318424224854,
      "learning_rate": 0.0002223607038123167,
      "loss": 0.9495,
      "step": 3687
    },
    {
      "epoch": 0.5773324984345648,
      "grad_norm": 2.484539031982422,
      "learning_rate": 0.00022233626588465298,
      "loss": 0.862,
      "step": 3688
    },
    {
      "epoch": 0.5774890419536631,
      "grad_norm": 3.069092035293579,
      "learning_rate": 0.00022231182795698923,
      "loss": 1.6457,
      "step": 3689
    },
    {
      "epoch": 0.5776455854727615,
      "grad_norm": 2.6385087966918945,
      "learning_rate": 0.0002222873900293255,
      "loss": 0.7699,
      "step": 3690
    },
    {
      "epoch": 0.5778021289918598,
      "grad_norm": 4.0282487869262695,
      "learning_rate": 0.00022226295210166176,
      "loss": 1.2982,
      "step": 3691
    },
    {
      "epoch": 0.577958672510958,
      "grad_norm": 1.7925729751586914,
      "learning_rate": 0.000222238514173998,
      "loss": 1.1737,
      "step": 3692
    },
    {
      "epoch": 0.5781152160300563,
      "grad_norm": 3.2553768157958984,
      "learning_rate": 0.0002222140762463343,
      "loss": 1.3953,
      "step": 3693
    },
    {
      "epoch": 0.5782717595491547,
      "grad_norm": 3.523599147796631,
      "learning_rate": 0.00022218963831867057,
      "loss": 0.9012,
      "step": 3694
    },
    {
      "epoch": 0.578428303068253,
      "grad_norm": 2.5558550357818604,
      "learning_rate": 0.00022216520039100682,
      "loss": 1.0458,
      "step": 3695
    },
    {
      "epoch": 0.5785848465873513,
      "grad_norm": 0.8834622502326965,
      "learning_rate": 0.0002221407624633431,
      "loss": 0.251,
      "step": 3696
    },
    {
      "epoch": 0.5787413901064495,
      "grad_norm": 1.9280049800872803,
      "learning_rate": 0.00022211632453567938,
      "loss": 0.653,
      "step": 3697
    },
    {
      "epoch": 0.5788979336255479,
      "grad_norm": 2.2830820083618164,
      "learning_rate": 0.0002220918866080156,
      "loss": 0.8649,
      "step": 3698
    },
    {
      "epoch": 0.5790544771446462,
      "grad_norm": 3.383026123046875,
      "learning_rate": 0.00022206744868035188,
      "loss": 0.7386,
      "step": 3699
    },
    {
      "epoch": 0.5792110206637445,
      "grad_norm": 2.898946523666382,
      "learning_rate": 0.00022204301075268816,
      "loss": 1.1781,
      "step": 3700
    },
    {
      "epoch": 0.5793675641828429,
      "grad_norm": 0.6618011593818665,
      "learning_rate": 0.0002220185728250244,
      "loss": 0.4284,
      "step": 3701
    },
    {
      "epoch": 0.5795241077019412,
      "grad_norm": 0.7165639996528625,
      "learning_rate": 0.00022199413489736069,
      "loss": 0.3873,
      "step": 3702
    },
    {
      "epoch": 0.5796806512210394,
      "grad_norm": 0.8477781414985657,
      "learning_rate": 0.00022196969696969696,
      "loss": 0.5621,
      "step": 3703
    },
    {
      "epoch": 0.5798371947401377,
      "grad_norm": 1.0597972869873047,
      "learning_rate": 0.00022194525904203322,
      "loss": 0.5527,
      "step": 3704
    },
    {
      "epoch": 0.5799937382592361,
      "grad_norm": 0.9503958225250244,
      "learning_rate": 0.0002219208211143695,
      "loss": 0.4481,
      "step": 3705
    },
    {
      "epoch": 0.5801502817783344,
      "grad_norm": 1.0318113565444946,
      "learning_rate": 0.00022189638318670577,
      "loss": 0.4143,
      "step": 3706
    },
    {
      "epoch": 0.5803068252974327,
      "grad_norm": 0.5940424203872681,
      "learning_rate": 0.000221871945259042,
      "loss": 0.3822,
      "step": 3707
    },
    {
      "epoch": 0.580463368816531,
      "grad_norm": 0.8478888869285583,
      "learning_rate": 0.00022184750733137828,
      "loss": 0.3572,
      "step": 3708
    },
    {
      "epoch": 0.5806199123356293,
      "grad_norm": 0.69258052110672,
      "learning_rate": 0.00022182306940371455,
      "loss": 0.2797,
      "step": 3709
    },
    {
      "epoch": 0.5807764558547276,
      "grad_norm": 0.7424476742744446,
      "learning_rate": 0.0002217986314760508,
      "loss": 0.4239,
      "step": 3710
    },
    {
      "epoch": 0.5809329993738259,
      "grad_norm": 1.1544733047485352,
      "learning_rate": 0.00022177419354838708,
      "loss": 0.4964,
      "step": 3711
    },
    {
      "epoch": 0.5810895428929242,
      "grad_norm": 1.1875557899475098,
      "learning_rate": 0.00022174975562072336,
      "loss": 0.5127,
      "step": 3712
    },
    {
      "epoch": 0.5812460864120226,
      "grad_norm": 1.065439224243164,
      "learning_rate": 0.0002217253176930596,
      "loss": 0.4603,
      "step": 3713
    },
    {
      "epoch": 0.5814026299311209,
      "grad_norm": 1.477260708808899,
      "learning_rate": 0.0002217008797653959,
      "loss": 0.6937,
      "step": 3714
    },
    {
      "epoch": 0.5815591734502191,
      "grad_norm": 1.7115600109100342,
      "learning_rate": 0.00022167644183773214,
      "loss": 0.6814,
      "step": 3715
    },
    {
      "epoch": 0.5817157169693175,
      "grad_norm": 1.2147424221038818,
      "learning_rate": 0.0002216520039100684,
      "loss": 0.5137,
      "step": 3716
    },
    {
      "epoch": 0.5818722604884158,
      "grad_norm": 1.544769287109375,
      "learning_rate": 0.00022162756598240467,
      "loss": 0.4772,
      "step": 3717
    },
    {
      "epoch": 0.5820288040075141,
      "grad_norm": 1.2717753648757935,
      "learning_rate": 0.00022160312805474095,
      "loss": 0.7419,
      "step": 3718
    },
    {
      "epoch": 0.5821853475266124,
      "grad_norm": 1.403295636177063,
      "learning_rate": 0.0002215786901270772,
      "loss": 0.8306,
      "step": 3719
    },
    {
      "epoch": 0.5823418910457107,
      "grad_norm": 2.2158901691436768,
      "learning_rate": 0.00022155425219941348,
      "loss": 0.8163,
      "step": 3720
    },
    {
      "epoch": 0.582498434564809,
      "grad_norm": 1.945855975151062,
      "learning_rate": 0.00022152981427174976,
      "loss": 0.7549,
      "step": 3721
    },
    {
      "epoch": 0.5826549780839073,
      "grad_norm": 2.006751775741577,
      "learning_rate": 0.00022150537634408598,
      "loss": 0.6222,
      "step": 3722
    },
    {
      "epoch": 0.5828115216030056,
      "grad_norm": 1.3016244173049927,
      "learning_rate": 0.00022148093841642226,
      "loss": 0.5719,
      "step": 3723
    },
    {
      "epoch": 0.582968065122104,
      "grad_norm": 1.7161877155303955,
      "learning_rate": 0.00022145650048875854,
      "loss": 0.8886,
      "step": 3724
    },
    {
      "epoch": 0.5831246086412023,
      "grad_norm": 1.5674386024475098,
      "learning_rate": 0.0002214320625610948,
      "loss": 0.8454,
      "step": 3725
    },
    {
      "epoch": 0.5832811521603005,
      "grad_norm": 2.3730225563049316,
      "learning_rate": 0.00022140762463343107,
      "loss": 1.0747,
      "step": 3726
    },
    {
      "epoch": 0.5834376956793988,
      "grad_norm": 2.1554417610168457,
      "learning_rate": 0.00022138318670576735,
      "loss": 0.6091,
      "step": 3727
    },
    {
      "epoch": 0.5835942391984972,
      "grad_norm": 1.7122255563735962,
      "learning_rate": 0.0002213587487781036,
      "loss": 0.4839,
      "step": 3728
    },
    {
      "epoch": 0.5837507827175955,
      "grad_norm": 1.168192744255066,
      "learning_rate": 0.00022133431085043988,
      "loss": 0.7182,
      "step": 3729
    },
    {
      "epoch": 0.5839073262366938,
      "grad_norm": 1.9943832159042358,
      "learning_rate": 0.00022130987292277615,
      "loss": 1.1607,
      "step": 3730
    },
    {
      "epoch": 0.5840638697557922,
      "grad_norm": 4.930220603942871,
      "learning_rate": 0.00022128543499511238,
      "loss": 0.8025,
      "step": 3731
    },
    {
      "epoch": 0.5842204132748904,
      "grad_norm": 1.833145260810852,
      "learning_rate": 0.00022126099706744866,
      "loss": 0.7035,
      "step": 3732
    },
    {
      "epoch": 0.5843769567939887,
      "grad_norm": 2.07147479057312,
      "learning_rate": 0.00022123655913978494,
      "loss": 1.1647,
      "step": 3733
    },
    {
      "epoch": 0.584533500313087,
      "grad_norm": 2.026843309402466,
      "learning_rate": 0.0002212121212121212,
      "loss": 0.7464,
      "step": 3734
    },
    {
      "epoch": 0.5846900438321854,
      "grad_norm": 1.1421523094177246,
      "learning_rate": 0.00022118768328445747,
      "loss": 1.1607,
      "step": 3735
    },
    {
      "epoch": 0.5848465873512837,
      "grad_norm": 3.0231730937957764,
      "learning_rate": 0.00022116324535679374,
      "loss": 1.0897,
      "step": 3736
    },
    {
      "epoch": 0.5850031308703819,
      "grad_norm": 1.649326205253601,
      "learning_rate": 0.00022113880742913,
      "loss": 0.869,
      "step": 3737
    },
    {
      "epoch": 0.5851596743894802,
      "grad_norm": 3.129502296447754,
      "learning_rate": 0.00022111436950146625,
      "loss": 1.1095,
      "step": 3738
    },
    {
      "epoch": 0.5853162179085786,
      "grad_norm": 2.2908878326416016,
      "learning_rate": 0.00022108993157380252,
      "loss": 0.7454,
      "step": 3739
    },
    {
      "epoch": 0.5854727614276769,
      "grad_norm": 2.386479377746582,
      "learning_rate": 0.00022106549364613878,
      "loss": 1.3902,
      "step": 3740
    },
    {
      "epoch": 0.5856293049467752,
      "grad_norm": 2.256403923034668,
      "learning_rate": 0.00022104105571847505,
      "loss": 1.2762,
      "step": 3741
    },
    {
      "epoch": 0.5857858484658736,
      "grad_norm": 3.2702090740203857,
      "learning_rate": 0.00022101661779081133,
      "loss": 1.1909,
      "step": 3742
    },
    {
      "epoch": 0.5859423919849718,
      "grad_norm": 1.4588786363601685,
      "learning_rate": 0.00022099217986314758,
      "loss": 1.6311,
      "step": 3743
    },
    {
      "epoch": 0.5860989355040701,
      "grad_norm": 1.102078914642334,
      "learning_rate": 0.00022096774193548386,
      "loss": 0.6543,
      "step": 3744
    },
    {
      "epoch": 0.5862554790231684,
      "grad_norm": 2.0560879707336426,
      "learning_rate": 0.00022094330400782014,
      "loss": 1.4536,
      "step": 3745
    },
    {
      "epoch": 0.5864120225422668,
      "grad_norm": 1.8703267574310303,
      "learning_rate": 0.00022091886608015636,
      "loss": 1.5638,
      "step": 3746
    },
    {
      "epoch": 0.5865685660613651,
      "grad_norm": 1.5793174505233765,
      "learning_rate": 0.00022089442815249264,
      "loss": 0.919,
      "step": 3747
    },
    {
      "epoch": 0.5867251095804634,
      "grad_norm": 3.379425287246704,
      "learning_rate": 0.00022086999022482892,
      "loss": 1.7933,
      "step": 3748
    },
    {
      "epoch": 0.5868816530995616,
      "grad_norm": 2.630582571029663,
      "learning_rate": 0.00022084555229716517,
      "loss": 0.6506,
      "step": 3749
    },
    {
      "epoch": 0.58703819661866,
      "grad_norm": 1.3981788158416748,
      "learning_rate": 0.00022082111436950145,
      "loss": 1.0893,
      "step": 3750
    },
    {
      "epoch": 0.5871947401377583,
      "grad_norm": 0.4805038273334503,
      "learning_rate": 0.00022079667644183773,
      "loss": 0.3598,
      "step": 3751
    },
    {
      "epoch": 0.5873512836568566,
      "grad_norm": 0.3983316719532013,
      "learning_rate": 0.00022077223851417398,
      "loss": 0.3031,
      "step": 3752
    },
    {
      "epoch": 0.587507827175955,
      "grad_norm": 1.9718865156173706,
      "learning_rate": 0.00022074780058651026,
      "loss": 1.3055,
      "step": 3753
    },
    {
      "epoch": 0.5876643706950532,
      "grad_norm": 0.5567598938941956,
      "learning_rate": 0.00022072336265884654,
      "loss": 0.3184,
      "step": 3754
    },
    {
      "epoch": 0.5878209142141515,
      "grad_norm": 0.6025465130805969,
      "learning_rate": 0.00022069892473118276,
      "loss": 0.3816,
      "step": 3755
    },
    {
      "epoch": 0.5879774577332498,
      "grad_norm": 0.9688357710838318,
      "learning_rate": 0.00022067448680351904,
      "loss": 0.4014,
      "step": 3756
    },
    {
      "epoch": 0.5881340012523482,
      "grad_norm": 0.6536909341812134,
      "learning_rate": 0.00022065004887585532,
      "loss": 0.4133,
      "step": 3757
    },
    {
      "epoch": 0.5882905447714465,
      "grad_norm": 0.6324391961097717,
      "learning_rate": 0.00022062561094819157,
      "loss": 0.3517,
      "step": 3758
    },
    {
      "epoch": 0.5884470882905448,
      "grad_norm": 0.7064645886421204,
      "learning_rate": 0.00022060117302052785,
      "loss": 0.404,
      "step": 3759
    },
    {
      "epoch": 0.588603631809643,
      "grad_norm": 0.8982453942298889,
      "learning_rate": 0.00022057673509286413,
      "loss": 0.517,
      "step": 3760
    },
    {
      "epoch": 0.5887601753287414,
      "grad_norm": 1.20693838596344,
      "learning_rate": 0.00022055229716520038,
      "loss": 0.4526,
      "step": 3761
    },
    {
      "epoch": 0.5889167188478397,
      "grad_norm": 0.7581743597984314,
      "learning_rate": 0.00022052785923753663,
      "loss": 0.3104,
      "step": 3762
    },
    {
      "epoch": 0.589073262366938,
      "grad_norm": 0.5927846431732178,
      "learning_rate": 0.0002205034213098729,
      "loss": 0.2852,
      "step": 3763
    },
    {
      "epoch": 0.5892298058860364,
      "grad_norm": 1.2253514528274536,
      "learning_rate": 0.00022047898338220916,
      "loss": 0.3494,
      "step": 3764
    },
    {
      "epoch": 0.5893863494051347,
      "grad_norm": 0.9083871245384216,
      "learning_rate": 0.00022045454545454544,
      "loss": 0.3888,
      "step": 3765
    },
    {
      "epoch": 0.5895428929242329,
      "grad_norm": 0.8482705950737,
      "learning_rate": 0.00022043010752688171,
      "loss": 0.487,
      "step": 3766
    },
    {
      "epoch": 0.5896994364433312,
      "grad_norm": 0.903735876083374,
      "learning_rate": 0.00022040566959921797,
      "loss": 0.4058,
      "step": 3767
    },
    {
      "epoch": 0.5898559799624296,
      "grad_norm": 1.6391104459762573,
      "learning_rate": 0.00022038123167155424,
      "loss": 0.7215,
      "step": 3768
    },
    {
      "epoch": 0.5900125234815279,
      "grad_norm": 1.211521863937378,
      "learning_rate": 0.00022035679374389052,
      "loss": 0.3585,
      "step": 3769
    },
    {
      "epoch": 0.5901690670006262,
      "grad_norm": 1.2085306644439697,
      "learning_rate": 0.00022033235581622675,
      "loss": 0.5602,
      "step": 3770
    },
    {
      "epoch": 0.5903256105197244,
      "grad_norm": 1.251050591468811,
      "learning_rate": 0.00022030791788856303,
      "loss": 0.6018,
      "step": 3771
    },
    {
      "epoch": 0.5904821540388228,
      "grad_norm": 3.4225316047668457,
      "learning_rate": 0.0002202834799608993,
      "loss": 1.5824,
      "step": 3772
    },
    {
      "epoch": 0.5906386975579211,
      "grad_norm": 1.3945788145065308,
      "learning_rate": 0.00022025904203323555,
      "loss": 0.6043,
      "step": 3773
    },
    {
      "epoch": 0.5907952410770194,
      "grad_norm": 2.437675952911377,
      "learning_rate": 0.00022023460410557183,
      "loss": 0.7169,
      "step": 3774
    },
    {
      "epoch": 0.5909517845961177,
      "grad_norm": 1.3050737380981445,
      "learning_rate": 0.0002202101661779081,
      "loss": 0.7835,
      "step": 3775
    },
    {
      "epoch": 0.5911083281152161,
      "grad_norm": 2.816662549972534,
      "learning_rate": 0.00022018572825024436,
      "loss": 0.6321,
      "step": 3776
    },
    {
      "epoch": 0.5912648716343143,
      "grad_norm": 1.4117590188980103,
      "learning_rate": 0.00022016129032258064,
      "loss": 0.479,
      "step": 3777
    },
    {
      "epoch": 0.5914214151534126,
      "grad_norm": 1.9365350008010864,
      "learning_rate": 0.00022013685239491692,
      "loss": 0.8931,
      "step": 3778
    },
    {
      "epoch": 0.591577958672511,
      "grad_norm": 1.9675894975662231,
      "learning_rate": 0.00022011241446725314,
      "loss": 0.6348,
      "step": 3779
    },
    {
      "epoch": 0.5917345021916093,
      "grad_norm": 1.806617259979248,
      "learning_rate": 0.00022008797653958942,
      "loss": 0.7128,
      "step": 3780
    },
    {
      "epoch": 0.5918910457107076,
      "grad_norm": 2.15679669380188,
      "learning_rate": 0.0002200635386119257,
      "loss": 0.7107,
      "step": 3781
    },
    {
      "epoch": 0.5920475892298059,
      "grad_norm": 1.5504449605941772,
      "learning_rate": 0.00022003910068426195,
      "loss": 0.6923,
      "step": 3782
    },
    {
      "epoch": 0.5922041327489042,
      "grad_norm": 2.190402030944824,
      "learning_rate": 0.00022001466275659823,
      "loss": 0.9243,
      "step": 3783
    },
    {
      "epoch": 0.5923606762680025,
      "grad_norm": 2.572727918624878,
      "learning_rate": 0.0002199902248289345,
      "loss": 1.3498,
      "step": 3784
    },
    {
      "epoch": 0.5925172197871008,
      "grad_norm": 2.3958308696746826,
      "learning_rate": 0.00021996578690127076,
      "loss": 1.3511,
      "step": 3785
    },
    {
      "epoch": 0.5926737633061991,
      "grad_norm": 3.290600538253784,
      "learning_rate": 0.000219941348973607,
      "loss": 1.1582,
      "step": 3786
    },
    {
      "epoch": 0.5928303068252975,
      "grad_norm": 2.0067548751831055,
      "learning_rate": 0.0002199169110459433,
      "loss": 0.6797,
      "step": 3787
    },
    {
      "epoch": 0.5929868503443957,
      "grad_norm": 2.599518060684204,
      "learning_rate": 0.00021989247311827954,
      "loss": 1.1569,
      "step": 3788
    },
    {
      "epoch": 0.593143393863494,
      "grad_norm": 3.275362968444824,
      "learning_rate": 0.00021986803519061582,
      "loss": 1.0037,
      "step": 3789
    },
    {
      "epoch": 0.5932999373825923,
      "grad_norm": 1.9998888969421387,
      "learning_rate": 0.0002198435972629521,
      "loss": 1.6154,
      "step": 3790
    },
    {
      "epoch": 0.5934564809016907,
      "grad_norm": 2.5467236042022705,
      "learning_rate": 0.00021981915933528835,
      "loss": 1.5551,
      "step": 3791
    },
    {
      "epoch": 0.593613024420789,
      "grad_norm": 5.1116042137146,
      "learning_rate": 0.00021979472140762463,
      "loss": 1.3565,
      "step": 3792
    },
    {
      "epoch": 0.5937695679398873,
      "grad_norm": 1.5216659307479858,
      "learning_rate": 0.0002197702834799609,
      "loss": 1.0188,
      "step": 3793
    },
    {
      "epoch": 0.5939261114589856,
      "grad_norm": 2.700352191925049,
      "learning_rate": 0.00021974584555229713,
      "loss": 1.9111,
      "step": 3794
    },
    {
      "epoch": 0.5940826549780839,
      "grad_norm": 2.896486759185791,
      "learning_rate": 0.0002197214076246334,
      "loss": 1.1349,
      "step": 3795
    },
    {
      "epoch": 0.5942391984971822,
      "grad_norm": 2.902263879776001,
      "learning_rate": 0.00021969696969696969,
      "loss": 0.7021,
      "step": 3796
    },
    {
      "epoch": 0.5943957420162805,
      "grad_norm": 1.9411492347717285,
      "learning_rate": 0.00021967253176930594,
      "loss": 0.9319,
      "step": 3797
    },
    {
      "epoch": 0.5945522855353789,
      "grad_norm": 1.5800654888153076,
      "learning_rate": 0.00021964809384164222,
      "loss": 0.6419,
      "step": 3798
    },
    {
      "epoch": 0.5947088290544772,
      "grad_norm": 2.5261833667755127,
      "learning_rate": 0.0002196236559139785,
      "loss": 1.0969,
      "step": 3799
    },
    {
      "epoch": 0.5948653725735754,
      "grad_norm": 3.4451301097869873,
      "learning_rate": 0.00021959921798631475,
      "loss": 1.2766,
      "step": 3800
    },
    {
      "epoch": 0.5950219160926737,
      "grad_norm": 0.6500614285469055,
      "learning_rate": 0.00021957478005865102,
      "loss": 0.461,
      "step": 3801
    },
    {
      "epoch": 0.5951784596117721,
      "grad_norm": 0.5970913767814636,
      "learning_rate": 0.0002195503421309873,
      "loss": 0.4256,
      "step": 3802
    },
    {
      "epoch": 0.5953350031308704,
      "grad_norm": 0.7132553458213806,
      "learning_rate": 0.00021952590420332353,
      "loss": 0.3076,
      "step": 3803
    },
    {
      "epoch": 0.5954915466499687,
      "grad_norm": 0.6507532000541687,
      "learning_rate": 0.0002195014662756598,
      "loss": 0.3146,
      "step": 3804
    },
    {
      "epoch": 0.595648090169067,
      "grad_norm": 0.8440534472465515,
      "learning_rate": 0.00021947702834799608,
      "loss": 0.3291,
      "step": 3805
    },
    {
      "epoch": 0.5958046336881653,
      "grad_norm": 0.7817156314849854,
      "learning_rate": 0.00021945259042033233,
      "loss": 0.2867,
      "step": 3806
    },
    {
      "epoch": 0.5959611772072636,
      "grad_norm": 0.5882556438446045,
      "learning_rate": 0.0002194281524926686,
      "loss": 0.3247,
      "step": 3807
    },
    {
      "epoch": 0.5961177207263619,
      "grad_norm": 2.025453805923462,
      "learning_rate": 0.0002194037145650049,
      "loss": 0.4633,
      "step": 3808
    },
    {
      "epoch": 0.5962742642454603,
      "grad_norm": 0.8439894318580627,
      "learning_rate": 0.00021937927663734111,
      "loss": 0.4273,
      "step": 3809
    },
    {
      "epoch": 0.5964308077645586,
      "grad_norm": 4.892188549041748,
      "learning_rate": 0.0002193548387096774,
      "loss": 0.6922,
      "step": 3810
    },
    {
      "epoch": 0.5965873512836568,
      "grad_norm": 1.2125036716461182,
      "learning_rate": 0.00021933040078201367,
      "loss": 0.4494,
      "step": 3811
    },
    {
      "epoch": 0.5967438948027551,
      "grad_norm": 1.2884728908538818,
      "learning_rate": 0.00021930596285434992,
      "loss": 0.4823,
      "step": 3812
    },
    {
      "epoch": 0.5969004383218535,
      "grad_norm": 1.1847399473190308,
      "learning_rate": 0.0002192815249266862,
      "loss": 0.5722,
      "step": 3813
    },
    {
      "epoch": 0.5970569818409518,
      "grad_norm": 0.9156528115272522,
      "learning_rate": 0.00021925708699902248,
      "loss": 0.3829,
      "step": 3814
    },
    {
      "epoch": 0.5972135253600501,
      "grad_norm": 1.0122822523117065,
      "learning_rate": 0.00021923264907135873,
      "loss": 0.364,
      "step": 3815
    },
    {
      "epoch": 0.5973700688791485,
      "grad_norm": 0.5517539978027344,
      "learning_rate": 0.000219208211143695,
      "loss": 0.2851,
      "step": 3816
    },
    {
      "epoch": 0.5975266123982467,
      "grad_norm": 0.777860164642334,
      "learning_rate": 0.0002191837732160313,
      "loss": 0.363,
      "step": 3817
    },
    {
      "epoch": 0.597683155917345,
      "grad_norm": 3.110654354095459,
      "learning_rate": 0.0002191593352883675,
      "loss": 0.9008,
      "step": 3818
    },
    {
      "epoch": 0.5978396994364433,
      "grad_norm": 1.7335413694381714,
      "learning_rate": 0.0002191348973607038,
      "loss": 0.5437,
      "step": 3819
    },
    {
      "epoch": 0.5979962429555417,
      "grad_norm": 1.2909818887710571,
      "learning_rate": 0.00021911045943304007,
      "loss": 0.3403,
      "step": 3820
    },
    {
      "epoch": 0.59815278647464,
      "grad_norm": 2.242083787918091,
      "learning_rate": 0.00021908602150537632,
      "loss": 0.6593,
      "step": 3821
    },
    {
      "epoch": 0.5983093299937383,
      "grad_norm": 1.45481538772583,
      "learning_rate": 0.0002190615835777126,
      "loss": 0.64,
      "step": 3822
    },
    {
      "epoch": 0.5984658735128365,
      "grad_norm": 1.841384768486023,
      "learning_rate": 0.00021903714565004888,
      "loss": 0.8033,
      "step": 3823
    },
    {
      "epoch": 0.5986224170319349,
      "grad_norm": 4.309696197509766,
      "learning_rate": 0.00021901270772238513,
      "loss": 0.9931,
      "step": 3824
    },
    {
      "epoch": 0.5987789605510332,
      "grad_norm": 2.3548316955566406,
      "learning_rate": 0.0002189882697947214,
      "loss": 0.7233,
      "step": 3825
    },
    {
      "epoch": 0.5989355040701315,
      "grad_norm": 2.9939208030700684,
      "learning_rate": 0.00021896383186705768,
      "loss": 1.0121,
      "step": 3826
    },
    {
      "epoch": 0.5990920475892298,
      "grad_norm": 1.7699570655822754,
      "learning_rate": 0.0002189393939393939,
      "loss": 0.6097,
      "step": 3827
    },
    {
      "epoch": 0.5992485911083281,
      "grad_norm": 2.700235605239868,
      "learning_rate": 0.0002189149560117302,
      "loss": 0.8177,
      "step": 3828
    },
    {
      "epoch": 0.5994051346274264,
      "grad_norm": 1.7766852378845215,
      "learning_rate": 0.00021889051808406647,
      "loss": 0.6307,
      "step": 3829
    },
    {
      "epoch": 0.5995616781465247,
      "grad_norm": 1.4954442977905273,
      "learning_rate": 0.00021886608015640272,
      "loss": 0.9373,
      "step": 3830
    },
    {
      "epoch": 0.599718221665623,
      "grad_norm": 1.9515974521636963,
      "learning_rate": 0.000218841642228739,
      "loss": 0.627,
      "step": 3831
    },
    {
      "epoch": 0.5998747651847214,
      "grad_norm": 1.922837734222412,
      "learning_rate": 0.00021881720430107527,
      "loss": 0.8141,
      "step": 3832
    },
    {
      "epoch": 0.6000313087038197,
      "grad_norm": 2.2533392906188965,
      "learning_rate": 0.0002187927663734115,
      "loss": 1.1225,
      "step": 3833
    },
    {
      "epoch": 0.6001878522229179,
      "grad_norm": 1.5181680917739868,
      "learning_rate": 0.00021876832844574778,
      "loss": 0.8138,
      "step": 3834
    },
    {
      "epoch": 0.6003443957420163,
      "grad_norm": 2.555852174758911,
      "learning_rate": 0.00021874389051808403,
      "loss": 1.0635,
      "step": 3835
    },
    {
      "epoch": 0.6005009392611146,
      "grad_norm": 1.9086858034133911,
      "learning_rate": 0.0002187194525904203,
      "loss": 1.2145,
      "step": 3836
    },
    {
      "epoch": 0.6006574827802129,
      "grad_norm": 2.0855298042297363,
      "learning_rate": 0.00021869501466275658,
      "loss": 0.4878,
      "step": 3837
    },
    {
      "epoch": 0.6008140262993112,
      "grad_norm": 2.812436819076538,
      "learning_rate": 0.00021867057673509283,
      "loss": 1.3879,
      "step": 3838
    },
    {
      "epoch": 0.6009705698184096,
      "grad_norm": 2.317641258239746,
      "learning_rate": 0.0002186461388074291,
      "loss": 1.0922,
      "step": 3839
    },
    {
      "epoch": 0.6011271133375078,
      "grad_norm": 1.9752837419509888,
      "learning_rate": 0.0002186217008797654,
      "loss": 1.0511,
      "step": 3840
    },
    {
      "epoch": 0.6012836568566061,
      "grad_norm": 3.892122745513916,
      "learning_rate": 0.00021859726295210162,
      "loss": 1.2536,
      "step": 3841
    },
    {
      "epoch": 0.6014402003757044,
      "grad_norm": 3.568765640258789,
      "learning_rate": 0.0002185728250244379,
      "loss": 1.7708,
      "step": 3842
    },
    {
      "epoch": 0.6015967438948028,
      "grad_norm": 4.544713497161865,
      "learning_rate": 0.00021854838709677417,
      "loss": 1.8441,
      "step": 3843
    },
    {
      "epoch": 0.6017532874139011,
      "grad_norm": 3.862272024154663,
      "learning_rate": 0.00021852394916911042,
      "loss": 1.3045,
      "step": 3844
    },
    {
      "epoch": 0.6019098309329993,
      "grad_norm": 2.4139466285705566,
      "learning_rate": 0.0002184995112414467,
      "loss": 0.8241,
      "step": 3845
    },
    {
      "epoch": 0.6020663744520977,
      "grad_norm": 2.8001489639282227,
      "learning_rate": 0.00021847507331378298,
      "loss": 1.4736,
      "step": 3846
    },
    {
      "epoch": 0.602222917971196,
      "grad_norm": 1.7694708108901978,
      "learning_rate": 0.00021845063538611923,
      "loss": 0.8626,
      "step": 3847
    },
    {
      "epoch": 0.6023794614902943,
      "grad_norm": 1.846111536026001,
      "learning_rate": 0.0002184261974584555,
      "loss": 0.8587,
      "step": 3848
    },
    {
      "epoch": 0.6025360050093926,
      "grad_norm": 2.2559170722961426,
      "learning_rate": 0.0002184017595307918,
      "loss": 0.8416,
      "step": 3849
    },
    {
      "epoch": 0.602692548528491,
      "grad_norm": 3.0026137828826904,
      "learning_rate": 0.000218377321603128,
      "loss": 0.8497,
      "step": 3850
    },
    {
      "epoch": 0.6028490920475892,
      "grad_norm": 2.2647368907928467,
      "learning_rate": 0.0002183528836754643,
      "loss": 0.4743,
      "step": 3851
    },
    {
      "epoch": 0.6030056355666875,
      "grad_norm": 0.49257951974868774,
      "learning_rate": 0.00021832844574780057,
      "loss": 0.3079,
      "step": 3852
    },
    {
      "epoch": 0.6031621790857858,
      "grad_norm": 0.505824089050293,
      "learning_rate": 0.00021830400782013682,
      "loss": 0.4189,
      "step": 3853
    },
    {
      "epoch": 0.6033187226048842,
      "grad_norm": 0.44036567211151123,
      "learning_rate": 0.0002182795698924731,
      "loss": 0.2503,
      "step": 3854
    },
    {
      "epoch": 0.6034752661239825,
      "grad_norm": 0.5498443245887756,
      "learning_rate": 0.00021825513196480938,
      "loss": 0.2296,
      "step": 3855
    },
    {
      "epoch": 0.6036318096430808,
      "grad_norm": 0.8247083425521851,
      "learning_rate": 0.00021823069403714563,
      "loss": 0.4274,
      "step": 3856
    },
    {
      "epoch": 0.603788353162179,
      "grad_norm": 0.6773415207862854,
      "learning_rate": 0.00021820625610948188,
      "loss": 0.2763,
      "step": 3857
    },
    {
      "epoch": 0.6039448966812774,
      "grad_norm": 0.810204803943634,
      "learning_rate": 0.00021818181818181816,
      "loss": 0.4864,
      "step": 3858
    },
    {
      "epoch": 0.6041014402003757,
      "grad_norm": 0.6132627129554749,
      "learning_rate": 0.0002181573802541544,
      "loss": 0.3392,
      "step": 3859
    },
    {
      "epoch": 0.604257983719474,
      "grad_norm": 0.5890900492668152,
      "learning_rate": 0.0002181329423264907,
      "loss": 0.3409,
      "step": 3860
    },
    {
      "epoch": 0.6044145272385724,
      "grad_norm": 1.143919587135315,
      "learning_rate": 0.00021810850439882697,
      "loss": 0.5649,
      "step": 3861
    },
    {
      "epoch": 0.6045710707576706,
      "grad_norm": 0.9753417372703552,
      "learning_rate": 0.00021808406647116322,
      "loss": 0.354,
      "step": 3862
    },
    {
      "epoch": 0.6047276142767689,
      "grad_norm": 0.9301658272743225,
      "learning_rate": 0.0002180596285434995,
      "loss": 0.2972,
      "step": 3863
    },
    {
      "epoch": 0.6048841577958672,
      "grad_norm": 0.7712864279747009,
      "learning_rate": 0.00021803519061583577,
      "loss": 0.3733,
      "step": 3864
    },
    {
      "epoch": 0.6050407013149656,
      "grad_norm": 1.1033275127410889,
      "learning_rate": 0.000218010752688172,
      "loss": 0.5233,
      "step": 3865
    },
    {
      "epoch": 0.6051972448340639,
      "grad_norm": 1.8046352863311768,
      "learning_rate": 0.00021798631476050828,
      "loss": 0.6654,
      "step": 3866
    },
    {
      "epoch": 0.6053537883531622,
      "grad_norm": 1.1659013032913208,
      "learning_rate": 0.00021796187683284455,
      "loss": 0.4882,
      "step": 3867
    },
    {
      "epoch": 0.6055103318722604,
      "grad_norm": 2.8485543727874756,
      "learning_rate": 0.0002179374389051808,
      "loss": 0.7751,
      "step": 3868
    },
    {
      "epoch": 0.6056668753913588,
      "grad_norm": 2.1419053077697754,
      "learning_rate": 0.00021791300097751708,
      "loss": 0.7052,
      "step": 3869
    },
    {
      "epoch": 0.6058234189104571,
      "grad_norm": 1.2547987699508667,
      "learning_rate": 0.00021788856304985336,
      "loss": 0.4797,
      "step": 3870
    },
    {
      "epoch": 0.6059799624295554,
      "grad_norm": 1.7469266653060913,
      "learning_rate": 0.00021786412512218961,
      "loss": 0.5729,
      "step": 3871
    },
    {
      "epoch": 0.6061365059486538,
      "grad_norm": 1.2163182497024536,
      "learning_rate": 0.0002178396871945259,
      "loss": 0.63,
      "step": 3872
    },
    {
      "epoch": 0.6062930494677521,
      "grad_norm": 1.283544898033142,
      "learning_rate": 0.00021781524926686217,
      "loss": 0.723,
      "step": 3873
    },
    {
      "epoch": 0.6064495929868503,
      "grad_norm": 1.9158495664596558,
      "learning_rate": 0.0002177908113391984,
      "loss": 0.7387,
      "step": 3874
    },
    {
      "epoch": 0.6066061365059486,
      "grad_norm": 2.14650821685791,
      "learning_rate": 0.00021776637341153467,
      "loss": 0.8682,
      "step": 3875
    },
    {
      "epoch": 0.606762680025047,
      "grad_norm": 1.1081241369247437,
      "learning_rate": 0.00021774193548387095,
      "loss": 0.5418,
      "step": 3876
    },
    {
      "epoch": 0.6069192235441453,
      "grad_norm": 1.6972672939300537,
      "learning_rate": 0.0002177174975562072,
      "loss": 0.9634,
      "step": 3877
    },
    {
      "epoch": 0.6070757670632436,
      "grad_norm": 2.2414941787719727,
      "learning_rate": 0.00021769305962854348,
      "loss": 0.9434,
      "step": 3878
    },
    {
      "epoch": 0.6072323105823418,
      "grad_norm": 1.9545202255249023,
      "learning_rate": 0.00021766862170087976,
      "loss": 0.5801,
      "step": 3879
    },
    {
      "epoch": 0.6073888541014402,
      "grad_norm": 2.0093443393707275,
      "learning_rate": 0.00021764418377321598,
      "loss": 0.9481,
      "step": 3880
    },
    {
      "epoch": 0.6075453976205385,
      "grad_norm": 1.2470927238464355,
      "learning_rate": 0.00021761974584555226,
      "loss": 0.3291,
      "step": 3881
    },
    {
      "epoch": 0.6077019411396368,
      "grad_norm": 3.9128541946411133,
      "learning_rate": 0.00021759530791788854,
      "loss": 0.4683,
      "step": 3882
    },
    {
      "epoch": 0.6078584846587352,
      "grad_norm": 1.703982949256897,
      "learning_rate": 0.0002175708699902248,
      "loss": 0.9931,
      "step": 3883
    },
    {
      "epoch": 0.6080150281778335,
      "grad_norm": 2.462613582611084,
      "learning_rate": 0.00021754643206256107,
      "loss": 1.0189,
      "step": 3884
    },
    {
      "epoch": 0.6081715716969317,
      "grad_norm": 5.5664591789245605,
      "learning_rate": 0.00021752199413489735,
      "loss": 0.7844,
      "step": 3885
    },
    {
      "epoch": 0.60832811521603,
      "grad_norm": 2.5077731609344482,
      "learning_rate": 0.0002174975562072336,
      "loss": 1.414,
      "step": 3886
    },
    {
      "epoch": 0.6084846587351284,
      "grad_norm": 2.240034341812134,
      "learning_rate": 0.00021747311827956988,
      "loss": 0.8179,
      "step": 3887
    },
    {
      "epoch": 0.6086412022542267,
      "grad_norm": 1.7230664491653442,
      "learning_rate": 0.00021744868035190616,
      "loss": 1.0263,
      "step": 3888
    },
    {
      "epoch": 0.608797745773325,
      "grad_norm": 3.0299770832061768,
      "learning_rate": 0.00021742424242424238,
      "loss": 1.0993,
      "step": 3889
    },
    {
      "epoch": 0.6089542892924233,
      "grad_norm": 4.0090413093566895,
      "learning_rate": 0.00021739980449657866,
      "loss": 1.0955,
      "step": 3890
    },
    {
      "epoch": 0.6091108328115216,
      "grad_norm": 3.151949644088745,
      "learning_rate": 0.00021737536656891494,
      "loss": 1.6831,
      "step": 3891
    },
    {
      "epoch": 0.6092673763306199,
      "grad_norm": 1.1991798877716064,
      "learning_rate": 0.0002173509286412512,
      "loss": 1.0109,
      "step": 3892
    },
    {
      "epoch": 0.6094239198497182,
      "grad_norm": 3.1792802810668945,
      "learning_rate": 0.00021732649071358747,
      "loss": 1.4323,
      "step": 3893
    },
    {
      "epoch": 0.6095804633688165,
      "grad_norm": 1.9545060396194458,
      "learning_rate": 0.00021730205278592374,
      "loss": 1.1116,
      "step": 3894
    },
    {
      "epoch": 0.6097370068879149,
      "grad_norm": 2.42977237701416,
      "learning_rate": 0.00021727761485826,
      "loss": 1.1406,
      "step": 3895
    },
    {
      "epoch": 0.6098935504070131,
      "grad_norm": 1.8895028829574585,
      "learning_rate": 0.00021725317693059627,
      "loss": 0.7086,
      "step": 3896
    },
    {
      "epoch": 0.6100500939261114,
      "grad_norm": 4.542184829711914,
      "learning_rate": 0.00021722873900293253,
      "loss": 1.2922,
      "step": 3897
    },
    {
      "epoch": 0.6102066374452098,
      "grad_norm": 1.5227937698364258,
      "learning_rate": 0.00021720430107526878,
      "loss": 0.777,
      "step": 3898
    },
    {
      "epoch": 0.6103631809643081,
      "grad_norm": 2.370554208755493,
      "learning_rate": 0.00021717986314760506,
      "loss": 0.5624,
      "step": 3899
    },
    {
      "epoch": 0.6105197244834064,
      "grad_norm": 1.582291603088379,
      "learning_rate": 0.00021715542521994133,
      "loss": 0.91,
      "step": 3900
    },
    {
      "epoch": 0.6106762680025047,
      "grad_norm": 0.5427595973014832,
      "learning_rate": 0.00021713098729227758,
      "loss": 0.4112,
      "step": 3901
    },
    {
      "epoch": 0.610832811521603,
      "grad_norm": 0.9115118980407715,
      "learning_rate": 0.00021710654936461386,
      "loss": 0.3359,
      "step": 3902
    },
    {
      "epoch": 0.6109893550407013,
      "grad_norm": 0.6298912763595581,
      "learning_rate": 0.00021708211143695014,
      "loss": 0.3834,
      "step": 3903
    },
    {
      "epoch": 0.6111458985597996,
      "grad_norm": 0.669316828250885,
      "learning_rate": 0.00021705767350928637,
      "loss": 0.5009,
      "step": 3904
    },
    {
      "epoch": 0.611302442078898,
      "grad_norm": 0.608873188495636,
      "learning_rate": 0.00021703323558162264,
      "loss": 0.2752,
      "step": 3905
    },
    {
      "epoch": 0.6114589855979963,
      "grad_norm": 0.4760167598724365,
      "learning_rate": 0.00021700879765395892,
      "loss": 0.3135,
      "step": 3906
    },
    {
      "epoch": 0.6116155291170946,
      "grad_norm": 0.5999681353569031,
      "learning_rate": 0.00021698435972629517,
      "loss": 0.4159,
      "step": 3907
    },
    {
      "epoch": 0.6117720726361928,
      "grad_norm": 0.8838676810264587,
      "learning_rate": 0.00021695992179863145,
      "loss": 0.5722,
      "step": 3908
    },
    {
      "epoch": 0.6119286161552911,
      "grad_norm": 0.8558535575866699,
      "learning_rate": 0.00021693548387096773,
      "loss": 0.5293,
      "step": 3909
    },
    {
      "epoch": 0.6120851596743895,
      "grad_norm": 0.6933575868606567,
      "learning_rate": 0.00021691104594330398,
      "loss": 0.3242,
      "step": 3910
    },
    {
      "epoch": 0.6122417031934878,
      "grad_norm": 0.6868442893028259,
      "learning_rate": 0.00021688660801564026,
      "loss": 0.288,
      "step": 3911
    },
    {
      "epoch": 0.6123982467125861,
      "grad_norm": 0.9867046475410461,
      "learning_rate": 0.00021686217008797654,
      "loss": 0.49,
      "step": 3912
    },
    {
      "epoch": 0.6125547902316844,
      "grad_norm": 0.8089808225631714,
      "learning_rate": 0.00021683773216031276,
      "loss": 0.4821,
      "step": 3913
    },
    {
      "epoch": 0.6127113337507827,
      "grad_norm": 1.2595003843307495,
      "learning_rate": 0.00021681329423264904,
      "loss": 0.5644,
      "step": 3914
    },
    {
      "epoch": 0.612867877269881,
      "grad_norm": 1.136533260345459,
      "learning_rate": 0.00021678885630498532,
      "loss": 0.3876,
      "step": 3915
    },
    {
      "epoch": 0.6130244207889793,
      "grad_norm": 0.9469622373580933,
      "learning_rate": 0.00021676441837732157,
      "loss": 0.3581,
      "step": 3916
    },
    {
      "epoch": 0.6131809643080777,
      "grad_norm": 1.390121340751648,
      "learning_rate": 0.00021673998044965785,
      "loss": 0.9731,
      "step": 3917
    },
    {
      "epoch": 0.613337507827176,
      "grad_norm": 0.9338005185127258,
      "learning_rate": 0.00021671554252199413,
      "loss": 0.5445,
      "step": 3918
    },
    {
      "epoch": 0.6134940513462742,
      "grad_norm": 1.3326369524002075,
      "learning_rate": 0.00021669110459433038,
      "loss": 0.5307,
      "step": 3919
    },
    {
      "epoch": 0.6136505948653725,
      "grad_norm": 1.399505615234375,
      "learning_rate": 0.00021666666666666666,
      "loss": 0.7355,
      "step": 3920
    },
    {
      "epoch": 0.6138071383844709,
      "grad_norm": 0.6472836136817932,
      "learning_rate": 0.0002166422287390029,
      "loss": 0.3582,
      "step": 3921
    },
    {
      "epoch": 0.6139636819035692,
      "grad_norm": 1.1896679401397705,
      "learning_rate": 0.00021661779081133916,
      "loss": 0.5519,
      "step": 3922
    },
    {
      "epoch": 0.6141202254226675,
      "grad_norm": 1.0599162578582764,
      "learning_rate": 0.00021659335288367544,
      "loss": 0.5286,
      "step": 3923
    },
    {
      "epoch": 0.6142767689417659,
      "grad_norm": 4.264036178588867,
      "learning_rate": 0.00021656891495601172,
      "loss": 0.5916,
      "step": 3924
    },
    {
      "epoch": 0.6144333124608641,
      "grad_norm": 5.218554973602295,
      "learning_rate": 0.00021654447702834797,
      "loss": 1.2145,
      "step": 3925
    },
    {
      "epoch": 0.6145898559799624,
      "grad_norm": 1.6409800052642822,
      "learning_rate": 0.00021652003910068425,
      "loss": 0.6093,
      "step": 3926
    },
    {
      "epoch": 0.6147463994990607,
      "grad_norm": 1.286957025527954,
      "learning_rate": 0.00021649560117302052,
      "loss": 0.7161,
      "step": 3927
    },
    {
      "epoch": 0.6149029430181591,
      "grad_norm": 2.0003950595855713,
      "learning_rate": 0.00021647116324535675,
      "loss": 0.7036,
      "step": 3928
    },
    {
      "epoch": 0.6150594865372574,
      "grad_norm": 1.561766505241394,
      "learning_rate": 0.00021644672531769303,
      "loss": 0.683,
      "step": 3929
    },
    {
      "epoch": 0.6152160300563556,
      "grad_norm": 1.2200226783752441,
      "learning_rate": 0.0002164222873900293,
      "loss": 0.5201,
      "step": 3930
    },
    {
      "epoch": 0.6153725735754539,
      "grad_norm": 3.1129469871520996,
      "learning_rate": 0.00021639784946236556,
      "loss": 0.7908,
      "step": 3931
    },
    {
      "epoch": 0.6155291170945523,
      "grad_norm": 2.578989028930664,
      "learning_rate": 0.00021637341153470183,
      "loss": 1.0388,
      "step": 3932
    },
    {
      "epoch": 0.6156856606136506,
      "grad_norm": 1.8467810153961182,
      "learning_rate": 0.0002163489736070381,
      "loss": 0.6982,
      "step": 3933
    },
    {
      "epoch": 0.6158422041327489,
      "grad_norm": 2.9458441734313965,
      "learning_rate": 0.00021632453567937436,
      "loss": 0.8299,
      "step": 3934
    },
    {
      "epoch": 0.6159987476518473,
      "grad_norm": 4.411853313446045,
      "learning_rate": 0.00021630009775171064,
      "loss": 0.9103,
      "step": 3935
    },
    {
      "epoch": 0.6161552911709455,
      "grad_norm": 1.8267143964767456,
      "learning_rate": 0.00021627565982404692,
      "loss": 0.9272,
      "step": 3936
    },
    {
      "epoch": 0.6163118346900438,
      "grad_norm": 5.815080165863037,
      "learning_rate": 0.00021625122189638314,
      "loss": 2.282,
      "step": 3937
    },
    {
      "epoch": 0.6164683782091421,
      "grad_norm": 3.6911022663116455,
      "learning_rate": 0.00021622678396871942,
      "loss": 1.254,
      "step": 3938
    },
    {
      "epoch": 0.6166249217282405,
      "grad_norm": 2.4619340896606445,
      "learning_rate": 0.0002162023460410557,
      "loss": 1.1136,
      "step": 3939
    },
    {
      "epoch": 0.6167814652473388,
      "grad_norm": 6.795989036560059,
      "learning_rate": 0.00021617790811339195,
      "loss": 1.067,
      "step": 3940
    },
    {
      "epoch": 0.6169380087664371,
      "grad_norm": 4.4798126220703125,
      "learning_rate": 0.00021615347018572823,
      "loss": 1.8717,
      "step": 3941
    },
    {
      "epoch": 0.6170945522855353,
      "grad_norm": 5.152458667755127,
      "learning_rate": 0.0002161290322580645,
      "loss": 1.3237,
      "step": 3942
    },
    {
      "epoch": 0.6172510958046337,
      "grad_norm": 4.771958827972412,
      "learning_rate": 0.00021610459433040076,
      "loss": 1.6574,
      "step": 3943
    },
    {
      "epoch": 0.617407639323732,
      "grad_norm": 2.8596343994140625,
      "learning_rate": 0.00021608015640273704,
      "loss": 2.0387,
      "step": 3944
    },
    {
      "epoch": 0.6175641828428303,
      "grad_norm": 2.4196653366088867,
      "learning_rate": 0.0002160557184750733,
      "loss": 1.2219,
      "step": 3945
    },
    {
      "epoch": 0.6177207263619287,
      "grad_norm": 1.575811505317688,
      "learning_rate": 0.00021603128054740954,
      "loss": 0.9666,
      "step": 3946
    },
    {
      "epoch": 0.617877269881027,
      "grad_norm": 1.7998102903366089,
      "learning_rate": 0.00021600684261974582,
      "loss": 0.6596,
      "step": 3947
    },
    {
      "epoch": 0.6180338134001252,
      "grad_norm": 2.5774621963500977,
      "learning_rate": 0.0002159824046920821,
      "loss": 0.7692,
      "step": 3948
    },
    {
      "epoch": 0.6181903569192235,
      "grad_norm": 2.7812228202819824,
      "learning_rate": 0.00021595796676441835,
      "loss": 1.069,
      "step": 3949
    },
    {
      "epoch": 0.6183469004383219,
      "grad_norm": 3.195280075073242,
      "learning_rate": 0.00021593352883675463,
      "loss": 1.0517,
      "step": 3950
    },
    {
      "epoch": 0.6185034439574202,
      "grad_norm": 0.4578677713871002,
      "learning_rate": 0.0002159090909090909,
      "loss": 0.2721,
      "step": 3951
    },
    {
      "epoch": 0.6186599874765185,
      "grad_norm": 0.7027708292007446,
      "learning_rate": 0.00021588465298142713,
      "loss": 0.3176,
      "step": 3952
    },
    {
      "epoch": 0.6188165309956167,
      "grad_norm": 0.6740389466285706,
      "learning_rate": 0.0002158602150537634,
      "loss": 0.416,
      "step": 3953
    },
    {
      "epoch": 0.6189730745147151,
      "grad_norm": 0.6556975245475769,
      "learning_rate": 0.0002158357771260997,
      "loss": 0.3664,
      "step": 3954
    },
    {
      "epoch": 0.6191296180338134,
      "grad_norm": 0.4339908957481384,
      "learning_rate": 0.00021581133919843594,
      "loss": 0.3161,
      "step": 3955
    },
    {
      "epoch": 0.6192861615529117,
      "grad_norm": 0.7102330923080444,
      "learning_rate": 0.00021578690127077222,
      "loss": 0.4315,
      "step": 3956
    },
    {
      "epoch": 0.61944270507201,
      "grad_norm": 0.7351087331771851,
      "learning_rate": 0.0002157624633431085,
      "loss": 0.3424,
      "step": 3957
    },
    {
      "epoch": 0.6195992485911084,
      "grad_norm": 0.7561622858047485,
      "learning_rate": 0.00021573802541544475,
      "loss": 0.4188,
      "step": 3958
    },
    {
      "epoch": 0.6197557921102066,
      "grad_norm": 1.3164188861846924,
      "learning_rate": 0.00021571358748778102,
      "loss": 0.4811,
      "step": 3959
    },
    {
      "epoch": 0.6199123356293049,
      "grad_norm": 0.9041202664375305,
      "learning_rate": 0.0002156891495601173,
      "loss": 0.3488,
      "step": 3960
    },
    {
      "epoch": 0.6200688791484033,
      "grad_norm": 1.003924012184143,
      "learning_rate": 0.00021566471163245353,
      "loss": 0.3205,
      "step": 3961
    },
    {
      "epoch": 0.6202254226675016,
      "grad_norm": 0.5604303479194641,
      "learning_rate": 0.0002156402737047898,
      "loss": 0.2906,
      "step": 3962
    },
    {
      "epoch": 0.6203819661865999,
      "grad_norm": 0.638725221157074,
      "learning_rate": 0.00021561583577712608,
      "loss": 0.3154,
      "step": 3963
    },
    {
      "epoch": 0.6205385097056982,
      "grad_norm": 1.5510947704315186,
      "learning_rate": 0.00021559139784946234,
      "loss": 0.5327,
      "step": 3964
    },
    {
      "epoch": 0.6206950532247965,
      "grad_norm": 1.3755923509597778,
      "learning_rate": 0.0002155669599217986,
      "loss": 0.5919,
      "step": 3965
    },
    {
      "epoch": 0.6208515967438948,
      "grad_norm": 0.9352520704269409,
      "learning_rate": 0.0002155425219941349,
      "loss": 0.2272,
      "step": 3966
    },
    {
      "epoch": 0.6210081402629931,
      "grad_norm": 1.7107229232788086,
      "learning_rate": 0.00021551808406647114,
      "loss": 0.352,
      "step": 3967
    },
    {
      "epoch": 0.6211646837820914,
      "grad_norm": 1.7082419395446777,
      "learning_rate": 0.0002154936461388074,
      "loss": 0.7203,
      "step": 3968
    },
    {
      "epoch": 0.6213212273011898,
      "grad_norm": 2.179602861404419,
      "learning_rate": 0.00021546920821114367,
      "loss": 0.8749,
      "step": 3969
    },
    {
      "epoch": 0.621477770820288,
      "grad_norm": 1.580485224723816,
      "learning_rate": 0.00021544477028347992,
      "loss": 0.6856,
      "step": 3970
    },
    {
      "epoch": 0.6216343143393863,
      "grad_norm": 1.842761754989624,
      "learning_rate": 0.0002154203323558162,
      "loss": 0.4401,
      "step": 3971
    },
    {
      "epoch": 0.6217908578584846,
      "grad_norm": 1.207093596458435,
      "learning_rate": 0.00021539589442815248,
      "loss": 0.6815,
      "step": 3972
    },
    {
      "epoch": 0.621947401377583,
      "grad_norm": 1.4717131853103638,
      "learning_rate": 0.00021537145650048873,
      "loss": 0.5138,
      "step": 3973
    },
    {
      "epoch": 0.6221039448966813,
      "grad_norm": 1.9231157302856445,
      "learning_rate": 0.000215347018572825,
      "loss": 0.6397,
      "step": 3974
    },
    {
      "epoch": 0.6222604884157796,
      "grad_norm": 1.4516040086746216,
      "learning_rate": 0.0002153225806451613,
      "loss": 0.6357,
      "step": 3975
    },
    {
      "epoch": 0.6224170319348779,
      "grad_norm": 1.3185113668441772,
      "learning_rate": 0.0002152981427174975,
      "loss": 0.497,
      "step": 3976
    },
    {
      "epoch": 0.6225735754539762,
      "grad_norm": 1.5022931098937988,
      "learning_rate": 0.0002152737047898338,
      "loss": 0.6394,
      "step": 3977
    },
    {
      "epoch": 0.6227301189730745,
      "grad_norm": 2.147803544998169,
      "learning_rate": 0.00021524926686217007,
      "loss": 0.6296,
      "step": 3978
    },
    {
      "epoch": 0.6228866624921728,
      "grad_norm": 2.3835220336914062,
      "learning_rate": 0.00021522482893450632,
      "loss": 1.1217,
      "step": 3979
    },
    {
      "epoch": 0.6230432060112712,
      "grad_norm": 2.3003957271575928,
      "learning_rate": 0.0002152003910068426,
      "loss": 0.9852,
      "step": 3980
    },
    {
      "epoch": 0.6231997495303695,
      "grad_norm": 2.8181819915771484,
      "learning_rate": 0.00021517595307917888,
      "loss": 0.9064,
      "step": 3981
    },
    {
      "epoch": 0.6233562930494677,
      "grad_norm": 1.4563791751861572,
      "learning_rate": 0.00021515151515151513,
      "loss": 0.8785,
      "step": 3982
    },
    {
      "epoch": 0.623512836568566,
      "grad_norm": 1.4717764854431152,
      "learning_rate": 0.0002151270772238514,
      "loss": 0.708,
      "step": 3983
    },
    {
      "epoch": 0.6236693800876644,
      "grad_norm": 1.8528285026550293,
      "learning_rate": 0.00021510263929618769,
      "loss": 1.0,
      "step": 3984
    },
    {
      "epoch": 0.6238259236067627,
      "grad_norm": 1.295715093612671,
      "learning_rate": 0.0002150782013685239,
      "loss": 0.6012,
      "step": 3985
    },
    {
      "epoch": 0.623982467125861,
      "grad_norm": 3.2727482318878174,
      "learning_rate": 0.0002150537634408602,
      "loss": 1.2052,
      "step": 3986
    },
    {
      "epoch": 0.6241390106449592,
      "grad_norm": 2.074563980102539,
      "learning_rate": 0.00021502932551319647,
      "loss": 1.0637,
      "step": 3987
    },
    {
      "epoch": 0.6242955541640576,
      "grad_norm": 2.350457191467285,
      "learning_rate": 0.00021500488758553272,
      "loss": 0.8989,
      "step": 3988
    },
    {
      "epoch": 0.6244520976831559,
      "grad_norm": 2.035271406173706,
      "learning_rate": 0.000214980449657869,
      "loss": 0.3382,
      "step": 3989
    },
    {
      "epoch": 0.6246086412022542,
      "grad_norm": 2.9534759521484375,
      "learning_rate": 0.00021495601173020527,
      "loss": 1.3783,
      "step": 3990
    },
    {
      "epoch": 0.6247651847213526,
      "grad_norm": 1.612892508506775,
      "learning_rate": 0.00021493157380254153,
      "loss": 1.3872,
      "step": 3991
    },
    {
      "epoch": 0.6249217282404509,
      "grad_norm": 1.9635756015777588,
      "learning_rate": 0.00021490713587487778,
      "loss": 1.1709,
      "step": 3992
    },
    {
      "epoch": 0.6250782717595491,
      "grad_norm": 3.8679616451263428,
      "learning_rate": 0.00021488269794721405,
      "loss": 1.7638,
      "step": 3993
    },
    {
      "epoch": 0.6252348152786474,
      "grad_norm": 2.8190555572509766,
      "learning_rate": 0.0002148582600195503,
      "loss": 1.1448,
      "step": 3994
    },
    {
      "epoch": 0.6253913587977458,
      "grad_norm": 1.001844882965088,
      "learning_rate": 0.00021483382209188658,
      "loss": 0.749,
      "step": 3995
    },
    {
      "epoch": 0.6255479023168441,
      "grad_norm": 1.1124273538589478,
      "learning_rate": 0.00021480938416422286,
      "loss": 0.5163,
      "step": 3996
    },
    {
      "epoch": 0.6257044458359424,
      "grad_norm": 2.1196939945220947,
      "learning_rate": 0.00021478494623655911,
      "loss": 0.2587,
      "step": 3997
    },
    {
      "epoch": 0.6258609893550408,
      "grad_norm": 3.0413784980773926,
      "learning_rate": 0.0002147605083088954,
      "loss": 0.7465,
      "step": 3998
    },
    {
      "epoch": 0.626017532874139,
      "grad_norm": 2.728447437286377,
      "learning_rate": 0.00021473607038123167,
      "loss": 0.7421,
      "step": 3999
    },
    {
      "epoch": 0.6261740763932373,
      "grad_norm": 1.2172927856445312,
      "learning_rate": 0.0002147116324535679,
      "loss": 0.5418,
      "step": 4000
    },
    {
      "epoch": 0.6261740763932373,
      "eval_loss": 0.6170632839202881,
      "eval_runtime": 206.2566,
      "eval_samples_per_second": 60.037,
      "eval_steps_per_second": 3.753,
      "eval_wer": 0.37072061542384765,
      "step": 4000
    },
    {
      "epoch": 0.6263306199123356,
      "grad_norm": 0.6230515837669373,
      "learning_rate": 0.00021468719452590417,
      "loss": 0.3052,
      "step": 4001
    },
    {
      "epoch": 0.626487163431434,
      "grad_norm": 0.36038628220558167,
      "learning_rate": 0.00021466275659824045,
      "loss": 0.253,
      "step": 4002
    },
    {
      "epoch": 0.6266437069505323,
      "grad_norm": 0.40060994029045105,
      "learning_rate": 0.0002146383186705767,
      "loss": 0.1601,
      "step": 4003
    },
    {
      "epoch": 0.6268002504696305,
      "grad_norm": 0.9217357635498047,
      "learning_rate": 0.00021461388074291298,
      "loss": 0.3302,
      "step": 4004
    },
    {
      "epoch": 0.6269567939887288,
      "grad_norm": 0.5147327780723572,
      "learning_rate": 0.00021458944281524926,
      "loss": 0.3022,
      "step": 4005
    },
    {
      "epoch": 0.6271133375078272,
      "grad_norm": 0.6115538477897644,
      "learning_rate": 0.0002145650048875855,
      "loss": 0.3489,
      "step": 4006
    },
    {
      "epoch": 0.6272698810269255,
      "grad_norm": 0.9012541174888611,
      "learning_rate": 0.0002145405669599218,
      "loss": 0.2962,
      "step": 4007
    },
    {
      "epoch": 0.6274264245460238,
      "grad_norm": 0.6350597143173218,
      "learning_rate": 0.00021451612903225807,
      "loss": 0.3945,
      "step": 4008
    },
    {
      "epoch": 0.6275829680651221,
      "grad_norm": 0.8227857947349548,
      "learning_rate": 0.0002144916911045943,
      "loss": 0.411,
      "step": 4009
    },
    {
      "epoch": 0.6277395115842204,
      "grad_norm": 0.618043839931488,
      "learning_rate": 0.00021446725317693057,
      "loss": 0.3542,
      "step": 4010
    },
    {
      "epoch": 0.6278960551033187,
      "grad_norm": 1.9552290439605713,
      "learning_rate": 0.00021444281524926685,
      "loss": 0.9303,
      "step": 4011
    },
    {
      "epoch": 0.628052598622417,
      "grad_norm": 0.5457572937011719,
      "learning_rate": 0.0002144183773216031,
      "loss": 0.3111,
      "step": 4012
    },
    {
      "epoch": 0.6282091421415154,
      "grad_norm": 1.2322858572006226,
      "learning_rate": 0.00021439393939393938,
      "loss": 0.3717,
      "step": 4013
    },
    {
      "epoch": 0.6283656856606137,
      "grad_norm": 1.5767748355865479,
      "learning_rate": 0.00021436950146627566,
      "loss": 0.4199,
      "step": 4014
    },
    {
      "epoch": 0.628522229179712,
      "grad_norm": 1.3694690465927124,
      "learning_rate": 0.0002143450635386119,
      "loss": 0.416,
      "step": 4015
    },
    {
      "epoch": 0.6286787726988102,
      "grad_norm": 0.9982991814613342,
      "learning_rate": 0.00021432062561094816,
      "loss": 0.4201,
      "step": 4016
    },
    {
      "epoch": 0.6288353162179086,
      "grad_norm": 1.6237698793411255,
      "learning_rate": 0.00021429618768328444,
      "loss": 0.677,
      "step": 4017
    },
    {
      "epoch": 0.6289918597370069,
      "grad_norm": 1.1065876483917236,
      "learning_rate": 0.0002142717497556207,
      "loss": 0.5136,
      "step": 4018
    },
    {
      "epoch": 0.6291484032561052,
      "grad_norm": 0.8187892436981201,
      "learning_rate": 0.00021424731182795697,
      "loss": 0.4697,
      "step": 4019
    },
    {
      "epoch": 0.6293049467752035,
      "grad_norm": 0.7985676527023315,
      "learning_rate": 0.00021422287390029325,
      "loss": 0.3248,
      "step": 4020
    },
    {
      "epoch": 0.6294614902943018,
      "grad_norm": 2.245656967163086,
      "learning_rate": 0.0002141984359726295,
      "loss": 0.788,
      "step": 4021
    },
    {
      "epoch": 0.6296180338134001,
      "grad_norm": 2.0072994232177734,
      "learning_rate": 0.00021417399804496577,
      "loss": 0.9463,
      "step": 4022
    },
    {
      "epoch": 0.6297745773324984,
      "grad_norm": 2.0363926887512207,
      "learning_rate": 0.00021414956011730205,
      "loss": 0.7412,
      "step": 4023
    },
    {
      "epoch": 0.6299311208515967,
      "grad_norm": 2.1571948528289795,
      "learning_rate": 0.00021412512218963828,
      "loss": 0.435,
      "step": 4024
    },
    {
      "epoch": 0.6300876643706951,
      "grad_norm": 1.8495596647262573,
      "learning_rate": 0.00021410068426197456,
      "loss": 0.7009,
      "step": 4025
    },
    {
      "epoch": 0.6302442078897934,
      "grad_norm": 1.2187695503234863,
      "learning_rate": 0.00021407624633431083,
      "loss": 0.5252,
      "step": 4026
    },
    {
      "epoch": 0.6304007514088916,
      "grad_norm": 1.3533210754394531,
      "learning_rate": 0.00021405180840664709,
      "loss": 0.586,
      "step": 4027
    },
    {
      "epoch": 0.63055729492799,
      "grad_norm": 3.1232190132141113,
      "learning_rate": 0.00021402737047898336,
      "loss": 0.9532,
      "step": 4028
    },
    {
      "epoch": 0.6307138384470883,
      "grad_norm": 1.3464336395263672,
      "learning_rate": 0.00021400293255131964,
      "loss": 0.7114,
      "step": 4029
    },
    {
      "epoch": 0.6308703819661866,
      "grad_norm": 1.8784873485565186,
      "learning_rate": 0.0002139784946236559,
      "loss": 1.0843,
      "step": 4030
    },
    {
      "epoch": 0.6310269254852849,
      "grad_norm": 1.593939185142517,
      "learning_rate": 0.00021395405669599217,
      "loss": 0.8251,
      "step": 4031
    },
    {
      "epoch": 0.6311834690043833,
      "grad_norm": 3.432223320007324,
      "learning_rate": 0.00021392961876832845,
      "loss": 0.9525,
      "step": 4032
    },
    {
      "epoch": 0.6313400125234815,
      "grad_norm": 2.074753522872925,
      "learning_rate": 0.00021390518084066467,
      "loss": 0.5878,
      "step": 4033
    },
    {
      "epoch": 0.6314965560425798,
      "grad_norm": 1.5874582529067993,
      "learning_rate": 0.00021388074291300095,
      "loss": 0.6233,
      "step": 4034
    },
    {
      "epoch": 0.6316530995616781,
      "grad_norm": 1.6352641582489014,
      "learning_rate": 0.00021385630498533723,
      "loss": 0.9886,
      "step": 4035
    },
    {
      "epoch": 0.6318096430807765,
      "grad_norm": 2.6665916442871094,
      "learning_rate": 0.00021383186705767348,
      "loss": 1.6576,
      "step": 4036
    },
    {
      "epoch": 0.6319661865998748,
      "grad_norm": 1.7960261106491089,
      "learning_rate": 0.00021380742913000976,
      "loss": 0.6991,
      "step": 4037
    },
    {
      "epoch": 0.632122730118973,
      "grad_norm": 1.923484206199646,
      "learning_rate": 0.00021378299120234604,
      "loss": 0.8864,
      "step": 4038
    },
    {
      "epoch": 0.6322792736380713,
      "grad_norm": 1.8930014371871948,
      "learning_rate": 0.00021375855327468226,
      "loss": 0.9939,
      "step": 4039
    },
    {
      "epoch": 0.6324358171571697,
      "grad_norm": 3.678338050842285,
      "learning_rate": 0.00021373411534701854,
      "loss": 1.4059,
      "step": 4040
    },
    {
      "epoch": 0.632592360676268,
      "grad_norm": 2.632375955581665,
      "learning_rate": 0.00021370967741935482,
      "loss": 1.4417,
      "step": 4041
    },
    {
      "epoch": 0.6327489041953663,
      "grad_norm": 2.788484573364258,
      "learning_rate": 0.00021368523949169107,
      "loss": 0.9454,
      "step": 4042
    },
    {
      "epoch": 0.6329054477144647,
      "grad_norm": 2.319645643234253,
      "learning_rate": 0.00021366080156402735,
      "loss": 1.4575,
      "step": 4043
    },
    {
      "epoch": 0.6330619912335629,
      "grad_norm": 3.663972854614258,
      "learning_rate": 0.00021363636363636363,
      "loss": 1.7115,
      "step": 4044
    },
    {
      "epoch": 0.6332185347526612,
      "grad_norm": 1.7628999948501587,
      "learning_rate": 0.00021361192570869988,
      "loss": 0.8049,
      "step": 4045
    },
    {
      "epoch": 0.6333750782717595,
      "grad_norm": 1.7270902395248413,
      "learning_rate": 0.00021358748778103616,
      "loss": 0.9988,
      "step": 4046
    },
    {
      "epoch": 0.6335316217908579,
      "grad_norm": 2.860292911529541,
      "learning_rate": 0.00021356304985337244,
      "loss": 0.9895,
      "step": 4047
    },
    {
      "epoch": 0.6336881653099562,
      "grad_norm": 3.5345377922058105,
      "learning_rate": 0.00021353861192570866,
      "loss": 1.4816,
      "step": 4048
    },
    {
      "epoch": 0.6338447088290545,
      "grad_norm": 3.5948832035064697,
      "learning_rate": 0.00021351417399804494,
      "loss": 1.0884,
      "step": 4049
    },
    {
      "epoch": 0.6340012523481527,
      "grad_norm": 3.8039002418518066,
      "learning_rate": 0.00021348973607038122,
      "loss": 1.7915,
      "step": 4050
    },
    {
      "epoch": 0.6341577958672511,
      "grad_norm": 0.4440103769302368,
      "learning_rate": 0.00021346529814271747,
      "loss": 0.313,
      "step": 4051
    },
    {
      "epoch": 0.6343143393863494,
      "grad_norm": 0.5781520009040833,
      "learning_rate": 0.00021344086021505375,
      "loss": 0.3249,
      "step": 4052
    },
    {
      "epoch": 0.6344708829054477,
      "grad_norm": 0.5911738276481628,
      "learning_rate": 0.00021341642228739002,
      "loss": 0.3927,
      "step": 4053
    },
    {
      "epoch": 0.6346274264245461,
      "grad_norm": 0.4731442332267761,
      "learning_rate": 0.00021339198435972628,
      "loss": 0.243,
      "step": 4054
    },
    {
      "epoch": 0.6347839699436444,
      "grad_norm": 0.5945417881011963,
      "learning_rate": 0.00021336754643206255,
      "loss": 0.3089,
      "step": 4055
    },
    {
      "epoch": 0.6349405134627426,
      "grad_norm": 0.6877188086509705,
      "learning_rate": 0.0002133431085043988,
      "loss": 0.4119,
      "step": 4056
    },
    {
      "epoch": 0.6350970569818409,
      "grad_norm": 0.788280725479126,
      "learning_rate": 0.00021331867057673506,
      "loss": 0.3965,
      "step": 4057
    },
    {
      "epoch": 0.6352536005009393,
      "grad_norm": 0.7950897216796875,
      "learning_rate": 0.00021329423264907133,
      "loss": 0.4293,
      "step": 4058
    },
    {
      "epoch": 0.6354101440200376,
      "grad_norm": 0.7185248732566833,
      "learning_rate": 0.0002132697947214076,
      "loss": 0.3425,
      "step": 4059
    },
    {
      "epoch": 0.6355666875391359,
      "grad_norm": 1.1782954931259155,
      "learning_rate": 0.00021324535679374386,
      "loss": 0.3744,
      "step": 4060
    },
    {
      "epoch": 0.6357232310582341,
      "grad_norm": 1.3076963424682617,
      "learning_rate": 0.00021322091886608014,
      "loss": 0.7161,
      "step": 4061
    },
    {
      "epoch": 0.6358797745773325,
      "grad_norm": 0.9288998246192932,
      "learning_rate": 0.00021319648093841642,
      "loss": 0.3778,
      "step": 4062
    },
    {
      "epoch": 0.6360363180964308,
      "grad_norm": 1.1923285722732544,
      "learning_rate": 0.00021317204301075265,
      "loss": 0.5499,
      "step": 4063
    },
    {
      "epoch": 0.6361928616155291,
      "grad_norm": 0.8304926156997681,
      "learning_rate": 0.00021314760508308892,
      "loss": 0.4468,
      "step": 4064
    },
    {
      "epoch": 0.6363494051346275,
      "grad_norm": 0.9042999744415283,
      "learning_rate": 0.0002131231671554252,
      "loss": 0.4214,
      "step": 4065
    },
    {
      "epoch": 0.6365059486537258,
      "grad_norm": 1.1802743673324585,
      "learning_rate": 0.00021309872922776145,
      "loss": 0.4863,
      "step": 4066
    },
    {
      "epoch": 0.636662492172824,
      "grad_norm": 0.9650530219078064,
      "learning_rate": 0.00021307429130009773,
      "loss": 0.5456,
      "step": 4067
    },
    {
      "epoch": 0.6368190356919223,
      "grad_norm": 1.8889888525009155,
      "learning_rate": 0.000213049853372434,
      "loss": 0.4523,
      "step": 4068
    },
    {
      "epoch": 0.6369755792110207,
      "grad_norm": 1.3778254985809326,
      "learning_rate": 0.00021302541544477026,
      "loss": 0.5295,
      "step": 4069
    },
    {
      "epoch": 0.637132122730119,
      "grad_norm": 1.3689180612564087,
      "learning_rate": 0.00021300097751710654,
      "loss": 0.8532,
      "step": 4070
    },
    {
      "epoch": 0.6372886662492173,
      "grad_norm": 1.6118444204330444,
      "learning_rate": 0.00021297653958944282,
      "loss": 0.5106,
      "step": 4071
    },
    {
      "epoch": 0.6374452097683156,
      "grad_norm": 2.750725507736206,
      "learning_rate": 0.00021295210166177904,
      "loss": 0.8254,
      "step": 4072
    },
    {
      "epoch": 0.6376017532874139,
      "grad_norm": 2.30176043510437,
      "learning_rate": 0.00021292766373411532,
      "loss": 0.8185,
      "step": 4073
    },
    {
      "epoch": 0.6377582968065122,
      "grad_norm": 2.448899745941162,
      "learning_rate": 0.0002129032258064516,
      "loss": 0.7568,
      "step": 4074
    },
    {
      "epoch": 0.6379148403256105,
      "grad_norm": 1.9214074611663818,
      "learning_rate": 0.00021287878787878785,
      "loss": 0.7279,
      "step": 4075
    },
    {
      "epoch": 0.6380713838447089,
      "grad_norm": 1.4484351873397827,
      "learning_rate": 0.00021285434995112413,
      "loss": 0.8629,
      "step": 4076
    },
    {
      "epoch": 0.6382279273638072,
      "grad_norm": 1.9307438135147095,
      "learning_rate": 0.0002128299120234604,
      "loss": 0.7699,
      "step": 4077
    },
    {
      "epoch": 0.6383844708829054,
      "grad_norm": 1.6077808141708374,
      "learning_rate": 0.00021280547409579666,
      "loss": 0.5525,
      "step": 4078
    },
    {
      "epoch": 0.6385410144020037,
      "grad_norm": 2.4951558113098145,
      "learning_rate": 0.00021278103616813294,
      "loss": 0.688,
      "step": 4079
    },
    {
      "epoch": 0.638697557921102,
      "grad_norm": 3.6255581378936768,
      "learning_rate": 0.0002127565982404692,
      "loss": 1.2918,
      "step": 4080
    },
    {
      "epoch": 0.6388541014402004,
      "grad_norm": 2.62573504447937,
      "learning_rate": 0.00021273216031280544,
      "loss": 0.8204,
      "step": 4081
    },
    {
      "epoch": 0.6390106449592987,
      "grad_norm": 1.5834779739379883,
      "learning_rate": 0.00021270772238514172,
      "loss": 0.5305,
      "step": 4082
    },
    {
      "epoch": 0.639167188478397,
      "grad_norm": 2.354691505432129,
      "learning_rate": 0.000212683284457478,
      "loss": 1.1743,
      "step": 4083
    },
    {
      "epoch": 0.6393237319974953,
      "grad_norm": 1.5539332628250122,
      "learning_rate": 0.00021265884652981425,
      "loss": 0.5419,
      "step": 4084
    },
    {
      "epoch": 0.6394802755165936,
      "grad_norm": 1.5284450054168701,
      "learning_rate": 0.00021263440860215052,
      "loss": 1.0443,
      "step": 4085
    },
    {
      "epoch": 0.6396368190356919,
      "grad_norm": 3.700737476348877,
      "learning_rate": 0.0002126099706744868,
      "loss": 1.6458,
      "step": 4086
    },
    {
      "epoch": 0.6397933625547902,
      "grad_norm": 2.4430134296417236,
      "learning_rate": 0.00021258553274682303,
      "loss": 0.8026,
      "step": 4087
    },
    {
      "epoch": 0.6399499060738886,
      "grad_norm": 2.9838156700134277,
      "learning_rate": 0.0002125610948191593,
      "loss": 1.4549,
      "step": 4088
    },
    {
      "epoch": 0.6401064495929869,
      "grad_norm": 1.9823230504989624,
      "learning_rate": 0.00021253665689149558,
      "loss": 1.2308,
      "step": 4089
    },
    {
      "epoch": 0.6402629931120851,
      "grad_norm": 2.805058717727661,
      "learning_rate": 0.00021251221896383184,
      "loss": 1.7271,
      "step": 4090
    },
    {
      "epoch": 0.6404195366311835,
      "grad_norm": 3.0151147842407227,
      "learning_rate": 0.00021248778103616811,
      "loss": 1.0577,
      "step": 4091
    },
    {
      "epoch": 0.6405760801502818,
      "grad_norm": 2.8304717540740967,
      "learning_rate": 0.0002124633431085044,
      "loss": 0.9225,
      "step": 4092
    },
    {
      "epoch": 0.6407326236693801,
      "grad_norm": 1.2722463607788086,
      "learning_rate": 0.00021243890518084064,
      "loss": 1.0837,
      "step": 4093
    },
    {
      "epoch": 0.6408891671884784,
      "grad_norm": 2.478703498840332,
      "learning_rate": 0.00021241446725317692,
      "loss": 1.4426,
      "step": 4094
    },
    {
      "epoch": 0.6410457107075767,
      "grad_norm": 1.236244559288025,
      "learning_rate": 0.0002123900293255132,
      "loss": 0.5695,
      "step": 4095
    },
    {
      "epoch": 0.641202254226675,
      "grad_norm": 1.9802755117416382,
      "learning_rate": 0.00021236559139784942,
      "loss": 0.6294,
      "step": 4096
    },
    {
      "epoch": 0.6413587977457733,
      "grad_norm": 2.97894549369812,
      "learning_rate": 0.0002123411534701857,
      "loss": 1.5215,
      "step": 4097
    },
    {
      "epoch": 0.6415153412648716,
      "grad_norm": 7.575010776519775,
      "learning_rate": 0.00021231671554252198,
      "loss": 1.141,
      "step": 4098
    },
    {
      "epoch": 0.64167188478397,
      "grad_norm": 2.812551975250244,
      "learning_rate": 0.00021229227761485823,
      "loss": 1.2502,
      "step": 4099
    },
    {
      "epoch": 0.6418284283030683,
      "grad_norm": 1.490632176399231,
      "learning_rate": 0.0002122678396871945,
      "loss": 1.0142,
      "step": 4100
    },
    {
      "epoch": 0.6419849718221665,
      "grad_norm": 0.6560826897621155,
      "learning_rate": 0.0002122434017595308,
      "loss": 0.479,
      "step": 4101
    },
    {
      "epoch": 0.6421415153412648,
      "grad_norm": 0.4517360329627991,
      "learning_rate": 0.00021221896383186704,
      "loss": 0.2937,
      "step": 4102
    },
    {
      "epoch": 0.6422980588603632,
      "grad_norm": 0.5508907437324524,
      "learning_rate": 0.00021219452590420332,
      "loss": 0.2687,
      "step": 4103
    },
    {
      "epoch": 0.6424546023794615,
      "grad_norm": 0.5405100584030151,
      "learning_rate": 0.00021217008797653957,
      "loss": 0.3279,
      "step": 4104
    },
    {
      "epoch": 0.6426111458985598,
      "grad_norm": 0.8774932622909546,
      "learning_rate": 0.00021214565004887582,
      "loss": 0.4322,
      "step": 4105
    },
    {
      "epoch": 0.6427676894176582,
      "grad_norm": 0.5635179281234741,
      "learning_rate": 0.0002121212121212121,
      "loss": 0.2838,
      "step": 4106
    },
    {
      "epoch": 0.6429242329367564,
      "grad_norm": 2.6404831409454346,
      "learning_rate": 0.00021209677419354838,
      "loss": 0.3582,
      "step": 4107
    },
    {
      "epoch": 0.6430807764558547,
      "grad_norm": 0.9229952692985535,
      "learning_rate": 0.00021207233626588463,
      "loss": 0.4306,
      "step": 4108
    },
    {
      "epoch": 0.643237319974953,
      "grad_norm": 1.216314435005188,
      "learning_rate": 0.0002120478983382209,
      "loss": 0.4203,
      "step": 4109
    },
    {
      "epoch": 0.6433938634940514,
      "grad_norm": 0.6581313014030457,
      "learning_rate": 0.00021202346041055719,
      "loss": 0.3948,
      "step": 4110
    },
    {
      "epoch": 0.6435504070131497,
      "grad_norm": 0.8522962331771851,
      "learning_rate": 0.0002119990224828934,
      "loss": 0.4591,
      "step": 4111
    },
    {
      "epoch": 0.6437069505322479,
      "grad_norm": 1.1675865650177002,
      "learning_rate": 0.0002119745845552297,
      "loss": 0.581,
      "step": 4112
    },
    {
      "epoch": 0.6438634940513462,
      "grad_norm": 0.7074406147003174,
      "learning_rate": 0.00021195014662756597,
      "loss": 0.3727,
      "step": 4113
    },
    {
      "epoch": 0.6440200375704446,
      "grad_norm": 1.105228066444397,
      "learning_rate": 0.00021192570869990222,
      "loss": 0.4141,
      "step": 4114
    },
    {
      "epoch": 0.6441765810895429,
      "grad_norm": 1.1589969396591187,
      "learning_rate": 0.0002119012707722385,
      "loss": 0.4892,
      "step": 4115
    },
    {
      "epoch": 0.6443331246086412,
      "grad_norm": 2.829336404800415,
      "learning_rate": 0.00021187683284457477,
      "loss": 0.861,
      "step": 4116
    },
    {
      "epoch": 0.6444896681277396,
      "grad_norm": 0.7126592993736267,
      "learning_rate": 0.00021185239491691103,
      "loss": 0.3913,
      "step": 4117
    },
    {
      "epoch": 0.6446462116468378,
      "grad_norm": 1.3239237070083618,
      "learning_rate": 0.0002118279569892473,
      "loss": 0.7084,
      "step": 4118
    },
    {
      "epoch": 0.6448027551659361,
      "grad_norm": 0.9522157311439514,
      "learning_rate": 0.00021180351906158358,
      "loss": 0.4494,
      "step": 4119
    },
    {
      "epoch": 0.6449592986850344,
      "grad_norm": 1.3508566617965698,
      "learning_rate": 0.0002117790811339198,
      "loss": 0.6281,
      "step": 4120
    },
    {
      "epoch": 0.6451158422041328,
      "grad_norm": 2.4964592456817627,
      "learning_rate": 0.00021175464320625608,
      "loss": 0.5852,
      "step": 4121
    },
    {
      "epoch": 0.6452723857232311,
      "grad_norm": 1.3371846675872803,
      "learning_rate": 0.00021173020527859236,
      "loss": 0.5809,
      "step": 4122
    },
    {
      "epoch": 0.6454289292423294,
      "grad_norm": 1.801464319229126,
      "learning_rate": 0.00021170576735092861,
      "loss": 0.5822,
      "step": 4123
    },
    {
      "epoch": 0.6455854727614276,
      "grad_norm": 1.4076706171035767,
      "learning_rate": 0.0002116813294232649,
      "loss": 0.6492,
      "step": 4124
    },
    {
      "epoch": 0.645742016280526,
      "grad_norm": 1.2247923612594604,
      "learning_rate": 0.00021165689149560117,
      "loss": 0.7113,
      "step": 4125
    },
    {
      "epoch": 0.6458985597996243,
      "grad_norm": 2.9736526012420654,
      "learning_rate": 0.00021163245356793742,
      "loss": 0.9173,
      "step": 4126
    },
    {
      "epoch": 0.6460551033187226,
      "grad_norm": 1.522207498550415,
      "learning_rate": 0.00021160801564027367,
      "loss": 0.717,
      "step": 4127
    },
    {
      "epoch": 0.646211646837821,
      "grad_norm": 2.8256654739379883,
      "learning_rate": 0.00021158357771260995,
      "loss": 0.7599,
      "step": 4128
    },
    {
      "epoch": 0.6463681903569192,
      "grad_norm": 2.5539469718933105,
      "learning_rate": 0.0002115591397849462,
      "loss": 0.9544,
      "step": 4129
    },
    {
      "epoch": 0.6465247338760175,
      "grad_norm": 1.8205986022949219,
      "learning_rate": 0.00021153470185728248,
      "loss": 0.9117,
      "step": 4130
    },
    {
      "epoch": 0.6466812773951158,
      "grad_norm": 3.9707062244415283,
      "learning_rate": 0.00021151026392961876,
      "loss": 1.1332,
      "step": 4131
    },
    {
      "epoch": 0.6468378209142142,
      "grad_norm": 1.7716293334960938,
      "learning_rate": 0.000211485826001955,
      "loss": 0.5995,
      "step": 4132
    },
    {
      "epoch": 0.6469943644333125,
      "grad_norm": 2.0093953609466553,
      "learning_rate": 0.0002114613880742913,
      "loss": 1.1797,
      "step": 4133
    },
    {
      "epoch": 0.6471509079524108,
      "grad_norm": 1.3525505065917969,
      "learning_rate": 0.00021143695014662757,
      "loss": 0.5744,
      "step": 4134
    },
    {
      "epoch": 0.647307451471509,
      "grad_norm": 0.9893065690994263,
      "learning_rate": 0.0002114125122189638,
      "loss": 0.8285,
      "step": 4135
    },
    {
      "epoch": 0.6474639949906074,
      "grad_norm": 3.0879127979278564,
      "learning_rate": 0.00021138807429130007,
      "loss": 1.8302,
      "step": 4136
    },
    {
      "epoch": 0.6476205385097057,
      "grad_norm": 1.6415681838989258,
      "learning_rate": 0.00021136363636363635,
      "loss": 0.8313,
      "step": 4137
    },
    {
      "epoch": 0.647777082028804,
      "grad_norm": 2.8045177459716797,
      "learning_rate": 0.0002113391984359726,
      "loss": 0.9433,
      "step": 4138
    },
    {
      "epoch": 0.6479336255479023,
      "grad_norm": 1.968576431274414,
      "learning_rate": 0.00021131476050830888,
      "loss": 0.7906,
      "step": 4139
    },
    {
      "epoch": 0.6480901690670007,
      "grad_norm": 1.4854947328567505,
      "learning_rate": 0.00021129032258064516,
      "loss": 1.2082,
      "step": 4140
    },
    {
      "epoch": 0.6482467125860989,
      "grad_norm": 3.7475380897521973,
      "learning_rate": 0.0002112658846529814,
      "loss": 1.9279,
      "step": 4141
    },
    {
      "epoch": 0.6484032561051972,
      "grad_norm": 1.9504555463790894,
      "learning_rate": 0.00021124144672531769,
      "loss": 1.4245,
      "step": 4142
    },
    {
      "epoch": 0.6485597996242956,
      "grad_norm": 2.5647873878479004,
      "learning_rate": 0.00021121700879765396,
      "loss": 1.1829,
      "step": 4143
    },
    {
      "epoch": 0.6487163431433939,
      "grad_norm": 1.7342371940612793,
      "learning_rate": 0.0002111925708699902,
      "loss": 1.759,
      "step": 4144
    },
    {
      "epoch": 0.6488728866624922,
      "grad_norm": 1.559617519378662,
      "learning_rate": 0.00021116813294232647,
      "loss": 1.0318,
      "step": 4145
    },
    {
      "epoch": 0.6490294301815904,
      "grad_norm": 1.5105706453323364,
      "learning_rate": 0.00021114369501466275,
      "loss": 1.1075,
      "step": 4146
    },
    {
      "epoch": 0.6491859737006888,
      "grad_norm": 1.054640769958496,
      "learning_rate": 0.000211119257086999,
      "loss": 0.7295,
      "step": 4147
    },
    {
      "epoch": 0.6493425172197871,
      "grad_norm": 1.3204190731048584,
      "learning_rate": 0.00021109481915933528,
      "loss": 0.6747,
      "step": 4148
    },
    {
      "epoch": 0.6494990607388854,
      "grad_norm": 1.6427069902420044,
      "learning_rate": 0.00021107038123167155,
      "loss": 1.0632,
      "step": 4149
    },
    {
      "epoch": 0.6496556042579837,
      "grad_norm": 2.6483640670776367,
      "learning_rate": 0.0002110459433040078,
      "loss": 1.724,
      "step": 4150
    },
    {
      "epoch": 0.6498121477770821,
      "grad_norm": 0.44321414828300476,
      "learning_rate": 0.00021102150537634406,
      "loss": 0.2836,
      "step": 4151
    },
    {
      "epoch": 0.6499686912961803,
      "grad_norm": 0.5823266506195068,
      "learning_rate": 0.00021099706744868033,
      "loss": 0.2743,
      "step": 4152
    },
    {
      "epoch": 0.6501252348152786,
      "grad_norm": 0.6924779415130615,
      "learning_rate": 0.00021097262952101659,
      "loss": 0.2912,
      "step": 4153
    },
    {
      "epoch": 0.650281778334377,
      "grad_norm": 0.7143043279647827,
      "learning_rate": 0.00021094819159335286,
      "loss": 0.4401,
      "step": 4154
    },
    {
      "epoch": 0.6504383218534753,
      "grad_norm": 1.3243781328201294,
      "learning_rate": 0.00021092375366568914,
      "loss": 0.3095,
      "step": 4155
    },
    {
      "epoch": 0.6505948653725736,
      "grad_norm": 0.8137784600257874,
      "learning_rate": 0.0002108993157380254,
      "loss": 0.3707,
      "step": 4156
    },
    {
      "epoch": 0.6507514088916719,
      "grad_norm": 0.5884976983070374,
      "learning_rate": 0.00021087487781036167,
      "loss": 0.2581,
      "step": 4157
    },
    {
      "epoch": 0.6509079524107702,
      "grad_norm": 0.9753175973892212,
      "learning_rate": 0.00021085043988269795,
      "loss": 0.4062,
      "step": 4158
    },
    {
      "epoch": 0.6510644959298685,
      "grad_norm": 0.8055758476257324,
      "learning_rate": 0.00021082600195503417,
      "loss": 0.4112,
      "step": 4159
    },
    {
      "epoch": 0.6512210394489668,
      "grad_norm": 0.7871150374412537,
      "learning_rate": 0.00021080156402737045,
      "loss": 0.5229,
      "step": 4160
    },
    {
      "epoch": 0.6513775829680651,
      "grad_norm": 0.666612982749939,
      "learning_rate": 0.00021077712609970673,
      "loss": 0.2359,
      "step": 4161
    },
    {
      "epoch": 0.6515341264871635,
      "grad_norm": 4.991723537445068,
      "learning_rate": 0.00021075268817204298,
      "loss": 1.1647,
      "step": 4162
    },
    {
      "epoch": 0.6516906700062617,
      "grad_norm": 0.6733570098876953,
      "learning_rate": 0.00021072825024437926,
      "loss": 0.2791,
      "step": 4163
    },
    {
      "epoch": 0.65184721352536,
      "grad_norm": 1.0828582048416138,
      "learning_rate": 0.00021070381231671554,
      "loss": 0.4028,
      "step": 4164
    },
    {
      "epoch": 0.6520037570444583,
      "grad_norm": 1.031664252281189,
      "learning_rate": 0.0002106793743890518,
      "loss": 0.4959,
      "step": 4165
    },
    {
      "epoch": 0.6521603005635567,
      "grad_norm": 1.7756761312484741,
      "learning_rate": 0.00021065493646138807,
      "loss": 0.5123,
      "step": 4166
    },
    {
      "epoch": 0.652316844082655,
      "grad_norm": 1.409739375114441,
      "learning_rate": 0.00021063049853372435,
      "loss": 0.5411,
      "step": 4167
    },
    {
      "epoch": 0.6524733876017533,
      "grad_norm": 0.8608100414276123,
      "learning_rate": 0.00021060606060606057,
      "loss": 0.5408,
      "step": 4168
    },
    {
      "epoch": 0.6526299311208515,
      "grad_norm": 1.9020010232925415,
      "learning_rate": 0.00021058162267839685,
      "loss": 0.5228,
      "step": 4169
    },
    {
      "epoch": 0.6527864746399499,
      "grad_norm": 1.2698709964752197,
      "learning_rate": 0.00021055718475073313,
      "loss": 0.4451,
      "step": 4170
    },
    {
      "epoch": 0.6529430181590482,
      "grad_norm": 1.3573118448257446,
      "learning_rate": 0.00021053274682306938,
      "loss": 0.5889,
      "step": 4171
    },
    {
      "epoch": 0.6530995616781465,
      "grad_norm": 1.1787749528884888,
      "learning_rate": 0.00021050830889540566,
      "loss": 0.4531,
      "step": 4172
    },
    {
      "epoch": 0.6532561051972449,
      "grad_norm": 1.463240146636963,
      "learning_rate": 0.00021048387096774194,
      "loss": 0.6581,
      "step": 4173
    },
    {
      "epoch": 0.6534126487163432,
      "grad_norm": 1.1334830522537231,
      "learning_rate": 0.0002104594330400782,
      "loss": 0.6735,
      "step": 4174
    },
    {
      "epoch": 0.6535691922354414,
      "grad_norm": 0.9221059083938599,
      "learning_rate": 0.00021043499511241444,
      "loss": 0.5811,
      "step": 4175
    },
    {
      "epoch": 0.6537257357545397,
      "grad_norm": 0.7749282717704773,
      "learning_rate": 0.00021041055718475072,
      "loss": 0.3367,
      "step": 4176
    },
    {
      "epoch": 0.6538822792736381,
      "grad_norm": 2.028270721435547,
      "learning_rate": 0.00021038611925708697,
      "loss": 0.744,
      "step": 4177
    },
    {
      "epoch": 0.6540388227927364,
      "grad_norm": 2.1838300228118896,
      "learning_rate": 0.00021036168132942325,
      "loss": 0.7978,
      "step": 4178
    },
    {
      "epoch": 0.6541953663118347,
      "grad_norm": 2.014901876449585,
      "learning_rate": 0.00021033724340175952,
      "loss": 0.7741,
      "step": 4179
    },
    {
      "epoch": 0.654351909830933,
      "grad_norm": 2.365983009338379,
      "learning_rate": 0.00021031280547409578,
      "loss": 0.8107,
      "step": 4180
    },
    {
      "epoch": 0.6545084533500313,
      "grad_norm": 2.7603774070739746,
      "learning_rate": 0.00021028836754643205,
      "loss": 1.2278,
      "step": 4181
    },
    {
      "epoch": 0.6546649968691296,
      "grad_norm": 1.7093403339385986,
      "learning_rate": 0.00021026392961876833,
      "loss": 0.8463,
      "step": 4182
    },
    {
      "epoch": 0.6548215403882279,
      "grad_norm": 1.8434052467346191,
      "learning_rate": 0.00021023949169110456,
      "loss": 0.8286,
      "step": 4183
    },
    {
      "epoch": 0.6549780839073263,
      "grad_norm": 1.2847257852554321,
      "learning_rate": 0.00021021505376344084,
      "loss": 0.7018,
      "step": 4184
    },
    {
      "epoch": 0.6551346274264246,
      "grad_norm": 1.887803077697754,
      "learning_rate": 0.0002101906158357771,
      "loss": 0.8022,
      "step": 4185
    },
    {
      "epoch": 0.6552911709455228,
      "grad_norm": 2.067678928375244,
      "learning_rate": 0.00021016617790811336,
      "loss": 0.5651,
      "step": 4186
    },
    {
      "epoch": 0.6554477144646211,
      "grad_norm": 1.9013502597808838,
      "learning_rate": 0.00021014173998044964,
      "loss": 0.9908,
      "step": 4187
    },
    {
      "epoch": 0.6556042579837195,
      "grad_norm": 2.773240566253662,
      "learning_rate": 0.00021011730205278592,
      "loss": 0.9834,
      "step": 4188
    },
    {
      "epoch": 0.6557608015028178,
      "grad_norm": 2.2640483379364014,
      "learning_rate": 0.00021009286412512217,
      "loss": 1.1264,
      "step": 4189
    },
    {
      "epoch": 0.6559173450219161,
      "grad_norm": 1.875777006149292,
      "learning_rate": 0.00021006842619745845,
      "loss": 1.1565,
      "step": 4190
    },
    {
      "epoch": 0.6560738885410144,
      "grad_norm": 4.020966529846191,
      "learning_rate": 0.00021004398826979473,
      "loss": 1.6043,
      "step": 4191
    },
    {
      "epoch": 0.6562304320601127,
      "grad_norm": 2.4041779041290283,
      "learning_rate": 0.00021001955034213095,
      "loss": 1.0613,
      "step": 4192
    },
    {
      "epoch": 0.656386975579211,
      "grad_norm": 2.9603583812713623,
      "learning_rate": 0.00020999511241446723,
      "loss": 1.2259,
      "step": 4193
    },
    {
      "epoch": 0.6565435190983093,
      "grad_norm": 3.0527658462524414,
      "learning_rate": 0.0002099706744868035,
      "loss": 0.9003,
      "step": 4194
    },
    {
      "epoch": 0.6567000626174077,
      "grad_norm": 2.1919569969177246,
      "learning_rate": 0.00020994623655913976,
      "loss": 1.3881,
      "step": 4195
    },
    {
      "epoch": 0.656856606136506,
      "grad_norm": 2.547144889831543,
      "learning_rate": 0.00020992179863147604,
      "loss": 0.6151,
      "step": 4196
    },
    {
      "epoch": 0.6570131496556043,
      "grad_norm": 1.940132975578308,
      "learning_rate": 0.00020989736070381232,
      "loss": 0.7662,
      "step": 4197
    },
    {
      "epoch": 0.6571696931747025,
      "grad_norm": 1.2711700201034546,
      "learning_rate": 0.00020987292277614854,
      "loss": 0.6982,
      "step": 4198
    },
    {
      "epoch": 0.6573262366938009,
      "grad_norm": 1.6666027307510376,
      "learning_rate": 0.00020984848484848482,
      "loss": 0.4992,
      "step": 4199
    },
    {
      "epoch": 0.6574827802128992,
      "grad_norm": 4.075043201446533,
      "learning_rate": 0.0002098240469208211,
      "loss": 1.2499,
      "step": 4200
    },
    {
      "epoch": 0.6576393237319975,
      "grad_norm": 0.5925846695899963,
      "learning_rate": 0.00020979960899315735,
      "loss": 0.2544,
      "step": 4201
    },
    {
      "epoch": 0.6577958672510958,
      "grad_norm": 0.90831458568573,
      "learning_rate": 0.00020977517106549363,
      "loss": 0.3609,
      "step": 4202
    },
    {
      "epoch": 0.6579524107701941,
      "grad_norm": 0.6981661319732666,
      "learning_rate": 0.0002097507331378299,
      "loss": 0.2574,
      "step": 4203
    },
    {
      "epoch": 0.6581089542892924,
      "grad_norm": 0.8402687311172485,
      "learning_rate": 0.00020972629521016616,
      "loss": 0.2842,
      "step": 4204
    },
    {
      "epoch": 0.6582654978083907,
      "grad_norm": 0.7427016496658325,
      "learning_rate": 0.00020970185728250244,
      "loss": 0.3959,
      "step": 4205
    },
    {
      "epoch": 0.658422041327489,
      "grad_norm": 0.7101203799247742,
      "learning_rate": 0.00020967741935483871,
      "loss": 0.4118,
      "step": 4206
    },
    {
      "epoch": 0.6585785848465874,
      "grad_norm": 0.7009015083312988,
      "learning_rate": 0.00020965298142717494,
      "loss": 0.3418,
      "step": 4207
    },
    {
      "epoch": 0.6587351283656857,
      "grad_norm": 0.8235505223274231,
      "learning_rate": 0.00020962854349951122,
      "loss": 0.3614,
      "step": 4208
    },
    {
      "epoch": 0.6588916718847839,
      "grad_norm": 0.49669182300567627,
      "learning_rate": 0.0002096041055718475,
      "loss": 0.3866,
      "step": 4209
    },
    {
      "epoch": 0.6590482154038823,
      "grad_norm": 8.812310218811035,
      "learning_rate": 0.00020957966764418375,
      "loss": 0.8613,
      "step": 4210
    },
    {
      "epoch": 0.6592047589229806,
      "grad_norm": 6.276821613311768,
      "learning_rate": 0.00020955522971652003,
      "loss": 1.7947,
      "step": 4211
    },
    {
      "epoch": 0.6593613024420789,
      "grad_norm": 2.7206177711486816,
      "learning_rate": 0.0002095307917888563,
      "loss": 0.5293,
      "step": 4212
    },
    {
      "epoch": 0.6595178459611772,
      "grad_norm": 2.726612091064453,
      "learning_rate": 0.00020950635386119255,
      "loss": 0.832,
      "step": 4213
    },
    {
      "epoch": 0.6596743894802756,
      "grad_norm": 0.86571204662323,
      "learning_rate": 0.00020948191593352883,
      "loss": 0.556,
      "step": 4214
    },
    {
      "epoch": 0.6598309329993738,
      "grad_norm": 1.1770457029342651,
      "learning_rate": 0.00020945747800586508,
      "loss": 0.5059,
      "step": 4215
    },
    {
      "epoch": 0.6599874765184721,
      "grad_norm": 1.2586264610290527,
      "learning_rate": 0.00020943304007820134,
      "loss": 0.6573,
      "step": 4216
    },
    {
      "epoch": 0.6601440200375704,
      "grad_norm": 1.2371019124984741,
      "learning_rate": 0.00020940860215053761,
      "loss": 0.471,
      "step": 4217
    },
    {
      "epoch": 0.6603005635566688,
      "grad_norm": 0.7235432863235474,
      "learning_rate": 0.0002093841642228739,
      "loss": 0.4126,
      "step": 4218
    },
    {
      "epoch": 0.6604571070757671,
      "grad_norm": 1.9548500776290894,
      "learning_rate": 0.00020935972629521014,
      "loss": 0.9866,
      "step": 4219
    },
    {
      "epoch": 0.6606136505948653,
      "grad_norm": 1.3150765895843506,
      "learning_rate": 0.00020933528836754642,
      "loss": 0.5777,
      "step": 4220
    },
    {
      "epoch": 0.6607701941139636,
      "grad_norm": 1.9192475080490112,
      "learning_rate": 0.0002093108504398827,
      "loss": 0.7398,
      "step": 4221
    },
    {
      "epoch": 0.660926737633062,
      "grad_norm": 2.385049819946289,
      "learning_rate": 0.00020928641251221892,
      "loss": 0.5958,
      "step": 4222
    },
    {
      "epoch": 0.6610832811521603,
      "grad_norm": 1.309512734413147,
      "learning_rate": 0.0002092619745845552,
      "loss": 0.6067,
      "step": 4223
    },
    {
      "epoch": 0.6612398246712586,
      "grad_norm": 1.5893750190734863,
      "learning_rate": 0.00020923753665689148,
      "loss": 0.7334,
      "step": 4224
    },
    {
      "epoch": 0.661396368190357,
      "grad_norm": 2.3424973487854004,
      "learning_rate": 0.00020921309872922773,
      "loss": 0.8073,
      "step": 4225
    },
    {
      "epoch": 0.6615529117094552,
      "grad_norm": 1.3689500093460083,
      "learning_rate": 0.000209188660801564,
      "loss": 0.6511,
      "step": 4226
    },
    {
      "epoch": 0.6617094552285535,
      "grad_norm": 1.4567039012908936,
      "learning_rate": 0.0002091642228739003,
      "loss": 0.8244,
      "step": 4227
    },
    {
      "epoch": 0.6618659987476518,
      "grad_norm": 5.747929573059082,
      "learning_rate": 0.00020913978494623654,
      "loss": 0.774,
      "step": 4228
    },
    {
      "epoch": 0.6620225422667502,
      "grad_norm": 1.2225399017333984,
      "learning_rate": 0.00020911534701857282,
      "loss": 0.6109,
      "step": 4229
    },
    {
      "epoch": 0.6621790857858485,
      "grad_norm": 1.25154709815979,
      "learning_rate": 0.0002090909090909091,
      "loss": 0.8214,
      "step": 4230
    },
    {
      "epoch": 0.6623356293049468,
      "grad_norm": 2.5937745571136475,
      "learning_rate": 0.00020906647116324532,
      "loss": 0.9792,
      "step": 4231
    },
    {
      "epoch": 0.662492172824045,
      "grad_norm": 4.251043796539307,
      "learning_rate": 0.0002090420332355816,
      "loss": 0.6997,
      "step": 4232
    },
    {
      "epoch": 0.6626487163431434,
      "grad_norm": 2.7848551273345947,
      "learning_rate": 0.00020901759530791788,
      "loss": 0.6984,
      "step": 4233
    },
    {
      "epoch": 0.6628052598622417,
      "grad_norm": 3.419072151184082,
      "learning_rate": 0.00020899315738025413,
      "loss": 0.8251,
      "step": 4234
    },
    {
      "epoch": 0.66296180338134,
      "grad_norm": 4.448485374450684,
      "learning_rate": 0.0002089687194525904,
      "loss": 0.9583,
      "step": 4235
    },
    {
      "epoch": 0.6631183469004384,
      "grad_norm": 3.649228572845459,
      "learning_rate": 0.00020894428152492669,
      "loss": 1.222,
      "step": 4236
    },
    {
      "epoch": 0.6632748904195366,
      "grad_norm": 2.3916118144989014,
      "learning_rate": 0.00020891984359726294,
      "loss": 1.2783,
      "step": 4237
    },
    {
      "epoch": 0.6634314339386349,
      "grad_norm": 2.919701337814331,
      "learning_rate": 0.00020889540566959922,
      "loss": 0.9857,
      "step": 4238
    },
    {
      "epoch": 0.6635879774577332,
      "grad_norm": 2.3915016651153564,
      "learning_rate": 0.00020887096774193547,
      "loss": 1.1059,
      "step": 4239
    },
    {
      "epoch": 0.6637445209768316,
      "grad_norm": 4.138210773468018,
      "learning_rate": 0.00020884652981427172,
      "loss": 1.2545,
      "step": 4240
    },
    {
      "epoch": 0.6639010644959299,
      "grad_norm": 4.440970420837402,
      "learning_rate": 0.000208822091886608,
      "loss": 1.5261,
      "step": 4241
    },
    {
      "epoch": 0.6640576080150282,
      "grad_norm": 3.668529748916626,
      "learning_rate": 0.00020879765395894427,
      "loss": 1.3974,
      "step": 4242
    },
    {
      "epoch": 0.6642141515341264,
      "grad_norm": 2.6934828758239746,
      "learning_rate": 0.00020877321603128053,
      "loss": 1.0031,
      "step": 4243
    },
    {
      "epoch": 0.6643706950532248,
      "grad_norm": 3.754547119140625,
      "learning_rate": 0.0002087487781036168,
      "loss": 1.5351,
      "step": 4244
    },
    {
      "epoch": 0.6645272385723231,
      "grad_norm": 2.1990721225738525,
      "learning_rate": 0.00020872434017595308,
      "loss": 1.2908,
      "step": 4245
    },
    {
      "epoch": 0.6646837820914214,
      "grad_norm": 1.8534988164901733,
      "learning_rate": 0.0002086999022482893,
      "loss": 0.7223,
      "step": 4246
    },
    {
      "epoch": 0.6648403256105198,
      "grad_norm": 1.668362021446228,
      "learning_rate": 0.00020867546432062559,
      "loss": 0.7743,
      "step": 4247
    },
    {
      "epoch": 0.6649968691296181,
      "grad_norm": 3.3219528198242188,
      "learning_rate": 0.00020865102639296186,
      "loss": 1.1719,
      "step": 4248
    },
    {
      "epoch": 0.6651534126487163,
      "grad_norm": 1.7605839967727661,
      "learning_rate": 0.00020862658846529811,
      "loss": 0.5382,
      "step": 4249
    },
    {
      "epoch": 0.6653099561678146,
      "grad_norm": 2.431262731552124,
      "learning_rate": 0.0002086021505376344,
      "loss": 1.0354,
      "step": 4250
    },
    {
      "epoch": 0.665466499686913,
      "grad_norm": 0.6644131541252136,
      "learning_rate": 0.00020857771260997067,
      "loss": 0.3289,
      "step": 4251
    },
    {
      "epoch": 0.6656230432060113,
      "grad_norm": 0.4004727303981781,
      "learning_rate": 0.00020855327468230692,
      "loss": 0.2793,
      "step": 4252
    },
    {
      "epoch": 0.6657795867251096,
      "grad_norm": 0.5396392345428467,
      "learning_rate": 0.0002085288367546432,
      "loss": 0.2703,
      "step": 4253
    },
    {
      "epoch": 0.6659361302442078,
      "grad_norm": 0.4889543056488037,
      "learning_rate": 0.00020850439882697948,
      "loss": 0.2565,
      "step": 4254
    },
    {
      "epoch": 0.6660926737633062,
      "grad_norm": 0.5937919616699219,
      "learning_rate": 0.0002084799608993157,
      "loss": 0.294,
      "step": 4255
    },
    {
      "epoch": 0.6662492172824045,
      "grad_norm": 1.0246959924697876,
      "learning_rate": 0.00020845552297165198,
      "loss": 0.4547,
      "step": 4256
    },
    {
      "epoch": 0.6664057608015028,
      "grad_norm": 0.8385992646217346,
      "learning_rate": 0.00020843108504398826,
      "loss": 0.558,
      "step": 4257
    },
    {
      "epoch": 0.6665623043206012,
      "grad_norm": 0.6174805760383606,
      "learning_rate": 0.0002084066471163245,
      "loss": 0.3394,
      "step": 4258
    },
    {
      "epoch": 0.6667188478396995,
      "grad_norm": 0.82181715965271,
      "learning_rate": 0.0002083822091886608,
      "loss": 0.2792,
      "step": 4259
    },
    {
      "epoch": 0.6668753913587977,
      "grad_norm": 0.5891256332397461,
      "learning_rate": 0.00020835777126099707,
      "loss": 0.3115,
      "step": 4260
    },
    {
      "epoch": 0.667031934877896,
      "grad_norm": 0.9402687549591064,
      "learning_rate": 0.00020833333333333332,
      "loss": 0.3959,
      "step": 4261
    },
    {
      "epoch": 0.6671884783969944,
      "grad_norm": 0.887904942035675,
      "learning_rate": 0.0002083088954056696,
      "loss": 0.3962,
      "step": 4262
    },
    {
      "epoch": 0.6673450219160927,
      "grad_norm": 1.0026453733444214,
      "learning_rate": 0.00020828445747800585,
      "loss": 0.4687,
      "step": 4263
    },
    {
      "epoch": 0.667501565435191,
      "grad_norm": 0.8495490550994873,
      "learning_rate": 0.0002082600195503421,
      "loss": 0.3504,
      "step": 4264
    },
    {
      "epoch": 0.6676581089542893,
      "grad_norm": 1.067874789237976,
      "learning_rate": 0.00020823558162267838,
      "loss": 0.4712,
      "step": 4265
    },
    {
      "epoch": 0.6678146524733876,
      "grad_norm": 1.5945181846618652,
      "learning_rate": 0.00020821114369501466,
      "loss": 0.5911,
      "step": 4266
    },
    {
      "epoch": 0.6679711959924859,
      "grad_norm": 1.3359369039535522,
      "learning_rate": 0.0002081867057673509,
      "loss": 0.5162,
      "step": 4267
    },
    {
      "epoch": 0.6681277395115842,
      "grad_norm": 1.5492264032363892,
      "learning_rate": 0.0002081622678396872,
      "loss": 0.5094,
      "step": 4268
    },
    {
      "epoch": 0.6682842830306825,
      "grad_norm": 1.4912946224212646,
      "learning_rate": 0.00020813782991202347,
      "loss": 0.5471,
      "step": 4269
    },
    {
      "epoch": 0.6684408265497809,
      "grad_norm": 0.7569031715393066,
      "learning_rate": 0.0002081133919843597,
      "loss": 0.3852,
      "step": 4270
    },
    {
      "epoch": 0.6685973700688791,
      "grad_norm": 1.4911766052246094,
      "learning_rate": 0.00020808895405669597,
      "loss": 0.5473,
      "step": 4271
    },
    {
      "epoch": 0.6687539135879774,
      "grad_norm": 1.7154449224472046,
      "learning_rate": 0.00020806451612903225,
      "loss": 0.4659,
      "step": 4272
    },
    {
      "epoch": 0.6689104571070758,
      "grad_norm": 1.293839693069458,
      "learning_rate": 0.0002080400782013685,
      "loss": 0.8146,
      "step": 4273
    },
    {
      "epoch": 0.6690670006261741,
      "grad_norm": 1.7432725429534912,
      "learning_rate": 0.00020801564027370478,
      "loss": 0.5831,
      "step": 4274
    },
    {
      "epoch": 0.6692235441452724,
      "grad_norm": 1.6460049152374268,
      "learning_rate": 0.00020799120234604105,
      "loss": 0.7064,
      "step": 4275
    },
    {
      "epoch": 0.6693800876643707,
      "grad_norm": 1.205126404762268,
      "learning_rate": 0.0002079667644183773,
      "loss": 0.6556,
      "step": 4276
    },
    {
      "epoch": 0.669536631183469,
      "grad_norm": 2.2884862422943115,
      "learning_rate": 0.00020794232649071358,
      "loss": 0.5786,
      "step": 4277
    },
    {
      "epoch": 0.6696931747025673,
      "grad_norm": 1.5183991193771362,
      "learning_rate": 0.00020791788856304986,
      "loss": 0.916,
      "step": 4278
    },
    {
      "epoch": 0.6698497182216656,
      "grad_norm": 1.8654146194458008,
      "learning_rate": 0.00020789345063538609,
      "loss": 0.6673,
      "step": 4279
    },
    {
      "epoch": 0.6700062617407639,
      "grad_norm": 1.886267066001892,
      "learning_rate": 0.00020786901270772236,
      "loss": 0.6066,
      "step": 4280
    },
    {
      "epoch": 0.6701628052598623,
      "grad_norm": 1.4289623498916626,
      "learning_rate": 0.00020784457478005864,
      "loss": 0.8144,
      "step": 4281
    },
    {
      "epoch": 0.6703193487789606,
      "grad_norm": 2.9784719944000244,
      "learning_rate": 0.0002078201368523949,
      "loss": 0.879,
      "step": 4282
    },
    {
      "epoch": 0.6704758922980588,
      "grad_norm": 3.7526068687438965,
      "learning_rate": 0.00020779569892473117,
      "loss": 1.1636,
      "step": 4283
    },
    {
      "epoch": 0.6706324358171571,
      "grad_norm": 2.4929449558258057,
      "learning_rate": 0.00020777126099706745,
      "loss": 0.9424,
      "step": 4284
    },
    {
      "epoch": 0.6707889793362555,
      "grad_norm": 3.227921962738037,
      "learning_rate": 0.0002077468230694037,
      "loss": 0.8243,
      "step": 4285
    },
    {
      "epoch": 0.6709455228553538,
      "grad_norm": 1.7560867071151733,
      "learning_rate": 0.00020772238514173995,
      "loss": 0.8561,
      "step": 4286
    },
    {
      "epoch": 0.6711020663744521,
      "grad_norm": 2.7342591285705566,
      "learning_rate": 0.00020769794721407623,
      "loss": 1.1016,
      "step": 4287
    },
    {
      "epoch": 0.6712586098935505,
      "grad_norm": 1.8622885942459106,
      "learning_rate": 0.00020767350928641248,
      "loss": 1.1822,
      "step": 4288
    },
    {
      "epoch": 0.6714151534126487,
      "grad_norm": 3.4083244800567627,
      "learning_rate": 0.00020764907135874876,
      "loss": 1.1509,
      "step": 4289
    },
    {
      "epoch": 0.671571696931747,
      "grad_norm": 2.9127471446990967,
      "learning_rate": 0.00020762463343108504,
      "loss": 0.8396,
      "step": 4290
    },
    {
      "epoch": 0.6717282404508453,
      "grad_norm": 2.4927799701690674,
      "learning_rate": 0.0002076001955034213,
      "loss": 1.2386,
      "step": 4291
    },
    {
      "epoch": 0.6718847839699437,
      "grad_norm": 2.8656435012817383,
      "learning_rate": 0.00020757575757575757,
      "loss": 1.2362,
      "step": 4292
    },
    {
      "epoch": 0.672041327489042,
      "grad_norm": 3.2088873386383057,
      "learning_rate": 0.00020755131964809385,
      "loss": 1.3128,
      "step": 4293
    },
    {
      "epoch": 0.6721978710081402,
      "grad_norm": 3.3077950477600098,
      "learning_rate": 0.00020752688172043007,
      "loss": 1.6932,
      "step": 4294
    },
    {
      "epoch": 0.6723544145272385,
      "grad_norm": 4.320693016052246,
      "learning_rate": 0.00020750244379276635,
      "loss": 1.554,
      "step": 4295
    },
    {
      "epoch": 0.6725109580463369,
      "grad_norm": 3.5774495601654053,
      "learning_rate": 0.00020747800586510263,
      "loss": 1.0525,
      "step": 4296
    },
    {
      "epoch": 0.6726675015654352,
      "grad_norm": 2.9771318435668945,
      "learning_rate": 0.00020745356793743888,
      "loss": 0.9957,
      "step": 4297
    },
    {
      "epoch": 0.6728240450845335,
      "grad_norm": 2.046910524368286,
      "learning_rate": 0.00020742913000977516,
      "loss": 0.903,
      "step": 4298
    },
    {
      "epoch": 0.6729805886036319,
      "grad_norm": 1.8642005920410156,
      "learning_rate": 0.00020740469208211144,
      "loss": 0.8665,
      "step": 4299
    },
    {
      "epoch": 0.6731371321227301,
      "grad_norm": 1.3706653118133545,
      "learning_rate": 0.0002073802541544477,
      "loss": 0.5211,
      "step": 4300
    },
    {
      "epoch": 0.6732936756418284,
      "grad_norm": 0.5348394513130188,
      "learning_rate": 0.00020735581622678397,
      "loss": 0.4051,
      "step": 4301
    },
    {
      "epoch": 0.6734502191609267,
      "grad_norm": 0.43285202980041504,
      "learning_rate": 0.00020733137829912024,
      "loss": 0.2932,
      "step": 4302
    },
    {
      "epoch": 0.6736067626800251,
      "grad_norm": 0.6725158095359802,
      "learning_rate": 0.00020730694037145647,
      "loss": 0.4606,
      "step": 4303
    },
    {
      "epoch": 0.6737633061991234,
      "grad_norm": 0.676227867603302,
      "learning_rate": 0.00020728250244379275,
      "loss": 0.3906,
      "step": 4304
    },
    {
      "epoch": 0.6739198497182217,
      "grad_norm": 0.6109701991081238,
      "learning_rate": 0.00020725806451612903,
      "loss": 0.5082,
      "step": 4305
    },
    {
      "epoch": 0.6740763932373199,
      "grad_norm": 0.6717798113822937,
      "learning_rate": 0.00020723362658846528,
      "loss": 0.3191,
      "step": 4306
    },
    {
      "epoch": 0.6742329367564183,
      "grad_norm": 0.5643997192382812,
      "learning_rate": 0.00020720918866080155,
      "loss": 0.5007,
      "step": 4307
    },
    {
      "epoch": 0.6743894802755166,
      "grad_norm": 0.6461310386657715,
      "learning_rate": 0.00020718475073313783,
      "loss": 0.3413,
      "step": 4308
    },
    {
      "epoch": 0.6745460237946149,
      "grad_norm": 0.7936927080154419,
      "learning_rate": 0.00020716031280547408,
      "loss": 0.4067,
      "step": 4309
    },
    {
      "epoch": 0.6747025673137133,
      "grad_norm": 0.7589799761772156,
      "learning_rate": 0.00020713587487781034,
      "loss": 0.4756,
      "step": 4310
    },
    {
      "epoch": 0.6748591108328115,
      "grad_norm": 0.9731733202934265,
      "learning_rate": 0.00020711143695014661,
      "loss": 0.3809,
      "step": 4311
    },
    {
      "epoch": 0.6750156543519098,
      "grad_norm": 0.774403989315033,
      "learning_rate": 0.00020708699902248287,
      "loss": 0.4063,
      "step": 4312
    },
    {
      "epoch": 0.6751721978710081,
      "grad_norm": 1.026542067527771,
      "learning_rate": 0.00020706256109481914,
      "loss": 0.3735,
      "step": 4313
    },
    {
      "epoch": 0.6753287413901065,
      "grad_norm": 1.497208595275879,
      "learning_rate": 0.00020703812316715542,
      "loss": 0.501,
      "step": 4314
    },
    {
      "epoch": 0.6754852849092048,
      "grad_norm": 1.3015997409820557,
      "learning_rate": 0.00020701368523949167,
      "loss": 0.607,
      "step": 4315
    },
    {
      "epoch": 0.6756418284283031,
      "grad_norm": 1.392624855041504,
      "learning_rate": 0.00020698924731182795,
      "loss": 0.7354,
      "step": 4316
    },
    {
      "epoch": 0.6757983719474013,
      "grad_norm": 1.2120251655578613,
      "learning_rate": 0.00020696480938416423,
      "loss": 0.3351,
      "step": 4317
    },
    {
      "epoch": 0.6759549154664997,
      "grad_norm": 1.3814586400985718,
      "learning_rate": 0.00020694037145650045,
      "loss": 0.5827,
      "step": 4318
    },
    {
      "epoch": 0.676111458985598,
      "grad_norm": 1.2726234197616577,
      "learning_rate": 0.00020691593352883673,
      "loss": 0.4963,
      "step": 4319
    },
    {
      "epoch": 0.6762680025046963,
      "grad_norm": 2.120060682296753,
      "learning_rate": 0.000206891495601173,
      "loss": 0.6605,
      "step": 4320
    },
    {
      "epoch": 0.6764245460237946,
      "grad_norm": 1.4398250579833984,
      "learning_rate": 0.00020686705767350926,
      "loss": 0.7302,
      "step": 4321
    },
    {
      "epoch": 0.676581089542893,
      "grad_norm": 1.0549432039260864,
      "learning_rate": 0.00020684261974584554,
      "loss": 0.5445,
      "step": 4322
    },
    {
      "epoch": 0.6767376330619912,
      "grad_norm": 1.3273248672485352,
      "learning_rate": 0.00020681818181818182,
      "loss": 0.6731,
      "step": 4323
    },
    {
      "epoch": 0.6768941765810895,
      "grad_norm": 1.9823793172836304,
      "learning_rate": 0.00020679374389051807,
      "loss": 1.0642,
      "step": 4324
    },
    {
      "epoch": 0.6770507201001879,
      "grad_norm": 1.5976402759552002,
      "learning_rate": 0.00020676930596285435,
      "loss": 0.897,
      "step": 4325
    },
    {
      "epoch": 0.6772072636192862,
      "grad_norm": 1.3136533498764038,
      "learning_rate": 0.00020674486803519063,
      "loss": 0.5817,
      "step": 4326
    },
    {
      "epoch": 0.6773638071383845,
      "grad_norm": 3.931715726852417,
      "learning_rate": 0.00020672043010752685,
      "loss": 0.67,
      "step": 4327
    },
    {
      "epoch": 0.6775203506574827,
      "grad_norm": 1.4402580261230469,
      "learning_rate": 0.00020669599217986313,
      "loss": 0.573,
      "step": 4328
    },
    {
      "epoch": 0.6776768941765811,
      "grad_norm": 3.5042459964752197,
      "learning_rate": 0.0002066715542521994,
      "loss": 0.6495,
      "step": 4329
    },
    {
      "epoch": 0.6778334376956794,
      "grad_norm": 1.4907385110855103,
      "learning_rate": 0.00020664711632453566,
      "loss": 0.689,
      "step": 4330
    },
    {
      "epoch": 0.6779899812147777,
      "grad_norm": 1.5746352672576904,
      "learning_rate": 0.00020662267839687194,
      "loss": 0.8336,
      "step": 4331
    },
    {
      "epoch": 0.678146524733876,
      "grad_norm": 1.0189085006713867,
      "learning_rate": 0.00020659824046920822,
      "loss": 0.8657,
      "step": 4332
    },
    {
      "epoch": 0.6783030682529744,
      "grad_norm": 2.78194522857666,
      "learning_rate": 0.00020657380254154447,
      "loss": 1.1317,
      "step": 4333
    },
    {
      "epoch": 0.6784596117720726,
      "grad_norm": 1.732728362083435,
      "learning_rate": 0.00020654936461388072,
      "loss": 0.7439,
      "step": 4334
    },
    {
      "epoch": 0.6786161552911709,
      "grad_norm": 3.1215643882751465,
      "learning_rate": 0.000206524926686217,
      "loss": 1.0976,
      "step": 4335
    },
    {
      "epoch": 0.6787726988102692,
      "grad_norm": 2.303900957107544,
      "learning_rate": 0.00020650048875855325,
      "loss": 0.7992,
      "step": 4336
    },
    {
      "epoch": 0.6789292423293676,
      "grad_norm": 2.1387338638305664,
      "learning_rate": 0.00020647605083088953,
      "loss": 1.4244,
      "step": 4337
    },
    {
      "epoch": 0.6790857858484659,
      "grad_norm": 2.9875400066375732,
      "learning_rate": 0.0002064516129032258,
      "loss": 1.5699,
      "step": 4338
    },
    {
      "epoch": 0.6792423293675642,
      "grad_norm": 1.6865546703338623,
      "learning_rate": 0.00020642717497556206,
      "loss": 0.9713,
      "step": 4339
    },
    {
      "epoch": 0.6793988728866625,
      "grad_norm": 2.4787464141845703,
      "learning_rate": 0.00020640273704789833,
      "loss": 1.4294,
      "step": 4340
    },
    {
      "epoch": 0.6795554164057608,
      "grad_norm": 2.0415139198303223,
      "learning_rate": 0.0002063782991202346,
      "loss": 0.7225,
      "step": 4341
    },
    {
      "epoch": 0.6797119599248591,
      "grad_norm": 3.089674234390259,
      "learning_rate": 0.00020635386119257084,
      "loss": 1.3035,
      "step": 4342
    },
    {
      "epoch": 0.6798685034439574,
      "grad_norm": 1.6210919618606567,
      "learning_rate": 0.00020632942326490711,
      "loss": 1.1855,
      "step": 4343
    },
    {
      "epoch": 0.6800250469630558,
      "grad_norm": 3.708653450012207,
      "learning_rate": 0.0002063049853372434,
      "loss": 1.2739,
      "step": 4344
    },
    {
      "epoch": 0.680181590482154,
      "grad_norm": 2.1804747581481934,
      "learning_rate": 0.00020628054740957964,
      "loss": 1.0516,
      "step": 4345
    },
    {
      "epoch": 0.6803381340012523,
      "grad_norm": 2.3872623443603516,
      "learning_rate": 0.00020625610948191592,
      "loss": 0.8218,
      "step": 4346
    },
    {
      "epoch": 0.6804946775203506,
      "grad_norm": 2.274266242980957,
      "learning_rate": 0.00020623167155425217,
      "loss": 0.6364,
      "step": 4347
    },
    {
      "epoch": 0.680651221039449,
      "grad_norm": 4.519956588745117,
      "learning_rate": 0.00020620723362658845,
      "loss": 0.6342,
      "step": 4348
    },
    {
      "epoch": 0.6808077645585473,
      "grad_norm": 1.9020459651947021,
      "learning_rate": 0.00020618279569892473,
      "loss": 1.1602,
      "step": 4349
    },
    {
      "epoch": 0.6809643080776456,
      "grad_norm": 2.6886696815490723,
      "learning_rate": 0.00020615835777126095,
      "loss": 1.2329,
      "step": 4350
    },
    {
      "epoch": 0.6811208515967438,
      "grad_norm": 0.6094191670417786,
      "learning_rate": 0.00020613391984359723,
      "loss": 0.2924,
      "step": 4351
    },
    {
      "epoch": 0.6812773951158422,
      "grad_norm": 0.5536388158798218,
      "learning_rate": 0.0002061094819159335,
      "loss": 0.3504,
      "step": 4352
    },
    {
      "epoch": 0.6814339386349405,
      "grad_norm": 0.5976225137710571,
      "learning_rate": 0.00020608504398826976,
      "loss": 0.3809,
      "step": 4353
    },
    {
      "epoch": 0.6815904821540388,
      "grad_norm": 0.694050133228302,
      "learning_rate": 0.00020606060606060604,
      "loss": 0.4363,
      "step": 4354
    },
    {
      "epoch": 0.6817470256731372,
      "grad_norm": 0.606235682964325,
      "learning_rate": 0.00020603616813294232,
      "loss": 0.2675,
      "step": 4355
    },
    {
      "epoch": 0.6819035691922355,
      "grad_norm": 2.016716957092285,
      "learning_rate": 0.00020601173020527857,
      "loss": 0.4803,
      "step": 4356
    },
    {
      "epoch": 0.6820601127113337,
      "grad_norm": 0.9147260785102844,
      "learning_rate": 0.00020598729227761482,
      "loss": 0.5548,
      "step": 4357
    },
    {
      "epoch": 0.682216656230432,
      "grad_norm": 0.6617628931999207,
      "learning_rate": 0.0002059628543499511,
      "loss": 0.4761,
      "step": 4358
    },
    {
      "epoch": 0.6823731997495304,
      "grad_norm": 0.9042291045188904,
      "learning_rate": 0.00020593841642228735,
      "loss": 0.3996,
      "step": 4359
    },
    {
      "epoch": 0.6825297432686287,
      "grad_norm": 1.5656142234802246,
      "learning_rate": 0.00020591397849462363,
      "loss": 0.6516,
      "step": 4360
    },
    {
      "epoch": 0.682686286787727,
      "grad_norm": 0.7753491401672363,
      "learning_rate": 0.0002058895405669599,
      "loss": 0.3016,
      "step": 4361
    },
    {
      "epoch": 0.6828428303068252,
      "grad_norm": 0.5845422148704529,
      "learning_rate": 0.00020586510263929616,
      "loss": 0.2986,
      "step": 4362
    },
    {
      "epoch": 0.6829993738259236,
      "grad_norm": 2.0732033252716064,
      "learning_rate": 0.00020584066471163244,
      "loss": 0.7511,
      "step": 4363
    },
    {
      "epoch": 0.6831559173450219,
      "grad_norm": 1.0156924724578857,
      "learning_rate": 0.00020581622678396872,
      "loss": 0.5104,
      "step": 4364
    },
    {
      "epoch": 0.6833124608641202,
      "grad_norm": 1.2725284099578857,
      "learning_rate": 0.00020579178885630494,
      "loss": 0.2841,
      "step": 4365
    },
    {
      "epoch": 0.6834690043832186,
      "grad_norm": 0.7375205159187317,
      "learning_rate": 0.00020576735092864122,
      "loss": 0.3947,
      "step": 4366
    },
    {
      "epoch": 0.6836255479023169,
      "grad_norm": 1.245803952217102,
      "learning_rate": 0.0002057429130009775,
      "loss": 0.5679,
      "step": 4367
    },
    {
      "epoch": 0.6837820914214151,
      "grad_norm": 1.113433599472046,
      "learning_rate": 0.00020571847507331375,
      "loss": 0.3995,
      "step": 4368
    },
    {
      "epoch": 0.6839386349405134,
      "grad_norm": 1.771336555480957,
      "learning_rate": 0.00020569403714565003,
      "loss": 0.5824,
      "step": 4369
    },
    {
      "epoch": 0.6840951784596118,
      "grad_norm": 0.9144507050514221,
      "learning_rate": 0.0002056695992179863,
      "loss": 0.576,
      "step": 4370
    },
    {
      "epoch": 0.6842517219787101,
      "grad_norm": 1.0565403699874878,
      "learning_rate": 0.00020564516129032256,
      "loss": 0.4776,
      "step": 4371
    },
    {
      "epoch": 0.6844082654978084,
      "grad_norm": 2.380849838256836,
      "learning_rate": 0.00020562072336265883,
      "loss": 0.9487,
      "step": 4372
    },
    {
      "epoch": 0.6845648090169068,
      "grad_norm": 1.3692196607589722,
      "learning_rate": 0.0002055962854349951,
      "loss": 0.5143,
      "step": 4373
    },
    {
      "epoch": 0.684721352536005,
      "grad_norm": 2.105653762817383,
      "learning_rate": 0.00020557184750733134,
      "loss": 0.5443,
      "step": 4374
    },
    {
      "epoch": 0.6848778960551033,
      "grad_norm": 1.329337239265442,
      "learning_rate": 0.00020554740957966762,
      "loss": 0.7253,
      "step": 4375
    },
    {
      "epoch": 0.6850344395742016,
      "grad_norm": 0.9854122400283813,
      "learning_rate": 0.0002055229716520039,
      "loss": 0.5775,
      "step": 4376
    },
    {
      "epoch": 0.6851909830933,
      "grad_norm": 1.4902323484420776,
      "learning_rate": 0.00020549853372434014,
      "loss": 0.6427,
      "step": 4377
    },
    {
      "epoch": 0.6853475266123983,
      "grad_norm": 2.347221851348877,
      "learning_rate": 0.00020547409579667642,
      "loss": 0.8651,
      "step": 4378
    },
    {
      "epoch": 0.6855040701314965,
      "grad_norm": 1.5970897674560547,
      "learning_rate": 0.0002054496578690127,
      "loss": 0.8242,
      "step": 4379
    },
    {
      "epoch": 0.6856606136505948,
      "grad_norm": 1.8963264226913452,
      "learning_rate": 0.00020542521994134895,
      "loss": 0.6639,
      "step": 4380
    },
    {
      "epoch": 0.6858171571696932,
      "grad_norm": 1.861736536026001,
      "learning_rate": 0.0002054007820136852,
      "loss": 0.882,
      "step": 4381
    },
    {
      "epoch": 0.6859737006887915,
      "grad_norm": 2.6222732067108154,
      "learning_rate": 0.00020537634408602148,
      "loss": 0.5547,
      "step": 4382
    },
    {
      "epoch": 0.6861302442078898,
      "grad_norm": 7.720530033111572,
      "learning_rate": 0.00020535190615835773,
      "loss": 0.9057,
      "step": 4383
    },
    {
      "epoch": 0.6862867877269881,
      "grad_norm": 2.147437810897827,
      "learning_rate": 0.000205327468230694,
      "loss": 1.5026,
      "step": 4384
    },
    {
      "epoch": 0.6864433312460864,
      "grad_norm": 1.9895732402801514,
      "learning_rate": 0.0002053030303030303,
      "loss": 1.0872,
      "step": 4385
    },
    {
      "epoch": 0.6865998747651847,
      "grad_norm": 3.379164218902588,
      "learning_rate": 0.00020527859237536654,
      "loss": 1.4365,
      "step": 4386
    },
    {
      "epoch": 0.686756418284283,
      "grad_norm": 3.2199456691741943,
      "learning_rate": 0.00020525415444770282,
      "loss": 0.855,
      "step": 4387
    },
    {
      "epoch": 0.6869129618033814,
      "grad_norm": 4.4102911949157715,
      "learning_rate": 0.0002052297165200391,
      "loss": 1.2534,
      "step": 4388
    },
    {
      "epoch": 0.6870695053224797,
      "grad_norm": 1.3978711366653442,
      "learning_rate": 0.00020520527859237532,
      "loss": 0.9027,
      "step": 4389
    },
    {
      "epoch": 0.687226048841578,
      "grad_norm": 2.3685648441314697,
      "learning_rate": 0.0002051808406647116,
      "loss": 0.8522,
      "step": 4390
    },
    {
      "epoch": 0.6873825923606762,
      "grad_norm": 2.080855369567871,
      "learning_rate": 0.00020515640273704788,
      "loss": 1.2799,
      "step": 4391
    },
    {
      "epoch": 0.6875391358797746,
      "grad_norm": 2.1453280448913574,
      "learning_rate": 0.00020513196480938413,
      "loss": 1.1435,
      "step": 4392
    },
    {
      "epoch": 0.6876956793988729,
      "grad_norm": 2.5470387935638428,
      "learning_rate": 0.0002051075268817204,
      "loss": 1.9692,
      "step": 4393
    },
    {
      "epoch": 0.6878522229179712,
      "grad_norm": 2.8885648250579834,
      "learning_rate": 0.0002050830889540567,
      "loss": 1.7749,
      "step": 4394
    },
    {
      "epoch": 0.6880087664370695,
      "grad_norm": 2.181378126144409,
      "learning_rate": 0.00020505865102639294,
      "loss": 1.5839,
      "step": 4395
    },
    {
      "epoch": 0.6881653099561679,
      "grad_norm": 4.625649452209473,
      "learning_rate": 0.00020503421309872922,
      "loss": 1.4834,
      "step": 4396
    },
    {
      "epoch": 0.6883218534752661,
      "grad_norm": 2.8574166297912598,
      "learning_rate": 0.0002050097751710655,
      "loss": 1.4019,
      "step": 4397
    },
    {
      "epoch": 0.6884783969943644,
      "grad_norm": 2.1024327278137207,
      "learning_rate": 0.00020498533724340172,
      "loss": 1.1237,
      "step": 4398
    },
    {
      "epoch": 0.6886349405134627,
      "grad_norm": 1.9445034265518188,
      "learning_rate": 0.000204960899315738,
      "loss": 0.9635,
      "step": 4399
    },
    {
      "epoch": 0.6887914840325611,
      "grad_norm": 2.038120746612549,
      "learning_rate": 0.00020493646138807428,
      "loss": 1.1126,
      "step": 4400
    },
    {
      "epoch": 0.6889480275516594,
      "grad_norm": 0.9593400955200195,
      "learning_rate": 0.00020491202346041053,
      "loss": 0.8911,
      "step": 4401
    },
    {
      "epoch": 0.6891045710707576,
      "grad_norm": 0.4626724421977997,
      "learning_rate": 0.0002048875855327468,
      "loss": 0.306,
      "step": 4402
    },
    {
      "epoch": 0.689261114589856,
      "grad_norm": 0.5017557144165039,
      "learning_rate": 0.00020486314760508308,
      "loss": 0.387,
      "step": 4403
    },
    {
      "epoch": 0.6894176581089543,
      "grad_norm": 0.7594949007034302,
      "learning_rate": 0.0002048387096774193,
      "loss": 0.2952,
      "step": 4404
    },
    {
      "epoch": 0.6895742016280526,
      "grad_norm": 0.5855520367622375,
      "learning_rate": 0.00020481427174975559,
      "loss": 0.3796,
      "step": 4405
    },
    {
      "epoch": 0.6897307451471509,
      "grad_norm": 0.7784630656242371,
      "learning_rate": 0.00020478983382209186,
      "loss": 0.3505,
      "step": 4406
    },
    {
      "epoch": 0.6898872886662493,
      "grad_norm": 0.7261280417442322,
      "learning_rate": 0.00020476539589442812,
      "loss": 0.3887,
      "step": 4407
    },
    {
      "epoch": 0.6900438321853475,
      "grad_norm": 0.5250700116157532,
      "learning_rate": 0.0002047409579667644,
      "loss": 0.3455,
      "step": 4408
    },
    {
      "epoch": 0.6902003757044458,
      "grad_norm": 1.4963864088058472,
      "learning_rate": 0.00020471652003910067,
      "loss": 0.4713,
      "step": 4409
    },
    {
      "epoch": 0.6903569192235441,
      "grad_norm": 0.7633784413337708,
      "learning_rate": 0.00020469208211143692,
      "loss": 0.3757,
      "step": 4410
    },
    {
      "epoch": 0.6905134627426425,
      "grad_norm": 0.6645981669425964,
      "learning_rate": 0.0002046676441837732,
      "loss": 0.3691,
      "step": 4411
    },
    {
      "epoch": 0.6906700062617408,
      "grad_norm": 0.977664589881897,
      "learning_rate": 0.00020464320625610948,
      "loss": 0.4197,
      "step": 4412
    },
    {
      "epoch": 0.6908265497808391,
      "grad_norm": 0.827433168888092,
      "learning_rate": 0.0002046187683284457,
      "loss": 0.4272,
      "step": 4413
    },
    {
      "epoch": 0.6909830932999373,
      "grad_norm": 1.0397074222564697,
      "learning_rate": 0.00020459433040078198,
      "loss": 0.546,
      "step": 4414
    },
    {
      "epoch": 0.6911396368190357,
      "grad_norm": 0.9832563400268555,
      "learning_rate": 0.00020456989247311826,
      "loss": 0.4227,
      "step": 4415
    },
    {
      "epoch": 0.691296180338134,
      "grad_norm": 1.1884939670562744,
      "learning_rate": 0.0002045454545454545,
      "loss": 0.4485,
      "step": 4416
    },
    {
      "epoch": 0.6914527238572323,
      "grad_norm": 1.7477940320968628,
      "learning_rate": 0.0002045210166177908,
      "loss": 0.5891,
      "step": 4417
    },
    {
      "epoch": 0.6916092673763307,
      "grad_norm": 1.1390234231948853,
      "learning_rate": 0.00020449657869012707,
      "loss": 0.4141,
      "step": 4418
    },
    {
      "epoch": 0.6917658108954289,
      "grad_norm": 2.4235641956329346,
      "learning_rate": 0.00020447214076246332,
      "loss": 0.5264,
      "step": 4419
    },
    {
      "epoch": 0.6919223544145272,
      "grad_norm": 2.1330161094665527,
      "learning_rate": 0.0002044477028347996,
      "loss": 0.7413,
      "step": 4420
    },
    {
      "epoch": 0.6920788979336255,
      "grad_norm": 1.950913429260254,
      "learning_rate": 0.00020442326490713588,
      "loss": 0.5941,
      "step": 4421
    },
    {
      "epoch": 0.6922354414527239,
      "grad_norm": 1.4877159595489502,
      "learning_rate": 0.0002043988269794721,
      "loss": 0.6298,
      "step": 4422
    },
    {
      "epoch": 0.6923919849718222,
      "grad_norm": 2.8081557750701904,
      "learning_rate": 0.00020437438905180838,
      "loss": 0.6594,
      "step": 4423
    },
    {
      "epoch": 0.6925485284909205,
      "grad_norm": 3.2152159214019775,
      "learning_rate": 0.00020434995112414466,
      "loss": 0.8644,
      "step": 4424
    },
    {
      "epoch": 0.6927050720100187,
      "grad_norm": 1.4229108095169067,
      "learning_rate": 0.0002043255131964809,
      "loss": 0.5714,
      "step": 4425
    },
    {
      "epoch": 0.6928616155291171,
      "grad_norm": 2.301750421524048,
      "learning_rate": 0.0002043010752688172,
      "loss": 0.7934,
      "step": 4426
    },
    {
      "epoch": 0.6930181590482154,
      "grad_norm": 1.235784888267517,
      "learning_rate": 0.00020427663734115347,
      "loss": 0.7241,
      "step": 4427
    },
    {
      "epoch": 0.6931747025673137,
      "grad_norm": 1.9009183645248413,
      "learning_rate": 0.0002042521994134897,
      "loss": 1.0616,
      "step": 4428
    },
    {
      "epoch": 0.6933312460864121,
      "grad_norm": 1.6725027561187744,
      "learning_rate": 0.00020422776148582597,
      "loss": 0.6872,
      "step": 4429
    },
    {
      "epoch": 0.6934877896055104,
      "grad_norm": 2.6750926971435547,
      "learning_rate": 0.00020420332355816225,
      "loss": 1.1431,
      "step": 4430
    },
    {
      "epoch": 0.6936443331246086,
      "grad_norm": 3.524231195449829,
      "learning_rate": 0.0002041788856304985,
      "loss": 1.205,
      "step": 4431
    },
    {
      "epoch": 0.6938008766437069,
      "grad_norm": 1.7456165552139282,
      "learning_rate": 0.00020415444770283478,
      "loss": 0.9005,
      "step": 4432
    },
    {
      "epoch": 0.6939574201628053,
      "grad_norm": 1.8678324222564697,
      "learning_rate": 0.00020413000977517106,
      "loss": 0.6964,
      "step": 4433
    },
    {
      "epoch": 0.6941139636819036,
      "grad_norm": 2.4247958660125732,
      "learning_rate": 0.0002041055718475073,
      "loss": 1.397,
      "step": 4434
    },
    {
      "epoch": 0.6942705072010019,
      "grad_norm": 3.6125667095184326,
      "learning_rate": 0.00020408113391984358,
      "loss": 0.9847,
      "step": 4435
    },
    {
      "epoch": 0.6944270507201001,
      "grad_norm": 2.9348018169403076,
      "learning_rate": 0.00020405669599217986,
      "loss": 1.3243,
      "step": 4436
    },
    {
      "epoch": 0.6945835942391985,
      "grad_norm": 1.7903168201446533,
      "learning_rate": 0.0002040322580645161,
      "loss": 1.0472,
      "step": 4437
    },
    {
      "epoch": 0.6947401377582968,
      "grad_norm": 1.7623295783996582,
      "learning_rate": 0.00020400782013685237,
      "loss": 0.8571,
      "step": 4438
    },
    {
      "epoch": 0.6948966812773951,
      "grad_norm": 1.318037986755371,
      "learning_rate": 0.00020398338220918864,
      "loss": 0.6425,
      "step": 4439
    },
    {
      "epoch": 0.6950532247964935,
      "grad_norm": 2.563493013381958,
      "learning_rate": 0.0002039589442815249,
      "loss": 1.0581,
      "step": 4440
    },
    {
      "epoch": 0.6952097683155918,
      "grad_norm": 2.8916120529174805,
      "learning_rate": 0.00020393450635386117,
      "loss": 1.1032,
      "step": 4441
    },
    {
      "epoch": 0.69536631183469,
      "grad_norm": 2.9763169288635254,
      "learning_rate": 0.00020391006842619745,
      "loss": 1.2579,
      "step": 4442
    },
    {
      "epoch": 0.6955228553537883,
      "grad_norm": 1.8792319297790527,
      "learning_rate": 0.0002038856304985337,
      "loss": 1.4755,
      "step": 4443
    },
    {
      "epoch": 0.6956793988728867,
      "grad_norm": 2.4958150386810303,
      "learning_rate": 0.00020386119257086998,
      "loss": 1.6926,
      "step": 4444
    },
    {
      "epoch": 0.695835942391985,
      "grad_norm": 3.446523904800415,
      "learning_rate": 0.00020383675464320623,
      "loss": 0.8777,
      "step": 4445
    },
    {
      "epoch": 0.6959924859110833,
      "grad_norm": 1.1154353618621826,
      "learning_rate": 0.00020381231671554248,
      "loss": 0.5229,
      "step": 4446
    },
    {
      "epoch": 0.6961490294301816,
      "grad_norm": 1.8287882804870605,
      "learning_rate": 0.00020378787878787876,
      "loss": 0.7689,
      "step": 4447
    },
    {
      "epoch": 0.6963055729492799,
      "grad_norm": 2.325324535369873,
      "learning_rate": 0.00020376344086021504,
      "loss": 0.6508,
      "step": 4448
    },
    {
      "epoch": 0.6964621164683782,
      "grad_norm": 1.61253821849823,
      "learning_rate": 0.0002037390029325513,
      "loss": 0.4842,
      "step": 4449
    },
    {
      "epoch": 0.6966186599874765,
      "grad_norm": 3.248868703842163,
      "learning_rate": 0.00020371456500488757,
      "loss": 1.2176,
      "step": 4450
    },
    {
      "epoch": 0.6967752035065748,
      "grad_norm": 0.74549800157547,
      "learning_rate": 0.00020369012707722385,
      "loss": 0.4581,
      "step": 4451
    },
    {
      "epoch": 0.6969317470256732,
      "grad_norm": 0.5620818734169006,
      "learning_rate": 0.00020366568914956007,
      "loss": 0.2706,
      "step": 4452
    },
    {
      "epoch": 0.6970882905447714,
      "grad_norm": 0.6187224984169006,
      "learning_rate": 0.00020364125122189635,
      "loss": 0.2828,
      "step": 4453
    },
    {
      "epoch": 0.6972448340638697,
      "grad_norm": 1.0300192832946777,
      "learning_rate": 0.00020361681329423263,
      "loss": 0.5353,
      "step": 4454
    },
    {
      "epoch": 0.697401377582968,
      "grad_norm": 0.5869678854942322,
      "learning_rate": 0.00020359237536656888,
      "loss": 0.3123,
      "step": 4455
    },
    {
      "epoch": 0.6975579211020664,
      "grad_norm": 0.944622814655304,
      "learning_rate": 0.00020356793743890516,
      "loss": 0.4034,
      "step": 4456
    },
    {
      "epoch": 0.6977144646211647,
      "grad_norm": 0.7487533688545227,
      "learning_rate": 0.00020354349951124144,
      "loss": 0.5087,
      "step": 4457
    },
    {
      "epoch": 0.697871008140263,
      "grad_norm": 0.7087098956108093,
      "learning_rate": 0.0002035190615835777,
      "loss": 0.2688,
      "step": 4458
    },
    {
      "epoch": 0.6980275516593613,
      "grad_norm": 1.030664324760437,
      "learning_rate": 0.00020349462365591397,
      "loss": 0.4653,
      "step": 4459
    },
    {
      "epoch": 0.6981840951784596,
      "grad_norm": 1.0674512386322021,
      "learning_rate": 0.00020347018572825025,
      "loss": 0.4403,
      "step": 4460
    },
    {
      "epoch": 0.6983406386975579,
      "grad_norm": 0.9911973476409912,
      "learning_rate": 0.00020344574780058647,
      "loss": 0.322,
      "step": 4461
    },
    {
      "epoch": 0.6984971822166562,
      "grad_norm": 2.1932692527770996,
      "learning_rate": 0.00020342130987292275,
      "loss": 0.4332,
      "step": 4462
    },
    {
      "epoch": 0.6986537257357546,
      "grad_norm": 0.7170542478561401,
      "learning_rate": 0.00020339687194525903,
      "loss": 0.3078,
      "step": 4463
    },
    {
      "epoch": 0.6988102692548529,
      "grad_norm": 1.0266473293304443,
      "learning_rate": 0.00020337243401759528,
      "loss": 0.3968,
      "step": 4464
    },
    {
      "epoch": 0.6989668127739511,
      "grad_norm": 0.8614751100540161,
      "learning_rate": 0.00020334799608993156,
      "loss": 0.2991,
      "step": 4465
    },
    {
      "epoch": 0.6991233562930494,
      "grad_norm": 0.9565255641937256,
      "learning_rate": 0.00020332355816226783,
      "loss": 0.6496,
      "step": 4466
    },
    {
      "epoch": 0.6992798998121478,
      "grad_norm": 1.1029607057571411,
      "learning_rate": 0.00020329912023460409,
      "loss": 0.569,
      "step": 4467
    },
    {
      "epoch": 0.6994364433312461,
      "grad_norm": 1.1075776815414429,
      "learning_rate": 0.00020327468230694036,
      "loss": 0.5768,
      "step": 4468
    },
    {
      "epoch": 0.6995929868503444,
      "grad_norm": 1.6944200992584229,
      "learning_rate": 0.00020325024437927661,
      "loss": 0.9883,
      "step": 4469
    },
    {
      "epoch": 0.6997495303694427,
      "grad_norm": 1.856938362121582,
      "learning_rate": 0.00020322580645161287,
      "loss": 0.6218,
      "step": 4470
    },
    {
      "epoch": 0.699906073888541,
      "grad_norm": 1.5838960409164429,
      "learning_rate": 0.00020320136852394914,
      "loss": 0.6972,
      "step": 4471
    },
    {
      "epoch": 0.7000626174076393,
      "grad_norm": 1.5133699178695679,
      "learning_rate": 0.00020317693059628542,
      "loss": 0.8786,
      "step": 4472
    },
    {
      "epoch": 0.7002191609267376,
      "grad_norm": 0.9104510545730591,
      "learning_rate": 0.00020315249266862167,
      "loss": 0.4001,
      "step": 4473
    },
    {
      "epoch": 0.700375704445836,
      "grad_norm": 1.3532549142837524,
      "learning_rate": 0.00020312805474095795,
      "loss": 0.4596,
      "step": 4474
    },
    {
      "epoch": 0.7005322479649343,
      "grad_norm": 0.9898942708969116,
      "learning_rate": 0.00020310361681329423,
      "loss": 0.6143,
      "step": 4475
    },
    {
      "epoch": 0.7006887914840325,
      "grad_norm": 1.7784004211425781,
      "learning_rate": 0.00020307917888563045,
      "loss": 0.9761,
      "step": 4476
    },
    {
      "epoch": 0.7008453350031308,
      "grad_norm": 2.9220640659332275,
      "learning_rate": 0.00020305474095796673,
      "loss": 1.0946,
      "step": 4477
    },
    {
      "epoch": 0.7010018785222292,
      "grad_norm": 1.6523720026016235,
      "learning_rate": 0.000203030303030303,
      "loss": 0.6745,
      "step": 4478
    },
    {
      "epoch": 0.7011584220413275,
      "grad_norm": 2.706207275390625,
      "learning_rate": 0.00020300586510263926,
      "loss": 0.5833,
      "step": 4479
    },
    {
      "epoch": 0.7013149655604258,
      "grad_norm": 1.879623532295227,
      "learning_rate": 0.00020298142717497554,
      "loss": 0.752,
      "step": 4480
    },
    {
      "epoch": 0.7014715090795242,
      "grad_norm": 2.1002285480499268,
      "learning_rate": 0.00020295698924731182,
      "loss": 0.7268,
      "step": 4481
    },
    {
      "epoch": 0.7016280525986224,
      "grad_norm": 2.6373350620269775,
      "learning_rate": 0.00020293255131964807,
      "loss": 0.9332,
      "step": 4482
    },
    {
      "epoch": 0.7017845961177207,
      "grad_norm": 3.0177090167999268,
      "learning_rate": 0.00020290811339198435,
      "loss": 1.4244,
      "step": 4483
    },
    {
      "epoch": 0.701941139636819,
      "grad_norm": 2.546877384185791,
      "learning_rate": 0.00020288367546432063,
      "loss": 0.6665,
      "step": 4484
    },
    {
      "epoch": 0.7020976831559174,
      "grad_norm": 1.9773567914962769,
      "learning_rate": 0.00020285923753665685,
      "loss": 1.2049,
      "step": 4485
    },
    {
      "epoch": 0.7022542266750157,
      "grad_norm": 3.617976188659668,
      "learning_rate": 0.00020283479960899313,
      "loss": 1.2953,
      "step": 4486
    },
    {
      "epoch": 0.7024107701941139,
      "grad_norm": 2.691190719604492,
      "learning_rate": 0.0002028103616813294,
      "loss": 0.722,
      "step": 4487
    },
    {
      "epoch": 0.7025673137132122,
      "grad_norm": 2.3003101348876953,
      "learning_rate": 0.00020278592375366566,
      "loss": 0.8476,
      "step": 4488
    },
    {
      "epoch": 0.7027238572323106,
      "grad_norm": 2.354837656021118,
      "learning_rate": 0.00020276148582600194,
      "loss": 0.6616,
      "step": 4489
    },
    {
      "epoch": 0.7028804007514089,
      "grad_norm": 2.914757013320923,
      "learning_rate": 0.00020273704789833822,
      "loss": 1.2246,
      "step": 4490
    },
    {
      "epoch": 0.7030369442705072,
      "grad_norm": 2.373718500137329,
      "learning_rate": 0.00020271260997067447,
      "loss": 1.2729,
      "step": 4491
    },
    {
      "epoch": 0.7031934877896056,
      "grad_norm": 2.7361257076263428,
      "learning_rate": 0.00020268817204301075,
      "loss": 1.2303,
      "step": 4492
    },
    {
      "epoch": 0.7033500313087038,
      "grad_norm": 1.035384178161621,
      "learning_rate": 0.000202663734115347,
      "loss": 0.6874,
      "step": 4493
    },
    {
      "epoch": 0.7035065748278021,
      "grad_norm": 3.0830559730529785,
      "learning_rate": 0.00020263929618768325,
      "loss": 1.0254,
      "step": 4494
    },
    {
      "epoch": 0.7036631183469004,
      "grad_norm": 5.6683478355407715,
      "learning_rate": 0.00020261485826001953,
      "loss": 0.7345,
      "step": 4495
    },
    {
      "epoch": 0.7038196618659988,
      "grad_norm": 2.4761593341827393,
      "learning_rate": 0.0002025904203323558,
      "loss": 0.9819,
      "step": 4496
    },
    {
      "epoch": 0.7039762053850971,
      "grad_norm": 4.851938724517822,
      "learning_rate": 0.00020256598240469206,
      "loss": 0.9002,
      "step": 4497
    },
    {
      "epoch": 0.7041327489041954,
      "grad_norm": 1.542232871055603,
      "learning_rate": 0.00020254154447702833,
      "loss": 0.7226,
      "step": 4498
    },
    {
      "epoch": 0.7042892924232936,
      "grad_norm": 1.4405412673950195,
      "learning_rate": 0.0002025171065493646,
      "loss": 1.0137,
      "step": 4499
    },
    {
      "epoch": 0.704445835942392,
      "grad_norm": 1.6948432922363281,
      "learning_rate": 0.00020249266862170084,
      "loss": 0.725,
      "step": 4500
    },
    {
      "epoch": 0.7046023794614903,
      "grad_norm": 0.5752367973327637,
      "learning_rate": 0.00020246823069403712,
      "loss": 0.2645,
      "step": 4501
    },
    {
      "epoch": 0.7047589229805886,
      "grad_norm": 0.5872207880020142,
      "learning_rate": 0.0002024437927663734,
      "loss": 0.3297,
      "step": 4502
    },
    {
      "epoch": 0.704915466499687,
      "grad_norm": 0.5640816688537598,
      "learning_rate": 0.00020241935483870965,
      "loss": 0.379,
      "step": 4503
    },
    {
      "epoch": 0.7050720100187852,
      "grad_norm": 0.9761353731155396,
      "learning_rate": 0.00020239491691104592,
      "loss": 0.3643,
      "step": 4504
    },
    {
      "epoch": 0.7052285535378835,
      "grad_norm": 0.6274861693382263,
      "learning_rate": 0.0002023704789833822,
      "loss": 0.3448,
      "step": 4505
    },
    {
      "epoch": 0.7053850970569818,
      "grad_norm": 0.8410617709159851,
      "learning_rate": 0.00020234604105571845,
      "loss": 0.3694,
      "step": 4506
    },
    {
      "epoch": 0.7055416405760802,
      "grad_norm": 0.7062875032424927,
      "learning_rate": 0.00020232160312805473,
      "loss": 0.3699,
      "step": 4507
    },
    {
      "epoch": 0.7056981840951785,
      "grad_norm": 0.783191442489624,
      "learning_rate": 0.000202297165200391,
      "loss": 0.478,
      "step": 4508
    },
    {
      "epoch": 0.7058547276142768,
      "grad_norm": 0.8087450265884399,
      "learning_rate": 0.00020227272727272723,
      "loss": 0.3735,
      "step": 4509
    },
    {
      "epoch": 0.706011271133375,
      "grad_norm": 0.7015910148620605,
      "learning_rate": 0.0002022482893450635,
      "loss": 0.4296,
      "step": 4510
    },
    {
      "epoch": 0.7061678146524734,
      "grad_norm": 1.2119005918502808,
      "learning_rate": 0.0002022238514173998,
      "loss": 0.4157,
      "step": 4511
    },
    {
      "epoch": 0.7063243581715717,
      "grad_norm": 0.7165002226829529,
      "learning_rate": 0.00020219941348973604,
      "loss": 0.3681,
      "step": 4512
    },
    {
      "epoch": 0.70648090169067,
      "grad_norm": 0.9335975050926208,
      "learning_rate": 0.00020217497556207232,
      "loss": 0.472,
      "step": 4513
    },
    {
      "epoch": 0.7066374452097683,
      "grad_norm": 1.0434401035308838,
      "learning_rate": 0.0002021505376344086,
      "loss": 0.6244,
      "step": 4514
    },
    {
      "epoch": 0.7067939887288667,
      "grad_norm": 1.2325775623321533,
      "learning_rate": 0.00020212609970674485,
      "loss": 0.5675,
      "step": 4515
    },
    {
      "epoch": 0.7069505322479649,
      "grad_norm": 1.6948950290679932,
      "learning_rate": 0.0002021016617790811,
      "loss": 0.6639,
      "step": 4516
    },
    {
      "epoch": 0.7071070757670632,
      "grad_norm": 1.0508677959442139,
      "learning_rate": 0.00020207722385141738,
      "loss": 0.6657,
      "step": 4517
    },
    {
      "epoch": 0.7072636192861615,
      "grad_norm": 3.625051259994507,
      "learning_rate": 0.00020205278592375363,
      "loss": 0.5776,
      "step": 4518
    },
    {
      "epoch": 0.7074201628052599,
      "grad_norm": 3.57965087890625,
      "learning_rate": 0.0002020283479960899,
      "loss": 0.7766,
      "step": 4519
    },
    {
      "epoch": 0.7075767063243582,
      "grad_norm": 3.801496744155884,
      "learning_rate": 0.0002020039100684262,
      "loss": 1.1872,
      "step": 4520
    },
    {
      "epoch": 0.7077332498434565,
      "grad_norm": 1.3635119199752808,
      "learning_rate": 0.00020197947214076244,
      "loss": 0.9458,
      "step": 4521
    },
    {
      "epoch": 0.7078897933625548,
      "grad_norm": 3.277301788330078,
      "learning_rate": 0.00020195503421309872,
      "loss": 0.8063,
      "step": 4522
    },
    {
      "epoch": 0.7080463368816531,
      "grad_norm": 2.7568883895874023,
      "learning_rate": 0.000201930596285435,
      "loss": 0.552,
      "step": 4523
    },
    {
      "epoch": 0.7082028804007514,
      "grad_norm": 1.8860586881637573,
      "learning_rate": 0.00020190615835777122,
      "loss": 0.6122,
      "step": 4524
    },
    {
      "epoch": 0.7083594239198497,
      "grad_norm": 1.4238359928131104,
      "learning_rate": 0.0002018817204301075,
      "loss": 0.7285,
      "step": 4525
    },
    {
      "epoch": 0.7085159674389481,
      "grad_norm": 0.9961578845977783,
      "learning_rate": 0.00020185728250244378,
      "loss": 0.4278,
      "step": 4526
    },
    {
      "epoch": 0.7086725109580463,
      "grad_norm": 1.8736709356307983,
      "learning_rate": 0.00020183284457478003,
      "loss": 1.2713,
      "step": 4527
    },
    {
      "epoch": 0.7088290544771446,
      "grad_norm": 1.6112703084945679,
      "learning_rate": 0.0002018084066471163,
      "loss": 0.7608,
      "step": 4528
    },
    {
      "epoch": 0.7089855979962429,
      "grad_norm": 1.8438599109649658,
      "learning_rate": 0.00020178396871945258,
      "loss": 0.6969,
      "step": 4529
    },
    {
      "epoch": 0.7091421415153413,
      "grad_norm": 2.1331958770751953,
      "learning_rate": 0.00020175953079178884,
      "loss": 0.7564,
      "step": 4530
    },
    {
      "epoch": 0.7092986850344396,
      "grad_norm": 1.4046313762664795,
      "learning_rate": 0.00020173509286412511,
      "loss": 0.5279,
      "step": 4531
    },
    {
      "epoch": 0.7094552285535379,
      "grad_norm": 2.5683090686798096,
      "learning_rate": 0.0002017106549364614,
      "loss": 1.1941,
      "step": 4532
    },
    {
      "epoch": 0.7096117720726361,
      "grad_norm": 2.2568979263305664,
      "learning_rate": 0.00020168621700879762,
      "loss": 1.0081,
      "step": 4533
    },
    {
      "epoch": 0.7097683155917345,
      "grad_norm": 2.451312780380249,
      "learning_rate": 0.0002016617790811339,
      "loss": 1.0977,
      "step": 4534
    },
    {
      "epoch": 0.7099248591108328,
      "grad_norm": 2.566115379333496,
      "learning_rate": 0.00020163734115347017,
      "loss": 0.9091,
      "step": 4535
    },
    {
      "epoch": 0.7100814026299311,
      "grad_norm": 1.651110053062439,
      "learning_rate": 0.00020161290322580642,
      "loss": 1.1188,
      "step": 4536
    },
    {
      "epoch": 0.7102379461490295,
      "grad_norm": 1.3296421766281128,
      "learning_rate": 0.0002015884652981427,
      "loss": 0.9653,
      "step": 4537
    },
    {
      "epoch": 0.7103944896681278,
      "grad_norm": 4.0526862144470215,
      "learning_rate": 0.00020156402737047898,
      "loss": 1.0826,
      "step": 4538
    },
    {
      "epoch": 0.710551033187226,
      "grad_norm": 2.9230265617370605,
      "learning_rate": 0.00020153958944281523,
      "loss": 1.4499,
      "step": 4539
    },
    {
      "epoch": 0.7107075767063243,
      "grad_norm": 3.018975257873535,
      "learning_rate": 0.00020151515151515148,
      "loss": 0.9125,
      "step": 4540
    },
    {
      "epoch": 0.7108641202254227,
      "grad_norm": 1.6214097738265991,
      "learning_rate": 0.00020149071358748776,
      "loss": 1.3237,
      "step": 4541
    },
    {
      "epoch": 0.711020663744521,
      "grad_norm": 2.642343521118164,
      "learning_rate": 0.000201466275659824,
      "loss": 1.8457,
      "step": 4542
    },
    {
      "epoch": 0.7111772072636193,
      "grad_norm": 2.0636792182922363,
      "learning_rate": 0.0002014418377321603,
      "loss": 1.279,
      "step": 4543
    },
    {
      "epoch": 0.7113337507827175,
      "grad_norm": 3.2463300228118896,
      "learning_rate": 0.00020141739980449657,
      "loss": 1.4645,
      "step": 4544
    },
    {
      "epoch": 0.7114902943018159,
      "grad_norm": 2.8252651691436768,
      "learning_rate": 0.00020139296187683282,
      "loss": 1.3413,
      "step": 4545
    },
    {
      "epoch": 0.7116468378209142,
      "grad_norm": 0.8941448330879211,
      "learning_rate": 0.0002013685239491691,
      "loss": 0.7135,
      "step": 4546
    },
    {
      "epoch": 0.7118033813400125,
      "grad_norm": 2.6736679077148438,
      "learning_rate": 0.00020134408602150538,
      "loss": 0.9046,
      "step": 4547
    },
    {
      "epoch": 0.7119599248591109,
      "grad_norm": 2.5625529289245605,
      "learning_rate": 0.0002013196480938416,
      "loss": 1.2271,
      "step": 4548
    },
    {
      "epoch": 0.7121164683782092,
      "grad_norm": 0.8302935361862183,
      "learning_rate": 0.00020129521016617788,
      "loss": 0.3858,
      "step": 4549
    },
    {
      "epoch": 0.7122730118973074,
      "grad_norm": 3.529597043991089,
      "learning_rate": 0.00020127077223851416,
      "loss": 1.4059,
      "step": 4550
    },
    {
      "epoch": 0.7124295554164057,
      "grad_norm": 1.3789228200912476,
      "learning_rate": 0.0002012463343108504,
      "loss": 0.7232,
      "step": 4551
    },
    {
      "epoch": 0.7125860989355041,
      "grad_norm": 0.34966397285461426,
      "learning_rate": 0.0002012218963831867,
      "loss": 0.2367,
      "step": 4552
    },
    {
      "epoch": 0.7127426424546024,
      "grad_norm": 0.5851748585700989,
      "learning_rate": 0.00020119745845552297,
      "loss": 0.2883,
      "step": 4553
    },
    {
      "epoch": 0.7128991859737007,
      "grad_norm": 0.4715137779712677,
      "learning_rate": 0.00020117302052785922,
      "loss": 0.3685,
      "step": 4554
    },
    {
      "epoch": 0.713055729492799,
      "grad_norm": 0.7511829733848572,
      "learning_rate": 0.0002011485826001955,
      "loss": 0.3928,
      "step": 4555
    },
    {
      "epoch": 0.7132122730118973,
      "grad_norm": 0.8073253035545349,
      "learning_rate": 0.00020112414467253177,
      "loss": 0.3393,
      "step": 4556
    },
    {
      "epoch": 0.7133688165309956,
      "grad_norm": 1.0048470497131348,
      "learning_rate": 0.000201099706744868,
      "loss": 0.3951,
      "step": 4557
    },
    {
      "epoch": 0.7135253600500939,
      "grad_norm": 0.5629849433898926,
      "learning_rate": 0.00020107526881720428,
      "loss": 0.3419,
      "step": 4558
    },
    {
      "epoch": 0.7136819035691923,
      "grad_norm": 1.2211352586746216,
      "learning_rate": 0.00020105083088954056,
      "loss": 0.3908,
      "step": 4559
    },
    {
      "epoch": 0.7138384470882906,
      "grad_norm": 0.7062564492225647,
      "learning_rate": 0.0002010263929618768,
      "loss": 0.395,
      "step": 4560
    },
    {
      "epoch": 0.7139949906073888,
      "grad_norm": 1.036588430404663,
      "learning_rate": 0.00020100195503421309,
      "loss": 0.4042,
      "step": 4561
    },
    {
      "epoch": 0.7141515341264871,
      "grad_norm": 1.1843818426132202,
      "learning_rate": 0.00020097751710654936,
      "loss": 0.7025,
      "step": 4562
    },
    {
      "epoch": 0.7143080776455855,
      "grad_norm": 1.298744559288025,
      "learning_rate": 0.0002009530791788856,
      "loss": 0.6033,
      "step": 4563
    },
    {
      "epoch": 0.7144646211646838,
      "grad_norm": 1.004136085510254,
      "learning_rate": 0.00020092864125122187,
      "loss": 0.4368,
      "step": 4564
    },
    {
      "epoch": 0.7146211646837821,
      "grad_norm": 0.6956773996353149,
      "learning_rate": 0.00020090420332355814,
      "loss": 0.2452,
      "step": 4565
    },
    {
      "epoch": 0.7147777082028804,
      "grad_norm": 1.565514087677002,
      "learning_rate": 0.0002008797653958944,
      "loss": 0.7385,
      "step": 4566
    },
    {
      "epoch": 0.7149342517219787,
      "grad_norm": 1.181185007095337,
      "learning_rate": 0.00020085532746823067,
      "loss": 0.5778,
      "step": 4567
    },
    {
      "epoch": 0.715090795241077,
      "grad_norm": 1.010599970817566,
      "learning_rate": 0.00020083088954056695,
      "loss": 0.6435,
      "step": 4568
    },
    {
      "epoch": 0.7152473387601753,
      "grad_norm": 1.560499668121338,
      "learning_rate": 0.0002008064516129032,
      "loss": 0.6996,
      "step": 4569
    },
    {
      "epoch": 0.7154038822792737,
      "grad_norm": 2.0916194915771484,
      "learning_rate": 0.00020078201368523948,
      "loss": 0.5439,
      "step": 4570
    },
    {
      "epoch": 0.715560425798372,
      "grad_norm": 2.5257811546325684,
      "learning_rate": 0.00020075757575757576,
      "loss": 0.8417,
      "step": 4571
    },
    {
      "epoch": 0.7157169693174703,
      "grad_norm": 1.5999330282211304,
      "learning_rate": 0.00020073313782991198,
      "loss": 0.7435,
      "step": 4572
    },
    {
      "epoch": 0.7158735128365685,
      "grad_norm": 1.6199560165405273,
      "learning_rate": 0.00020070869990224826,
      "loss": 0.5395,
      "step": 4573
    },
    {
      "epoch": 0.7160300563556669,
      "grad_norm": 6.264113426208496,
      "learning_rate": 0.00020068426197458454,
      "loss": 1.2339,
      "step": 4574
    },
    {
      "epoch": 0.7161865998747652,
      "grad_norm": 1.9628108739852905,
      "learning_rate": 0.0002006598240469208,
      "loss": 0.6285,
      "step": 4575
    },
    {
      "epoch": 0.7163431433938635,
      "grad_norm": 2.0946178436279297,
      "learning_rate": 0.00020063538611925707,
      "loss": 0.5511,
      "step": 4576
    },
    {
      "epoch": 0.7164996869129618,
      "grad_norm": 1.8806025981903076,
      "learning_rate": 0.00020061094819159335,
      "loss": 0.8856,
      "step": 4577
    },
    {
      "epoch": 0.7166562304320601,
      "grad_norm": 1.6844736337661743,
      "learning_rate": 0.0002005865102639296,
      "loss": 0.9052,
      "step": 4578
    },
    {
      "epoch": 0.7168127739511584,
      "grad_norm": 1.8848754167556763,
      "learning_rate": 0.00020056207233626588,
      "loss": 0.7984,
      "step": 4579
    },
    {
      "epoch": 0.7169693174702567,
      "grad_norm": 1.9630393981933594,
      "learning_rate": 0.00020053763440860216,
      "loss": 0.7998,
      "step": 4580
    },
    {
      "epoch": 0.717125860989355,
      "grad_norm": 1.6129907369613647,
      "learning_rate": 0.00020051319648093838,
      "loss": 0.6739,
      "step": 4581
    },
    {
      "epoch": 0.7172824045084534,
      "grad_norm": 1.1391911506652832,
      "learning_rate": 0.00020048875855327466,
      "loss": 0.8227,
      "step": 4582
    },
    {
      "epoch": 0.7174389480275517,
      "grad_norm": 2.1303088665008545,
      "learning_rate": 0.00020046432062561094,
      "loss": 0.717,
      "step": 4583
    },
    {
      "epoch": 0.7175954915466499,
      "grad_norm": 2.0801308155059814,
      "learning_rate": 0.0002004398826979472,
      "loss": 1.046,
      "step": 4584
    },
    {
      "epoch": 0.7177520350657483,
      "grad_norm": 3.0255331993103027,
      "learning_rate": 0.00020041544477028347,
      "loss": 1.8212,
      "step": 4585
    },
    {
      "epoch": 0.7179085785848466,
      "grad_norm": 1.748745322227478,
      "learning_rate": 0.00020039100684261975,
      "loss": 0.8383,
      "step": 4586
    },
    {
      "epoch": 0.7180651221039449,
      "grad_norm": 2.0003836154937744,
      "learning_rate": 0.00020036656891495597,
      "loss": 0.6246,
      "step": 4587
    },
    {
      "epoch": 0.7182216656230432,
      "grad_norm": 2.874161720275879,
      "learning_rate": 0.00020034213098729225,
      "loss": 0.9589,
      "step": 4588
    },
    {
      "epoch": 0.7183782091421416,
      "grad_norm": 3.0715625286102295,
      "learning_rate": 0.00020031769305962853,
      "loss": 1.201,
      "step": 4589
    },
    {
      "epoch": 0.7185347526612398,
      "grad_norm": 2.9870784282684326,
      "learning_rate": 0.00020029325513196478,
      "loss": 1.5703,
      "step": 4590
    },
    {
      "epoch": 0.7186912961803381,
      "grad_norm": 2.393979549407959,
      "learning_rate": 0.00020026881720430106,
      "loss": 0.8619,
      "step": 4591
    },
    {
      "epoch": 0.7188478396994364,
      "grad_norm": 3.002680540084839,
      "learning_rate": 0.00020024437927663733,
      "loss": 1.3913,
      "step": 4592
    },
    {
      "epoch": 0.7190043832185348,
      "grad_norm": 2.3127334117889404,
      "learning_rate": 0.00020021994134897359,
      "loss": 1.3388,
      "step": 4593
    },
    {
      "epoch": 0.7191609267376331,
      "grad_norm": 1.3608214855194092,
      "learning_rate": 0.00020019550342130986,
      "loss": 0.948,
      "step": 4594
    },
    {
      "epoch": 0.7193174702567313,
      "grad_norm": 3.9520657062530518,
      "learning_rate": 0.00020017106549364614,
      "loss": 1.2869,
      "step": 4595
    },
    {
      "epoch": 0.7194740137758296,
      "grad_norm": 2.2881007194519043,
      "learning_rate": 0.00020014662756598237,
      "loss": 1.098,
      "step": 4596
    },
    {
      "epoch": 0.719630557294928,
      "grad_norm": 1.3503066301345825,
      "learning_rate": 0.00020012218963831864,
      "loss": 0.5111,
      "step": 4597
    },
    {
      "epoch": 0.7197871008140263,
      "grad_norm": 2.778672695159912,
      "learning_rate": 0.00020009775171065492,
      "loss": 0.8472,
      "step": 4598
    },
    {
      "epoch": 0.7199436443331246,
      "grad_norm": 3.3491392135620117,
      "learning_rate": 0.00020007331378299117,
      "loss": 0.7971,
      "step": 4599
    },
    {
      "epoch": 0.720100187852223,
      "grad_norm": 2.288565158843994,
      "learning_rate": 0.00020004887585532745,
      "loss": 0.8447,
      "step": 4600
    },
    {
      "epoch": 0.7202567313713212,
      "grad_norm": 0.5646255612373352,
      "learning_rate": 0.00020002443792766373,
      "loss": 0.3159,
      "step": 4601
    },
    {
      "epoch": 0.7204132748904195,
      "grad_norm": 0.8426042199134827,
      "learning_rate": 0.00019999999999999998,
      "loss": 0.3544,
      "step": 4602
    },
    {
      "epoch": 0.7205698184095178,
      "grad_norm": 0.6321278214454651,
      "learning_rate": 0.00019997556207233626,
      "loss": 0.2904,
      "step": 4603
    },
    {
      "epoch": 0.7207263619286162,
      "grad_norm": 0.8084971904754639,
      "learning_rate": 0.0001999511241446725,
      "loss": 0.274,
      "step": 4604
    },
    {
      "epoch": 0.7208829054477145,
      "grad_norm": 0.5742786526679993,
      "learning_rate": 0.00019992668621700876,
      "loss": 0.2366,
      "step": 4605
    },
    {
      "epoch": 0.7210394489668128,
      "grad_norm": 1.111324667930603,
      "learning_rate": 0.00019990224828934504,
      "loss": 0.324,
      "step": 4606
    },
    {
      "epoch": 0.721195992485911,
      "grad_norm": 0.9967480301856995,
      "learning_rate": 0.00019987781036168132,
      "loss": 0.3936,
      "step": 4607
    },
    {
      "epoch": 0.7213525360050094,
      "grad_norm": 0.802562952041626,
      "learning_rate": 0.00019985337243401757,
      "loss": 0.3368,
      "step": 4608
    },
    {
      "epoch": 0.7215090795241077,
      "grad_norm": 0.7982587218284607,
      "learning_rate": 0.00019982893450635385,
      "loss": 0.342,
      "step": 4609
    },
    {
      "epoch": 0.721665623043206,
      "grad_norm": 0.8451456427574158,
      "learning_rate": 0.00019980449657869013,
      "loss": 0.3102,
      "step": 4610
    },
    {
      "epoch": 0.7218221665623044,
      "grad_norm": 2.590470314025879,
      "learning_rate": 0.00019978005865102635,
      "loss": 0.8298,
      "step": 4611
    },
    {
      "epoch": 0.7219787100814026,
      "grad_norm": 1.0624793767929077,
      "learning_rate": 0.00019975562072336263,
      "loss": 0.5124,
      "step": 4612
    },
    {
      "epoch": 0.7221352536005009,
      "grad_norm": 0.9378711581230164,
      "learning_rate": 0.0001997311827956989,
      "loss": 0.4405,
      "step": 4613
    },
    {
      "epoch": 0.7222917971195992,
      "grad_norm": 1.3992924690246582,
      "learning_rate": 0.00019970674486803516,
      "loss": 0.4996,
      "step": 4614
    },
    {
      "epoch": 0.7224483406386976,
      "grad_norm": 0.9326929450035095,
      "learning_rate": 0.00019968230694037144,
      "loss": 0.5062,
      "step": 4615
    },
    {
      "epoch": 0.7226048841577959,
      "grad_norm": 1.211406946182251,
      "learning_rate": 0.00019965786901270772,
      "loss": 0.5575,
      "step": 4616
    },
    {
      "epoch": 0.7227614276768942,
      "grad_norm": 0.5172481536865234,
      "learning_rate": 0.00019963343108504397,
      "loss": 0.2156,
      "step": 4617
    },
    {
      "epoch": 0.7229179711959924,
      "grad_norm": 1.6718294620513916,
      "learning_rate": 0.00019960899315738025,
      "loss": 0.8636,
      "step": 4618
    },
    {
      "epoch": 0.7230745147150908,
      "grad_norm": 1.0531017780303955,
      "learning_rate": 0.00019958455522971652,
      "loss": 0.4851,
      "step": 4619
    },
    {
      "epoch": 0.7232310582341891,
      "grad_norm": 1.2837929725646973,
      "learning_rate": 0.00019956011730205275,
      "loss": 0.4803,
      "step": 4620
    },
    {
      "epoch": 0.7233876017532874,
      "grad_norm": 1.3181400299072266,
      "learning_rate": 0.00019953567937438903,
      "loss": 0.5956,
      "step": 4621
    },
    {
      "epoch": 0.7235441452723858,
      "grad_norm": 1.248836874961853,
      "learning_rate": 0.0001995112414467253,
      "loss": 0.6616,
      "step": 4622
    },
    {
      "epoch": 0.7237006887914841,
      "grad_norm": 1.2771812677383423,
      "learning_rate": 0.00019948680351906156,
      "loss": 0.2288,
      "step": 4623
    },
    {
      "epoch": 0.7238572323105823,
      "grad_norm": 1.3296245336532593,
      "learning_rate": 0.00019946236559139784,
      "loss": 0.6624,
      "step": 4624
    },
    {
      "epoch": 0.7240137758296806,
      "grad_norm": 1.7813029289245605,
      "learning_rate": 0.0001994379276637341,
      "loss": 0.6186,
      "step": 4625
    },
    {
      "epoch": 0.724170319348779,
      "grad_norm": 1.5180063247680664,
      "learning_rate": 0.00019941348973607036,
      "loss": 0.4384,
      "step": 4626
    },
    {
      "epoch": 0.7243268628678773,
      "grad_norm": 1.4584667682647705,
      "learning_rate": 0.00019938905180840664,
      "loss": 0.6051,
      "step": 4627
    },
    {
      "epoch": 0.7244834063869756,
      "grad_norm": 2.218289852142334,
      "learning_rate": 0.0001993646138807429,
      "loss": 0.5041,
      "step": 4628
    },
    {
      "epoch": 0.7246399499060739,
      "grad_norm": 2.1602261066436768,
      "learning_rate": 0.00019934017595307915,
      "loss": 0.7868,
      "step": 4629
    },
    {
      "epoch": 0.7247964934251722,
      "grad_norm": 1.7537661790847778,
      "learning_rate": 0.00019931573802541542,
      "loss": 0.72,
      "step": 4630
    },
    {
      "epoch": 0.7249530369442705,
      "grad_norm": 3.2519748210906982,
      "learning_rate": 0.0001992913000977517,
      "loss": 1.0954,
      "step": 4631
    },
    {
      "epoch": 0.7251095804633688,
      "grad_norm": 2.7691283226013184,
      "learning_rate": 0.00019926686217008795,
      "loss": 1.229,
      "step": 4632
    },
    {
      "epoch": 0.7252661239824671,
      "grad_norm": 2.6838536262512207,
      "learning_rate": 0.00019924242424242423,
      "loss": 0.9008,
      "step": 4633
    },
    {
      "epoch": 0.7254226675015655,
      "grad_norm": 2.501833915710449,
      "learning_rate": 0.0001992179863147605,
      "loss": 1.4006,
      "step": 4634
    },
    {
      "epoch": 0.7255792110206637,
      "grad_norm": 3.2722952365875244,
      "learning_rate": 0.00019919354838709673,
      "loss": 0.8523,
      "step": 4635
    },
    {
      "epoch": 0.725735754539762,
      "grad_norm": 2.0954771041870117,
      "learning_rate": 0.000199169110459433,
      "loss": 0.9759,
      "step": 4636
    },
    {
      "epoch": 0.7258922980588604,
      "grad_norm": 3.6169707775115967,
      "learning_rate": 0.0001991446725317693,
      "loss": 0.8356,
      "step": 4637
    },
    {
      "epoch": 0.7260488415779587,
      "grad_norm": 1.3619242906570435,
      "learning_rate": 0.00019912023460410554,
      "loss": 0.3973,
      "step": 4638
    },
    {
      "epoch": 0.726205385097057,
      "grad_norm": 2.543351650238037,
      "learning_rate": 0.00019909579667644182,
      "loss": 1.3629,
      "step": 4639
    },
    {
      "epoch": 0.7263619286161553,
      "grad_norm": 2.047996759414673,
      "learning_rate": 0.0001990713587487781,
      "loss": 0.9438,
      "step": 4640
    },
    {
      "epoch": 0.7265184721352536,
      "grad_norm": 1.984495997428894,
      "learning_rate": 0.00019904692082111435,
      "loss": 1.1399,
      "step": 4641
    },
    {
      "epoch": 0.7266750156543519,
      "grad_norm": 2.7449281215667725,
      "learning_rate": 0.00019902248289345063,
      "loss": 0.7897,
      "step": 4642
    },
    {
      "epoch": 0.7268315591734502,
      "grad_norm": 2.5032405853271484,
      "learning_rate": 0.0001989980449657869,
      "loss": 0.9229,
      "step": 4643
    },
    {
      "epoch": 0.7269881026925485,
      "grad_norm": 2.4411966800689697,
      "learning_rate": 0.00019897360703812313,
      "loss": 1.2118,
      "step": 4644
    },
    {
      "epoch": 0.7271446462116469,
      "grad_norm": 2.0347883701324463,
      "learning_rate": 0.0001989491691104594,
      "loss": 0.4972,
      "step": 4645
    },
    {
      "epoch": 0.7273011897307452,
      "grad_norm": 2.888296604156494,
      "learning_rate": 0.0001989247311827957,
      "loss": 1.3068,
      "step": 4646
    },
    {
      "epoch": 0.7274577332498434,
      "grad_norm": 1.6684255599975586,
      "learning_rate": 0.00019890029325513194,
      "loss": 0.9385,
      "step": 4647
    },
    {
      "epoch": 0.7276142767689417,
      "grad_norm": 3.1219706535339355,
      "learning_rate": 0.00019887585532746822,
      "loss": 1.1736,
      "step": 4648
    },
    {
      "epoch": 0.7277708202880401,
      "grad_norm": 2.7629895210266113,
      "learning_rate": 0.0001988514173998045,
      "loss": 0.7565,
      "step": 4649
    },
    {
      "epoch": 0.7279273638071384,
      "grad_norm": 2.8487377166748047,
      "learning_rate": 0.00019882697947214075,
      "loss": 1.5535,
      "step": 4650
    },
    {
      "epoch": 0.7280839073262367,
      "grad_norm": 0.5729645490646362,
      "learning_rate": 0.00019880254154447703,
      "loss": 0.3549,
      "step": 4651
    },
    {
      "epoch": 0.728240450845335,
      "grad_norm": 0.7285500764846802,
      "learning_rate": 0.00019877810361681328,
      "loss": 0.2756,
      "step": 4652
    },
    {
      "epoch": 0.7283969943644333,
      "grad_norm": 0.9222913384437561,
      "learning_rate": 0.00019875366568914953,
      "loss": 0.3588,
      "step": 4653
    },
    {
      "epoch": 0.7285535378835316,
      "grad_norm": 0.7123448848724365,
      "learning_rate": 0.0001987292277614858,
      "loss": 0.3541,
      "step": 4654
    },
    {
      "epoch": 0.7287100814026299,
      "grad_norm": 0.6087605357170105,
      "learning_rate": 0.00019870478983382208,
      "loss": 0.339,
      "step": 4655
    },
    {
      "epoch": 0.7288666249217283,
      "grad_norm": 0.6646626591682434,
      "learning_rate": 0.00019868035190615834,
      "loss": 0.5141,
      "step": 4656
    },
    {
      "epoch": 0.7290231684408266,
      "grad_norm": 0.6121235489845276,
      "learning_rate": 0.00019865591397849461,
      "loss": 0.3261,
      "step": 4657
    },
    {
      "epoch": 0.7291797119599248,
      "grad_norm": 0.7372736930847168,
      "learning_rate": 0.0001986314760508309,
      "loss": 0.3063,
      "step": 4658
    },
    {
      "epoch": 0.7293362554790231,
      "grad_norm": 2.8387651443481445,
      "learning_rate": 0.00019860703812316712,
      "loss": 0.4006,
      "step": 4659
    },
    {
      "epoch": 0.7294927989981215,
      "grad_norm": 0.7166696190834045,
      "learning_rate": 0.0001985826001955034,
      "loss": 0.3821,
      "step": 4660
    },
    {
      "epoch": 0.7296493425172198,
      "grad_norm": 0.9727258682250977,
      "learning_rate": 0.00019855816226783967,
      "loss": 0.3255,
      "step": 4661
    },
    {
      "epoch": 0.7298058860363181,
      "grad_norm": 0.5781895518302917,
      "learning_rate": 0.00019853372434017592,
      "loss": 0.3748,
      "step": 4662
    },
    {
      "epoch": 0.7299624295554165,
      "grad_norm": 0.8979054093360901,
      "learning_rate": 0.0001985092864125122,
      "loss": 0.3613,
      "step": 4663
    },
    {
      "epoch": 0.7301189730745147,
      "grad_norm": 1.2119108438491821,
      "learning_rate": 0.00019848484848484848,
      "loss": 0.5699,
      "step": 4664
    },
    {
      "epoch": 0.730275516593613,
      "grad_norm": 1.14756441116333,
      "learning_rate": 0.00019846041055718473,
      "loss": 0.4611,
      "step": 4665
    },
    {
      "epoch": 0.7304320601127113,
      "grad_norm": 0.9472888708114624,
      "learning_rate": 0.000198435972629521,
      "loss": 0.3666,
      "step": 4666
    },
    {
      "epoch": 0.7305886036318097,
      "grad_norm": 4.8855695724487305,
      "learning_rate": 0.0001984115347018573,
      "loss": 2.9419,
      "step": 4667
    },
    {
      "epoch": 0.730745147150908,
      "grad_norm": 1.7084994316101074,
      "learning_rate": 0.0001983870967741935,
      "loss": 0.7695,
      "step": 4668
    },
    {
      "epoch": 0.7309016906700062,
      "grad_norm": 1.1830832958221436,
      "learning_rate": 0.0001983626588465298,
      "loss": 0.5826,
      "step": 4669
    },
    {
      "epoch": 0.7310582341891045,
      "grad_norm": 1.2845669984817505,
      "learning_rate": 0.00019833822091886607,
      "loss": 0.6607,
      "step": 4670
    },
    {
      "epoch": 0.7312147777082029,
      "grad_norm": 1.136726975440979,
      "learning_rate": 0.00019831378299120232,
      "loss": 0.7845,
      "step": 4671
    },
    {
      "epoch": 0.7313713212273012,
      "grad_norm": 1.3713886737823486,
      "learning_rate": 0.0001982893450635386,
      "loss": 0.5098,
      "step": 4672
    },
    {
      "epoch": 0.7315278647463995,
      "grad_norm": 1.7736725807189941,
      "learning_rate": 0.00019826490713587488,
      "loss": 0.6415,
      "step": 4673
    },
    {
      "epoch": 0.7316844082654979,
      "grad_norm": 1.0796916484832764,
      "learning_rate": 0.00019824046920821113,
      "loss": 0.5017,
      "step": 4674
    },
    {
      "epoch": 0.7318409517845961,
      "grad_norm": 1.7601935863494873,
      "learning_rate": 0.00019821603128054738,
      "loss": 0.7534,
      "step": 4675
    },
    {
      "epoch": 0.7319974953036944,
      "grad_norm": 1.5599554777145386,
      "learning_rate": 0.00019819159335288366,
      "loss": 0.7588,
      "step": 4676
    },
    {
      "epoch": 0.7321540388227927,
      "grad_norm": 3.366999626159668,
      "learning_rate": 0.0001981671554252199,
      "loss": 0.9686,
      "step": 4677
    },
    {
      "epoch": 0.7323105823418911,
      "grad_norm": 2.5471243858337402,
      "learning_rate": 0.0001981427174975562,
      "loss": 0.8723,
      "step": 4678
    },
    {
      "epoch": 0.7324671258609894,
      "grad_norm": 1.7258960008621216,
      "learning_rate": 0.00019811827956989247,
      "loss": 0.6776,
      "step": 4679
    },
    {
      "epoch": 0.7326236693800877,
      "grad_norm": 2.3678998947143555,
      "learning_rate": 0.00019809384164222872,
      "loss": 0.821,
      "step": 4680
    },
    {
      "epoch": 0.7327802128991859,
      "grad_norm": 2.2252848148345947,
      "learning_rate": 0.000198069403714565,
      "loss": 1.0205,
      "step": 4681
    },
    {
      "epoch": 0.7329367564182843,
      "grad_norm": 1.9169201850891113,
      "learning_rate": 0.00019804496578690127,
      "loss": 0.8085,
      "step": 4682
    },
    {
      "epoch": 0.7330932999373826,
      "grad_norm": 2.3197524547576904,
      "learning_rate": 0.0001980205278592375,
      "loss": 0.6308,
      "step": 4683
    },
    {
      "epoch": 0.7332498434564809,
      "grad_norm": 2.232663631439209,
      "learning_rate": 0.00019799608993157378,
      "loss": 0.9189,
      "step": 4684
    },
    {
      "epoch": 0.7334063869755792,
      "grad_norm": 3.014984130859375,
      "learning_rate": 0.00019797165200391006,
      "loss": 0.8978,
      "step": 4685
    },
    {
      "epoch": 0.7335629304946775,
      "grad_norm": 2.799757480621338,
      "learning_rate": 0.0001979472140762463,
      "loss": 0.7204,
      "step": 4686
    },
    {
      "epoch": 0.7337194740137758,
      "grad_norm": 4.892414569854736,
      "learning_rate": 0.00019792277614858259,
      "loss": 0.8103,
      "step": 4687
    },
    {
      "epoch": 0.7338760175328741,
      "grad_norm": 2.7434020042419434,
      "learning_rate": 0.00019789833822091886,
      "loss": 1.2881,
      "step": 4688
    },
    {
      "epoch": 0.7340325610519725,
      "grad_norm": 2.714355945587158,
      "learning_rate": 0.00019787390029325511,
      "loss": 1.6406,
      "step": 4689
    },
    {
      "epoch": 0.7341891045710708,
      "grad_norm": 2.0118868350982666,
      "learning_rate": 0.0001978494623655914,
      "loss": 1.2144,
      "step": 4690
    },
    {
      "epoch": 0.7343456480901691,
      "grad_norm": 1.1020734310150146,
      "learning_rate": 0.00019782502443792767,
      "loss": 0.742,
      "step": 4691
    },
    {
      "epoch": 0.7345021916092673,
      "grad_norm": 4.144432544708252,
      "learning_rate": 0.0001978005865102639,
      "loss": 1.6325,
      "step": 4692
    },
    {
      "epoch": 0.7346587351283657,
      "grad_norm": 1.8533861637115479,
      "learning_rate": 0.00019777614858260017,
      "loss": 0.9883,
      "step": 4693
    },
    {
      "epoch": 0.734815278647464,
      "grad_norm": 2.328321695327759,
      "learning_rate": 0.00019775171065493645,
      "loss": 1.4588,
      "step": 4694
    },
    {
      "epoch": 0.7349718221665623,
      "grad_norm": 1.688111424446106,
      "learning_rate": 0.0001977272727272727,
      "loss": 0.7339,
      "step": 4695
    },
    {
      "epoch": 0.7351283656856606,
      "grad_norm": 1.9149725437164307,
      "learning_rate": 0.00019770283479960898,
      "loss": 0.7413,
      "step": 4696
    },
    {
      "epoch": 0.735284909204759,
      "grad_norm": 1.6234050989151,
      "learning_rate": 0.00019767839687194526,
      "loss": 0.5104,
      "step": 4697
    },
    {
      "epoch": 0.7354414527238572,
      "grad_norm": 2.461130142211914,
      "learning_rate": 0.0001976539589442815,
      "loss": 1.3501,
      "step": 4698
    },
    {
      "epoch": 0.7355979962429555,
      "grad_norm": 1.6503095626831055,
      "learning_rate": 0.00019762952101661776,
      "loss": 0.9382,
      "step": 4699
    },
    {
      "epoch": 0.7357545397620538,
      "grad_norm": 3.062802314758301,
      "learning_rate": 0.00019760508308895404,
      "loss": 1.4864,
      "step": 4700
    },
    {
      "epoch": 0.7359110832811522,
      "grad_norm": 0.5734681487083435,
      "learning_rate": 0.0001975806451612903,
      "loss": 0.2212,
      "step": 4701
    },
    {
      "epoch": 0.7360676268002505,
      "grad_norm": 0.5626726150512695,
      "learning_rate": 0.00019755620723362657,
      "loss": 0.2945,
      "step": 4702
    },
    {
      "epoch": 0.7362241703193487,
      "grad_norm": 0.5201683044433594,
      "learning_rate": 0.00019753176930596285,
      "loss": 0.2841,
      "step": 4703
    },
    {
      "epoch": 0.736380713838447,
      "grad_norm": 0.7159972190856934,
      "learning_rate": 0.0001975073313782991,
      "loss": 0.5189,
      "step": 4704
    },
    {
      "epoch": 0.7365372573575454,
      "grad_norm": 0.732814610004425,
      "learning_rate": 0.00019748289345063538,
      "loss": 0.3791,
      "step": 4705
    },
    {
      "epoch": 0.7366938008766437,
      "grad_norm": 0.7638669610023499,
      "learning_rate": 0.00019745845552297166,
      "loss": 0.4011,
      "step": 4706
    },
    {
      "epoch": 0.736850344395742,
      "grad_norm": 0.7000142931938171,
      "learning_rate": 0.00019743401759530788,
      "loss": 0.3322,
      "step": 4707
    },
    {
      "epoch": 0.7370068879148404,
      "grad_norm": 0.7838113307952881,
      "learning_rate": 0.00019740957966764416,
      "loss": 0.2646,
      "step": 4708
    },
    {
      "epoch": 0.7371634314339386,
      "grad_norm": 0.7795481085777283,
      "learning_rate": 0.00019738514173998044,
      "loss": 0.4024,
      "step": 4709
    },
    {
      "epoch": 0.7373199749530369,
      "grad_norm": 0.6289814114570618,
      "learning_rate": 0.0001973607038123167,
      "loss": 0.381,
      "step": 4710
    },
    {
      "epoch": 0.7374765184721352,
      "grad_norm": 1.3623909950256348,
      "learning_rate": 0.00019733626588465297,
      "loss": 0.4522,
      "step": 4711
    },
    {
      "epoch": 0.7376330619912336,
      "grad_norm": 1.1188793182373047,
      "learning_rate": 0.00019731182795698925,
      "loss": 0.4841,
      "step": 4712
    },
    {
      "epoch": 0.7377896055103319,
      "grad_norm": 1.1815077066421509,
      "learning_rate": 0.0001972873900293255,
      "loss": 0.8086,
      "step": 4713
    },
    {
      "epoch": 0.7379461490294302,
      "grad_norm": 1.2694467306137085,
      "learning_rate": 0.00019726295210166178,
      "loss": 0.6665,
      "step": 4714
    },
    {
      "epoch": 0.7381026925485284,
      "grad_norm": 1.3122878074645996,
      "learning_rate": 0.00019723851417399805,
      "loss": 0.5785,
      "step": 4715
    },
    {
      "epoch": 0.7382592360676268,
      "grad_norm": 1.089698314666748,
      "learning_rate": 0.00019721407624633428,
      "loss": 0.4062,
      "step": 4716
    },
    {
      "epoch": 0.7384157795867251,
      "grad_norm": 0.9487230181694031,
      "learning_rate": 0.00019718963831867056,
      "loss": 0.455,
      "step": 4717
    },
    {
      "epoch": 0.7385723231058234,
      "grad_norm": 1.9743801355361938,
      "learning_rate": 0.00019716520039100683,
      "loss": 0.6051,
      "step": 4718
    },
    {
      "epoch": 0.7387288666249218,
      "grad_norm": 1.2036941051483154,
      "learning_rate": 0.00019714076246334309,
      "loss": 0.5583,
      "step": 4719
    },
    {
      "epoch": 0.73888541014402,
      "grad_norm": 1.425595760345459,
      "learning_rate": 0.00019711632453567936,
      "loss": 0.562,
      "step": 4720
    },
    {
      "epoch": 0.7390419536631183,
      "grad_norm": 0.8784242272377014,
      "learning_rate": 0.00019709188660801564,
      "loss": 0.4964,
      "step": 4721
    },
    {
      "epoch": 0.7391984971822166,
      "grad_norm": 3.1630759239196777,
      "learning_rate": 0.00019706744868035187,
      "loss": 0.411,
      "step": 4722
    },
    {
      "epoch": 0.739355040701315,
      "grad_norm": 1.366860270500183,
      "learning_rate": 0.00019704301075268815,
      "loss": 0.2949,
      "step": 4723
    },
    {
      "epoch": 0.7395115842204133,
      "grad_norm": 1.7974003553390503,
      "learning_rate": 0.00019701857282502442,
      "loss": 0.7963,
      "step": 4724
    },
    {
      "epoch": 0.7396681277395116,
      "grad_norm": 2.272477388381958,
      "learning_rate": 0.00019699413489736067,
      "loss": 0.7553,
      "step": 4725
    },
    {
      "epoch": 0.7398246712586098,
      "grad_norm": 1.5582921504974365,
      "learning_rate": 0.00019696969696969695,
      "loss": 0.7692,
      "step": 4726
    },
    {
      "epoch": 0.7399812147777082,
      "grad_norm": 2.499539613723755,
      "learning_rate": 0.00019694525904203323,
      "loss": 0.6417,
      "step": 4727
    },
    {
      "epoch": 0.7401377582968065,
      "grad_norm": 2.7339725494384766,
      "learning_rate": 0.00019692082111436948,
      "loss": 0.8532,
      "step": 4728
    },
    {
      "epoch": 0.7402943018159048,
      "grad_norm": 1.615174651145935,
      "learning_rate": 0.00019689638318670576,
      "loss": 0.6366,
      "step": 4729
    },
    {
      "epoch": 0.7404508453350032,
      "grad_norm": 2.133688449859619,
      "learning_rate": 0.00019687194525904204,
      "loss": 1.0497,
      "step": 4730
    },
    {
      "epoch": 0.7406073888541015,
      "grad_norm": 1.7480201721191406,
      "learning_rate": 0.00019684750733137826,
      "loss": 0.7695,
      "step": 4731
    },
    {
      "epoch": 0.7407639323731997,
      "grad_norm": 3.059267282485962,
      "learning_rate": 0.00019682306940371454,
      "loss": 1.0978,
      "step": 4732
    },
    {
      "epoch": 0.740920475892298,
      "grad_norm": 2.2188549041748047,
      "learning_rate": 0.00019679863147605082,
      "loss": 1.1363,
      "step": 4733
    },
    {
      "epoch": 0.7410770194113964,
      "grad_norm": 1.8795181512832642,
      "learning_rate": 0.00019677419354838707,
      "loss": 0.6612,
      "step": 4734
    },
    {
      "epoch": 0.7412335629304947,
      "grad_norm": 1.7549196481704712,
      "learning_rate": 0.00019674975562072335,
      "loss": 0.987,
      "step": 4735
    },
    {
      "epoch": 0.741390106449593,
      "grad_norm": 2.692260265350342,
      "learning_rate": 0.00019672531769305963,
      "loss": 1.1045,
      "step": 4736
    },
    {
      "epoch": 0.7415466499686914,
      "grad_norm": 2.230609893798828,
      "learning_rate": 0.00019670087976539588,
      "loss": 1.2261,
      "step": 4737
    },
    {
      "epoch": 0.7417031934877896,
      "grad_norm": 3.0220465660095215,
      "learning_rate": 0.00019667644183773216,
      "loss": 1.0071,
      "step": 4738
    },
    {
      "epoch": 0.7418597370068879,
      "grad_norm": 1.864869236946106,
      "learning_rate": 0.00019665200391006844,
      "loss": 1.1151,
      "step": 4739
    },
    {
      "epoch": 0.7420162805259862,
      "grad_norm": 5.262318134307861,
      "learning_rate": 0.00019662756598240466,
      "loss": 1.2847,
      "step": 4740
    },
    {
      "epoch": 0.7421728240450846,
      "grad_norm": 2.782365083694458,
      "learning_rate": 0.00019660312805474094,
      "loss": 1.131,
      "step": 4741
    },
    {
      "epoch": 0.7423293675641829,
      "grad_norm": 1.8691052198410034,
      "learning_rate": 0.00019657869012707722,
      "loss": 1.2404,
      "step": 4742
    },
    {
      "epoch": 0.7424859110832811,
      "grad_norm": 1.3026301860809326,
      "learning_rate": 0.00019655425219941347,
      "loss": 0.7896,
      "step": 4743
    },
    {
      "epoch": 0.7426424546023794,
      "grad_norm": 2.631027936935425,
      "learning_rate": 0.00019652981427174975,
      "loss": 1.0242,
      "step": 4744
    },
    {
      "epoch": 0.7427989981214778,
      "grad_norm": 2.3272483348846436,
      "learning_rate": 0.00019650537634408603,
      "loss": 1.4005,
      "step": 4745
    },
    {
      "epoch": 0.7429555416405761,
      "grad_norm": 2.3887293338775635,
      "learning_rate": 0.00019648093841642225,
      "loss": 0.7136,
      "step": 4746
    },
    {
      "epoch": 0.7431120851596744,
      "grad_norm": 1.8265565633773804,
      "learning_rate": 0.00019645650048875853,
      "loss": 0.6281,
      "step": 4747
    },
    {
      "epoch": 0.7432686286787727,
      "grad_norm": 2.172302007675171,
      "learning_rate": 0.0001964320625610948,
      "loss": 1.017,
      "step": 4748
    },
    {
      "epoch": 0.743425172197871,
      "grad_norm": 2.7405848503112793,
      "learning_rate": 0.00019640762463343106,
      "loss": 1.1234,
      "step": 4749
    },
    {
      "epoch": 0.7435817157169693,
      "grad_norm": 1.8919093608856201,
      "learning_rate": 0.00019638318670576734,
      "loss": 0.8493,
      "step": 4750
    },
    {
      "epoch": 0.7437382592360676,
      "grad_norm": 0.5198003053665161,
      "learning_rate": 0.00019635874877810361,
      "loss": 0.338,
      "step": 4751
    },
    {
      "epoch": 0.743894802755166,
      "grad_norm": 0.49822479486465454,
      "learning_rate": 0.00019633431085043987,
      "loss": 0.253,
      "step": 4752
    },
    {
      "epoch": 0.7440513462742643,
      "grad_norm": 0.4993680417537689,
      "learning_rate": 0.00019630987292277614,
      "loss": 0.2911,
      "step": 4753
    },
    {
      "epoch": 0.7442078897933626,
      "grad_norm": 0.7572706341743469,
      "learning_rate": 0.00019628543499511242,
      "loss": 0.3687,
      "step": 4754
    },
    {
      "epoch": 0.7443644333124608,
      "grad_norm": 0.4129246175289154,
      "learning_rate": 0.00019626099706744865,
      "loss": 0.2211,
      "step": 4755
    },
    {
      "epoch": 0.7445209768315592,
      "grad_norm": 0.4937713146209717,
      "learning_rate": 0.00019623655913978492,
      "loss": 0.3346,
      "step": 4756
    },
    {
      "epoch": 0.7446775203506575,
      "grad_norm": 0.7381402850151062,
      "learning_rate": 0.0001962121212121212,
      "loss": 0.3343,
      "step": 4757
    },
    {
      "epoch": 0.7448340638697558,
      "grad_norm": 0.9552004933357239,
      "learning_rate": 0.00019618768328445745,
      "loss": 0.3558,
      "step": 4758
    },
    {
      "epoch": 0.7449906073888541,
      "grad_norm": 0.7461534142494202,
      "learning_rate": 0.00019616324535679373,
      "loss": 0.3339,
      "step": 4759
    },
    {
      "epoch": 0.7451471509079524,
      "grad_norm": 0.7386569976806641,
      "learning_rate": 0.00019613880742913,
      "loss": 0.4365,
      "step": 4760
    },
    {
      "epoch": 0.7453036944270507,
      "grad_norm": 0.824906587600708,
      "learning_rate": 0.00019611436950146626,
      "loss": 0.3719,
      "step": 4761
    },
    {
      "epoch": 0.745460237946149,
      "grad_norm": 2.553926467895508,
      "learning_rate": 0.00019608993157380254,
      "loss": 0.3857,
      "step": 4762
    },
    {
      "epoch": 0.7456167814652473,
      "grad_norm": 1.7398136854171753,
      "learning_rate": 0.0001960654936461388,
      "loss": 0.6648,
      "step": 4763
    },
    {
      "epoch": 0.7457733249843457,
      "grad_norm": 1.5790472030639648,
      "learning_rate": 0.00019604105571847504,
      "loss": 0.5358,
      "step": 4764
    },
    {
      "epoch": 0.745929868503444,
      "grad_norm": 0.9220511317253113,
      "learning_rate": 0.00019601661779081132,
      "loss": 0.3689,
      "step": 4765
    },
    {
      "epoch": 0.7460864120225422,
      "grad_norm": 2.0262274742126465,
      "learning_rate": 0.0001959921798631476,
      "loss": 1.2123,
      "step": 4766
    },
    {
      "epoch": 0.7462429555416406,
      "grad_norm": 0.8574962615966797,
      "learning_rate": 0.00019596774193548385,
      "loss": 0.5389,
      "step": 4767
    },
    {
      "epoch": 0.7463994990607389,
      "grad_norm": 0.822786271572113,
      "learning_rate": 0.00019594330400782013,
      "loss": 0.5364,
      "step": 4768
    },
    {
      "epoch": 0.7465560425798372,
      "grad_norm": 1.0409291982650757,
      "learning_rate": 0.0001959188660801564,
      "loss": 0.8335,
      "step": 4769
    },
    {
      "epoch": 0.7467125860989355,
      "grad_norm": 0.9173699617385864,
      "learning_rate": 0.00019589442815249263,
      "loss": 0.5052,
      "step": 4770
    },
    {
      "epoch": 0.7468691296180339,
      "grad_norm": 1.4677929878234863,
      "learning_rate": 0.0001958699902248289,
      "loss": 0.6807,
      "step": 4771
    },
    {
      "epoch": 0.7470256731371321,
      "grad_norm": 1.5798442363739014,
      "learning_rate": 0.0001958455522971652,
      "loss": 0.6056,
      "step": 4772
    },
    {
      "epoch": 0.7471822166562304,
      "grad_norm": 1.009047508239746,
      "learning_rate": 0.00019582111436950144,
      "loss": 0.5881,
      "step": 4773
    },
    {
      "epoch": 0.7473387601753287,
      "grad_norm": 2.4256644248962402,
      "learning_rate": 0.00019579667644183772,
      "loss": 0.5304,
      "step": 4774
    },
    {
      "epoch": 0.7474953036944271,
      "grad_norm": 2.2748332023620605,
      "learning_rate": 0.000195772238514174,
      "loss": 0.6864,
      "step": 4775
    },
    {
      "epoch": 0.7476518472135254,
      "grad_norm": 2.184377431869507,
      "learning_rate": 0.00019574780058651025,
      "loss": 0.7061,
      "step": 4776
    },
    {
      "epoch": 0.7478083907326236,
      "grad_norm": 2.4698145389556885,
      "learning_rate": 0.00019572336265884653,
      "loss": 0.8436,
      "step": 4777
    },
    {
      "epoch": 0.747964934251722,
      "grad_norm": 0.8937560319900513,
      "learning_rate": 0.0001956989247311828,
      "loss": 0.4022,
      "step": 4778
    },
    {
      "epoch": 0.7481214777708203,
      "grad_norm": 1.8450487852096558,
      "learning_rate": 0.00019567448680351903,
      "loss": 0.6108,
      "step": 4779
    },
    {
      "epoch": 0.7482780212899186,
      "grad_norm": 1.5226956605911255,
      "learning_rate": 0.0001956500488758553,
      "loss": 0.7764,
      "step": 4780
    },
    {
      "epoch": 0.7484345648090169,
      "grad_norm": 3.9673829078674316,
      "learning_rate": 0.00019562561094819159,
      "loss": 0.9701,
      "step": 4781
    },
    {
      "epoch": 0.7485911083281153,
      "grad_norm": 1.7222696542739868,
      "learning_rate": 0.00019560117302052784,
      "loss": 0.887,
      "step": 4782
    },
    {
      "epoch": 0.7487476518472135,
      "grad_norm": 3.377755641937256,
      "learning_rate": 0.00019557673509286411,
      "loss": 0.993,
      "step": 4783
    },
    {
      "epoch": 0.7489041953663118,
      "grad_norm": 2.456941604614258,
      "learning_rate": 0.0001955522971652004,
      "loss": 1.0556,
      "step": 4784
    },
    {
      "epoch": 0.7490607388854101,
      "grad_norm": 3.0926427841186523,
      "learning_rate": 0.00019552785923753664,
      "loss": 1.115,
      "step": 4785
    },
    {
      "epoch": 0.7492172824045085,
      "grad_norm": 5.375258445739746,
      "learning_rate": 0.00019550342130987292,
      "loss": 1.5124,
      "step": 4786
    },
    {
      "epoch": 0.7493738259236068,
      "grad_norm": 2.7964589595794678,
      "learning_rate": 0.00019547898338220917,
      "loss": 1.1732,
      "step": 4787
    },
    {
      "epoch": 0.7495303694427051,
      "grad_norm": 1.498062014579773,
      "learning_rate": 0.00019545454545454543,
      "loss": 1.0657,
      "step": 4788
    },
    {
      "epoch": 0.7496869129618033,
      "grad_norm": 6.085139751434326,
      "learning_rate": 0.0001954301075268817,
      "loss": 1.1077,
      "step": 4789
    },
    {
      "epoch": 0.7498434564809017,
      "grad_norm": 3.144690990447998,
      "learning_rate": 0.00019540566959921798,
      "loss": 1.4877,
      "step": 4790
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.460711717605591,
      "learning_rate": 0.00019538123167155423,
      "loss": 1.4845,
      "step": 4791
    },
    {
      "epoch": 0.7501565435190983,
      "grad_norm": 2.4659714698791504,
      "learning_rate": 0.0001953567937438905,
      "loss": 1.4173,
      "step": 4792
    },
    {
      "epoch": 0.7503130870381967,
      "grad_norm": 2.4354169368743896,
      "learning_rate": 0.0001953323558162268,
      "loss": 1.5278,
      "step": 4793
    },
    {
      "epoch": 0.7504696305572949,
      "grad_norm": 4.406135559082031,
      "learning_rate": 0.00019530791788856301,
      "loss": 1.5933,
      "step": 4794
    },
    {
      "epoch": 0.7506261740763932,
      "grad_norm": 2.1102359294891357,
      "learning_rate": 0.0001952834799608993,
      "loss": 1.278,
      "step": 4795
    },
    {
      "epoch": 0.7507827175954915,
      "grad_norm": 1.0757032632827759,
      "learning_rate": 0.00019525904203323557,
      "loss": 0.439,
      "step": 4796
    },
    {
      "epoch": 0.7509392611145899,
      "grad_norm": 1.6576645374298096,
      "learning_rate": 0.00019523460410557182,
      "loss": 0.5876,
      "step": 4797
    },
    {
      "epoch": 0.7510958046336882,
      "grad_norm": 2.0612480640411377,
      "learning_rate": 0.0001952101661779081,
      "loss": 1.1348,
      "step": 4798
    },
    {
      "epoch": 0.7512523481527865,
      "grad_norm": 4.78619384765625,
      "learning_rate": 0.00019518572825024438,
      "loss": 1.1104,
      "step": 4799
    },
    {
      "epoch": 0.7514088916718847,
      "grad_norm": 2.2526986598968506,
      "learning_rate": 0.00019516129032258063,
      "loss": 1.3444,
      "step": 4800
    },
    {
      "epoch": 0.7515654351909831,
      "grad_norm": 0.41874295473098755,
      "learning_rate": 0.0001951368523949169,
      "loss": 0.322,
      "step": 4801
    },
    {
      "epoch": 0.7517219787100814,
      "grad_norm": 0.5169926285743713,
      "learning_rate": 0.0001951124144672532,
      "loss": 0.2551,
      "step": 4802
    },
    {
      "epoch": 0.7518785222291797,
      "grad_norm": 0.8094984292984009,
      "learning_rate": 0.0001950879765395894,
      "loss": 0.3832,
      "step": 4803
    },
    {
      "epoch": 0.752035065748278,
      "grad_norm": 0.5564207434654236,
      "learning_rate": 0.0001950635386119257,
      "loss": 0.2254,
      "step": 4804
    },
    {
      "epoch": 0.7521916092673764,
      "grad_norm": 0.42472612857818604,
      "learning_rate": 0.00019503910068426197,
      "loss": 0.3473,
      "step": 4805
    },
    {
      "epoch": 0.7523481527864746,
      "grad_norm": 0.7204293012619019,
      "learning_rate": 0.00019501466275659822,
      "loss": 0.3752,
      "step": 4806
    },
    {
      "epoch": 0.7525046963055729,
      "grad_norm": 1.2864282131195068,
      "learning_rate": 0.0001949902248289345,
      "loss": 0.4892,
      "step": 4807
    },
    {
      "epoch": 0.7526612398246713,
      "grad_norm": 0.8378409743309021,
      "learning_rate": 0.00019496578690127078,
      "loss": 0.4192,
      "step": 4808
    },
    {
      "epoch": 0.7528177833437696,
      "grad_norm": 0.6147698163986206,
      "learning_rate": 0.00019494134897360703,
      "loss": 0.3945,
      "step": 4809
    },
    {
      "epoch": 0.7529743268628679,
      "grad_norm": 0.4717334508895874,
      "learning_rate": 0.0001949169110459433,
      "loss": 0.2257,
      "step": 4810
    },
    {
      "epoch": 0.7531308703819661,
      "grad_norm": 1.3184202909469604,
      "learning_rate": 0.00019489247311827956,
      "loss": 0.597,
      "step": 4811
    },
    {
      "epoch": 0.7532874139010645,
      "grad_norm": 1.2259258031845093,
      "learning_rate": 0.0001948680351906158,
      "loss": 0.4335,
      "step": 4812
    },
    {
      "epoch": 0.7534439574201628,
      "grad_norm": 0.6783806085586548,
      "learning_rate": 0.00019484359726295209,
      "loss": 0.3843,
      "step": 4813
    },
    {
      "epoch": 0.7536005009392611,
      "grad_norm": 1.3408708572387695,
      "learning_rate": 0.00019481915933528836,
      "loss": 0.3306,
      "step": 4814
    },
    {
      "epoch": 0.7537570444583594,
      "grad_norm": 0.9002304673194885,
      "learning_rate": 0.00019479472140762462,
      "loss": 0.4586,
      "step": 4815
    },
    {
      "epoch": 0.7539135879774578,
      "grad_norm": 1.12655770778656,
      "learning_rate": 0.0001947702834799609,
      "loss": 0.5046,
      "step": 4816
    },
    {
      "epoch": 0.754070131496556,
      "grad_norm": 0.9550591111183167,
      "learning_rate": 0.00019474584555229717,
      "loss": 0.5049,
      "step": 4817
    },
    {
      "epoch": 0.7542266750156543,
      "grad_norm": 1.1619162559509277,
      "learning_rate": 0.0001947214076246334,
      "loss": 0.5001,
      "step": 4818
    },
    {
      "epoch": 0.7543832185347527,
      "grad_norm": 0.7153265476226807,
      "learning_rate": 0.00019469696969696967,
      "loss": 0.3228,
      "step": 4819
    },
    {
      "epoch": 0.754539762053851,
      "grad_norm": 0.8718234300613403,
      "learning_rate": 0.00019467253176930595,
      "loss": 0.3079,
      "step": 4820
    },
    {
      "epoch": 0.7546963055729493,
      "grad_norm": 1.3940988779067993,
      "learning_rate": 0.0001946480938416422,
      "loss": 0.5131,
      "step": 4821
    },
    {
      "epoch": 0.7548528490920476,
      "grad_norm": 1.4768911600112915,
      "learning_rate": 0.00019462365591397848,
      "loss": 0.6426,
      "step": 4822
    },
    {
      "epoch": 0.7550093926111459,
      "grad_norm": 1.1347852945327759,
      "learning_rate": 0.00019459921798631476,
      "loss": 0.5,
      "step": 4823
    },
    {
      "epoch": 0.7551659361302442,
      "grad_norm": 2.5851755142211914,
      "learning_rate": 0.000194574780058651,
      "loss": 0.7089,
      "step": 4824
    },
    {
      "epoch": 0.7553224796493425,
      "grad_norm": 1.3372689485549927,
      "learning_rate": 0.0001945503421309873,
      "loss": 0.7053,
      "step": 4825
    },
    {
      "epoch": 0.7554790231684408,
      "grad_norm": 2.5522773265838623,
      "learning_rate": 0.00019452590420332357,
      "loss": 0.6847,
      "step": 4826
    },
    {
      "epoch": 0.7556355666875392,
      "grad_norm": 1.1774206161499023,
      "learning_rate": 0.0001945014662756598,
      "loss": 0.5898,
      "step": 4827
    },
    {
      "epoch": 0.7557921102066374,
      "grad_norm": 1.0767072439193726,
      "learning_rate": 0.00019447702834799607,
      "loss": 0.529,
      "step": 4828
    },
    {
      "epoch": 0.7559486537257357,
      "grad_norm": 1.9729124307632446,
      "learning_rate": 0.00019445259042033235,
      "loss": 0.4727,
      "step": 4829
    },
    {
      "epoch": 0.756105197244834,
      "grad_norm": 5.122827053070068,
      "learning_rate": 0.0001944281524926686,
      "loss": 1.1988,
      "step": 4830
    },
    {
      "epoch": 0.7562617407639324,
      "grad_norm": 1.9116922616958618,
      "learning_rate": 0.00019440371456500488,
      "loss": 0.8946,
      "step": 4831
    },
    {
      "epoch": 0.7564182842830307,
      "grad_norm": 2.4748942852020264,
      "learning_rate": 0.00019437927663734116,
      "loss": 1.0248,
      "step": 4832
    },
    {
      "epoch": 0.756574827802129,
      "grad_norm": 2.4819397926330566,
      "learning_rate": 0.0001943548387096774,
      "loss": 1.1779,
      "step": 4833
    },
    {
      "epoch": 0.7567313713212273,
      "grad_norm": 2.278400421142578,
      "learning_rate": 0.00019433040078201366,
      "loss": 0.7033,
      "step": 4834
    },
    {
      "epoch": 0.7568879148403256,
      "grad_norm": 2.6924045085906982,
      "learning_rate": 0.00019430596285434994,
      "loss": 0.8813,
      "step": 4835
    },
    {
      "epoch": 0.7570444583594239,
      "grad_norm": 1.8537712097167969,
      "learning_rate": 0.0001942815249266862,
      "loss": 0.9923,
      "step": 4836
    },
    {
      "epoch": 0.7572010018785222,
      "grad_norm": 3.0504987239837646,
      "learning_rate": 0.00019425708699902247,
      "loss": 0.8471,
      "step": 4837
    },
    {
      "epoch": 0.7573575453976206,
      "grad_norm": 3.1439967155456543,
      "learning_rate": 0.00019423264907135875,
      "loss": 1.0216,
      "step": 4838
    },
    {
      "epoch": 0.7575140889167189,
      "grad_norm": 4.802983283996582,
      "learning_rate": 0.000194208211143695,
      "loss": 0.8951,
      "step": 4839
    },
    {
      "epoch": 0.7576706324358171,
      "grad_norm": 2.6184773445129395,
      "learning_rate": 0.00019418377321603128,
      "loss": 1.1434,
      "step": 4840
    },
    {
      "epoch": 0.7578271759549154,
      "grad_norm": 4.851845741271973,
      "learning_rate": 0.00019415933528836755,
      "loss": 1.3546,
      "step": 4841
    },
    {
      "epoch": 0.7579837194740138,
      "grad_norm": 2.406771421432495,
      "learning_rate": 0.00019413489736070378,
      "loss": 1.5022,
      "step": 4842
    },
    {
      "epoch": 0.7581402629931121,
      "grad_norm": 3.594590663909912,
      "learning_rate": 0.00019411045943304006,
      "loss": 1.6779,
      "step": 4843
    },
    {
      "epoch": 0.7582968065122104,
      "grad_norm": 2.279587984085083,
      "learning_rate": 0.00019408602150537634,
      "loss": 1.8704,
      "step": 4844
    },
    {
      "epoch": 0.7584533500313086,
      "grad_norm": 2.240976333618164,
      "learning_rate": 0.0001940615835777126,
      "loss": 1.3369,
      "step": 4845
    },
    {
      "epoch": 0.758609893550407,
      "grad_norm": 2.4125685691833496,
      "learning_rate": 0.00019403714565004886,
      "loss": 0.6641,
      "step": 4846
    },
    {
      "epoch": 0.7587664370695053,
      "grad_norm": 3.437535285949707,
      "learning_rate": 0.00019401270772238514,
      "loss": 0.9713,
      "step": 4847
    },
    {
      "epoch": 0.7589229805886036,
      "grad_norm": 5.112056732177734,
      "learning_rate": 0.0001939882697947214,
      "loss": 0.6326,
      "step": 4848
    },
    {
      "epoch": 0.759079524107702,
      "grad_norm": 3.5413787364959717,
      "learning_rate": 0.00019396383186705767,
      "loss": 1.0366,
      "step": 4849
    },
    {
      "epoch": 0.7592360676268003,
      "grad_norm": 1.493313193321228,
      "learning_rate": 0.00019393939393939395,
      "loss": 0.7756,
      "step": 4850
    },
    {
      "epoch": 0.7593926111458985,
      "grad_norm": 0.4888933598995209,
      "learning_rate": 0.00019391495601173018,
      "loss": 0.2853,
      "step": 4851
    },
    {
      "epoch": 0.7595491546649968,
      "grad_norm": 0.5343114137649536,
      "learning_rate": 0.00019389051808406645,
      "loss": 0.3158,
      "step": 4852
    },
    {
      "epoch": 0.7597056981840952,
      "grad_norm": 0.571465790271759,
      "learning_rate": 0.00019386608015640273,
      "loss": 0.322,
      "step": 4853
    },
    {
      "epoch": 0.7598622417031935,
      "grad_norm": 0.4364553689956665,
      "learning_rate": 0.00019384164222873898,
      "loss": 0.2126,
      "step": 4854
    },
    {
      "epoch": 0.7600187852222918,
      "grad_norm": 1.6458876132965088,
      "learning_rate": 0.00019381720430107526,
      "loss": 0.5031,
      "step": 4855
    },
    {
      "epoch": 0.7601753287413902,
      "grad_norm": 0.8052377104759216,
      "learning_rate": 0.00019379276637341154,
      "loss": 0.3538,
      "step": 4856
    },
    {
      "epoch": 0.7603318722604884,
      "grad_norm": 0.7143148183822632,
      "learning_rate": 0.0001937683284457478,
      "loss": 0.2654,
      "step": 4857
    },
    {
      "epoch": 0.7604884157795867,
      "grad_norm": 1.215345859527588,
      "learning_rate": 0.00019374389051808404,
      "loss": 0.3274,
      "step": 4858
    },
    {
      "epoch": 0.760644959298685,
      "grad_norm": 1.0219749212265015,
      "learning_rate": 0.0001937194525904203,
      "loss": 0.4571,
      "step": 4859
    },
    {
      "epoch": 0.7608015028177834,
      "grad_norm": 1.1042121648788452,
      "learning_rate": 0.00019369501466275657,
      "loss": 0.469,
      "step": 4860
    },
    {
      "epoch": 0.7609580463368817,
      "grad_norm": 0.9635404944419861,
      "learning_rate": 0.00019367057673509285,
      "loss": 0.5121,
      "step": 4861
    },
    {
      "epoch": 0.76111458985598,
      "grad_norm": 1.2368090152740479,
      "learning_rate": 0.0001936461388074291,
      "loss": 0.4804,
      "step": 4862
    },
    {
      "epoch": 0.7612711333750782,
      "grad_norm": 1.2510402202606201,
      "learning_rate": 0.00019362170087976538,
      "loss": 0.5622,
      "step": 4863
    },
    {
      "epoch": 0.7614276768941766,
      "grad_norm": 0.9782634973526001,
      "learning_rate": 0.00019359726295210166,
      "loss": 0.495,
      "step": 4864
    },
    {
      "epoch": 0.7615842204132749,
      "grad_norm": 1.6762943267822266,
      "learning_rate": 0.00019357282502443788,
      "loss": 0.5429,
      "step": 4865
    },
    {
      "epoch": 0.7617407639323732,
      "grad_norm": 0.6857059597969055,
      "learning_rate": 0.00019354838709677416,
      "loss": 0.3212,
      "step": 4866
    },
    {
      "epoch": 0.7618973074514716,
      "grad_norm": 1.0604544878005981,
      "learning_rate": 0.00019352394916911044,
      "loss": 0.4868,
      "step": 4867
    },
    {
      "epoch": 0.7620538509705698,
      "grad_norm": 1.5608930587768555,
      "learning_rate": 0.0001934995112414467,
      "loss": 0.7897,
      "step": 4868
    },
    {
      "epoch": 0.7622103944896681,
      "grad_norm": 1.5711655616760254,
      "learning_rate": 0.00019347507331378297,
      "loss": 0.6437,
      "step": 4869
    },
    {
      "epoch": 0.7623669380087664,
      "grad_norm": 1.5553734302520752,
      "learning_rate": 0.00019345063538611925,
      "loss": 0.4996,
      "step": 4870
    },
    {
      "epoch": 0.7625234815278648,
      "grad_norm": 1.692755937576294,
      "learning_rate": 0.0001934261974584555,
      "loss": 0.5887,
      "step": 4871
    },
    {
      "epoch": 0.7626800250469631,
      "grad_norm": 1.0537468194961548,
      "learning_rate": 0.00019340175953079178,
      "loss": 0.4608,
      "step": 4872
    },
    {
      "epoch": 0.7628365685660614,
      "grad_norm": 1.4170693159103394,
      "learning_rate": 0.00019337732160312806,
      "loss": 0.4771,
      "step": 4873
    },
    {
      "epoch": 0.7629931120851596,
      "grad_norm": 2.4024274349212646,
      "learning_rate": 0.00019335288367546428,
      "loss": 0.8474,
      "step": 4874
    },
    {
      "epoch": 0.763149655604258,
      "grad_norm": 1.5059177875518799,
      "learning_rate": 0.00019332844574780056,
      "loss": 0.559,
      "step": 4875
    },
    {
      "epoch": 0.7633061991233563,
      "grad_norm": 1.0502533912658691,
      "learning_rate": 0.00019330400782013684,
      "loss": 0.4221,
      "step": 4876
    },
    {
      "epoch": 0.7634627426424546,
      "grad_norm": 2.7585763931274414,
      "learning_rate": 0.0001932795698924731,
      "loss": 0.8188,
      "step": 4877
    },
    {
      "epoch": 0.763619286161553,
      "grad_norm": 1.459269642829895,
      "learning_rate": 0.00019325513196480937,
      "loss": 0.584,
      "step": 4878
    },
    {
      "epoch": 0.7637758296806513,
      "grad_norm": 3.582939624786377,
      "learning_rate": 0.00019323069403714564,
      "loss": 1.2836,
      "step": 4879
    },
    {
      "epoch": 0.7639323731997495,
      "grad_norm": 2.4551584720611572,
      "learning_rate": 0.0001932062561094819,
      "loss": 1.0656,
      "step": 4880
    },
    {
      "epoch": 0.7640889167188478,
      "grad_norm": 1.6925396919250488,
      "learning_rate": 0.00019318181818181815,
      "loss": 0.8687,
      "step": 4881
    },
    {
      "epoch": 0.7642454602379462,
      "grad_norm": 2.231828451156616,
      "learning_rate": 0.00019315738025415442,
      "loss": 1.13,
      "step": 4882
    },
    {
      "epoch": 0.7644020037570445,
      "grad_norm": 2.1356468200683594,
      "learning_rate": 0.00019313294232649068,
      "loss": 1.1292,
      "step": 4883
    },
    {
      "epoch": 0.7645585472761428,
      "grad_norm": 2.5488100051879883,
      "learning_rate": 0.00019310850439882695,
      "loss": 0.9482,
      "step": 4884
    },
    {
      "epoch": 0.764715090795241,
      "grad_norm": 2.787428617477417,
      "learning_rate": 0.00019308406647116323,
      "loss": 1.275,
      "step": 4885
    },
    {
      "epoch": 0.7648716343143394,
      "grad_norm": 2.909212112426758,
      "learning_rate": 0.00019305962854349948,
      "loss": 1.1905,
      "step": 4886
    },
    {
      "epoch": 0.7650281778334377,
      "grad_norm": 1.801330804824829,
      "learning_rate": 0.00019303519061583576,
      "loss": 0.9273,
      "step": 4887
    },
    {
      "epoch": 0.765184721352536,
      "grad_norm": 1.8224399089813232,
      "learning_rate": 0.00019301075268817204,
      "loss": 1.3595,
      "step": 4888
    },
    {
      "epoch": 0.7653412648716343,
      "grad_norm": 1.63428795337677,
      "learning_rate": 0.00019298631476050826,
      "loss": 1.1251,
      "step": 4889
    },
    {
      "epoch": 0.7654978083907327,
      "grad_norm": 2.024421215057373,
      "learning_rate": 0.00019296187683284454,
      "loss": 1.0573,
      "step": 4890
    },
    {
      "epoch": 0.7656543519098309,
      "grad_norm": 2.4601943492889404,
      "learning_rate": 0.00019293743890518082,
      "loss": 1.2419,
      "step": 4891
    },
    {
      "epoch": 0.7658108954289292,
      "grad_norm": 1.7658562660217285,
      "learning_rate": 0.00019291300097751707,
      "loss": 0.7137,
      "step": 4892
    },
    {
      "epoch": 0.7659674389480275,
      "grad_norm": 2.124553680419922,
      "learning_rate": 0.00019288856304985335,
      "loss": 1.0605,
      "step": 4893
    },
    {
      "epoch": 0.7661239824671259,
      "grad_norm": 1.8370262384414673,
      "learning_rate": 0.00019286412512218963,
      "loss": 1.3344,
      "step": 4894
    },
    {
      "epoch": 0.7662805259862242,
      "grad_norm": 1.633594036102295,
      "learning_rate": 0.00019283968719452588,
      "loss": 1.9377,
      "step": 4895
    },
    {
      "epoch": 0.7664370695053225,
      "grad_norm": 1.5041015148162842,
      "learning_rate": 0.00019281524926686216,
      "loss": 0.7699,
      "step": 4896
    },
    {
      "epoch": 0.7665936130244208,
      "grad_norm": NaN,
      "learning_rate": 0.00019281524926686216,
      "loss": 0.0,
      "step": 4897
    },
    {
      "epoch": 0.7667501565435191,
      "grad_norm": 2.5734617710113525,
      "learning_rate": 0.00019279081133919844,
      "loss": 0.6776,
      "step": 4898
    },
    {
      "epoch": 0.7669067000626174,
      "grad_norm": 2.468716859817505,
      "learning_rate": 0.00019276637341153466,
      "loss": 0.8875,
      "step": 4899
    },
    {
      "epoch": 0.7670632435817157,
      "grad_norm": 2.735874891281128,
      "learning_rate": 0.00019274193548387094,
      "loss": 0.6044,
      "step": 4900
    },
    {
      "epoch": 0.7672197871008141,
      "grad_norm": 0.6753793954849243,
      "learning_rate": 0.00019271749755620722,
      "loss": 0.3667,
      "step": 4901
    },
    {
      "epoch": 0.7673763306199123,
      "grad_norm": 0.5195480585098267,
      "learning_rate": 0.00019269305962854347,
      "loss": 0.3415,
      "step": 4902
    },
    {
      "epoch": 0.7675328741390106,
      "grad_norm": 0.5045526623725891,
      "learning_rate": 0.00019266862170087975,
      "loss": 0.3182,
      "step": 4903
    },
    {
      "epoch": 0.7676894176581089,
      "grad_norm": 2.755859136581421,
      "learning_rate": 0.00019264418377321603,
      "loss": 0.5985,
      "step": 4904
    },
    {
      "epoch": 0.7678459611772073,
      "grad_norm": 0.4260087013244629,
      "learning_rate": 0.00019261974584555228,
      "loss": 0.2805,
      "step": 4905
    },
    {
      "epoch": 0.7680025046963056,
      "grad_norm": 0.8395310044288635,
      "learning_rate": 0.00019259530791788853,
      "loss": 0.4325,
      "step": 4906
    },
    {
      "epoch": 0.7681590482154039,
      "grad_norm": 0.5676664710044861,
      "learning_rate": 0.0001925708699902248,
      "loss": 0.2764,
      "step": 4907
    },
    {
      "epoch": 0.7683155917345021,
      "grad_norm": 0.5876914262771606,
      "learning_rate": 0.00019254643206256106,
      "loss": 0.3626,
      "step": 4908
    },
    {
      "epoch": 0.7684721352536005,
      "grad_norm": 0.6870219707489014,
      "learning_rate": 0.00019252199413489734,
      "loss": 0.3014,
      "step": 4909
    },
    {
      "epoch": 0.7686286787726988,
      "grad_norm": 0.8017184138298035,
      "learning_rate": 0.00019249755620723362,
      "loss": 0.5405,
      "step": 4910
    },
    {
      "epoch": 0.7687852222917971,
      "grad_norm": 0.7471238374710083,
      "learning_rate": 0.00019247311827956987,
      "loss": 0.4282,
      "step": 4911
    },
    {
      "epoch": 0.7689417658108955,
      "grad_norm": 0.7291784286499023,
      "learning_rate": 0.00019244868035190614,
      "loss": 0.466,
      "step": 4912
    },
    {
      "epoch": 0.7690983093299938,
      "grad_norm": 0.8841153383255005,
      "learning_rate": 0.00019242424242424242,
      "loss": 0.4488,
      "step": 4913
    },
    {
      "epoch": 0.769254852849092,
      "grad_norm": 0.8288737535476685,
      "learning_rate": 0.00019239980449657865,
      "loss": 0.3224,
      "step": 4914
    },
    {
      "epoch": 0.7694113963681903,
      "grad_norm": 2.8311057090759277,
      "learning_rate": 0.00019237536656891493,
      "loss": 0.5345,
      "step": 4915
    },
    {
      "epoch": 0.7695679398872887,
      "grad_norm": 0.9499149322509766,
      "learning_rate": 0.0001923509286412512,
      "loss": 0.3726,
      "step": 4916
    },
    {
      "epoch": 0.769724483406387,
      "grad_norm": 1.6338462829589844,
      "learning_rate": 0.00019232649071358746,
      "loss": 0.4571,
      "step": 4917
    },
    {
      "epoch": 0.7698810269254853,
      "grad_norm": 1.347562313079834,
      "learning_rate": 0.00019230205278592373,
      "loss": 0.6984,
      "step": 4918
    },
    {
      "epoch": 0.7700375704445835,
      "grad_norm": 1.349605679512024,
      "learning_rate": 0.00019227761485826,
      "loss": 0.5362,
      "step": 4919
    },
    {
      "epoch": 0.7701941139636819,
      "grad_norm": 0.9548114538192749,
      "learning_rate": 0.00019225317693059626,
      "loss": 0.3536,
      "step": 4920
    },
    {
      "epoch": 0.7703506574827802,
      "grad_norm": 1.3998003005981445,
      "learning_rate": 0.00019222873900293254,
      "loss": 0.593,
      "step": 4921
    },
    {
      "epoch": 0.7705072010018785,
      "grad_norm": 2.3617327213287354,
      "learning_rate": 0.00019220430107526882,
      "loss": 0.652,
      "step": 4922
    },
    {
      "epoch": 0.7706637445209769,
      "grad_norm": 2.4933969974517822,
      "learning_rate": 0.00019217986314760504,
      "loss": 0.8062,
      "step": 4923
    },
    {
      "epoch": 0.7708202880400752,
      "grad_norm": 1.7823599576950073,
      "learning_rate": 0.00019215542521994132,
      "loss": 0.545,
      "step": 4924
    },
    {
      "epoch": 0.7709768315591734,
      "grad_norm": 1.9176454544067383,
      "learning_rate": 0.0001921309872922776,
      "loss": 0.4175,
      "step": 4925
    },
    {
      "epoch": 0.7711333750782717,
      "grad_norm": 1.5695958137512207,
      "learning_rate": 0.00019210654936461385,
      "loss": 0.7757,
      "step": 4926
    },
    {
      "epoch": 0.7712899185973701,
      "grad_norm": 1.9228142499923706,
      "learning_rate": 0.00019208211143695013,
      "loss": 0.9986,
      "step": 4927
    },
    {
      "epoch": 0.7714464621164684,
      "grad_norm": 2.260345458984375,
      "learning_rate": 0.0001920576735092864,
      "loss": 1.1804,
      "step": 4928
    },
    {
      "epoch": 0.7716030056355667,
      "grad_norm": 1.4471839666366577,
      "learning_rate": 0.00019203323558162266,
      "loss": 0.7815,
      "step": 4929
    },
    {
      "epoch": 0.771759549154665,
      "grad_norm": 2.359945774078369,
      "learning_rate": 0.0001920087976539589,
      "loss": 0.6585,
      "step": 4930
    },
    {
      "epoch": 0.7719160926737633,
      "grad_norm": 2.4294509887695312,
      "learning_rate": 0.0001919843597262952,
      "loss": 0.8588,
      "step": 4931
    },
    {
      "epoch": 0.7720726361928616,
      "grad_norm": 2.964313268661499,
      "learning_rate": 0.00019195992179863144,
      "loss": 1.2107,
      "step": 4932
    },
    {
      "epoch": 0.7722291797119599,
      "grad_norm": 2.60548996925354,
      "learning_rate": 0.00019193548387096772,
      "loss": 0.5851,
      "step": 4933
    },
    {
      "epoch": 0.7723857232310583,
      "grad_norm": 2.0704550743103027,
      "learning_rate": 0.000191911045943304,
      "loss": 0.9809,
      "step": 4934
    },
    {
      "epoch": 0.7725422667501566,
      "grad_norm": 3.5938777923583984,
      "learning_rate": 0.00019188660801564025,
      "loss": 0.8783,
      "step": 4935
    },
    {
      "epoch": 0.7726988102692548,
      "grad_norm": 1.8894193172454834,
      "learning_rate": 0.00019186217008797653,
      "loss": 0.856,
      "step": 4936
    },
    {
      "epoch": 0.7728553537883531,
      "grad_norm": 2.43099308013916,
      "learning_rate": 0.0001918377321603128,
      "loss": 1.0566,
      "step": 4937
    },
    {
      "epoch": 0.7730118973074515,
      "grad_norm": 2.183492660522461,
      "learning_rate": 0.00019181329423264903,
      "loss": 1.2289,
      "step": 4938
    },
    {
      "epoch": 0.7731684408265498,
      "grad_norm": 1.2107994556427002,
      "learning_rate": 0.0001917888563049853,
      "loss": 0.7877,
      "step": 4939
    },
    {
      "epoch": 0.7733249843456481,
      "grad_norm": 1.4150274991989136,
      "learning_rate": 0.00019176441837732159,
      "loss": 0.9841,
      "step": 4940
    },
    {
      "epoch": 0.7734815278647464,
      "grad_norm": 1.461196780204773,
      "learning_rate": 0.00019173998044965784,
      "loss": 1.1294,
      "step": 4941
    },
    {
      "epoch": 0.7736380713838447,
      "grad_norm": 2.6352620124816895,
      "learning_rate": 0.00019171554252199412,
      "loss": 1.2065,
      "step": 4942
    },
    {
      "epoch": 0.773794614902943,
      "grad_norm": 2.3676352500915527,
      "learning_rate": 0.0001916911045943304,
      "loss": 1.1227,
      "step": 4943
    },
    {
      "epoch": 0.7739511584220413,
      "grad_norm": 2.5080063343048096,
      "learning_rate": 0.00019166666666666665,
      "loss": 1.377,
      "step": 4944
    },
    {
      "epoch": 0.7741077019411396,
      "grad_norm": 2.701669216156006,
      "learning_rate": 0.00019164222873900292,
      "loss": 1.71,
      "step": 4945
    },
    {
      "epoch": 0.774264245460238,
      "grad_norm": 1.4313629865646362,
      "learning_rate": 0.0001916177908113392,
      "loss": 0.5902,
      "step": 4946
    },
    {
      "epoch": 0.7744207889793363,
      "grad_norm": 2.563730478286743,
      "learning_rate": 0.00019159335288367543,
      "loss": 1.0311,
      "step": 4947
    },
    {
      "epoch": 0.7745773324984345,
      "grad_norm": 2.521282434463501,
      "learning_rate": 0.0001915689149560117,
      "loss": 0.8272,
      "step": 4948
    },
    {
      "epoch": 0.7747338760175329,
      "grad_norm": 2.894930601119995,
      "learning_rate": 0.00019154447702834798,
      "loss": 0.8662,
      "step": 4949
    },
    {
      "epoch": 0.7748904195366312,
      "grad_norm": 1.747538685798645,
      "learning_rate": 0.00019152003910068423,
      "loss": 0.5643,
      "step": 4950
    },
    {
      "epoch": 0.7750469630557295,
      "grad_norm": 0.750394880771637,
      "learning_rate": 0.0001914956011730205,
      "loss": 0.3981,
      "step": 4951
    },
    {
      "epoch": 0.7752035065748278,
      "grad_norm": 0.8471237421035767,
      "learning_rate": 0.0001914711632453568,
      "loss": 0.711,
      "step": 4952
    },
    {
      "epoch": 0.7753600500939261,
      "grad_norm": 0.6519607901573181,
      "learning_rate": 0.00019144672531769302,
      "loss": 0.4321,
      "step": 4953
    },
    {
      "epoch": 0.7755165936130244,
      "grad_norm": 0.5569426417350769,
      "learning_rate": 0.0001914222873900293,
      "loss": 0.2508,
      "step": 4954
    },
    {
      "epoch": 0.7756731371321227,
      "grad_norm": 0.8768931031227112,
      "learning_rate": 0.00019139784946236557,
      "loss": 0.5589,
      "step": 4955
    },
    {
      "epoch": 0.775829680651221,
      "grad_norm": 0.6277607083320618,
      "learning_rate": 0.00019137341153470182,
      "loss": 0.4621,
      "step": 4956
    },
    {
      "epoch": 0.7759862241703194,
      "grad_norm": 0.6822944283485413,
      "learning_rate": 0.0001913489736070381,
      "loss": 0.4333,
      "step": 4957
    },
    {
      "epoch": 0.7761427676894177,
      "grad_norm": 0.45079341530799866,
      "learning_rate": 0.00019132453567937438,
      "loss": 0.3288,
      "step": 4958
    },
    {
      "epoch": 0.7762993112085159,
      "grad_norm": 0.6848064661026001,
      "learning_rate": 0.00019130009775171063,
      "loss": 0.4235,
      "step": 4959
    },
    {
      "epoch": 0.7764558547276142,
      "grad_norm": 0.524940013885498,
      "learning_rate": 0.0001912756598240469,
      "loss": 0.2753,
      "step": 4960
    },
    {
      "epoch": 0.7766123982467126,
      "grad_norm": 1.9908454418182373,
      "learning_rate": 0.0001912512218963832,
      "loss": 0.421,
      "step": 4961
    },
    {
      "epoch": 0.7767689417658109,
      "grad_norm": 0.7081311345100403,
      "learning_rate": 0.0001912267839687194,
      "loss": 0.383,
      "step": 4962
    },
    {
      "epoch": 0.7769254852849092,
      "grad_norm": 1.0572980642318726,
      "learning_rate": 0.0001912023460410557,
      "loss": 0.5341,
      "step": 4963
    },
    {
      "epoch": 0.7770820288040076,
      "grad_norm": 7.10352087020874,
      "learning_rate": 0.00019117790811339197,
      "loss": 1.4755,
      "step": 4964
    },
    {
      "epoch": 0.7772385723231058,
      "grad_norm": 1.0377283096313477,
      "learning_rate": 0.00019115347018572822,
      "loss": 0.5192,
      "step": 4965
    },
    {
      "epoch": 0.7773951158422041,
      "grad_norm": 1.1436519622802734,
      "learning_rate": 0.0001911290322580645,
      "loss": 0.426,
      "step": 4966
    },
    {
      "epoch": 0.7775516593613024,
      "grad_norm": 2.1015119552612305,
      "learning_rate": 0.00019110459433040078,
      "loss": 0.6584,
      "step": 4967
    },
    {
      "epoch": 0.7777082028804008,
      "grad_norm": 0.8442532420158386,
      "learning_rate": 0.00019108015640273703,
      "loss": 0.3944,
      "step": 4968
    },
    {
      "epoch": 0.7778647463994991,
      "grad_norm": 1.5063536167144775,
      "learning_rate": 0.0001910557184750733,
      "loss": 0.5206,
      "step": 4969
    },
    {
      "epoch": 0.7780212899185974,
      "grad_norm": 1.2549412250518799,
      "learning_rate": 0.00019103128054740958,
      "loss": 0.492,
      "step": 4970
    },
    {
      "epoch": 0.7781778334376956,
      "grad_norm": 1.9004099369049072,
      "learning_rate": 0.0001910068426197458,
      "loss": 0.6607,
      "step": 4971
    },
    {
      "epoch": 0.778334376956794,
      "grad_norm": 0.9570836424827576,
      "learning_rate": 0.0001909824046920821,
      "loss": 0.4599,
      "step": 4972
    },
    {
      "epoch": 0.7784909204758923,
      "grad_norm": 1.3211475610733032,
      "learning_rate": 0.00019095796676441837,
      "loss": 0.4858,
      "step": 4973
    },
    {
      "epoch": 0.7786474639949906,
      "grad_norm": 1.0831687450408936,
      "learning_rate": 0.00019093352883675462,
      "loss": 0.4007,
      "step": 4974
    },
    {
      "epoch": 0.778804007514089,
      "grad_norm": 1.4682847261428833,
      "learning_rate": 0.0001909090909090909,
      "loss": 0.5252,
      "step": 4975
    },
    {
      "epoch": 0.7789605510331872,
      "grad_norm": 3.620468854904175,
      "learning_rate": 0.00019088465298142717,
      "loss": 1.4791,
      "step": 4976
    },
    {
      "epoch": 0.7791170945522855,
      "grad_norm": 2.0373001098632812,
      "learning_rate": 0.0001908602150537634,
      "loss": 0.7857,
      "step": 4977
    },
    {
      "epoch": 0.7792736380713838,
      "grad_norm": 2.083617925643921,
      "learning_rate": 0.00019083577712609968,
      "loss": 0.8352,
      "step": 4978
    },
    {
      "epoch": 0.7794301815904822,
      "grad_norm": 1.7250630855560303,
      "learning_rate": 0.00019081133919843595,
      "loss": 1.4002,
      "step": 4979
    },
    {
      "epoch": 0.7795867251095805,
      "grad_norm": 1.6049273014068604,
      "learning_rate": 0.0001907869012707722,
      "loss": 0.588,
      "step": 4980
    },
    {
      "epoch": 0.7797432686286788,
      "grad_norm": 1.8046834468841553,
      "learning_rate": 0.00019076246334310848,
      "loss": 0.9482,
      "step": 4981
    },
    {
      "epoch": 0.779899812147777,
      "grad_norm": 2.0542006492614746,
      "learning_rate": 0.00019073802541544476,
      "loss": 0.9811,
      "step": 4982
    },
    {
      "epoch": 0.7800563556668754,
      "grad_norm": 1.8561007976531982,
      "learning_rate": 0.000190713587487781,
      "loss": 0.5584,
      "step": 4983
    },
    {
      "epoch": 0.7802128991859737,
      "grad_norm": 4.412578582763672,
      "learning_rate": 0.0001906891495601173,
      "loss": 1.2283,
      "step": 4984
    },
    {
      "epoch": 0.780369442705072,
      "grad_norm": 2.7626380920410156,
      "learning_rate": 0.00019066471163245357,
      "loss": 0.8703,
      "step": 4985
    },
    {
      "epoch": 0.7805259862241704,
      "grad_norm": 3.2905735969543457,
      "learning_rate": 0.0001906402737047898,
      "loss": 1.096,
      "step": 4986
    },
    {
      "epoch": 0.7806825297432687,
      "grad_norm": 1.68156099319458,
      "learning_rate": 0.00019061583577712607,
      "loss": 1.0584,
      "step": 4987
    },
    {
      "epoch": 0.7808390732623669,
      "grad_norm": 3.365527629852295,
      "learning_rate": 0.00019059139784946235,
      "loss": 1.3791,
      "step": 4988
    },
    {
      "epoch": 0.7809956167814652,
      "grad_norm": 3.2215206623077393,
      "learning_rate": 0.0001905669599217986,
      "loss": 1.0837,
      "step": 4989
    },
    {
      "epoch": 0.7811521603005636,
      "grad_norm": 2.5578112602233887,
      "learning_rate": 0.00019054252199413488,
      "loss": 1.0125,
      "step": 4990
    },
    {
      "epoch": 0.7813087038196619,
      "grad_norm": 4.676730632781982,
      "learning_rate": 0.00019051808406647116,
      "loss": 1.3229,
      "step": 4991
    },
    {
      "epoch": 0.7814652473387602,
      "grad_norm": 3.1872496604919434,
      "learning_rate": 0.0001904936461388074,
      "loss": 1.6003,
      "step": 4992
    },
    {
      "epoch": 0.7816217908578584,
      "grad_norm": 2.7677407264709473,
      "learning_rate": 0.0001904692082111437,
      "loss": 0.9255,
      "step": 4993
    },
    {
      "epoch": 0.7817783343769568,
      "grad_norm": 1.499570608139038,
      "learning_rate": 0.00019044477028347994,
      "loss": 1.177,
      "step": 4994
    },
    {
      "epoch": 0.7819348778960551,
      "grad_norm": 2.0705745220184326,
      "learning_rate": 0.0001904203323558162,
      "loss": 1.2654,
      "step": 4995
    },
    {
      "epoch": 0.7820914214151534,
      "grad_norm": 3.193979263305664,
      "learning_rate": 0.00019039589442815247,
      "loss": 0.7452,
      "step": 4996
    },
    {
      "epoch": 0.7822479649342517,
      "grad_norm": 2.0789194107055664,
      "learning_rate": 0.00019037145650048875,
      "loss": 0.5481,
      "step": 4997
    },
    {
      "epoch": 0.7824045084533501,
      "grad_norm": 2.25919771194458,
      "learning_rate": 0.000190347018572825,
      "loss": 1.0294,
      "step": 4998
    },
    {
      "epoch": 0.7825610519724483,
      "grad_norm": 1.2901616096496582,
      "learning_rate": 0.00019032258064516128,
      "loss": 1.025,
      "step": 4999
    },
    {
      "epoch": 0.7827175954915466,
      "grad_norm": 1.560736894607544,
      "learning_rate": 0.00019029814271749756,
      "loss": 0.8492,
      "step": 5000
    },
    {
      "epoch": 0.7827175954915466,
      "eval_loss": 0.5757540464401245,
      "eval_runtime": 205.5697,
      "eval_samples_per_second": 60.237,
      "eval_steps_per_second": 3.765,
      "eval_wer": 0.36954119472129543,
      "step": 5000
    },
    {
      "epoch": 0.782874139010645,
      "grad_norm": 1.0069044828414917,
      "learning_rate": 0.00019027370478983378,
      "loss": 0.5492,
      "step": 5001
    },
    {
      "epoch": 0.7830306825297433,
      "grad_norm": 0.5260990262031555,
      "learning_rate": 0.00019024926686217006,
      "loss": 0.2823,
      "step": 5002
    },
    {
      "epoch": 0.7831872260488416,
      "grad_norm": 0.48100632429122925,
      "learning_rate": 0.00019022482893450634,
      "loss": 0.2471,
      "step": 5003
    },
    {
      "epoch": 0.7833437695679399,
      "grad_norm": 0.6281126141548157,
      "learning_rate": 0.0001902003910068426,
      "loss": 0.2494,
      "step": 5004
    },
    {
      "epoch": 0.7835003130870382,
      "grad_norm": 0.869832456111908,
      "learning_rate": 0.00019017595307917887,
      "loss": 0.3811,
      "step": 5005
    },
    {
      "epoch": 0.7836568566061365,
      "grad_norm": 0.6189180612564087,
      "learning_rate": 0.00019015151515151514,
      "loss": 0.2062,
      "step": 5006
    },
    {
      "epoch": 0.7838134001252348,
      "grad_norm": 1.0378501415252686,
      "learning_rate": 0.0001901270772238514,
      "loss": 0.3194,
      "step": 5007
    },
    {
      "epoch": 0.7839699436443331,
      "grad_norm": 0.7968530654907227,
      "learning_rate": 0.00019010263929618767,
      "loss": 0.3416,
      "step": 5008
    },
    {
      "epoch": 0.7841264871634315,
      "grad_norm": 1.4084452390670776,
      "learning_rate": 0.00019007820136852395,
      "loss": 0.4465,
      "step": 5009
    },
    {
      "epoch": 0.7842830306825297,
      "grad_norm": 0.9723837375640869,
      "learning_rate": 0.00019005376344086018,
      "loss": 0.3285,
      "step": 5010
    },
    {
      "epoch": 0.784439574201628,
      "grad_norm": 1.0934810638427734,
      "learning_rate": 0.00019002932551319645,
      "loss": 0.3684,
      "step": 5011
    },
    {
      "epoch": 0.7845961177207263,
      "grad_norm": 1.6922805309295654,
      "learning_rate": 0.00019000488758553273,
      "loss": 0.7612,
      "step": 5012
    },
    {
      "epoch": 0.7847526612398247,
      "grad_norm": 1.0072715282440186,
      "learning_rate": 0.00018998044965786898,
      "loss": 0.7402,
      "step": 5013
    },
    {
      "epoch": 0.784909204758923,
      "grad_norm": 1.367684245109558,
      "learning_rate": 0.00018995601173020526,
      "loss": 0.4451,
      "step": 5014
    },
    {
      "epoch": 0.7850657482780213,
      "grad_norm": 1.189140796661377,
      "learning_rate": 0.00018993157380254154,
      "loss": 0.4766,
      "step": 5015
    },
    {
      "epoch": 0.7852222917971196,
      "grad_norm": 1.3974438905715942,
      "learning_rate": 0.0001899071358748778,
      "loss": 0.6411,
      "step": 5016
    },
    {
      "epoch": 0.7853788353162179,
      "grad_norm": 1.6327006816864014,
      "learning_rate": 0.00018988269794721407,
      "loss": 0.3786,
      "step": 5017
    },
    {
      "epoch": 0.7855353788353162,
      "grad_norm": 0.9266005754470825,
      "learning_rate": 0.00018985826001955032,
      "loss": 0.4677,
      "step": 5018
    },
    {
      "epoch": 0.7856919223544145,
      "grad_norm": 0.8187036514282227,
      "learning_rate": 0.00018983382209188657,
      "loss": 0.5805,
      "step": 5019
    },
    {
      "epoch": 0.7858484658735129,
      "grad_norm": 1.5598969459533691,
      "learning_rate": 0.00018980938416422285,
      "loss": 0.6537,
      "step": 5020
    },
    {
      "epoch": 0.7860050093926112,
      "grad_norm": 1.5645040273666382,
      "learning_rate": 0.00018978494623655913,
      "loss": 0.6621,
      "step": 5021
    },
    {
      "epoch": 0.7861615529117094,
      "grad_norm": 1.342915654182434,
      "learning_rate": 0.00018976050830889538,
      "loss": 0.4445,
      "step": 5022
    },
    {
      "epoch": 0.7863180964308077,
      "grad_norm": 1.3394932746887207,
      "learning_rate": 0.00018973607038123166,
      "loss": 0.7038,
      "step": 5023
    },
    {
      "epoch": 0.7864746399499061,
      "grad_norm": 1.4908233880996704,
      "learning_rate": 0.00018971163245356794,
      "loss": 0.5493,
      "step": 5024
    },
    {
      "epoch": 0.7866311834690044,
      "grad_norm": 1.5220710039138794,
      "learning_rate": 0.00018968719452590416,
      "loss": 0.6794,
      "step": 5025
    },
    {
      "epoch": 0.7867877269881027,
      "grad_norm": 6.862061023712158,
      "learning_rate": 0.00018966275659824044,
      "loss": 1.0358,
      "step": 5026
    },
    {
      "epoch": 0.786944270507201,
      "grad_norm": 1.3390004634857178,
      "learning_rate": 0.00018963831867057672,
      "loss": 0.6504,
      "step": 5027
    },
    {
      "epoch": 0.7871008140262993,
      "grad_norm": 1.7340364456176758,
      "learning_rate": 0.00018961388074291297,
      "loss": 0.9927,
      "step": 5028
    },
    {
      "epoch": 0.7872573575453976,
      "grad_norm": 2.199751377105713,
      "learning_rate": 0.00018958944281524925,
      "loss": 0.7016,
      "step": 5029
    },
    {
      "epoch": 0.7874139010644959,
      "grad_norm": 1.6952979564666748,
      "learning_rate": 0.00018956500488758553,
      "loss": 0.3266,
      "step": 5030
    },
    {
      "epoch": 0.7875704445835943,
      "grad_norm": 2.159717082977295,
      "learning_rate": 0.00018954056695992178,
      "loss": 0.6416,
      "step": 5031
    },
    {
      "epoch": 0.7877269881026926,
      "grad_norm": 1.6389358043670654,
      "learning_rate": 0.00018951612903225806,
      "loss": 0.8744,
      "step": 5032
    },
    {
      "epoch": 0.7878835316217908,
      "grad_norm": 2.425839424133301,
      "learning_rate": 0.00018949169110459433,
      "loss": 1.0329,
      "step": 5033
    },
    {
      "epoch": 0.7880400751408891,
      "grad_norm": 3.5907609462738037,
      "learning_rate": 0.00018946725317693056,
      "loss": 0.8958,
      "step": 5034
    },
    {
      "epoch": 0.7881966186599875,
      "grad_norm": 1.7229177951812744,
      "learning_rate": 0.00018944281524926684,
      "loss": 0.7585,
      "step": 5035
    },
    {
      "epoch": 0.7883531621790858,
      "grad_norm": 5.789181232452393,
      "learning_rate": 0.00018941837732160312,
      "loss": 1.1827,
      "step": 5036
    },
    {
      "epoch": 0.7885097056981841,
      "grad_norm": 6.507972717285156,
      "learning_rate": 0.00018939393939393937,
      "loss": 1.1366,
      "step": 5037
    },
    {
      "epoch": 0.7886662492172825,
      "grad_norm": 2.509685754776001,
      "learning_rate": 0.00018936950146627565,
      "loss": 0.904,
      "step": 5038
    },
    {
      "epoch": 0.7888227927363807,
      "grad_norm": 3.2579421997070312,
      "learning_rate": 0.00018934506353861192,
      "loss": 0.7319,
      "step": 5039
    },
    {
      "epoch": 0.788979336255479,
      "grad_norm": 2.354316234588623,
      "learning_rate": 0.00018932062561094817,
      "loss": 1.0652,
      "step": 5040
    },
    {
      "epoch": 0.7891358797745773,
      "grad_norm": 1.7672300338745117,
      "learning_rate": 0.00018929618768328443,
      "loss": 1.0612,
      "step": 5041
    },
    {
      "epoch": 0.7892924232936757,
      "grad_norm": 2.5716187953948975,
      "learning_rate": 0.0001892717497556207,
      "loss": 1.1822,
      "step": 5042
    },
    {
      "epoch": 0.789448966812774,
      "grad_norm": 1.6928919553756714,
      "learning_rate": 0.00018924731182795696,
      "loss": 1.7569,
      "step": 5043
    },
    {
      "epoch": 0.7896055103318722,
      "grad_norm": 2.918978452682495,
      "learning_rate": 0.00018922287390029323,
      "loss": 1.1707,
      "step": 5044
    },
    {
      "epoch": 0.7897620538509705,
      "grad_norm": 2.933793783187866,
      "learning_rate": 0.0001891984359726295,
      "loss": 1.4186,
      "step": 5045
    },
    {
      "epoch": 0.7899185973700689,
      "grad_norm": 3.004098653793335,
      "learning_rate": 0.00018917399804496576,
      "loss": 1.0802,
      "step": 5046
    },
    {
      "epoch": 0.7900751408891672,
      "grad_norm": 4.242359638214111,
      "learning_rate": 0.00018914956011730204,
      "loss": 1.2112,
      "step": 5047
    },
    {
      "epoch": 0.7902316844082655,
      "grad_norm": 1.2052658796310425,
      "learning_rate": 0.00018912512218963832,
      "loss": 0.6445,
      "step": 5048
    },
    {
      "epoch": 0.7903882279273639,
      "grad_norm": 2.339385986328125,
      "learning_rate": 0.00018910068426197454,
      "loss": 1.21,
      "step": 5049
    },
    {
      "epoch": 0.7905447714464621,
      "grad_norm": 1.4080181121826172,
      "learning_rate": 0.00018907624633431082,
      "loss": 0.8448,
      "step": 5050
    },
    {
      "epoch": 0.7907013149655604,
      "grad_norm": 0.7067837119102478,
      "learning_rate": 0.0001890518084066471,
      "loss": 0.2988,
      "step": 5051
    },
    {
      "epoch": 0.7908578584846587,
      "grad_norm": 0.7016892433166504,
      "learning_rate": 0.00018902737047898335,
      "loss": 0.2903,
      "step": 5052
    },
    {
      "epoch": 0.7910144020037571,
      "grad_norm": 0.7686540484428406,
      "learning_rate": 0.00018900293255131963,
      "loss": 0.3338,
      "step": 5053
    },
    {
      "epoch": 0.7911709455228554,
      "grad_norm": 0.7467711567878723,
      "learning_rate": 0.0001889784946236559,
      "loss": 0.4192,
      "step": 5054
    },
    {
      "epoch": 0.7913274890419537,
      "grad_norm": 1.1654859781265259,
      "learning_rate": 0.00018895405669599216,
      "loss": 0.3853,
      "step": 5055
    },
    {
      "epoch": 0.7914840325610519,
      "grad_norm": 0.7265309691429138,
      "learning_rate": 0.00018892961876832844,
      "loss": 0.3844,
      "step": 5056
    },
    {
      "epoch": 0.7916405760801503,
      "grad_norm": 0.5267289280891418,
      "learning_rate": 0.00018890518084066472,
      "loss": 0.3022,
      "step": 5057
    },
    {
      "epoch": 0.7917971195992486,
      "grad_norm": 0.6257277727127075,
      "learning_rate": 0.00018888074291300094,
      "loss": 0.4393,
      "step": 5058
    },
    {
      "epoch": 0.7919536631183469,
      "grad_norm": 2.329874277114868,
      "learning_rate": 0.00018885630498533722,
      "loss": 0.8329,
      "step": 5059
    },
    {
      "epoch": 0.7921102066374452,
      "grad_norm": 0.7630550861358643,
      "learning_rate": 0.0001888318670576735,
      "loss": 0.3971,
      "step": 5060
    },
    {
      "epoch": 0.7922667501565435,
      "grad_norm": 0.726796567440033,
      "learning_rate": 0.00018880742913000975,
      "loss": 0.3248,
      "step": 5061
    },
    {
      "epoch": 0.7924232936756418,
      "grad_norm": 0.9759939312934875,
      "learning_rate": 0.00018878299120234603,
      "loss": 0.3606,
      "step": 5062
    },
    {
      "epoch": 0.7925798371947401,
      "grad_norm": 1.23695969581604,
      "learning_rate": 0.0001887585532746823,
      "loss": 0.5576,
      "step": 5063
    },
    {
      "epoch": 0.7927363807138385,
      "grad_norm": 1.5264012813568115,
      "learning_rate": 0.00018873411534701856,
      "loss": 0.7945,
      "step": 5064
    },
    {
      "epoch": 0.7928929242329368,
      "grad_norm": 0.8853788375854492,
      "learning_rate": 0.0001887096774193548,
      "loss": 0.5012,
      "step": 5065
    },
    {
      "epoch": 0.7930494677520351,
      "grad_norm": 0.9085557460784912,
      "learning_rate": 0.0001886852394916911,
      "loss": 0.4011,
      "step": 5066
    },
    {
      "epoch": 0.7932060112711333,
      "grad_norm": 0.9665614366531372,
      "learning_rate": 0.00018866080156402734,
      "loss": 0.3467,
      "step": 5067
    },
    {
      "epoch": 0.7933625547902317,
      "grad_norm": 1.1219550371170044,
      "learning_rate": 0.00018863636363636362,
      "loss": 0.6318,
      "step": 5068
    },
    {
      "epoch": 0.79351909830933,
      "grad_norm": 1.8163543939590454,
      "learning_rate": 0.0001886119257086999,
      "loss": 0.5886,
      "step": 5069
    },
    {
      "epoch": 0.7936756418284283,
      "grad_norm": 1.5584337711334229,
      "learning_rate": 0.00018858748778103615,
      "loss": 0.764,
      "step": 5070
    },
    {
      "epoch": 0.7938321853475266,
      "grad_norm": 2.26932692527771,
      "learning_rate": 0.00018856304985337242,
      "loss": 1.0698,
      "step": 5071
    },
    {
      "epoch": 0.793988728866625,
      "grad_norm": 1.0103601217269897,
      "learning_rate": 0.0001885386119257087,
      "loss": 0.4875,
      "step": 5072
    },
    {
      "epoch": 0.7941452723857232,
      "grad_norm": 2.160799980163574,
      "learning_rate": 0.00018851417399804493,
      "loss": 0.7125,
      "step": 5073
    },
    {
      "epoch": 0.7943018159048215,
      "grad_norm": 1.7695913314819336,
      "learning_rate": 0.0001884897360703812,
      "loss": 0.7405,
      "step": 5074
    },
    {
      "epoch": 0.7944583594239198,
      "grad_norm": 1.6188089847564697,
      "learning_rate": 0.00018846529814271748,
      "loss": 0.6548,
      "step": 5075
    },
    {
      "epoch": 0.7946149029430182,
      "grad_norm": 2.61811900138855,
      "learning_rate": 0.00018844086021505373,
      "loss": 1.2027,
      "step": 5076
    },
    {
      "epoch": 0.7947714464621165,
      "grad_norm": 2.0806033611297607,
      "learning_rate": 0.00018841642228739,
      "loss": 0.9647,
      "step": 5077
    },
    {
      "epoch": 0.7949279899812148,
      "grad_norm": 1.6380743980407715,
      "learning_rate": 0.0001883919843597263,
      "loss": 0.73,
      "step": 5078
    },
    {
      "epoch": 0.795084533500313,
      "grad_norm": 1.2052991390228271,
      "learning_rate": 0.00018836754643206254,
      "loss": 0.5505,
      "step": 5079
    },
    {
      "epoch": 0.7952410770194114,
      "grad_norm": 1.2171603441238403,
      "learning_rate": 0.00018834310850439882,
      "loss": 0.6914,
      "step": 5080
    },
    {
      "epoch": 0.7953976205385097,
      "grad_norm": 1.6889578104019165,
      "learning_rate": 0.0001883186705767351,
      "loss": 0.6407,
      "step": 5081
    },
    {
      "epoch": 0.795554164057608,
      "grad_norm": 2.487717866897583,
      "learning_rate": 0.00018829423264907132,
      "loss": 0.9845,
      "step": 5082
    },
    {
      "epoch": 0.7957107075767064,
      "grad_norm": 2.156898021697998,
      "learning_rate": 0.0001882697947214076,
      "loss": 0.7382,
      "step": 5083
    },
    {
      "epoch": 0.7958672510958046,
      "grad_norm": 2.150857448577881,
      "learning_rate": 0.00018824535679374388,
      "loss": 0.6193,
      "step": 5084
    },
    {
      "epoch": 0.7960237946149029,
      "grad_norm": 2.0828707218170166,
      "learning_rate": 0.00018822091886608013,
      "loss": 0.6914,
      "step": 5085
    },
    {
      "epoch": 0.7961803381340012,
      "grad_norm": 2.405583381652832,
      "learning_rate": 0.0001881964809384164,
      "loss": 0.9627,
      "step": 5086
    },
    {
      "epoch": 0.7963368816530996,
      "grad_norm": 2.8909096717834473,
      "learning_rate": 0.0001881720430107527,
      "loss": 1.0168,
      "step": 5087
    },
    {
      "epoch": 0.7964934251721979,
      "grad_norm": 2.1869802474975586,
      "learning_rate": 0.00018814760508308894,
      "loss": 0.8169,
      "step": 5088
    },
    {
      "epoch": 0.7966499686912962,
      "grad_norm": 2.3138229846954346,
      "learning_rate": 0.0001881231671554252,
      "loss": 1.6036,
      "step": 5089
    },
    {
      "epoch": 0.7968065122103944,
      "grad_norm": 5.03582763671875,
      "learning_rate": 0.00018809872922776147,
      "loss": 1.9535,
      "step": 5090
    },
    {
      "epoch": 0.7969630557294928,
      "grad_norm": 2.7218527793884277,
      "learning_rate": 0.00018807429130009772,
      "loss": 1.4739,
      "step": 5091
    },
    {
      "epoch": 0.7971195992485911,
      "grad_norm": 2.0012664794921875,
      "learning_rate": 0.000188049853372434,
      "loss": 0.9925,
      "step": 5092
    },
    {
      "epoch": 0.7972761427676894,
      "grad_norm": 3.092926502227783,
      "learning_rate": 0.00018802541544477028,
      "loss": 1.289,
      "step": 5093
    },
    {
      "epoch": 0.7974326862867878,
      "grad_norm": 1.7731401920318604,
      "learning_rate": 0.00018800097751710653,
      "loss": 1.1165,
      "step": 5094
    },
    {
      "epoch": 0.7975892298058861,
      "grad_norm": 3.9922702312469482,
      "learning_rate": 0.0001879765395894428,
      "loss": 1.2458,
      "step": 5095
    },
    {
      "epoch": 0.7977457733249843,
      "grad_norm": 1.2292232513427734,
      "learning_rate": 0.00018795210166177908,
      "loss": 0.6469,
      "step": 5096
    },
    {
      "epoch": 0.7979023168440826,
      "grad_norm": 1.4839969873428345,
      "learning_rate": 0.0001879276637341153,
      "loss": 0.4321,
      "step": 5097
    },
    {
      "epoch": 0.798058860363181,
      "grad_norm": 1.8727073669433594,
      "learning_rate": 0.0001879032258064516,
      "loss": 0.7269,
      "step": 5098
    },
    {
      "epoch": 0.7982154038822793,
      "grad_norm": 1.444756269454956,
      "learning_rate": 0.00018787878787878787,
      "loss": 0.4632,
      "step": 5099
    },
    {
      "epoch": 0.7983719474013776,
      "grad_norm": 1.8207427263259888,
      "learning_rate": 0.00018785434995112412,
      "loss": 0.7119,
      "step": 5100
    },
    {
      "epoch": 0.7985284909204758,
      "grad_norm": 2.3908092975616455,
      "learning_rate": 0.0001878299120234604,
      "loss": 0.7385,
      "step": 5101
    },
    {
      "epoch": 0.7986850344395742,
      "grad_norm": 0.8618811964988708,
      "learning_rate": 0.00018780547409579667,
      "loss": 0.4004,
      "step": 5102
    },
    {
      "epoch": 0.7988415779586725,
      "grad_norm": 0.7508620023727417,
      "learning_rate": 0.00018778103616813292,
      "loss": 0.3438,
      "step": 5103
    },
    {
      "epoch": 0.7989981214777708,
      "grad_norm": 0.8002655506134033,
      "learning_rate": 0.0001877565982404692,
      "loss": 0.3142,
      "step": 5104
    },
    {
      "epoch": 0.7991546649968692,
      "grad_norm": 0.8202148675918579,
      "learning_rate": 0.00018773216031280548,
      "loss": 0.2769,
      "step": 5105
    },
    {
      "epoch": 0.7993112085159675,
      "grad_norm": 0.5834996700286865,
      "learning_rate": 0.0001877077223851417,
      "loss": 0.2992,
      "step": 5106
    },
    {
      "epoch": 0.7994677520350657,
      "grad_norm": 0.6148617267608643,
      "learning_rate": 0.00018768328445747798,
      "loss": 0.2415,
      "step": 5107
    },
    {
      "epoch": 0.799624295554164,
      "grad_norm": 1.0197445154190063,
      "learning_rate": 0.00018765884652981426,
      "loss": 0.3005,
      "step": 5108
    },
    {
      "epoch": 0.7997808390732624,
      "grad_norm": 0.8099470138549805,
      "learning_rate": 0.00018763440860215051,
      "loss": 0.4335,
      "step": 5109
    },
    {
      "epoch": 0.7999373825923607,
      "grad_norm": 0.9061375260353088,
      "learning_rate": 0.0001876099706744868,
      "loss": 0.4103,
      "step": 5110
    },
    {
      "epoch": 0.800093926111459,
      "grad_norm": 0.8547300696372986,
      "learning_rate": 0.00018758553274682307,
      "loss": 0.4025,
      "step": 5111
    },
    {
      "epoch": 0.8002504696305573,
      "grad_norm": 5.366735935211182,
      "learning_rate": 0.0001875610948191593,
      "loss": 0.8325,
      "step": 5112
    },
    {
      "epoch": 0.8004070131496556,
      "grad_norm": 0.9143383502960205,
      "learning_rate": 0.00018753665689149557,
      "loss": 0.3426,
      "step": 5113
    },
    {
      "epoch": 0.8005635566687539,
      "grad_norm": 0.7643350958824158,
      "learning_rate": 0.00018751221896383185,
      "loss": 0.3594,
      "step": 5114
    },
    {
      "epoch": 0.8007201001878522,
      "grad_norm": 1.4406952857971191,
      "learning_rate": 0.0001874877810361681,
      "loss": 0.3918,
      "step": 5115
    },
    {
      "epoch": 0.8008766437069506,
      "grad_norm": 1.629372477531433,
      "learning_rate": 0.00018746334310850438,
      "loss": 0.441,
      "step": 5116
    },
    {
      "epoch": 0.8010331872260489,
      "grad_norm": 2.073367118835449,
      "learning_rate": 0.00018743890518084066,
      "loss": 0.5043,
      "step": 5117
    },
    {
      "epoch": 0.8011897307451471,
      "grad_norm": 1.0843485593795776,
      "learning_rate": 0.0001874144672531769,
      "loss": 0.5772,
      "step": 5118
    },
    {
      "epoch": 0.8013462742642454,
      "grad_norm": 1.2174686193466187,
      "learning_rate": 0.0001873900293255132,
      "loss": 0.4383,
      "step": 5119
    },
    {
      "epoch": 0.8015028177833438,
      "grad_norm": 1.9818603992462158,
      "learning_rate": 0.00018736559139784947,
      "loss": 0.9853,
      "step": 5120
    },
    {
      "epoch": 0.8016593613024421,
      "grad_norm": 1.621567726135254,
      "learning_rate": 0.0001873411534701857,
      "loss": 0.7447,
      "step": 5121
    },
    {
      "epoch": 0.8018159048215404,
      "grad_norm": 1.1117850542068481,
      "learning_rate": 0.00018731671554252197,
      "loss": 0.745,
      "step": 5122
    },
    {
      "epoch": 0.8019724483406387,
      "grad_norm": 1.9802541732788086,
      "learning_rate": 0.00018729227761485825,
      "loss": 0.673,
      "step": 5123
    },
    {
      "epoch": 0.802128991859737,
      "grad_norm": 2.39848256111145,
      "learning_rate": 0.0001872678396871945,
      "loss": 0.7147,
      "step": 5124
    },
    {
      "epoch": 0.8022855353788353,
      "grad_norm": 2.0519936084747314,
      "learning_rate": 0.00018724340175953078,
      "loss": 0.8749,
      "step": 5125
    },
    {
      "epoch": 0.8024420788979336,
      "grad_norm": 3.485672950744629,
      "learning_rate": 0.00018721896383186706,
      "loss": 0.6292,
      "step": 5126
    },
    {
      "epoch": 0.802598622417032,
      "grad_norm": 1.4344536066055298,
      "learning_rate": 0.0001871945259042033,
      "loss": 0.6966,
      "step": 5127
    },
    {
      "epoch": 0.8027551659361303,
      "grad_norm": 1.8065955638885498,
      "learning_rate": 0.00018717008797653959,
      "loss": 0.7884,
      "step": 5128
    },
    {
      "epoch": 0.8029117094552286,
      "grad_norm": 3.6765334606170654,
      "learning_rate": 0.00018714565004887586,
      "loss": 1.0263,
      "step": 5129
    },
    {
      "epoch": 0.8030682529743268,
      "grad_norm": 2.0539987087249756,
      "learning_rate": 0.0001871212121212121,
      "loss": 0.5238,
      "step": 5130
    },
    {
      "epoch": 0.8032247964934252,
      "grad_norm": 2.6825201511383057,
      "learning_rate": 0.00018709677419354837,
      "loss": 1.0184,
      "step": 5131
    },
    {
      "epoch": 0.8033813400125235,
      "grad_norm": 3.3649113178253174,
      "learning_rate": 0.00018707233626588464,
      "loss": 0.8083,
      "step": 5132
    },
    {
      "epoch": 0.8035378835316218,
      "grad_norm": 2.283419370651245,
      "learning_rate": 0.0001870478983382209,
      "loss": 0.7746,
      "step": 5133
    },
    {
      "epoch": 0.8036944270507201,
      "grad_norm": 2.9016916751861572,
      "learning_rate": 0.00018702346041055717,
      "loss": 0.9586,
      "step": 5134
    },
    {
      "epoch": 0.8038509705698184,
      "grad_norm": 3.5919530391693115,
      "learning_rate": 0.00018699902248289345,
      "loss": 1.083,
      "step": 5135
    },
    {
      "epoch": 0.8040075140889167,
      "grad_norm": 5.343000888824463,
      "learning_rate": 0.00018697458455522968,
      "loss": 1.187,
      "step": 5136
    },
    {
      "epoch": 0.804164057608015,
      "grad_norm": 2.6350221633911133,
      "learning_rate": 0.00018695014662756596,
      "loss": 1.5941,
      "step": 5137
    },
    {
      "epoch": 0.8043206011271133,
      "grad_norm": 1.6021567583084106,
      "learning_rate": 0.00018692570869990223,
      "loss": 0.6295,
      "step": 5138
    },
    {
      "epoch": 0.8044771446462117,
      "grad_norm": 4.3423075675964355,
      "learning_rate": 0.00018690127077223848,
      "loss": 1.3454,
      "step": 5139
    },
    {
      "epoch": 0.80463368816531,
      "grad_norm": 3.026441812515259,
      "learning_rate": 0.00018687683284457476,
      "loss": 1.0692,
      "step": 5140
    },
    {
      "epoch": 0.8047902316844082,
      "grad_norm": 2.199367046356201,
      "learning_rate": 0.00018685239491691104,
      "loss": 0.7581,
      "step": 5141
    },
    {
      "epoch": 0.8049467752035065,
      "grad_norm": 1.787598967552185,
      "learning_rate": 0.0001868279569892473,
      "loss": 0.8282,
      "step": 5142
    },
    {
      "epoch": 0.8051033187226049,
      "grad_norm": 3.308527708053589,
      "learning_rate": 0.00018680351906158357,
      "loss": 1.3213,
      "step": 5143
    },
    {
      "epoch": 0.8052598622417032,
      "grad_norm": 2.4407522678375244,
      "learning_rate": 0.00018677908113391985,
      "loss": 1.6773,
      "step": 5144
    },
    {
      "epoch": 0.8054164057608015,
      "grad_norm": 0.9789965748786926,
      "learning_rate": 0.00018675464320625607,
      "loss": 0.6245,
      "step": 5145
    },
    {
      "epoch": 0.8055729492798999,
      "grad_norm": 1.3967782258987427,
      "learning_rate": 0.00018673020527859235,
      "loss": 0.976,
      "step": 5146
    },
    {
      "epoch": 0.8057294927989981,
      "grad_norm": 1.6715110540390015,
      "learning_rate": 0.00018670576735092863,
      "loss": 0.8646,
      "step": 5147
    },
    {
      "epoch": 0.8058860363180964,
      "grad_norm": 2.1324195861816406,
      "learning_rate": 0.00018668132942326488,
      "loss": 0.6652,
      "step": 5148
    },
    {
      "epoch": 0.8060425798371947,
      "grad_norm": 1.6442357301712036,
      "learning_rate": 0.00018665689149560116,
      "loss": 0.8102,
      "step": 5149
    },
    {
      "epoch": 0.8061991233562931,
      "grad_norm": 2.52543044090271,
      "learning_rate": 0.00018663245356793744,
      "loss": 0.9856,
      "step": 5150
    },
    {
      "epoch": 0.8063556668753914,
      "grad_norm": 0.6359297633171082,
      "learning_rate": 0.0001866080156402737,
      "loss": 0.284,
      "step": 5151
    },
    {
      "epoch": 0.8065122103944896,
      "grad_norm": 0.6601283550262451,
      "learning_rate": 0.00018658357771260997,
      "loss": 0.4191,
      "step": 5152
    },
    {
      "epoch": 0.8066687539135879,
      "grad_norm": 0.5493868589401245,
      "learning_rate": 0.00018655913978494622,
      "loss": 0.3558,
      "step": 5153
    },
    {
      "epoch": 0.8068252974326863,
      "grad_norm": 0.6157817244529724,
      "learning_rate": 0.00018653470185728247,
      "loss": 0.355,
      "step": 5154
    },
    {
      "epoch": 0.8069818409517846,
      "grad_norm": 0.5704903602600098,
      "learning_rate": 0.00018651026392961875,
      "loss": 0.3663,
      "step": 5155
    },
    {
      "epoch": 0.8071383844708829,
      "grad_norm": 1.178697943687439,
      "learning_rate": 0.00018648582600195503,
      "loss": 0.5071,
      "step": 5156
    },
    {
      "epoch": 0.8072949279899813,
      "grad_norm": 0.8343028426170349,
      "learning_rate": 0.00018646138807429128,
      "loss": 0.4386,
      "step": 5157
    },
    {
      "epoch": 0.8074514715090795,
      "grad_norm": 0.592948317527771,
      "learning_rate": 0.00018643695014662756,
      "loss": 0.2441,
      "step": 5158
    },
    {
      "epoch": 0.8076080150281778,
      "grad_norm": 0.7243725061416626,
      "learning_rate": 0.00018641251221896383,
      "loss": 0.3678,
      "step": 5159
    },
    {
      "epoch": 0.8077645585472761,
      "grad_norm": 0.8186553120613098,
      "learning_rate": 0.00018638807429130006,
      "loss": 0.3722,
      "step": 5160
    },
    {
      "epoch": 0.8079211020663745,
      "grad_norm": 1.4765008687973022,
      "learning_rate": 0.00018636363636363634,
      "loss": 0.2819,
      "step": 5161
    },
    {
      "epoch": 0.8080776455854728,
      "grad_norm": 1.4610421657562256,
      "learning_rate": 0.00018633919843597262,
      "loss": 0.3065,
      "step": 5162
    },
    {
      "epoch": 0.8082341891045711,
      "grad_norm": 1.0413181781768799,
      "learning_rate": 0.00018631476050830887,
      "loss": 0.3565,
      "step": 5163
    },
    {
      "epoch": 0.8083907326236693,
      "grad_norm": 9.378120422363281,
      "learning_rate": 0.00018629032258064515,
      "loss": 0.9926,
      "step": 5164
    },
    {
      "epoch": 0.8085472761427677,
      "grad_norm": 1.3265831470489502,
      "learning_rate": 0.00018626588465298142,
      "loss": 0.5953,
      "step": 5165
    },
    {
      "epoch": 0.808703819661866,
      "grad_norm": 2.425917863845825,
      "learning_rate": 0.00018624144672531768,
      "loss": 0.4746,
      "step": 5166
    },
    {
      "epoch": 0.8088603631809643,
      "grad_norm": 1.6301259994506836,
      "learning_rate": 0.00018621700879765395,
      "loss": 0.4585,
      "step": 5167
    },
    {
      "epoch": 0.8090169067000627,
      "grad_norm": 3.910301446914673,
      "learning_rate": 0.00018619257086999023,
      "loss": 1.0197,
      "step": 5168
    },
    {
      "epoch": 0.8091734502191609,
      "grad_norm": 1.4134447574615479,
      "learning_rate": 0.00018616813294232646,
      "loss": 0.5156,
      "step": 5169
    },
    {
      "epoch": 0.8093299937382592,
      "grad_norm": 1.712280035018921,
      "learning_rate": 0.00018614369501466273,
      "loss": 1.0762,
      "step": 5170
    },
    {
      "epoch": 0.8094865372573575,
      "grad_norm": 1.478226900100708,
      "learning_rate": 0.000186119257086999,
      "loss": 0.4157,
      "step": 5171
    },
    {
      "epoch": 0.8096430807764559,
      "grad_norm": 1.3933120965957642,
      "learning_rate": 0.00018609481915933526,
      "loss": 0.4726,
      "step": 5172
    },
    {
      "epoch": 0.8097996242955542,
      "grad_norm": 1.6714564561843872,
      "learning_rate": 0.00018607038123167154,
      "loss": 0.4534,
      "step": 5173
    },
    {
      "epoch": 0.8099561678146525,
      "grad_norm": 1.5796337127685547,
      "learning_rate": 0.00018604594330400782,
      "loss": 0.689,
      "step": 5174
    },
    {
      "epoch": 0.8101127113337507,
      "grad_norm": 1.966103196144104,
      "learning_rate": 0.00018602150537634407,
      "loss": 0.8875,
      "step": 5175
    },
    {
      "epoch": 0.8102692548528491,
      "grad_norm": 1.9266483783721924,
      "learning_rate": 0.00018599706744868035,
      "loss": 0.7517,
      "step": 5176
    },
    {
      "epoch": 0.8104257983719474,
      "grad_norm": 2.138370990753174,
      "learning_rate": 0.0001859726295210166,
      "loss": 0.6377,
      "step": 5177
    },
    {
      "epoch": 0.8105823418910457,
      "grad_norm": 2.1694908142089844,
      "learning_rate": 0.00018594819159335285,
      "loss": 0.7177,
      "step": 5178
    },
    {
      "epoch": 0.810738885410144,
      "grad_norm": 2.8311069011688232,
      "learning_rate": 0.00018592375366568913,
      "loss": 0.7425,
      "step": 5179
    },
    {
      "epoch": 0.8108954289292424,
      "grad_norm": 3.3369836807250977,
      "learning_rate": 0.0001858993157380254,
      "loss": 1.0659,
      "step": 5180
    },
    {
      "epoch": 0.8110519724483406,
      "grad_norm": 1.3343548774719238,
      "learning_rate": 0.00018587487781036166,
      "loss": 0.6171,
      "step": 5181
    },
    {
      "epoch": 0.8112085159674389,
      "grad_norm": 3.1208863258361816,
      "learning_rate": 0.00018585043988269794,
      "loss": 1.0402,
      "step": 5182
    },
    {
      "epoch": 0.8113650594865373,
      "grad_norm": 2.674014091491699,
      "learning_rate": 0.00018582600195503422,
      "loss": 0.8502,
      "step": 5183
    },
    {
      "epoch": 0.8115216030056356,
      "grad_norm": 2.035822868347168,
      "learning_rate": 0.00018580156402737044,
      "loss": 0.7894,
      "step": 5184
    },
    {
      "epoch": 0.8116781465247339,
      "grad_norm": 3.007331371307373,
      "learning_rate": 0.00018577712609970672,
      "loss": 1.1385,
      "step": 5185
    },
    {
      "epoch": 0.8118346900438321,
      "grad_norm": 1.874494194984436,
      "learning_rate": 0.000185752688172043,
      "loss": 0.7571,
      "step": 5186
    },
    {
      "epoch": 0.8119912335629305,
      "grad_norm": 2.905369758605957,
      "learning_rate": 0.00018572825024437925,
      "loss": 1.2572,
      "step": 5187
    },
    {
      "epoch": 0.8121477770820288,
      "grad_norm": 3.7571399211883545,
      "learning_rate": 0.00018570381231671553,
      "loss": 1.2583,
      "step": 5188
    },
    {
      "epoch": 0.8123043206011271,
      "grad_norm": 2.9104902744293213,
      "learning_rate": 0.0001856793743890518,
      "loss": 1.8115,
      "step": 5189
    },
    {
      "epoch": 0.8124608641202254,
      "grad_norm": 2.058563709259033,
      "learning_rate": 0.00018565493646138806,
      "loss": 1.3185,
      "step": 5190
    },
    {
      "epoch": 0.8126174076393238,
      "grad_norm": 3.4244906902313232,
      "learning_rate": 0.00018563049853372434,
      "loss": 1.1396,
      "step": 5191
    },
    {
      "epoch": 0.812773951158422,
      "grad_norm": 1.9940974712371826,
      "learning_rate": 0.00018560606060606061,
      "loss": 1.322,
      "step": 5192
    },
    {
      "epoch": 0.8129304946775203,
      "grad_norm": 2.131615161895752,
      "learning_rate": 0.00018558162267839684,
      "loss": 1.2332,
      "step": 5193
    },
    {
      "epoch": 0.8130870381966186,
      "grad_norm": 2.615813970565796,
      "learning_rate": 0.00018555718475073312,
      "loss": 0.9671,
      "step": 5194
    },
    {
      "epoch": 0.813243581715717,
      "grad_norm": 1.4775357246398926,
      "learning_rate": 0.0001855327468230694,
      "loss": 1.0884,
      "step": 5195
    },
    {
      "epoch": 0.8134001252348153,
      "grad_norm": 5.091252326965332,
      "learning_rate": 0.00018550830889540565,
      "loss": 0.7112,
      "step": 5196
    },
    {
      "epoch": 0.8135566687539136,
      "grad_norm": 1.498961329460144,
      "learning_rate": 0.00018548387096774192,
      "loss": 0.4947,
      "step": 5197
    },
    {
      "epoch": 0.8137132122730119,
      "grad_norm": 2.671917676925659,
      "learning_rate": 0.0001854594330400782,
      "loss": 0.8125,
      "step": 5198
    },
    {
      "epoch": 0.8138697557921102,
      "grad_norm": 1.2073392868041992,
      "learning_rate": 0.00018543499511241445,
      "loss": 0.6217,
      "step": 5199
    },
    {
      "epoch": 0.8140262993112085,
      "grad_norm": 2.6718733310699463,
      "learning_rate": 0.0001854105571847507,
      "loss": 1.5128,
      "step": 5200
    },
    {
      "epoch": 0.8141828428303068,
      "grad_norm": 0.4015403687953949,
      "learning_rate": 0.00018538611925708698,
      "loss": 0.2958,
      "step": 5201
    },
    {
      "epoch": 0.8143393863494052,
      "grad_norm": 0.7443885803222656,
      "learning_rate": 0.00018536168132942323,
      "loss": 0.211,
      "step": 5202
    },
    {
      "epoch": 0.8144959298685035,
      "grad_norm": 0.7921191453933716,
      "learning_rate": 0.0001853372434017595,
      "loss": 0.3133,
      "step": 5203
    },
    {
      "epoch": 0.8146524733876017,
      "grad_norm": 0.7802867293357849,
      "learning_rate": 0.0001853128054740958,
      "loss": 0.2662,
      "step": 5204
    },
    {
      "epoch": 0.8148090169067,
      "grad_norm": 0.7072981595993042,
      "learning_rate": 0.00018528836754643204,
      "loss": 0.3204,
      "step": 5205
    },
    {
      "epoch": 0.8149655604257984,
      "grad_norm": 0.6935706734657288,
      "learning_rate": 0.00018526392961876832,
      "loss": 0.2573,
      "step": 5206
    },
    {
      "epoch": 0.8151221039448967,
      "grad_norm": 0.6448320746421814,
      "learning_rate": 0.0001852394916911046,
      "loss": 0.305,
      "step": 5207
    },
    {
      "epoch": 0.815278647463995,
      "grad_norm": 0.5979820489883423,
      "learning_rate": 0.00018521505376344082,
      "loss": 0.2683,
      "step": 5208
    },
    {
      "epoch": 0.8154351909830932,
      "grad_norm": 0.9508888125419617,
      "learning_rate": 0.0001851906158357771,
      "loss": 0.3316,
      "step": 5209
    },
    {
      "epoch": 0.8155917345021916,
      "grad_norm": 0.6390425562858582,
      "learning_rate": 0.00018516617790811338,
      "loss": 0.2239,
      "step": 5210
    },
    {
      "epoch": 0.8157482780212899,
      "grad_norm": 0.9599819779396057,
      "learning_rate": 0.00018514173998044963,
      "loss": 0.4099,
      "step": 5211
    },
    {
      "epoch": 0.8159048215403882,
      "grad_norm": 1.198017954826355,
      "learning_rate": 0.0001851173020527859,
      "loss": 0.3733,
      "step": 5212
    },
    {
      "epoch": 0.8160613650594866,
      "grad_norm": 1.6387499570846558,
      "learning_rate": 0.0001850928641251222,
      "loss": 0.3879,
      "step": 5213
    },
    {
      "epoch": 0.8162179085785849,
      "grad_norm": 1.1071934700012207,
      "learning_rate": 0.00018506842619745844,
      "loss": 0.6261,
      "step": 5214
    },
    {
      "epoch": 0.8163744520976831,
      "grad_norm": 0.7644991278648376,
      "learning_rate": 0.00018504398826979472,
      "loss": 0.2503,
      "step": 5215
    },
    {
      "epoch": 0.8165309956167814,
      "grad_norm": 0.9139405488967896,
      "learning_rate": 0.000185019550342131,
      "loss": 0.5133,
      "step": 5216
    },
    {
      "epoch": 0.8166875391358798,
      "grad_norm": 0.9907679557800293,
      "learning_rate": 0.00018499511241446722,
      "loss": 0.4415,
      "step": 5217
    },
    {
      "epoch": 0.8168440826549781,
      "grad_norm": 0.9434964060783386,
      "learning_rate": 0.0001849706744868035,
      "loss": 0.491,
      "step": 5218
    },
    {
      "epoch": 0.8170006261740764,
      "grad_norm": 1.1994105577468872,
      "learning_rate": 0.00018494623655913978,
      "loss": 0.673,
      "step": 5219
    },
    {
      "epoch": 0.8171571696931748,
      "grad_norm": 1.1028125286102295,
      "learning_rate": 0.00018492179863147603,
      "loss": 0.5795,
      "step": 5220
    },
    {
      "epoch": 0.817313713212273,
      "grad_norm": 0.9780545234680176,
      "learning_rate": 0.0001848973607038123,
      "loss": 0.5664,
      "step": 5221
    },
    {
      "epoch": 0.8174702567313713,
      "grad_norm": 2.6875715255737305,
      "learning_rate": 0.00018487292277614859,
      "loss": 0.6478,
      "step": 5222
    },
    {
      "epoch": 0.8176268002504696,
      "grad_norm": 1.439847469329834,
      "learning_rate": 0.00018484848484848484,
      "loss": 0.6378,
      "step": 5223
    },
    {
      "epoch": 0.817783343769568,
      "grad_norm": 1.5552678108215332,
      "learning_rate": 0.0001848240469208211,
      "loss": 0.8491,
      "step": 5224
    },
    {
      "epoch": 0.8179398872886663,
      "grad_norm": 2.3292577266693115,
      "learning_rate": 0.00018479960899315737,
      "loss": 0.8801,
      "step": 5225
    },
    {
      "epoch": 0.8180964308077645,
      "grad_norm": 1.8641091585159302,
      "learning_rate": 0.00018477517106549362,
      "loss": 0.7684,
      "step": 5226
    },
    {
      "epoch": 0.8182529743268628,
      "grad_norm": 1.5867455005645752,
      "learning_rate": 0.0001847507331378299,
      "loss": 0.5417,
      "step": 5227
    },
    {
      "epoch": 0.8184095178459612,
      "grad_norm": 3.227484941482544,
      "learning_rate": 0.00018472629521016617,
      "loss": 1.1114,
      "step": 5228
    },
    {
      "epoch": 0.8185660613650595,
      "grad_norm": 1.3463890552520752,
      "learning_rate": 0.00018470185728250243,
      "loss": 0.6013,
      "step": 5229
    },
    {
      "epoch": 0.8187226048841578,
      "grad_norm": 2.245208501815796,
      "learning_rate": 0.0001846774193548387,
      "loss": 0.82,
      "step": 5230
    },
    {
      "epoch": 0.8188791484032562,
      "grad_norm": 1.4095137119293213,
      "learning_rate": 0.00018465298142717498,
      "loss": 0.6866,
      "step": 5231
    },
    {
      "epoch": 0.8190356919223544,
      "grad_norm": 2.2786121368408203,
      "learning_rate": 0.0001846285434995112,
      "loss": 0.7222,
      "step": 5232
    },
    {
      "epoch": 0.8191922354414527,
      "grad_norm": 1.5390690565109253,
      "learning_rate": 0.00018460410557184748,
      "loss": 0.7086,
      "step": 5233
    },
    {
      "epoch": 0.819348778960551,
      "grad_norm": 1.4006104469299316,
      "learning_rate": 0.00018457966764418376,
      "loss": 0.9285,
      "step": 5234
    },
    {
      "epoch": 0.8195053224796494,
      "grad_norm": 1.5036606788635254,
      "learning_rate": 0.00018455522971652001,
      "loss": 0.7367,
      "step": 5235
    },
    {
      "epoch": 0.8196618659987477,
      "grad_norm": 1.538991928100586,
      "learning_rate": 0.0001845307917888563,
      "loss": 0.7435,
      "step": 5236
    },
    {
      "epoch": 0.819818409517846,
      "grad_norm": 1.5713046789169312,
      "learning_rate": 0.00018450635386119257,
      "loss": 0.7751,
      "step": 5237
    },
    {
      "epoch": 0.8199749530369442,
      "grad_norm": 2.501213788986206,
      "learning_rate": 0.00018448191593352882,
      "loss": 1.1718,
      "step": 5238
    },
    {
      "epoch": 0.8201314965560426,
      "grad_norm": 2.6545050144195557,
      "learning_rate": 0.0001844574780058651,
      "loss": 1.362,
      "step": 5239
    },
    {
      "epoch": 0.8202880400751409,
      "grad_norm": 2.392634153366089,
      "learning_rate": 0.00018443304007820138,
      "loss": 1.4286,
      "step": 5240
    },
    {
      "epoch": 0.8204445835942392,
      "grad_norm": 1.6317548751831055,
      "learning_rate": 0.0001844086021505376,
      "loss": 1.2579,
      "step": 5241
    },
    {
      "epoch": 0.8206011271133375,
      "grad_norm": 8.814602851867676,
      "learning_rate": 0.00018438416422287388,
      "loss": 1.1427,
      "step": 5242
    },
    {
      "epoch": 0.8207576706324358,
      "grad_norm": 3.1028459072113037,
      "learning_rate": 0.00018435972629521016,
      "loss": 1.6293,
      "step": 5243
    },
    {
      "epoch": 0.8209142141515341,
      "grad_norm": 2.7763304710388184,
      "learning_rate": 0.0001843352883675464,
      "loss": 1.5323,
      "step": 5244
    },
    {
      "epoch": 0.8210707576706324,
      "grad_norm": 3.5093159675598145,
      "learning_rate": 0.0001843108504398827,
      "loss": 2.2438,
      "step": 5245
    },
    {
      "epoch": 0.8212273011897308,
      "grad_norm": 2.848104953765869,
      "learning_rate": 0.00018428641251221897,
      "loss": 1.2866,
      "step": 5246
    },
    {
      "epoch": 0.8213838447088291,
      "grad_norm": 1.3957146406173706,
      "learning_rate": 0.00018426197458455522,
      "loss": 0.6522,
      "step": 5247
    },
    {
      "epoch": 0.8215403882279274,
      "grad_norm": 4.863888740539551,
      "learning_rate": 0.00018423753665689147,
      "loss": 0.663,
      "step": 5248
    },
    {
      "epoch": 0.8216969317470256,
      "grad_norm": 2.6501758098602295,
      "learning_rate": 0.00018421309872922775,
      "loss": 1.2547,
      "step": 5249
    },
    {
      "epoch": 0.821853475266124,
      "grad_norm": 1.6998775005340576,
      "learning_rate": 0.000184188660801564,
      "loss": 1.0807,
      "step": 5250
    },
    {
      "epoch": 0.8220100187852223,
      "grad_norm": 0.7901161909103394,
      "learning_rate": 0.00018416422287390028,
      "loss": 0.3135,
      "step": 5251
    },
    {
      "epoch": 0.8221665623043206,
      "grad_norm": 0.7049028277397156,
      "learning_rate": 0.00018413978494623656,
      "loss": 0.4211,
      "step": 5252
    },
    {
      "epoch": 0.8223231058234189,
      "grad_norm": 0.6811177134513855,
      "learning_rate": 0.0001841153470185728,
      "loss": 0.2857,
      "step": 5253
    },
    {
      "epoch": 0.8224796493425173,
      "grad_norm": 0.7868577241897583,
      "learning_rate": 0.00018409090909090909,
      "loss": 0.35,
      "step": 5254
    },
    {
      "epoch": 0.8226361928616155,
      "grad_norm": 0.7949536442756653,
      "learning_rate": 0.00018406647116324536,
      "loss": 0.4374,
      "step": 5255
    },
    {
      "epoch": 0.8227927363807138,
      "grad_norm": 0.6903027296066284,
      "learning_rate": 0.0001840420332355816,
      "loss": 0.427,
      "step": 5256
    },
    {
      "epoch": 0.8229492798998121,
      "grad_norm": 0.6659723520278931,
      "learning_rate": 0.00018401759530791787,
      "loss": 0.5036,
      "step": 5257
    },
    {
      "epoch": 0.8231058234189105,
      "grad_norm": 0.5568656325340271,
      "learning_rate": 0.00018399315738025415,
      "loss": 0.3561,
      "step": 5258
    },
    {
      "epoch": 0.8232623669380088,
      "grad_norm": 0.8612244129180908,
      "learning_rate": 0.0001839687194525904,
      "loss": 0.3908,
      "step": 5259
    },
    {
      "epoch": 0.823418910457107,
      "grad_norm": 0.7384498119354248,
      "learning_rate": 0.00018394428152492667,
      "loss": 0.5201,
      "step": 5260
    },
    {
      "epoch": 0.8235754539762054,
      "grad_norm": 0.9855079054832458,
      "learning_rate": 0.00018391984359726295,
      "loss": 0.413,
      "step": 5261
    },
    {
      "epoch": 0.8237319974953037,
      "grad_norm": 0.768484890460968,
      "learning_rate": 0.0001838954056695992,
      "loss": 0.4253,
      "step": 5262
    },
    {
      "epoch": 0.823888541014402,
      "grad_norm": 0.6024149060249329,
      "learning_rate": 0.00018387096774193548,
      "loss": 0.2572,
      "step": 5263
    },
    {
      "epoch": 0.8240450845335003,
      "grad_norm": 0.5036283135414124,
      "learning_rate": 0.00018384652981427176,
      "loss": 0.2692,
      "step": 5264
    },
    {
      "epoch": 0.8242016280525987,
      "grad_norm": 1.0418510437011719,
      "learning_rate": 0.00018382209188660799,
      "loss": 0.4799,
      "step": 5265
    },
    {
      "epoch": 0.8243581715716969,
      "grad_norm": 0.663507342338562,
      "learning_rate": 0.00018379765395894426,
      "loss": 0.3185,
      "step": 5266
    },
    {
      "epoch": 0.8245147150907952,
      "grad_norm": 0.8213881254196167,
      "learning_rate": 0.00018377321603128054,
      "loss": 0.337,
      "step": 5267
    },
    {
      "epoch": 0.8246712586098935,
      "grad_norm": 1.1728672981262207,
      "learning_rate": 0.0001837487781036168,
      "loss": 0.4151,
      "step": 5268
    },
    {
      "epoch": 0.8248278021289919,
      "grad_norm": 1.2628549337387085,
      "learning_rate": 0.00018372434017595307,
      "loss": 0.6478,
      "step": 5269
    },
    {
      "epoch": 0.8249843456480902,
      "grad_norm": 1.121236801147461,
      "learning_rate": 0.00018369990224828935,
      "loss": 0.4379,
      "step": 5270
    },
    {
      "epoch": 0.8251408891671885,
      "grad_norm": 1.1426113843917847,
      "learning_rate": 0.00018367546432062557,
      "loss": 0.461,
      "step": 5271
    },
    {
      "epoch": 0.8252974326862867,
      "grad_norm": 1.555210828781128,
      "learning_rate": 0.00018365102639296185,
      "loss": 0.493,
      "step": 5272
    },
    {
      "epoch": 0.8254539762053851,
      "grad_norm": 1.8274154663085938,
      "learning_rate": 0.00018362658846529813,
      "loss": 0.5785,
      "step": 5273
    },
    {
      "epoch": 0.8256105197244834,
      "grad_norm": 1.771406888961792,
      "learning_rate": 0.00018360215053763438,
      "loss": 0.71,
      "step": 5274
    },
    {
      "epoch": 0.8257670632435817,
      "grad_norm": 3.0321593284606934,
      "learning_rate": 0.00018357771260997066,
      "loss": 0.8284,
      "step": 5275
    },
    {
      "epoch": 0.8259236067626801,
      "grad_norm": 4.053110599517822,
      "learning_rate": 0.00018355327468230694,
      "loss": 1.0345,
      "step": 5276
    },
    {
      "epoch": 0.8260801502817783,
      "grad_norm": 1.7022113800048828,
      "learning_rate": 0.0001835288367546432,
      "loss": 0.8215,
      "step": 5277
    },
    {
      "epoch": 0.8262366938008766,
      "grad_norm": 2.7974987030029297,
      "learning_rate": 0.00018350439882697947,
      "loss": 1.186,
      "step": 5278
    },
    {
      "epoch": 0.8263932373199749,
      "grad_norm": 2.123547077178955,
      "learning_rate": 0.00018347996089931575,
      "loss": 0.7569,
      "step": 5279
    },
    {
      "epoch": 0.8265497808390733,
      "grad_norm": 2.535295009613037,
      "learning_rate": 0.00018345552297165197,
      "loss": 1.1509,
      "step": 5280
    },
    {
      "epoch": 0.8267063243581716,
      "grad_norm": 2.005593776702881,
      "learning_rate": 0.00018343108504398825,
      "loss": 0.6985,
      "step": 5281
    },
    {
      "epoch": 0.8268628678772699,
      "grad_norm": 2.139389753341675,
      "learning_rate": 0.00018340664711632453,
      "loss": 1.401,
      "step": 5282
    },
    {
      "epoch": 0.8270194113963681,
      "grad_norm": 1.5043542385101318,
      "learning_rate": 0.00018338220918866078,
      "loss": 0.829,
      "step": 5283
    },
    {
      "epoch": 0.8271759549154665,
      "grad_norm": 1.6007672548294067,
      "learning_rate": 0.00018335777126099706,
      "loss": 0.8851,
      "step": 5284
    },
    {
      "epoch": 0.8273324984345648,
      "grad_norm": 2.1892812252044678,
      "learning_rate": 0.00018333333333333334,
      "loss": 1.3312,
      "step": 5285
    },
    {
      "epoch": 0.8274890419536631,
      "grad_norm": 1.2320849895477295,
      "learning_rate": 0.0001833088954056696,
      "loss": 0.6038,
      "step": 5286
    },
    {
      "epoch": 0.8276455854727615,
      "grad_norm": 4.166670322418213,
      "learning_rate": 0.00018328445747800586,
      "loss": 0.669,
      "step": 5287
    },
    {
      "epoch": 0.8278021289918598,
      "grad_norm": 2.1780524253845215,
      "learning_rate": 0.00018326001955034214,
      "loss": 1.1425,
      "step": 5288
    },
    {
      "epoch": 0.827958672510958,
      "grad_norm": 2.1687676906585693,
      "learning_rate": 0.00018323558162267837,
      "loss": 1.1492,
      "step": 5289
    },
    {
      "epoch": 0.8281152160300563,
      "grad_norm": 1.6996123790740967,
      "learning_rate": 0.00018321114369501465,
      "loss": 1.0186,
      "step": 5290
    },
    {
      "epoch": 0.8282717595491547,
      "grad_norm": 3.4165749549865723,
      "learning_rate": 0.00018318670576735092,
      "loss": 1.1482,
      "step": 5291
    },
    {
      "epoch": 0.828428303068253,
      "grad_norm": 2.1775505542755127,
      "learning_rate": 0.00018316226783968718,
      "loss": 1.3908,
      "step": 5292
    },
    {
      "epoch": 0.8285848465873513,
      "grad_norm": 2.812617063522339,
      "learning_rate": 0.00018313782991202345,
      "loss": 1.1503,
      "step": 5293
    },
    {
      "epoch": 0.8287413901064495,
      "grad_norm": 1.5233439207077026,
      "learning_rate": 0.00018311339198435973,
      "loss": 1.3862,
      "step": 5294
    },
    {
      "epoch": 0.8288979336255479,
      "grad_norm": 1.8722327947616577,
      "learning_rate": 0.00018308895405669596,
      "loss": 1.5704,
      "step": 5295
    },
    {
      "epoch": 0.8290544771446462,
      "grad_norm": 2.6162047386169434,
      "learning_rate": 0.00018306451612903223,
      "loss": 1.0679,
      "step": 5296
    },
    {
      "epoch": 0.8292110206637445,
      "grad_norm": 1.2465320825576782,
      "learning_rate": 0.0001830400782013685,
      "loss": 0.5608,
      "step": 5297
    },
    {
      "epoch": 0.8293675641828429,
      "grad_norm": 1.294814109802246,
      "learning_rate": 0.00018301564027370476,
      "loss": 0.8224,
      "step": 5298
    },
    {
      "epoch": 0.8295241077019412,
      "grad_norm": 2.261251926422119,
      "learning_rate": 0.00018299120234604104,
      "loss": 1.3591,
      "step": 5299
    },
    {
      "epoch": 0.8296806512210394,
      "grad_norm": 1.4905081987380981,
      "learning_rate": 0.00018296676441837732,
      "loss": 1.4874,
      "step": 5300
    },
    {
      "epoch": 0.8298371947401377,
      "grad_norm": 0.5138626098632812,
      "learning_rate": 0.00018294232649071357,
      "loss": 0.2785,
      "step": 5301
    },
    {
      "epoch": 0.8299937382592361,
      "grad_norm": 0.5530992150306702,
      "learning_rate": 0.00018291788856304985,
      "loss": 0.3083,
      "step": 5302
    },
    {
      "epoch": 0.8301502817783344,
      "grad_norm": 0.40813249349594116,
      "learning_rate": 0.00018289345063538613,
      "loss": 0.3063,
      "step": 5303
    },
    {
      "epoch": 0.8303068252974327,
      "grad_norm": 0.5321076512336731,
      "learning_rate": 0.00018286901270772235,
      "loss": 0.3811,
      "step": 5304
    },
    {
      "epoch": 0.830463368816531,
      "grad_norm": 0.5852525234222412,
      "learning_rate": 0.00018284457478005863,
      "loss": 0.3492,
      "step": 5305
    },
    {
      "epoch": 0.8306199123356293,
      "grad_norm": 0.7372064590454102,
      "learning_rate": 0.0001828201368523949,
      "loss": 0.4506,
      "step": 5306
    },
    {
      "epoch": 0.8307764558547276,
      "grad_norm": 0.6820610761642456,
      "learning_rate": 0.00018279569892473116,
      "loss": 0.3694,
      "step": 5307
    },
    {
      "epoch": 0.8309329993738259,
      "grad_norm": 1.2214698791503906,
      "learning_rate": 0.00018277126099706744,
      "loss": 0.4984,
      "step": 5308
    },
    {
      "epoch": 0.8310895428929242,
      "grad_norm": 1.288681983947754,
      "learning_rate": 0.00018274682306940372,
      "loss": 0.6551,
      "step": 5309
    },
    {
      "epoch": 0.8312460864120226,
      "grad_norm": 1.0131791830062866,
      "learning_rate": 0.00018272238514173997,
      "loss": 0.3613,
      "step": 5310
    },
    {
      "epoch": 0.8314026299311209,
      "grad_norm": 2.0954971313476562,
      "learning_rate": 0.00018269794721407625,
      "loss": 0.5802,
      "step": 5311
    },
    {
      "epoch": 0.8315591734502191,
      "grad_norm": 0.8063324093818665,
      "learning_rate": 0.0001826735092864125,
      "loss": 0.4414,
      "step": 5312
    },
    {
      "epoch": 0.8317157169693175,
      "grad_norm": 1.2960774898529053,
      "learning_rate": 0.00018264907135874875,
      "loss": 0.485,
      "step": 5313
    },
    {
      "epoch": 0.8318722604884158,
      "grad_norm": 0.9982731342315674,
      "learning_rate": 0.00018262463343108503,
      "loss": 0.5702,
      "step": 5314
    },
    {
      "epoch": 0.8320288040075141,
      "grad_norm": 2.1627659797668457,
      "learning_rate": 0.0001826001955034213,
      "loss": 0.5258,
      "step": 5315
    },
    {
      "epoch": 0.8321853475266124,
      "grad_norm": 1.2846755981445312,
      "learning_rate": 0.00018257575757575756,
      "loss": 0.6247,
      "step": 5316
    },
    {
      "epoch": 0.8323418910457107,
      "grad_norm": 1.6041959524154663,
      "learning_rate": 0.00018255131964809384,
      "loss": 0.5156,
      "step": 5317
    },
    {
      "epoch": 0.832498434564809,
      "grad_norm": 0.9407545328140259,
      "learning_rate": 0.00018252688172043011,
      "loss": 0.3831,
      "step": 5318
    },
    {
      "epoch": 0.8326549780839073,
      "grad_norm": 1.5402854681015015,
      "learning_rate": 0.00018250244379276634,
      "loss": 0.9489,
      "step": 5319
    },
    {
      "epoch": 0.8328115216030056,
      "grad_norm": 0.9646233916282654,
      "learning_rate": 0.00018247800586510262,
      "loss": 0.5643,
      "step": 5320
    },
    {
      "epoch": 0.832968065122104,
      "grad_norm": 1.2722687721252441,
      "learning_rate": 0.0001824535679374389,
      "loss": 0.5555,
      "step": 5321
    },
    {
      "epoch": 0.8331246086412023,
      "grad_norm": 2.5088601112365723,
      "learning_rate": 0.00018242913000977515,
      "loss": 0.904,
      "step": 5322
    },
    {
      "epoch": 0.8332811521603005,
      "grad_norm": 1.9677549600601196,
      "learning_rate": 0.00018240469208211142,
      "loss": 0.7769,
      "step": 5323
    },
    {
      "epoch": 0.8334376956793988,
      "grad_norm": 2.1917471885681152,
      "learning_rate": 0.0001823802541544477,
      "loss": 0.8852,
      "step": 5324
    },
    {
      "epoch": 0.8335942391984972,
      "grad_norm": 2.5052521228790283,
      "learning_rate": 0.00018235581622678395,
      "loss": 0.8834,
      "step": 5325
    },
    {
      "epoch": 0.8337507827175955,
      "grad_norm": 1.5879862308502197,
      "learning_rate": 0.00018233137829912023,
      "loss": 0.8438,
      "step": 5326
    },
    {
      "epoch": 0.8339073262366938,
      "grad_norm": 2.1139187812805176,
      "learning_rate": 0.0001823069403714565,
      "loss": 0.5548,
      "step": 5327
    },
    {
      "epoch": 0.8340638697557922,
      "grad_norm": 1.9790080785751343,
      "learning_rate": 0.00018228250244379274,
      "loss": 0.9604,
      "step": 5328
    },
    {
      "epoch": 0.8342204132748904,
      "grad_norm": 2.8384530544281006,
      "learning_rate": 0.00018225806451612901,
      "loss": 0.7413,
      "step": 5329
    },
    {
      "epoch": 0.8343769567939887,
      "grad_norm": 2.2223446369171143,
      "learning_rate": 0.0001822336265884653,
      "loss": 0.5664,
      "step": 5330
    },
    {
      "epoch": 0.834533500313087,
      "grad_norm": 2.0191867351531982,
      "learning_rate": 0.00018220918866080154,
      "loss": 0.6546,
      "step": 5331
    },
    {
      "epoch": 0.8346900438321854,
      "grad_norm": 2.5596179962158203,
      "learning_rate": 0.00018218475073313782,
      "loss": 0.7337,
      "step": 5332
    },
    {
      "epoch": 0.8348465873512837,
      "grad_norm": 1.8105648756027222,
      "learning_rate": 0.0001821603128054741,
      "loss": 0.6803,
      "step": 5333
    },
    {
      "epoch": 0.8350031308703819,
      "grad_norm": 1.6906538009643555,
      "learning_rate": 0.00018213587487781035,
      "loss": 0.8126,
      "step": 5334
    },
    {
      "epoch": 0.8351596743894802,
      "grad_norm": 1.9950919151306152,
      "learning_rate": 0.00018211143695014663,
      "loss": 0.7731,
      "step": 5335
    },
    {
      "epoch": 0.8353162179085786,
      "grad_norm": 1.5140913724899292,
      "learning_rate": 0.00018208699902248288,
      "loss": 0.7505,
      "step": 5336
    },
    {
      "epoch": 0.8354727614276769,
      "grad_norm": 4.230485439300537,
      "learning_rate": 0.00018206256109481913,
      "loss": 1.1235,
      "step": 5337
    },
    {
      "epoch": 0.8356293049467752,
      "grad_norm": 2.215449094772339,
      "learning_rate": 0.0001820381231671554,
      "loss": 1.1191,
      "step": 5338
    },
    {
      "epoch": 0.8357858484658736,
      "grad_norm": 2.6537933349609375,
      "learning_rate": 0.0001820136852394917,
      "loss": 0.9407,
      "step": 5339
    },
    {
      "epoch": 0.8359423919849718,
      "grad_norm": 1.676081895828247,
      "learning_rate": 0.00018198924731182794,
      "loss": 0.6755,
      "step": 5340
    },
    {
      "epoch": 0.8360989355040701,
      "grad_norm": 3.5831100940704346,
      "learning_rate": 0.00018196480938416422,
      "loss": 0.9189,
      "step": 5341
    },
    {
      "epoch": 0.8362554790231684,
      "grad_norm": 2.116856575012207,
      "learning_rate": 0.0001819403714565005,
      "loss": 0.8016,
      "step": 5342
    },
    {
      "epoch": 0.8364120225422668,
      "grad_norm": 4.248983383178711,
      "learning_rate": 0.00018191593352883672,
      "loss": 1.1484,
      "step": 5343
    },
    {
      "epoch": 0.8365685660613651,
      "grad_norm": 2.514638900756836,
      "learning_rate": 0.000181891495601173,
      "loss": 1.1032,
      "step": 5344
    },
    {
      "epoch": 0.8367251095804634,
      "grad_norm": 2.494845390319824,
      "learning_rate": 0.00018186705767350928,
      "loss": 0.8824,
      "step": 5345
    },
    {
      "epoch": 0.8368816530995616,
      "grad_norm": 1.30543053150177,
      "learning_rate": 0.00018184261974584553,
      "loss": 0.596,
      "step": 5346
    },
    {
      "epoch": 0.83703819661866,
      "grad_norm": 2.425416946411133,
      "learning_rate": 0.0001818181818181818,
      "loss": 1.4802,
      "step": 5347
    },
    {
      "epoch": 0.8371947401377583,
      "grad_norm": 1.7893593311309814,
      "learning_rate": 0.00018179374389051809,
      "loss": 0.4148,
      "step": 5348
    },
    {
      "epoch": 0.8373512836568566,
      "grad_norm": 3.8236687183380127,
      "learning_rate": 0.00018176930596285434,
      "loss": 0.7683,
      "step": 5349
    },
    {
      "epoch": 0.837507827175955,
      "grad_norm": 1.503875494003296,
      "learning_rate": 0.00018174486803519062,
      "loss": 0.886,
      "step": 5350
    },
    {
      "epoch": 0.8376643706950532,
      "grad_norm": 0.4786723256111145,
      "learning_rate": 0.0001817204301075269,
      "loss": 0.2861,
      "step": 5351
    },
    {
      "epoch": 0.8378209142141515,
      "grad_norm": 0.8675291538238525,
      "learning_rate": 0.00018169599217986312,
      "loss": 0.4324,
      "step": 5352
    },
    {
      "epoch": 0.8379774577332498,
      "grad_norm": 0.8914163708686829,
      "learning_rate": 0.0001816715542521994,
      "loss": 0.3441,
      "step": 5353
    },
    {
      "epoch": 0.8381340012523482,
      "grad_norm": 0.8245097994804382,
      "learning_rate": 0.00018164711632453567,
      "loss": 0.3934,
      "step": 5354
    },
    {
      "epoch": 0.8382905447714465,
      "grad_norm": 0.8394196033477783,
      "learning_rate": 0.00018162267839687193,
      "loss": 0.2701,
      "step": 5355
    },
    {
      "epoch": 0.8384470882905448,
      "grad_norm": 1.792810320854187,
      "learning_rate": 0.0001815982404692082,
      "loss": 0.4637,
      "step": 5356
    },
    {
      "epoch": 0.838603631809643,
      "grad_norm": 0.979446291923523,
      "learning_rate": 0.00018157380254154448,
      "loss": 0.2882,
      "step": 5357
    },
    {
      "epoch": 0.8387601753287414,
      "grad_norm": 0.9786679148674011,
      "learning_rate": 0.00018154936461388073,
      "loss": 0.3115,
      "step": 5358
    },
    {
      "epoch": 0.8389167188478397,
      "grad_norm": 0.618453860282898,
      "learning_rate": 0.00018152492668621698,
      "loss": 0.3731,
      "step": 5359
    },
    {
      "epoch": 0.839073262366938,
      "grad_norm": 0.9989930391311646,
      "learning_rate": 0.00018150048875855326,
      "loss": 0.3479,
      "step": 5360
    },
    {
      "epoch": 0.8392298058860364,
      "grad_norm": 0.7567367553710938,
      "learning_rate": 0.00018147605083088951,
      "loss": 0.3715,
      "step": 5361
    },
    {
      "epoch": 0.8393863494051347,
      "grad_norm": 1.233569622039795,
      "learning_rate": 0.0001814516129032258,
      "loss": 0.4816,
      "step": 5362
    },
    {
      "epoch": 0.8395428929242329,
      "grad_norm": 1.9397759437561035,
      "learning_rate": 0.00018142717497556207,
      "loss": 0.6499,
      "step": 5363
    },
    {
      "epoch": 0.8396994364433312,
      "grad_norm": 0.8263518810272217,
      "learning_rate": 0.00018140273704789832,
      "loss": 0.3907,
      "step": 5364
    },
    {
      "epoch": 0.8398559799624296,
      "grad_norm": 0.7921581864356995,
      "learning_rate": 0.0001813782991202346,
      "loss": 0.3535,
      "step": 5365
    },
    {
      "epoch": 0.8400125234815279,
      "grad_norm": 1.2115155458450317,
      "learning_rate": 0.00018135386119257088,
      "loss": 0.5176,
      "step": 5366
    },
    {
      "epoch": 0.8401690670006262,
      "grad_norm": 1.769970417022705,
      "learning_rate": 0.0001813294232649071,
      "loss": 0.6016,
      "step": 5367
    },
    {
      "epoch": 0.8403256105197244,
      "grad_norm": 0.8580262660980225,
      "learning_rate": 0.00018130498533724338,
      "loss": 0.4667,
      "step": 5368
    },
    {
      "epoch": 0.8404821540388228,
      "grad_norm": 0.9332250356674194,
      "learning_rate": 0.00018128054740957966,
      "loss": 0.4062,
      "step": 5369
    },
    {
      "epoch": 0.8406386975579211,
      "grad_norm": 2.4739444255828857,
      "learning_rate": 0.0001812561094819159,
      "loss": 0.5315,
      "step": 5370
    },
    {
      "epoch": 0.8407952410770194,
      "grad_norm": 1.3356553316116333,
      "learning_rate": 0.0001812316715542522,
      "loss": 0.4582,
      "step": 5371
    },
    {
      "epoch": 0.8409517845961177,
      "grad_norm": 1.6975020170211792,
      "learning_rate": 0.00018120723362658844,
      "loss": 0.7177,
      "step": 5372
    },
    {
      "epoch": 0.8411083281152161,
      "grad_norm": 1.566121220588684,
      "learning_rate": 0.00018118279569892472,
      "loss": 0.6598,
      "step": 5373
    },
    {
      "epoch": 0.8412648716343143,
      "grad_norm": 1.5858402252197266,
      "learning_rate": 0.000181158357771261,
      "loss": 1.0747,
      "step": 5374
    },
    {
      "epoch": 0.8414214151534126,
      "grad_norm": 0.9164276719093323,
      "learning_rate": 0.00018113391984359722,
      "loss": 0.4249,
      "step": 5375
    },
    {
      "epoch": 0.841577958672511,
      "grad_norm": 1.1517704725265503,
      "learning_rate": 0.0001811094819159335,
      "loss": 0.4706,
      "step": 5376
    },
    {
      "epoch": 0.8417345021916093,
      "grad_norm": 1.679603099822998,
      "learning_rate": 0.00018108504398826978,
      "loss": 0.6945,
      "step": 5377
    },
    {
      "epoch": 0.8418910457107076,
      "grad_norm": 1.76631760597229,
      "learning_rate": 0.00018106060606060603,
      "loss": 0.8951,
      "step": 5378
    },
    {
      "epoch": 0.8420475892298059,
      "grad_norm": 1.7927947044372559,
      "learning_rate": 0.0001810361681329423,
      "loss": 0.4552,
      "step": 5379
    },
    {
      "epoch": 0.8422041327489042,
      "grad_norm": 2.7323668003082275,
      "learning_rate": 0.00018101173020527859,
      "loss": 0.6823,
      "step": 5380
    },
    {
      "epoch": 0.8423606762680025,
      "grad_norm": 1.3663641214370728,
      "learning_rate": 0.00018098729227761484,
      "loss": 0.9579,
      "step": 5381
    },
    {
      "epoch": 0.8425172197871008,
      "grad_norm": 1.6697648763656616,
      "learning_rate": 0.00018096285434995112,
      "loss": 0.8778,
      "step": 5382
    },
    {
      "epoch": 0.8426737633061991,
      "grad_norm": 1.3440797328948975,
      "learning_rate": 0.00018093841642228737,
      "loss": 0.7617,
      "step": 5383
    },
    {
      "epoch": 0.8428303068252975,
      "grad_norm": 2.0381476879119873,
      "learning_rate": 0.00018091397849462362,
      "loss": 0.6647,
      "step": 5384
    },
    {
      "epoch": 0.8429868503443957,
      "grad_norm": 2.4340994358062744,
      "learning_rate": 0.0001808895405669599,
      "loss": 0.8858,
      "step": 5385
    },
    {
      "epoch": 0.843143393863494,
      "grad_norm": 3.9241204261779785,
      "learning_rate": 0.00018086510263929618,
      "loss": 1.9136,
      "step": 5386
    },
    {
      "epoch": 0.8432999373825923,
      "grad_norm": 1.6787124872207642,
      "learning_rate": 0.00018084066471163243,
      "loss": 1.0138,
      "step": 5387
    },
    {
      "epoch": 0.8434564809016907,
      "grad_norm": 2.1270811557769775,
      "learning_rate": 0.0001808162267839687,
      "loss": 1.1926,
      "step": 5388
    },
    {
      "epoch": 0.843613024420789,
      "grad_norm": 2.4386167526245117,
      "learning_rate": 0.00018079178885630498,
      "loss": 0.8631,
      "step": 5389
    },
    {
      "epoch": 0.8437695679398873,
      "grad_norm": 2.620981216430664,
      "learning_rate": 0.0001807673509286412,
      "loss": 1.6441,
      "step": 5390
    },
    {
      "epoch": 0.8439261114589856,
      "grad_norm": 2.2242493629455566,
      "learning_rate": 0.00018074291300097749,
      "loss": 1.1166,
      "step": 5391
    },
    {
      "epoch": 0.8440826549780839,
      "grad_norm": 2.703012466430664,
      "learning_rate": 0.00018071847507331376,
      "loss": 1.421,
      "step": 5392
    },
    {
      "epoch": 0.8442391984971822,
      "grad_norm": 2.006281852722168,
      "learning_rate": 0.00018069403714565002,
      "loss": 1.2213,
      "step": 5393
    },
    {
      "epoch": 0.8443957420162805,
      "grad_norm": 2.2452259063720703,
      "learning_rate": 0.0001806695992179863,
      "loss": 0.8721,
      "step": 5394
    },
    {
      "epoch": 0.8445522855353789,
      "grad_norm": 2.2399497032165527,
      "learning_rate": 0.00018064516129032257,
      "loss": 0.7445,
      "step": 5395
    },
    {
      "epoch": 0.8447088290544772,
      "grad_norm": 1.1072652339935303,
      "learning_rate": 0.00018062072336265882,
      "loss": 0.4617,
      "step": 5396
    },
    {
      "epoch": 0.8448653725735754,
      "grad_norm": 4.509149074554443,
      "learning_rate": 0.0001805962854349951,
      "loss": 1.5535,
      "step": 5397
    },
    {
      "epoch": 0.8450219160926737,
      "grad_norm": 4.23785924911499,
      "learning_rate": 0.00018057184750733138,
      "loss": 0.9141,
      "step": 5398
    },
    {
      "epoch": 0.8451784596117721,
      "grad_norm": 2.642498254776001,
      "learning_rate": 0.0001805474095796676,
      "loss": 1.5931,
      "step": 5399
    },
    {
      "epoch": 0.8453350031308704,
      "grad_norm": 2.1803855895996094,
      "learning_rate": 0.00018052297165200388,
      "loss": 0.9882,
      "step": 5400
    },
    {
      "epoch": 0.8454915466499687,
      "grad_norm": 0.598574161529541,
      "learning_rate": 0.00018049853372434016,
      "loss": 0.2871,
      "step": 5401
    },
    {
      "epoch": 0.845648090169067,
      "grad_norm": 0.48999595642089844,
      "learning_rate": 0.0001804740957966764,
      "loss": 0.2676,
      "step": 5402
    },
    {
      "epoch": 0.8458046336881653,
      "grad_norm": 0.6198676228523254,
      "learning_rate": 0.0001804496578690127,
      "loss": 0.2758,
      "step": 5403
    },
    {
      "epoch": 0.8459611772072636,
      "grad_norm": 1.454820990562439,
      "learning_rate": 0.00018042521994134897,
      "loss": 0.3911,
      "step": 5404
    },
    {
      "epoch": 0.8461177207263619,
      "grad_norm": 0.7595102787017822,
      "learning_rate": 0.00018040078201368522,
      "loss": 0.2041,
      "step": 5405
    },
    {
      "epoch": 0.8462742642454603,
      "grad_norm": 0.6085957288742065,
      "learning_rate": 0.0001803763440860215,
      "loss": 0.3308,
      "step": 5406
    },
    {
      "epoch": 0.8464308077645586,
      "grad_norm": 0.7562193274497986,
      "learning_rate": 0.00018035190615835775,
      "loss": 0.2946,
      "step": 5407
    },
    {
      "epoch": 0.8465873512836568,
      "grad_norm": 0.44967907667160034,
      "learning_rate": 0.000180327468230694,
      "loss": 0.2354,
      "step": 5408
    },
    {
      "epoch": 0.8467438948027551,
      "grad_norm": 0.9210385680198669,
      "learning_rate": 0.00018030303030303028,
      "loss": 0.5657,
      "step": 5409
    },
    {
      "epoch": 0.8469004383218535,
      "grad_norm": 0.6817724704742432,
      "learning_rate": 0.00018027859237536656,
      "loss": 0.3728,
      "step": 5410
    },
    {
      "epoch": 0.8470569818409518,
      "grad_norm": 0.7915209531784058,
      "learning_rate": 0.0001802541544477028,
      "loss": 0.322,
      "step": 5411
    },
    {
      "epoch": 0.8472135253600501,
      "grad_norm": 1.467758297920227,
      "learning_rate": 0.0001802297165200391,
      "loss": 0.3134,
      "step": 5412
    },
    {
      "epoch": 0.8473700688791485,
      "grad_norm": 1.2615087032318115,
      "learning_rate": 0.00018020527859237537,
      "loss": 0.4713,
      "step": 5413
    },
    {
      "epoch": 0.8475266123982467,
      "grad_norm": 0.9758779406547546,
      "learning_rate": 0.0001801808406647116,
      "loss": 0.3469,
      "step": 5414
    },
    {
      "epoch": 0.847683155917345,
      "grad_norm": 1.8404748439788818,
      "learning_rate": 0.00018015640273704787,
      "loss": 0.3852,
      "step": 5415
    },
    {
      "epoch": 0.8478396994364433,
      "grad_norm": 1.1955705881118774,
      "learning_rate": 0.00018013196480938415,
      "loss": 0.9143,
      "step": 5416
    },
    {
      "epoch": 0.8479962429555417,
      "grad_norm": 1.832085132598877,
      "learning_rate": 0.0001801075268817204,
      "loss": 0.6122,
      "step": 5417
    },
    {
      "epoch": 0.84815278647464,
      "grad_norm": 1.7228481769561768,
      "learning_rate": 0.00018008308895405668,
      "loss": 0.6208,
      "step": 5418
    },
    {
      "epoch": 0.8483093299937383,
      "grad_norm": 1.7654926776885986,
      "learning_rate": 0.00018005865102639295,
      "loss": 0.8098,
      "step": 5419
    },
    {
      "epoch": 0.8484658735128365,
      "grad_norm": 0.9912778735160828,
      "learning_rate": 0.0001800342130987292,
      "loss": 0.4749,
      "step": 5420
    },
    {
      "epoch": 0.8486224170319349,
      "grad_norm": 0.9740948677062988,
      "learning_rate": 0.00018000977517106548,
      "loss": 0.456,
      "step": 5421
    },
    {
      "epoch": 0.8487789605510332,
      "grad_norm": 1.4612542390823364,
      "learning_rate": 0.00017998533724340176,
      "loss": 0.6931,
      "step": 5422
    },
    {
      "epoch": 0.8489355040701315,
      "grad_norm": 1.1655532121658325,
      "learning_rate": 0.00017996089931573799,
      "loss": 0.4385,
      "step": 5423
    },
    {
      "epoch": 0.8490920475892298,
      "grad_norm": 1.642591953277588,
      "learning_rate": 0.00017993646138807426,
      "loss": 0.7765,
      "step": 5424
    },
    {
      "epoch": 0.8492485911083281,
      "grad_norm": 4.441413879394531,
      "learning_rate": 0.00017991202346041054,
      "loss": 0.8041,
      "step": 5425
    },
    {
      "epoch": 0.8494051346274264,
      "grad_norm": 4.761545181274414,
      "learning_rate": 0.0001798875855327468,
      "loss": 1.3471,
      "step": 5426
    },
    {
      "epoch": 0.8495616781465247,
      "grad_norm": 1.85892915725708,
      "learning_rate": 0.00017986314760508307,
      "loss": 0.9073,
      "step": 5427
    },
    {
      "epoch": 0.849718221665623,
      "grad_norm": 1.2642240524291992,
      "learning_rate": 0.00017983870967741935,
      "loss": 0.5756,
      "step": 5428
    },
    {
      "epoch": 0.8498747651847214,
      "grad_norm": 1.6355515718460083,
      "learning_rate": 0.0001798142717497556,
      "loss": 0.622,
      "step": 5429
    },
    {
      "epoch": 0.8500313087038197,
      "grad_norm": 2.2661969661712646,
      "learning_rate": 0.00017978983382209185,
      "loss": 0.9791,
      "step": 5430
    },
    {
      "epoch": 0.8501878522229179,
      "grad_norm": 2.119992256164551,
      "learning_rate": 0.00017976539589442813,
      "loss": 0.9954,
      "step": 5431
    },
    {
      "epoch": 0.8503443957420163,
      "grad_norm": 1.942277431488037,
      "learning_rate": 0.00017974095796676438,
      "loss": 0.7139,
      "step": 5432
    },
    {
      "epoch": 0.8505009392611146,
      "grad_norm": 2.0111358165740967,
      "learning_rate": 0.00017971652003910066,
      "loss": 1.0877,
      "step": 5433
    },
    {
      "epoch": 0.8506574827802129,
      "grad_norm": 20.57202911376953,
      "learning_rate": 0.00017969208211143694,
      "loss": 1.0568,
      "step": 5434
    },
    {
      "epoch": 0.8508140262993112,
      "grad_norm": 6.058287143707275,
      "learning_rate": 0.0001796676441837732,
      "loss": 1.0449,
      "step": 5435
    },
    {
      "epoch": 0.8509705698184096,
      "grad_norm": 2.6762993335723877,
      "learning_rate": 0.00017964320625610947,
      "loss": 0.7661,
      "step": 5436
    },
    {
      "epoch": 0.8511271133375078,
      "grad_norm": 3.0614171028137207,
      "learning_rate": 0.00017961876832844575,
      "loss": 0.7657,
      "step": 5437
    },
    {
      "epoch": 0.8512836568566061,
      "grad_norm": 2.681861162185669,
      "learning_rate": 0.00017959433040078197,
      "loss": 1.1165,
      "step": 5438
    },
    {
      "epoch": 0.8514402003757044,
      "grad_norm": 3.7926366329193115,
      "learning_rate": 0.00017956989247311825,
      "loss": 1.3606,
      "step": 5439
    },
    {
      "epoch": 0.8515967438948028,
      "grad_norm": 2.6977744102478027,
      "learning_rate": 0.00017954545454545453,
      "loss": 1.2335,
      "step": 5440
    },
    {
      "epoch": 0.8517532874139011,
      "grad_norm": 1.4805278778076172,
      "learning_rate": 0.00017952101661779078,
      "loss": 0.9316,
      "step": 5441
    },
    {
      "epoch": 0.8519098309329993,
      "grad_norm": 2.5783205032348633,
      "learning_rate": 0.00017949657869012706,
      "loss": 1.1716,
      "step": 5442
    },
    {
      "epoch": 0.8520663744520977,
      "grad_norm": 2.976900100708008,
      "learning_rate": 0.00017947214076246334,
      "loss": 1.5495,
      "step": 5443
    },
    {
      "epoch": 0.852222917971196,
      "grad_norm": 3.093869924545288,
      "learning_rate": 0.0001794477028347996,
      "loss": 1.321,
      "step": 5444
    },
    {
      "epoch": 0.8523794614902943,
      "grad_norm": 1.297680139541626,
      "learning_rate": 0.00017942326490713587,
      "loss": 0.7588,
      "step": 5445
    },
    {
      "epoch": 0.8525360050093926,
      "grad_norm": 3.9038898944854736,
      "learning_rate": 0.00017939882697947214,
      "loss": 1.2392,
      "step": 5446
    },
    {
      "epoch": 0.852692548528491,
      "grad_norm": 1.2751731872558594,
      "learning_rate": 0.00017937438905180837,
      "loss": 0.3811,
      "step": 5447
    },
    {
      "epoch": 0.8528490920475892,
      "grad_norm": 1.0307965278625488,
      "learning_rate": 0.00017934995112414465,
      "loss": 0.382,
      "step": 5448
    },
    {
      "epoch": 0.8530056355666875,
      "grad_norm": 2.1279537677764893,
      "learning_rate": 0.00017932551319648093,
      "loss": 0.8235,
      "step": 5449
    },
    {
      "epoch": 0.8531621790857858,
      "grad_norm": 2.3851094245910645,
      "learning_rate": 0.00017930107526881718,
      "loss": 1.0698,
      "step": 5450
    },
    {
      "epoch": 0.8533187226048842,
      "grad_norm": 0.3719303011894226,
      "learning_rate": 0.00017927663734115345,
      "loss": 0.2337,
      "step": 5451
    },
    {
      "epoch": 0.8534752661239825,
      "grad_norm": 0.48134690523147583,
      "learning_rate": 0.00017925219941348973,
      "loss": 0.4041,
      "step": 5452
    },
    {
      "epoch": 0.8536318096430808,
      "grad_norm": 0.40173661708831787,
      "learning_rate": 0.00017922776148582598,
      "loss": 0.2524,
      "step": 5453
    },
    {
      "epoch": 0.853788353162179,
      "grad_norm": 0.5400435328483582,
      "learning_rate": 0.00017920332355816224,
      "loss": 0.2702,
      "step": 5454
    },
    {
      "epoch": 0.8539448966812774,
      "grad_norm": 0.8261874318122864,
      "learning_rate": 0.00017917888563049851,
      "loss": 0.4001,
      "step": 5455
    },
    {
      "epoch": 0.8541014402003757,
      "grad_norm": 0.846582293510437,
      "learning_rate": 0.00017915444770283477,
      "loss": 0.5371,
      "step": 5456
    },
    {
      "epoch": 0.854257983719474,
      "grad_norm": 0.961149275302887,
      "learning_rate": 0.00017913000977517104,
      "loss": 0.5499,
      "step": 5457
    },
    {
      "epoch": 0.8544145272385724,
      "grad_norm": 0.5829175710678101,
      "learning_rate": 0.00017910557184750732,
      "loss": 0.4273,
      "step": 5458
    },
    {
      "epoch": 0.8545710707576706,
      "grad_norm": 0.8650907874107361,
      "learning_rate": 0.00017908113391984357,
      "loss": 0.2947,
      "step": 5459
    },
    {
      "epoch": 0.8547276142767689,
      "grad_norm": 0.7858892679214478,
      "learning_rate": 0.00017905669599217985,
      "loss": 0.4431,
      "step": 5460
    },
    {
      "epoch": 0.8548841577958672,
      "grad_norm": 0.6087132692337036,
      "learning_rate": 0.00017903225806451613,
      "loss": 0.4506,
      "step": 5461
    },
    {
      "epoch": 0.8550407013149656,
      "grad_norm": 0.840796709060669,
      "learning_rate": 0.00017900782013685235,
      "loss": 0.4651,
      "step": 5462
    },
    {
      "epoch": 0.8551972448340639,
      "grad_norm": 2.8522396087646484,
      "learning_rate": 0.00017898338220918863,
      "loss": 0.5114,
      "step": 5463
    },
    {
      "epoch": 0.8553537883531622,
      "grad_norm": 1.8388530015945435,
      "learning_rate": 0.0001789589442815249,
      "loss": 0.7611,
      "step": 5464
    },
    {
      "epoch": 0.8555103318722604,
      "grad_norm": 1.0927659273147583,
      "learning_rate": 0.00017893450635386116,
      "loss": 0.4343,
      "step": 5465
    },
    {
      "epoch": 0.8556668753913588,
      "grad_norm": 1.3488267660140991,
      "learning_rate": 0.00017891006842619744,
      "loss": 0.591,
      "step": 5466
    },
    {
      "epoch": 0.8558234189104571,
      "grad_norm": 2.569788932800293,
      "learning_rate": 0.00017888563049853372,
      "loss": 0.5916,
      "step": 5467
    },
    {
      "epoch": 0.8559799624295554,
      "grad_norm": 1.4221092462539673,
      "learning_rate": 0.00017886119257086997,
      "loss": 0.7243,
      "step": 5468
    },
    {
      "epoch": 0.8561365059486538,
      "grad_norm": 0.8458523750305176,
      "learning_rate": 0.00017883675464320625,
      "loss": 0.3696,
      "step": 5469
    },
    {
      "epoch": 0.8562930494677521,
      "grad_norm": 1.2293727397918701,
      "learning_rate": 0.00017881231671554253,
      "loss": 0.6373,
      "step": 5470
    },
    {
      "epoch": 0.8564495929868503,
      "grad_norm": 1.3807203769683838,
      "learning_rate": 0.00017878787878787875,
      "loss": 0.6213,
      "step": 5471
    },
    {
      "epoch": 0.8566061365059486,
      "grad_norm": 3.1636455059051514,
      "learning_rate": 0.00017876344086021503,
      "loss": 1.1287,
      "step": 5472
    },
    {
      "epoch": 0.856762680025047,
      "grad_norm": 1.5414478778839111,
      "learning_rate": 0.0001787390029325513,
      "loss": 0.6642,
      "step": 5473
    },
    {
      "epoch": 0.8569192235441453,
      "grad_norm": 1.7993489503860474,
      "learning_rate": 0.00017871456500488756,
      "loss": 0.5733,
      "step": 5474
    },
    {
      "epoch": 0.8570757670632436,
      "grad_norm": 2.5660245418548584,
      "learning_rate": 0.00017869012707722384,
      "loss": 0.976,
      "step": 5475
    },
    {
      "epoch": 0.8572323105823418,
      "grad_norm": 1.549904465675354,
      "learning_rate": 0.00017866568914956012,
      "loss": 1.0276,
      "step": 5476
    },
    {
      "epoch": 0.8573888541014402,
      "grad_norm": 1.6813697814941406,
      "learning_rate": 0.00017864125122189634,
      "loss": 0.6649,
      "step": 5477
    },
    {
      "epoch": 0.8575453976205385,
      "grad_norm": 1.3475943803787231,
      "learning_rate": 0.00017861681329423262,
      "loss": 0.7519,
      "step": 5478
    },
    {
      "epoch": 0.8577019411396368,
      "grad_norm": 2.2717173099517822,
      "learning_rate": 0.0001785923753665689,
      "loss": 0.4629,
      "step": 5479
    },
    {
      "epoch": 0.8578584846587352,
      "grad_norm": 1.826530933380127,
      "learning_rate": 0.00017856793743890515,
      "loss": 0.8551,
      "step": 5480
    },
    {
      "epoch": 0.8580150281778335,
      "grad_norm": 2.976663112640381,
      "learning_rate": 0.00017854349951124143,
      "loss": 0.8756,
      "step": 5481
    },
    {
      "epoch": 0.8581715716969317,
      "grad_norm": 1.6091177463531494,
      "learning_rate": 0.0001785190615835777,
      "loss": 0.8143,
      "step": 5482
    },
    {
      "epoch": 0.85832811521603,
      "grad_norm": 6.574666500091553,
      "learning_rate": 0.00017849462365591396,
      "loss": 1.8195,
      "step": 5483
    },
    {
      "epoch": 0.8584846587351284,
      "grad_norm": 4.467514514923096,
      "learning_rate": 0.00017847018572825023,
      "loss": 1.5207,
      "step": 5484
    },
    {
      "epoch": 0.8586412022542267,
      "grad_norm": 2.5957164764404297,
      "learning_rate": 0.0001784457478005865,
      "loss": 0.6295,
      "step": 5485
    },
    {
      "epoch": 0.858797745773325,
      "grad_norm": 1.4115712642669678,
      "learning_rate": 0.00017842130987292274,
      "loss": 1.1869,
      "step": 5486
    },
    {
      "epoch": 0.8589542892924233,
      "grad_norm": 3.7182257175445557,
      "learning_rate": 0.00017839687194525901,
      "loss": 1.3938,
      "step": 5487
    },
    {
      "epoch": 0.8591108328115216,
      "grad_norm": 2.587322235107422,
      "learning_rate": 0.0001783724340175953,
      "loss": 1.2754,
      "step": 5488
    },
    {
      "epoch": 0.8592673763306199,
      "grad_norm": 2.736600160598755,
      "learning_rate": 0.00017834799608993154,
      "loss": 1.1982,
      "step": 5489
    },
    {
      "epoch": 0.8594239198497182,
      "grad_norm": 5.915489673614502,
      "learning_rate": 0.00017832355816226782,
      "loss": 1.8069,
      "step": 5490
    },
    {
      "epoch": 0.8595804633688165,
      "grad_norm": 4.513766765594482,
      "learning_rate": 0.0001782991202346041,
      "loss": 1.2607,
      "step": 5491
    },
    {
      "epoch": 0.8597370068879149,
      "grad_norm": 2.3064467906951904,
      "learning_rate": 0.00017827468230694035,
      "loss": 0.6809,
      "step": 5492
    },
    {
      "epoch": 0.8598935504070131,
      "grad_norm": 1.9814224243164062,
      "learning_rate": 0.00017825024437927663,
      "loss": 0.4819,
      "step": 5493
    },
    {
      "epoch": 0.8600500939261114,
      "grad_norm": 2.532914400100708,
      "learning_rate": 0.0001782258064516129,
      "loss": 1.5882,
      "step": 5494
    },
    {
      "epoch": 0.8602066374452098,
      "grad_norm": 1.8483891487121582,
      "learning_rate": 0.00017820136852394913,
      "loss": 0.8876,
      "step": 5495
    },
    {
      "epoch": 0.8603631809643081,
      "grad_norm": 3.369330644607544,
      "learning_rate": 0.0001781769305962854,
      "loss": 1.2741,
      "step": 5496
    },
    {
      "epoch": 0.8605197244834064,
      "grad_norm": 2.431166648864746,
      "learning_rate": 0.0001781524926686217,
      "loss": 0.5309,
      "step": 5497
    },
    {
      "epoch": 0.8606762680025047,
      "grad_norm": 2.442338466644287,
      "learning_rate": 0.00017812805474095794,
      "loss": 0.8135,
      "step": 5498
    },
    {
      "epoch": 0.860832811521603,
      "grad_norm": 2.9823551177978516,
      "learning_rate": 0.00017810361681329422,
      "loss": 0.6185,
      "step": 5499
    },
    {
      "epoch": 0.8609893550407013,
      "grad_norm": 3.414311170578003,
      "learning_rate": 0.0001780791788856305,
      "loss": 1.6136,
      "step": 5500
    },
    {
      "epoch": 0.8611458985597996,
      "grad_norm": 0.6202065348625183,
      "learning_rate": 0.00017805474095796672,
      "loss": 0.3127,
      "step": 5501
    },
    {
      "epoch": 0.861302442078898,
      "grad_norm": 0.6588072776794434,
      "learning_rate": 0.000178030303030303,
      "loss": 0.2606,
      "step": 5502
    },
    {
      "epoch": 0.8614589855979963,
      "grad_norm": 0.9525992274284363,
      "learning_rate": 0.00017800586510263928,
      "loss": 0.267,
      "step": 5503
    },
    {
      "epoch": 0.8616155291170946,
      "grad_norm": 0.8494582176208496,
      "learning_rate": 0.00017798142717497553,
      "loss": 0.501,
      "step": 5504
    },
    {
      "epoch": 0.8617720726361928,
      "grad_norm": 0.632400393486023,
      "learning_rate": 0.0001779569892473118,
      "loss": 0.4369,
      "step": 5505
    },
    {
      "epoch": 0.8619286161552911,
      "grad_norm": 0.5401912331581116,
      "learning_rate": 0.0001779325513196481,
      "loss": 0.3133,
      "step": 5506
    },
    {
      "epoch": 0.8620851596743895,
      "grad_norm": 0.6302811503410339,
      "learning_rate": 0.00017790811339198434,
      "loss": 0.3823,
      "step": 5507
    },
    {
      "epoch": 0.8622417031934878,
      "grad_norm": 0.8073775768280029,
      "learning_rate": 0.00017788367546432062,
      "loss": 0.2837,
      "step": 5508
    },
    {
      "epoch": 0.8623982467125861,
      "grad_norm": 0.7132818698883057,
      "learning_rate": 0.0001778592375366569,
      "loss": 0.2525,
      "step": 5509
    },
    {
      "epoch": 0.8625547902316844,
      "grad_norm": 1.0062077045440674,
      "learning_rate": 0.00017783479960899312,
      "loss": 0.3663,
      "step": 5510
    },
    {
      "epoch": 0.8627113337507827,
      "grad_norm": 0.8359341621398926,
      "learning_rate": 0.0001778103616813294,
      "loss": 0.351,
      "step": 5511
    },
    {
      "epoch": 0.862867877269881,
      "grad_norm": 0.8639108538627625,
      "learning_rate": 0.00017778592375366568,
      "loss": 0.3803,
      "step": 5512
    },
    {
      "epoch": 0.8630244207889793,
      "grad_norm": 0.7320135831832886,
      "learning_rate": 0.00017776148582600193,
      "loss": 0.3996,
      "step": 5513
    },
    {
      "epoch": 0.8631809643080777,
      "grad_norm": 0.9680222868919373,
      "learning_rate": 0.0001777370478983382,
      "loss": 0.3478,
      "step": 5514
    },
    {
      "epoch": 0.863337507827176,
      "grad_norm": 0.8415281176567078,
      "learning_rate": 0.00017771260997067448,
      "loss": 0.4855,
      "step": 5515
    },
    {
      "epoch": 0.8634940513462742,
      "grad_norm": 1.6651570796966553,
      "learning_rate": 0.00017768817204301073,
      "loss": 0.6531,
      "step": 5516
    },
    {
      "epoch": 0.8636505948653725,
      "grad_norm": 1.7974225282669067,
      "learning_rate": 0.000177663734115347,
      "loss": 0.7548,
      "step": 5517
    },
    {
      "epoch": 0.8638071383844709,
      "grad_norm": 1.254573106765747,
      "learning_rate": 0.00017763929618768326,
      "loss": 0.6231,
      "step": 5518
    },
    {
      "epoch": 0.8639636819035692,
      "grad_norm": 0.7575573921203613,
      "learning_rate": 0.00017761485826001952,
      "loss": 0.4347,
      "step": 5519
    },
    {
      "epoch": 0.8641202254226675,
      "grad_norm": 1.291746973991394,
      "learning_rate": 0.0001775904203323558,
      "loss": 0.6743,
      "step": 5520
    },
    {
      "epoch": 0.8642767689417659,
      "grad_norm": 1.3523181676864624,
      "learning_rate": 0.00017756598240469207,
      "loss": 0.5223,
      "step": 5521
    },
    {
      "epoch": 0.8644333124608641,
      "grad_norm": 1.1703637838363647,
      "learning_rate": 0.00017754154447702832,
      "loss": 0.6335,
      "step": 5522
    },
    {
      "epoch": 0.8645898559799624,
      "grad_norm": 2.5771422386169434,
      "learning_rate": 0.0001775171065493646,
      "loss": 0.5819,
      "step": 5523
    },
    {
      "epoch": 0.8647463994990607,
      "grad_norm": 1.401025414466858,
      "learning_rate": 0.00017749266862170088,
      "loss": 0.7858,
      "step": 5524
    },
    {
      "epoch": 0.8649029430181591,
      "grad_norm": 1.5256321430206299,
      "learning_rate": 0.0001774682306940371,
      "loss": 0.55,
      "step": 5525
    },
    {
      "epoch": 0.8650594865372574,
      "grad_norm": 1.255598545074463,
      "learning_rate": 0.00017744379276637338,
      "loss": 0.4361,
      "step": 5526
    },
    {
      "epoch": 0.8652160300563556,
      "grad_norm": 2.342947483062744,
      "learning_rate": 0.00017741935483870966,
      "loss": 1.0666,
      "step": 5527
    },
    {
      "epoch": 0.8653725735754539,
      "grad_norm": 2.2198994159698486,
      "learning_rate": 0.0001773949169110459,
      "loss": 1.0796,
      "step": 5528
    },
    {
      "epoch": 0.8655291170945523,
      "grad_norm": 2.544025182723999,
      "learning_rate": 0.0001773704789833822,
      "loss": 1.0962,
      "step": 5529
    },
    {
      "epoch": 0.8656856606136506,
      "grad_norm": 3.0264079570770264,
      "learning_rate": 0.00017734604105571847,
      "loss": 0.7408,
      "step": 5530
    },
    {
      "epoch": 0.8658422041327489,
      "grad_norm": 1.2119104862213135,
      "learning_rate": 0.00017732160312805472,
      "loss": 0.618,
      "step": 5531
    },
    {
      "epoch": 0.8659987476518473,
      "grad_norm": 1.5257172584533691,
      "learning_rate": 0.000177297165200391,
      "loss": 0.631,
      "step": 5532
    },
    {
      "epoch": 0.8661552911709455,
      "grad_norm": 2.0830912590026855,
      "learning_rate": 0.00017727272727272728,
      "loss": 0.7618,
      "step": 5533
    },
    {
      "epoch": 0.8663118346900438,
      "grad_norm": 2.5756418704986572,
      "learning_rate": 0.0001772482893450635,
      "loss": 0.8881,
      "step": 5534
    },
    {
      "epoch": 0.8664683782091421,
      "grad_norm": 2.78877329826355,
      "learning_rate": 0.00017722385141739978,
      "loss": 0.7217,
      "step": 5535
    },
    {
      "epoch": 0.8666249217282405,
      "grad_norm": 2.5432016849517822,
      "learning_rate": 0.00017719941348973606,
      "loss": 0.9808,
      "step": 5536
    },
    {
      "epoch": 0.8667814652473388,
      "grad_norm": 1.7844754457473755,
      "learning_rate": 0.0001771749755620723,
      "loss": 0.8685,
      "step": 5537
    },
    {
      "epoch": 0.8669380087664371,
      "grad_norm": 2.5685477256774902,
      "learning_rate": 0.0001771505376344086,
      "loss": 1.4534,
      "step": 5538
    },
    {
      "epoch": 0.8670945522855353,
      "grad_norm": 2.9484968185424805,
      "learning_rate": 0.00017712609970674487,
      "loss": 1.3919,
      "step": 5539
    },
    {
      "epoch": 0.8672510958046337,
      "grad_norm": 3.510503053665161,
      "learning_rate": 0.00017710166177908112,
      "loss": 0.8989,
      "step": 5540
    },
    {
      "epoch": 0.867407639323732,
      "grad_norm": 3.30690860748291,
      "learning_rate": 0.0001770772238514174,
      "loss": 1.7287,
      "step": 5541
    },
    {
      "epoch": 0.8675641828428303,
      "grad_norm": 4.295806407928467,
      "learning_rate": 0.00017705278592375365,
      "loss": 1.2677,
      "step": 5542
    },
    {
      "epoch": 0.8677207263619287,
      "grad_norm": 1.898603081703186,
      "learning_rate": 0.0001770283479960899,
      "loss": 1.0414,
      "step": 5543
    },
    {
      "epoch": 0.867877269881027,
      "grad_norm": 1.397929310798645,
      "learning_rate": 0.00017700391006842618,
      "loss": 1.0474,
      "step": 5544
    },
    {
      "epoch": 0.8680338134001252,
      "grad_norm": 3.208662986755371,
      "learning_rate": 0.00017697947214076245,
      "loss": 1.4236,
      "step": 5545
    },
    {
      "epoch": 0.8681903569192235,
      "grad_norm": 4.296112537384033,
      "learning_rate": 0.0001769550342130987,
      "loss": 1.4281,
      "step": 5546
    },
    {
      "epoch": 0.8683469004383219,
      "grad_norm": 5.432612895965576,
      "learning_rate": 0.00017693059628543498,
      "loss": 0.9672,
      "step": 5547
    },
    {
      "epoch": 0.8685034439574202,
      "grad_norm": 2.7205517292022705,
      "learning_rate": 0.00017690615835777126,
      "loss": 0.863,
      "step": 5548
    },
    {
      "epoch": 0.8686599874765185,
      "grad_norm": 2.6513023376464844,
      "learning_rate": 0.0001768817204301075,
      "loss": 0.9332,
      "step": 5549
    },
    {
      "epoch": 0.8688165309956167,
      "grad_norm": 1.974501132965088,
      "learning_rate": 0.00017685728250244376,
      "loss": 0.7272,
      "step": 5550
    },
    {
      "epoch": 0.8689730745147151,
      "grad_norm": 0.6594939827919006,
      "learning_rate": 0.00017683284457478004,
      "loss": 0.3761,
      "step": 5551
    },
    {
      "epoch": 0.8691296180338134,
      "grad_norm": 1.849006175994873,
      "learning_rate": 0.0001768084066471163,
      "loss": 0.4447,
      "step": 5552
    },
    {
      "epoch": 0.8692861615529117,
      "grad_norm": 0.555202841758728,
      "learning_rate": 0.00017678396871945257,
      "loss": 0.2471,
      "step": 5553
    },
    {
      "epoch": 0.86944270507201,
      "grad_norm": 0.5816833972930908,
      "learning_rate": 0.00017675953079178885,
      "loss": 0.3585,
      "step": 5554
    },
    {
      "epoch": 0.8695992485911084,
      "grad_norm": 0.6719066500663757,
      "learning_rate": 0.0001767350928641251,
      "loss": 0.363,
      "step": 5555
    },
    {
      "epoch": 0.8697557921102066,
      "grad_norm": 0.6604121327400208,
      "learning_rate": 0.00017671065493646138,
      "loss": 0.3488,
      "step": 5556
    },
    {
      "epoch": 0.8699123356293049,
      "grad_norm": 0.6314492225646973,
      "learning_rate": 0.00017668621700879766,
      "loss": 0.2469,
      "step": 5557
    },
    {
      "epoch": 0.8700688791484033,
      "grad_norm": 0.6827051639556885,
      "learning_rate": 0.00017666177908113388,
      "loss": 0.3692,
      "step": 5558
    },
    {
      "epoch": 0.8702254226675016,
      "grad_norm": 0.6794304251670837,
      "learning_rate": 0.00017663734115347016,
      "loss": 0.2522,
      "step": 5559
    },
    {
      "epoch": 0.8703819661865999,
      "grad_norm": 0.6768459677696228,
      "learning_rate": 0.00017661290322580644,
      "loss": 0.4427,
      "step": 5560
    },
    {
      "epoch": 0.8705385097056982,
      "grad_norm": 0.9314690232276917,
      "learning_rate": 0.0001765884652981427,
      "loss": 0.2897,
      "step": 5561
    },
    {
      "epoch": 0.8706950532247965,
      "grad_norm": 0.586393415927887,
      "learning_rate": 0.00017656402737047897,
      "loss": 0.2953,
      "step": 5562
    },
    {
      "epoch": 0.8708515967438948,
      "grad_norm": 1.349017858505249,
      "learning_rate": 0.00017653958944281525,
      "loss": 0.4265,
      "step": 5563
    },
    {
      "epoch": 0.8710081402629931,
      "grad_norm": 0.8151659965515137,
      "learning_rate": 0.0001765151515151515,
      "loss": 0.4198,
      "step": 5564
    },
    {
      "epoch": 0.8711646837820914,
      "grad_norm": 0.9918164014816284,
      "learning_rate": 0.00017649071358748778,
      "loss": 0.6098,
      "step": 5565
    },
    {
      "epoch": 0.8713212273011898,
      "grad_norm": 0.7317390441894531,
      "learning_rate": 0.00017646627565982403,
      "loss": 0.4211,
      "step": 5566
    },
    {
      "epoch": 0.871477770820288,
      "grad_norm": 1.2176004648208618,
      "learning_rate": 0.00017644183773216028,
      "loss": 0.5503,
      "step": 5567
    },
    {
      "epoch": 0.8716343143393863,
      "grad_norm": 1.647218108177185,
      "learning_rate": 0.00017641739980449656,
      "loss": 0.8489,
      "step": 5568
    },
    {
      "epoch": 0.8717908578584846,
      "grad_norm": 0.9232958555221558,
      "learning_rate": 0.00017639296187683284,
      "loss": 0.4466,
      "step": 5569
    },
    {
      "epoch": 0.871947401377583,
      "grad_norm": 0.800988495349884,
      "learning_rate": 0.0001763685239491691,
      "loss": 0.511,
      "step": 5570
    },
    {
      "epoch": 0.8721039448966813,
      "grad_norm": 1.9028470516204834,
      "learning_rate": 0.00017634408602150537,
      "loss": 0.6131,
      "step": 5571
    },
    {
      "epoch": 0.8722604884157796,
      "grad_norm": 1.8957113027572632,
      "learning_rate": 0.00017631964809384164,
      "loss": 0.5967,
      "step": 5572
    },
    {
      "epoch": 0.8724170319348779,
      "grad_norm": 2.135235071182251,
      "learning_rate": 0.00017629521016617787,
      "loss": 0.7269,
      "step": 5573
    },
    {
      "epoch": 0.8725735754539762,
      "grad_norm": 3.0502326488494873,
      "learning_rate": 0.00017627077223851415,
      "loss": 0.482,
      "step": 5574
    },
    {
      "epoch": 0.8727301189730745,
      "grad_norm": 1.7193087339401245,
      "learning_rate": 0.00017624633431085043,
      "loss": 0.7517,
      "step": 5575
    },
    {
      "epoch": 0.8728866624921728,
      "grad_norm": 2.7776575088500977,
      "learning_rate": 0.00017622189638318668,
      "loss": 0.5461,
      "step": 5576
    },
    {
      "epoch": 0.8730432060112712,
      "grad_norm": 1.1250964403152466,
      "learning_rate": 0.00017619745845552296,
      "loss": 0.4677,
      "step": 5577
    },
    {
      "epoch": 0.8731997495303695,
      "grad_norm": 2.022542953491211,
      "learning_rate": 0.00017617302052785923,
      "loss": 0.7262,
      "step": 5578
    },
    {
      "epoch": 0.8733562930494677,
      "grad_norm": 2.4576961994171143,
      "learning_rate": 0.00017614858260019548,
      "loss": 1.2791,
      "step": 5579
    },
    {
      "epoch": 0.873512836568566,
      "grad_norm": 2.275444984436035,
      "learning_rate": 0.00017612414467253176,
      "loss": 0.6484,
      "step": 5580
    },
    {
      "epoch": 0.8736693800876644,
      "grad_norm": 2.2250964641571045,
      "learning_rate": 0.00017609970674486804,
      "loss": 0.6692,
      "step": 5581
    },
    {
      "epoch": 0.8738259236067627,
      "grad_norm": 3.6571662425994873,
      "learning_rate": 0.00017607526881720427,
      "loss": 1.1964,
      "step": 5582
    },
    {
      "epoch": 0.873982467125861,
      "grad_norm": 2.21502947807312,
      "learning_rate": 0.00017605083088954054,
      "loss": 0.9287,
      "step": 5583
    },
    {
      "epoch": 0.8741390106449592,
      "grad_norm": 1.8925524950027466,
      "learning_rate": 0.00017602639296187682,
      "loss": 1.0618,
      "step": 5584
    },
    {
      "epoch": 0.8742955541640576,
      "grad_norm": 1.7226258516311646,
      "learning_rate": 0.00017600195503421307,
      "loss": 0.9194,
      "step": 5585
    },
    {
      "epoch": 0.8744520976831559,
      "grad_norm": 2.176206111907959,
      "learning_rate": 0.00017597751710654935,
      "loss": 0.6395,
      "step": 5586
    },
    {
      "epoch": 0.8746086412022542,
      "grad_norm": 2.134561538696289,
      "learning_rate": 0.00017595307917888563,
      "loss": 1.4345,
      "step": 5587
    },
    {
      "epoch": 0.8747651847213526,
      "grad_norm": 3.149474859237671,
      "learning_rate": 0.00017592864125122188,
      "loss": 0.8548,
      "step": 5588
    },
    {
      "epoch": 0.8749217282404509,
      "grad_norm": 1.696314811706543,
      "learning_rate": 0.00017590420332355813,
      "loss": 0.9873,
      "step": 5589
    },
    {
      "epoch": 0.8750782717595491,
      "grad_norm": 2.3394696712493896,
      "learning_rate": 0.0001758797653958944,
      "loss": 1.0674,
      "step": 5590
    },
    {
      "epoch": 0.8752348152786474,
      "grad_norm": 1.807666301727295,
      "learning_rate": 0.00017585532746823066,
      "loss": 1.3712,
      "step": 5591
    },
    {
      "epoch": 0.8753913587977458,
      "grad_norm": 4.91036319732666,
      "learning_rate": 0.00017583088954056694,
      "loss": 1.8955,
      "step": 5592
    },
    {
      "epoch": 0.8755479023168441,
      "grad_norm": 3.2543954849243164,
      "learning_rate": 0.00017580645161290322,
      "loss": 1.4988,
      "step": 5593
    },
    {
      "epoch": 0.8757044458359424,
      "grad_norm": 1.3064311742782593,
      "learning_rate": 0.00017578201368523947,
      "loss": 0.6692,
      "step": 5594
    },
    {
      "epoch": 0.8758609893550408,
      "grad_norm": 3.3625075817108154,
      "learning_rate": 0.00017575757575757575,
      "loss": 1.0202,
      "step": 5595
    },
    {
      "epoch": 0.876017532874139,
      "grad_norm": 1.9790335893630981,
      "learning_rate": 0.00017573313782991203,
      "loss": 1.3143,
      "step": 5596
    },
    {
      "epoch": 0.8761740763932373,
      "grad_norm": 3.206354856491089,
      "learning_rate": 0.00017570869990224825,
      "loss": 0.8278,
      "step": 5597
    },
    {
      "epoch": 0.8763306199123356,
      "grad_norm": 3.371476650238037,
      "learning_rate": 0.00017568426197458453,
      "loss": 0.5511,
      "step": 5598
    },
    {
      "epoch": 0.876487163431434,
      "grad_norm": 3.493743419647217,
      "learning_rate": 0.0001756598240469208,
      "loss": 1.5164,
      "step": 5599
    },
    {
      "epoch": 0.8766437069505323,
      "grad_norm": 2.634359359741211,
      "learning_rate": 0.00017563538611925706,
      "loss": 1.186,
      "step": 5600
    },
    {
      "epoch": 0.8768002504696305,
      "grad_norm": 0.7865440845489502,
      "learning_rate": 0.00017561094819159334,
      "loss": 0.3571,
      "step": 5601
    },
    {
      "epoch": 0.8769567939887288,
      "grad_norm": 0.7957772612571716,
      "learning_rate": 0.00017558651026392962,
      "loss": 0.3469,
      "step": 5602
    },
    {
      "epoch": 0.8771133375078272,
      "grad_norm": 0.6495577692985535,
      "learning_rate": 0.00017556207233626587,
      "loss": 0.3274,
      "step": 5603
    },
    {
      "epoch": 0.8772698810269255,
      "grad_norm": 0.5228533744812012,
      "learning_rate": 0.00017553763440860215,
      "loss": 0.2298,
      "step": 5604
    },
    {
      "epoch": 0.8774264245460238,
      "grad_norm": 1.175572156906128,
      "learning_rate": 0.00017551319648093842,
      "loss": 0.5251,
      "step": 5605
    },
    {
      "epoch": 0.8775829680651221,
      "grad_norm": 0.658734142780304,
      "learning_rate": 0.00017548875855327465,
      "loss": 0.3617,
      "step": 5606
    },
    {
      "epoch": 0.8777395115842204,
      "grad_norm": 0.7342488765716553,
      "learning_rate": 0.00017546432062561093,
      "loss": 0.4297,
      "step": 5607
    },
    {
      "epoch": 0.8778960551033187,
      "grad_norm": 1.1647714376449585,
      "learning_rate": 0.0001754398826979472,
      "loss": 0.4839,
      "step": 5608
    },
    {
      "epoch": 0.878052598622417,
      "grad_norm": 1.1591604948043823,
      "learning_rate": 0.00017541544477028346,
      "loss": 0.4779,
      "step": 5609
    },
    {
      "epoch": 0.8782091421415154,
      "grad_norm": 0.5184983015060425,
      "learning_rate": 0.00017539100684261973,
      "loss": 0.2418,
      "step": 5610
    },
    {
      "epoch": 0.8783656856606137,
      "grad_norm": 0.9003580808639526,
      "learning_rate": 0.000175366568914956,
      "loss": 0.4329,
      "step": 5611
    },
    {
      "epoch": 0.878522229179712,
      "grad_norm": 0.6690883040428162,
      "learning_rate": 0.00017534213098729226,
      "loss": 0.337,
      "step": 5612
    },
    {
      "epoch": 0.8786787726988102,
      "grad_norm": 1.1468678712844849,
      "learning_rate": 0.00017531769305962852,
      "loss": 0.3636,
      "step": 5613
    },
    {
      "epoch": 0.8788353162179086,
      "grad_norm": 0.7906423807144165,
      "learning_rate": 0.0001752932551319648,
      "loss": 0.3264,
      "step": 5614
    },
    {
      "epoch": 0.8789918597370069,
      "grad_norm": 0.82657390832901,
      "learning_rate": 0.00017526881720430104,
      "loss": 0.4039,
      "step": 5615
    },
    {
      "epoch": 0.8791484032561052,
      "grad_norm": 3.476700782775879,
      "learning_rate": 0.00017524437927663732,
      "loss": 0.411,
      "step": 5616
    },
    {
      "epoch": 0.8793049467752035,
      "grad_norm": 1.0398757457733154,
      "learning_rate": 0.0001752199413489736,
      "loss": 0.6646,
      "step": 5617
    },
    {
      "epoch": 0.8794614902943018,
      "grad_norm": 1.3956527709960938,
      "learning_rate": 0.00017519550342130985,
      "loss": 0.7714,
      "step": 5618
    },
    {
      "epoch": 0.8796180338134001,
      "grad_norm": 1.2822397947311401,
      "learning_rate": 0.00017517106549364613,
      "loss": 0.5916,
      "step": 5619
    },
    {
      "epoch": 0.8797745773324984,
      "grad_norm": 1.0409700870513916,
      "learning_rate": 0.0001751466275659824,
      "loss": 0.4558,
      "step": 5620
    },
    {
      "epoch": 0.8799311208515967,
      "grad_norm": 1.218854308128357,
      "learning_rate": 0.00017512218963831863,
      "loss": 0.4444,
      "step": 5621
    },
    {
      "epoch": 0.8800876643706951,
      "grad_norm": 2.0905308723449707,
      "learning_rate": 0.0001750977517106549,
      "loss": 0.336,
      "step": 5622
    },
    {
      "epoch": 0.8802442078897934,
      "grad_norm": 1.4038474559783936,
      "learning_rate": 0.0001750733137829912,
      "loss": 0.5637,
      "step": 5623
    },
    {
      "epoch": 0.8804007514088916,
      "grad_norm": 1.5574206113815308,
      "learning_rate": 0.00017504887585532744,
      "loss": 0.6929,
      "step": 5624
    },
    {
      "epoch": 0.88055729492799,
      "grad_norm": 2.9178993701934814,
      "learning_rate": 0.00017502443792766372,
      "loss": 0.572,
      "step": 5625
    },
    {
      "epoch": 0.8807138384470883,
      "grad_norm": 1.9534040689468384,
      "learning_rate": 0.000175,
      "loss": 0.538,
      "step": 5626
    },
    {
      "epoch": 0.8808703819661866,
      "grad_norm": 2.420355796813965,
      "learning_rate": 0.00017497556207233625,
      "loss": 0.7121,
      "step": 5627
    },
    {
      "epoch": 0.8810269254852849,
      "grad_norm": 2.449112892150879,
      "learning_rate": 0.00017495112414467253,
      "loss": 0.8366,
      "step": 5628
    },
    {
      "epoch": 0.8811834690043833,
      "grad_norm": 1.790299415588379,
      "learning_rate": 0.0001749266862170088,
      "loss": 0.3996,
      "step": 5629
    },
    {
      "epoch": 0.8813400125234815,
      "grad_norm": 2.5560543537139893,
      "learning_rate": 0.00017490224828934503,
      "loss": 0.7785,
      "step": 5630
    },
    {
      "epoch": 0.8814965560425798,
      "grad_norm": 1.6959267854690552,
      "learning_rate": 0.0001748778103616813,
      "loss": 0.6992,
      "step": 5631
    },
    {
      "epoch": 0.8816530995616781,
      "grad_norm": 2.7933099269866943,
      "learning_rate": 0.0001748533724340176,
      "loss": 1.0714,
      "step": 5632
    },
    {
      "epoch": 0.8818096430807765,
      "grad_norm": 3.0920727252960205,
      "learning_rate": 0.00017482893450635384,
      "loss": 0.8542,
      "step": 5633
    },
    {
      "epoch": 0.8819661865998748,
      "grad_norm": 1.7544466257095337,
      "learning_rate": 0.00017480449657869012,
      "loss": 0.8824,
      "step": 5634
    },
    {
      "epoch": 0.882122730118973,
      "grad_norm": 3.3472537994384766,
      "learning_rate": 0.0001747800586510264,
      "loss": 0.7097,
      "step": 5635
    },
    {
      "epoch": 0.8822792736380713,
      "grad_norm": 2.1331112384796143,
      "learning_rate": 0.00017475562072336262,
      "loss": 1.0795,
      "step": 5636
    },
    {
      "epoch": 0.8824358171571697,
      "grad_norm": 3.316182851791382,
      "learning_rate": 0.0001747311827956989,
      "loss": 0.8523,
      "step": 5637
    },
    {
      "epoch": 0.882592360676268,
      "grad_norm": 3.732465982437134,
      "learning_rate": 0.00017470674486803518,
      "loss": 0.9487,
      "step": 5638
    },
    {
      "epoch": 0.8827489041953663,
      "grad_norm": 2.216749668121338,
      "learning_rate": 0.00017468230694037143,
      "loss": 0.7443,
      "step": 5639
    },
    {
      "epoch": 0.8829054477144647,
      "grad_norm": 2.096297025680542,
      "learning_rate": 0.0001746578690127077,
      "loss": 1.0188,
      "step": 5640
    },
    {
      "epoch": 0.8830619912335629,
      "grad_norm": 1.9539072513580322,
      "learning_rate": 0.00017463343108504398,
      "loss": 1.7533,
      "step": 5641
    },
    {
      "epoch": 0.8832185347526612,
      "grad_norm": 1.9574966430664062,
      "learning_rate": 0.00017460899315738024,
      "loss": 1.3721,
      "step": 5642
    },
    {
      "epoch": 0.8833750782717595,
      "grad_norm": 2.3340044021606445,
      "learning_rate": 0.0001745845552297165,
      "loss": 1.3333,
      "step": 5643
    },
    {
      "epoch": 0.8835316217908579,
      "grad_norm": 2.5659821033477783,
      "learning_rate": 0.0001745601173020528,
      "loss": 1.2622,
      "step": 5644
    },
    {
      "epoch": 0.8836881653099562,
      "grad_norm": 2.5627105236053467,
      "learning_rate": 0.00017453567937438902,
      "loss": 1.5452,
      "step": 5645
    },
    {
      "epoch": 0.8838447088290545,
      "grad_norm": 2.1331045627593994,
      "learning_rate": 0.0001745112414467253,
      "loss": 0.7612,
      "step": 5646
    },
    {
      "epoch": 0.8840012523481527,
      "grad_norm": 1.764351725578308,
      "learning_rate": 0.00017448680351906157,
      "loss": 0.5008,
      "step": 5647
    },
    {
      "epoch": 0.8841577958672511,
      "grad_norm": 1.369174838066101,
      "learning_rate": 0.00017446236559139782,
      "loss": 0.4137,
      "step": 5648
    },
    {
      "epoch": 0.8843143393863494,
      "grad_norm": 2.9028494358062744,
      "learning_rate": 0.0001744379276637341,
      "loss": 0.7086,
      "step": 5649
    },
    {
      "epoch": 0.8844708829054477,
      "grad_norm": 3.7896909713745117,
      "learning_rate": 0.00017441348973607038,
      "loss": 0.9732,
      "step": 5650
    },
    {
      "epoch": 0.8846274264245461,
      "grad_norm": 1.121634840965271,
      "learning_rate": 0.00017438905180840663,
      "loss": 0.2748,
      "step": 5651
    },
    {
      "epoch": 0.8847839699436444,
      "grad_norm": 0.823630154132843,
      "learning_rate": 0.0001743646138807429,
      "loss": 0.2495,
      "step": 5652
    },
    {
      "epoch": 0.8849405134627426,
      "grad_norm": 0.7519258260726929,
      "learning_rate": 0.0001743401759530792,
      "loss": 0.2669,
      "step": 5653
    },
    {
      "epoch": 0.8850970569818409,
      "grad_norm": 0.6291525363922119,
      "learning_rate": 0.0001743157380254154,
      "loss": 0.3282,
      "step": 5654
    },
    {
      "epoch": 0.8852536005009393,
      "grad_norm": 0.8792591691017151,
      "learning_rate": 0.0001742913000977517,
      "loss": 0.5075,
      "step": 5655
    },
    {
      "epoch": 0.8854101440200376,
      "grad_norm": 0.6888756155967712,
      "learning_rate": 0.00017426686217008797,
      "loss": 0.4798,
      "step": 5656
    },
    {
      "epoch": 0.8855666875391359,
      "grad_norm": 0.6959285140037537,
      "learning_rate": 0.00017424242424242422,
      "loss": 0.2821,
      "step": 5657
    },
    {
      "epoch": 0.8857232310582341,
      "grad_norm": 0.9070468544960022,
      "learning_rate": 0.0001742179863147605,
      "loss": 0.2513,
      "step": 5658
    },
    {
      "epoch": 0.8858797745773325,
      "grad_norm": 1.2020875215530396,
      "learning_rate": 0.00017419354838709678,
      "loss": 0.4034,
      "step": 5659
    },
    {
      "epoch": 0.8860363180964308,
      "grad_norm": 0.8414081931114197,
      "learning_rate": 0.000174169110459433,
      "loss": 0.2479,
      "step": 5660
    },
    {
      "epoch": 0.8861928616155291,
      "grad_norm": 0.5406526327133179,
      "learning_rate": 0.00017414467253176928,
      "loss": 0.3122,
      "step": 5661
    },
    {
      "epoch": 0.8863494051346275,
      "grad_norm": 0.6893888711929321,
      "learning_rate": 0.00017412023460410556,
      "loss": 0.3185,
      "step": 5662
    },
    {
      "epoch": 0.8865059486537258,
      "grad_norm": 0.9592669606208801,
      "learning_rate": 0.0001740957966764418,
      "loss": 0.4614,
      "step": 5663
    },
    {
      "epoch": 0.886662492172824,
      "grad_norm": 1.71309232711792,
      "learning_rate": 0.0001740713587487781,
      "loss": 0.5796,
      "step": 5664
    },
    {
      "epoch": 0.8868190356919223,
      "grad_norm": 1.557362675666809,
      "learning_rate": 0.00017404692082111437,
      "loss": 0.5306,
      "step": 5665
    },
    {
      "epoch": 0.8869755792110207,
      "grad_norm": 1.2218095064163208,
      "learning_rate": 0.00017402248289345062,
      "loss": 0.532,
      "step": 5666
    },
    {
      "epoch": 0.887132122730119,
      "grad_norm": 1.2000401020050049,
      "learning_rate": 0.0001739980449657869,
      "loss": 0.5144,
      "step": 5667
    },
    {
      "epoch": 0.8872886662492173,
      "grad_norm": 1.0709697008132935,
      "learning_rate": 0.00017397360703812317,
      "loss": 0.5698,
      "step": 5668
    },
    {
      "epoch": 0.8874452097683156,
      "grad_norm": 0.7280417680740356,
      "learning_rate": 0.0001739491691104594,
      "loss": 0.4509,
      "step": 5669
    },
    {
      "epoch": 0.8876017532874139,
      "grad_norm": 1.5254422426223755,
      "learning_rate": 0.00017392473118279568,
      "loss": 0.6166,
      "step": 5670
    },
    {
      "epoch": 0.8877582968065122,
      "grad_norm": 1.4855180978775024,
      "learning_rate": 0.00017390029325513195,
      "loss": 0.4885,
      "step": 5671
    },
    {
      "epoch": 0.8879148403256105,
      "grad_norm": 1.855657935142517,
      "learning_rate": 0.0001738758553274682,
      "loss": 0.7829,
      "step": 5672
    },
    {
      "epoch": 0.8880713838447089,
      "grad_norm": 1.7714046239852905,
      "learning_rate": 0.00017385141739980448,
      "loss": 1.2026,
      "step": 5673
    },
    {
      "epoch": 0.8882279273638072,
      "grad_norm": 2.0689687728881836,
      "learning_rate": 0.00017382697947214076,
      "loss": 0.8146,
      "step": 5674
    },
    {
      "epoch": 0.8883844708829054,
      "grad_norm": 2.669292688369751,
      "learning_rate": 0.00017380254154447701,
      "loss": 0.8094,
      "step": 5675
    },
    {
      "epoch": 0.8885410144020037,
      "grad_norm": 1.2255752086639404,
      "learning_rate": 0.0001737781036168133,
      "loss": 0.486,
      "step": 5676
    },
    {
      "epoch": 0.888697557921102,
      "grad_norm": 0.9265427589416504,
      "learning_rate": 0.00017375366568914954,
      "loss": 0.5691,
      "step": 5677
    },
    {
      "epoch": 0.8888541014402004,
      "grad_norm": 1.7718168497085571,
      "learning_rate": 0.0001737292277614858,
      "loss": 0.6208,
      "step": 5678
    },
    {
      "epoch": 0.8890106449592987,
      "grad_norm": 3.1587021350860596,
      "learning_rate": 0.00017370478983382207,
      "loss": 1.0041,
      "step": 5679
    },
    {
      "epoch": 0.889167188478397,
      "grad_norm": 1.6209098100662231,
      "learning_rate": 0.00017368035190615835,
      "loss": 0.4903,
      "step": 5680
    },
    {
      "epoch": 0.8893237319974953,
      "grad_norm": 1.9299261569976807,
      "learning_rate": 0.0001736559139784946,
      "loss": 0.6506,
      "step": 5681
    },
    {
      "epoch": 0.8894802755165936,
      "grad_norm": 1.872786521911621,
      "learning_rate": 0.00017363147605083088,
      "loss": 1.0693,
      "step": 5682
    },
    {
      "epoch": 0.8896368190356919,
      "grad_norm": 2.118117094039917,
      "learning_rate": 0.00017360703812316716,
      "loss": 0.9645,
      "step": 5683
    },
    {
      "epoch": 0.8897933625547902,
      "grad_norm": 2.271125078201294,
      "learning_rate": 0.00017358260019550338,
      "loss": 0.895,
      "step": 5684
    },
    {
      "epoch": 0.8899499060738886,
      "grad_norm": 2.594882011413574,
      "learning_rate": 0.00017355816226783966,
      "loss": 1.2066,
      "step": 5685
    },
    {
      "epoch": 0.8901064495929869,
      "grad_norm": 1.8685270547866821,
      "learning_rate": 0.00017353372434017594,
      "loss": 0.6995,
      "step": 5686
    },
    {
      "epoch": 0.8902629931120851,
      "grad_norm": 3.1582865715026855,
      "learning_rate": 0.0001735092864125122,
      "loss": 1.3999,
      "step": 5687
    },
    {
      "epoch": 0.8904195366311835,
      "grad_norm": 8.269997596740723,
      "learning_rate": 0.00017348484848484847,
      "loss": 1.6178,
      "step": 5688
    },
    {
      "epoch": 0.8905760801502818,
      "grad_norm": 4.720138072967529,
      "learning_rate": 0.00017346041055718475,
      "loss": 1.1089,
      "step": 5689
    },
    {
      "epoch": 0.8907326236693801,
      "grad_norm": 1.907230257987976,
      "learning_rate": 0.000173435972629521,
      "loss": 1.2031,
      "step": 5690
    },
    {
      "epoch": 0.8908891671884784,
      "grad_norm": 2.7072951793670654,
      "learning_rate": 0.00017341153470185728,
      "loss": 1.6695,
      "step": 5691
    },
    {
      "epoch": 0.8910457107075767,
      "grad_norm": 4.086122989654541,
      "learning_rate": 0.00017338709677419356,
      "loss": 1.0969,
      "step": 5692
    },
    {
      "epoch": 0.891202254226675,
      "grad_norm": 1.9636390209197998,
      "learning_rate": 0.00017336265884652978,
      "loss": 0.7329,
      "step": 5693
    },
    {
      "epoch": 0.8913587977457733,
      "grad_norm": 2.4966297149658203,
      "learning_rate": 0.00017333822091886606,
      "loss": 0.7581,
      "step": 5694
    },
    {
      "epoch": 0.8915153412648716,
      "grad_norm": 1.2184866666793823,
      "learning_rate": 0.00017331378299120234,
      "loss": 0.4329,
      "step": 5695
    },
    {
      "epoch": 0.89167188478397,
      "grad_norm": 3.0518410205841064,
      "learning_rate": 0.0001732893450635386,
      "loss": 0.9643,
      "step": 5696
    },
    {
      "epoch": 0.8918284283030683,
      "grad_norm": 2.205357074737549,
      "learning_rate": 0.00017326490713587487,
      "loss": 1.076,
      "step": 5697
    },
    {
      "epoch": 0.8919849718221665,
      "grad_norm": 1.6067478656768799,
      "learning_rate": 0.00017324046920821115,
      "loss": 0.2366,
      "step": 5698
    },
    {
      "epoch": 0.8921415153412648,
      "grad_norm": 1.8827284574508667,
      "learning_rate": 0.0001732160312805474,
      "loss": 0.9772,
      "step": 5699
    },
    {
      "epoch": 0.8922980588603632,
      "grad_norm": 3.9518826007843018,
      "learning_rate": 0.00017319159335288367,
      "loss": 1.1912,
      "step": 5700
    },
    {
      "epoch": 0.8924546023794615,
      "grad_norm": 0.5631269812583923,
      "learning_rate": 0.00017316715542521993,
      "loss": 0.2854,
      "step": 5701
    },
    {
      "epoch": 0.8926111458985598,
      "grad_norm": 0.7066730856895447,
      "learning_rate": 0.00017314271749755618,
      "loss": 0.4003,
      "step": 5702
    },
    {
      "epoch": 0.8927676894176582,
      "grad_norm": 0.7470924854278564,
      "learning_rate": 0.00017311827956989246,
      "loss": 0.2347,
      "step": 5703
    },
    {
      "epoch": 0.8929242329367564,
      "grad_norm": 0.9745728969573975,
      "learning_rate": 0.00017309384164222873,
      "loss": 0.4271,
      "step": 5704
    },
    {
      "epoch": 0.8930807764558547,
      "grad_norm": 0.6492823362350464,
      "learning_rate": 0.00017306940371456499,
      "loss": 0.3031,
      "step": 5705
    },
    {
      "epoch": 0.893237319974953,
      "grad_norm": 0.7017116546630859,
      "learning_rate": 0.00017304496578690126,
      "loss": 0.2764,
      "step": 5706
    },
    {
      "epoch": 0.8933938634940514,
      "grad_norm": 0.9077380895614624,
      "learning_rate": 0.00017302052785923754,
      "loss": 0.3132,
      "step": 5707
    },
    {
      "epoch": 0.8935504070131497,
      "grad_norm": 0.9337749481201172,
      "learning_rate": 0.00017299608993157377,
      "loss": 0.4168,
      "step": 5708
    },
    {
      "epoch": 0.8937069505322479,
      "grad_norm": 1.3660385608673096,
      "learning_rate": 0.00017297165200391004,
      "loss": 0.2644,
      "step": 5709
    },
    {
      "epoch": 0.8938634940513462,
      "grad_norm": 2.1170074939727783,
      "learning_rate": 0.00017294721407624632,
      "loss": 0.4336,
      "step": 5710
    },
    {
      "epoch": 0.8940200375704446,
      "grad_norm": 0.9623836278915405,
      "learning_rate": 0.00017292277614858257,
      "loss": 0.5736,
      "step": 5711
    },
    {
      "epoch": 0.8941765810895429,
      "grad_norm": 1.502590298652649,
      "learning_rate": 0.00017289833822091885,
      "loss": 0.4649,
      "step": 5712
    },
    {
      "epoch": 0.8943331246086412,
      "grad_norm": 1.0726498365402222,
      "learning_rate": 0.00017287390029325513,
      "loss": 0.3547,
      "step": 5713
    },
    {
      "epoch": 0.8944896681277396,
      "grad_norm": 1.2182965278625488,
      "learning_rate": 0.00017284946236559138,
      "loss": 0.4567,
      "step": 5714
    },
    {
      "epoch": 0.8946462116468378,
      "grad_norm": 1.0065919160842896,
      "learning_rate": 0.00017282502443792766,
      "loss": 0.5393,
      "step": 5715
    },
    {
      "epoch": 0.8948027551659361,
      "grad_norm": 0.693427562713623,
      "learning_rate": 0.00017280058651026394,
      "loss": 0.3155,
      "step": 5716
    },
    {
      "epoch": 0.8949592986850344,
      "grad_norm": 1.1155869960784912,
      "learning_rate": 0.00017277614858260016,
      "loss": 0.5297,
      "step": 5717
    },
    {
      "epoch": 0.8951158422041328,
      "grad_norm": 4.776678562164307,
      "learning_rate": 0.00017275171065493644,
      "loss": 1.0583,
      "step": 5718
    },
    {
      "epoch": 0.8952723857232311,
      "grad_norm": 3.4916136264801025,
      "learning_rate": 0.00017272727272727272,
      "loss": 0.7256,
      "step": 5719
    },
    {
      "epoch": 0.8954289292423294,
      "grad_norm": 1.9992774724960327,
      "learning_rate": 0.00017270283479960897,
      "loss": 0.5441,
      "step": 5720
    },
    {
      "epoch": 0.8955854727614276,
      "grad_norm": 1.5791343450546265,
      "learning_rate": 0.00017267839687194525,
      "loss": 0.8787,
      "step": 5721
    },
    {
      "epoch": 0.895742016280526,
      "grad_norm": 1.3216490745544434,
      "learning_rate": 0.00017265395894428153,
      "loss": 0.4736,
      "step": 5722
    },
    {
      "epoch": 0.8958985597996243,
      "grad_norm": 1.6248879432678223,
      "learning_rate": 0.00017262952101661778,
      "loss": 0.4765,
      "step": 5723
    },
    {
      "epoch": 0.8960551033187226,
      "grad_norm": 1.6858633756637573,
      "learning_rate": 0.00017260508308895406,
      "loss": 0.8183,
      "step": 5724
    },
    {
      "epoch": 0.896211646837821,
      "grad_norm": 1.9334176778793335,
      "learning_rate": 0.0001725806451612903,
      "loss": 0.622,
      "step": 5725
    },
    {
      "epoch": 0.8963681903569192,
      "grad_norm": 1.8685994148254395,
      "learning_rate": 0.00017255620723362656,
      "loss": 0.5704,
      "step": 5726
    },
    {
      "epoch": 0.8965247338760175,
      "grad_norm": 2.087904453277588,
      "learning_rate": 0.00017253176930596284,
      "loss": 0.8175,
      "step": 5727
    },
    {
      "epoch": 0.8966812773951158,
      "grad_norm": 1.4752204418182373,
      "learning_rate": 0.00017250733137829912,
      "loss": 0.509,
      "step": 5728
    },
    {
      "epoch": 0.8968378209142142,
      "grad_norm": 1.6020748615264893,
      "learning_rate": 0.00017248289345063537,
      "loss": 0.9958,
      "step": 5729
    },
    {
      "epoch": 0.8969943644333125,
      "grad_norm": 2.548856258392334,
      "learning_rate": 0.00017245845552297165,
      "loss": 0.8474,
      "step": 5730
    },
    {
      "epoch": 0.8971509079524108,
      "grad_norm": 1.2991943359375,
      "learning_rate": 0.00017243401759530792,
      "loss": 0.7504,
      "step": 5731
    },
    {
      "epoch": 0.897307451471509,
      "grad_norm": 3.308283805847168,
      "learning_rate": 0.00017240957966764415,
      "loss": 0.9691,
      "step": 5732
    },
    {
      "epoch": 0.8974639949906074,
      "grad_norm": 2.927663803100586,
      "learning_rate": 0.00017238514173998043,
      "loss": 1.2162,
      "step": 5733
    },
    {
      "epoch": 0.8976205385097057,
      "grad_norm": 1.6149358749389648,
      "learning_rate": 0.0001723607038123167,
      "loss": 0.6097,
      "step": 5734
    },
    {
      "epoch": 0.897777082028804,
      "grad_norm": 2.015613555908203,
      "learning_rate": 0.00017233626588465296,
      "loss": 0.9498,
      "step": 5735
    },
    {
      "epoch": 0.8979336255479023,
      "grad_norm": 3.3470332622528076,
      "learning_rate": 0.00017231182795698923,
      "loss": 1.2403,
      "step": 5736
    },
    {
      "epoch": 0.8980901690670007,
      "grad_norm": 3.9462878704071045,
      "learning_rate": 0.0001722873900293255,
      "loss": 0.6295,
      "step": 5737
    },
    {
      "epoch": 0.8982467125860989,
      "grad_norm": 1.829472303390503,
      "learning_rate": 0.00017226295210166176,
      "loss": 0.6039,
      "step": 5738
    },
    {
      "epoch": 0.8984032561051972,
      "grad_norm": 1.9444429874420166,
      "learning_rate": 0.00017223851417399804,
      "loss": 1.0498,
      "step": 5739
    },
    {
      "epoch": 0.8985597996242956,
      "grad_norm": 2.516277551651001,
      "learning_rate": 0.00017221407624633432,
      "loss": 1.4808,
      "step": 5740
    },
    {
      "epoch": 0.8987163431433939,
      "grad_norm": 2.0812556743621826,
      "learning_rate": 0.00017218963831867055,
      "loss": 1.1102,
      "step": 5741
    },
    {
      "epoch": 0.8988728866624922,
      "grad_norm": 3.566479206085205,
      "learning_rate": 0.00017216520039100682,
      "loss": 1.3883,
      "step": 5742
    },
    {
      "epoch": 0.8990294301815904,
      "grad_norm": 4.617666244506836,
      "learning_rate": 0.0001721407624633431,
      "loss": 1.3953,
      "step": 5743
    },
    {
      "epoch": 0.8991859737006888,
      "grad_norm": 2.720006227493286,
      "learning_rate": 0.00017211632453567935,
      "loss": 1.2931,
      "step": 5744
    },
    {
      "epoch": 0.8993425172197871,
      "grad_norm": 3.672116756439209,
      "learning_rate": 0.00017209188660801563,
      "loss": 1.467,
      "step": 5745
    },
    {
      "epoch": 0.8994990607388854,
      "grad_norm": 5.463393211364746,
      "learning_rate": 0.0001720674486803519,
      "loss": 0.8112,
      "step": 5746
    },
    {
      "epoch": 0.8996556042579837,
      "grad_norm": 3.226026773452759,
      "learning_rate": 0.00017204301075268816,
      "loss": 0.3882,
      "step": 5747
    },
    {
      "epoch": 0.8998121477770821,
      "grad_norm": 1.810815453529358,
      "learning_rate": 0.0001720185728250244,
      "loss": 0.3974,
      "step": 5748
    },
    {
      "epoch": 0.8999686912961803,
      "grad_norm": 4.585860729217529,
      "learning_rate": 0.0001719941348973607,
      "loss": 1.265,
      "step": 5749
    },
    {
      "epoch": 0.9001252348152786,
      "grad_norm": 1.762209415435791,
      "learning_rate": 0.00017196969696969694,
      "loss": 0.8456,
      "step": 5750
    },
    {
      "epoch": 0.900281778334377,
      "grad_norm": 0.5263707041740417,
      "learning_rate": 0.00017194525904203322,
      "loss": 0.3196,
      "step": 5751
    },
    {
      "epoch": 0.9004383218534753,
      "grad_norm": 0.41388189792633057,
      "learning_rate": 0.0001719208211143695,
      "loss": 0.1791,
      "step": 5752
    },
    {
      "epoch": 0.9005948653725736,
      "grad_norm": 1.194480061531067,
      "learning_rate": 0.00017189638318670575,
      "loss": 0.2162,
      "step": 5753
    },
    {
      "epoch": 0.9007514088916719,
      "grad_norm": 0.5579673647880554,
      "learning_rate": 0.00017187194525904203,
      "loss": 0.3029,
      "step": 5754
    },
    {
      "epoch": 0.9009079524107702,
      "grad_norm": 0.6641569137573242,
      "learning_rate": 0.0001718475073313783,
      "loss": 0.3157,
      "step": 5755
    },
    {
      "epoch": 0.9010644959298685,
      "grad_norm": 0.6336182355880737,
      "learning_rate": 0.00017182306940371453,
      "loss": 0.3372,
      "step": 5756
    },
    {
      "epoch": 0.9012210394489668,
      "grad_norm": 0.9077296853065491,
      "learning_rate": 0.0001717986314760508,
      "loss": 0.4165,
      "step": 5757
    },
    {
      "epoch": 0.9013775829680651,
      "grad_norm": 1.123456597328186,
      "learning_rate": 0.0001717741935483871,
      "loss": 0.3048,
      "step": 5758
    },
    {
      "epoch": 0.9015341264871635,
      "grad_norm": 0.6466776132583618,
      "learning_rate": 0.00017174975562072334,
      "loss": 0.4485,
      "step": 5759
    },
    {
      "epoch": 0.9016906700062617,
      "grad_norm": 0.5841704607009888,
      "learning_rate": 0.00017172531769305962,
      "loss": 0.3242,
      "step": 5760
    },
    {
      "epoch": 0.90184721352536,
      "grad_norm": 0.8340890407562256,
      "learning_rate": 0.0001717008797653959,
      "loss": 0.4687,
      "step": 5761
    },
    {
      "epoch": 0.9020037570444583,
      "grad_norm": 1.2464579343795776,
      "learning_rate": 0.00017167644183773215,
      "loss": 0.4741,
      "step": 5762
    },
    {
      "epoch": 0.9021603005635567,
      "grad_norm": 0.7806826233863831,
      "learning_rate": 0.00017165200391006842,
      "loss": 0.4531,
      "step": 5763
    },
    {
      "epoch": 0.902316844082655,
      "grad_norm": 1.0296310186386108,
      "learning_rate": 0.0001716275659824047,
      "loss": 0.4653,
      "step": 5764
    },
    {
      "epoch": 0.9024733876017533,
      "grad_norm": 0.7613905072212219,
      "learning_rate": 0.00017160312805474093,
      "loss": 0.5479,
      "step": 5765
    },
    {
      "epoch": 0.9026299311208515,
      "grad_norm": 1.586656093597412,
      "learning_rate": 0.0001715786901270772,
      "loss": 0.7166,
      "step": 5766
    },
    {
      "epoch": 0.9027864746399499,
      "grad_norm": 1.1600213050842285,
      "learning_rate": 0.00017155425219941348,
      "loss": 0.4473,
      "step": 5767
    },
    {
      "epoch": 0.9029430181590482,
      "grad_norm": 1.4801439046859741,
      "learning_rate": 0.00017152981427174974,
      "loss": 0.5886,
      "step": 5768
    },
    {
      "epoch": 0.9030995616781465,
      "grad_norm": 1.5481951236724854,
      "learning_rate": 0.00017150537634408601,
      "loss": 0.5857,
      "step": 5769
    },
    {
      "epoch": 0.9032561051972449,
      "grad_norm": 1.2954734563827515,
      "learning_rate": 0.0001714809384164223,
      "loss": 0.8743,
      "step": 5770
    },
    {
      "epoch": 0.9034126487163432,
      "grad_norm": 2.4921019077301025,
      "learning_rate": 0.00017145650048875854,
      "loss": 0.6394,
      "step": 5771
    },
    {
      "epoch": 0.9035691922354414,
      "grad_norm": 1.4190541505813599,
      "learning_rate": 0.0001714320625610948,
      "loss": 0.6264,
      "step": 5772
    },
    {
      "epoch": 0.9037257357545397,
      "grad_norm": 0.9357963800430298,
      "learning_rate": 0.00017140762463343107,
      "loss": 0.4692,
      "step": 5773
    },
    {
      "epoch": 0.9038822792736381,
      "grad_norm": 3.0093657970428467,
      "learning_rate": 0.00017138318670576732,
      "loss": 0.6029,
      "step": 5774
    },
    {
      "epoch": 0.9040388227927364,
      "grad_norm": 2.1367123126983643,
      "learning_rate": 0.0001713587487781036,
      "loss": 0.5638,
      "step": 5775
    },
    {
      "epoch": 0.9041953663118347,
      "grad_norm": 3.115868091583252,
      "learning_rate": 0.00017133431085043988,
      "loss": 0.9128,
      "step": 5776
    },
    {
      "epoch": 0.904351909830933,
      "grad_norm": 1.7553256750106812,
      "learning_rate": 0.00017130987292277613,
      "loss": 0.893,
      "step": 5777
    },
    {
      "epoch": 0.9045084533500313,
      "grad_norm": 1.6324225664138794,
      "learning_rate": 0.0001712854349951124,
      "loss": 0.8314,
      "step": 5778
    },
    {
      "epoch": 0.9046649968691296,
      "grad_norm": 2.084900379180908,
      "learning_rate": 0.0001712609970674487,
      "loss": 0.4471,
      "step": 5779
    },
    {
      "epoch": 0.9048215403882279,
      "grad_norm": 1.5276941061019897,
      "learning_rate": 0.0001712365591397849,
      "loss": 0.8395,
      "step": 5780
    },
    {
      "epoch": 0.9049780839073263,
      "grad_norm": 1.490938663482666,
      "learning_rate": 0.0001712121212121212,
      "loss": 0.8451,
      "step": 5781
    },
    {
      "epoch": 0.9051346274264246,
      "grad_norm": 2.2541961669921875,
      "learning_rate": 0.00017118768328445747,
      "loss": 1.2903,
      "step": 5782
    },
    {
      "epoch": 0.9052911709455228,
      "grad_norm": 2.1898505687713623,
      "learning_rate": 0.00017116324535679372,
      "loss": 1.0065,
      "step": 5783
    },
    {
      "epoch": 0.9054477144646211,
      "grad_norm": 1.505751609802246,
      "learning_rate": 0.00017113880742913,
      "loss": 0.8832,
      "step": 5784
    },
    {
      "epoch": 0.9056042579837195,
      "grad_norm": 2.1820976734161377,
      "learning_rate": 0.00017111436950146628,
      "loss": 0.6737,
      "step": 5785
    },
    {
      "epoch": 0.9057608015028178,
      "grad_norm": 2.6969125270843506,
      "learning_rate": 0.00017108993157380253,
      "loss": 0.9483,
      "step": 5786
    },
    {
      "epoch": 0.9059173450219161,
      "grad_norm": 4.440044403076172,
      "learning_rate": 0.0001710654936461388,
      "loss": 1.0961,
      "step": 5787
    },
    {
      "epoch": 0.9060738885410144,
      "grad_norm": 2.3744637966156006,
      "learning_rate": 0.00017104105571847509,
      "loss": 1.0894,
      "step": 5788
    },
    {
      "epoch": 0.9062304320601127,
      "grad_norm": 2.1773595809936523,
      "learning_rate": 0.0001710166177908113,
      "loss": 0.8032,
      "step": 5789
    },
    {
      "epoch": 0.906386975579211,
      "grad_norm": 2.6673295497894287,
      "learning_rate": 0.0001709921798631476,
      "loss": 1.3428,
      "step": 5790
    },
    {
      "epoch": 0.9065435190983093,
      "grad_norm": 2.424062967300415,
      "learning_rate": 0.00017096774193548387,
      "loss": 1.0168,
      "step": 5791
    },
    {
      "epoch": 0.9067000626174077,
      "grad_norm": 1.6261483430862427,
      "learning_rate": 0.00017094330400782012,
      "loss": 0.843,
      "step": 5792
    },
    {
      "epoch": 0.906856606136506,
      "grad_norm": 2.254560947418213,
      "learning_rate": 0.0001709188660801564,
      "loss": 1.4431,
      "step": 5793
    },
    {
      "epoch": 0.9070131496556043,
      "grad_norm": 1.4824719429016113,
      "learning_rate": 0.00017089442815249267,
      "loss": 0.788,
      "step": 5794
    },
    {
      "epoch": 0.9071696931747025,
      "grad_norm": 1.547086477279663,
      "learning_rate": 0.0001708699902248289,
      "loss": 0.9419,
      "step": 5795
    },
    {
      "epoch": 0.9073262366938009,
      "grad_norm": 2.4336373805999756,
      "learning_rate": 0.00017084555229716518,
      "loss": 0.899,
      "step": 5796
    },
    {
      "epoch": 0.9074827802128992,
      "grad_norm": 1.5645617246627808,
      "learning_rate": 0.00017082111436950146,
      "loss": 0.3773,
      "step": 5797
    },
    {
      "epoch": 0.9076393237319975,
      "grad_norm": 3.626685380935669,
      "learning_rate": 0.0001707966764418377,
      "loss": 1.37,
      "step": 5798
    },
    {
      "epoch": 0.9077958672510958,
      "grad_norm": 2.090294122695923,
      "learning_rate": 0.00017077223851417398,
      "loss": 0.4822,
      "step": 5799
    },
    {
      "epoch": 0.9079524107701941,
      "grad_norm": 3.3169608116149902,
      "learning_rate": 0.00017074780058651026,
      "loss": 1.7751,
      "step": 5800
    },
    {
      "epoch": 0.9081089542892924,
      "grad_norm": 0.6831493377685547,
      "learning_rate": 0.00017072336265884651,
      "loss": 0.4204,
      "step": 5801
    },
    {
      "epoch": 0.9082654978083907,
      "grad_norm": 0.5136266350746155,
      "learning_rate": 0.0001706989247311828,
      "loss": 0.2246,
      "step": 5802
    },
    {
      "epoch": 0.908422041327489,
      "grad_norm": 0.6040349006652832,
      "learning_rate": 0.00017067448680351907,
      "loss": 0.2548,
      "step": 5803
    },
    {
      "epoch": 0.9085785848465874,
      "grad_norm": 0.5405697226524353,
      "learning_rate": 0.0001706500488758553,
      "loss": 0.3375,
      "step": 5804
    },
    {
      "epoch": 0.9087351283656857,
      "grad_norm": 0.4443010985851288,
      "learning_rate": 0.00017062561094819157,
      "loss": 0.3044,
      "step": 5805
    },
    {
      "epoch": 0.9088916718847839,
      "grad_norm": 0.9515902400016785,
      "learning_rate": 0.00017060117302052785,
      "loss": 0.5221,
      "step": 5806
    },
    {
      "epoch": 0.9090482154038823,
      "grad_norm": 0.5636407732963562,
      "learning_rate": 0.0001705767350928641,
      "loss": 0.3327,
      "step": 5807
    },
    {
      "epoch": 0.9092047589229806,
      "grad_norm": 1.611080288887024,
      "learning_rate": 0.00017055229716520038,
      "loss": 0.3531,
      "step": 5808
    },
    {
      "epoch": 0.9093613024420789,
      "grad_norm": 0.7491126656532288,
      "learning_rate": 0.00017052785923753666,
      "loss": 0.3976,
      "step": 5809
    },
    {
      "epoch": 0.9095178459611772,
      "grad_norm": 0.7487491965293884,
      "learning_rate": 0.0001705034213098729,
      "loss": 0.2023,
      "step": 5810
    },
    {
      "epoch": 0.9096743894802756,
      "grad_norm": 0.7374340295791626,
      "learning_rate": 0.0001704789833822092,
      "loss": 0.4256,
      "step": 5811
    },
    {
      "epoch": 0.9098309329993738,
      "grad_norm": 0.7710371017456055,
      "learning_rate": 0.00017045454545454547,
      "loss": 0.3024,
      "step": 5812
    },
    {
      "epoch": 0.9099874765184721,
      "grad_norm": 0.7972769737243652,
      "learning_rate": 0.0001704301075268817,
      "loss": 0.4539,
      "step": 5813
    },
    {
      "epoch": 0.9101440200375704,
      "grad_norm": 2.1603457927703857,
      "learning_rate": 0.00017040566959921797,
      "loss": 0.4388,
      "step": 5814
    },
    {
      "epoch": 0.9103005635566688,
      "grad_norm": 1.0259324312210083,
      "learning_rate": 0.00017038123167155425,
      "loss": 0.5987,
      "step": 5815
    },
    {
      "epoch": 0.9104571070757671,
      "grad_norm": 1.267693281173706,
      "learning_rate": 0.0001703567937438905,
      "loss": 0.6287,
      "step": 5816
    },
    {
      "epoch": 0.9106136505948653,
      "grad_norm": 0.7310298085212708,
      "learning_rate": 0.00017033235581622678,
      "loss": 0.3516,
      "step": 5817
    },
    {
      "epoch": 0.9107701941139636,
      "grad_norm": 0.9882366061210632,
      "learning_rate": 0.00017030791788856306,
      "loss": 0.6303,
      "step": 5818
    },
    {
      "epoch": 0.910926737633062,
      "grad_norm": 1.0292433500289917,
      "learning_rate": 0.00017028347996089928,
      "loss": 0.508,
      "step": 5819
    },
    {
      "epoch": 0.9110832811521603,
      "grad_norm": 2.08266019821167,
      "learning_rate": 0.00017025904203323556,
      "loss": 0.986,
      "step": 5820
    },
    {
      "epoch": 0.9112398246712586,
      "grad_norm": 1.249509572982788,
      "learning_rate": 0.00017023460410557184,
      "loss": 0.558,
      "step": 5821
    },
    {
      "epoch": 0.911396368190357,
      "grad_norm": 1.4662624597549438,
      "learning_rate": 0.0001702101661779081,
      "loss": 0.5545,
      "step": 5822
    },
    {
      "epoch": 0.9115529117094552,
      "grad_norm": 1.311246395111084,
      "learning_rate": 0.00017018572825024437,
      "loss": 0.5318,
      "step": 5823
    },
    {
      "epoch": 0.9117094552285535,
      "grad_norm": 1.9841166734695435,
      "learning_rate": 0.00017016129032258065,
      "loss": 0.8669,
      "step": 5824
    },
    {
      "epoch": 0.9118659987476518,
      "grad_norm": 1.4273552894592285,
      "learning_rate": 0.0001701368523949169,
      "loss": 0.3847,
      "step": 5825
    },
    {
      "epoch": 0.9120225422667502,
      "grad_norm": 1.7364933490753174,
      "learning_rate": 0.00017011241446725318,
      "loss": 0.5714,
      "step": 5826
    },
    {
      "epoch": 0.9121790857858485,
      "grad_norm": 1.4556832313537598,
      "learning_rate": 0.00017008797653958945,
      "loss": 0.7064,
      "step": 5827
    },
    {
      "epoch": 0.9123356293049468,
      "grad_norm": 1.4327338933944702,
      "learning_rate": 0.00017006353861192568,
      "loss": 0.6752,
      "step": 5828
    },
    {
      "epoch": 0.912492172824045,
      "grad_norm": 1.3399438858032227,
      "learning_rate": 0.00017003910068426196,
      "loss": 0.6105,
      "step": 5829
    },
    {
      "epoch": 0.9126487163431434,
      "grad_norm": 1.6405707597732544,
      "learning_rate": 0.00017001466275659823,
      "loss": 0.954,
      "step": 5830
    },
    {
      "epoch": 0.9128052598622417,
      "grad_norm": 2.615856170654297,
      "learning_rate": 0.00016999022482893449,
      "loss": 0.7187,
      "step": 5831
    },
    {
      "epoch": 0.91296180338134,
      "grad_norm": 2.1250553131103516,
      "learning_rate": 0.00016996578690127076,
      "loss": 1.097,
      "step": 5832
    },
    {
      "epoch": 0.9131183469004384,
      "grad_norm": 1.6402782201766968,
      "learning_rate": 0.00016994134897360704,
      "loss": 0.6268,
      "step": 5833
    },
    {
      "epoch": 0.9132748904195366,
      "grad_norm": 2.8018393516540527,
      "learning_rate": 0.0001699169110459433,
      "loss": 0.8648,
      "step": 5834
    },
    {
      "epoch": 0.9134314339386349,
      "grad_norm": 2.0957236289978027,
      "learning_rate": 0.00016989247311827957,
      "loss": 0.9006,
      "step": 5835
    },
    {
      "epoch": 0.9135879774577332,
      "grad_norm": 4.053585529327393,
      "learning_rate": 0.00016986803519061582,
      "loss": 1.3675,
      "step": 5836
    },
    {
      "epoch": 0.9137445209768316,
      "grad_norm": 1.946283221244812,
      "learning_rate": 0.00016984359726295207,
      "loss": 0.8759,
      "step": 5837
    },
    {
      "epoch": 0.9139010644959299,
      "grad_norm": 1.3934121131896973,
      "learning_rate": 0.00016981915933528835,
      "loss": 1.0044,
      "step": 5838
    },
    {
      "epoch": 0.9140576080150282,
      "grad_norm": 4.167563438415527,
      "learning_rate": 0.00016979472140762463,
      "loss": 0.9486,
      "step": 5839
    },
    {
      "epoch": 0.9142141515341264,
      "grad_norm": 3.6018738746643066,
      "learning_rate": 0.00016977028347996088,
      "loss": 0.9403,
      "step": 5840
    },
    {
      "epoch": 0.9143706950532248,
      "grad_norm": 2.0300357341766357,
      "learning_rate": 0.00016974584555229716,
      "loss": 1.152,
      "step": 5841
    },
    {
      "epoch": 0.9145272385723231,
      "grad_norm": 2.6860861778259277,
      "learning_rate": 0.00016972140762463344,
      "loss": 0.8827,
      "step": 5842
    },
    {
      "epoch": 0.9146837820914214,
      "grad_norm": 1.8871678113937378,
      "learning_rate": 0.00016969696969696966,
      "loss": 1.57,
      "step": 5843
    },
    {
      "epoch": 0.9148403256105198,
      "grad_norm": 2.2716727256774902,
      "learning_rate": 0.00016967253176930594,
      "loss": 1.4632,
      "step": 5844
    },
    {
      "epoch": 0.9149968691296181,
      "grad_norm": 3.0800204277038574,
      "learning_rate": 0.00016964809384164222,
      "loss": 1.4531,
      "step": 5845
    },
    {
      "epoch": 0.9151534126487163,
      "grad_norm": 4.430184841156006,
      "learning_rate": 0.00016962365591397847,
      "loss": 0.8666,
      "step": 5846
    },
    {
      "epoch": 0.9153099561678146,
      "grad_norm": 2.232022285461426,
      "learning_rate": 0.00016959921798631475,
      "loss": 0.6021,
      "step": 5847
    },
    {
      "epoch": 0.915466499686913,
      "grad_norm": 2.0096168518066406,
      "learning_rate": 0.00016957478005865103,
      "loss": 0.9608,
      "step": 5848
    },
    {
      "epoch": 0.9156230432060113,
      "grad_norm": 3.4531753063201904,
      "learning_rate": 0.00016955034213098728,
      "loss": 0.7093,
      "step": 5849
    },
    {
      "epoch": 0.9157795867251096,
      "grad_norm": 5.444746971130371,
      "learning_rate": 0.00016952590420332356,
      "loss": 0.777,
      "step": 5850
    },
    {
      "epoch": 0.9159361302442078,
      "grad_norm": 0.5307621359825134,
      "learning_rate": 0.00016950146627565984,
      "loss": 0.3685,
      "step": 5851
    },
    {
      "epoch": 0.9160926737633062,
      "grad_norm": 0.5665816068649292,
      "learning_rate": 0.00016947702834799606,
      "loss": 0.3451,
      "step": 5852
    },
    {
      "epoch": 0.9162492172824045,
      "grad_norm": 1.0723793506622314,
      "learning_rate": 0.00016945259042033234,
      "loss": 0.5412,
      "step": 5853
    },
    {
      "epoch": 0.9164057608015028,
      "grad_norm": 0.6314266324043274,
      "learning_rate": 0.00016942815249266862,
      "loss": 0.3838,
      "step": 5854
    },
    {
      "epoch": 0.9165623043206012,
      "grad_norm": 0.65800541639328,
      "learning_rate": 0.00016940371456500487,
      "loss": 0.4242,
      "step": 5855
    },
    {
      "epoch": 0.9167188478396995,
      "grad_norm": 0.8351601958274841,
      "learning_rate": 0.00016937927663734115,
      "loss": 0.3196,
      "step": 5856
    },
    {
      "epoch": 0.9168753913587977,
      "grad_norm": 0.6536497473716736,
      "learning_rate": 0.00016935483870967742,
      "loss": 0.3413,
      "step": 5857
    },
    {
      "epoch": 0.917031934877896,
      "grad_norm": 0.8502795696258545,
      "learning_rate": 0.00016933040078201368,
      "loss": 0.4184,
      "step": 5858
    },
    {
      "epoch": 0.9171884783969944,
      "grad_norm": 0.9803659319877625,
      "learning_rate": 0.00016930596285434995,
      "loss": 0.3015,
      "step": 5859
    },
    {
      "epoch": 0.9173450219160927,
      "grad_norm": 1.2949978113174438,
      "learning_rate": 0.0001692815249266862,
      "loss": 0.3153,
      "step": 5860
    },
    {
      "epoch": 0.917501565435191,
      "grad_norm": 0.9877235889434814,
      "learning_rate": 0.00016925708699902246,
      "loss": 0.3463,
      "step": 5861
    },
    {
      "epoch": 0.9176581089542893,
      "grad_norm": 1.161659598350525,
      "learning_rate": 0.00016923264907135874,
      "loss": 0.5856,
      "step": 5862
    },
    {
      "epoch": 0.9178146524733876,
      "grad_norm": 0.7833372950553894,
      "learning_rate": 0.000169208211143695,
      "loss": 0.3127,
      "step": 5863
    },
    {
      "epoch": 0.9179711959924859,
      "grad_norm": 1.466927170753479,
      "learning_rate": 0.00016918377321603126,
      "loss": 0.4847,
      "step": 5864
    },
    {
      "epoch": 0.9181277395115842,
      "grad_norm": 1.4383978843688965,
      "learning_rate": 0.00016915933528836754,
      "loss": 0.5227,
      "step": 5865
    },
    {
      "epoch": 0.9182842830306825,
      "grad_norm": 1.0085314512252808,
      "learning_rate": 0.00016913489736070382,
      "loss": 0.3953,
      "step": 5866
    },
    {
      "epoch": 0.9184408265497809,
      "grad_norm": 1.5321907997131348,
      "learning_rate": 0.00016911045943304005,
      "loss": 0.6083,
      "step": 5867
    },
    {
      "epoch": 0.9185973700688791,
      "grad_norm": 4.02199649810791,
      "learning_rate": 0.00016908602150537632,
      "loss": 0.7558,
      "step": 5868
    },
    {
      "epoch": 0.9187539135879774,
      "grad_norm": 1.296752691268921,
      "learning_rate": 0.0001690615835777126,
      "loss": 0.3596,
      "step": 5869
    },
    {
      "epoch": 0.9189104571070758,
      "grad_norm": 2.0699713230133057,
      "learning_rate": 0.00016903714565004885,
      "loss": 0.6962,
      "step": 5870
    },
    {
      "epoch": 0.9190670006261741,
      "grad_norm": 1.0956624746322632,
      "learning_rate": 0.00016901270772238513,
      "loss": 0.3088,
      "step": 5871
    },
    {
      "epoch": 0.9192235441452724,
      "grad_norm": 1.2984445095062256,
      "learning_rate": 0.0001689882697947214,
      "loss": 0.5232,
      "step": 5872
    },
    {
      "epoch": 0.9193800876643707,
      "grad_norm": 1.7365597486495972,
      "learning_rate": 0.00016896383186705766,
      "loss": 0.9313,
      "step": 5873
    },
    {
      "epoch": 0.919536631183469,
      "grad_norm": 1.6211953163146973,
      "learning_rate": 0.00016893939393939394,
      "loss": 0.5194,
      "step": 5874
    },
    {
      "epoch": 0.9196931747025673,
      "grad_norm": 1.4481335878372192,
      "learning_rate": 0.00016891495601173022,
      "loss": 0.4233,
      "step": 5875
    },
    {
      "epoch": 0.9198497182216656,
      "grad_norm": 2.125072956085205,
      "learning_rate": 0.00016889051808406644,
      "loss": 0.5753,
      "step": 5876
    },
    {
      "epoch": 0.9200062617407639,
      "grad_norm": 4.071976184844971,
      "learning_rate": 0.00016886608015640272,
      "loss": 0.6559,
      "step": 5877
    },
    {
      "epoch": 0.9201628052598623,
      "grad_norm": 2.427960157394409,
      "learning_rate": 0.000168841642228739,
      "loss": 1.0354,
      "step": 5878
    },
    {
      "epoch": 0.9203193487789606,
      "grad_norm": 1.5165985822677612,
      "learning_rate": 0.00016881720430107525,
      "loss": 0.813,
      "step": 5879
    },
    {
      "epoch": 0.9204758922980588,
      "grad_norm": 2.0573408603668213,
      "learning_rate": 0.00016879276637341153,
      "loss": 1.2233,
      "step": 5880
    },
    {
      "epoch": 0.9206324358171571,
      "grad_norm": 2.256502866744995,
      "learning_rate": 0.0001687683284457478,
      "loss": 0.6152,
      "step": 5881
    },
    {
      "epoch": 0.9207889793362555,
      "grad_norm": 2.5055789947509766,
      "learning_rate": 0.00016874389051808406,
      "loss": 0.7555,
      "step": 5882
    },
    {
      "epoch": 0.9209455228553538,
      "grad_norm": 1.9413427114486694,
      "learning_rate": 0.00016871945259042034,
      "loss": 1.0083,
      "step": 5883
    },
    {
      "epoch": 0.9211020663744521,
      "grad_norm": 1.8647050857543945,
      "learning_rate": 0.00016869501466275656,
      "loss": 1.0279,
      "step": 5884
    },
    {
      "epoch": 0.9212586098935505,
      "grad_norm": 2.933109998703003,
      "learning_rate": 0.00016867057673509284,
      "loss": 0.8053,
      "step": 5885
    },
    {
      "epoch": 0.9214151534126487,
      "grad_norm": 2.2728209495544434,
      "learning_rate": 0.00016864613880742912,
      "loss": 0.9038,
      "step": 5886
    },
    {
      "epoch": 0.921571696931747,
      "grad_norm": 2.305570125579834,
      "learning_rate": 0.00016862170087976537,
      "loss": 0.9049,
      "step": 5887
    },
    {
      "epoch": 0.9217282404508453,
      "grad_norm": 3.410182476043701,
      "learning_rate": 0.00016859726295210165,
      "loss": 1.1697,
      "step": 5888
    },
    {
      "epoch": 0.9218847839699437,
      "grad_norm": 3.4939680099487305,
      "learning_rate": 0.00016857282502443793,
      "loss": 1.5584,
      "step": 5889
    },
    {
      "epoch": 0.922041327489042,
      "grad_norm": 1.5268841981887817,
      "learning_rate": 0.00016854838709677415,
      "loss": 1.1331,
      "step": 5890
    },
    {
      "epoch": 0.9221978710081402,
      "grad_norm": 1.2586132287979126,
      "learning_rate": 0.00016852394916911043,
      "loss": 1.0516,
      "step": 5891
    },
    {
      "epoch": 0.9223544145272385,
      "grad_norm": 2.3908731937408447,
      "learning_rate": 0.0001684995112414467,
      "loss": 1.0746,
      "step": 5892
    },
    {
      "epoch": 0.9225109580463369,
      "grad_norm": 2.6460394859313965,
      "learning_rate": 0.00016847507331378296,
      "loss": 1.5005,
      "step": 5893
    },
    {
      "epoch": 0.9226675015654352,
      "grad_norm": 1.6901088953018188,
      "learning_rate": 0.00016845063538611924,
      "loss": 0.7368,
      "step": 5894
    },
    {
      "epoch": 0.9228240450845335,
      "grad_norm": 1.7683452367782593,
      "learning_rate": 0.00016842619745845551,
      "loss": 0.6092,
      "step": 5895
    },
    {
      "epoch": 0.9229805886036319,
      "grad_norm": 3.026430606842041,
      "learning_rate": 0.00016840175953079177,
      "loss": 0.4762,
      "step": 5896
    },
    {
      "epoch": 0.9231371321227301,
      "grad_norm": 2.5244085788726807,
      "learning_rate": 0.00016837732160312804,
      "loss": 1.0575,
      "step": 5897
    },
    {
      "epoch": 0.9232936756418284,
      "grad_norm": 1.5700312852859497,
      "learning_rate": 0.00016835288367546432,
      "loss": 0.9384,
      "step": 5898
    },
    {
      "epoch": 0.9234502191609267,
      "grad_norm": 1.4944273233413696,
      "learning_rate": 0.00016832844574780055,
      "loss": 0.3906,
      "step": 5899
    },
    {
      "epoch": 0.9236067626800251,
      "grad_norm": 4.99465799331665,
      "learning_rate": 0.00016830400782013682,
      "loss": 0.9438,
      "step": 5900
    },
    {
      "epoch": 0.9237633061991234,
      "grad_norm": 0.4812981188297272,
      "learning_rate": 0.0001682795698924731,
      "loss": 0.3414,
      "step": 5901
    },
    {
      "epoch": 0.9239198497182217,
      "grad_norm": 0.5931546092033386,
      "learning_rate": 0.00016825513196480935,
      "loss": 0.3613,
      "step": 5902
    },
    {
      "epoch": 0.9240763932373199,
      "grad_norm": 0.6139108538627625,
      "learning_rate": 0.00016823069403714563,
      "loss": 0.2214,
      "step": 5903
    },
    {
      "epoch": 0.9242329367564183,
      "grad_norm": 0.5874446630477905,
      "learning_rate": 0.0001682062561094819,
      "loss": 0.3274,
      "step": 5904
    },
    {
      "epoch": 0.9243894802755166,
      "grad_norm": 0.4940461218357086,
      "learning_rate": 0.00016818181818181816,
      "loss": 0.2925,
      "step": 5905
    },
    {
      "epoch": 0.9245460237946149,
      "grad_norm": 0.5036855340003967,
      "learning_rate": 0.00016815738025415444,
      "loss": 0.2816,
      "step": 5906
    },
    {
      "epoch": 0.9247025673137133,
      "grad_norm": 0.6490256190299988,
      "learning_rate": 0.0001681329423264907,
      "loss": 0.2061,
      "step": 5907
    },
    {
      "epoch": 0.9248591108328115,
      "grad_norm": 1.324135184288025,
      "learning_rate": 0.00016810850439882694,
      "loss": 0.5117,
      "step": 5908
    },
    {
      "epoch": 0.9250156543519098,
      "grad_norm": 0.9895930290222168,
      "learning_rate": 0.00016808406647116322,
      "loss": 0.3738,
      "step": 5909
    },
    {
      "epoch": 0.9251721978710081,
      "grad_norm": 0.761587917804718,
      "learning_rate": 0.0001680596285434995,
      "loss": 0.3548,
      "step": 5910
    },
    {
      "epoch": 0.9253287413901065,
      "grad_norm": 0.905089259147644,
      "learning_rate": 0.00016803519061583575,
      "loss": 0.3428,
      "step": 5911
    },
    {
      "epoch": 0.9254852849092048,
      "grad_norm": 1.7721465826034546,
      "learning_rate": 0.00016801075268817203,
      "loss": 0.3537,
      "step": 5912
    },
    {
      "epoch": 0.9256418284283031,
      "grad_norm": 0.9065170288085938,
      "learning_rate": 0.0001679863147605083,
      "loss": 0.7074,
      "step": 5913
    },
    {
      "epoch": 0.9257983719474013,
      "grad_norm": 1.3769489526748657,
      "learning_rate": 0.00016796187683284453,
      "loss": 0.4449,
      "step": 5914
    },
    {
      "epoch": 0.9259549154664997,
      "grad_norm": 0.8921677470207214,
      "learning_rate": 0.0001679374389051808,
      "loss": 0.4508,
      "step": 5915
    },
    {
      "epoch": 0.926111458985598,
      "grad_norm": 0.827818751335144,
      "learning_rate": 0.0001679130009775171,
      "loss": 0.6338,
      "step": 5916
    },
    {
      "epoch": 0.9262680025046963,
      "grad_norm": 1.2946507930755615,
      "learning_rate": 0.00016788856304985334,
      "loss": 0.6249,
      "step": 5917
    },
    {
      "epoch": 0.9264245460237946,
      "grad_norm": 0.9976163506507874,
      "learning_rate": 0.00016786412512218962,
      "loss": 0.2552,
      "step": 5918
    },
    {
      "epoch": 0.926581089542893,
      "grad_norm": 0.9220026731491089,
      "learning_rate": 0.0001678396871945259,
      "loss": 0.3495,
      "step": 5919
    },
    {
      "epoch": 0.9267376330619912,
      "grad_norm": 1.1707324981689453,
      "learning_rate": 0.00016781524926686215,
      "loss": 0.5967,
      "step": 5920
    },
    {
      "epoch": 0.9268941765810895,
      "grad_norm": 1.8587265014648438,
      "learning_rate": 0.00016779081133919843,
      "loss": 0.512,
      "step": 5921
    },
    {
      "epoch": 0.9270507201001879,
      "grad_norm": 1.1185016632080078,
      "learning_rate": 0.0001677663734115347,
      "loss": 0.4552,
      "step": 5922
    },
    {
      "epoch": 0.9272072636192862,
      "grad_norm": 7.0849127769470215,
      "learning_rate": 0.00016774193548387093,
      "loss": 0.5696,
      "step": 5923
    },
    {
      "epoch": 0.9273638071383845,
      "grad_norm": 1.0725510120391846,
      "learning_rate": 0.0001677174975562072,
      "loss": 0.5536,
      "step": 5924
    },
    {
      "epoch": 0.9275203506574827,
      "grad_norm": 1.6228716373443604,
      "learning_rate": 0.00016769305962854349,
      "loss": 0.4286,
      "step": 5925
    },
    {
      "epoch": 0.9276768941765811,
      "grad_norm": 1.596377968788147,
      "learning_rate": 0.00016766862170087974,
      "loss": 0.6547,
      "step": 5926
    },
    {
      "epoch": 0.9278334376956794,
      "grad_norm": 1.7780789136886597,
      "learning_rate": 0.00016764418377321601,
      "loss": 0.8426,
      "step": 5927
    },
    {
      "epoch": 0.9279899812147777,
      "grad_norm": 1.8917478322982788,
      "learning_rate": 0.0001676197458455523,
      "loss": 0.6604,
      "step": 5928
    },
    {
      "epoch": 0.928146524733876,
      "grad_norm": 1.4411269426345825,
      "learning_rate": 0.00016759530791788854,
      "loss": 0.5667,
      "step": 5929
    },
    {
      "epoch": 0.9283030682529744,
      "grad_norm": 2.5890650749206543,
      "learning_rate": 0.00016757086999022482,
      "loss": 0.4491,
      "step": 5930
    },
    {
      "epoch": 0.9284596117720726,
      "grad_norm": 1.4896984100341797,
      "learning_rate": 0.00016754643206256107,
      "loss": 0.7193,
      "step": 5931
    },
    {
      "epoch": 0.9286161552911709,
      "grad_norm": 2.9723289012908936,
      "learning_rate": 0.00016752199413489733,
      "loss": 0.7311,
      "step": 5932
    },
    {
      "epoch": 0.9287726988102692,
      "grad_norm": 1.3109147548675537,
      "learning_rate": 0.0001674975562072336,
      "loss": 0.9025,
      "step": 5933
    },
    {
      "epoch": 0.9289292423293676,
      "grad_norm": 2.1195662021636963,
      "learning_rate": 0.00016747311827956988,
      "loss": 0.8071,
      "step": 5934
    },
    {
      "epoch": 0.9290857858484659,
      "grad_norm": 2.0129497051239014,
      "learning_rate": 0.00016744868035190613,
      "loss": 0.6614,
      "step": 5935
    },
    {
      "epoch": 0.9292423293675642,
      "grad_norm": 3.3154754638671875,
      "learning_rate": 0.0001674242424242424,
      "loss": 1.3507,
      "step": 5936
    },
    {
      "epoch": 0.9293988728866625,
      "grad_norm": 2.2003567218780518,
      "learning_rate": 0.0001673998044965787,
      "loss": 1.2579,
      "step": 5937
    },
    {
      "epoch": 0.9295554164057608,
      "grad_norm": 2.3545401096343994,
      "learning_rate": 0.00016737536656891491,
      "loss": 0.9919,
      "step": 5938
    },
    {
      "epoch": 0.9297119599248591,
      "grad_norm": 3.617964744567871,
      "learning_rate": 0.0001673509286412512,
      "loss": 1.6852,
      "step": 5939
    },
    {
      "epoch": 0.9298685034439574,
      "grad_norm": 2.4754271507263184,
      "learning_rate": 0.00016732649071358747,
      "loss": 0.9846,
      "step": 5940
    },
    {
      "epoch": 0.9300250469630558,
      "grad_norm": 2.2454872131347656,
      "learning_rate": 0.00016730205278592372,
      "loss": 0.7251,
      "step": 5941
    },
    {
      "epoch": 0.930181590482154,
      "grad_norm": 2.1622064113616943,
      "learning_rate": 0.00016727761485826,
      "loss": 0.998,
      "step": 5942
    },
    {
      "epoch": 0.9303381340012523,
      "grad_norm": 3.354268789291382,
      "learning_rate": 0.00016725317693059628,
      "loss": 1.5063,
      "step": 5943
    },
    {
      "epoch": 0.9304946775203506,
      "grad_norm": 1.39380943775177,
      "learning_rate": 0.00016722873900293253,
      "loss": 1.1028,
      "step": 5944
    },
    {
      "epoch": 0.930651221039449,
      "grad_norm": 2.941927194595337,
      "learning_rate": 0.0001672043010752688,
      "loss": 1.2247,
      "step": 5945
    },
    {
      "epoch": 0.9308077645585473,
      "grad_norm": 6.12858772277832,
      "learning_rate": 0.0001671798631476051,
      "loss": 1.3599,
      "step": 5946
    },
    {
      "epoch": 0.9309643080776456,
      "grad_norm": 1.9864230155944824,
      "learning_rate": 0.0001671554252199413,
      "loss": 1.2749,
      "step": 5947
    },
    {
      "epoch": 0.9311208515967438,
      "grad_norm": 1.012682318687439,
      "learning_rate": 0.0001671309872922776,
      "loss": 0.5183,
      "step": 5948
    },
    {
      "epoch": 0.9312773951158422,
      "grad_norm": 3.2158727645874023,
      "learning_rate": 0.00016710654936461387,
      "loss": 0.7957,
      "step": 5949
    },
    {
      "epoch": 0.9314339386349405,
      "grad_norm": 2.2417471408843994,
      "learning_rate": 0.00016708211143695012,
      "loss": 0.8408,
      "step": 5950
    },
    {
      "epoch": 0.9315904821540388,
      "grad_norm": 0.5746961236000061,
      "learning_rate": 0.0001670576735092864,
      "loss": 0.4067,
      "step": 5951
    },
    {
      "epoch": 0.9317470256731372,
      "grad_norm": 0.5741134881973267,
      "learning_rate": 0.00016703323558162268,
      "loss": 0.2959,
      "step": 5952
    },
    {
      "epoch": 0.9319035691922355,
      "grad_norm": 0.4729560911655426,
      "learning_rate": 0.00016700879765395893,
      "loss": 0.232,
      "step": 5953
    },
    {
      "epoch": 0.9320601127113337,
      "grad_norm": 0.6859723329544067,
      "learning_rate": 0.00016698435972629518,
      "loss": 0.3221,
      "step": 5954
    },
    {
      "epoch": 0.932216656230432,
      "grad_norm": 0.8597524166107178,
      "learning_rate": 0.00016695992179863146,
      "loss": 0.4589,
      "step": 5955
    },
    {
      "epoch": 0.9323731997495304,
      "grad_norm": 0.4495220184326172,
      "learning_rate": 0.0001669354838709677,
      "loss": 0.3401,
      "step": 5956
    },
    {
      "epoch": 0.9325297432686287,
      "grad_norm": 3.076122999191284,
      "learning_rate": 0.00016691104594330399,
      "loss": 1.0121,
      "step": 5957
    },
    {
      "epoch": 0.932686286787727,
      "grad_norm": 0.5101608633995056,
      "learning_rate": 0.00016688660801564026,
      "loss": 0.2799,
      "step": 5958
    },
    {
      "epoch": 0.9328428303068252,
      "grad_norm": 0.4766397476196289,
      "learning_rate": 0.00016686217008797652,
      "loss": 0.2636,
      "step": 5959
    },
    {
      "epoch": 0.9329993738259236,
      "grad_norm": 1.034203052520752,
      "learning_rate": 0.0001668377321603128,
      "loss": 0.4569,
      "step": 5960
    },
    {
      "epoch": 0.9331559173450219,
      "grad_norm": 0.8217408061027527,
      "learning_rate": 0.00016681329423264907,
      "loss": 0.3696,
      "step": 5961
    },
    {
      "epoch": 0.9333124608641202,
      "grad_norm": 1.199652075767517,
      "learning_rate": 0.0001667888563049853,
      "loss": 0.5895,
      "step": 5962
    },
    {
      "epoch": 0.9334690043832186,
      "grad_norm": 0.9428247809410095,
      "learning_rate": 0.00016676441837732157,
      "loss": 0.3166,
      "step": 5963
    },
    {
      "epoch": 0.9336255479023169,
      "grad_norm": 1.7076395750045776,
      "learning_rate": 0.00016673998044965785,
      "loss": 0.663,
      "step": 5964
    },
    {
      "epoch": 0.9337820914214151,
      "grad_norm": 3.6879310607910156,
      "learning_rate": 0.0001667155425219941,
      "loss": 1.2153,
      "step": 5965
    },
    {
      "epoch": 0.9339386349405134,
      "grad_norm": 0.7371928691864014,
      "learning_rate": 0.00016669110459433038,
      "loss": 0.2916,
      "step": 5966
    },
    {
      "epoch": 0.9340951784596118,
      "grad_norm": 0.8231880068778992,
      "learning_rate": 0.00016666666666666666,
      "loss": 0.5749,
      "step": 5967
    },
    {
      "epoch": 0.9342517219787101,
      "grad_norm": 0.9370611310005188,
      "learning_rate": 0.0001666422287390029,
      "loss": 0.3539,
      "step": 5968
    },
    {
      "epoch": 0.9344082654978084,
      "grad_norm": 1.1584241390228271,
      "learning_rate": 0.0001666177908113392,
      "loss": 0.5032,
      "step": 5969
    },
    {
      "epoch": 0.9345648090169068,
      "grad_norm": 1.1886309385299683,
      "learning_rate": 0.00016659335288367547,
      "loss": 0.4978,
      "step": 5970
    },
    {
      "epoch": 0.934721352536005,
      "grad_norm": 1.277079463005066,
      "learning_rate": 0.0001665689149560117,
      "loss": 0.5764,
      "step": 5971
    },
    {
      "epoch": 0.9348778960551033,
      "grad_norm": 1.9561655521392822,
      "learning_rate": 0.00016654447702834797,
      "loss": 0.478,
      "step": 5972
    },
    {
      "epoch": 0.9350344395742016,
      "grad_norm": 2.3838415145874023,
      "learning_rate": 0.00016652003910068425,
      "loss": 0.6815,
      "step": 5973
    },
    {
      "epoch": 0.9351909830933,
      "grad_norm": 1.1179418563842773,
      "learning_rate": 0.0001664956011730205,
      "loss": 0.6301,
      "step": 5974
    },
    {
      "epoch": 0.9353475266123983,
      "grad_norm": 1.8033316135406494,
      "learning_rate": 0.00016647116324535678,
      "loss": 0.6859,
      "step": 5975
    },
    {
      "epoch": 0.9355040701314965,
      "grad_norm": 1.8072056770324707,
      "learning_rate": 0.00016644672531769306,
      "loss": 0.9525,
      "step": 5976
    },
    {
      "epoch": 0.9356606136505948,
      "grad_norm": 1.693436861038208,
      "learning_rate": 0.0001664222873900293,
      "loss": 0.645,
      "step": 5977
    },
    {
      "epoch": 0.9358171571696932,
      "grad_norm": 2.3530519008636475,
      "learning_rate": 0.00016639784946236556,
      "loss": 0.8476,
      "step": 5978
    },
    {
      "epoch": 0.9359737006887915,
      "grad_norm": 1.2091078758239746,
      "learning_rate": 0.00016637341153470184,
      "loss": 0.6442,
      "step": 5979
    },
    {
      "epoch": 0.9361302442078898,
      "grad_norm": 2.406554937362671,
      "learning_rate": 0.0001663489736070381,
      "loss": 0.937,
      "step": 5980
    },
    {
      "epoch": 0.9362867877269881,
      "grad_norm": 1.2704511880874634,
      "learning_rate": 0.00016632453567937437,
      "loss": 0.8456,
      "step": 5981
    },
    {
      "epoch": 0.9364433312460864,
      "grad_norm": 1.3322428464889526,
      "learning_rate": 0.00016630009775171065,
      "loss": 0.4806,
      "step": 5982
    },
    {
      "epoch": 0.9365998747651847,
      "grad_norm": 1.9257659912109375,
      "learning_rate": 0.0001662756598240469,
      "loss": 1.2242,
      "step": 5983
    },
    {
      "epoch": 0.936756418284283,
      "grad_norm": 5.305508136749268,
      "learning_rate": 0.00016625122189638318,
      "loss": 0.8894,
      "step": 5984
    },
    {
      "epoch": 0.9369129618033814,
      "grad_norm": 2.0064687728881836,
      "learning_rate": 0.00016622678396871945,
      "loss": 1.1084,
      "step": 5985
    },
    {
      "epoch": 0.9370695053224797,
      "grad_norm": 0.8927270174026489,
      "learning_rate": 0.00016620234604105568,
      "loss": 0.3418,
      "step": 5986
    },
    {
      "epoch": 0.937226048841578,
      "grad_norm": 2.652418851852417,
      "learning_rate": 0.00016617790811339196,
      "loss": 1.2848,
      "step": 5987
    },
    {
      "epoch": 0.9373825923606762,
      "grad_norm": 1.884332537651062,
      "learning_rate": 0.00016615347018572824,
      "loss": 0.8169,
      "step": 5988
    },
    {
      "epoch": 0.9375391358797746,
      "grad_norm": 2.335824728012085,
      "learning_rate": 0.0001661290322580645,
      "loss": 1.0861,
      "step": 5989
    },
    {
      "epoch": 0.9376956793988729,
      "grad_norm": 2.0490598678588867,
      "learning_rate": 0.00016610459433040077,
      "loss": 0.7576,
      "step": 5990
    },
    {
      "epoch": 0.9378522229179712,
      "grad_norm": 1.8792589902877808,
      "learning_rate": 0.00016608015640273704,
      "loss": 1.1315,
      "step": 5991
    },
    {
      "epoch": 0.9380087664370695,
      "grad_norm": 2.0781798362731934,
      "learning_rate": 0.0001660557184750733,
      "loss": 0.4345,
      "step": 5992
    },
    {
      "epoch": 0.9381653099561679,
      "grad_norm": 1.9264535903930664,
      "learning_rate": 0.00016603128054740957,
      "loss": 0.8727,
      "step": 5993
    },
    {
      "epoch": 0.9383218534752661,
      "grad_norm": 2.6917431354522705,
      "learning_rate": 0.00016600684261974585,
      "loss": 0.7562,
      "step": 5994
    },
    {
      "epoch": 0.9384783969943644,
      "grad_norm": 2.0042667388916016,
      "learning_rate": 0.00016598240469208208,
      "loss": 0.9416,
      "step": 5995
    },
    {
      "epoch": 0.9386349405134627,
      "grad_norm": 3.084071397781372,
      "learning_rate": 0.00016595796676441835,
      "loss": 0.9873,
      "step": 5996
    },
    {
      "epoch": 0.9387914840325611,
      "grad_norm": 2.0100440979003906,
      "learning_rate": 0.00016593352883675463,
      "loss": 0.7261,
      "step": 5997
    },
    {
      "epoch": 0.9389480275516594,
      "grad_norm": 2.626487970352173,
      "learning_rate": 0.00016590909090909088,
      "loss": 0.9508,
      "step": 5998
    },
    {
      "epoch": 0.9391045710707576,
      "grad_norm": 4.288865089416504,
      "learning_rate": 0.00016588465298142716,
      "loss": 1.564,
      "step": 5999
    },
    {
      "epoch": 0.939261114589856,
      "grad_norm": 6.611445903778076,
      "learning_rate": 0.00016586021505376344,
      "loss": 1.4826,
      "step": 6000
    },
    {
      "epoch": 0.939261114589856,
      "eval_loss": 0.580058753490448,
      "eval_runtime": 206.0189,
      "eval_samples_per_second": 60.106,
      "eval_steps_per_second": 3.757,
      "eval_wer": 0.35448498629321884,
      "step": 6000
    },
    {
      "epoch": 0.9394176581089543,
      "grad_norm": 0.8797839283943176,
      "learning_rate": 0.0001658357771260997,
      "loss": 0.2883,
      "step": 6001
    },
    {
      "epoch": 0.9395742016280526,
      "grad_norm": 0.6887585520744324,
      "learning_rate": 0.00016581133919843594,
      "loss": 0.3189,
      "step": 6002
    },
    {
      "epoch": 0.9397307451471509,
      "grad_norm": 0.4749625325202942,
      "learning_rate": 0.00016578690127077222,
      "loss": 0.3,
      "step": 6003
    },
    {
      "epoch": 0.9398872886662493,
      "grad_norm": 0.7343231439590454,
      "learning_rate": 0.00016576246334310847,
      "loss": 0.2879,
      "step": 6004
    },
    {
      "epoch": 0.9400438321853475,
      "grad_norm": 0.631679892539978,
      "learning_rate": 0.00016573802541544475,
      "loss": 0.4164,
      "step": 6005
    },
    {
      "epoch": 0.9402003757044458,
      "grad_norm": 1.4770556688308716,
      "learning_rate": 0.00016571358748778103,
      "loss": 0.3442,
      "step": 6006
    },
    {
      "epoch": 0.9403569192235441,
      "grad_norm": 0.5342082977294922,
      "learning_rate": 0.00016568914956011728,
      "loss": 0.2752,
      "step": 6007
    },
    {
      "epoch": 0.9405134627426425,
      "grad_norm": 0.7217116355895996,
      "learning_rate": 0.00016566471163245356,
      "loss": 0.3402,
      "step": 6008
    },
    {
      "epoch": 0.9406700062617408,
      "grad_norm": 0.3949846625328064,
      "learning_rate": 0.00016564027370478984,
      "loss": 0.249,
      "step": 6009
    },
    {
      "epoch": 0.9408265497808391,
      "grad_norm": 0.6596114635467529,
      "learning_rate": 0.00016561583577712606,
      "loss": 0.3487,
      "step": 6010
    },
    {
      "epoch": 0.9409830932999373,
      "grad_norm": 0.5770506858825684,
      "learning_rate": 0.00016559139784946234,
      "loss": 0.3152,
      "step": 6011
    },
    {
      "epoch": 0.9411396368190357,
      "grad_norm": 0.6533299088478088,
      "learning_rate": 0.00016556695992179862,
      "loss": 0.3005,
      "step": 6012
    },
    {
      "epoch": 0.941296180338134,
      "grad_norm": 1.0162473917007446,
      "learning_rate": 0.00016554252199413487,
      "loss": 0.5086,
      "step": 6013
    },
    {
      "epoch": 0.9414527238572323,
      "grad_norm": 4.207003116607666,
      "learning_rate": 0.00016551808406647115,
      "loss": 0.4517,
      "step": 6014
    },
    {
      "epoch": 0.9416092673763307,
      "grad_norm": 1.691664695739746,
      "learning_rate": 0.00016549364613880743,
      "loss": 0.4643,
      "step": 6015
    },
    {
      "epoch": 0.9417658108954289,
      "grad_norm": 1.8534256219863892,
      "learning_rate": 0.00016546920821114368,
      "loss": 0.4732,
      "step": 6016
    },
    {
      "epoch": 0.9419223544145272,
      "grad_norm": 1.1564279794692993,
      "learning_rate": 0.00016544477028347996,
      "loss": 0.4391,
      "step": 6017
    },
    {
      "epoch": 0.9420788979336255,
      "grad_norm": 1.8195252418518066,
      "learning_rate": 0.00016542033235581623,
      "loss": 0.7114,
      "step": 6018
    },
    {
      "epoch": 0.9422354414527239,
      "grad_norm": 1.6135766506195068,
      "learning_rate": 0.00016539589442815246,
      "loss": 0.6024,
      "step": 6019
    },
    {
      "epoch": 0.9423919849718222,
      "grad_norm": 2.4532477855682373,
      "learning_rate": 0.00016537145650048874,
      "loss": 0.5552,
      "step": 6020
    },
    {
      "epoch": 0.9425485284909205,
      "grad_norm": 0.8328799605369568,
      "learning_rate": 0.00016534701857282501,
      "loss": 0.3148,
      "step": 6021
    },
    {
      "epoch": 0.9427050720100187,
      "grad_norm": 1.1934763193130493,
      "learning_rate": 0.00016532258064516127,
      "loss": 0.4579,
      "step": 6022
    },
    {
      "epoch": 0.9428616155291171,
      "grad_norm": 1.3018206357955933,
      "learning_rate": 0.00016529814271749754,
      "loss": 0.4106,
      "step": 6023
    },
    {
      "epoch": 0.9430181590482154,
      "grad_norm": 2.117647647857666,
      "learning_rate": 0.00016527370478983382,
      "loss": 0.8004,
      "step": 6024
    },
    {
      "epoch": 0.9431747025673137,
      "grad_norm": 1.9799567461013794,
      "learning_rate": 0.00016524926686217005,
      "loss": 0.4452,
      "step": 6025
    },
    {
      "epoch": 0.9433312460864121,
      "grad_norm": 1.8844330310821533,
      "learning_rate": 0.00016522482893450632,
      "loss": 0.7205,
      "step": 6026
    },
    {
      "epoch": 0.9434877896055104,
      "grad_norm": 2.3716540336608887,
      "learning_rate": 0.0001652003910068426,
      "loss": 0.7866,
      "step": 6027
    },
    {
      "epoch": 0.9436443331246086,
      "grad_norm": 1.5887513160705566,
      "learning_rate": 0.00016517595307917885,
      "loss": 0.8434,
      "step": 6028
    },
    {
      "epoch": 0.9438008766437069,
      "grad_norm": 1.6907641887664795,
      "learning_rate": 0.00016515151515151513,
      "loss": 0.8462,
      "step": 6029
    },
    {
      "epoch": 0.9439574201628053,
      "grad_norm": 4.261603355407715,
      "learning_rate": 0.0001651270772238514,
      "loss": 0.732,
      "step": 6030
    },
    {
      "epoch": 0.9441139636819036,
      "grad_norm": 2.2813000679016113,
      "learning_rate": 0.00016510263929618766,
      "loss": 1.302,
      "step": 6031
    },
    {
      "epoch": 0.9442705072010019,
      "grad_norm": 1.3129147291183472,
      "learning_rate": 0.00016507820136852394,
      "loss": 0.7006,
      "step": 6032
    },
    {
      "epoch": 0.9444270507201001,
      "grad_norm": 3.804840564727783,
      "learning_rate": 0.00016505376344086022,
      "loss": 1.5246,
      "step": 6033
    },
    {
      "epoch": 0.9445835942391985,
      "grad_norm": 6.360110759735107,
      "learning_rate": 0.00016502932551319644,
      "loss": 0.651,
      "step": 6034
    },
    {
      "epoch": 0.9447401377582968,
      "grad_norm": 1.6431469917297363,
      "learning_rate": 0.00016500488758553272,
      "loss": 0.8854,
      "step": 6035
    },
    {
      "epoch": 0.9448966812773951,
      "grad_norm": 2.868042469024658,
      "learning_rate": 0.000164980449657869,
      "loss": 1.4713,
      "step": 6036
    },
    {
      "epoch": 0.9450532247964935,
      "grad_norm": 1.8416922092437744,
      "learning_rate": 0.00016495601173020525,
      "loss": 1.0,
      "step": 6037
    },
    {
      "epoch": 0.9452097683155918,
      "grad_norm": 2.1547200679779053,
      "learning_rate": 0.00016493157380254153,
      "loss": 1.1594,
      "step": 6038
    },
    {
      "epoch": 0.94536631183469,
      "grad_norm": 3.3095543384552,
      "learning_rate": 0.0001649071358748778,
      "loss": 1.5753,
      "step": 6039
    },
    {
      "epoch": 0.9455228553537883,
      "grad_norm": 3.2236099243164062,
      "learning_rate": 0.00016488269794721406,
      "loss": 1.079,
      "step": 6040
    },
    {
      "epoch": 0.9456793988728867,
      "grad_norm": 1.9828205108642578,
      "learning_rate": 0.00016485826001955034,
      "loss": 0.7609,
      "step": 6041
    },
    {
      "epoch": 0.945835942391985,
      "grad_norm": 2.30930757522583,
      "learning_rate": 0.00016483382209188662,
      "loss": 1.2285,
      "step": 6042
    },
    {
      "epoch": 0.9459924859110833,
      "grad_norm": 2.148881435394287,
      "learning_rate": 0.00016480938416422284,
      "loss": 1.3376,
      "step": 6043
    },
    {
      "epoch": 0.9461490294301816,
      "grad_norm": 2.2145028114318848,
      "learning_rate": 0.00016478494623655912,
      "loss": 1.3568,
      "step": 6044
    },
    {
      "epoch": 0.9463055729492799,
      "grad_norm": 3.270275831222534,
      "learning_rate": 0.0001647605083088954,
      "loss": 1.349,
      "step": 6045
    },
    {
      "epoch": 0.9464621164683782,
      "grad_norm": 1.849869728088379,
      "learning_rate": 0.00016473607038123165,
      "loss": 0.9733,
      "step": 6046
    },
    {
      "epoch": 0.9466186599874765,
      "grad_norm": 2.6626498699188232,
      "learning_rate": 0.00016471163245356793,
      "loss": 0.6026,
      "step": 6047
    },
    {
      "epoch": 0.9467752035065748,
      "grad_norm": 2.1792116165161133,
      "learning_rate": 0.0001646871945259042,
      "loss": 0.9302,
      "step": 6048
    },
    {
      "epoch": 0.9469317470256732,
      "grad_norm": 4.216753959655762,
      "learning_rate": 0.00016466275659824043,
      "loss": 1.1798,
      "step": 6049
    },
    {
      "epoch": 0.9470882905447714,
      "grad_norm": 1.9251540899276733,
      "learning_rate": 0.0001646383186705767,
      "loss": 0.8717,
      "step": 6050
    },
    {
      "epoch": 0.9472448340638697,
      "grad_norm": 0.6738594174385071,
      "learning_rate": 0.00016461388074291299,
      "loss": 0.4177,
      "step": 6051
    },
    {
      "epoch": 0.947401377582968,
      "grad_norm": 0.584973156452179,
      "learning_rate": 0.00016458944281524924,
      "loss": 0.3434,
      "step": 6052
    },
    {
      "epoch": 0.9475579211020664,
      "grad_norm": 0.5951725840568542,
      "learning_rate": 0.00016456500488758552,
      "loss": 0.3325,
      "step": 6053
    },
    {
      "epoch": 0.9477144646211647,
      "grad_norm": 0.6660220623016357,
      "learning_rate": 0.0001645405669599218,
      "loss": 0.2946,
      "step": 6054
    },
    {
      "epoch": 0.947871008140263,
      "grad_norm": 0.7421219348907471,
      "learning_rate": 0.00016451612903225804,
      "loss": 0.416,
      "step": 6055
    },
    {
      "epoch": 0.9480275516593613,
      "grad_norm": 0.5771187543869019,
      "learning_rate": 0.00016449169110459432,
      "loss": 0.288,
      "step": 6056
    },
    {
      "epoch": 0.9481840951784596,
      "grad_norm": 0.9346206784248352,
      "learning_rate": 0.0001644672531769306,
      "loss": 0.4272,
      "step": 6057
    },
    {
      "epoch": 0.9483406386975579,
      "grad_norm": 0.5371845364570618,
      "learning_rate": 0.00016444281524926683,
      "loss": 0.3013,
      "step": 6058
    },
    {
      "epoch": 0.9484971822166562,
      "grad_norm": 0.6031778454780579,
      "learning_rate": 0.0001644183773216031,
      "loss": 0.2977,
      "step": 6059
    },
    {
      "epoch": 0.9486537257357546,
      "grad_norm": 0.7967322468757629,
      "learning_rate": 0.00016439393939393938,
      "loss": 0.2708,
      "step": 6060
    },
    {
      "epoch": 0.9488102692548529,
      "grad_norm": 0.7594047784805298,
      "learning_rate": 0.00016436950146627563,
      "loss": 0.3233,
      "step": 6061
    },
    {
      "epoch": 0.9489668127739511,
      "grad_norm": 0.7103455066680908,
      "learning_rate": 0.0001643450635386119,
      "loss": 0.4047,
      "step": 6062
    },
    {
      "epoch": 0.9491233562930494,
      "grad_norm": 0.8710353970527649,
      "learning_rate": 0.0001643206256109482,
      "loss": 0.3425,
      "step": 6063
    },
    {
      "epoch": 0.9492798998121478,
      "grad_norm": 0.7795736193656921,
      "learning_rate": 0.00016429618768328444,
      "loss": 0.4108,
      "step": 6064
    },
    {
      "epoch": 0.9494364433312461,
      "grad_norm": 1.0512408018112183,
      "learning_rate": 0.00016427174975562072,
      "loss": 0.4573,
      "step": 6065
    },
    {
      "epoch": 0.9495929868503444,
      "grad_norm": 2.8175127506256104,
      "learning_rate": 0.00016424731182795697,
      "loss": 0.5165,
      "step": 6066
    },
    {
      "epoch": 0.9497495303694427,
      "grad_norm": 1.3486829996109009,
      "learning_rate": 0.00016422287390029322,
      "loss": 0.5704,
      "step": 6067
    },
    {
      "epoch": 0.949906073888541,
      "grad_norm": 1.54037344455719,
      "learning_rate": 0.0001641984359726295,
      "loss": 0.4442,
      "step": 6068
    },
    {
      "epoch": 0.9500626174076393,
      "grad_norm": 0.9929664134979248,
      "learning_rate": 0.00016417399804496578,
      "loss": 0.5064,
      "step": 6069
    },
    {
      "epoch": 0.9502191609267376,
      "grad_norm": 1.431089997291565,
      "learning_rate": 0.00016414956011730203,
      "loss": 0.5722,
      "step": 6070
    },
    {
      "epoch": 0.950375704445836,
      "grad_norm": 1.4068636894226074,
      "learning_rate": 0.0001641251221896383,
      "loss": 0.6155,
      "step": 6071
    },
    {
      "epoch": 0.9505322479649343,
      "grad_norm": 1.600339651107788,
      "learning_rate": 0.0001641006842619746,
      "loss": 0.5878,
      "step": 6072
    },
    {
      "epoch": 0.9506887914840325,
      "grad_norm": 1.6295877695083618,
      "learning_rate": 0.0001640762463343108,
      "loss": 0.6764,
      "step": 6073
    },
    {
      "epoch": 0.9508453350031308,
      "grad_norm": 1.2888362407684326,
      "learning_rate": 0.0001640518084066471,
      "loss": 0.5213,
      "step": 6074
    },
    {
      "epoch": 0.9510018785222292,
      "grad_norm": 1.4205604791641235,
      "learning_rate": 0.00016402737047898337,
      "loss": 0.6027,
      "step": 6075
    },
    {
      "epoch": 0.9511584220413275,
      "grad_norm": 1.4100985527038574,
      "learning_rate": 0.00016400293255131962,
      "loss": 0.6991,
      "step": 6076
    },
    {
      "epoch": 0.9513149655604258,
      "grad_norm": 2.3864524364471436,
      "learning_rate": 0.0001639784946236559,
      "loss": 0.6062,
      "step": 6077
    },
    {
      "epoch": 0.9514715090795242,
      "grad_norm": 1.8597694635391235,
      "learning_rate": 0.00016395405669599218,
      "loss": 0.5013,
      "step": 6078
    },
    {
      "epoch": 0.9516280525986224,
      "grad_norm": 1.8252570629119873,
      "learning_rate": 0.00016392961876832843,
      "loss": 0.8579,
      "step": 6079
    },
    {
      "epoch": 0.9517845961177207,
      "grad_norm": 1.7800219058990479,
      "learning_rate": 0.0001639051808406647,
      "loss": 0.6946,
      "step": 6080
    },
    {
      "epoch": 0.951941139636819,
      "grad_norm": 1.8645308017730713,
      "learning_rate": 0.00016388074291300098,
      "loss": 1.0379,
      "step": 6081
    },
    {
      "epoch": 0.9520976831559174,
      "grad_norm": 2.3539228439331055,
      "learning_rate": 0.0001638563049853372,
      "loss": 0.9241,
      "step": 6082
    },
    {
      "epoch": 0.9522542266750157,
      "grad_norm": 5.200006484985352,
      "learning_rate": 0.00016383186705767349,
      "loss": 1.2359,
      "step": 6083
    },
    {
      "epoch": 0.9524107701941139,
      "grad_norm": 1.8422434329986572,
      "learning_rate": 0.00016380742913000976,
      "loss": 1.3175,
      "step": 6084
    },
    {
      "epoch": 0.9525673137132122,
      "grad_norm": 3.778296709060669,
      "learning_rate": 0.00016378299120234602,
      "loss": 0.979,
      "step": 6085
    },
    {
      "epoch": 0.9527238572323106,
      "grad_norm": 2.5928831100463867,
      "learning_rate": 0.0001637585532746823,
      "loss": 1.2131,
      "step": 6086
    },
    {
      "epoch": 0.9528804007514089,
      "grad_norm": 0.75013267993927,
      "learning_rate": 0.00016373411534701857,
      "loss": 0.6146,
      "step": 6087
    },
    {
      "epoch": 0.9530369442705072,
      "grad_norm": 3.9112038612365723,
      "learning_rate": 0.00016370967741935482,
      "loss": 0.7088,
      "step": 6088
    },
    {
      "epoch": 0.9531934877896056,
      "grad_norm": 1.6564382314682007,
      "learning_rate": 0.0001636852394916911,
      "loss": 1.0111,
      "step": 6089
    },
    {
      "epoch": 0.9533500313087038,
      "grad_norm": 4.030745506286621,
      "learning_rate": 0.00016366080156402735,
      "loss": 1.3784,
      "step": 6090
    },
    {
      "epoch": 0.9535065748278021,
      "grad_norm": 3.7477824687957764,
      "learning_rate": 0.0001636363636363636,
      "loss": 1.2618,
      "step": 6091
    },
    {
      "epoch": 0.9536631183469004,
      "grad_norm": 4.694358825683594,
      "learning_rate": 0.00016361192570869988,
      "loss": 1.029,
      "step": 6092
    },
    {
      "epoch": 0.9538196618659988,
      "grad_norm": 2.626904010772705,
      "learning_rate": 0.00016358748778103616,
      "loss": 1.1996,
      "step": 6093
    },
    {
      "epoch": 0.9539762053850971,
      "grad_norm": 1.2721118927001953,
      "learning_rate": 0.0001635630498533724,
      "loss": 0.6772,
      "step": 6094
    },
    {
      "epoch": 0.9541327489041954,
      "grad_norm": 2.771449327468872,
      "learning_rate": 0.0001635386119257087,
      "loss": 1.4234,
      "step": 6095
    },
    {
      "epoch": 0.9542892924232936,
      "grad_norm": 3.96138858795166,
      "learning_rate": 0.00016351417399804497,
      "loss": 1.236,
      "step": 6096
    },
    {
      "epoch": 0.954445835942392,
      "grad_norm": 1.7881155014038086,
      "learning_rate": 0.0001634897360703812,
      "loss": 0.3081,
      "step": 6097
    },
    {
      "epoch": 0.9546023794614903,
      "grad_norm": 3.0272154808044434,
      "learning_rate": 0.00016346529814271747,
      "loss": 1.5357,
      "step": 6098
    },
    {
      "epoch": 0.9547589229805886,
      "grad_norm": 1.5907299518585205,
      "learning_rate": 0.00016344086021505375,
      "loss": 0.485,
      "step": 6099
    },
    {
      "epoch": 0.954915466499687,
      "grad_norm": 2.836285352706909,
      "learning_rate": 0.00016341642228739,
      "loss": 1.5678,
      "step": 6100
    },
    {
      "epoch": 0.9550720100187852,
      "grad_norm": 0.5653170347213745,
      "learning_rate": 0.00016339198435972628,
      "loss": 0.3355,
      "step": 6101
    },
    {
      "epoch": 0.9552285535378835,
      "grad_norm": 0.646906316280365,
      "learning_rate": 0.00016336754643206256,
      "loss": 0.3155,
      "step": 6102
    },
    {
      "epoch": 0.9553850970569818,
      "grad_norm": 0.7139442563056946,
      "learning_rate": 0.0001633431085043988,
      "loss": 0.3738,
      "step": 6103
    },
    {
      "epoch": 0.9555416405760802,
      "grad_norm": 0.612605094909668,
      "learning_rate": 0.0001633186705767351,
      "loss": 0.2725,
      "step": 6104
    },
    {
      "epoch": 0.9556981840951785,
      "grad_norm": 0.4651452898979187,
      "learning_rate": 0.00016329423264907137,
      "loss": 0.291,
      "step": 6105
    },
    {
      "epoch": 0.9558547276142768,
      "grad_norm": 0.6952741146087646,
      "learning_rate": 0.0001632697947214076,
      "loss": 0.2871,
      "step": 6106
    },
    {
      "epoch": 0.956011271133375,
      "grad_norm": 0.5448809266090393,
      "learning_rate": 0.00016324535679374387,
      "loss": 0.2545,
      "step": 6107
    },
    {
      "epoch": 0.9561678146524734,
      "grad_norm": 0.9530249834060669,
      "learning_rate": 0.00016322091886608015,
      "loss": 0.3625,
      "step": 6108
    },
    {
      "epoch": 0.9563243581715717,
      "grad_norm": 0.5093350410461426,
      "learning_rate": 0.0001631964809384164,
      "loss": 0.3335,
      "step": 6109
    },
    {
      "epoch": 0.95648090169067,
      "grad_norm": 0.7554910778999329,
      "learning_rate": 0.00016317204301075268,
      "loss": 0.3436,
      "step": 6110
    },
    {
      "epoch": 0.9566374452097683,
      "grad_norm": 0.7633286118507385,
      "learning_rate": 0.00016314760508308895,
      "loss": 0.3616,
      "step": 6111
    },
    {
      "epoch": 0.9567939887288667,
      "grad_norm": 0.7461150288581848,
      "learning_rate": 0.0001631231671554252,
      "loss": 0.3472,
      "step": 6112
    },
    {
      "epoch": 0.9569505322479649,
      "grad_norm": 1.1075865030288696,
      "learning_rate": 0.00016309872922776146,
      "loss": 0.4187,
      "step": 6113
    },
    {
      "epoch": 0.9571070757670632,
      "grad_norm": 1.0741170644760132,
      "learning_rate": 0.00016307429130009774,
      "loss": 0.5136,
      "step": 6114
    },
    {
      "epoch": 0.9572636192861615,
      "grad_norm": 0.8717575073242188,
      "learning_rate": 0.000163049853372434,
      "loss": 0.317,
      "step": 6115
    },
    {
      "epoch": 0.9574201628052599,
      "grad_norm": 2.2658355236053467,
      "learning_rate": 0.00016302541544477027,
      "loss": 0.4035,
      "step": 6116
    },
    {
      "epoch": 0.9575767063243582,
      "grad_norm": 1.5720932483673096,
      "learning_rate": 0.00016300097751710654,
      "loss": 0.7772,
      "step": 6117
    },
    {
      "epoch": 0.9577332498434565,
      "grad_norm": 2.046311140060425,
      "learning_rate": 0.0001629765395894428,
      "loss": 0.4958,
      "step": 6118
    },
    {
      "epoch": 0.9578897933625548,
      "grad_norm": 1.5623672008514404,
      "learning_rate": 0.00016295210166177907,
      "loss": 0.6284,
      "step": 6119
    },
    {
      "epoch": 0.9580463368816531,
      "grad_norm": 2.6061389446258545,
      "learning_rate": 0.00016292766373411535,
      "loss": 0.9757,
      "step": 6120
    },
    {
      "epoch": 0.9582028804007514,
      "grad_norm": 1.6166800260543823,
      "learning_rate": 0.00016290322580645158,
      "loss": 0.5318,
      "step": 6121
    },
    {
      "epoch": 0.9583594239198497,
      "grad_norm": 1.0401281118392944,
      "learning_rate": 0.00016287878787878785,
      "loss": 0.6959,
      "step": 6122
    },
    {
      "epoch": 0.9585159674389481,
      "grad_norm": 2.1870529651641846,
      "learning_rate": 0.00016285434995112413,
      "loss": 0.6678,
      "step": 6123
    },
    {
      "epoch": 0.9586725109580463,
      "grad_norm": 0.9855921864509583,
      "learning_rate": 0.00016282991202346038,
      "loss": 0.3948,
      "step": 6124
    },
    {
      "epoch": 0.9588290544771446,
      "grad_norm": 1.3813424110412598,
      "learning_rate": 0.00016280547409579666,
      "loss": 1.0385,
      "step": 6125
    },
    {
      "epoch": 0.9589855979962429,
      "grad_norm": 1.2955299615859985,
      "learning_rate": 0.00016278103616813294,
      "loss": 0.4331,
      "step": 6126
    },
    {
      "epoch": 0.9591421415153413,
      "grad_norm": 2.7574148178100586,
      "learning_rate": 0.0001627565982404692,
      "loss": 1.1984,
      "step": 6127
    },
    {
      "epoch": 0.9592986850344396,
      "grad_norm": 2.4210774898529053,
      "learning_rate": 0.00016273216031280547,
      "loss": 0.7977,
      "step": 6128
    },
    {
      "epoch": 0.9594552285535379,
      "grad_norm": 4.516207695007324,
      "learning_rate": 0.00016270772238514175,
      "loss": 0.9785,
      "step": 6129
    },
    {
      "epoch": 0.9596117720726361,
      "grad_norm": 1.7034456729888916,
      "learning_rate": 0.00016268328445747797,
      "loss": 0.8274,
      "step": 6130
    },
    {
      "epoch": 0.9597683155917345,
      "grad_norm": 1.6179447174072266,
      "learning_rate": 0.00016265884652981425,
      "loss": 0.8648,
      "step": 6131
    },
    {
      "epoch": 0.9599248591108328,
      "grad_norm": 1.179610013961792,
      "learning_rate": 0.00016263440860215053,
      "loss": 0.49,
      "step": 6132
    },
    {
      "epoch": 0.9600814026299311,
      "grad_norm": 2.1187167167663574,
      "learning_rate": 0.00016260997067448678,
      "loss": 0.9443,
      "step": 6133
    },
    {
      "epoch": 0.9602379461490295,
      "grad_norm": 2.0016579627990723,
      "learning_rate": 0.00016258553274682306,
      "loss": 0.9851,
      "step": 6134
    },
    {
      "epoch": 0.9603944896681278,
      "grad_norm": 2.0684750080108643,
      "learning_rate": 0.00016256109481915934,
      "loss": 0.9574,
      "step": 6135
    },
    {
      "epoch": 0.960551033187226,
      "grad_norm": 2.0303995609283447,
      "learning_rate": 0.0001625366568914956,
      "loss": 1.1024,
      "step": 6136
    },
    {
      "epoch": 0.9607075767063243,
      "grad_norm": 3.4601995944976807,
      "learning_rate": 0.00016251221896383184,
      "loss": 0.6025,
      "step": 6137
    },
    {
      "epoch": 0.9608641202254227,
      "grad_norm": 3.948913097381592,
      "learning_rate": 0.00016248778103616812,
      "loss": 0.6272,
      "step": 6138
    },
    {
      "epoch": 0.961020663744521,
      "grad_norm": 9.379945755004883,
      "learning_rate": 0.00016246334310850437,
      "loss": 1.3707,
      "step": 6139
    },
    {
      "epoch": 0.9611772072636193,
      "grad_norm": 1.712647557258606,
      "learning_rate": 0.00016243890518084065,
      "loss": 1.0695,
      "step": 6140
    },
    {
      "epoch": 0.9613337507827175,
      "grad_norm": 2.5404181480407715,
      "learning_rate": 0.00016241446725317693,
      "loss": 1.3814,
      "step": 6141
    },
    {
      "epoch": 0.9614902943018159,
      "grad_norm": 1.7662317752838135,
      "learning_rate": 0.00016239002932551318,
      "loss": 1.2451,
      "step": 6142
    },
    {
      "epoch": 0.9616468378209142,
      "grad_norm": 3.7080023288726807,
      "learning_rate": 0.00016236559139784946,
      "loss": 1.4876,
      "step": 6143
    },
    {
      "epoch": 0.9618033813400125,
      "grad_norm": 2.5112531185150146,
      "learning_rate": 0.00016234115347018573,
      "loss": 1.0143,
      "step": 6144
    },
    {
      "epoch": 0.9619599248591109,
      "grad_norm": 2.770322322845459,
      "learning_rate": 0.00016231671554252196,
      "loss": 1.2868,
      "step": 6145
    },
    {
      "epoch": 0.9621164683782092,
      "grad_norm": 1.875613808631897,
      "learning_rate": 0.00016229227761485824,
      "loss": 1.1641,
      "step": 6146
    },
    {
      "epoch": 0.9622730118973074,
      "grad_norm": 3.305671215057373,
      "learning_rate": 0.00016226783968719451,
      "loss": 0.8963,
      "step": 6147
    },
    {
      "epoch": 0.9624295554164057,
      "grad_norm": 1.5809119939804077,
      "learning_rate": 0.00016224340175953077,
      "loss": 0.6262,
      "step": 6148
    },
    {
      "epoch": 0.9625860989355041,
      "grad_norm": 2.6327903270721436,
      "learning_rate": 0.00016221896383186704,
      "loss": 0.4588,
      "step": 6149
    },
    {
      "epoch": 0.9627426424546024,
      "grad_norm": 1.2948060035705566,
      "learning_rate": 0.00016219452590420332,
      "loss": 1.0183,
      "step": 6150
    },
    {
      "epoch": 0.9628991859737007,
      "grad_norm": 0.6153829097747803,
      "learning_rate": 0.00016217008797653957,
      "loss": 0.2905,
      "step": 6151
    },
    {
      "epoch": 0.963055729492799,
      "grad_norm": 0.7002934813499451,
      "learning_rate": 0.00016214565004887585,
      "loss": 0.3681,
      "step": 6152
    },
    {
      "epoch": 0.9632122730118973,
      "grad_norm": 0.4247320294380188,
      "learning_rate": 0.00016212121212121213,
      "loss": 0.2522,
      "step": 6153
    },
    {
      "epoch": 0.9633688165309956,
      "grad_norm": 1.023521900177002,
      "learning_rate": 0.00016209677419354835,
      "loss": 0.4101,
      "step": 6154
    },
    {
      "epoch": 0.9635253600500939,
      "grad_norm": 0.9346298575401306,
      "learning_rate": 0.00016207233626588463,
      "loss": 0.4846,
      "step": 6155
    },
    {
      "epoch": 0.9636819035691923,
      "grad_norm": 1.02970290184021,
      "learning_rate": 0.0001620478983382209,
      "loss": 0.526,
      "step": 6156
    },
    {
      "epoch": 0.9638384470882906,
      "grad_norm": 0.6798651218414307,
      "learning_rate": 0.00016202346041055716,
      "loss": 0.346,
      "step": 6157
    },
    {
      "epoch": 0.9639949906073888,
      "grad_norm": 0.7402242422103882,
      "learning_rate": 0.00016199902248289344,
      "loss": 0.3458,
      "step": 6158
    },
    {
      "epoch": 0.9641515341264871,
      "grad_norm": 0.9456111788749695,
      "learning_rate": 0.00016197458455522972,
      "loss": 0.328,
      "step": 6159
    },
    {
      "epoch": 0.9643080776455855,
      "grad_norm": 0.9947827458381653,
      "learning_rate": 0.00016195014662756597,
      "loss": 0.4049,
      "step": 6160
    },
    {
      "epoch": 0.9644646211646838,
      "grad_norm": 0.9443894028663635,
      "learning_rate": 0.00016192570869990222,
      "loss": 0.2766,
      "step": 6161
    },
    {
      "epoch": 0.9646211646837821,
      "grad_norm": 0.9167144298553467,
      "learning_rate": 0.0001619012707722385,
      "loss": 0.4539,
      "step": 6162
    },
    {
      "epoch": 0.9647777082028804,
      "grad_norm": 1.195816993713379,
      "learning_rate": 0.00016187683284457475,
      "loss": 0.608,
      "step": 6163
    },
    {
      "epoch": 0.9649342517219787,
      "grad_norm": 1.4440490007400513,
      "learning_rate": 0.00016185239491691103,
      "loss": 0.5446,
      "step": 6164
    },
    {
      "epoch": 0.965090795241077,
      "grad_norm": 0.7405814528465271,
      "learning_rate": 0.0001618279569892473,
      "loss": 0.3242,
      "step": 6165
    },
    {
      "epoch": 0.9652473387601753,
      "grad_norm": 1.427616000175476,
      "learning_rate": 0.00016180351906158356,
      "loss": 0.4028,
      "step": 6166
    },
    {
      "epoch": 0.9654038822792737,
      "grad_norm": 1.6844470500946045,
      "learning_rate": 0.00016177908113391984,
      "loss": 0.6549,
      "step": 6167
    },
    {
      "epoch": 0.965560425798372,
      "grad_norm": 2.0406758785247803,
      "learning_rate": 0.00016175464320625612,
      "loss": 0.4393,
      "step": 6168
    },
    {
      "epoch": 0.9657169693174703,
      "grad_norm": 1.3769197463989258,
      "learning_rate": 0.00016173020527859234,
      "loss": 0.4588,
      "step": 6169
    },
    {
      "epoch": 0.9658735128365685,
      "grad_norm": 1.0385569334030151,
      "learning_rate": 0.00016170576735092862,
      "loss": 0.4175,
      "step": 6170
    },
    {
      "epoch": 0.9660300563556669,
      "grad_norm": 1.2535983324050903,
      "learning_rate": 0.0001616813294232649,
      "loss": 0.4402,
      "step": 6171
    },
    {
      "epoch": 0.9661865998747652,
      "grad_norm": 1.0669677257537842,
      "learning_rate": 0.00016165689149560115,
      "loss": 0.3781,
      "step": 6172
    },
    {
      "epoch": 0.9663431433938635,
      "grad_norm": 1.9178048372268677,
      "learning_rate": 0.00016163245356793743,
      "loss": 0.5489,
      "step": 6173
    },
    {
      "epoch": 0.9664996869129618,
      "grad_norm": 1.401336908340454,
      "learning_rate": 0.0001616080156402737,
      "loss": 0.6525,
      "step": 6174
    },
    {
      "epoch": 0.9666562304320601,
      "grad_norm": 1.5251128673553467,
      "learning_rate": 0.00016158357771260996,
      "loss": 0.8831,
      "step": 6175
    },
    {
      "epoch": 0.9668127739511584,
      "grad_norm": 2.3394248485565186,
      "learning_rate": 0.00016155913978494623,
      "loss": 0.906,
      "step": 6176
    },
    {
      "epoch": 0.9669693174702567,
      "grad_norm": 1.7218315601348877,
      "learning_rate": 0.0001615347018572825,
      "loss": 0.7642,
      "step": 6177
    },
    {
      "epoch": 0.967125860989355,
      "grad_norm": 1.3168402910232544,
      "learning_rate": 0.00016151026392961874,
      "loss": 0.6152,
      "step": 6178
    },
    {
      "epoch": 0.9672824045084534,
      "grad_norm": 1.864134430885315,
      "learning_rate": 0.00016148582600195502,
      "loss": 1.0541,
      "step": 6179
    },
    {
      "epoch": 0.9674389480275517,
      "grad_norm": 2.145655870437622,
      "learning_rate": 0.0001614613880742913,
      "loss": 0.9664,
      "step": 6180
    },
    {
      "epoch": 0.9675954915466499,
      "grad_norm": 1.7807810306549072,
      "learning_rate": 0.00016143695014662755,
      "loss": 0.7563,
      "step": 6181
    },
    {
      "epoch": 0.9677520350657483,
      "grad_norm": 2.374579906463623,
      "learning_rate": 0.00016141251221896382,
      "loss": 0.7937,
      "step": 6182
    },
    {
      "epoch": 0.9679085785848466,
      "grad_norm": 2.7262957096099854,
      "learning_rate": 0.0001613880742913001,
      "loss": 1.1843,
      "step": 6183
    },
    {
      "epoch": 0.9680651221039449,
      "grad_norm": 2.082878589630127,
      "learning_rate": 0.00016136363636363633,
      "loss": 1.443,
      "step": 6184
    },
    {
      "epoch": 0.9682216656230432,
      "grad_norm": 3.2135493755340576,
      "learning_rate": 0.0001613391984359726,
      "loss": 0.9447,
      "step": 6185
    },
    {
      "epoch": 0.9683782091421416,
      "grad_norm": 1.5901567935943604,
      "learning_rate": 0.00016131476050830888,
      "loss": 0.7122,
      "step": 6186
    },
    {
      "epoch": 0.9685347526612398,
      "grad_norm": 1.4855965375900269,
      "learning_rate": 0.00016129032258064513,
      "loss": 1.1293,
      "step": 6187
    },
    {
      "epoch": 0.9686912961803381,
      "grad_norm": 3.177372694015503,
      "learning_rate": 0.0001612658846529814,
      "loss": 1.2107,
      "step": 6188
    },
    {
      "epoch": 0.9688478396994364,
      "grad_norm": 2.5584800243377686,
      "learning_rate": 0.0001612414467253177,
      "loss": 0.9807,
      "step": 6189
    },
    {
      "epoch": 0.9690043832185348,
      "grad_norm": 1.333678960800171,
      "learning_rate": 0.00016121700879765394,
      "loss": 0.876,
      "step": 6190
    },
    {
      "epoch": 0.9691609267376331,
      "grad_norm": 2.623502254486084,
      "learning_rate": 0.00016119257086999022,
      "loss": 0.8509,
      "step": 6191
    },
    {
      "epoch": 0.9693174702567313,
      "grad_norm": 1.4758764505386353,
      "learning_rate": 0.0001611681329423265,
      "loss": 1.3429,
      "step": 6192
    },
    {
      "epoch": 0.9694740137758296,
      "grad_norm": 1.8458406925201416,
      "learning_rate": 0.00016114369501466272,
      "loss": 1.4386,
      "step": 6193
    },
    {
      "epoch": 0.969630557294928,
      "grad_norm": 1.940486192703247,
      "learning_rate": 0.000161119257086999,
      "loss": 0.7314,
      "step": 6194
    },
    {
      "epoch": 0.9697871008140263,
      "grad_norm": 2.4906039237976074,
      "learning_rate": 0.00016109481915933528,
      "loss": 1.3775,
      "step": 6195
    },
    {
      "epoch": 0.9699436443331246,
      "grad_norm": 1.3360508680343628,
      "learning_rate": 0.00016107038123167153,
      "loss": 0.4396,
      "step": 6196
    },
    {
      "epoch": 0.970100187852223,
      "grad_norm": 3.2190940380096436,
      "learning_rate": 0.0001610459433040078,
      "loss": 0.7913,
      "step": 6197
    },
    {
      "epoch": 0.9702567313713212,
      "grad_norm": 1.5043747425079346,
      "learning_rate": 0.0001610215053763441,
      "loss": 0.5736,
      "step": 6198
    },
    {
      "epoch": 0.9704132748904195,
      "grad_norm": 2.0739946365356445,
      "learning_rate": 0.00016099706744868034,
      "loss": 0.775,
      "step": 6199
    },
    {
      "epoch": 0.9705698184095178,
      "grad_norm": 3.4089341163635254,
      "learning_rate": 0.00016097262952101662,
      "loss": 1.0023,
      "step": 6200
    },
    {
      "epoch": 0.9707263619286162,
      "grad_norm": 0.49759766459465027,
      "learning_rate": 0.0001609481915933529,
      "loss": 0.3349,
      "step": 6201
    },
    {
      "epoch": 0.9708829054477145,
      "grad_norm": 0.44160133600234985,
      "learning_rate": 0.00016092375366568912,
      "loss": 0.2845,
      "step": 6202
    },
    {
      "epoch": 0.9710394489668128,
      "grad_norm": 0.6713477969169617,
      "learning_rate": 0.0001608993157380254,
      "loss": 0.3082,
      "step": 6203
    },
    {
      "epoch": 0.971195992485911,
      "grad_norm": 0.48395052552223206,
      "learning_rate": 0.00016087487781036168,
      "loss": 0.2274,
      "step": 6204
    },
    {
      "epoch": 0.9713525360050094,
      "grad_norm": 0.5824487805366516,
      "learning_rate": 0.00016085043988269793,
      "loss": 0.3094,
      "step": 6205
    },
    {
      "epoch": 0.9715090795241077,
      "grad_norm": 0.6694998145103455,
      "learning_rate": 0.0001608260019550342,
      "loss": 0.2379,
      "step": 6206
    },
    {
      "epoch": 0.971665623043206,
      "grad_norm": 0.701755166053772,
      "learning_rate": 0.00016080156402737048,
      "loss": 0.2673,
      "step": 6207
    },
    {
      "epoch": 0.9718221665623044,
      "grad_norm": 0.9201617240905762,
      "learning_rate": 0.0001607771260997067,
      "loss": 0.403,
      "step": 6208
    },
    {
      "epoch": 0.9719787100814026,
      "grad_norm": 0.9466022849082947,
      "learning_rate": 0.000160752688172043,
      "loss": 0.3029,
      "step": 6209
    },
    {
      "epoch": 0.9721352536005009,
      "grad_norm": 1.3907926082611084,
      "learning_rate": 0.00016072825024437927,
      "loss": 0.3822,
      "step": 6210
    },
    {
      "epoch": 0.9722917971195992,
      "grad_norm": 0.9519656300544739,
      "learning_rate": 0.00016070381231671552,
      "loss": 0.5414,
      "step": 6211
    },
    {
      "epoch": 0.9724483406386976,
      "grad_norm": 1.2506569623947144,
      "learning_rate": 0.0001606793743890518,
      "loss": 0.5634,
      "step": 6212
    },
    {
      "epoch": 0.9726048841577959,
      "grad_norm": 0.7912557125091553,
      "learning_rate": 0.00016065493646138807,
      "loss": 0.4921,
      "step": 6213
    },
    {
      "epoch": 0.9727614276768942,
      "grad_norm": 1.3480660915374756,
      "learning_rate": 0.00016063049853372432,
      "loss": 0.3606,
      "step": 6214
    },
    {
      "epoch": 0.9729179711959924,
      "grad_norm": 1.24032461643219,
      "learning_rate": 0.0001606060606060606,
      "loss": 0.4046,
      "step": 6215
    },
    {
      "epoch": 0.9730745147150908,
      "grad_norm": 1.4148228168487549,
      "learning_rate": 0.00016058162267839688,
      "loss": 0.6278,
      "step": 6216
    },
    {
      "epoch": 0.9732310582341891,
      "grad_norm": 2.0846927165985107,
      "learning_rate": 0.0001605571847507331,
      "loss": 0.6703,
      "step": 6217
    },
    {
      "epoch": 0.9733876017532874,
      "grad_norm": 2.4335579872131348,
      "learning_rate": 0.00016053274682306938,
      "loss": 0.6913,
      "step": 6218
    },
    {
      "epoch": 0.9735441452723858,
      "grad_norm": 1.5219314098358154,
      "learning_rate": 0.00016050830889540566,
      "loss": 0.4337,
      "step": 6219
    },
    {
      "epoch": 0.9737006887914841,
      "grad_norm": 3.503244400024414,
      "learning_rate": 0.0001604838709677419,
      "loss": 0.3213,
      "step": 6220
    },
    {
      "epoch": 0.9738572323105823,
      "grad_norm": 1.0507615804672241,
      "learning_rate": 0.0001604594330400782,
      "loss": 0.3756,
      "step": 6221
    },
    {
      "epoch": 0.9740137758296806,
      "grad_norm": 1.3683404922485352,
      "learning_rate": 0.00016043499511241447,
      "loss": 0.808,
      "step": 6222
    },
    {
      "epoch": 0.974170319348779,
      "grad_norm": 3.0311946868896484,
      "learning_rate": 0.00016041055718475072,
      "loss": 0.7998,
      "step": 6223
    },
    {
      "epoch": 0.9743268628678773,
      "grad_norm": 1.0792346000671387,
      "learning_rate": 0.000160386119257087,
      "loss": 0.473,
      "step": 6224
    },
    {
      "epoch": 0.9744834063869756,
      "grad_norm": 1.5082753896713257,
      "learning_rate": 0.00016036168132942325,
      "loss": 0.4845,
      "step": 6225
    },
    {
      "epoch": 0.9746399499060739,
      "grad_norm": 1.629244089126587,
      "learning_rate": 0.0001603372434017595,
      "loss": 0.4439,
      "step": 6226
    },
    {
      "epoch": 0.9747964934251722,
      "grad_norm": 2.6280176639556885,
      "learning_rate": 0.00016031280547409578,
      "loss": 0.7095,
      "step": 6227
    },
    {
      "epoch": 0.9749530369442705,
      "grad_norm": 1.1485744714736938,
      "learning_rate": 0.00016028836754643206,
      "loss": 0.6459,
      "step": 6228
    },
    {
      "epoch": 0.9751095804633688,
      "grad_norm": 3.294517755508423,
      "learning_rate": 0.0001602639296187683,
      "loss": 0.4601,
      "step": 6229
    },
    {
      "epoch": 0.9752661239824671,
      "grad_norm": 2.4806206226348877,
      "learning_rate": 0.0001602394916911046,
      "loss": 0.9412,
      "step": 6230
    },
    {
      "epoch": 0.9754226675015655,
      "grad_norm": 1.6524327993392944,
      "learning_rate": 0.00016021505376344087,
      "loss": 0.6169,
      "step": 6231
    },
    {
      "epoch": 0.9755792110206637,
      "grad_norm": 1.4262551069259644,
      "learning_rate": 0.0001601906158357771,
      "loss": 0.7836,
      "step": 6232
    },
    {
      "epoch": 0.975735754539762,
      "grad_norm": 1.37894868850708,
      "learning_rate": 0.00016016617790811337,
      "loss": 0.3791,
      "step": 6233
    },
    {
      "epoch": 0.9758922980588604,
      "grad_norm": 3.260524272918701,
      "learning_rate": 0.00016014173998044965,
      "loss": 1.2778,
      "step": 6234
    },
    {
      "epoch": 0.9760488415779587,
      "grad_norm": 1.1274797916412354,
      "learning_rate": 0.0001601173020527859,
      "loss": 0.5008,
      "step": 6235
    },
    {
      "epoch": 0.976205385097057,
      "grad_norm": 3.6673662662506104,
      "learning_rate": 0.00016009286412512218,
      "loss": 1.2082,
      "step": 6236
    },
    {
      "epoch": 0.9763619286161553,
      "grad_norm": 2.9804086685180664,
      "learning_rate": 0.00016006842619745846,
      "loss": 0.8466,
      "step": 6237
    },
    {
      "epoch": 0.9765184721352536,
      "grad_norm": 2.351480007171631,
      "learning_rate": 0.0001600439882697947,
      "loss": 1.1366,
      "step": 6238
    },
    {
      "epoch": 0.9766750156543519,
      "grad_norm": 2.855189561843872,
      "learning_rate": 0.00016001955034213098,
      "loss": 0.8895,
      "step": 6239
    },
    {
      "epoch": 0.9768315591734502,
      "grad_norm": 3.6381757259368896,
      "learning_rate": 0.00015999511241446726,
      "loss": 1.5563,
      "step": 6240
    },
    {
      "epoch": 0.9769881026925485,
      "grad_norm": 1.6015832424163818,
      "learning_rate": 0.0001599706744868035,
      "loss": 1.0543,
      "step": 6241
    },
    {
      "epoch": 0.9771446462116469,
      "grad_norm": 1.6288787126541138,
      "learning_rate": 0.00015994623655913977,
      "loss": 0.8863,
      "step": 6242
    },
    {
      "epoch": 0.9773011897307452,
      "grad_norm": 3.075251579284668,
      "learning_rate": 0.00015992179863147604,
      "loss": 1.1518,
      "step": 6243
    },
    {
      "epoch": 0.9774577332498434,
      "grad_norm": 1.8009350299835205,
      "learning_rate": 0.0001598973607038123,
      "loss": 1.1322,
      "step": 6244
    },
    {
      "epoch": 0.9776142767689417,
      "grad_norm": 1.083701491355896,
      "learning_rate": 0.00015987292277614857,
      "loss": 0.5735,
      "step": 6245
    },
    {
      "epoch": 0.9777708202880401,
      "grad_norm": 3.801417589187622,
      "learning_rate": 0.00015984848484848485,
      "loss": 1.359,
      "step": 6246
    },
    {
      "epoch": 0.9779273638071384,
      "grad_norm": 3.566204309463501,
      "learning_rate": 0.0001598240469208211,
      "loss": 1.027,
      "step": 6247
    },
    {
      "epoch": 0.9780839073262367,
      "grad_norm": 1.9416583776474,
      "learning_rate": 0.00015979960899315738,
      "loss": 1.3133,
      "step": 6248
    },
    {
      "epoch": 0.978240450845335,
      "grad_norm": 4.796366214752197,
      "learning_rate": 0.00015977517106549363,
      "loss": 1.1721,
      "step": 6249
    },
    {
      "epoch": 0.9783969943644333,
      "grad_norm": 2.206047296524048,
      "learning_rate": 0.00015975073313782988,
      "loss": 1.2208,
      "step": 6250
    },
    {
      "epoch": 0.9785535378835316,
      "grad_norm": 0.7884498834609985,
      "learning_rate": 0.00015972629521016616,
      "loss": 0.3513,
      "step": 6251
    },
    {
      "epoch": 0.9787100814026299,
      "grad_norm": 0.5849615931510925,
      "learning_rate": 0.00015970185728250244,
      "loss": 0.2797,
      "step": 6252
    },
    {
      "epoch": 0.9788666249217283,
      "grad_norm": 0.5109837651252747,
      "learning_rate": 0.0001596774193548387,
      "loss": 0.3183,
      "step": 6253
    },
    {
      "epoch": 0.9790231684408266,
      "grad_norm": 0.5798192620277405,
      "learning_rate": 0.00015965298142717497,
      "loss": 0.2323,
      "step": 6254
    },
    {
      "epoch": 0.9791797119599248,
      "grad_norm": 0.5868884921073914,
      "learning_rate": 0.00015962854349951125,
      "loss": 0.259,
      "step": 6255
    },
    {
      "epoch": 0.9793362554790231,
      "grad_norm": 0.7878360748291016,
      "learning_rate": 0.00015960410557184747,
      "loss": 0.2945,
      "step": 6256
    },
    {
      "epoch": 0.9794927989981215,
      "grad_norm": 0.6286665201187134,
      "learning_rate": 0.00015957966764418375,
      "loss": 0.2745,
      "step": 6257
    },
    {
      "epoch": 0.9796493425172198,
      "grad_norm": 1.0530362129211426,
      "learning_rate": 0.00015955522971652003,
      "loss": 0.4147,
      "step": 6258
    },
    {
      "epoch": 0.9798058860363181,
      "grad_norm": 1.6230087280273438,
      "learning_rate": 0.00015953079178885628,
      "loss": 0.3675,
      "step": 6259
    },
    {
      "epoch": 0.9799624295554165,
      "grad_norm": 0.9370027184486389,
      "learning_rate": 0.00015950635386119256,
      "loss": 0.4494,
      "step": 6260
    },
    {
      "epoch": 0.9801189730745147,
      "grad_norm": 0.6508410573005676,
      "learning_rate": 0.00015948191593352884,
      "loss": 0.3247,
      "step": 6261
    },
    {
      "epoch": 0.980275516593613,
      "grad_norm": 0.7678089141845703,
      "learning_rate": 0.0001594574780058651,
      "loss": 0.4053,
      "step": 6262
    },
    {
      "epoch": 0.9804320601127113,
      "grad_norm": 1.5006393194198608,
      "learning_rate": 0.00015943304007820137,
      "loss": 0.4773,
      "step": 6263
    },
    {
      "epoch": 0.9805886036318097,
      "grad_norm": 1.076756238937378,
      "learning_rate": 0.00015940860215053765,
      "loss": 0.4695,
      "step": 6264
    },
    {
      "epoch": 0.980745147150908,
      "grad_norm": 1.24697744846344,
      "learning_rate": 0.00015938416422287387,
      "loss": 0.4956,
      "step": 6265
    },
    {
      "epoch": 0.9809016906700062,
      "grad_norm": 1.1420774459838867,
      "learning_rate": 0.00015935972629521015,
      "loss": 0.4614,
      "step": 6266
    },
    {
      "epoch": 0.9810582341891045,
      "grad_norm": 1.9122599363327026,
      "learning_rate": 0.00015933528836754643,
      "loss": 0.512,
      "step": 6267
    },
    {
      "epoch": 0.9812147777082029,
      "grad_norm": 1.2293574810028076,
      "learning_rate": 0.00015931085043988268,
      "loss": 0.4518,
      "step": 6268
    },
    {
      "epoch": 0.9813713212273012,
      "grad_norm": 2.1564345359802246,
      "learning_rate": 0.00015928641251221896,
      "loss": 0.5496,
      "step": 6269
    },
    {
      "epoch": 0.9815278647463995,
      "grad_norm": 1.1697052717208862,
      "learning_rate": 0.00015926197458455523,
      "loss": 0.6363,
      "step": 6270
    },
    {
      "epoch": 0.9816844082654979,
      "grad_norm": 1.1160368919372559,
      "learning_rate": 0.00015923753665689149,
      "loss": 0.3766,
      "step": 6271
    },
    {
      "epoch": 0.9818409517845961,
      "grad_norm": 1.3110209703445435,
      "learning_rate": 0.00015921309872922774,
      "loss": 0.7845,
      "step": 6272
    },
    {
      "epoch": 0.9819974953036944,
      "grad_norm": 1.1179189682006836,
      "learning_rate": 0.00015918866080156402,
      "loss": 0.3716,
      "step": 6273
    },
    {
      "epoch": 0.9821540388227927,
      "grad_norm": 0.8737648129463196,
      "learning_rate": 0.00015916422287390027,
      "loss": 0.4372,
      "step": 6274
    },
    {
      "epoch": 0.9823105823418911,
      "grad_norm": 1.6135969161987305,
      "learning_rate": 0.00015913978494623654,
      "loss": 0.7372,
      "step": 6275
    },
    {
      "epoch": 0.9824671258609894,
      "grad_norm": 2.5550239086151123,
      "learning_rate": 0.00015911534701857282,
      "loss": 0.713,
      "step": 6276
    },
    {
      "epoch": 0.9826236693800877,
      "grad_norm": 2.2706096172332764,
      "learning_rate": 0.00015909090909090907,
      "loss": 0.967,
      "step": 6277
    },
    {
      "epoch": 0.9827802128991859,
      "grad_norm": 2.025991678237915,
      "learning_rate": 0.00015906647116324535,
      "loss": 0.706,
      "step": 6278
    },
    {
      "epoch": 0.9829367564182843,
      "grad_norm": 1.7579374313354492,
      "learning_rate": 0.00015904203323558163,
      "loss": 0.5235,
      "step": 6279
    },
    {
      "epoch": 0.9830932999373826,
      "grad_norm": 1.4392457008361816,
      "learning_rate": 0.00015901759530791786,
      "loss": 0.608,
      "step": 6280
    },
    {
      "epoch": 0.9832498434564809,
      "grad_norm": 1.8196216821670532,
      "learning_rate": 0.00015899315738025413,
      "loss": 1.2177,
      "step": 6281
    },
    {
      "epoch": 0.9834063869755792,
      "grad_norm": 4.727586269378662,
      "learning_rate": 0.0001589687194525904,
      "loss": 0.6056,
      "step": 6282
    },
    {
      "epoch": 0.9835629304946775,
      "grad_norm": 1.8476710319519043,
      "learning_rate": 0.00015894428152492666,
      "loss": 0.7133,
      "step": 6283
    },
    {
      "epoch": 0.9837194740137758,
      "grad_norm": 2.2918593883514404,
      "learning_rate": 0.00015891984359726294,
      "loss": 1.0066,
      "step": 6284
    },
    {
      "epoch": 0.9838760175328741,
      "grad_norm": 1.7023167610168457,
      "learning_rate": 0.00015889540566959922,
      "loss": 0.7832,
      "step": 6285
    },
    {
      "epoch": 0.9840325610519725,
      "grad_norm": 1.6587579250335693,
      "learning_rate": 0.00015887096774193547,
      "loss": 0.8247,
      "step": 6286
    },
    {
      "epoch": 0.9841891045710708,
      "grad_norm": 2.7378196716308594,
      "learning_rate": 0.00015884652981427175,
      "loss": 1.1478,
      "step": 6287
    },
    {
      "epoch": 0.9843456480901691,
      "grad_norm": 5.176706314086914,
      "learning_rate": 0.00015882209188660803,
      "loss": 0.6136,
      "step": 6288
    },
    {
      "epoch": 0.9845021916092673,
      "grad_norm": 2.3430936336517334,
      "learning_rate": 0.00015879765395894425,
      "loss": 0.8779,
      "step": 6289
    },
    {
      "epoch": 0.9846587351283657,
      "grad_norm": 2.579249382019043,
      "learning_rate": 0.00015877321603128053,
      "loss": 1.3216,
      "step": 6290
    },
    {
      "epoch": 0.984815278647464,
      "grad_norm": 2.4976139068603516,
      "learning_rate": 0.0001587487781036168,
      "loss": 1.1508,
      "step": 6291
    },
    {
      "epoch": 0.9849718221665623,
      "grad_norm": 1.357665777206421,
      "learning_rate": 0.00015872434017595306,
      "loss": 0.8853,
      "step": 6292
    },
    {
      "epoch": 0.9851283656856606,
      "grad_norm": 2.1828815937042236,
      "learning_rate": 0.00015869990224828934,
      "loss": 0.814,
      "step": 6293
    },
    {
      "epoch": 0.985284909204759,
      "grad_norm": 3.8649942874908447,
      "learning_rate": 0.00015867546432062562,
      "loss": 1.0242,
      "step": 6294
    },
    {
      "epoch": 0.9854414527238572,
      "grad_norm": 2.397716760635376,
      "learning_rate": 0.00015865102639296187,
      "loss": 1.4216,
      "step": 6295
    },
    {
      "epoch": 0.9855979962429555,
      "grad_norm": 2.9856514930725098,
      "learning_rate": 0.00015862658846529812,
      "loss": 0.9754,
      "step": 6296
    },
    {
      "epoch": 0.9857545397620538,
      "grad_norm": 4.637380123138428,
      "learning_rate": 0.0001586021505376344,
      "loss": 1.3687,
      "step": 6297
    },
    {
      "epoch": 0.9859110832811522,
      "grad_norm": 3.106740713119507,
      "learning_rate": 0.00015857771260997065,
      "loss": 1.0291,
      "step": 6298
    },
    {
      "epoch": 0.9860676268002505,
      "grad_norm": 2.850269317626953,
      "learning_rate": 0.00015855327468230693,
      "loss": 0.6318,
      "step": 6299
    },
    {
      "epoch": 0.9862241703193487,
      "grad_norm": 1.4052451848983765,
      "learning_rate": 0.0001585288367546432,
      "loss": 0.6223,
      "step": 6300
    },
    {
      "epoch": 0.986380713838447,
      "grad_norm": 0.5287259817123413,
      "learning_rate": 0.00015850439882697946,
      "loss": 0.3596,
      "step": 6301
    },
    {
      "epoch": 0.9865372573575454,
      "grad_norm": 0.48303574323654175,
      "learning_rate": 0.00015847996089931574,
      "loss": 0.2857,
      "step": 6302
    },
    {
      "epoch": 0.9866938008766437,
      "grad_norm": 0.5090639591217041,
      "learning_rate": 0.000158455522971652,
      "loss": 0.3439,
      "step": 6303
    },
    {
      "epoch": 0.986850344395742,
      "grad_norm": 0.6524388790130615,
      "learning_rate": 0.00015843108504398824,
      "loss": 0.2502,
      "step": 6304
    },
    {
      "epoch": 0.9870068879148404,
      "grad_norm": 0.5750242471694946,
      "learning_rate": 0.00015840664711632452,
      "loss": 0.2383,
      "step": 6305
    },
    {
      "epoch": 0.9871634314339386,
      "grad_norm": 1.1469411849975586,
      "learning_rate": 0.0001583822091886608,
      "loss": 0.3987,
      "step": 6306
    },
    {
      "epoch": 0.9873199749530369,
      "grad_norm": 0.6706929206848145,
      "learning_rate": 0.00015835777126099705,
      "loss": 0.4182,
      "step": 6307
    },
    {
      "epoch": 0.9874765184721352,
      "grad_norm": 0.9753531813621521,
      "learning_rate": 0.00015833333333333332,
      "loss": 0.3501,
      "step": 6308
    },
    {
      "epoch": 0.9876330619912336,
      "grad_norm": 0.6613764762878418,
      "learning_rate": 0.0001583088954056696,
      "loss": 0.3439,
      "step": 6309
    },
    {
      "epoch": 0.9877896055103319,
      "grad_norm": 0.5925769209861755,
      "learning_rate": 0.00015828445747800585,
      "loss": 0.3267,
      "step": 6310
    },
    {
      "epoch": 0.9879461490294302,
      "grad_norm": 1.1225662231445312,
      "learning_rate": 0.00015826001955034213,
      "loss": 0.2879,
      "step": 6311
    },
    {
      "epoch": 0.9881026925485284,
      "grad_norm": 1.3221330642700195,
      "learning_rate": 0.0001582355816226784,
      "loss": 0.517,
      "step": 6312
    },
    {
      "epoch": 0.9882592360676268,
      "grad_norm": 0.802402675151825,
      "learning_rate": 0.00015821114369501463,
      "loss": 0.4133,
      "step": 6313
    },
    {
      "epoch": 0.9884157795867251,
      "grad_norm": 1.222743272781372,
      "learning_rate": 0.0001581867057673509,
      "loss": 0.54,
      "step": 6314
    },
    {
      "epoch": 0.9885723231058234,
      "grad_norm": 0.8106583952903748,
      "learning_rate": 0.0001581622678396872,
      "loss": 0.4293,
      "step": 6315
    },
    {
      "epoch": 0.9887288666249218,
      "grad_norm": 1.7479103803634644,
      "learning_rate": 0.00015813782991202344,
      "loss": 0.5579,
      "step": 6316
    },
    {
      "epoch": 0.98888541014402,
      "grad_norm": 0.8788287043571472,
      "learning_rate": 0.00015811339198435972,
      "loss": 0.2549,
      "step": 6317
    },
    {
      "epoch": 0.9890419536631183,
      "grad_norm": 4.599166393280029,
      "learning_rate": 0.000158088954056696,
      "loss": 1.0898,
      "step": 6318
    },
    {
      "epoch": 0.9891984971822166,
      "grad_norm": 1.6982215642929077,
      "learning_rate": 0.00015806451612903225,
      "loss": 0.5713,
      "step": 6319
    },
    {
      "epoch": 0.989355040701315,
      "grad_norm": 1.5406105518341064,
      "learning_rate": 0.0001580400782013685,
      "loss": 0.3639,
      "step": 6320
    },
    {
      "epoch": 0.9895115842204133,
      "grad_norm": 0.9149010181427002,
      "learning_rate": 0.00015801564027370478,
      "loss": 0.4535,
      "step": 6321
    },
    {
      "epoch": 0.9896681277395116,
      "grad_norm": 2.5321741104125977,
      "learning_rate": 0.00015799120234604103,
      "loss": 1.0454,
      "step": 6322
    },
    {
      "epoch": 0.9898246712586098,
      "grad_norm": 1.6757642030715942,
      "learning_rate": 0.0001579667644183773,
      "loss": 0.6215,
      "step": 6323
    },
    {
      "epoch": 0.9899812147777082,
      "grad_norm": 4.210154056549072,
      "learning_rate": 0.0001579423264907136,
      "loss": 0.5722,
      "step": 6324
    },
    {
      "epoch": 0.9901377582968065,
      "grad_norm": 2.43436598777771,
      "learning_rate": 0.00015791788856304984,
      "loss": 0.558,
      "step": 6325
    },
    {
      "epoch": 0.9902943018159048,
      "grad_norm": 1.8142644166946411,
      "learning_rate": 0.00015789345063538612,
      "loss": 0.4534,
      "step": 6326
    },
    {
      "epoch": 0.9904508453350032,
      "grad_norm": 1.3764431476593018,
      "learning_rate": 0.0001578690127077224,
      "loss": 0.4666,
      "step": 6327
    },
    {
      "epoch": 0.9906073888541015,
      "grad_norm": 1.3100895881652832,
      "learning_rate": 0.00015784457478005862,
      "loss": 0.8198,
      "step": 6328
    },
    {
      "epoch": 0.9907639323731997,
      "grad_norm": 1.911555528640747,
      "learning_rate": 0.0001578201368523949,
      "loss": 0.7068,
      "step": 6329
    },
    {
      "epoch": 0.990920475892298,
      "grad_norm": 2.208777666091919,
      "learning_rate": 0.00015779569892473118,
      "loss": 0.8239,
      "step": 6330
    },
    {
      "epoch": 0.9910770194113964,
      "grad_norm": 2.607349157333374,
      "learning_rate": 0.00015777126099706743,
      "loss": 0.3357,
      "step": 6331
    },
    {
      "epoch": 0.9912335629304947,
      "grad_norm": 2.01904559135437,
      "learning_rate": 0.0001577468230694037,
      "loss": 0.6867,
      "step": 6332
    },
    {
      "epoch": 0.991390106449593,
      "grad_norm": 3.4520325660705566,
      "learning_rate": 0.00015772238514173998,
      "loss": 1.0985,
      "step": 6333
    },
    {
      "epoch": 0.9915466499686914,
      "grad_norm": 1.0999062061309814,
      "learning_rate": 0.00015769794721407624,
      "loss": 0.4234,
      "step": 6334
    },
    {
      "epoch": 0.9917031934877896,
      "grad_norm": 6.374632358551025,
      "learning_rate": 0.00015767350928641251,
      "loss": 0.865,
      "step": 6335
    },
    {
      "epoch": 0.9918597370068879,
      "grad_norm": 2.246000051498413,
      "learning_rate": 0.0001576490713587488,
      "loss": 0.8057,
      "step": 6336
    },
    {
      "epoch": 0.9920162805259862,
      "grad_norm": 2.388260841369629,
      "learning_rate": 0.00015762463343108502,
      "loss": 0.8103,
      "step": 6337
    },
    {
      "epoch": 0.9921728240450846,
      "grad_norm": 2.392380952835083,
      "learning_rate": 0.0001576001955034213,
      "loss": 1.3118,
      "step": 6338
    },
    {
      "epoch": 0.9923293675641829,
      "grad_norm": 2.8762900829315186,
      "learning_rate": 0.00015757575757575757,
      "loss": 1.4997,
      "step": 6339
    },
    {
      "epoch": 0.9924859110832811,
      "grad_norm": 2.6236648559570312,
      "learning_rate": 0.00015755131964809382,
      "loss": 1.0189,
      "step": 6340
    },
    {
      "epoch": 0.9926424546023794,
      "grad_norm": 2.0896835327148438,
      "learning_rate": 0.0001575268817204301,
      "loss": 1.0106,
      "step": 6341
    },
    {
      "epoch": 0.9927989981214778,
      "grad_norm": 1.5402086973190308,
      "learning_rate": 0.00015750244379276638,
      "loss": 0.7675,
      "step": 6342
    },
    {
      "epoch": 0.9929555416405761,
      "grad_norm": 5.9841179847717285,
      "learning_rate": 0.0001574780058651026,
      "loss": 1.0019,
      "step": 6343
    },
    {
      "epoch": 0.9931120851596744,
      "grad_norm": 2.565718412399292,
      "learning_rate": 0.00015745356793743888,
      "loss": 1.1758,
      "step": 6344
    },
    {
      "epoch": 0.9932686286787727,
      "grad_norm": 2.391500473022461,
      "learning_rate": 0.00015742913000977516,
      "loss": 1.1052,
      "step": 6345
    },
    {
      "epoch": 0.993425172197871,
      "grad_norm": 1.9955416917800903,
      "learning_rate": 0.0001574046920821114,
      "loss": 0.9472,
      "step": 6346
    },
    {
      "epoch": 0.9935817157169693,
      "grad_norm": 2.492854356765747,
      "learning_rate": 0.0001573802541544477,
      "loss": 0.6311,
      "step": 6347
    },
    {
      "epoch": 0.9937382592360676,
      "grad_norm": 1.4435681104660034,
      "learning_rate": 0.00015735581622678397,
      "loss": 0.3773,
      "step": 6348
    },
    {
      "epoch": 0.993894802755166,
      "grad_norm": 3.8596134185791016,
      "learning_rate": 0.00015733137829912022,
      "loss": 0.9125,
      "step": 6349
    },
    {
      "epoch": 0.9940513462742643,
      "grad_norm": 2.726219415664673,
      "learning_rate": 0.0001573069403714565,
      "loss": 1.5958,
      "step": 6350
    },
    {
      "epoch": 0.9942078897933626,
      "grad_norm": 0.686896562576294,
      "learning_rate": 0.00015728250244379278,
      "loss": 0.3087,
      "step": 6351
    },
    {
      "epoch": 0.9943644333124608,
      "grad_norm": 0.6529414057731628,
      "learning_rate": 0.000157258064516129,
      "loss": 0.3721,
      "step": 6352
    },
    {
      "epoch": 0.9945209768315592,
      "grad_norm": 0.6401044726371765,
      "learning_rate": 0.00015723362658846528,
      "loss": 0.3123,
      "step": 6353
    },
    {
      "epoch": 0.9946775203506575,
      "grad_norm": 0.6423803567886353,
      "learning_rate": 0.00015720918866080156,
      "loss": 0.2896,
      "step": 6354
    },
    {
      "epoch": 0.9948340638697558,
      "grad_norm": 0.576648473739624,
      "learning_rate": 0.0001571847507331378,
      "loss": 0.2859,
      "step": 6355
    },
    {
      "epoch": 0.9949906073888541,
      "grad_norm": 0.7672856450080872,
      "learning_rate": 0.0001571603128054741,
      "loss": 0.3422,
      "step": 6356
    },
    {
      "epoch": 0.9951471509079524,
      "grad_norm": 1.0555219650268555,
      "learning_rate": 0.00015713587487781037,
      "loss": 0.294,
      "step": 6357
    },
    {
      "epoch": 0.9953036944270507,
      "grad_norm": 0.8917555212974548,
      "learning_rate": 0.00015711143695014662,
      "loss": 0.3567,
      "step": 6358
    },
    {
      "epoch": 0.995460237946149,
      "grad_norm": 0.863348126411438,
      "learning_rate": 0.0001570869990224829,
      "loss": 0.3465,
      "step": 6359
    },
    {
      "epoch": 0.9956167814652473,
      "grad_norm": 1.0300931930541992,
      "learning_rate": 0.00015706256109481917,
      "loss": 0.4107,
      "step": 6360
    },
    {
      "epoch": 0.9957733249843457,
      "grad_norm": 2.140244483947754,
      "learning_rate": 0.0001570381231671554,
      "loss": 0.5715,
      "step": 6361
    },
    {
      "epoch": 0.995929868503444,
      "grad_norm": 1.4137368202209473,
      "learning_rate": 0.00015701368523949168,
      "loss": 0.6483,
      "step": 6362
    },
    {
      "epoch": 0.9960864120225422,
      "grad_norm": 1.0269410610198975,
      "learning_rate": 0.00015698924731182796,
      "loss": 0.4882,
      "step": 6363
    },
    {
      "epoch": 0.9962429555416406,
      "grad_norm": 2.0102436542510986,
      "learning_rate": 0.0001569648093841642,
      "loss": 0.4823,
      "step": 6364
    },
    {
      "epoch": 0.9963994990607389,
      "grad_norm": 2.118197441101074,
      "learning_rate": 0.00015694037145650049,
      "loss": 0.5051,
      "step": 6365
    },
    {
      "epoch": 0.9965560425798372,
      "grad_norm": 1.0090569257736206,
      "learning_rate": 0.00015691593352883676,
      "loss": 0.2953,
      "step": 6366
    },
    {
      "epoch": 0.9967125860989355,
      "grad_norm": 2.38740873336792,
      "learning_rate": 0.000156891495601173,
      "loss": 0.6598,
      "step": 6367
    },
    {
      "epoch": 0.9968691296180339,
      "grad_norm": 1.2782179117202759,
      "learning_rate": 0.00015686705767350927,
      "loss": 0.5072,
      "step": 6368
    },
    {
      "epoch": 0.9970256731371321,
      "grad_norm": 1.9059410095214844,
      "learning_rate": 0.00015684261974584554,
      "loss": 0.5702,
      "step": 6369
    },
    {
      "epoch": 0.9971822166562304,
      "grad_norm": 2.062696933746338,
      "learning_rate": 0.0001568181818181818,
      "loss": 0.6943,
      "step": 6370
    },
    {
      "epoch": 0.9973387601753287,
      "grad_norm": 1.2728642225265503,
      "learning_rate": 0.00015679374389051807,
      "loss": 0.5178,
      "step": 6371
    },
    {
      "epoch": 0.9974953036944271,
      "grad_norm": 2.6865758895874023,
      "learning_rate": 0.00015676930596285435,
      "loss": 1.1807,
      "step": 6372
    },
    {
      "epoch": 0.9976518472135254,
      "grad_norm": 2.504348039627075,
      "learning_rate": 0.0001567448680351906,
      "loss": 1.0252,
      "step": 6373
    },
    {
      "epoch": 0.9978083907326236,
      "grad_norm": 3.697204351425171,
      "learning_rate": 0.00015672043010752688,
      "loss": 1.1543,
      "step": 6374
    },
    {
      "epoch": 0.997964934251722,
      "grad_norm": 1.612819790840149,
      "learning_rate": 0.00015669599217986316,
      "loss": 0.8737,
      "step": 6375
    },
    {
      "epoch": 0.9981214777708203,
      "grad_norm": 3.1941633224487305,
      "learning_rate": 0.00015667155425219938,
      "loss": 0.8725,
      "step": 6376
    },
    {
      "epoch": 0.9982780212899186,
      "grad_norm": 2.2976832389831543,
      "learning_rate": 0.00015664711632453566,
      "loss": 1.1326,
      "step": 6377
    },
    {
      "epoch": 0.9984345648090169,
      "grad_norm": 3.410414457321167,
      "learning_rate": 0.00015662267839687194,
      "loss": 0.5624,
      "step": 6378
    },
    {
      "epoch": 0.9985911083281153,
      "grad_norm": 3.318864107131958,
      "learning_rate": 0.0001565982404692082,
      "loss": 0.9388,
      "step": 6379
    },
    {
      "epoch": 0.9987476518472135,
      "grad_norm": 1.7484426498413086,
      "learning_rate": 0.00015657380254154447,
      "loss": 0.9887,
      "step": 6380
    },
    {
      "epoch": 0.9989041953663118,
      "grad_norm": 3.5230112075805664,
      "learning_rate": 0.00015654936461388075,
      "loss": 1.4972,
      "step": 6381
    },
    {
      "epoch": 0.9990607388854101,
      "grad_norm": 5.785534381866455,
      "learning_rate": 0.000156524926686217,
      "loss": 1.1754,
      "step": 6382
    },
    {
      "epoch": 0.9992172824045085,
      "grad_norm": 2.6626009941101074,
      "learning_rate": 0.00015650048875855328,
      "loss": 1.2539,
      "step": 6383
    },
    {
      "epoch": 0.9993738259236068,
      "grad_norm": 1.2295653820037842,
      "learning_rate": 0.00015647605083088953,
      "loss": 1.0544,
      "step": 6384
    },
    {
      "epoch": 0.9995303694427051,
      "grad_norm": 1.510327935218811,
      "learning_rate": 0.00015645161290322578,
      "loss": 0.5378,
      "step": 6385
    },
    {
      "epoch": 0.9996869129618033,
      "grad_norm": 1.7747565507888794,
      "learning_rate": 0.00015642717497556206,
      "loss": 0.8904,
      "step": 6386
    },
    {
      "epoch": 0.9998434564809017,
      "grad_norm": 6.381943225860596,
      "learning_rate": 0.00015640273704789834,
      "loss": 1.7167,
      "step": 6387
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.6135759353637695,
      "learning_rate": 0.0001563782991202346,
      "loss": 1.0733,
      "step": 6388
    },
    {
      "epoch": 1.0001565435190982,
      "grad_norm": 0.6401867866516113,
      "learning_rate": 0.00015635386119257087,
      "loss": 0.3242,
      "step": 6389
    },
    {
      "epoch": 1.0003130870381967,
      "grad_norm": 0.46278223395347595,
      "learning_rate": 0.00015632942326490715,
      "loss": 0.2188,
      "step": 6390
    },
    {
      "epoch": 1.0004696305572949,
      "grad_norm": 0.5220637917518616,
      "learning_rate": 0.00015630498533724337,
      "loss": 0.2602,
      "step": 6391
    },
    {
      "epoch": 1.0006261740763933,
      "grad_norm": 0.6111396551132202,
      "learning_rate": 0.00015628054740957965,
      "loss": 0.3363,
      "step": 6392
    },
    {
      "epoch": 1.0007827175954915,
      "grad_norm": 0.6467208862304688,
      "learning_rate": 0.00015625610948191593,
      "loss": 0.2684,
      "step": 6393
    },
    {
      "epoch": 1.0009392611145898,
      "grad_norm": 0.5766490697860718,
      "learning_rate": 0.00015623167155425218,
      "loss": 0.2703,
      "step": 6394
    },
    {
      "epoch": 1.0010958046336882,
      "grad_norm": 0.6601141095161438,
      "learning_rate": 0.00015620723362658846,
      "loss": 0.3272,
      "step": 6395
    },
    {
      "epoch": 1.0012523481527864,
      "grad_norm": 0.61875981092453,
      "learning_rate": 0.0001561827956989247,
      "loss": 0.2942,
      "step": 6396
    },
    {
      "epoch": 1.0014088916718848,
      "grad_norm": 0.9664865732192993,
      "learning_rate": 0.00015615835777126099,
      "loss": 0.3894,
      "step": 6397
    },
    {
      "epoch": 1.001565435190983,
      "grad_norm": 0.5463569760322571,
      "learning_rate": 0.00015613391984359726,
      "loss": 0.1922,
      "step": 6398
    },
    {
      "epoch": 1.0017219787100815,
      "grad_norm": 1.0034434795379639,
      "learning_rate": 0.0001561094819159335,
      "loss": 0.3852,
      "step": 6399
    },
    {
      "epoch": 1.0018785222291797,
      "grad_norm": 0.9108190536499023,
      "learning_rate": 0.00015608504398826977,
      "loss": 0.3026,
      "step": 6400
    },
    {
      "epoch": 1.002035065748278,
      "grad_norm": 1.1288667917251587,
      "learning_rate": 0.00015606060606060605,
      "loss": 0.4889,
      "step": 6401
    },
    {
      "epoch": 1.0021916092673764,
      "grad_norm": 0.8430051207542419,
      "learning_rate": 0.0001560361681329423,
      "loss": 0.5495,
      "step": 6402
    },
    {
      "epoch": 1.0023481527864746,
      "grad_norm": 1.0893449783325195,
      "learning_rate": 0.00015601173020527857,
      "loss": 0.3109,
      "step": 6403
    },
    {
      "epoch": 1.002504696305573,
      "grad_norm": 1.9297972917556763,
      "learning_rate": 0.00015598729227761485,
      "loss": 0.5138,
      "step": 6404
    },
    {
      "epoch": 1.0026612398246713,
      "grad_norm": 1.0670981407165527,
      "learning_rate": 0.0001559628543499511,
      "loss": 0.3637,
      "step": 6405
    },
    {
      "epoch": 1.0028177833437695,
      "grad_norm": 1.307648777961731,
      "learning_rate": 0.00015593841642228738,
      "loss": 0.4379,
      "step": 6406
    },
    {
      "epoch": 1.002974326862868,
      "grad_norm": 1.2096155881881714,
      "learning_rate": 0.00015591397849462366,
      "loss": 0.4889,
      "step": 6407
    },
    {
      "epoch": 1.0031308703819661,
      "grad_norm": 1.7205982208251953,
      "learning_rate": 0.00015588954056695989,
      "loss": 0.2895,
      "step": 6408
    },
    {
      "epoch": 1.0032874139010646,
      "grad_norm": 1.0032813549041748,
      "learning_rate": 0.00015586510263929616,
      "loss": 0.4063,
      "step": 6409
    },
    {
      "epoch": 1.0034439574201628,
      "grad_norm": 2.1352264881134033,
      "learning_rate": 0.00015584066471163244,
      "loss": 0.4581,
      "step": 6410
    },
    {
      "epoch": 1.003600500939261,
      "grad_norm": 1.522994041442871,
      "learning_rate": 0.0001558162267839687,
      "loss": 0.4092,
      "step": 6411
    },
    {
      "epoch": 1.0037570444583594,
      "grad_norm": 1.0274169445037842,
      "learning_rate": 0.00015579178885630497,
      "loss": 0.3602,
      "step": 6412
    },
    {
      "epoch": 1.0039135879774577,
      "grad_norm": 1.5904802083969116,
      "learning_rate": 0.00015576735092864125,
      "loss": 0.7179,
      "step": 6413
    },
    {
      "epoch": 1.004070131496556,
      "grad_norm": 1.4016789197921753,
      "learning_rate": 0.00015574291300097747,
      "loss": 0.5017,
      "step": 6414
    },
    {
      "epoch": 1.0042266750156543,
      "grad_norm": 5.12330436706543,
      "learning_rate": 0.00015571847507331375,
      "loss": 0.8605,
      "step": 6415
    },
    {
      "epoch": 1.0043832185347528,
      "grad_norm": 2.253363609313965,
      "learning_rate": 0.00015569403714565003,
      "loss": 0.49,
      "step": 6416
    },
    {
      "epoch": 1.004539762053851,
      "grad_norm": 8.99915599822998,
      "learning_rate": 0.00015566959921798628,
      "loss": 0.6846,
      "step": 6417
    },
    {
      "epoch": 1.0046963055729492,
      "grad_norm": 2.108546018600464,
      "learning_rate": 0.00015564516129032256,
      "loss": 0.7329,
      "step": 6418
    },
    {
      "epoch": 1.0048528490920476,
      "grad_norm": 2.5911080837249756,
      "learning_rate": 0.00015562072336265884,
      "loss": 0.7537,
      "step": 6419
    },
    {
      "epoch": 1.0050093926111459,
      "grad_norm": 9.908422470092773,
      "learning_rate": 0.0001555962854349951,
      "loss": 1.2488,
      "step": 6420
    },
    {
      "epoch": 1.0051659361302443,
      "grad_norm": 2.6049554347991943,
      "learning_rate": 0.00015557184750733137,
      "loss": 0.872,
      "step": 6421
    },
    {
      "epoch": 1.0053224796493425,
      "grad_norm": 4.485533237457275,
      "learning_rate": 0.00015554740957966765,
      "loss": 1.3471,
      "step": 6422
    },
    {
      "epoch": 1.0054790231684407,
      "grad_norm": 3.122943162918091,
      "learning_rate": 0.00015552297165200387,
      "loss": 0.7273,
      "step": 6423
    },
    {
      "epoch": 1.0056355666875392,
      "grad_norm": 2.615659475326538,
      "learning_rate": 0.00015549853372434015,
      "loss": 1.014,
      "step": 6424
    },
    {
      "epoch": 1.0057921102066374,
      "grad_norm": 1.3930341005325317,
      "learning_rate": 0.00015547409579667643,
      "loss": 0.3932,
      "step": 6425
    },
    {
      "epoch": 1.0059486537257358,
      "grad_norm": 2.3153088092803955,
      "learning_rate": 0.00015544965786901268,
      "loss": 1.1018,
      "step": 6426
    },
    {
      "epoch": 1.006105197244834,
      "grad_norm": 4.314839839935303,
      "learning_rate": 0.00015542521994134896,
      "loss": 1.1748,
      "step": 6427
    },
    {
      "epoch": 1.0062617407639323,
      "grad_norm": 3.0916011333465576,
      "learning_rate": 0.00015540078201368524,
      "loss": 1.2124,
      "step": 6428
    },
    {
      "epoch": 1.0064182842830307,
      "grad_norm": 4.510741233825684,
      "learning_rate": 0.0001553763440860215,
      "loss": 1.0817,
      "step": 6429
    },
    {
      "epoch": 1.006574827802129,
      "grad_norm": 5.305693626403809,
      "learning_rate": 0.00015535190615835777,
      "loss": 1.6207,
      "step": 6430
    },
    {
      "epoch": 1.0067313713212274,
      "grad_norm": 2.222895383834839,
      "learning_rate": 0.00015532746823069402,
      "loss": 1.512,
      "step": 6431
    },
    {
      "epoch": 1.0068879148403256,
      "grad_norm": 3.7936456203460693,
      "learning_rate": 0.00015530303030303027,
      "loss": 1.712,
      "step": 6432
    },
    {
      "epoch": 1.007044458359424,
      "grad_norm": 1.699407696723938,
      "learning_rate": 0.00015527859237536655,
      "loss": 1.0667,
      "step": 6433
    },
    {
      "epoch": 1.0072010018785222,
      "grad_norm": 2.3237552642822266,
      "learning_rate": 0.00015525415444770282,
      "loss": 1.1835,
      "step": 6434
    },
    {
      "epoch": 1.0073575453976205,
      "grad_norm": 0.950505793094635,
      "learning_rate": 0.00015522971652003908,
      "loss": 0.3155,
      "step": 6435
    },
    {
      "epoch": 1.007514088916719,
      "grad_norm": 3.0012030601501465,
      "learning_rate": 0.00015520527859237535,
      "loss": 0.5834,
      "step": 6436
    },
    {
      "epoch": 1.0076706324358171,
      "grad_norm": 1.9200234413146973,
      "learning_rate": 0.00015518084066471163,
      "loss": 0.8096,
      "step": 6437
    },
    {
      "epoch": 1.0078271759549156,
      "grad_norm": 3.209449291229248,
      "learning_rate": 0.00015515640273704786,
      "loss": 1.6,
      "step": 6438
    },
    {
      "epoch": 1.0079837194740138,
      "grad_norm": 0.4958963990211487,
      "learning_rate": 0.00015513196480938413,
      "loss": 0.2785,
      "step": 6439
    },
    {
      "epoch": 1.008140262993112,
      "grad_norm": 0.461821585893631,
      "learning_rate": 0.0001551075268817204,
      "loss": 0.2305,
      "step": 6440
    },
    {
      "epoch": 1.0082968065122104,
      "grad_norm": 0.5686135292053223,
      "learning_rate": 0.00015508308895405666,
      "loss": 0.1765,
      "step": 6441
    },
    {
      "epoch": 1.0084533500313086,
      "grad_norm": 0.5753898024559021,
      "learning_rate": 0.00015505865102639294,
      "loss": 0.1702,
      "step": 6442
    },
    {
      "epoch": 1.008609893550407,
      "grad_norm": 0.2971111536026001,
      "learning_rate": 0.00015503421309872922,
      "loss": 0.1378,
      "step": 6443
    },
    {
      "epoch": 1.0087664370695053,
      "grad_norm": 0.6010518074035645,
      "learning_rate": 0.00015500977517106547,
      "loss": 0.2316,
      "step": 6444
    },
    {
      "epoch": 1.0089229805886035,
      "grad_norm": 0.7052372097969055,
      "learning_rate": 0.00015498533724340175,
      "loss": 0.1997,
      "step": 6445
    },
    {
      "epoch": 1.009079524107702,
      "grad_norm": 0.728542685508728,
      "learning_rate": 0.00015496089931573803,
      "loss": 0.2774,
      "step": 6446
    },
    {
      "epoch": 1.0092360676268002,
      "grad_norm": 0.6369313597679138,
      "learning_rate": 0.00015493646138807425,
      "loss": 0.2283,
      "step": 6447
    },
    {
      "epoch": 1.0093926111458986,
      "grad_norm": 0.7698948383331299,
      "learning_rate": 0.00015491202346041053,
      "loss": 0.336,
      "step": 6448
    },
    {
      "epoch": 1.0095491546649968,
      "grad_norm": 0.579069197177887,
      "learning_rate": 0.0001548875855327468,
      "loss": 0.275,
      "step": 6449
    },
    {
      "epoch": 1.0097056981840953,
      "grad_norm": 0.8608638644218445,
      "learning_rate": 0.00015486314760508306,
      "loss": 0.2888,
      "step": 6450
    },
    {
      "epoch": 1.0098622417031935,
      "grad_norm": 0.651512622833252,
      "learning_rate": 0.00015483870967741934,
      "loss": 0.2552,
      "step": 6451
    },
    {
      "epoch": 1.0100187852222917,
      "grad_norm": 1.6202661991119385,
      "learning_rate": 0.00015481427174975562,
      "loss": 0.5846,
      "step": 6452
    },
    {
      "epoch": 1.0101753287413902,
      "grad_norm": 1.3742421865463257,
      "learning_rate": 0.00015478983382209187,
      "loss": 0.29,
      "step": 6453
    },
    {
      "epoch": 1.0103318722604884,
      "grad_norm": 1.502272605895996,
      "learning_rate": 0.00015476539589442815,
      "loss": 0.4051,
      "step": 6454
    },
    {
      "epoch": 1.0104884157795868,
      "grad_norm": 0.8512781262397766,
      "learning_rate": 0.0001547409579667644,
      "loss": 0.4372,
      "step": 6455
    },
    {
      "epoch": 1.010644959298685,
      "grad_norm": 1.3594355583190918,
      "learning_rate": 0.00015471652003910065,
      "loss": 0.4414,
      "step": 6456
    },
    {
      "epoch": 1.0108015028177832,
      "grad_norm": 1.8934544324874878,
      "learning_rate": 0.00015469208211143693,
      "loss": 0.5294,
      "step": 6457
    },
    {
      "epoch": 1.0109580463368817,
      "grad_norm": 1.089806318283081,
      "learning_rate": 0.0001546676441837732,
      "loss": 0.2968,
      "step": 6458
    },
    {
      "epoch": 1.01111458985598,
      "grad_norm": 1.3797799348831177,
      "learning_rate": 0.00015464320625610946,
      "loss": 0.2608,
      "step": 6459
    },
    {
      "epoch": 1.0112711333750783,
      "grad_norm": 1.3458391427993774,
      "learning_rate": 0.00015461876832844574,
      "loss": 0.4816,
      "step": 6460
    },
    {
      "epoch": 1.0114276768941766,
      "grad_norm": 1.3416376113891602,
      "learning_rate": 0.00015459433040078201,
      "loss": 0.5628,
      "step": 6461
    },
    {
      "epoch": 1.0115842204132748,
      "grad_norm": 1.862665057182312,
      "learning_rate": 0.00015456989247311824,
      "loss": 0.4349,
      "step": 6462
    },
    {
      "epoch": 1.0117407639323732,
      "grad_norm": 1.254166603088379,
      "learning_rate": 0.00015454545454545452,
      "loss": 0.4393,
      "step": 6463
    },
    {
      "epoch": 1.0118973074514714,
      "grad_norm": 1.7121856212615967,
      "learning_rate": 0.0001545210166177908,
      "loss": 0.5046,
      "step": 6464
    },
    {
      "epoch": 1.0120538509705699,
      "grad_norm": 1.5799607038497925,
      "learning_rate": 0.00015449657869012705,
      "loss": 0.5843,
      "step": 6465
    },
    {
      "epoch": 1.012210394489668,
      "grad_norm": 2.453345775604248,
      "learning_rate": 0.00015447214076246333,
      "loss": 1.143,
      "step": 6466
    },
    {
      "epoch": 1.0123669380087665,
      "grad_norm": 2.5280091762542725,
      "learning_rate": 0.0001544477028347996,
      "loss": 0.6528,
      "step": 6467
    },
    {
      "epoch": 1.0125234815278648,
      "grad_norm": 1.8263561725616455,
      "learning_rate": 0.00015442326490713585,
      "loss": 0.8534,
      "step": 6468
    },
    {
      "epoch": 1.012680025046963,
      "grad_norm": 1.4700437784194946,
      "learning_rate": 0.00015439882697947213,
      "loss": 0.7371,
      "step": 6469
    },
    {
      "epoch": 1.0128365685660614,
      "grad_norm": 2.2002007961273193,
      "learning_rate": 0.0001543743890518084,
      "loss": 0.9834,
      "step": 6470
    },
    {
      "epoch": 1.0129931120851596,
      "grad_norm": 2.333889961242676,
      "learning_rate": 0.00015434995112414464,
      "loss": 0.7919,
      "step": 6471
    },
    {
      "epoch": 1.013149655604258,
      "grad_norm": 5.423836708068848,
      "learning_rate": 0.00015432551319648091,
      "loss": 0.8916,
      "step": 6472
    },
    {
      "epoch": 1.0133061991233563,
      "grad_norm": 1.438730001449585,
      "learning_rate": 0.0001543010752688172,
      "loss": 0.7162,
      "step": 6473
    },
    {
      "epoch": 1.0134627426424545,
      "grad_norm": 1.9986294507980347,
      "learning_rate": 0.00015427663734115344,
      "loss": 1.0373,
      "step": 6474
    },
    {
      "epoch": 1.013619286161553,
      "grad_norm": 4.634669303894043,
      "learning_rate": 0.00015425219941348972,
      "loss": 1.1194,
      "step": 6475
    },
    {
      "epoch": 1.0137758296806512,
      "grad_norm": 2.2568159103393555,
      "learning_rate": 0.000154227761485826,
      "loss": 0.9643,
      "step": 6476
    },
    {
      "epoch": 1.0139323731997496,
      "grad_norm": 1.3911081552505493,
      "learning_rate": 0.00015420332355816225,
      "loss": 0.6398,
      "step": 6477
    },
    {
      "epoch": 1.0140889167188478,
      "grad_norm": 2.0544090270996094,
      "learning_rate": 0.00015417888563049853,
      "loss": 1.4445,
      "step": 6478
    },
    {
      "epoch": 1.014245460237946,
      "grad_norm": 1.972231388092041,
      "learning_rate": 0.00015415444770283478,
      "loss": 0.7207,
      "step": 6479
    },
    {
      "epoch": 1.0144020037570445,
      "grad_norm": 2.4303290843963623,
      "learning_rate": 0.00015413000977517103,
      "loss": 1.3453,
      "step": 6480
    },
    {
      "epoch": 1.0145585472761427,
      "grad_norm": 5.570343017578125,
      "learning_rate": 0.0001541055718475073,
      "loss": 1.5838,
      "step": 6481
    },
    {
      "epoch": 1.0147150907952411,
      "grad_norm": 1.2938313484191895,
      "learning_rate": 0.0001540811339198436,
      "loss": 1.1335,
      "step": 6482
    },
    {
      "epoch": 1.0148716343143394,
      "grad_norm": 2.1745591163635254,
      "learning_rate": 0.00015405669599217984,
      "loss": 0.6899,
      "step": 6483
    },
    {
      "epoch": 1.0150281778334378,
      "grad_norm": 3.9206926822662354,
      "learning_rate": 0.00015403225806451612,
      "loss": 0.8181,
      "step": 6484
    },
    {
      "epoch": 1.015184721352536,
      "grad_norm": 3.452662706375122,
      "learning_rate": 0.0001540078201368524,
      "loss": 0.2675,
      "step": 6485
    },
    {
      "epoch": 1.0153412648716342,
      "grad_norm": 1.481506109237671,
      "learning_rate": 0.00015398338220918862,
      "loss": 0.7693,
      "step": 6486
    },
    {
      "epoch": 1.0154978083907327,
      "grad_norm": 3.136730670928955,
      "learning_rate": 0.0001539589442815249,
      "loss": 0.4043,
      "step": 6487
    },
    {
      "epoch": 1.0156543519098309,
      "grad_norm": 4.050405502319336,
      "learning_rate": 0.00015393450635386118,
      "loss": 1.0601,
      "step": 6488
    },
    {
      "epoch": 1.0158108954289293,
      "grad_norm": 0.4689655601978302,
      "learning_rate": 0.00015391006842619743,
      "loss": 0.312,
      "step": 6489
    },
    {
      "epoch": 1.0159674389480275,
      "grad_norm": 0.32348453998565674,
      "learning_rate": 0.0001538856304985337,
      "loss": 0.1755,
      "step": 6490
    },
    {
      "epoch": 1.0161239824671258,
      "grad_norm": 0.456193745136261,
      "learning_rate": 0.00015386119257086999,
      "loss": 0.2209,
      "step": 6491
    },
    {
      "epoch": 1.0162805259862242,
      "grad_norm": 0.6613138914108276,
      "learning_rate": 0.00015383675464320624,
      "loss": 0.3865,
      "step": 6492
    },
    {
      "epoch": 1.0164370695053224,
      "grad_norm": 0.45760709047317505,
      "learning_rate": 0.00015381231671554252,
      "loss": 0.2318,
      "step": 6493
    },
    {
      "epoch": 1.0165936130244209,
      "grad_norm": 0.4187608063220978,
      "learning_rate": 0.0001537878787878788,
      "loss": 0.2009,
      "step": 6494
    },
    {
      "epoch": 1.016750156543519,
      "grad_norm": 0.9235534071922302,
      "learning_rate": 0.00015376344086021502,
      "loss": 0.3943,
      "step": 6495
    },
    {
      "epoch": 1.0169067000626173,
      "grad_norm": 0.9163877367973328,
      "learning_rate": 0.0001537390029325513,
      "loss": 0.3704,
      "step": 6496
    },
    {
      "epoch": 1.0170632435817157,
      "grad_norm": 0.5193113684654236,
      "learning_rate": 0.00015371456500488757,
      "loss": 0.2252,
      "step": 6497
    },
    {
      "epoch": 1.017219787100814,
      "grad_norm": 0.8574435710906982,
      "learning_rate": 0.00015369012707722383,
      "loss": 0.3135,
      "step": 6498
    },
    {
      "epoch": 1.0173763306199124,
      "grad_norm": 1.2782654762268066,
      "learning_rate": 0.0001536656891495601,
      "loss": 0.3573,
      "step": 6499
    },
    {
      "epoch": 1.0175328741390106,
      "grad_norm": 4.555233955383301,
      "learning_rate": 0.00015364125122189638,
      "loss": 0.9878,
      "step": 6500
    },
    {
      "epoch": 1.017689417658109,
      "grad_norm": 0.6440590620040894,
      "learning_rate": 0.00015361681329423263,
      "loss": 0.212,
      "step": 6501
    },
    {
      "epoch": 1.0178459611772073,
      "grad_norm": 1.6863186359405518,
      "learning_rate": 0.00015359237536656889,
      "loss": 0.5993,
      "step": 6502
    },
    {
      "epoch": 1.0180025046963055,
      "grad_norm": 1.924282431602478,
      "learning_rate": 0.00015356793743890516,
      "loss": 0.3741,
      "step": 6503
    },
    {
      "epoch": 1.018159048215404,
      "grad_norm": 0.9348188638687134,
      "learning_rate": 0.00015354349951124141,
      "loss": 0.4833,
      "step": 6504
    },
    {
      "epoch": 1.0183155917345021,
      "grad_norm": 1.7575799226760864,
      "learning_rate": 0.0001535190615835777,
      "loss": 0.3112,
      "step": 6505
    },
    {
      "epoch": 1.0184721352536006,
      "grad_norm": 1.014967679977417,
      "learning_rate": 0.00015349462365591397,
      "loss": 0.4143,
      "step": 6506
    },
    {
      "epoch": 1.0186286787726988,
      "grad_norm": 1.0853968858718872,
      "learning_rate": 0.00015347018572825022,
      "loss": 0.4754,
      "step": 6507
    },
    {
      "epoch": 1.018785222291797,
      "grad_norm": 1.5835812091827393,
      "learning_rate": 0.0001534457478005865,
      "loss": 0.3803,
      "step": 6508
    },
    {
      "epoch": 1.0189417658108955,
      "grad_norm": 0.9677704572677612,
      "learning_rate": 0.00015342130987292278,
      "loss": 0.4094,
      "step": 6509
    },
    {
      "epoch": 1.0190983093299937,
      "grad_norm": 1.162556767463684,
      "learning_rate": 0.000153396871945259,
      "loss": 0.4622,
      "step": 6510
    },
    {
      "epoch": 1.0192548528490921,
      "grad_norm": 1.3809646368026733,
      "learning_rate": 0.00015337243401759528,
      "loss": 0.361,
      "step": 6511
    },
    {
      "epoch": 1.0194113963681903,
      "grad_norm": 1.8124253749847412,
      "learning_rate": 0.00015334799608993156,
      "loss": 0.4641,
      "step": 6512
    },
    {
      "epoch": 1.0195679398872888,
      "grad_norm": 1.3627941608428955,
      "learning_rate": 0.0001533235581622678,
      "loss": 0.3851,
      "step": 6513
    },
    {
      "epoch": 1.019724483406387,
      "grad_norm": 4.081453800201416,
      "learning_rate": 0.0001532991202346041,
      "loss": 1.1882,
      "step": 6514
    },
    {
      "epoch": 1.0198810269254852,
      "grad_norm": 1.3441665172576904,
      "learning_rate": 0.00015327468230694037,
      "loss": 0.4868,
      "step": 6515
    },
    {
      "epoch": 1.0200375704445837,
      "grad_norm": 3.412243366241455,
      "learning_rate": 0.00015325024437927662,
      "loss": 0.6887,
      "step": 6516
    },
    {
      "epoch": 1.0201941139636819,
      "grad_norm": 0.8700874447822571,
      "learning_rate": 0.0001532258064516129,
      "loss": 0.3311,
      "step": 6517
    },
    {
      "epoch": 1.0203506574827803,
      "grad_norm": 1.8043177127838135,
      "learning_rate": 0.00015320136852394918,
      "loss": 0.607,
      "step": 6518
    },
    {
      "epoch": 1.0205072010018785,
      "grad_norm": 2.9217605590820312,
      "learning_rate": 0.0001531769305962854,
      "loss": 0.8962,
      "step": 6519
    },
    {
      "epoch": 1.0206637445209767,
      "grad_norm": 1.854831337928772,
      "learning_rate": 0.00015315249266862168,
      "loss": 0.4919,
      "step": 6520
    },
    {
      "epoch": 1.0208202880400752,
      "grad_norm": 2.2908477783203125,
      "learning_rate": 0.00015312805474095796,
      "loss": 0.9187,
      "step": 6521
    },
    {
      "epoch": 1.0209768315591734,
      "grad_norm": 3.0967938899993896,
      "learning_rate": 0.0001531036168132942,
      "loss": 0.4282,
      "step": 6522
    },
    {
      "epoch": 1.0211333750782718,
      "grad_norm": 1.6227751970291138,
      "learning_rate": 0.0001530791788856305,
      "loss": 0.6401,
      "step": 6523
    },
    {
      "epoch": 1.02128991859737,
      "grad_norm": 1.9775352478027344,
      "learning_rate": 0.00015305474095796676,
      "loss": 1.1405,
      "step": 6524
    },
    {
      "epoch": 1.0214464621164683,
      "grad_norm": 2.77632212638855,
      "learning_rate": 0.00015303030303030302,
      "loss": 0.6276,
      "step": 6525
    },
    {
      "epoch": 1.0216030056355667,
      "grad_norm": 5.503850936889648,
      "learning_rate": 0.00015300586510263927,
      "loss": 1.3446,
      "step": 6526
    },
    {
      "epoch": 1.021759549154665,
      "grad_norm": 4.683707237243652,
      "learning_rate": 0.00015298142717497555,
      "loss": 1.1199,
      "step": 6527
    },
    {
      "epoch": 1.0219160926737634,
      "grad_norm": 2.9078924655914307,
      "learning_rate": 0.0001529569892473118,
      "loss": 1.3954,
      "step": 6528
    },
    {
      "epoch": 1.0220726361928616,
      "grad_norm": 1.9573535919189453,
      "learning_rate": 0.00015293255131964808,
      "loss": 0.5525,
      "step": 6529
    },
    {
      "epoch": 1.0222291797119598,
      "grad_norm": 2.339132308959961,
      "learning_rate": 0.00015290811339198435,
      "loss": 1.9003,
      "step": 6530
    },
    {
      "epoch": 1.0223857232310583,
      "grad_norm": 2.6952342987060547,
      "learning_rate": 0.0001528836754643206,
      "loss": 1.2449,
      "step": 6531
    },
    {
      "epoch": 1.0225422667501565,
      "grad_norm": 6.4942216873168945,
      "learning_rate": 0.00015285923753665688,
      "loss": 2.0955,
      "step": 6532
    },
    {
      "epoch": 1.022698810269255,
      "grad_norm": 2.6404001712799072,
      "learning_rate": 0.00015283479960899316,
      "loss": 1.3895,
      "step": 6533
    },
    {
      "epoch": 1.0228553537883531,
      "grad_norm": 1.230292558670044,
      "learning_rate": 0.00015281036168132939,
      "loss": 0.4467,
      "step": 6534
    },
    {
      "epoch": 1.0230118973074516,
      "grad_norm": 1.6380767822265625,
      "learning_rate": 0.00015278592375366566,
      "loss": 0.5228,
      "step": 6535
    },
    {
      "epoch": 1.0231684408265498,
      "grad_norm": 1.788313388824463,
      "learning_rate": 0.00015276148582600194,
      "loss": 0.4542,
      "step": 6536
    },
    {
      "epoch": 1.023324984345648,
      "grad_norm": 2.3561313152313232,
      "learning_rate": 0.0001527370478983382,
      "loss": 0.8112,
      "step": 6537
    },
    {
      "epoch": 1.0234815278647464,
      "grad_norm": 2.4350478649139404,
      "learning_rate": 0.00015271260997067447,
      "loss": 1.1279,
      "step": 6538
    },
    {
      "epoch": 1.0236380713838447,
      "grad_norm": 1.0181922912597656,
      "learning_rate": 0.00015268817204301075,
      "loss": 0.3027,
      "step": 6539
    },
    {
      "epoch": 1.023794614902943,
      "grad_norm": 1.0185344219207764,
      "learning_rate": 0.000152663734115347,
      "loss": 0.2241,
      "step": 6540
    },
    {
      "epoch": 1.0239511584220413,
      "grad_norm": 1.063295841217041,
      "learning_rate": 0.00015263929618768328,
      "loss": 0.3597,
      "step": 6541
    },
    {
      "epoch": 1.0241077019411395,
      "grad_norm": 0.785956621170044,
      "learning_rate": 0.00015261485826001956,
      "loss": 0.252,
      "step": 6542
    },
    {
      "epoch": 1.024264245460238,
      "grad_norm": 0.5662931203842163,
      "learning_rate": 0.00015259042033235578,
      "loss": 0.2251,
      "step": 6543
    },
    {
      "epoch": 1.0244207889793362,
      "grad_norm": 0.4908407926559448,
      "learning_rate": 0.00015256598240469206,
      "loss": 0.1643,
      "step": 6544
    },
    {
      "epoch": 1.0245773324984346,
      "grad_norm": 0.8578366041183472,
      "learning_rate": 0.00015254154447702834,
      "loss": 0.2976,
      "step": 6545
    },
    {
      "epoch": 1.0247338760175329,
      "grad_norm": 0.5953052639961243,
      "learning_rate": 0.0001525171065493646,
      "loss": 0.2358,
      "step": 6546
    },
    {
      "epoch": 1.0248904195366313,
      "grad_norm": 0.401744544506073,
      "learning_rate": 0.00015249266862170087,
      "loss": 0.1725,
      "step": 6547
    },
    {
      "epoch": 1.0250469630557295,
      "grad_norm": 0.6543681621551514,
      "learning_rate": 0.00015246823069403715,
      "loss": 0.3137,
      "step": 6548
    },
    {
      "epoch": 1.0252035065748277,
      "grad_norm": 0.9384623765945435,
      "learning_rate": 0.0001524437927663734,
      "loss": 0.2601,
      "step": 6549
    },
    {
      "epoch": 1.0253600500939262,
      "grad_norm": 1.2204899787902832,
      "learning_rate": 0.00015241935483870965,
      "loss": 0.3343,
      "step": 6550
    },
    {
      "epoch": 1.0255165936130244,
      "grad_norm": 0.8273910880088806,
      "learning_rate": 0.00015239491691104593,
      "loss": 0.3626,
      "step": 6551
    },
    {
      "epoch": 1.0256731371321228,
      "grad_norm": 0.8227601051330566,
      "learning_rate": 0.00015237047898338218,
      "loss": 0.2757,
      "step": 6552
    },
    {
      "epoch": 1.025829680651221,
      "grad_norm": 0.7043795585632324,
      "learning_rate": 0.00015234604105571846,
      "loss": 0.3225,
      "step": 6553
    },
    {
      "epoch": 1.0259862241703193,
      "grad_norm": 1.460331916809082,
      "learning_rate": 0.00015232160312805474,
      "loss": 0.2981,
      "step": 6554
    },
    {
      "epoch": 1.0261427676894177,
      "grad_norm": 1.7429531812667847,
      "learning_rate": 0.000152297165200391,
      "loss": 0.7302,
      "step": 6555
    },
    {
      "epoch": 1.026299311208516,
      "grad_norm": 0.9695984125137329,
      "learning_rate": 0.00015227272727272727,
      "loss": 0.4463,
      "step": 6556
    },
    {
      "epoch": 1.0264558547276144,
      "grad_norm": 0.9565237164497375,
      "learning_rate": 0.00015224828934506354,
      "loss": 0.4064,
      "step": 6557
    },
    {
      "epoch": 1.0266123982467126,
      "grad_norm": 1.8992421627044678,
      "learning_rate": 0.00015222385141739977,
      "loss": 0.3796,
      "step": 6558
    },
    {
      "epoch": 1.0267689417658108,
      "grad_norm": 1.4258291721343994,
      "learning_rate": 0.00015219941348973605,
      "loss": 0.3354,
      "step": 6559
    },
    {
      "epoch": 1.0269254852849092,
      "grad_norm": 2.1398561000823975,
      "learning_rate": 0.00015217497556207232,
      "loss": 0.4635,
      "step": 6560
    },
    {
      "epoch": 1.0270820288040075,
      "grad_norm": 1.383025884628296,
      "learning_rate": 0.00015215053763440858,
      "loss": 0.5687,
      "step": 6561
    },
    {
      "epoch": 1.027238572323106,
      "grad_norm": 1.5760375261306763,
      "learning_rate": 0.00015212609970674485,
      "loss": 0.5766,
      "step": 6562
    },
    {
      "epoch": 1.027395115842204,
      "grad_norm": 1.6190121173858643,
      "learning_rate": 0.00015210166177908113,
      "loss": 0.5502,
      "step": 6563
    },
    {
      "epoch": 1.0275516593613025,
      "grad_norm": 1.2204431295394897,
      "learning_rate": 0.00015207722385141738,
      "loss": 0.6046,
      "step": 6564
    },
    {
      "epoch": 1.0277082028804008,
      "grad_norm": 1.0577452182769775,
      "learning_rate": 0.00015205278592375366,
      "loss": 0.3985,
      "step": 6565
    },
    {
      "epoch": 1.027864746399499,
      "grad_norm": 1.111060619354248,
      "learning_rate": 0.00015202834799608994,
      "loss": 0.4528,
      "step": 6566
    },
    {
      "epoch": 1.0280212899185974,
      "grad_norm": 2.2486462593078613,
      "learning_rate": 0.00015200391006842616,
      "loss": 0.8267,
      "step": 6567
    },
    {
      "epoch": 1.0281778334376956,
      "grad_norm": 1.4202734231948853,
      "learning_rate": 0.00015197947214076244,
      "loss": 0.8305,
      "step": 6568
    },
    {
      "epoch": 1.028334376956794,
      "grad_norm": 2.8072428703308105,
      "learning_rate": 0.00015195503421309872,
      "loss": 0.7312,
      "step": 6569
    },
    {
      "epoch": 1.0284909204758923,
      "grad_norm": 3.127408504486084,
      "learning_rate": 0.00015193059628543497,
      "loss": 0.8986,
      "step": 6570
    },
    {
      "epoch": 1.0286474639949905,
      "grad_norm": 1.911171793937683,
      "learning_rate": 0.00015190615835777125,
      "loss": 0.8522,
      "step": 6571
    },
    {
      "epoch": 1.028804007514089,
      "grad_norm": 2.793549060821533,
      "learning_rate": 0.00015188172043010753,
      "loss": 0.9098,
      "step": 6572
    },
    {
      "epoch": 1.0289605510331872,
      "grad_norm": 1.8251924514770508,
      "learning_rate": 0.00015185728250244375,
      "loss": 1.2318,
      "step": 6573
    },
    {
      "epoch": 1.0291170945522856,
      "grad_norm": 7.607204437255859,
      "learning_rate": 0.00015183284457478003,
      "loss": 0.9744,
      "step": 6574
    },
    {
      "epoch": 1.0292736380713838,
      "grad_norm": 3.5871646404266357,
      "learning_rate": 0.0001518084066471163,
      "loss": 1.8618,
      "step": 6575
    },
    {
      "epoch": 1.029430181590482,
      "grad_norm": 3.6757612228393555,
      "learning_rate": 0.00015178396871945256,
      "loss": 1.4547,
      "step": 6576
    },
    {
      "epoch": 1.0295867251095805,
      "grad_norm": 2.2011477947235107,
      "learning_rate": 0.00015175953079178884,
      "loss": 0.8764,
      "step": 6577
    },
    {
      "epoch": 1.0297432686286787,
      "grad_norm": 2.396167278289795,
      "learning_rate": 0.00015173509286412512,
      "loss": 0.9784,
      "step": 6578
    },
    {
      "epoch": 1.0298998121477771,
      "grad_norm": 2.8914072513580322,
      "learning_rate": 0.00015171065493646137,
      "loss": 1.4503,
      "step": 6579
    },
    {
      "epoch": 1.0300563556668754,
      "grad_norm": 2.1105411052703857,
      "learning_rate": 0.00015168621700879765,
      "loss": 1.3224,
      "step": 6580
    },
    {
      "epoch": 1.0302128991859738,
      "grad_norm": 1.6223315000534058,
      "learning_rate": 0.00015166177908113393,
      "loss": 0.9473,
      "step": 6581
    },
    {
      "epoch": 1.030369442705072,
      "grad_norm": 2.1070399284362793,
      "learning_rate": 0.00015163734115347015,
      "loss": 1.2726,
      "step": 6582
    },
    {
      "epoch": 1.0305259862241702,
      "grad_norm": 2.866163492202759,
      "learning_rate": 0.00015161290322580643,
      "loss": 1.6022,
      "step": 6583
    },
    {
      "epoch": 1.0306825297432687,
      "grad_norm": 1.343387484550476,
      "learning_rate": 0.0001515884652981427,
      "loss": 0.6449,
      "step": 6584
    },
    {
      "epoch": 1.030839073262367,
      "grad_norm": 2.003796339035034,
      "learning_rate": 0.00015156402737047896,
      "loss": 0.6767,
      "step": 6585
    },
    {
      "epoch": 1.0309956167814653,
      "grad_norm": 1.329639196395874,
      "learning_rate": 0.00015153958944281524,
      "loss": 0.8415,
      "step": 6586
    },
    {
      "epoch": 1.0311521603005636,
      "grad_norm": 2.692793607711792,
      "learning_rate": 0.00015151515151515152,
      "loss": 0.717,
      "step": 6587
    },
    {
      "epoch": 1.0313087038196618,
      "grad_norm": 1.7069381475448608,
      "learning_rate": 0.00015149071358748777,
      "loss": 0.7249,
      "step": 6588
    },
    {
      "epoch": 1.0314652473387602,
      "grad_norm": 0.498267263174057,
      "learning_rate": 0.00015146627565982404,
      "loss": 0.2442,
      "step": 6589
    },
    {
      "epoch": 1.0316217908578584,
      "grad_norm": 0.7013705372810364,
      "learning_rate": 0.0001514418377321603,
      "loss": 0.3577,
      "step": 6590
    },
    {
      "epoch": 1.0317783343769569,
      "grad_norm": 1.0541874170303345,
      "learning_rate": 0.00015141739980449655,
      "loss": 0.2867,
      "step": 6591
    },
    {
      "epoch": 1.031934877896055,
      "grad_norm": 1.2763069868087769,
      "learning_rate": 0.00015139296187683283,
      "loss": 0.2876,
      "step": 6592
    },
    {
      "epoch": 1.0320914214151533,
      "grad_norm": 0.702283501625061,
      "learning_rate": 0.0001513685239491691,
      "loss": 0.3681,
      "step": 6593
    },
    {
      "epoch": 1.0322479649342517,
      "grad_norm": 0.6756613850593567,
      "learning_rate": 0.00015134408602150536,
      "loss": 0.2794,
      "step": 6594
    },
    {
      "epoch": 1.03240450845335,
      "grad_norm": 1.0558266639709473,
      "learning_rate": 0.00015131964809384163,
      "loss": 0.2589,
      "step": 6595
    },
    {
      "epoch": 1.0325610519724484,
      "grad_norm": 0.8309182524681091,
      "learning_rate": 0.0001512952101661779,
      "loss": 0.4377,
      "step": 6596
    },
    {
      "epoch": 1.0327175954915466,
      "grad_norm": 0.7502772212028503,
      "learning_rate": 0.00015127077223851414,
      "loss": 0.2772,
      "step": 6597
    },
    {
      "epoch": 1.032874139010645,
      "grad_norm": 1.0841580629348755,
      "learning_rate": 0.00015124633431085041,
      "loss": 0.3705,
      "step": 6598
    },
    {
      "epoch": 1.0330306825297433,
      "grad_norm": 0.5633625388145447,
      "learning_rate": 0.0001512218963831867,
      "loss": 0.2433,
      "step": 6599
    },
    {
      "epoch": 1.0331872260488415,
      "grad_norm": 0.6718013882637024,
      "learning_rate": 0.00015119745845552294,
      "loss": 0.2835,
      "step": 6600
    },
    {
      "epoch": 1.03334376956794,
      "grad_norm": 0.8712942600250244,
      "learning_rate": 0.00015117302052785922,
      "loss": 0.3426,
      "step": 6601
    },
    {
      "epoch": 1.0335003130870382,
      "grad_norm": 0.9348687529563904,
      "learning_rate": 0.0001511485826001955,
      "loss": 0.311,
      "step": 6602
    },
    {
      "epoch": 1.0336568566061366,
      "grad_norm": 0.8929564356803894,
      "learning_rate": 0.00015112414467253175,
      "loss": 0.3267,
      "step": 6603
    },
    {
      "epoch": 1.0338134001252348,
      "grad_norm": 1.1071451902389526,
      "learning_rate": 0.00015109970674486803,
      "loss": 0.491,
      "step": 6604
    },
    {
      "epoch": 1.033969943644333,
      "grad_norm": 0.7876406908035278,
      "learning_rate": 0.0001510752688172043,
      "loss": 0.2884,
      "step": 6605
    },
    {
      "epoch": 1.0341264871634315,
      "grad_norm": 1.126440405845642,
      "learning_rate": 0.00015105083088954053,
      "loss": 0.4125,
      "step": 6606
    },
    {
      "epoch": 1.0342830306825297,
      "grad_norm": 0.6488103866577148,
      "learning_rate": 0.0001510263929618768,
      "loss": 0.1574,
      "step": 6607
    },
    {
      "epoch": 1.0344395742016281,
      "grad_norm": 2.044353485107422,
      "learning_rate": 0.0001510019550342131,
      "loss": 0.3864,
      "step": 6608
    },
    {
      "epoch": 1.0345961177207263,
      "grad_norm": 1.4567679166793823,
      "learning_rate": 0.00015097751710654934,
      "loss": 0.4484,
      "step": 6609
    },
    {
      "epoch": 1.0347526612398246,
      "grad_norm": 2.9722204208374023,
      "learning_rate": 0.00015095307917888562,
      "loss": 0.9196,
      "step": 6610
    },
    {
      "epoch": 1.034909204758923,
      "grad_norm": 1.4675703048706055,
      "learning_rate": 0.0001509286412512219,
      "loss": 0.7483,
      "step": 6611
    },
    {
      "epoch": 1.0350657482780212,
      "grad_norm": 1.8821780681610107,
      "learning_rate": 0.00015090420332355815,
      "loss": 0.586,
      "step": 6612
    },
    {
      "epoch": 1.0352222917971197,
      "grad_norm": 2.960568428039551,
      "learning_rate": 0.00015087976539589443,
      "loss": 0.6357,
      "step": 6613
    },
    {
      "epoch": 1.0353788353162179,
      "grad_norm": 1.3695021867752075,
      "learning_rate": 0.00015085532746823068,
      "loss": 0.5007,
      "step": 6614
    },
    {
      "epoch": 1.0355353788353163,
      "grad_norm": 1.4479131698608398,
      "learning_rate": 0.00015083088954056693,
      "loss": 0.5556,
      "step": 6615
    },
    {
      "epoch": 1.0356919223544145,
      "grad_norm": 1.7650444507598877,
      "learning_rate": 0.0001508064516129032,
      "loss": 0.9825,
      "step": 6616
    },
    {
      "epoch": 1.0358484658735128,
      "grad_norm": 2.19905424118042,
      "learning_rate": 0.00015078201368523949,
      "loss": 0.6728,
      "step": 6617
    },
    {
      "epoch": 1.0360050093926112,
      "grad_norm": 1.4833353757858276,
      "learning_rate": 0.00015075757575757574,
      "loss": 0.7472,
      "step": 6618
    },
    {
      "epoch": 1.0361615529117094,
      "grad_norm": 1.978433609008789,
      "learning_rate": 0.00015073313782991202,
      "loss": 0.7654,
      "step": 6619
    },
    {
      "epoch": 1.0363180964308079,
      "grad_norm": 1.5188682079315186,
      "learning_rate": 0.0001507086999022483,
      "loss": 1.1002,
      "step": 6620
    },
    {
      "epoch": 1.036474639949906,
      "grad_norm": 2.915623188018799,
      "learning_rate": 0.00015068426197458452,
      "loss": 0.746,
      "step": 6621
    },
    {
      "epoch": 1.0366311834690043,
      "grad_norm": 3.499290704727173,
      "learning_rate": 0.0001506598240469208,
      "loss": 0.6476,
      "step": 6622
    },
    {
      "epoch": 1.0367877269881027,
      "grad_norm": 2.9197161197662354,
      "learning_rate": 0.00015063538611925707,
      "loss": 0.8943,
      "step": 6623
    },
    {
      "epoch": 1.036944270507201,
      "grad_norm": 1.2932711839675903,
      "learning_rate": 0.00015061094819159333,
      "loss": 0.7917,
      "step": 6624
    },
    {
      "epoch": 1.0371008140262994,
      "grad_norm": 2.556731939315796,
      "learning_rate": 0.0001505865102639296,
      "loss": 0.6509,
      "step": 6625
    },
    {
      "epoch": 1.0372573575453976,
      "grad_norm": 3.067978620529175,
      "learning_rate": 0.00015056207233626588,
      "loss": 1.0172,
      "step": 6626
    },
    {
      "epoch": 1.0374139010644958,
      "grad_norm": 2.2281928062438965,
      "learning_rate": 0.00015053763440860213,
      "loss": 0.5996,
      "step": 6627
    },
    {
      "epoch": 1.0375704445835943,
      "grad_norm": 2.0016701221466064,
      "learning_rate": 0.0001505131964809384,
      "loss": 0.6033,
      "step": 6628
    },
    {
      "epoch": 1.0377269881026925,
      "grad_norm": 2.9594852924346924,
      "learning_rate": 0.0001504887585532747,
      "loss": 1.0168,
      "step": 6629
    },
    {
      "epoch": 1.037883531621791,
      "grad_norm": 2.735733985900879,
      "learning_rate": 0.00015046432062561091,
      "loss": 1.1409,
      "step": 6630
    },
    {
      "epoch": 1.0380400751408891,
      "grad_norm": 4.281361103057861,
      "learning_rate": 0.0001504398826979472,
      "loss": 1.7845,
      "step": 6631
    },
    {
      "epoch": 1.0381966186599876,
      "grad_norm": 4.471887111663818,
      "learning_rate": 0.00015041544477028347,
      "loss": 1.0803,
      "step": 6632
    },
    {
      "epoch": 1.0383531621790858,
      "grad_norm": 3.5582237243652344,
      "learning_rate": 0.00015039100684261972,
      "loss": 1.1523,
      "step": 6633
    },
    {
      "epoch": 1.038509705698184,
      "grad_norm": 2.951169490814209,
      "learning_rate": 0.000150366568914956,
      "loss": 0.6606,
      "step": 6634
    },
    {
      "epoch": 1.0386662492172825,
      "grad_norm": 1.8436496257781982,
      "learning_rate": 0.00015034213098729228,
      "loss": 0.6368,
      "step": 6635
    },
    {
      "epoch": 1.0388227927363807,
      "grad_norm": 2.94608473777771,
      "learning_rate": 0.00015031769305962853,
      "loss": 0.4588,
      "step": 6636
    },
    {
      "epoch": 1.0389793362554791,
      "grad_norm": 2.0353431701660156,
      "learning_rate": 0.0001502932551319648,
      "loss": 0.4628,
      "step": 6637
    },
    {
      "epoch": 1.0391358797745773,
      "grad_norm": 2.008845329284668,
      "learning_rate": 0.00015026881720430106,
      "loss": 0.7833,
      "step": 6638
    },
    {
      "epoch": 1.0392924232936755,
      "grad_norm": 0.664910078048706,
      "learning_rate": 0.0001502443792766373,
      "loss": 0.194,
      "step": 6639
    },
    {
      "epoch": 1.039448966812774,
      "grad_norm": 0.388094037771225,
      "learning_rate": 0.0001502199413489736,
      "loss": 0.1953,
      "step": 6640
    },
    {
      "epoch": 1.0396055103318722,
      "grad_norm": 0.3977845311164856,
      "learning_rate": 0.00015019550342130987,
      "loss": 0.1952,
      "step": 6641
    },
    {
      "epoch": 1.0397620538509706,
      "grad_norm": 0.4224265515804291,
      "learning_rate": 0.00015017106549364612,
      "loss": 0.1599,
      "step": 6642
    },
    {
      "epoch": 1.0399185973700689,
      "grad_norm": 0.7730469703674316,
      "learning_rate": 0.0001501466275659824,
      "loss": 0.2271,
      "step": 6643
    },
    {
      "epoch": 1.040075140889167,
      "grad_norm": 0.7396877408027649,
      "learning_rate": 0.00015012218963831868,
      "loss": 0.328,
      "step": 6644
    },
    {
      "epoch": 1.0402316844082655,
      "grad_norm": 0.7163776755332947,
      "learning_rate": 0.0001500977517106549,
      "loss": 0.2924,
      "step": 6645
    },
    {
      "epoch": 1.0403882279273637,
      "grad_norm": 0.7537462115287781,
      "learning_rate": 0.00015007331378299118,
      "loss": 0.2665,
      "step": 6646
    },
    {
      "epoch": 1.0405447714464622,
      "grad_norm": 4.295088768005371,
      "learning_rate": 0.00015004887585532746,
      "loss": 1.0396,
      "step": 6647
    },
    {
      "epoch": 1.0407013149655604,
      "grad_norm": 0.830767810344696,
      "learning_rate": 0.0001500244379276637,
      "loss": 0.2891,
      "step": 6648
    },
    {
      "epoch": 1.0408578584846588,
      "grad_norm": 0.9357022643089294,
      "learning_rate": 0.00015,
      "loss": 0.3978,
      "step": 6649
    },
    {
      "epoch": 1.041014402003757,
      "grad_norm": 0.6923136115074158,
      "learning_rate": 0.00014997556207233624,
      "loss": 0.2655,
      "step": 6650
    },
    {
      "epoch": 1.0411709455228553,
      "grad_norm": 1.0531424283981323,
      "learning_rate": 0.00014995112414467252,
      "loss": 0.2939,
      "step": 6651
    },
    {
      "epoch": 1.0413274890419537,
      "grad_norm": 0.8230469822883606,
      "learning_rate": 0.0001499266862170088,
      "loss": 0.4025,
      "step": 6652
    },
    {
      "epoch": 1.041484032561052,
      "grad_norm": 0.7676156163215637,
      "learning_rate": 0.00014990224828934505,
      "loss": 0.3075,
      "step": 6653
    },
    {
      "epoch": 1.0416405760801504,
      "grad_norm": 2.29850697517395,
      "learning_rate": 0.00014987781036168132,
      "loss": 0.5837,
      "step": 6654
    },
    {
      "epoch": 1.0417971195992486,
      "grad_norm": 0.798516571521759,
      "learning_rate": 0.00014985337243401758,
      "loss": 0.4561,
      "step": 6655
    },
    {
      "epoch": 1.0419536631183468,
      "grad_norm": 1.210540771484375,
      "learning_rate": 0.00014982893450635385,
      "loss": 0.5438,
      "step": 6656
    },
    {
      "epoch": 1.0421102066374452,
      "grad_norm": 1.6626873016357422,
      "learning_rate": 0.00014980449657869013,
      "loss": 0.6722,
      "step": 6657
    },
    {
      "epoch": 1.0422667501565435,
      "grad_norm": 0.6938304305076599,
      "learning_rate": 0.00014978005865102638,
      "loss": 0.5357,
      "step": 6658
    },
    {
      "epoch": 1.042423293675642,
      "grad_norm": 0.9663745164871216,
      "learning_rate": 0.00014975562072336263,
      "loss": 0.4106,
      "step": 6659
    },
    {
      "epoch": 1.0425798371947401,
      "grad_norm": 0.8333467245101929,
      "learning_rate": 0.0001497311827956989,
      "loss": 0.3207,
      "step": 6660
    },
    {
      "epoch": 1.0427363807138383,
      "grad_norm": 0.7671467065811157,
      "learning_rate": 0.00014970674486803516,
      "loss": 0.3437,
      "step": 6661
    },
    {
      "epoch": 1.0428929242329368,
      "grad_norm": 0.9945407509803772,
      "learning_rate": 0.00014968230694037144,
      "loss": 0.4651,
      "step": 6662
    },
    {
      "epoch": 1.043049467752035,
      "grad_norm": 1.5871999263763428,
      "learning_rate": 0.00014965786901270772,
      "loss": 0.5176,
      "step": 6663
    },
    {
      "epoch": 1.0432060112711334,
      "grad_norm": 3.3692855834960938,
      "learning_rate": 0.00014963343108504397,
      "loss": 0.7892,
      "step": 6664
    },
    {
      "epoch": 1.0433625547902317,
      "grad_norm": 2.7460215091705322,
      "learning_rate": 0.00014960899315738022,
      "loss": 0.4972,
      "step": 6665
    },
    {
      "epoch": 1.04351909830933,
      "grad_norm": 2.0710816383361816,
      "learning_rate": 0.0001495845552297165,
      "loss": 0.4056,
      "step": 6666
    },
    {
      "epoch": 1.0436756418284283,
      "grad_norm": 2.1053054332733154,
      "learning_rate": 0.00014956011730205278,
      "loss": 0.6009,
      "step": 6667
    },
    {
      "epoch": 1.0438321853475265,
      "grad_norm": 2.5301668643951416,
      "learning_rate": 0.00014953567937438903,
      "loss": 0.8748,
      "step": 6668
    },
    {
      "epoch": 1.043988728866625,
      "grad_norm": 1.9505361318588257,
      "learning_rate": 0.0001495112414467253,
      "loss": 0.6859,
      "step": 6669
    },
    {
      "epoch": 1.0441452723857232,
      "grad_norm": 1.7739278078079224,
      "learning_rate": 0.00014948680351906156,
      "loss": 0.4152,
      "step": 6670
    },
    {
      "epoch": 1.0443018159048216,
      "grad_norm": 2.2372188568115234,
      "learning_rate": 0.00014946236559139784,
      "loss": 0.5335,
      "step": 6671
    },
    {
      "epoch": 1.0444583594239198,
      "grad_norm": 3.2507851123809814,
      "learning_rate": 0.00014943792766373412,
      "loss": 0.9134,
      "step": 6672
    },
    {
      "epoch": 1.044614902943018,
      "grad_norm": 1.7486919164657593,
      "learning_rate": 0.00014941348973607037,
      "loss": 0.9072,
      "step": 6673
    },
    {
      "epoch": 1.0447714464621165,
      "grad_norm": 1.5472897291183472,
      "learning_rate": 0.00014938905180840662,
      "loss": 0.8984,
      "step": 6674
    },
    {
      "epoch": 1.0449279899812147,
      "grad_norm": 2.7646539211273193,
      "learning_rate": 0.0001493646138807429,
      "loss": 0.6659,
      "step": 6675
    },
    {
      "epoch": 1.0450845335003132,
      "grad_norm": 2.3693439960479736,
      "learning_rate": 0.00014934017595307918,
      "loss": 1.1551,
      "step": 6676
    },
    {
      "epoch": 1.0452410770194114,
      "grad_norm": 4.565879821777344,
      "learning_rate": 0.00014931573802541543,
      "loss": 1.4091,
      "step": 6677
    },
    {
      "epoch": 1.0453976205385098,
      "grad_norm": 2.8540971279144287,
      "learning_rate": 0.0001492913000977517,
      "loss": 1.1402,
      "step": 6678
    },
    {
      "epoch": 1.045554164057608,
      "grad_norm": 3.433985948562622,
      "learning_rate": 0.00014926686217008796,
      "loss": 1.2866,
      "step": 6679
    },
    {
      "epoch": 1.0457107075767063,
      "grad_norm": 1.5474061965942383,
      "learning_rate": 0.00014924242424242424,
      "loss": 0.4375,
      "step": 6680
    },
    {
      "epoch": 1.0458672510958047,
      "grad_norm": 1.9934204816818237,
      "learning_rate": 0.00014921798631476051,
      "loss": 0.9397,
      "step": 6681
    },
    {
      "epoch": 1.046023794614903,
      "grad_norm": 4.180722236633301,
      "learning_rate": 0.00014919354838709677,
      "loss": 1.5848,
      "step": 6682
    },
    {
      "epoch": 1.0461803381340014,
      "grad_norm": 2.291654348373413,
      "learning_rate": 0.00014916911045943302,
      "loss": 1.2605,
      "step": 6683
    },
    {
      "epoch": 1.0463368816530996,
      "grad_norm": 2.387073516845703,
      "learning_rate": 0.0001491446725317693,
      "loss": 0.4326,
      "step": 6684
    },
    {
      "epoch": 1.0464934251721978,
      "grad_norm": 2.3532493114471436,
      "learning_rate": 0.00014912023460410555,
      "loss": 0.4884,
      "step": 6685
    },
    {
      "epoch": 1.0466499686912962,
      "grad_norm": 2.2486379146575928,
      "learning_rate": 0.00014909579667644183,
      "loss": 1.0124,
      "step": 6686
    },
    {
      "epoch": 1.0468065122103944,
      "grad_norm": 2.661993980407715,
      "learning_rate": 0.0001490713587487781,
      "loss": 1.0389,
      "step": 6687
    },
    {
      "epoch": 1.0469630557294929,
      "grad_norm": 2.130023717880249,
      "learning_rate": 0.00014904692082111435,
      "loss": 0.4506,
      "step": 6688
    },
    {
      "epoch": 1.047119599248591,
      "grad_norm": 0.6266974806785583,
      "learning_rate": 0.0001490224828934506,
      "loss": 0.2949,
      "step": 6689
    },
    {
      "epoch": 1.0472761427676893,
      "grad_norm": 0.5856244564056396,
      "learning_rate": 0.00014899804496578688,
      "loss": 0.2782,
      "step": 6690
    },
    {
      "epoch": 1.0474326862867878,
      "grad_norm": 0.578836977481842,
      "learning_rate": 0.00014897360703812316,
      "loss": 0.244,
      "step": 6691
    },
    {
      "epoch": 1.047589229805886,
      "grad_norm": 0.5507057905197144,
      "learning_rate": 0.00014894916911045941,
      "loss": 0.2312,
      "step": 6692
    },
    {
      "epoch": 1.0477457733249844,
      "grad_norm": 0.4590769112110138,
      "learning_rate": 0.0001489247311827957,
      "loss": 0.2004,
      "step": 6693
    },
    {
      "epoch": 1.0479023168440826,
      "grad_norm": 0.5535408854484558,
      "learning_rate": 0.00014890029325513194,
      "loss": 0.2628,
      "step": 6694
    },
    {
      "epoch": 1.0480588603631809,
      "grad_norm": 2.307926654815674,
      "learning_rate": 0.00014887585532746822,
      "loss": 0.2208,
      "step": 6695
    },
    {
      "epoch": 1.0482154038822793,
      "grad_norm": 0.5684087872505188,
      "learning_rate": 0.0001488514173998045,
      "loss": 0.2184,
      "step": 6696
    },
    {
      "epoch": 1.0483719474013775,
      "grad_norm": 0.6311387419700623,
      "learning_rate": 0.00014882697947214075,
      "loss": 0.2584,
      "step": 6697
    },
    {
      "epoch": 1.048528490920476,
      "grad_norm": 0.6801841259002686,
      "learning_rate": 0.000148802541544477,
      "loss": 0.2507,
      "step": 6698
    },
    {
      "epoch": 1.0486850344395742,
      "grad_norm": 0.8191403150558472,
      "learning_rate": 0.00014877810361681328,
      "loss": 0.3224,
      "step": 6699
    },
    {
      "epoch": 1.0488415779586726,
      "grad_norm": 0.9720447063446045,
      "learning_rate": 0.00014875366568914956,
      "loss": 0.3435,
      "step": 6700
    },
    {
      "epoch": 1.0489981214777708,
      "grad_norm": 0.380145400762558,
      "learning_rate": 0.0001487292277614858,
      "loss": 0.1308,
      "step": 6701
    },
    {
      "epoch": 1.049154664996869,
      "grad_norm": 0.5885724425315857,
      "learning_rate": 0.0001487047898338221,
      "loss": 0.3108,
      "step": 6702
    },
    {
      "epoch": 1.0493112085159675,
      "grad_norm": 0.8687223792076111,
      "learning_rate": 0.00014868035190615834,
      "loss": 0.3114,
      "step": 6703
    },
    {
      "epoch": 1.0494677520350657,
      "grad_norm": 1.1362780332565308,
      "learning_rate": 0.00014865591397849462,
      "loss": 0.3741,
      "step": 6704
    },
    {
      "epoch": 1.0496242955541641,
      "grad_norm": 1.4467766284942627,
      "learning_rate": 0.00014863147605083087,
      "loss": 0.2473,
      "step": 6705
    },
    {
      "epoch": 1.0497808390732624,
      "grad_norm": 1.3377172946929932,
      "learning_rate": 0.00014860703812316715,
      "loss": 0.5426,
      "step": 6706
    },
    {
      "epoch": 1.0499373825923606,
      "grad_norm": 1.023105502128601,
      "learning_rate": 0.0001485826001955034,
      "loss": 0.4587,
      "step": 6707
    },
    {
      "epoch": 1.050093926111459,
      "grad_norm": 2.0178208351135254,
      "learning_rate": 0.00014855816226783968,
      "loss": 0.3052,
      "step": 6708
    },
    {
      "epoch": 1.0502504696305572,
      "grad_norm": 1.5888885259628296,
      "learning_rate": 0.00014853372434017593,
      "loss": 0.4834,
      "step": 6709
    },
    {
      "epoch": 1.0504070131496557,
      "grad_norm": 2.150106430053711,
      "learning_rate": 0.0001485092864125122,
      "loss": 0.6655,
      "step": 6710
    },
    {
      "epoch": 1.050563556668754,
      "grad_norm": 1.21843421459198,
      "learning_rate": 0.00014848484848484849,
      "loss": 0.6639,
      "step": 6711
    },
    {
      "epoch": 1.0507201001878523,
      "grad_norm": 1.0294426679611206,
      "learning_rate": 0.00014846041055718474,
      "loss": 0.3775,
      "step": 6712
    },
    {
      "epoch": 1.0508766437069506,
      "grad_norm": 1.3102911710739136,
      "learning_rate": 0.000148435972629521,
      "loss": 0.6612,
      "step": 6713
    },
    {
      "epoch": 1.0510331872260488,
      "grad_norm": 2.4257891178131104,
      "learning_rate": 0.00014841153470185727,
      "loss": 0.6148,
      "step": 6714
    },
    {
      "epoch": 1.0511897307451472,
      "grad_norm": 1.66291344165802,
      "learning_rate": 0.00014838709677419355,
      "loss": 0.7202,
      "step": 6715
    },
    {
      "epoch": 1.0513462742642454,
      "grad_norm": 2.0738868713378906,
      "learning_rate": 0.0001483626588465298,
      "loss": 0.7751,
      "step": 6716
    },
    {
      "epoch": 1.0515028177833439,
      "grad_norm": 1.5332388877868652,
      "learning_rate": 0.00014833822091886607,
      "loss": 0.5269,
      "step": 6717
    },
    {
      "epoch": 1.051659361302442,
      "grad_norm": 1.8209433555603027,
      "learning_rate": 0.00014831378299120233,
      "loss": 0.8477,
      "step": 6718
    },
    {
      "epoch": 1.0518159048215403,
      "grad_norm": 1.7813162803649902,
      "learning_rate": 0.0001482893450635386,
      "loss": 1.1148,
      "step": 6719
    },
    {
      "epoch": 1.0519724483406387,
      "grad_norm": 1.9501174688339233,
      "learning_rate": 0.00014826490713587488,
      "loss": 0.7127,
      "step": 6720
    },
    {
      "epoch": 1.052128991859737,
      "grad_norm": 2.0499088764190674,
      "learning_rate": 0.00014824046920821113,
      "loss": 1.0263,
      "step": 6721
    },
    {
      "epoch": 1.0522855353788354,
      "grad_norm": 2.545484781265259,
      "learning_rate": 0.00014821603128054739,
      "loss": 0.8959,
      "step": 6722
    },
    {
      "epoch": 1.0524420788979336,
      "grad_norm": 2.9357028007507324,
      "learning_rate": 0.00014819159335288366,
      "loss": 1.346,
      "step": 6723
    },
    {
      "epoch": 1.0525986224170318,
      "grad_norm": 1.6298185586929321,
      "learning_rate": 0.00014816715542521994,
      "loss": 1.0074,
      "step": 6724
    },
    {
      "epoch": 1.0527551659361303,
      "grad_norm": 2.8814504146575928,
      "learning_rate": 0.0001481427174975562,
      "loss": 0.636,
      "step": 6725
    },
    {
      "epoch": 1.0529117094552285,
      "grad_norm": 3.723740339279175,
      "learning_rate": 0.00014811827956989247,
      "loss": 0.7553,
      "step": 6726
    },
    {
      "epoch": 1.053068252974327,
      "grad_norm": 1.4765031337738037,
      "learning_rate": 0.00014809384164222872,
      "loss": 0.9195,
      "step": 6727
    },
    {
      "epoch": 1.0532247964934252,
      "grad_norm": 1.9450255632400513,
      "learning_rate": 0.000148069403714565,
      "loss": 1.3745,
      "step": 6728
    },
    {
      "epoch": 1.0533813400125234,
      "grad_norm": 2.2991535663604736,
      "learning_rate": 0.00014804496578690125,
      "loss": 1.0113,
      "step": 6729
    },
    {
      "epoch": 1.0535378835316218,
      "grad_norm": 2.7434890270233154,
      "learning_rate": 0.00014802052785923753,
      "loss": 1.1969,
      "step": 6730
    },
    {
      "epoch": 1.05369442705072,
      "grad_norm": 2.5354650020599365,
      "learning_rate": 0.00014799608993157378,
      "loss": 1.1237,
      "step": 6731
    },
    {
      "epoch": 1.0538509705698185,
      "grad_norm": 1.4999096393585205,
      "learning_rate": 0.00014797165200391006,
      "loss": 1.2223,
      "step": 6732
    },
    {
      "epoch": 1.0540075140889167,
      "grad_norm": 2.2378246784210205,
      "learning_rate": 0.0001479472140762463,
      "loss": 1.1544,
      "step": 6733
    },
    {
      "epoch": 1.0541640576080151,
      "grad_norm": 1.6880450248718262,
      "learning_rate": 0.0001479227761485826,
      "loss": 0.7526,
      "step": 6734
    },
    {
      "epoch": 1.0543206011271133,
      "grad_norm": 5.19423246383667,
      "learning_rate": 0.00014789833822091887,
      "loss": 1.2331,
      "step": 6735
    },
    {
      "epoch": 1.0544771446462116,
      "grad_norm": 3.7082395553588867,
      "learning_rate": 0.00014787390029325512,
      "loss": 0.4293,
      "step": 6736
    },
    {
      "epoch": 1.05463368816531,
      "grad_norm": 2.9954659938812256,
      "learning_rate": 0.00014784946236559137,
      "loss": 1.0767,
      "step": 6737
    },
    {
      "epoch": 1.0547902316844082,
      "grad_norm": 1.1557728052139282,
      "learning_rate": 0.00014782502443792765,
      "loss": 0.4366,
      "step": 6738
    },
    {
      "epoch": 1.0549467752035067,
      "grad_norm": 0.6128056049346924,
      "learning_rate": 0.00014780058651026393,
      "loss": 0.2985,
      "step": 6739
    },
    {
      "epoch": 1.0551033187226049,
      "grad_norm": 0.6102817058563232,
      "learning_rate": 0.00014777614858260018,
      "loss": 0.3021,
      "step": 6740
    },
    {
      "epoch": 1.055259862241703,
      "grad_norm": 0.8309397101402283,
      "learning_rate": 0.00014775171065493646,
      "loss": 0.2417,
      "step": 6741
    },
    {
      "epoch": 1.0554164057608015,
      "grad_norm": 0.7619971632957458,
      "learning_rate": 0.0001477272727272727,
      "loss": 0.2599,
      "step": 6742
    },
    {
      "epoch": 1.0555729492798998,
      "grad_norm": 0.6904311180114746,
      "learning_rate": 0.000147702834799609,
      "loss": 0.3468,
      "step": 6743
    },
    {
      "epoch": 1.0557294927989982,
      "grad_norm": 1.885140061378479,
      "learning_rate": 0.00014767839687194526,
      "loss": 0.412,
      "step": 6744
    },
    {
      "epoch": 1.0558860363180964,
      "grad_norm": 0.6122669577598572,
      "learning_rate": 0.00014765395894428152,
      "loss": 0.2116,
      "step": 6745
    },
    {
      "epoch": 1.0560425798371949,
      "grad_norm": 0.754728376865387,
      "learning_rate": 0.00014762952101661777,
      "loss": 0.3224,
      "step": 6746
    },
    {
      "epoch": 1.056199123356293,
      "grad_norm": 2.336873769760132,
      "learning_rate": 0.00014760508308895405,
      "loss": 0.3732,
      "step": 6747
    },
    {
      "epoch": 1.0563556668753913,
      "grad_norm": 1.1394752264022827,
      "learning_rate": 0.00014758064516129032,
      "loss": 0.4202,
      "step": 6748
    },
    {
      "epoch": 1.0565122103944897,
      "grad_norm": 0.6358504891395569,
      "learning_rate": 0.00014755620723362658,
      "loss": 0.2911,
      "step": 6749
    },
    {
      "epoch": 1.056668753913588,
      "grad_norm": 0.8128040432929993,
      "learning_rate": 0.00014753176930596285,
      "loss": 0.2091,
      "step": 6750
    },
    {
      "epoch": 1.0568252974326864,
      "grad_norm": 1.233589768409729,
      "learning_rate": 0.0001475073313782991,
      "loss": 0.2296,
      "step": 6751
    },
    {
      "epoch": 1.0569818409517846,
      "grad_norm": 1.3525103330612183,
      "learning_rate": 0.00014748289345063538,
      "loss": 0.4545,
      "step": 6752
    },
    {
      "epoch": 1.0571383844708828,
      "grad_norm": 1.5476621389389038,
      "learning_rate": 0.00014745845552297163,
      "loss": 0.4035,
      "step": 6753
    },
    {
      "epoch": 1.0572949279899813,
      "grad_norm": 1.2331606149673462,
      "learning_rate": 0.0001474340175953079,
      "loss": 0.2726,
      "step": 6754
    },
    {
      "epoch": 1.0574514715090795,
      "grad_norm": 1.3619499206542969,
      "learning_rate": 0.00014740957966764416,
      "loss": 0.5273,
      "step": 6755
    },
    {
      "epoch": 1.057608015028178,
      "grad_norm": 3.523498773574829,
      "learning_rate": 0.00014738514173998044,
      "loss": 0.5354,
      "step": 6756
    },
    {
      "epoch": 1.0577645585472761,
      "grad_norm": 1.602575659751892,
      "learning_rate": 0.0001473607038123167,
      "loss": 0.6084,
      "step": 6757
    },
    {
      "epoch": 1.0579211020663744,
      "grad_norm": 2.1576313972473145,
      "learning_rate": 0.00014733626588465297,
      "loss": 0.3558,
      "step": 6758
    },
    {
      "epoch": 1.0580776455854728,
      "grad_norm": 1.1257455348968506,
      "learning_rate": 0.00014731182795698925,
      "loss": 0.4306,
      "step": 6759
    },
    {
      "epoch": 1.058234189104571,
      "grad_norm": 1.3478970527648926,
      "learning_rate": 0.0001472873900293255,
      "loss": 0.5438,
      "step": 6760
    },
    {
      "epoch": 1.0583907326236695,
      "grad_norm": 2.616065263748169,
      "learning_rate": 0.00014726295210166175,
      "loss": 0.8583,
      "step": 6761
    },
    {
      "epoch": 1.0585472761427677,
      "grad_norm": 1.401318907737732,
      "learning_rate": 0.00014723851417399803,
      "loss": 0.2882,
      "step": 6762
    },
    {
      "epoch": 1.0587038196618659,
      "grad_norm": 2.02484130859375,
      "learning_rate": 0.0001472140762463343,
      "loss": 0.9521,
      "step": 6763
    },
    {
      "epoch": 1.0588603631809643,
      "grad_norm": 1.5256311893463135,
      "learning_rate": 0.00014718963831867056,
      "loss": 0.4258,
      "step": 6764
    },
    {
      "epoch": 1.0590169067000625,
      "grad_norm": 1.3191006183624268,
      "learning_rate": 0.00014716520039100684,
      "loss": 0.4773,
      "step": 6765
    },
    {
      "epoch": 1.059173450219161,
      "grad_norm": 2.7704293727874756,
      "learning_rate": 0.0001471407624633431,
      "loss": 0.6198,
      "step": 6766
    },
    {
      "epoch": 1.0593299937382592,
      "grad_norm": 1.2429702281951904,
      "learning_rate": 0.00014711632453567937,
      "loss": 0.646,
      "step": 6767
    },
    {
      "epoch": 1.0594865372573576,
      "grad_norm": 1.98096764087677,
      "learning_rate": 0.00014709188660801565,
      "loss": 0.6263,
      "step": 6768
    },
    {
      "epoch": 1.0596430807764559,
      "grad_norm": 5.404338836669922,
      "learning_rate": 0.0001470674486803519,
      "loss": 1.4979,
      "step": 6769
    },
    {
      "epoch": 1.059799624295554,
      "grad_norm": 2.1490466594696045,
      "learning_rate": 0.00014704301075268815,
      "loss": 0.7869,
      "step": 6770
    },
    {
      "epoch": 1.0599561678146525,
      "grad_norm": 1.6337779760360718,
      "learning_rate": 0.00014701857282502443,
      "loss": 0.8592,
      "step": 6771
    },
    {
      "epoch": 1.0601127113337507,
      "grad_norm": 3.1310315132141113,
      "learning_rate": 0.0001469941348973607,
      "loss": 0.9336,
      "step": 6772
    },
    {
      "epoch": 1.0602692548528492,
      "grad_norm": 2.745962619781494,
      "learning_rate": 0.00014696969696969696,
      "loss": 1.043,
      "step": 6773
    },
    {
      "epoch": 1.0604257983719474,
      "grad_norm": 2.7593672275543213,
      "learning_rate": 0.00014694525904203324,
      "loss": 1.1677,
      "step": 6774
    },
    {
      "epoch": 1.0605823418910456,
      "grad_norm": 5.965583801269531,
      "learning_rate": 0.0001469208211143695,
      "loss": 1.777,
      "step": 6775
    },
    {
      "epoch": 1.060738885410144,
      "grad_norm": 3.3984930515289307,
      "learning_rate": 0.00014689638318670574,
      "loss": 1.0561,
      "step": 6776
    },
    {
      "epoch": 1.0608954289292423,
      "grad_norm": 2.305185317993164,
      "learning_rate": 0.00014687194525904202,
      "loss": 0.6857,
      "step": 6777
    },
    {
      "epoch": 1.0610519724483407,
      "grad_norm": 2.9663422107696533,
      "learning_rate": 0.0001468475073313783,
      "loss": 0.8298,
      "step": 6778
    },
    {
      "epoch": 1.061208515967439,
      "grad_norm": 2.1952407360076904,
      "learning_rate": 0.00014682306940371455,
      "loss": 0.9201,
      "step": 6779
    },
    {
      "epoch": 1.0613650594865374,
      "grad_norm": 2.043056011199951,
      "learning_rate": 0.0001467986314760508,
      "loss": 1.1013,
      "step": 6780
    },
    {
      "epoch": 1.0615216030056356,
      "grad_norm": 1.94023597240448,
      "learning_rate": 0.00014677419354838708,
      "loss": 1.3418,
      "step": 6781
    },
    {
      "epoch": 1.0616781465247338,
      "grad_norm": 2.6357297897338867,
      "learning_rate": 0.00014674975562072335,
      "loss": 0.8855,
      "step": 6782
    },
    {
      "epoch": 1.0618346900438322,
      "grad_norm": 2.302753448486328,
      "learning_rate": 0.0001467253176930596,
      "loss": 0.9809,
      "step": 6783
    },
    {
      "epoch": 1.0619912335629305,
      "grad_norm": 2.690652370452881,
      "learning_rate": 0.00014670087976539588,
      "loss": 0.4587,
      "step": 6784
    },
    {
      "epoch": 1.062147777082029,
      "grad_norm": 3.5731654167175293,
      "learning_rate": 0.00014667644183773214,
      "loss": 1.0801,
      "step": 6785
    },
    {
      "epoch": 1.0623043206011271,
      "grad_norm": 2.150663375854492,
      "learning_rate": 0.00014665200391006841,
      "loss": 0.7417,
      "step": 6786
    },
    {
      "epoch": 1.0624608641202253,
      "grad_norm": 2.761242151260376,
      "learning_rate": 0.0001466275659824047,
      "loss": 1.2934,
      "step": 6787
    },
    {
      "epoch": 1.0626174076393238,
      "grad_norm": 2.148921251296997,
      "learning_rate": 0.00014660312805474094,
      "loss": 0.5649,
      "step": 6788
    },
    {
      "epoch": 1.062773951158422,
      "grad_norm": 0.691116452217102,
      "learning_rate": 0.0001465786901270772,
      "loss": 0.287,
      "step": 6789
    },
    {
      "epoch": 1.0629304946775204,
      "grad_norm": 0.4150255024433136,
      "learning_rate": 0.00014655425219941347,
      "loss": 0.3013,
      "step": 6790
    },
    {
      "epoch": 1.0630870381966186,
      "grad_norm": 0.48139479756355286,
      "learning_rate": 0.00014652981427174975,
      "loss": 0.3082,
      "step": 6791
    },
    {
      "epoch": 1.0632435817157169,
      "grad_norm": 0.808285117149353,
      "learning_rate": 0.000146505376344086,
      "loss": 0.2616,
      "step": 6792
    },
    {
      "epoch": 1.0634001252348153,
      "grad_norm": 0.7357711791992188,
      "learning_rate": 0.00014648093841642228,
      "loss": 0.2752,
      "step": 6793
    },
    {
      "epoch": 1.0635566687539135,
      "grad_norm": 0.6319572329521179,
      "learning_rate": 0.00014645650048875853,
      "loss": 0.26,
      "step": 6794
    },
    {
      "epoch": 1.063713212273012,
      "grad_norm": 0.698928952217102,
      "learning_rate": 0.0001464320625610948,
      "loss": 0.2896,
      "step": 6795
    },
    {
      "epoch": 1.0638697557921102,
      "grad_norm": 1.5302014350891113,
      "learning_rate": 0.0001464076246334311,
      "loss": 0.3453,
      "step": 6796
    },
    {
      "epoch": 1.0640262993112084,
      "grad_norm": 1.11781907081604,
      "learning_rate": 0.00014638318670576734,
      "loss": 0.3171,
      "step": 6797
    },
    {
      "epoch": 1.0641828428303068,
      "grad_norm": 1.107818365097046,
      "learning_rate": 0.0001463587487781036,
      "loss": 0.2568,
      "step": 6798
    },
    {
      "epoch": 1.064339386349405,
      "grad_norm": 1.2923407554626465,
      "learning_rate": 0.00014633431085043987,
      "loss": 0.2953,
      "step": 6799
    },
    {
      "epoch": 1.0644959298685035,
      "grad_norm": 1.0053560733795166,
      "learning_rate": 0.00014630987292277612,
      "loss": 0.4314,
      "step": 6800
    },
    {
      "epoch": 1.0646524733876017,
      "grad_norm": 0.73367840051651,
      "learning_rate": 0.0001462854349951124,
      "loss": 0.1903,
      "step": 6801
    },
    {
      "epoch": 1.0648090169067002,
      "grad_norm": 0.6966777443885803,
      "learning_rate": 0.00014626099706744868,
      "loss": 0.2807,
      "step": 6802
    },
    {
      "epoch": 1.0649655604257984,
      "grad_norm": 1.2056246995925903,
      "learning_rate": 0.00014623655913978493,
      "loss": 0.5498,
      "step": 6803
    },
    {
      "epoch": 1.0651221039448966,
      "grad_norm": 1.8048405647277832,
      "learning_rate": 0.00014621212121212118,
      "loss": 0.3544,
      "step": 6804
    },
    {
      "epoch": 1.065278647463995,
      "grad_norm": 1.2983676195144653,
      "learning_rate": 0.00014618768328445746,
      "loss": 0.3607,
      "step": 6805
    },
    {
      "epoch": 1.0654351909830932,
      "grad_norm": 0.5953453183174133,
      "learning_rate": 0.00014616324535679374,
      "loss": 0.2121,
      "step": 6806
    },
    {
      "epoch": 1.0655917345021917,
      "grad_norm": 1.4588236808776855,
      "learning_rate": 0.00014613880742913,
      "loss": 0.7027,
      "step": 6807
    },
    {
      "epoch": 1.06574827802129,
      "grad_norm": 1.0992096662521362,
      "learning_rate": 0.00014611436950146627,
      "loss": 0.4823,
      "step": 6808
    },
    {
      "epoch": 1.0659048215403881,
      "grad_norm": 1.8514833450317383,
      "learning_rate": 0.00014608993157380252,
      "loss": 0.5958,
      "step": 6809
    },
    {
      "epoch": 1.0660613650594866,
      "grad_norm": 1.5322425365447998,
      "learning_rate": 0.0001460654936461388,
      "loss": 0.495,
      "step": 6810
    },
    {
      "epoch": 1.0662179085785848,
      "grad_norm": 0.9982278347015381,
      "learning_rate": 0.00014604105571847507,
      "loss": 0.3159,
      "step": 6811
    },
    {
      "epoch": 1.0663744520976832,
      "grad_norm": 1.4766974449157715,
      "learning_rate": 0.00014601661779081133,
      "loss": 0.5013,
      "step": 6812
    },
    {
      "epoch": 1.0665309956167814,
      "grad_norm": 1.6511269807815552,
      "learning_rate": 0.00014599217986314758,
      "loss": 0.5875,
      "step": 6813
    },
    {
      "epoch": 1.0666875391358799,
      "grad_norm": 1.9279791116714478,
      "learning_rate": 0.00014596774193548386,
      "loss": 0.7228,
      "step": 6814
    },
    {
      "epoch": 1.066844082654978,
      "grad_norm": 2.1245806217193604,
      "learning_rate": 0.00014594330400782013,
      "loss": 0.6166,
      "step": 6815
    },
    {
      "epoch": 1.0670006261740763,
      "grad_norm": 2.9913034439086914,
      "learning_rate": 0.00014591886608015638,
      "loss": 0.6317,
      "step": 6816
    },
    {
      "epoch": 1.0671571696931748,
      "grad_norm": 1.126991868019104,
      "learning_rate": 0.00014589442815249266,
      "loss": 0.352,
      "step": 6817
    },
    {
      "epoch": 1.067313713212273,
      "grad_norm": 1.3619285821914673,
      "learning_rate": 0.00014586999022482891,
      "loss": 0.5466,
      "step": 6818
    },
    {
      "epoch": 1.0674702567313714,
      "grad_norm": 1.5446709394454956,
      "learning_rate": 0.0001458455522971652,
      "loss": 0.7798,
      "step": 6819
    },
    {
      "epoch": 1.0676268002504696,
      "grad_norm": 1.6549172401428223,
      "learning_rate": 0.00014582111436950144,
      "loss": 0.7226,
      "step": 6820
    },
    {
      "epoch": 1.0677833437695678,
      "grad_norm": 1.4883651733398438,
      "learning_rate": 0.00014579667644183772,
      "loss": 0.8854,
      "step": 6821
    },
    {
      "epoch": 1.0679398872886663,
      "grad_norm": 1.6428464651107788,
      "learning_rate": 0.00014577223851417397,
      "loss": 0.9569,
      "step": 6822
    },
    {
      "epoch": 1.0680964308077645,
      "grad_norm": 2.9931986331939697,
      "learning_rate": 0.00014574780058651025,
      "loss": 1.1195,
      "step": 6823
    },
    {
      "epoch": 1.068252974326863,
      "grad_norm": 1.3941583633422852,
      "learning_rate": 0.0001457233626588465,
      "loss": 0.6068,
      "step": 6824
    },
    {
      "epoch": 1.0684095178459612,
      "grad_norm": 1.757436752319336,
      "learning_rate": 0.00014569892473118278,
      "loss": 0.5655,
      "step": 6825
    },
    {
      "epoch": 1.0685660613650594,
      "grad_norm": 2.191127300262451,
      "learning_rate": 0.00014567448680351906,
      "loss": 0.8291,
      "step": 6826
    },
    {
      "epoch": 1.0687226048841578,
      "grad_norm": 3.134359121322632,
      "learning_rate": 0.0001456500488758553,
      "loss": 0.9857,
      "step": 6827
    },
    {
      "epoch": 1.068879148403256,
      "grad_norm": 1.7476389408111572,
      "learning_rate": 0.00014562561094819156,
      "loss": 0.7704,
      "step": 6828
    },
    {
      "epoch": 1.0690356919223545,
      "grad_norm": 2.1934890747070312,
      "learning_rate": 0.00014560117302052784,
      "loss": 1.404,
      "step": 6829
    },
    {
      "epoch": 1.0691922354414527,
      "grad_norm": 3.2194011211395264,
      "learning_rate": 0.00014557673509286412,
      "loss": 0.8811,
      "step": 6830
    },
    {
      "epoch": 1.069348778960551,
      "grad_norm": 2.342510461807251,
      "learning_rate": 0.00014555229716520037,
      "loss": 0.781,
      "step": 6831
    },
    {
      "epoch": 1.0695053224796494,
      "grad_norm": 2.317560911178589,
      "learning_rate": 0.00014552785923753665,
      "loss": 0.8468,
      "step": 6832
    },
    {
      "epoch": 1.0696618659987476,
      "grad_norm": 2.0507400035858154,
      "learning_rate": 0.0001455034213098729,
      "loss": 0.8117,
      "step": 6833
    },
    {
      "epoch": 1.069818409517846,
      "grad_norm": 2.625714063644409,
      "learning_rate": 0.00014547898338220918,
      "loss": 0.7835,
      "step": 6834
    },
    {
      "epoch": 1.0699749530369442,
      "grad_norm": 2.597598075866699,
      "learning_rate": 0.00014545454545454546,
      "loss": 0.5083,
      "step": 6835
    },
    {
      "epoch": 1.0701314965560427,
      "grad_norm": 2.5991411209106445,
      "learning_rate": 0.0001454301075268817,
      "loss": 1.1845,
      "step": 6836
    },
    {
      "epoch": 1.070288040075141,
      "grad_norm": 1.9454537630081177,
      "learning_rate": 0.00014540566959921796,
      "loss": 0.414,
      "step": 6837
    },
    {
      "epoch": 1.070444583594239,
      "grad_norm": 1.6982367038726807,
      "learning_rate": 0.00014538123167155424,
      "loss": 0.6621,
      "step": 6838
    },
    {
      "epoch": 1.0706011271133375,
      "grad_norm": 0.6477040648460388,
      "learning_rate": 0.00014535679374389052,
      "loss": 0.3462,
      "step": 6839
    },
    {
      "epoch": 1.0707576706324358,
      "grad_norm": 0.3390127420425415,
      "learning_rate": 0.00014533235581622677,
      "loss": 0.1706,
      "step": 6840
    },
    {
      "epoch": 1.0709142141515342,
      "grad_norm": 0.8520225882530212,
      "learning_rate": 0.00014530791788856305,
      "loss": 0.3586,
      "step": 6841
    },
    {
      "epoch": 1.0710707576706324,
      "grad_norm": 0.6063798069953918,
      "learning_rate": 0.0001452834799608993,
      "loss": 0.1433,
      "step": 6842
    },
    {
      "epoch": 1.0712273011897309,
      "grad_norm": 0.7696285247802734,
      "learning_rate": 0.00014525904203323557,
      "loss": 0.5062,
      "step": 6843
    },
    {
      "epoch": 1.071383844708829,
      "grad_norm": 0.47626444697380066,
      "learning_rate": 0.00014523460410557183,
      "loss": 0.2092,
      "step": 6844
    },
    {
      "epoch": 1.0715403882279273,
      "grad_norm": 0.7897009253501892,
      "learning_rate": 0.0001452101661779081,
      "loss": 0.295,
      "step": 6845
    },
    {
      "epoch": 1.0716969317470257,
      "grad_norm": 0.6091282963752747,
      "learning_rate": 0.00014518572825024436,
      "loss": 0.2313,
      "step": 6846
    },
    {
      "epoch": 1.071853475266124,
      "grad_norm": 0.9005686044692993,
      "learning_rate": 0.00014516129032258063,
      "loss": 0.2965,
      "step": 6847
    },
    {
      "epoch": 1.0720100187852224,
      "grad_norm": 0.9107126593589783,
      "learning_rate": 0.00014513685239491689,
      "loss": 0.2936,
      "step": 6848
    },
    {
      "epoch": 1.0721665623043206,
      "grad_norm": 0.5607098340988159,
      "learning_rate": 0.00014511241446725316,
      "loss": 0.2276,
      "step": 6849
    },
    {
      "epoch": 1.0723231058234188,
      "grad_norm": 13.283991813659668,
      "learning_rate": 0.00014508797653958944,
      "loss": 0.3193,
      "step": 6850
    },
    {
      "epoch": 1.0724796493425173,
      "grad_norm": 1.3404076099395752,
      "learning_rate": 0.0001450635386119257,
      "loss": 0.422,
      "step": 6851
    },
    {
      "epoch": 1.0726361928616155,
      "grad_norm": 0.9615288376808167,
      "learning_rate": 0.00014503910068426194,
      "loss": 0.4161,
      "step": 6852
    },
    {
      "epoch": 1.072792736380714,
      "grad_norm": 1.6815721988677979,
      "learning_rate": 0.00014501466275659822,
      "loss": 0.3811,
      "step": 6853
    },
    {
      "epoch": 1.0729492798998121,
      "grad_norm": 0.926059901714325,
      "learning_rate": 0.0001449902248289345,
      "loss": 0.5096,
      "step": 6854
    },
    {
      "epoch": 1.0731058234189104,
      "grad_norm": 0.8995077610015869,
      "learning_rate": 0.00014496578690127075,
      "loss": 0.4286,
      "step": 6855
    },
    {
      "epoch": 1.0732623669380088,
      "grad_norm": 0.9982707500457764,
      "learning_rate": 0.00014494134897360703,
      "loss": 0.5821,
      "step": 6856
    },
    {
      "epoch": 1.073418910457107,
      "grad_norm": 1.031527042388916,
      "learning_rate": 0.00014491691104594328,
      "loss": 0.3978,
      "step": 6857
    },
    {
      "epoch": 1.0735754539762055,
      "grad_norm": 0.7758069038391113,
      "learning_rate": 0.00014489247311827956,
      "loss": 0.3313,
      "step": 6858
    },
    {
      "epoch": 1.0737319974953037,
      "grad_norm": 1.3925272226333618,
      "learning_rate": 0.00014486803519061584,
      "loss": 0.2895,
      "step": 6859
    },
    {
      "epoch": 1.073888541014402,
      "grad_norm": 1.40436851978302,
      "learning_rate": 0.0001448435972629521,
      "loss": 0.4096,
      "step": 6860
    },
    {
      "epoch": 1.0740450845335003,
      "grad_norm": 1.4296326637268066,
      "learning_rate": 0.00014481915933528834,
      "loss": 0.4963,
      "step": 6861
    },
    {
      "epoch": 1.0742016280525986,
      "grad_norm": 1.4773507118225098,
      "learning_rate": 0.00014479472140762462,
      "loss": 0.4817,
      "step": 6862
    },
    {
      "epoch": 1.074358171571697,
      "grad_norm": 1.4051315784454346,
      "learning_rate": 0.0001447702834799609,
      "loss": 0.6741,
      "step": 6863
    },
    {
      "epoch": 1.0745147150907952,
      "grad_norm": 1.041042447090149,
      "learning_rate": 0.00014474584555229715,
      "loss": 0.2722,
      "step": 6864
    },
    {
      "epoch": 1.0746712586098937,
      "grad_norm": 2.3967373371124268,
      "learning_rate": 0.00014472140762463343,
      "loss": 0.764,
      "step": 6865
    },
    {
      "epoch": 1.0748278021289919,
      "grad_norm": 4.08719539642334,
      "learning_rate": 0.00014469696969696968,
      "loss": 0.4523,
      "step": 6866
    },
    {
      "epoch": 1.07498434564809,
      "grad_norm": 1.987392783164978,
      "learning_rate": 0.00014467253176930596,
      "loss": 0.5414,
      "step": 6867
    },
    {
      "epoch": 1.0751408891671885,
      "grad_norm": 3.570604085922241,
      "learning_rate": 0.0001446480938416422,
      "loss": 0.8434,
      "step": 6868
    },
    {
      "epoch": 1.0752974326862867,
      "grad_norm": 7.099319934844971,
      "learning_rate": 0.0001446236559139785,
      "loss": 0.7103,
      "step": 6869
    },
    {
      "epoch": 1.0754539762053852,
      "grad_norm": 3.188019037246704,
      "learning_rate": 0.00014459921798631474,
      "loss": 0.5604,
      "step": 6870
    },
    {
      "epoch": 1.0756105197244834,
      "grad_norm": 1.420037865638733,
      "learning_rate": 0.00014457478005865102,
      "loss": 0.5358,
      "step": 6871
    },
    {
      "epoch": 1.0757670632435816,
      "grad_norm": 1.5740822553634644,
      "learning_rate": 0.00014455034213098727,
      "loss": 0.6757,
      "step": 6872
    },
    {
      "epoch": 1.07592360676268,
      "grad_norm": 2.2425143718719482,
      "learning_rate": 0.00014452590420332355,
      "loss": 1.1542,
      "step": 6873
    },
    {
      "epoch": 1.0760801502817783,
      "grad_norm": 3.6810855865478516,
      "learning_rate": 0.00014450146627565982,
      "loss": 0.8133,
      "step": 6874
    },
    {
      "epoch": 1.0762366938008767,
      "grad_norm": 2.337358236312866,
      "learning_rate": 0.00014447702834799608,
      "loss": 0.8714,
      "step": 6875
    },
    {
      "epoch": 1.076393237319975,
      "grad_norm": 3.2725279331207275,
      "learning_rate": 0.00014445259042033233,
      "loss": 0.9102,
      "step": 6876
    },
    {
      "epoch": 1.0765497808390734,
      "grad_norm": 5.1807355880737305,
      "learning_rate": 0.0001444281524926686,
      "loss": 1.4275,
      "step": 6877
    },
    {
      "epoch": 1.0767063243581716,
      "grad_norm": 3.2198193073272705,
      "learning_rate": 0.00014440371456500488,
      "loss": 0.6962,
      "step": 6878
    },
    {
      "epoch": 1.0768628678772698,
      "grad_norm": 5.173696041107178,
      "learning_rate": 0.00014437927663734113,
      "loss": 1.4533,
      "step": 6879
    },
    {
      "epoch": 1.0770194113963683,
      "grad_norm": 2.3174338340759277,
      "learning_rate": 0.0001443548387096774,
      "loss": 1.139,
      "step": 6880
    },
    {
      "epoch": 1.0771759549154665,
      "grad_norm": 2.363914966583252,
      "learning_rate": 0.00014433040078201366,
      "loss": 1.0105,
      "step": 6881
    },
    {
      "epoch": 1.077332498434565,
      "grad_norm": 2.0799217224121094,
      "learning_rate": 0.00014430596285434994,
      "loss": 0.8286,
      "step": 6882
    },
    {
      "epoch": 1.0774890419536631,
      "grad_norm": 2.2222182750701904,
      "learning_rate": 0.00014428152492668622,
      "loss": 0.6723,
      "step": 6883
    },
    {
      "epoch": 1.0776455854727613,
      "grad_norm": 2.70257568359375,
      "learning_rate": 0.00014425708699902247,
      "loss": 0.8287,
      "step": 6884
    },
    {
      "epoch": 1.0778021289918598,
      "grad_norm": 2.3015530109405518,
      "learning_rate": 0.00014423264907135872,
      "loss": 0.3816,
      "step": 6885
    },
    {
      "epoch": 1.077958672510958,
      "grad_norm": 2.944084644317627,
      "learning_rate": 0.000144208211143695,
      "loss": 0.9525,
      "step": 6886
    },
    {
      "epoch": 1.0781152160300564,
      "grad_norm": 2.2739789485931396,
      "learning_rate": 0.00014418377321603128,
      "loss": 0.489,
      "step": 6887
    },
    {
      "epoch": 1.0782717595491547,
      "grad_norm": 6.7248969078063965,
      "learning_rate": 0.00014415933528836753,
      "loss": 0.9746,
      "step": 6888
    },
    {
      "epoch": 1.0784283030682529,
      "grad_norm": 0.4254947006702423,
      "learning_rate": 0.0001441348973607038,
      "loss": 0.2798,
      "step": 6889
    },
    {
      "epoch": 1.0785848465873513,
      "grad_norm": 0.3661378026008606,
      "learning_rate": 0.00014411045943304006,
      "loss": 0.1879,
      "step": 6890
    },
    {
      "epoch": 1.0787413901064495,
      "grad_norm": 0.5312639474868774,
      "learning_rate": 0.0001440860215053763,
      "loss": 0.1684,
      "step": 6891
    },
    {
      "epoch": 1.078897933625548,
      "grad_norm": 0.9988065361976624,
      "learning_rate": 0.0001440615835777126,
      "loss": 0.2838,
      "step": 6892
    },
    {
      "epoch": 1.0790544771446462,
      "grad_norm": 0.5143894553184509,
      "learning_rate": 0.00014403714565004887,
      "loss": 0.2015,
      "step": 6893
    },
    {
      "epoch": 1.0792110206637444,
      "grad_norm": 0.6396801471710205,
      "learning_rate": 0.00014401270772238512,
      "loss": 0.2814,
      "step": 6894
    },
    {
      "epoch": 1.0793675641828429,
      "grad_norm": 0.8477953672409058,
      "learning_rate": 0.0001439882697947214,
      "loss": 0.3326,
      "step": 6895
    },
    {
      "epoch": 1.079524107701941,
      "grad_norm": 0.47463440895080566,
      "learning_rate": 0.00014396383186705765,
      "loss": 0.1656,
      "step": 6896
    },
    {
      "epoch": 1.0796806512210395,
      "grad_norm": 0.7857781052589417,
      "learning_rate": 0.00014393939393939393,
      "loss": 0.2643,
      "step": 6897
    },
    {
      "epoch": 1.0798371947401377,
      "grad_norm": 0.7944388389587402,
      "learning_rate": 0.0001439149560117302,
      "loss": 0.3092,
      "step": 6898
    },
    {
      "epoch": 1.0799937382592362,
      "grad_norm": 1.1828808784484863,
      "learning_rate": 0.00014389051808406646,
      "loss": 0.2997,
      "step": 6899
    },
    {
      "epoch": 1.0801502817783344,
      "grad_norm": 0.6055023074150085,
      "learning_rate": 0.0001438660801564027,
      "loss": 0.2784,
      "step": 6900
    },
    {
      "epoch": 1.0803068252974326,
      "grad_norm": 0.6134934425354004,
      "learning_rate": 0.000143841642228739,
      "loss": 0.3366,
      "step": 6901
    },
    {
      "epoch": 1.080463368816531,
      "grad_norm": 1.3209537267684937,
      "learning_rate": 0.00014381720430107527,
      "loss": 0.5705,
      "step": 6902
    },
    {
      "epoch": 1.0806199123356293,
      "grad_norm": 0.7764045000076294,
      "learning_rate": 0.00014379276637341152,
      "loss": 0.3105,
      "step": 6903
    },
    {
      "epoch": 1.0807764558547277,
      "grad_norm": 1.349913477897644,
      "learning_rate": 0.0001437683284457478,
      "loss": 0.2681,
      "step": 6904
    },
    {
      "epoch": 1.080932999373826,
      "grad_norm": 0.920647919178009,
      "learning_rate": 0.00014374389051808405,
      "loss": 0.3003,
      "step": 6905
    },
    {
      "epoch": 1.0810895428929241,
      "grad_norm": 1.2609316110610962,
      "learning_rate": 0.00014371945259042033,
      "loss": 0.3485,
      "step": 6906
    },
    {
      "epoch": 1.0812460864120226,
      "grad_norm": 1.3429254293441772,
      "learning_rate": 0.0001436950146627566,
      "loss": 0.4673,
      "step": 6907
    },
    {
      "epoch": 1.0814026299311208,
      "grad_norm": 1.207157850265503,
      "learning_rate": 0.00014367057673509285,
      "loss": 0.4167,
      "step": 6908
    },
    {
      "epoch": 1.0815591734502192,
      "grad_norm": 0.8418388366699219,
      "learning_rate": 0.0001436461388074291,
      "loss": 0.4355,
      "step": 6909
    },
    {
      "epoch": 1.0817157169693175,
      "grad_norm": 1.242262363433838,
      "learning_rate": 0.00014362170087976538,
      "loss": 0.627,
      "step": 6910
    },
    {
      "epoch": 1.081872260488416,
      "grad_norm": 2.378530979156494,
      "learning_rate": 0.00014359726295210166,
      "loss": 0.9752,
      "step": 6911
    },
    {
      "epoch": 1.0820288040075141,
      "grad_norm": 2.7604053020477295,
      "learning_rate": 0.00014357282502443791,
      "loss": 0.8568,
      "step": 6912
    },
    {
      "epoch": 1.0821853475266123,
      "grad_norm": 1.2572295665740967,
      "learning_rate": 0.0001435483870967742,
      "loss": 0.6182,
      "step": 6913
    },
    {
      "epoch": 1.0823418910457108,
      "grad_norm": 1.6974389553070068,
      "learning_rate": 0.00014352394916911044,
      "loss": 0.6046,
      "step": 6914
    },
    {
      "epoch": 1.082498434564809,
      "grad_norm": 2.8349547386169434,
      "learning_rate": 0.0001434995112414467,
      "loss": 0.7052,
      "step": 6915
    },
    {
      "epoch": 1.0826549780839074,
      "grad_norm": 3.3643503189086914,
      "learning_rate": 0.00014347507331378297,
      "loss": 0.8496,
      "step": 6916
    },
    {
      "epoch": 1.0828115216030056,
      "grad_norm": 2.5238869190216064,
      "learning_rate": 0.00014345063538611925,
      "loss": 0.9502,
      "step": 6917
    },
    {
      "epoch": 1.0829680651221039,
      "grad_norm": 2.205315113067627,
      "learning_rate": 0.0001434261974584555,
      "loss": 1.0775,
      "step": 6918
    },
    {
      "epoch": 1.0831246086412023,
      "grad_norm": 5.002822399139404,
      "learning_rate": 0.00014340175953079178,
      "loss": 0.9507,
      "step": 6919
    },
    {
      "epoch": 1.0832811521603005,
      "grad_norm": 3.2781922817230225,
      "learning_rate": 0.00014337732160312803,
      "loss": 0.8607,
      "step": 6920
    },
    {
      "epoch": 1.083437695679399,
      "grad_norm": 1.792444109916687,
      "learning_rate": 0.0001433528836754643,
      "loss": 0.9261,
      "step": 6921
    },
    {
      "epoch": 1.0835942391984972,
      "grad_norm": 2.4402613639831543,
      "learning_rate": 0.0001433284457478006,
      "loss": 0.9687,
      "step": 6922
    },
    {
      "epoch": 1.0837507827175954,
      "grad_norm": 2.88875412940979,
      "learning_rate": 0.00014330400782013684,
      "loss": 1.0102,
      "step": 6923
    },
    {
      "epoch": 1.0839073262366938,
      "grad_norm": 2.4586126804351807,
      "learning_rate": 0.0001432795698924731,
      "loss": 0.962,
      "step": 6924
    },
    {
      "epoch": 1.084063869755792,
      "grad_norm": 2.9678328037261963,
      "learning_rate": 0.00014325513196480937,
      "loss": 1.0995,
      "step": 6925
    },
    {
      "epoch": 1.0842204132748905,
      "grad_norm": 2.3805899620056152,
      "learning_rate": 0.00014323069403714565,
      "loss": 0.9321,
      "step": 6926
    },
    {
      "epoch": 1.0843769567939887,
      "grad_norm": 2.5837199687957764,
      "learning_rate": 0.0001432062561094819,
      "loss": 0.7621,
      "step": 6927
    },
    {
      "epoch": 1.084533500313087,
      "grad_norm": 2.366360664367676,
      "learning_rate": 0.00014318181818181818,
      "loss": 0.9827,
      "step": 6928
    },
    {
      "epoch": 1.0846900438321854,
      "grad_norm": 3.093599796295166,
      "learning_rate": 0.00014315738025415443,
      "loss": 1.1201,
      "step": 6929
    },
    {
      "epoch": 1.0848465873512836,
      "grad_norm": 3.7076659202575684,
      "learning_rate": 0.0001431329423264907,
      "loss": 1.1055,
      "step": 6930
    },
    {
      "epoch": 1.085003130870382,
      "grad_norm": 3.0820627212524414,
      "learning_rate": 0.00014310850439882699,
      "loss": 1.3536,
      "step": 6931
    },
    {
      "epoch": 1.0851596743894802,
      "grad_norm": 8.682129859924316,
      "learning_rate": 0.00014308406647116324,
      "loss": 1.7647,
      "step": 6932
    },
    {
      "epoch": 1.0853162179085787,
      "grad_norm": 2.9856128692626953,
      "learning_rate": 0.0001430596285434995,
      "loss": 1.6023,
      "step": 6933
    },
    {
      "epoch": 1.085472761427677,
      "grad_norm": 2.074152708053589,
      "learning_rate": 0.00014303519061583577,
      "loss": 0.2357,
      "step": 6934
    },
    {
      "epoch": 1.0856293049467751,
      "grad_norm": 3.0086069107055664,
      "learning_rate": 0.00014301075268817202,
      "loss": 0.9179,
      "step": 6935
    },
    {
      "epoch": 1.0857858484658736,
      "grad_norm": 4.519247531890869,
      "learning_rate": 0.0001429863147605083,
      "loss": 1.0155,
      "step": 6936
    },
    {
      "epoch": 1.0859423919849718,
      "grad_norm": 1.4745041131973267,
      "learning_rate": 0.00014296187683284457,
      "loss": 0.6405,
      "step": 6937
    },
    {
      "epoch": 1.0860989355040702,
      "grad_norm": 3.711125135421753,
      "learning_rate": 0.00014293743890518083,
      "loss": 0.9687,
      "step": 6938
    },
    {
      "epoch": 1.0862554790231684,
      "grad_norm": 0.41256627440452576,
      "learning_rate": 0.00014291300097751708,
      "loss": 0.2144,
      "step": 6939
    },
    {
      "epoch": 1.0864120225422667,
      "grad_norm": 0.39318832755088806,
      "learning_rate": 0.00014288856304985336,
      "loss": 0.2536,
      "step": 6940
    },
    {
      "epoch": 1.086568566061365,
      "grad_norm": 0.5901644229888916,
      "learning_rate": 0.00014286412512218963,
      "loss": 0.2521,
      "step": 6941
    },
    {
      "epoch": 1.0867251095804633,
      "grad_norm": 0.5370203852653503,
      "learning_rate": 0.00014283968719452589,
      "loss": 0.2123,
      "step": 6942
    },
    {
      "epoch": 1.0868816530995618,
      "grad_norm": 0.546731173992157,
      "learning_rate": 0.00014281524926686216,
      "loss": 0.3049,
      "step": 6943
    },
    {
      "epoch": 1.08703819661866,
      "grad_norm": 0.36624133586883545,
      "learning_rate": 0.00014279081133919841,
      "loss": 0.1984,
      "step": 6944
    },
    {
      "epoch": 1.0871947401377584,
      "grad_norm": 0.6180028319358826,
      "learning_rate": 0.0001427663734115347,
      "loss": 0.2255,
      "step": 6945
    },
    {
      "epoch": 1.0873512836568566,
      "grad_norm": 0.9784983396530151,
      "learning_rate": 0.00014274193548387097,
      "loss": 0.4502,
      "step": 6946
    },
    {
      "epoch": 1.0875078271759548,
      "grad_norm": 0.5127384066581726,
      "learning_rate": 0.00014271749755620722,
      "loss": 0.27,
      "step": 6947
    },
    {
      "epoch": 1.0876643706950533,
      "grad_norm": 0.6251503825187683,
      "learning_rate": 0.00014269305962854347,
      "loss": 0.2674,
      "step": 6948
    },
    {
      "epoch": 1.0878209142141515,
      "grad_norm": 0.6730063557624817,
      "learning_rate": 0.00014266862170087975,
      "loss": 0.31,
      "step": 6949
    },
    {
      "epoch": 1.08797745773325,
      "grad_norm": 0.7769110798835754,
      "learning_rate": 0.00014264418377321603,
      "loss": 0.302,
      "step": 6950
    },
    {
      "epoch": 1.0881340012523482,
      "grad_norm": 1.0139259099960327,
      "learning_rate": 0.00014261974584555228,
      "loss": 0.3961,
      "step": 6951
    },
    {
      "epoch": 1.0882905447714464,
      "grad_norm": 0.7795376777648926,
      "learning_rate": 0.00014259530791788856,
      "loss": 0.2495,
      "step": 6952
    },
    {
      "epoch": 1.0884470882905448,
      "grad_norm": 0.770647406578064,
      "learning_rate": 0.0001425708699902248,
      "loss": 0.4041,
      "step": 6953
    },
    {
      "epoch": 1.088603631809643,
      "grad_norm": 0.9551957845687866,
      "learning_rate": 0.0001425464320625611,
      "loss": 0.2077,
      "step": 6954
    },
    {
      "epoch": 1.0887601753287415,
      "grad_norm": 0.8338924646377563,
      "learning_rate": 0.00014252199413489737,
      "loss": 0.4206,
      "step": 6955
    },
    {
      "epoch": 1.0889167188478397,
      "grad_norm": 0.7971686720848083,
      "learning_rate": 0.00014249755620723362,
      "loss": 0.31,
      "step": 6956
    },
    {
      "epoch": 1.089073262366938,
      "grad_norm": 1.553481101989746,
      "learning_rate": 0.00014247311827956987,
      "loss": 0.6022,
      "step": 6957
    },
    {
      "epoch": 1.0892298058860364,
      "grad_norm": 1.3536430597305298,
      "learning_rate": 0.00014244868035190615,
      "loss": 0.4746,
      "step": 6958
    },
    {
      "epoch": 1.0893863494051346,
      "grad_norm": 0.8568630814552307,
      "learning_rate": 0.0001424242424242424,
      "loss": 0.6675,
      "step": 6959
    },
    {
      "epoch": 1.089542892924233,
      "grad_norm": 0.9841766357421875,
      "learning_rate": 0.00014239980449657868,
      "loss": 0.56,
      "step": 6960
    },
    {
      "epoch": 1.0896994364433312,
      "grad_norm": 1.8160009384155273,
      "learning_rate": 0.00014237536656891496,
      "loss": 0.6362,
      "step": 6961
    },
    {
      "epoch": 1.0898559799624294,
      "grad_norm": 1.0871392488479614,
      "learning_rate": 0.0001423509286412512,
      "loss": 0.387,
      "step": 6962
    },
    {
      "epoch": 1.0900125234815279,
      "grad_norm": 2.788846731185913,
      "learning_rate": 0.00014232649071358746,
      "loss": 0.5088,
      "step": 6963
    },
    {
      "epoch": 1.090169067000626,
      "grad_norm": 2.411051034927368,
      "learning_rate": 0.00014230205278592374,
      "loss": 0.6828,
      "step": 6964
    },
    {
      "epoch": 1.0903256105197245,
      "grad_norm": 1.6987364292144775,
      "learning_rate": 0.00014227761485826002,
      "loss": 0.7937,
      "step": 6965
    },
    {
      "epoch": 1.0904821540388228,
      "grad_norm": 1.3906645774841309,
      "learning_rate": 0.00014225317693059627,
      "loss": 0.3764,
      "step": 6966
    },
    {
      "epoch": 1.0906386975579212,
      "grad_norm": 1.8853800296783447,
      "learning_rate": 0.00014222873900293255,
      "loss": 0.4234,
      "step": 6967
    },
    {
      "epoch": 1.0907952410770194,
      "grad_norm": 4.4653639793396,
      "learning_rate": 0.0001422043010752688,
      "loss": 1.001,
      "step": 6968
    },
    {
      "epoch": 1.0909517845961176,
      "grad_norm": 1.9520578384399414,
      "learning_rate": 0.00014217986314760508,
      "loss": 0.6155,
      "step": 6969
    },
    {
      "epoch": 1.091108328115216,
      "grad_norm": 2.5212900638580322,
      "learning_rate": 0.00014215542521994135,
      "loss": 0.5802,
      "step": 6970
    },
    {
      "epoch": 1.0912648716343143,
      "grad_norm": 3.541121006011963,
      "learning_rate": 0.0001421309872922776,
      "loss": 0.819,
      "step": 6971
    },
    {
      "epoch": 1.0914214151534127,
      "grad_norm": 4.24415922164917,
      "learning_rate": 0.00014210654936461386,
      "loss": 1.0289,
      "step": 6972
    },
    {
      "epoch": 1.091577958672511,
      "grad_norm": 2.449824094772339,
      "learning_rate": 0.00014208211143695013,
      "loss": 1.1641,
      "step": 6973
    },
    {
      "epoch": 1.0917345021916092,
      "grad_norm": 1.3512892723083496,
      "learning_rate": 0.0001420576735092864,
      "loss": 0.4655,
      "step": 6974
    },
    {
      "epoch": 1.0918910457107076,
      "grad_norm": 3.9181694984436035,
      "learning_rate": 0.00014203323558162266,
      "loss": 1.4388,
      "step": 6975
    },
    {
      "epoch": 1.0920475892298058,
      "grad_norm": 3.194889545440674,
      "learning_rate": 0.00014200879765395894,
      "loss": 1.5037,
      "step": 6976
    },
    {
      "epoch": 1.0922041327489043,
      "grad_norm": 3.483785629272461,
      "learning_rate": 0.0001419843597262952,
      "loss": 0.8793,
      "step": 6977
    },
    {
      "epoch": 1.0923606762680025,
      "grad_norm": 2.2053303718566895,
      "learning_rate": 0.00014195992179863147,
      "loss": 0.8187,
      "step": 6978
    },
    {
      "epoch": 1.092517219787101,
      "grad_norm": 3.390990734100342,
      "learning_rate": 0.00014193548387096772,
      "loss": 0.9644,
      "step": 6979
    },
    {
      "epoch": 1.0926737633061991,
      "grad_norm": 3.826531171798706,
      "learning_rate": 0.000141911045943304,
      "loss": 1.2942,
      "step": 6980
    },
    {
      "epoch": 1.0928303068252974,
      "grad_norm": 2.1854097843170166,
      "learning_rate": 0.00014188660801564025,
      "loss": 1.08,
      "step": 6981
    },
    {
      "epoch": 1.0929868503443958,
      "grad_norm": 1.7453972101211548,
      "learning_rate": 0.00014186217008797653,
      "loss": 1.066,
      "step": 6982
    },
    {
      "epoch": 1.093143393863494,
      "grad_norm": 1.322583556175232,
      "learning_rate": 0.00014183773216031278,
      "loss": 1.0875,
      "step": 6983
    },
    {
      "epoch": 1.0932999373825925,
      "grad_norm": 3.8100578784942627,
      "learning_rate": 0.00014181329423264906,
      "loss": 0.9858,
      "step": 6984
    },
    {
      "epoch": 1.0934564809016907,
      "grad_norm": 1.9491938352584839,
      "learning_rate": 0.00014178885630498534,
      "loss": 1.0154,
      "step": 6985
    },
    {
      "epoch": 1.093613024420789,
      "grad_norm": 1.2058466672897339,
      "learning_rate": 0.0001417644183773216,
      "loss": 0.5821,
      "step": 6986
    },
    {
      "epoch": 1.0937695679398873,
      "grad_norm": 1.4509241580963135,
      "learning_rate": 0.00014173998044965784,
      "loss": 0.3841,
      "step": 6987
    },
    {
      "epoch": 1.0939261114589856,
      "grad_norm": 1.7737778425216675,
      "learning_rate": 0.00014171554252199412,
      "loss": 0.772,
      "step": 6988
    },
    {
      "epoch": 1.094082654978084,
      "grad_norm": 0.6553102135658264,
      "learning_rate": 0.0001416911045943304,
      "loss": 0.2943,
      "step": 6989
    },
    {
      "epoch": 1.0942391984971822,
      "grad_norm": 0.4766181409358978,
      "learning_rate": 0.00014166666666666665,
      "loss": 0.1739,
      "step": 6990
    },
    {
      "epoch": 1.0943957420162804,
      "grad_norm": 0.475190132856369,
      "learning_rate": 0.00014164222873900293,
      "loss": 0.229,
      "step": 6991
    },
    {
      "epoch": 1.0945522855353789,
      "grad_norm": 0.4964744448661804,
      "learning_rate": 0.00014161779081133918,
      "loss": 0.2992,
      "step": 6992
    },
    {
      "epoch": 1.094708829054477,
      "grad_norm": 0.7098869681358337,
      "learning_rate": 0.00014159335288367546,
      "loss": 0.3533,
      "step": 6993
    },
    {
      "epoch": 1.0948653725735755,
      "grad_norm": 0.6949512958526611,
      "learning_rate": 0.00014156891495601174,
      "loss": 0.3115,
      "step": 6994
    },
    {
      "epoch": 1.0950219160926737,
      "grad_norm": 1.074985384941101,
      "learning_rate": 0.000141544477028348,
      "loss": 0.2106,
      "step": 6995
    },
    {
      "epoch": 1.095178459611772,
      "grad_norm": 1.048966884613037,
      "learning_rate": 0.00014152003910068424,
      "loss": 0.2574,
      "step": 6996
    },
    {
      "epoch": 1.0953350031308704,
      "grad_norm": 0.6164217591285706,
      "learning_rate": 0.00014149560117302052,
      "loss": 0.2964,
      "step": 6997
    },
    {
      "epoch": 1.0954915466499686,
      "grad_norm": 0.6958305239677429,
      "learning_rate": 0.0001414711632453568,
      "loss": 0.2955,
      "step": 6998
    },
    {
      "epoch": 1.095648090169067,
      "grad_norm": 0.7299469709396362,
      "learning_rate": 0.00014144672531769305,
      "loss": 0.332,
      "step": 6999
    },
    {
      "epoch": 1.0958046336881653,
      "grad_norm": 2.961721181869507,
      "learning_rate": 0.00014142228739002932,
      "loss": 0.3274,
      "step": 7000
    },
    {
      "epoch": 1.0958046336881653,
      "eval_loss": 0.524387538433075,
      "eval_runtime": 205.3671,
      "eval_samples_per_second": 60.297,
      "eval_steps_per_second": 3.769,
      "eval_wer": 0.33754807998809955,
      "step": 7000
    },
    {
      "epoch": 1.0959611772072637,
      "grad_norm": 0.8560186624526978,
      "learning_rate": 0.00014139784946236558,
      "loss": 0.2684,
      "step": 7001
    },
    {
      "epoch": 1.096117720726362,
      "grad_norm": 0.9131744503974915,
      "learning_rate": 0.00014137341153470185,
      "loss": 0.3631,
      "step": 7002
    },
    {
      "epoch": 1.0962742642454602,
      "grad_norm": 2.9472861289978027,
      "learning_rate": 0.0001413489736070381,
      "loss": 0.7318,
      "step": 7003
    },
    {
      "epoch": 1.0964308077645586,
      "grad_norm": 1.0932270288467407,
      "learning_rate": 0.00014132453567937438,
      "loss": 0.3798,
      "step": 7004
    },
    {
      "epoch": 1.0965873512836568,
      "grad_norm": 1.729344367980957,
      "learning_rate": 0.00014130009775171064,
      "loss": 0.6468,
      "step": 7005
    },
    {
      "epoch": 1.0967438948027552,
      "grad_norm": 2.079954147338867,
      "learning_rate": 0.00014127565982404691,
      "loss": 0.4306,
      "step": 7006
    },
    {
      "epoch": 1.0969004383218535,
      "grad_norm": 0.8679769039154053,
      "learning_rate": 0.00014125122189638316,
      "loss": 0.4454,
      "step": 7007
    },
    {
      "epoch": 1.0970569818409517,
      "grad_norm": 0.751891553401947,
      "learning_rate": 0.00014122678396871944,
      "loss": 0.3538,
      "step": 7008
    },
    {
      "epoch": 1.0972135253600501,
      "grad_norm": 2.451910972595215,
      "learning_rate": 0.00014120234604105572,
      "loss": 0.3732,
      "step": 7009
    },
    {
      "epoch": 1.0973700688791483,
      "grad_norm": 1.3220970630645752,
      "learning_rate": 0.00014117790811339197,
      "loss": 0.6134,
      "step": 7010
    },
    {
      "epoch": 1.0975266123982468,
      "grad_norm": 2.3990230560302734,
      "learning_rate": 0.00014115347018572822,
      "loss": 0.5123,
      "step": 7011
    },
    {
      "epoch": 1.097683155917345,
      "grad_norm": 3.5617446899414062,
      "learning_rate": 0.0001411290322580645,
      "loss": 0.8965,
      "step": 7012
    },
    {
      "epoch": 1.0978396994364434,
      "grad_norm": 1.2367515563964844,
      "learning_rate": 0.00014110459433040078,
      "loss": 0.4453,
      "step": 7013
    },
    {
      "epoch": 1.0979962429555417,
      "grad_norm": 1.6141263246536255,
      "learning_rate": 0.00014108015640273703,
      "loss": 0.6857,
      "step": 7014
    },
    {
      "epoch": 1.0981527864746399,
      "grad_norm": 1.6750285625457764,
      "learning_rate": 0.0001410557184750733,
      "loss": 0.8078,
      "step": 7015
    },
    {
      "epoch": 1.0983093299937383,
      "grad_norm": 1.4285156726837158,
      "learning_rate": 0.00014103128054740956,
      "loss": 0.3836,
      "step": 7016
    },
    {
      "epoch": 1.0984658735128365,
      "grad_norm": 2.9705681800842285,
      "learning_rate": 0.00014100684261974584,
      "loss": 0.4486,
      "step": 7017
    },
    {
      "epoch": 1.098622417031935,
      "grad_norm": 1.3870073556900024,
      "learning_rate": 0.00014098240469208212,
      "loss": 0.5095,
      "step": 7018
    },
    {
      "epoch": 1.0987789605510332,
      "grad_norm": 1.9252225160598755,
      "learning_rate": 0.00014095796676441837,
      "loss": 0.3421,
      "step": 7019
    },
    {
      "epoch": 1.0989355040701314,
      "grad_norm": 3.1899898052215576,
      "learning_rate": 0.00014093352883675462,
      "loss": 0.9082,
      "step": 7020
    },
    {
      "epoch": 1.0990920475892298,
      "grad_norm": 2.3937432765960693,
      "learning_rate": 0.0001409090909090909,
      "loss": 1.0056,
      "step": 7021
    },
    {
      "epoch": 1.099248591108328,
      "grad_norm": 1.9584646224975586,
      "learning_rate": 0.00014088465298142718,
      "loss": 0.7778,
      "step": 7022
    },
    {
      "epoch": 1.0994051346274265,
      "grad_norm": 1.8999119997024536,
      "learning_rate": 0.00014086021505376343,
      "loss": 0.7928,
      "step": 7023
    },
    {
      "epoch": 1.0995616781465247,
      "grad_norm": 1.100942611694336,
      "learning_rate": 0.0001408357771260997,
      "loss": 0.3812,
      "step": 7024
    },
    {
      "epoch": 1.099718221665623,
      "grad_norm": 3.0417444705963135,
      "learning_rate": 0.00014081133919843596,
      "loss": 0.8766,
      "step": 7025
    },
    {
      "epoch": 1.0998747651847214,
      "grad_norm": 3.2357630729675293,
      "learning_rate": 0.00014078690127077224,
      "loss": 1.1982,
      "step": 7026
    },
    {
      "epoch": 1.1000313087038196,
      "grad_norm": 1.8465913534164429,
      "learning_rate": 0.0001407624633431085,
      "loss": 1.1896,
      "step": 7027
    },
    {
      "epoch": 1.100187852222918,
      "grad_norm": 1.5905226469039917,
      "learning_rate": 0.00014073802541544477,
      "loss": 0.9382,
      "step": 7028
    },
    {
      "epoch": 1.1003443957420163,
      "grad_norm": 2.8563072681427,
      "learning_rate": 0.00014071358748778102,
      "loss": 1.0176,
      "step": 7029
    },
    {
      "epoch": 1.1005009392611145,
      "grad_norm": 3.9410860538482666,
      "learning_rate": 0.0001406891495601173,
      "loss": 1.497,
      "step": 7030
    },
    {
      "epoch": 1.100657482780213,
      "grad_norm": 2.426844358444214,
      "learning_rate": 0.00014066471163245355,
      "loss": 1.0797,
      "step": 7031
    },
    {
      "epoch": 1.1008140262993111,
      "grad_norm": 2.4825949668884277,
      "learning_rate": 0.00014064027370478983,
      "loss": 1.0159,
      "step": 7032
    },
    {
      "epoch": 1.1009705698184096,
      "grad_norm": 2.973639965057373,
      "learning_rate": 0.00014061583577712608,
      "loss": 1.1596,
      "step": 7033
    },
    {
      "epoch": 1.1011271133375078,
      "grad_norm": 2.815019369125366,
      "learning_rate": 0.00014059139784946236,
      "loss": 0.3454,
      "step": 7034
    },
    {
      "epoch": 1.1012836568566062,
      "grad_norm": 1.925430178642273,
      "learning_rate": 0.0001405669599217986,
      "loss": 0.6438,
      "step": 7035
    },
    {
      "epoch": 1.1014402003757044,
      "grad_norm": 4.631162166595459,
      "learning_rate": 0.00014054252199413488,
      "loss": 0.5383,
      "step": 7036
    },
    {
      "epoch": 1.1015967438948027,
      "grad_norm": 3.7617523670196533,
      "learning_rate": 0.00014051808406647116,
      "loss": 1.449,
      "step": 7037
    },
    {
      "epoch": 1.101753287413901,
      "grad_norm": 2.4548864364624023,
      "learning_rate": 0.00014049364613880741,
      "loss": 1.1796,
      "step": 7038
    },
    {
      "epoch": 1.1019098309329993,
      "grad_norm": 0.4076089560985565,
      "learning_rate": 0.00014046920821114367,
      "loss": 0.2382,
      "step": 7039
    },
    {
      "epoch": 1.1020663744520978,
      "grad_norm": 0.7186834216117859,
      "learning_rate": 0.00014044477028347994,
      "loss": 0.1617,
      "step": 7040
    },
    {
      "epoch": 1.102222917971196,
      "grad_norm": 0.419718861579895,
      "learning_rate": 0.00014042033235581622,
      "loss": 0.1957,
      "step": 7041
    },
    {
      "epoch": 1.1023794614902942,
      "grad_norm": 0.6931477189064026,
      "learning_rate": 0.00014039589442815247,
      "loss": 0.2296,
      "step": 7042
    },
    {
      "epoch": 1.1025360050093926,
      "grad_norm": 1.0758944749832153,
      "learning_rate": 0.00014037145650048875,
      "loss": 0.4824,
      "step": 7043
    },
    {
      "epoch": 1.1026925485284909,
      "grad_norm": 0.721367359161377,
      "learning_rate": 0.000140347018572825,
      "loss": 0.3532,
      "step": 7044
    },
    {
      "epoch": 1.1028490920475893,
      "grad_norm": 0.8995819091796875,
      "learning_rate": 0.00014032258064516128,
      "loss": 0.2786,
      "step": 7045
    },
    {
      "epoch": 1.1030056355666875,
      "grad_norm": 0.5277887582778931,
      "learning_rate": 0.00014029814271749756,
      "loss": 0.2787,
      "step": 7046
    },
    {
      "epoch": 1.103162179085786,
      "grad_norm": 0.6943791508674622,
      "learning_rate": 0.0001402737047898338,
      "loss": 0.2092,
      "step": 7047
    },
    {
      "epoch": 1.1033187226048842,
      "grad_norm": 1.1070524454116821,
      "learning_rate": 0.00014024926686217006,
      "loss": 0.3526,
      "step": 7048
    },
    {
      "epoch": 1.1034752661239824,
      "grad_norm": 1.660226583480835,
      "learning_rate": 0.00014022482893450634,
      "loss": 0.3538,
      "step": 7049
    },
    {
      "epoch": 1.1036318096430808,
      "grad_norm": 1.7828032970428467,
      "learning_rate": 0.0001402003910068426,
      "loss": 0.6886,
      "step": 7050
    },
    {
      "epoch": 1.103788353162179,
      "grad_norm": 1.3736459016799927,
      "learning_rate": 0.00014017595307917887,
      "loss": 0.3213,
      "step": 7051
    },
    {
      "epoch": 1.1039448966812775,
      "grad_norm": 1.7011991739273071,
      "learning_rate": 0.00014015151515151515,
      "loss": 0.3819,
      "step": 7052
    },
    {
      "epoch": 1.1041014402003757,
      "grad_norm": 1.1513324975967407,
      "learning_rate": 0.0001401270772238514,
      "loss": 0.3103,
      "step": 7053
    },
    {
      "epoch": 1.104257983719474,
      "grad_norm": 0.6520640254020691,
      "learning_rate": 0.00014010263929618765,
      "loss": 0.2689,
      "step": 7054
    },
    {
      "epoch": 1.1044145272385724,
      "grad_norm": 1.3298591375350952,
      "learning_rate": 0.00014007820136852393,
      "loss": 0.7632,
      "step": 7055
    },
    {
      "epoch": 1.1045710707576706,
      "grad_norm": 1.7201464176177979,
      "learning_rate": 0.0001400537634408602,
      "loss": 0.3654,
      "step": 7056
    },
    {
      "epoch": 1.104727614276769,
      "grad_norm": 0.8514449596405029,
      "learning_rate": 0.00014002932551319646,
      "loss": 0.3845,
      "step": 7057
    },
    {
      "epoch": 1.1048841577958672,
      "grad_norm": 1.0151549577713013,
      "learning_rate": 0.00014000488758553274,
      "loss": 0.4584,
      "step": 7058
    },
    {
      "epoch": 1.1050407013149655,
      "grad_norm": 0.9856629967689514,
      "learning_rate": 0.000139980449657869,
      "loss": 0.5458,
      "step": 7059
    },
    {
      "epoch": 1.105197244834064,
      "grad_norm": 0.9040552377700806,
      "learning_rate": 0.00013995601173020527,
      "loss": 0.2539,
      "step": 7060
    },
    {
      "epoch": 1.1053537883531621,
      "grad_norm": 1.2553309202194214,
      "learning_rate": 0.00013993157380254155,
      "loss": 0.6513,
      "step": 7061
    },
    {
      "epoch": 1.1055103318722606,
      "grad_norm": 3.3496036529541016,
      "learning_rate": 0.0001399071358748778,
      "loss": 0.6279,
      "step": 7062
    },
    {
      "epoch": 1.1056668753913588,
      "grad_norm": 2.5765726566314697,
      "learning_rate": 0.00013988269794721405,
      "loss": 0.5166,
      "step": 7063
    },
    {
      "epoch": 1.105823418910457,
      "grad_norm": 1.883778691291809,
      "learning_rate": 0.00013985826001955033,
      "loss": 0.4737,
      "step": 7064
    },
    {
      "epoch": 1.1059799624295554,
      "grad_norm": 0.8956629037857056,
      "learning_rate": 0.0001398338220918866,
      "loss": 0.3679,
      "step": 7065
    },
    {
      "epoch": 1.1061365059486536,
      "grad_norm": 1.2945466041564941,
      "learning_rate": 0.00013980938416422286,
      "loss": 0.5382,
      "step": 7066
    },
    {
      "epoch": 1.106293049467752,
      "grad_norm": 3.4074881076812744,
      "learning_rate": 0.00013978494623655913,
      "loss": 0.3423,
      "step": 7067
    },
    {
      "epoch": 1.1064495929868503,
      "grad_norm": 1.626989722251892,
      "learning_rate": 0.00013976050830889539,
      "loss": 0.6275,
      "step": 7068
    },
    {
      "epoch": 1.1066061365059487,
      "grad_norm": 1.9460794925689697,
      "learning_rate": 0.00013973607038123166,
      "loss": 0.7302,
      "step": 7069
    },
    {
      "epoch": 1.106762680025047,
      "grad_norm": 1.4970815181732178,
      "learning_rate": 0.00013971163245356794,
      "loss": 0.6552,
      "step": 7070
    },
    {
      "epoch": 1.1069192235441452,
      "grad_norm": 1.4219391345977783,
      "learning_rate": 0.0001396871945259042,
      "loss": 0.6406,
      "step": 7071
    },
    {
      "epoch": 1.1070757670632436,
      "grad_norm": 2.354487180709839,
      "learning_rate": 0.00013966275659824044,
      "loss": 0.8659,
      "step": 7072
    },
    {
      "epoch": 1.1072323105823418,
      "grad_norm": 3.323763370513916,
      "learning_rate": 0.00013963831867057672,
      "loss": 0.7318,
      "step": 7073
    },
    {
      "epoch": 1.1073888541014403,
      "grad_norm": 2.702415943145752,
      "learning_rate": 0.00013961388074291297,
      "loss": 1.239,
      "step": 7074
    },
    {
      "epoch": 1.1075453976205385,
      "grad_norm": 2.1730268001556396,
      "learning_rate": 0.00013958944281524925,
      "loss": 1.0215,
      "step": 7075
    },
    {
      "epoch": 1.107701941139637,
      "grad_norm": 1.8554517030715942,
      "learning_rate": 0.00013956500488758553,
      "loss": 0.7656,
      "step": 7076
    },
    {
      "epoch": 1.1078584846587352,
      "grad_norm": 3.0490665435791016,
      "learning_rate": 0.00013954056695992178,
      "loss": 0.793,
      "step": 7077
    },
    {
      "epoch": 1.1080150281778334,
      "grad_norm": 2.6447458267211914,
      "learning_rate": 0.00013951612903225803,
      "loss": 1.0286,
      "step": 7078
    },
    {
      "epoch": 1.1081715716969318,
      "grad_norm": 0.89915931224823,
      "learning_rate": 0.0001394916911045943,
      "loss": 0.5665,
      "step": 7079
    },
    {
      "epoch": 1.10832811521603,
      "grad_norm": 1.9024105072021484,
      "learning_rate": 0.0001394672531769306,
      "loss": 0.6443,
      "step": 7080
    },
    {
      "epoch": 1.1084846587351285,
      "grad_norm": 1.5853981971740723,
      "learning_rate": 0.00013944281524926684,
      "loss": 1.179,
      "step": 7081
    },
    {
      "epoch": 1.1086412022542267,
      "grad_norm": 3.08502197265625,
      "learning_rate": 0.00013941837732160312,
      "loss": 0.9776,
      "step": 7082
    },
    {
      "epoch": 1.108797745773325,
      "grad_norm": 3.0830800533294678,
      "learning_rate": 0.00013939393939393937,
      "loss": 1.2178,
      "step": 7083
    },
    {
      "epoch": 1.1089542892924233,
      "grad_norm": 2.8586432933807373,
      "learning_rate": 0.00013936950146627565,
      "loss": 1.0084,
      "step": 7084
    },
    {
      "epoch": 1.1091108328115216,
      "grad_norm": 1.6677206754684448,
      "learning_rate": 0.00013934506353861193,
      "loss": 0.5648,
      "step": 7085
    },
    {
      "epoch": 1.10926737633062,
      "grad_norm": 2.1359286308288574,
      "learning_rate": 0.00013932062561094818,
      "loss": 0.6297,
      "step": 7086
    },
    {
      "epoch": 1.1094239198497182,
      "grad_norm": 6.148825168609619,
      "learning_rate": 0.00013929618768328443,
      "loss": 0.9579,
      "step": 7087
    },
    {
      "epoch": 1.1095804633688164,
      "grad_norm": 1.7311344146728516,
      "learning_rate": 0.0001392717497556207,
      "loss": 0.6473,
      "step": 7088
    },
    {
      "epoch": 1.1097370068879149,
      "grad_norm": 0.3622504472732544,
      "learning_rate": 0.000139247311827957,
      "loss": 0.1864,
      "step": 7089
    },
    {
      "epoch": 1.109893550407013,
      "grad_norm": 0.5219933986663818,
      "learning_rate": 0.00013922287390029324,
      "loss": 0.189,
      "step": 7090
    },
    {
      "epoch": 1.1100500939261115,
      "grad_norm": 0.3946225345134735,
      "learning_rate": 0.00013919843597262952,
      "loss": 0.2183,
      "step": 7091
    },
    {
      "epoch": 1.1102066374452098,
      "grad_norm": 0.47833681106567383,
      "learning_rate": 0.00013917399804496577,
      "loss": 0.155,
      "step": 7092
    },
    {
      "epoch": 1.110363180964308,
      "grad_norm": 0.4084630608558655,
      "learning_rate": 0.00013914956011730205,
      "loss": 0.1816,
      "step": 7093
    },
    {
      "epoch": 1.1105197244834064,
      "grad_norm": 0.45835641026496887,
      "learning_rate": 0.0001391251221896383,
      "loss": 0.2926,
      "step": 7094
    },
    {
      "epoch": 1.1106762680025046,
      "grad_norm": 0.5102660655975342,
      "learning_rate": 0.00013910068426197458,
      "loss": 0.2767,
      "step": 7095
    },
    {
      "epoch": 1.110832811521603,
      "grad_norm": 0.914391040802002,
      "learning_rate": 0.00013907624633431083,
      "loss": 0.2147,
      "step": 7096
    },
    {
      "epoch": 1.1109893550407013,
      "grad_norm": 0.6715912222862244,
      "learning_rate": 0.0001390518084066471,
      "loss": 0.2518,
      "step": 7097
    },
    {
      "epoch": 1.1111458985597997,
      "grad_norm": 0.5033890604972839,
      "learning_rate": 0.00013902737047898336,
      "loss": 0.3206,
      "step": 7098
    },
    {
      "epoch": 1.111302442078898,
      "grad_norm": 0.7809054851531982,
      "learning_rate": 0.00013900293255131963,
      "loss": 0.2376,
      "step": 7099
    },
    {
      "epoch": 1.1114589855979962,
      "grad_norm": 0.5782076120376587,
      "learning_rate": 0.0001389784946236559,
      "loss": 0.1966,
      "step": 7100
    },
    {
      "epoch": 1.1116155291170946,
      "grad_norm": 1.2228553295135498,
      "learning_rate": 0.00013895405669599216,
      "loss": 0.3316,
      "step": 7101
    },
    {
      "epoch": 1.1117720726361928,
      "grad_norm": 2.8982632160186768,
      "learning_rate": 0.00013892961876832842,
      "loss": 0.5064,
      "step": 7102
    },
    {
      "epoch": 1.1119286161552913,
      "grad_norm": 1.0576456785202026,
      "learning_rate": 0.0001389051808406647,
      "loss": 0.3983,
      "step": 7103
    },
    {
      "epoch": 1.1120851596743895,
      "grad_norm": 1.5845671892166138,
      "learning_rate": 0.00013888074291300097,
      "loss": 0.5427,
      "step": 7104
    },
    {
      "epoch": 1.1122417031934877,
      "grad_norm": 0.8153006434440613,
      "learning_rate": 0.00013885630498533722,
      "loss": 0.4551,
      "step": 7105
    },
    {
      "epoch": 1.1123982467125861,
      "grad_norm": 1.2537267208099365,
      "learning_rate": 0.0001388318670576735,
      "loss": 0.5381,
      "step": 7106
    },
    {
      "epoch": 1.1125547902316844,
      "grad_norm": 1.0054357051849365,
      "learning_rate": 0.00013880742913000975,
      "loss": 0.3709,
      "step": 7107
    },
    {
      "epoch": 1.1127113337507828,
      "grad_norm": 3.644198179244995,
      "learning_rate": 0.00013878299120234603,
      "loss": 0.7605,
      "step": 7108
    },
    {
      "epoch": 1.112867877269881,
      "grad_norm": 1.2919872999191284,
      "learning_rate": 0.0001387585532746823,
      "loss": 0.2972,
      "step": 7109
    },
    {
      "epoch": 1.1130244207889795,
      "grad_norm": 1.0998269319534302,
      "learning_rate": 0.00013873411534701856,
      "loss": 0.4665,
      "step": 7110
    },
    {
      "epoch": 1.1131809643080777,
      "grad_norm": 3.0610947608947754,
      "learning_rate": 0.0001387096774193548,
      "loss": 0.6518,
      "step": 7111
    },
    {
      "epoch": 1.1133375078271759,
      "grad_norm": 2.023725986480713,
      "learning_rate": 0.0001386852394916911,
      "loss": 0.5192,
      "step": 7112
    },
    {
      "epoch": 1.1134940513462743,
      "grad_norm": 2.0220577716827393,
      "learning_rate": 0.00013866080156402737,
      "loss": 0.5305,
      "step": 7113
    },
    {
      "epoch": 1.1136505948653725,
      "grad_norm": 1.242210030555725,
      "learning_rate": 0.00013863636363636362,
      "loss": 0.4925,
      "step": 7114
    },
    {
      "epoch": 1.113807138384471,
      "grad_norm": 1.3182028532028198,
      "learning_rate": 0.0001386119257086999,
      "loss": 0.4868,
      "step": 7115
    },
    {
      "epoch": 1.1139636819035692,
      "grad_norm": 1.934834361076355,
      "learning_rate": 0.00013858748778103615,
      "loss": 0.4217,
      "step": 7116
    },
    {
      "epoch": 1.1141202254226674,
      "grad_norm": 1.3865692615509033,
      "learning_rate": 0.00013856304985337243,
      "loss": 0.5035,
      "step": 7117
    },
    {
      "epoch": 1.1142767689417659,
      "grad_norm": 1.6283689737319946,
      "learning_rate": 0.00013853861192570868,
      "loss": 1.0412,
      "step": 7118
    },
    {
      "epoch": 1.114433312460864,
      "grad_norm": 1.3820260763168335,
      "learning_rate": 0.00013851417399804496,
      "loss": 1.0473,
      "step": 7119
    },
    {
      "epoch": 1.1145898559799625,
      "grad_norm": 1.8468633890151978,
      "learning_rate": 0.0001384897360703812,
      "loss": 1.0301,
      "step": 7120
    },
    {
      "epoch": 1.1147463994990607,
      "grad_norm": 1.0981626510620117,
      "learning_rate": 0.0001384652981427175,
      "loss": 0.4976,
      "step": 7121
    },
    {
      "epoch": 1.114902943018159,
      "grad_norm": 1.9592245817184448,
      "learning_rate": 0.00013844086021505374,
      "loss": 0.6123,
      "step": 7122
    },
    {
      "epoch": 1.1150594865372574,
      "grad_norm": 4.474161624908447,
      "learning_rate": 0.00013841642228739002,
      "loss": 1.0584,
      "step": 7123
    },
    {
      "epoch": 1.1152160300563556,
      "grad_norm": 2.5803914070129395,
      "learning_rate": 0.0001383919843597263,
      "loss": 0.6644,
      "step": 7124
    },
    {
      "epoch": 1.115372573575454,
      "grad_norm": 1.8294001817703247,
      "learning_rate": 0.00013836754643206255,
      "loss": 0.8667,
      "step": 7125
    },
    {
      "epoch": 1.1155291170945523,
      "grad_norm": 3.3992698192596436,
      "learning_rate": 0.0001383431085043988,
      "loss": 0.8291,
      "step": 7126
    },
    {
      "epoch": 1.1156856606136505,
      "grad_norm": 2.8989360332489014,
      "learning_rate": 0.00013831867057673508,
      "loss": 0.9821,
      "step": 7127
    },
    {
      "epoch": 1.115842204132749,
      "grad_norm": 2.251101016998291,
      "learning_rate": 0.00013829423264907135,
      "loss": 0.6455,
      "step": 7128
    },
    {
      "epoch": 1.1159987476518471,
      "grad_norm": 4.921138763427734,
      "learning_rate": 0.0001382697947214076,
      "loss": 1.1246,
      "step": 7129
    },
    {
      "epoch": 1.1161552911709456,
      "grad_norm": 1.7035528421401978,
      "learning_rate": 0.00013824535679374388,
      "loss": 0.997,
      "step": 7130
    },
    {
      "epoch": 1.1163118346900438,
      "grad_norm": 5.489228248596191,
      "learning_rate": 0.00013822091886608014,
      "loss": 0.9107,
      "step": 7131
    },
    {
      "epoch": 1.1164683782091422,
      "grad_norm": 3.2178499698638916,
      "learning_rate": 0.00013819648093841641,
      "loss": 0.7538,
      "step": 7132
    },
    {
      "epoch": 1.1166249217282405,
      "grad_norm": 2.4438555240631104,
      "learning_rate": 0.0001381720430107527,
      "loss": 0.8375,
      "step": 7133
    },
    {
      "epoch": 1.1167814652473387,
      "grad_norm": 3.7935683727264404,
      "learning_rate": 0.00013814760508308894,
      "loss": 1.1526,
      "step": 7134
    },
    {
      "epoch": 1.1169380087664371,
      "grad_norm": 1.034727931022644,
      "learning_rate": 0.0001381231671554252,
      "loss": 0.2834,
      "step": 7135
    },
    {
      "epoch": 1.1170945522855353,
      "grad_norm": 5.797002792358398,
      "learning_rate": 0.00013809872922776147,
      "loss": 0.4101,
      "step": 7136
    },
    {
      "epoch": 1.1172510958046338,
      "grad_norm": 1.3776600360870361,
      "learning_rate": 0.00013807429130009775,
      "loss": 0.589,
      "step": 7137
    },
    {
      "epoch": 1.117407639323732,
      "grad_norm": 2.727193832397461,
      "learning_rate": 0.000138049853372434,
      "loss": 1.4797,
      "step": 7138
    },
    {
      "epoch": 1.1175641828428302,
      "grad_norm": 0.7956806421279907,
      "learning_rate": 0.00013802541544477028,
      "loss": 0.3785,
      "step": 7139
    },
    {
      "epoch": 1.1177207263619287,
      "grad_norm": 0.5914304256439209,
      "learning_rate": 0.00013800097751710653,
      "loss": 0.2568,
      "step": 7140
    },
    {
      "epoch": 1.1178772698810269,
      "grad_norm": 0.49332305788993835,
      "learning_rate": 0.00013797653958944278,
      "loss": 0.2745,
      "step": 7141
    },
    {
      "epoch": 1.1180338134001253,
      "grad_norm": 0.7752807140350342,
      "learning_rate": 0.00013795210166177906,
      "loss": 0.1904,
      "step": 7142
    },
    {
      "epoch": 1.1181903569192235,
      "grad_norm": 0.6126279234886169,
      "learning_rate": 0.00013792766373411534,
      "loss": 0.2937,
      "step": 7143
    },
    {
      "epoch": 1.118346900438322,
      "grad_norm": 0.710114061832428,
      "learning_rate": 0.0001379032258064516,
      "loss": 0.2494,
      "step": 7144
    },
    {
      "epoch": 1.1185034439574202,
      "grad_norm": 0.5817550420761108,
      "learning_rate": 0.00013787878787878787,
      "loss": 0.1989,
      "step": 7145
    },
    {
      "epoch": 1.1186599874765184,
      "grad_norm": 0.6137755513191223,
      "learning_rate": 0.00013785434995112412,
      "loss": 0.2376,
      "step": 7146
    },
    {
      "epoch": 1.1188165309956168,
      "grad_norm": 0.6869461536407471,
      "learning_rate": 0.0001378299120234604,
      "loss": 0.2195,
      "step": 7147
    },
    {
      "epoch": 1.118973074514715,
      "grad_norm": 0.5993041396141052,
      "learning_rate": 0.00013780547409579668,
      "loss": 0.1858,
      "step": 7148
    },
    {
      "epoch": 1.1191296180338135,
      "grad_norm": 0.7191832065582275,
      "learning_rate": 0.00013778103616813293,
      "loss": 0.3344,
      "step": 7149
    },
    {
      "epoch": 1.1192861615529117,
      "grad_norm": 8.964902877807617,
      "learning_rate": 0.00013775659824046918,
      "loss": 2.0161,
      "step": 7150
    },
    {
      "epoch": 1.11944270507201,
      "grad_norm": 0.8494324088096619,
      "learning_rate": 0.00013773216031280546,
      "loss": 0.3409,
      "step": 7151
    },
    {
      "epoch": 1.1195992485911084,
      "grad_norm": 2.661954402923584,
      "learning_rate": 0.00013770772238514174,
      "loss": 0.3471,
      "step": 7152
    },
    {
      "epoch": 1.1197557921102066,
      "grad_norm": 1.1184935569763184,
      "learning_rate": 0.000137683284457478,
      "loss": 0.3072,
      "step": 7153
    },
    {
      "epoch": 1.119912335629305,
      "grad_norm": 0.8442760109901428,
      "learning_rate": 0.00013765884652981427,
      "loss": 0.3564,
      "step": 7154
    },
    {
      "epoch": 1.1200688791484033,
      "grad_norm": 1.7543659210205078,
      "learning_rate": 0.00013763440860215052,
      "loss": 0.537,
      "step": 7155
    },
    {
      "epoch": 1.1202254226675015,
      "grad_norm": 4.0782060623168945,
      "learning_rate": 0.0001376099706744868,
      "loss": 0.5696,
      "step": 7156
    },
    {
      "epoch": 1.1203819661866,
      "grad_norm": 1.1545088291168213,
      "learning_rate": 0.00013758553274682307,
      "loss": 0.4484,
      "step": 7157
    },
    {
      "epoch": 1.1205385097056981,
      "grad_norm": 1.2286568880081177,
      "learning_rate": 0.00013756109481915933,
      "loss": 0.5164,
      "step": 7158
    },
    {
      "epoch": 1.1206950532247966,
      "grad_norm": 1.0465517044067383,
      "learning_rate": 0.00013753665689149558,
      "loss": 0.3714,
      "step": 7159
    },
    {
      "epoch": 1.1208515967438948,
      "grad_norm": 1.1569772958755493,
      "learning_rate": 0.00013751221896383186,
      "loss": 0.2982,
      "step": 7160
    },
    {
      "epoch": 1.121008140262993,
      "grad_norm": 2.364917755126953,
      "learning_rate": 0.00013748778103616813,
      "loss": 0.5142,
      "step": 7161
    },
    {
      "epoch": 1.1211646837820914,
      "grad_norm": 1.6103262901306152,
      "learning_rate": 0.00013746334310850439,
      "loss": 0.5519,
      "step": 7162
    },
    {
      "epoch": 1.1213212273011897,
      "grad_norm": 2.0225682258605957,
      "learning_rate": 0.00013743890518084066,
      "loss": 0.4566,
      "step": 7163
    },
    {
      "epoch": 1.121477770820288,
      "grad_norm": 1.9558186531066895,
      "learning_rate": 0.00013741446725317691,
      "loss": 0.4267,
      "step": 7164
    },
    {
      "epoch": 1.1216343143393863,
      "grad_norm": 3.2678587436676025,
      "learning_rate": 0.00013739002932551317,
      "loss": 0.6969,
      "step": 7165
    },
    {
      "epoch": 1.1217908578584848,
      "grad_norm": 1.560784101486206,
      "learning_rate": 0.00013736559139784944,
      "loss": 0.6032,
      "step": 7166
    },
    {
      "epoch": 1.121947401377583,
      "grad_norm": 1.4639748334884644,
      "learning_rate": 0.00013734115347018572,
      "loss": 0.3596,
      "step": 7167
    },
    {
      "epoch": 1.1221039448966812,
      "grad_norm": 1.3413069248199463,
      "learning_rate": 0.00013731671554252197,
      "loss": 0.7597,
      "step": 7168
    },
    {
      "epoch": 1.1222604884157796,
      "grad_norm": 2.400970935821533,
      "learning_rate": 0.00013729227761485825,
      "loss": 0.8581,
      "step": 7169
    },
    {
      "epoch": 1.1224170319348779,
      "grad_norm": 1.9984264373779297,
      "learning_rate": 0.0001372678396871945,
      "loss": 0.5273,
      "step": 7170
    },
    {
      "epoch": 1.1225735754539763,
      "grad_norm": 2.334470510482788,
      "learning_rate": 0.00013724340175953078,
      "loss": 1.0934,
      "step": 7171
    },
    {
      "epoch": 1.1227301189730745,
      "grad_norm": 6.798165321350098,
      "learning_rate": 0.00013721896383186706,
      "loss": 1.0502,
      "step": 7172
    },
    {
      "epoch": 1.1228866624921727,
      "grad_norm": 3.1501338481903076,
      "learning_rate": 0.0001371945259042033,
      "loss": 0.806,
      "step": 7173
    },
    {
      "epoch": 1.1230432060112712,
      "grad_norm": 5.227148056030273,
      "learning_rate": 0.00013717008797653956,
      "loss": 1.0714,
      "step": 7174
    },
    {
      "epoch": 1.1231997495303694,
      "grad_norm": 3.071563482284546,
      "learning_rate": 0.00013714565004887584,
      "loss": 1.4736,
      "step": 7175
    },
    {
      "epoch": 1.1233562930494678,
      "grad_norm": 4.112018585205078,
      "learning_rate": 0.00013712121212121212,
      "loss": 1.5653,
      "step": 7176
    },
    {
      "epoch": 1.123512836568566,
      "grad_norm": 3.3567934036254883,
      "learning_rate": 0.00013709677419354837,
      "loss": 0.8757,
      "step": 7177
    },
    {
      "epoch": 1.1236693800876645,
      "grad_norm": 2.7976365089416504,
      "learning_rate": 0.00013707233626588465,
      "loss": 0.7167,
      "step": 7178
    },
    {
      "epoch": 1.1238259236067627,
      "grad_norm": 3.194066286087036,
      "learning_rate": 0.0001370478983382209,
      "loss": 1.4257,
      "step": 7179
    },
    {
      "epoch": 1.123982467125861,
      "grad_norm": 5.53701114654541,
      "learning_rate": 0.00013702346041055718,
      "loss": 0.8894,
      "step": 7180
    },
    {
      "epoch": 1.1241390106449594,
      "grad_norm": 1.579353928565979,
      "learning_rate": 0.00013699902248289346,
      "loss": 1.0472,
      "step": 7181
    },
    {
      "epoch": 1.1242955541640576,
      "grad_norm": 2.461728811264038,
      "learning_rate": 0.0001369745845552297,
      "loss": 1.0859,
      "step": 7182
    },
    {
      "epoch": 1.124452097683156,
      "grad_norm": 4.658527851104736,
      "learning_rate": 0.00013695014662756596,
      "loss": 0.7208,
      "step": 7183
    },
    {
      "epoch": 1.1246086412022542,
      "grad_norm": 2.3477227687835693,
      "learning_rate": 0.00013692570869990224,
      "loss": 1.3767,
      "step": 7184
    },
    {
      "epoch": 1.1247651847213525,
      "grad_norm": 1.8970081806182861,
      "learning_rate": 0.00013690127077223852,
      "loss": 1.0941,
      "step": 7185
    },
    {
      "epoch": 1.124921728240451,
      "grad_norm": 5.8194355964660645,
      "learning_rate": 0.00013687683284457477,
      "loss": 0.7754,
      "step": 7186
    },
    {
      "epoch": 1.125078271759549,
      "grad_norm": 2.781679153442383,
      "learning_rate": 0.00013685239491691105,
      "loss": 1.0529,
      "step": 7187
    },
    {
      "epoch": 1.1252348152786475,
      "grad_norm": 3.368130683898926,
      "learning_rate": 0.0001368279569892473,
      "loss": 1.2293,
      "step": 7188
    },
    {
      "epoch": 1.1253913587977458,
      "grad_norm": 0.35694655776023865,
      "learning_rate": 0.00013680351906158355,
      "loss": 0.1819,
      "step": 7189
    },
    {
      "epoch": 1.125547902316844,
      "grad_norm": 0.6090511083602905,
      "learning_rate": 0.00013677908113391983,
      "loss": 0.4191,
      "step": 7190
    },
    {
      "epoch": 1.1257044458359424,
      "grad_norm": 0.5560696721076965,
      "learning_rate": 0.0001367546432062561,
      "loss": 0.202,
      "step": 7191
    },
    {
      "epoch": 1.1258609893550406,
      "grad_norm": 0.4044315218925476,
      "learning_rate": 0.00013673020527859236,
      "loss": 0.1424,
      "step": 7192
    },
    {
      "epoch": 1.126017532874139,
      "grad_norm": 0.5974329113960266,
      "learning_rate": 0.00013670576735092863,
      "loss": 0.2354,
      "step": 7193
    },
    {
      "epoch": 1.1261740763932373,
      "grad_norm": 0.4592570960521698,
      "learning_rate": 0.00013668132942326489,
      "loss": 0.3103,
      "step": 7194
    },
    {
      "epoch": 1.1263306199123355,
      "grad_norm": 0.8667699694633484,
      "learning_rate": 0.00013665689149560116,
      "loss": 0.6261,
      "step": 7195
    },
    {
      "epoch": 1.126487163431434,
      "grad_norm": 0.5696028470993042,
      "learning_rate": 0.00013663245356793744,
      "loss": 0.2512,
      "step": 7196
    },
    {
      "epoch": 1.1266437069505322,
      "grad_norm": 0.41011011600494385,
      "learning_rate": 0.0001366080156402737,
      "loss": 0.1274,
      "step": 7197
    },
    {
      "epoch": 1.1268002504696306,
      "grad_norm": 0.686440646648407,
      "learning_rate": 0.00013658357771260995,
      "loss": 0.3106,
      "step": 7198
    },
    {
      "epoch": 1.1269567939887288,
      "grad_norm": 0.823204755783081,
      "learning_rate": 0.00013655913978494622,
      "loss": 0.2809,
      "step": 7199
    },
    {
      "epoch": 1.127113337507827,
      "grad_norm": 0.6141254901885986,
      "learning_rate": 0.0001365347018572825,
      "loss": 0.263,
      "step": 7200
    },
    {
      "epoch": 1.1272698810269255,
      "grad_norm": 0.6794694066047668,
      "learning_rate": 0.00013651026392961875,
      "loss": 0.2794,
      "step": 7201
    },
    {
      "epoch": 1.1274264245460237,
      "grad_norm": 0.7313904166221619,
      "learning_rate": 0.00013648582600195503,
      "loss": 0.2181,
      "step": 7202
    },
    {
      "epoch": 1.1275829680651221,
      "grad_norm": 0.7755028009414673,
      "learning_rate": 0.00013646138807429128,
      "loss": 0.3791,
      "step": 7203
    },
    {
      "epoch": 1.1277395115842204,
      "grad_norm": 0.8282299041748047,
      "learning_rate": 0.00013643695014662756,
      "loss": 0.3037,
      "step": 7204
    },
    {
      "epoch": 1.1278960551033188,
      "grad_norm": 1.4426270723342896,
      "learning_rate": 0.00013641251221896384,
      "loss": 0.5826,
      "step": 7205
    },
    {
      "epoch": 1.128052598622417,
      "grad_norm": 1.3686167001724243,
      "learning_rate": 0.0001363880742913001,
      "loss": 0.3046,
      "step": 7206
    },
    {
      "epoch": 1.1282091421415155,
      "grad_norm": 1.07740318775177,
      "learning_rate": 0.00013636363636363634,
      "loss": 0.3509,
      "step": 7207
    },
    {
      "epoch": 1.1283656856606137,
      "grad_norm": 1.3811142444610596,
      "learning_rate": 0.00013633919843597262,
      "loss": 0.4518,
      "step": 7208
    },
    {
      "epoch": 1.128522229179712,
      "grad_norm": 1.2326154708862305,
      "learning_rate": 0.00013631476050830887,
      "loss": 0.4252,
      "step": 7209
    },
    {
      "epoch": 1.1286787726988103,
      "grad_norm": 1.8412114381790161,
      "learning_rate": 0.00013629032258064515,
      "loss": 0.5223,
      "step": 7210
    },
    {
      "epoch": 1.1288353162179086,
      "grad_norm": 1.9933035373687744,
      "learning_rate": 0.00013626588465298143,
      "loss": 0.5177,
      "step": 7211
    },
    {
      "epoch": 1.128991859737007,
      "grad_norm": 1.4402484893798828,
      "learning_rate": 0.00013624144672531768,
      "loss": 0.5853,
      "step": 7212
    },
    {
      "epoch": 1.1291484032561052,
      "grad_norm": 1.115515947341919,
      "learning_rate": 0.00013621700879765393,
      "loss": 0.4662,
      "step": 7213
    },
    {
      "epoch": 1.1293049467752034,
      "grad_norm": 2.486177444458008,
      "learning_rate": 0.0001361925708699902,
      "loss": 0.7741,
      "step": 7214
    },
    {
      "epoch": 1.1294614902943019,
      "grad_norm": 1.2071272134780884,
      "learning_rate": 0.0001361681329423265,
      "loss": 0.2729,
      "step": 7215
    },
    {
      "epoch": 1.1296180338134,
      "grad_norm": 5.3013529777526855,
      "learning_rate": 0.00013614369501466274,
      "loss": 0.9455,
      "step": 7216
    },
    {
      "epoch": 1.1297745773324985,
      "grad_norm": 2.959777593612671,
      "learning_rate": 0.00013611925708699902,
      "loss": 0.9704,
      "step": 7217
    },
    {
      "epoch": 1.1299311208515967,
      "grad_norm": 3.2113773822784424,
      "learning_rate": 0.00013609481915933527,
      "loss": 0.8733,
      "step": 7218
    },
    {
      "epoch": 1.130087664370695,
      "grad_norm": 1.5459614992141724,
      "learning_rate": 0.00013607038123167155,
      "loss": 0.616,
      "step": 7219
    },
    {
      "epoch": 1.1302442078897934,
      "grad_norm": 2.0902748107910156,
      "learning_rate": 0.00013604594330400782,
      "loss": 0.2964,
      "step": 7220
    },
    {
      "epoch": 1.1304007514088916,
      "grad_norm": 3.8055615425109863,
      "learning_rate": 0.00013602150537634408,
      "loss": 0.5847,
      "step": 7221
    },
    {
      "epoch": 1.13055729492799,
      "grad_norm": 2.048724412918091,
      "learning_rate": 0.00013599706744868033,
      "loss": 1.0193,
      "step": 7222
    },
    {
      "epoch": 1.1307138384470883,
      "grad_norm": 2.0475356578826904,
      "learning_rate": 0.0001359726295210166,
      "loss": 0.8073,
      "step": 7223
    },
    {
      "epoch": 1.1308703819661865,
      "grad_norm": 2.0801010131835938,
      "learning_rate": 0.00013594819159335288,
      "loss": 0.6983,
      "step": 7224
    },
    {
      "epoch": 1.131026925485285,
      "grad_norm": 2.2826006412506104,
      "learning_rate": 0.00013592375366568914,
      "loss": 0.78,
      "step": 7225
    },
    {
      "epoch": 1.1311834690043832,
      "grad_norm": 2.933180093765259,
      "learning_rate": 0.00013589931573802541,
      "loss": 0.6017,
      "step": 7226
    },
    {
      "epoch": 1.1313400125234816,
      "grad_norm": 2.438169240951538,
      "learning_rate": 0.00013587487781036166,
      "loss": 1.0778,
      "step": 7227
    },
    {
      "epoch": 1.1314965560425798,
      "grad_norm": 1.5362167358398438,
      "learning_rate": 0.00013585043988269794,
      "loss": 0.851,
      "step": 7228
    },
    {
      "epoch": 1.131653099561678,
      "grad_norm": 3.1869149208068848,
      "learning_rate": 0.00013582600195503422,
      "loss": 1.0545,
      "step": 7229
    },
    {
      "epoch": 1.1318096430807765,
      "grad_norm": 1.230353593826294,
      "learning_rate": 0.00013580156402737047,
      "loss": 0.9157,
      "step": 7230
    },
    {
      "epoch": 1.1319661865998747,
      "grad_norm": 1.9790470600128174,
      "learning_rate": 0.00013577712609970672,
      "loss": 1.0721,
      "step": 7231
    },
    {
      "epoch": 1.1321227301189731,
      "grad_norm": 2.1077675819396973,
      "learning_rate": 0.000135752688172043,
      "loss": 1.4632,
      "step": 7232
    },
    {
      "epoch": 1.1322792736380713,
      "grad_norm": 2.5034217834472656,
      "learning_rate": 0.00013572825024437925,
      "loss": 2.0202,
      "step": 7233
    },
    {
      "epoch": 1.1324358171571698,
      "grad_norm": 1.6274192333221436,
      "learning_rate": 0.00013570381231671553,
      "loss": 0.8832,
      "step": 7234
    },
    {
      "epoch": 1.132592360676268,
      "grad_norm": 1.7369862794876099,
      "learning_rate": 0.0001356793743890518,
      "loss": 0.5584,
      "step": 7235
    },
    {
      "epoch": 1.1327489041953662,
      "grad_norm": 2.9159839153289795,
      "learning_rate": 0.00013565493646138806,
      "loss": 0.8328,
      "step": 7236
    },
    {
      "epoch": 1.1329054477144647,
      "grad_norm": 1.95712411403656,
      "learning_rate": 0.0001356304985337243,
      "loss": 0.6125,
      "step": 7237
    },
    {
      "epoch": 1.1330619912335629,
      "grad_norm": 2.354888916015625,
      "learning_rate": 0.0001356060606060606,
      "loss": 1.3286,
      "step": 7238
    },
    {
      "epoch": 1.1332185347526613,
      "grad_norm": 0.7385460734367371,
      "learning_rate": 0.00013558162267839687,
      "loss": 0.303,
      "step": 7239
    },
    {
      "epoch": 1.1333750782717595,
      "grad_norm": 0.645324170589447,
      "learning_rate": 0.00013555718475073312,
      "loss": 0.2112,
      "step": 7240
    },
    {
      "epoch": 1.133531621790858,
      "grad_norm": 0.979313313961029,
      "learning_rate": 0.0001355327468230694,
      "loss": 0.4106,
      "step": 7241
    },
    {
      "epoch": 1.1336881653099562,
      "grad_norm": 0.601364254951477,
      "learning_rate": 0.00013550830889540565,
      "loss": 0.2252,
      "step": 7242
    },
    {
      "epoch": 1.1338447088290544,
      "grad_norm": 0.45826420187950134,
      "learning_rate": 0.00013548387096774193,
      "loss": 0.1746,
      "step": 7243
    },
    {
      "epoch": 1.1340012523481529,
      "grad_norm": 0.6122409105300903,
      "learning_rate": 0.0001354594330400782,
      "loss": 0.3237,
      "step": 7244
    },
    {
      "epoch": 1.134157795867251,
      "grad_norm": 0.7006223201751709,
      "learning_rate": 0.00013543499511241446,
      "loss": 0.2608,
      "step": 7245
    },
    {
      "epoch": 1.1343143393863495,
      "grad_norm": 0.7755613923072815,
      "learning_rate": 0.0001354105571847507,
      "loss": 0.2501,
      "step": 7246
    },
    {
      "epoch": 1.1344708829054477,
      "grad_norm": 0.7348597645759583,
      "learning_rate": 0.000135386119257087,
      "loss": 0.1775,
      "step": 7247
    },
    {
      "epoch": 1.134627426424546,
      "grad_norm": 1.9953992366790771,
      "learning_rate": 0.00013536168132942327,
      "loss": 0.3969,
      "step": 7248
    },
    {
      "epoch": 1.1347839699436444,
      "grad_norm": 0.8354098796844482,
      "learning_rate": 0.00013533724340175952,
      "loss": 0.3183,
      "step": 7249
    },
    {
      "epoch": 1.1349405134627426,
      "grad_norm": 0.8279335498809814,
      "learning_rate": 0.0001353128054740958,
      "loss": 0.3691,
      "step": 7250
    },
    {
      "epoch": 1.135097056981841,
      "grad_norm": 0.7575170397758484,
      "learning_rate": 0.00013528836754643205,
      "loss": 0.2994,
      "step": 7251
    },
    {
      "epoch": 1.1352536005009393,
      "grad_norm": 0.6239116787910461,
      "learning_rate": 0.00013526392961876833,
      "loss": 0.1918,
      "step": 7252
    },
    {
      "epoch": 1.1354101440200375,
      "grad_norm": 1.2877776622772217,
      "learning_rate": 0.00013523949169110458,
      "loss": 0.45,
      "step": 7253
    },
    {
      "epoch": 1.135566687539136,
      "grad_norm": 1.0643291473388672,
      "learning_rate": 0.00013521505376344086,
      "loss": 0.3337,
      "step": 7254
    },
    {
      "epoch": 1.1357232310582341,
      "grad_norm": 1.3853017091751099,
      "learning_rate": 0.0001351906158357771,
      "loss": 0.4557,
      "step": 7255
    },
    {
      "epoch": 1.1358797745773326,
      "grad_norm": 1.043006181716919,
      "learning_rate": 0.00013516617790811338,
      "loss": 0.3404,
      "step": 7256
    },
    {
      "epoch": 1.1360363180964308,
      "grad_norm": 1.343376636505127,
      "learning_rate": 0.00013514173998044964,
      "loss": 0.3246,
      "step": 7257
    },
    {
      "epoch": 1.136192861615529,
      "grad_norm": 1.1656047105789185,
      "learning_rate": 0.00013511730205278591,
      "loss": 0.3422,
      "step": 7258
    },
    {
      "epoch": 1.1363494051346275,
      "grad_norm": 2.2045726776123047,
      "learning_rate": 0.0001350928641251222,
      "loss": 0.7908,
      "step": 7259
    },
    {
      "epoch": 1.1365059486537257,
      "grad_norm": 1.6047229766845703,
      "learning_rate": 0.00013506842619745844,
      "loss": 0.3087,
      "step": 7260
    },
    {
      "epoch": 1.1366624921728241,
      "grad_norm": 1.924752116203308,
      "learning_rate": 0.0001350439882697947,
      "loss": 0.6155,
      "step": 7261
    },
    {
      "epoch": 1.1368190356919223,
      "grad_norm": 1.1636427640914917,
      "learning_rate": 0.00013501955034213097,
      "loss": 0.4651,
      "step": 7262
    },
    {
      "epoch": 1.1369755792110205,
      "grad_norm": 1.3303437232971191,
      "learning_rate": 0.00013499511241446725,
      "loss": 0.3138,
      "step": 7263
    },
    {
      "epoch": 1.137132122730119,
      "grad_norm": 1.4854735136032104,
      "learning_rate": 0.0001349706744868035,
      "loss": 0.5431,
      "step": 7264
    },
    {
      "epoch": 1.1372886662492172,
      "grad_norm": 1.4238706827163696,
      "learning_rate": 0.00013494623655913978,
      "loss": 0.7749,
      "step": 7265
    },
    {
      "epoch": 1.1374452097683156,
      "grad_norm": 1.6186683177947998,
      "learning_rate": 0.00013492179863147603,
      "loss": 0.4627,
      "step": 7266
    },
    {
      "epoch": 1.1376017532874139,
      "grad_norm": 2.7942728996276855,
      "learning_rate": 0.0001348973607038123,
      "loss": 0.6011,
      "step": 7267
    },
    {
      "epoch": 1.1377582968065123,
      "grad_norm": 1.4344183206558228,
      "learning_rate": 0.0001348729227761486,
      "loss": 0.5155,
      "step": 7268
    },
    {
      "epoch": 1.1379148403256105,
      "grad_norm": 1.8099091053009033,
      "learning_rate": 0.00013484848484848484,
      "loss": 0.5645,
      "step": 7269
    },
    {
      "epoch": 1.1380713838447087,
      "grad_norm": 2.336829900741577,
      "learning_rate": 0.0001348240469208211,
      "loss": 0.6922,
      "step": 7270
    },
    {
      "epoch": 1.1382279273638072,
      "grad_norm": 2.197136640548706,
      "learning_rate": 0.00013479960899315737,
      "loss": 0.6654,
      "step": 7271
    },
    {
      "epoch": 1.1383844708829054,
      "grad_norm": 1.8803507089614868,
      "learning_rate": 0.00013477517106549365,
      "loss": 0.6727,
      "step": 7272
    },
    {
      "epoch": 1.1385410144020038,
      "grad_norm": 2.4235281944274902,
      "learning_rate": 0.0001347507331378299,
      "loss": 0.6877,
      "step": 7273
    },
    {
      "epoch": 1.138697557921102,
      "grad_norm": 6.200526237487793,
      "learning_rate": 0.00013472629521016618,
      "loss": 0.9574,
      "step": 7274
    },
    {
      "epoch": 1.1388541014402005,
      "grad_norm": 4.244660377502441,
      "learning_rate": 0.00013470185728250243,
      "loss": 1.4685,
      "step": 7275
    },
    {
      "epoch": 1.1390106449592987,
      "grad_norm": 1.6314693689346313,
      "learning_rate": 0.0001346774193548387,
      "loss": 0.6895,
      "step": 7276
    },
    {
      "epoch": 1.139167188478397,
      "grad_norm": 2.2845335006713867,
      "learning_rate": 0.00013465298142717496,
      "loss": 1.0343,
      "step": 7277
    },
    {
      "epoch": 1.1393237319974954,
      "grad_norm": 2.063843250274658,
      "learning_rate": 0.00013462854349951124,
      "loss": 1.037,
      "step": 7278
    },
    {
      "epoch": 1.1394802755165936,
      "grad_norm": 3.8249752521514893,
      "learning_rate": 0.0001346041055718475,
      "loss": 1.1499,
      "step": 7279
    },
    {
      "epoch": 1.139636819035692,
      "grad_norm": 2.1687839031219482,
      "learning_rate": 0.00013457966764418377,
      "loss": 1.3537,
      "step": 7280
    },
    {
      "epoch": 1.1397933625547902,
      "grad_norm": 1.7228095531463623,
      "learning_rate": 0.00013455522971652002,
      "loss": 1.3049,
      "step": 7281
    },
    {
      "epoch": 1.1399499060738885,
      "grad_norm": 3.0871424674987793,
      "learning_rate": 0.0001345307917888563,
      "loss": 1.2728,
      "step": 7282
    },
    {
      "epoch": 1.140106449592987,
      "grad_norm": 1.9572391510009766,
      "learning_rate": 0.00013450635386119258,
      "loss": 0.8322,
      "step": 7283
    },
    {
      "epoch": 1.1402629931120851,
      "grad_norm": 1.1564244031906128,
      "learning_rate": 0.00013448191593352883,
      "loss": 0.7581,
      "step": 7284
    },
    {
      "epoch": 1.1404195366311836,
      "grad_norm": 2.321030378341675,
      "learning_rate": 0.00013445747800586508,
      "loss": 0.7736,
      "step": 7285
    },
    {
      "epoch": 1.1405760801502818,
      "grad_norm": 3.072721481323242,
      "learning_rate": 0.00013443304007820136,
      "loss": 0.3721,
      "step": 7286
    },
    {
      "epoch": 1.14073262366938,
      "grad_norm": 2.0024454593658447,
      "learning_rate": 0.00013440860215053763,
      "loss": 0.6595,
      "step": 7287
    },
    {
      "epoch": 1.1408891671884784,
      "grad_norm": 0.9508635401725769,
      "learning_rate": 0.00013438416422287389,
      "loss": 0.3166,
      "step": 7288
    },
    {
      "epoch": 1.1410457107075767,
      "grad_norm": 0.390267014503479,
      "learning_rate": 0.00013435972629521014,
      "loss": 0.2544,
      "step": 7289
    },
    {
      "epoch": 1.141202254226675,
      "grad_norm": 0.6170450448989868,
      "learning_rate": 0.00013433528836754642,
      "loss": 0.2775,
      "step": 7290
    },
    {
      "epoch": 1.1413587977457733,
      "grad_norm": 0.7190653681755066,
      "learning_rate": 0.0001343108504398827,
      "loss": 0.2613,
      "step": 7291
    },
    {
      "epoch": 1.1415153412648715,
      "grad_norm": 0.6273028254508972,
      "learning_rate": 0.00013428641251221894,
      "loss": 0.1957,
      "step": 7292
    },
    {
      "epoch": 1.14167188478397,
      "grad_norm": 1.515903115272522,
      "learning_rate": 0.00013426197458455522,
      "loss": 0.3157,
      "step": 7293
    },
    {
      "epoch": 1.1418284283030682,
      "grad_norm": 0.493855744600296,
      "learning_rate": 0.00013423753665689147,
      "loss": 0.2462,
      "step": 7294
    },
    {
      "epoch": 1.1419849718221666,
      "grad_norm": 1.0583348274230957,
      "learning_rate": 0.00013421309872922775,
      "loss": 0.3651,
      "step": 7295
    },
    {
      "epoch": 1.1421415153412648,
      "grad_norm": 0.5995838642120361,
      "learning_rate": 0.00013418866080156403,
      "loss": 0.2172,
      "step": 7296
    },
    {
      "epoch": 1.142298058860363,
      "grad_norm": 1.2257310152053833,
      "learning_rate": 0.00013416422287390028,
      "loss": 0.2177,
      "step": 7297
    },
    {
      "epoch": 1.1424546023794615,
      "grad_norm": 0.7221239805221558,
      "learning_rate": 0.00013413978494623653,
      "loss": 0.1958,
      "step": 7298
    },
    {
      "epoch": 1.1426111458985597,
      "grad_norm": 0.5078455209732056,
      "learning_rate": 0.0001341153470185728,
      "loss": 0.1939,
      "step": 7299
    },
    {
      "epoch": 1.1427676894176582,
      "grad_norm": 0.8579664826393127,
      "learning_rate": 0.00013409090909090906,
      "loss": 0.3449,
      "step": 7300
    },
    {
      "epoch": 1.1429242329367564,
      "grad_norm": 1.03110933303833,
      "learning_rate": 0.00013406647116324534,
      "loss": 0.3881,
      "step": 7301
    },
    {
      "epoch": 1.1430807764558548,
      "grad_norm": 0.700508713722229,
      "learning_rate": 0.00013404203323558162,
      "loss": 0.2653,
      "step": 7302
    },
    {
      "epoch": 1.143237319974953,
      "grad_norm": 1.0958391427993774,
      "learning_rate": 0.00013401759530791787,
      "loss": 0.4662,
      "step": 7303
    },
    {
      "epoch": 1.1433938634940513,
      "grad_norm": 1.1270815134048462,
      "learning_rate": 0.00013399315738025412,
      "loss": 0.2924,
      "step": 7304
    },
    {
      "epoch": 1.1435504070131497,
      "grad_norm": 1.0125340223312378,
      "learning_rate": 0.0001339687194525904,
      "loss": 0.4216,
      "step": 7305
    },
    {
      "epoch": 1.143706950532248,
      "grad_norm": 3.5537991523742676,
      "learning_rate": 0.00013394428152492668,
      "loss": 0.5891,
      "step": 7306
    },
    {
      "epoch": 1.1438634940513464,
      "grad_norm": 1.405709147453308,
      "learning_rate": 0.00013391984359726293,
      "loss": 0.4454,
      "step": 7307
    },
    {
      "epoch": 1.1440200375704446,
      "grad_norm": 1.4942262172698975,
      "learning_rate": 0.0001338954056695992,
      "loss": 0.6456,
      "step": 7308
    },
    {
      "epoch": 1.144176581089543,
      "grad_norm": 3.732208490371704,
      "learning_rate": 0.00013387096774193546,
      "loss": 0.3922,
      "step": 7309
    },
    {
      "epoch": 1.1443331246086412,
      "grad_norm": 1.3866475820541382,
      "learning_rate": 0.00013384652981427174,
      "loss": 0.7081,
      "step": 7310
    },
    {
      "epoch": 1.1444896681277394,
      "grad_norm": 0.8793286085128784,
      "learning_rate": 0.00013382209188660802,
      "loss": 0.2748,
      "step": 7311
    },
    {
      "epoch": 1.1446462116468379,
      "grad_norm": 1.271528959274292,
      "learning_rate": 0.00013379765395894427,
      "loss": 0.6008,
      "step": 7312
    },
    {
      "epoch": 1.144802755165936,
      "grad_norm": 1.1284598112106323,
      "learning_rate": 0.00013377321603128052,
      "loss": 0.5822,
      "step": 7313
    },
    {
      "epoch": 1.1449592986850345,
      "grad_norm": 1.5115302801132202,
      "learning_rate": 0.0001337487781036168,
      "loss": 0.4663,
      "step": 7314
    },
    {
      "epoch": 1.1451158422041328,
      "grad_norm": 1.522741675376892,
      "learning_rate": 0.00013372434017595308,
      "loss": 0.6408,
      "step": 7315
    },
    {
      "epoch": 1.145272385723231,
      "grad_norm": 3.165752649307251,
      "learning_rate": 0.00013369990224828933,
      "loss": 1.0592,
      "step": 7316
    },
    {
      "epoch": 1.1454289292423294,
      "grad_norm": 1.4550795555114746,
      "learning_rate": 0.0001336754643206256,
      "loss": 0.4573,
      "step": 7317
    },
    {
      "epoch": 1.1455854727614276,
      "grad_norm": 1.3216426372528076,
      "learning_rate": 0.00013365102639296186,
      "loss": 0.7879,
      "step": 7318
    },
    {
      "epoch": 1.145742016280526,
      "grad_norm": 3.195171356201172,
      "learning_rate": 0.00013362658846529814,
      "loss": 0.52,
      "step": 7319
    },
    {
      "epoch": 1.1458985597996243,
      "grad_norm": 2.3996787071228027,
      "learning_rate": 0.0001336021505376344,
      "loss": 0.7727,
      "step": 7320
    },
    {
      "epoch": 1.1460551033187225,
      "grad_norm": 1.751764178276062,
      "learning_rate": 0.00013357771260997066,
      "loss": 0.8789,
      "step": 7321
    },
    {
      "epoch": 1.146211646837821,
      "grad_norm": 1.3786290884017944,
      "learning_rate": 0.00013355327468230692,
      "loss": 0.3855,
      "step": 7322
    },
    {
      "epoch": 1.1463681903569192,
      "grad_norm": 1.4035483598709106,
      "learning_rate": 0.0001335288367546432,
      "loss": 0.7553,
      "step": 7323
    },
    {
      "epoch": 1.1465247338760176,
      "grad_norm": 2.2080776691436768,
      "learning_rate": 0.00013350439882697945,
      "loss": 1.1153,
      "step": 7324
    },
    {
      "epoch": 1.1466812773951158,
      "grad_norm": 3.5789425373077393,
      "learning_rate": 0.00013347996089931572,
      "loss": 0.7042,
      "step": 7325
    },
    {
      "epoch": 1.146837820914214,
      "grad_norm": 2.5934054851531982,
      "learning_rate": 0.000133455522971652,
      "loss": 1.1949,
      "step": 7326
    },
    {
      "epoch": 1.1469943644333125,
      "grad_norm": 1.6251215934753418,
      "learning_rate": 0.00013343108504398825,
      "loss": 0.7631,
      "step": 7327
    },
    {
      "epoch": 1.1471509079524107,
      "grad_norm": 2.1703507900238037,
      "learning_rate": 0.0001334066471163245,
      "loss": 0.8629,
      "step": 7328
    },
    {
      "epoch": 1.1473074514715091,
      "grad_norm": 2.7989585399627686,
      "learning_rate": 0.00013338220918866078,
      "loss": 1.0605,
      "step": 7329
    },
    {
      "epoch": 1.1474639949906074,
      "grad_norm": 2.1508922576904297,
      "learning_rate": 0.00013335777126099706,
      "loss": 0.7024,
      "step": 7330
    },
    {
      "epoch": 1.1476205385097056,
      "grad_norm": 3.7469875812530518,
      "learning_rate": 0.0001333333333333333,
      "loss": 1.3555,
      "step": 7331
    },
    {
      "epoch": 1.147777082028804,
      "grad_norm": 2.172651767730713,
      "learning_rate": 0.0001333088954056696,
      "loss": 1.0349,
      "step": 7332
    },
    {
      "epoch": 1.1479336255479022,
      "grad_norm": 3.3047091960906982,
      "learning_rate": 0.00013328445747800584,
      "loss": 1.2347,
      "step": 7333
    },
    {
      "epoch": 1.1480901690670007,
      "grad_norm": 2.5102274417877197,
      "learning_rate": 0.00013326001955034212,
      "loss": 1.2959,
      "step": 7334
    },
    {
      "epoch": 1.148246712586099,
      "grad_norm": 3.89743971824646,
      "learning_rate": 0.0001332355816226784,
      "loss": 0.487,
      "step": 7335
    },
    {
      "epoch": 1.1484032561051973,
      "grad_norm": 2.789970874786377,
      "learning_rate": 0.00013321114369501465,
      "loss": 0.7017,
      "step": 7336
    },
    {
      "epoch": 1.1485597996242956,
      "grad_norm": 2.094162702560425,
      "learning_rate": 0.0001331867057673509,
      "loss": 0.3889,
      "step": 7337
    },
    {
      "epoch": 1.1487163431433938,
      "grad_norm": 2.4462733268737793,
      "learning_rate": 0.00013316226783968718,
      "loss": 1.0959,
      "step": 7338
    },
    {
      "epoch": 1.1488728866624922,
      "grad_norm": 0.5095207095146179,
      "learning_rate": 0.00013313782991202346,
      "loss": 0.2746,
      "step": 7339
    },
    {
      "epoch": 1.1490294301815904,
      "grad_norm": 0.5467092394828796,
      "learning_rate": 0.0001331133919843597,
      "loss": 0.196,
      "step": 7340
    },
    {
      "epoch": 1.1491859737006889,
      "grad_norm": 0.3943612277507782,
      "learning_rate": 0.000133088954056696,
      "loss": 0.2072,
      "step": 7341
    },
    {
      "epoch": 1.149342517219787,
      "grad_norm": 0.613156259059906,
      "learning_rate": 0.00013306451612903224,
      "loss": 0.2318,
      "step": 7342
    },
    {
      "epoch": 1.1494990607388855,
      "grad_norm": 0.4345967173576355,
      "learning_rate": 0.00013304007820136852,
      "loss": 0.1638,
      "step": 7343
    },
    {
      "epoch": 1.1496556042579837,
      "grad_norm": 0.9504840970039368,
      "learning_rate": 0.0001330156402737048,
      "loss": 0.2764,
      "step": 7344
    },
    {
      "epoch": 1.149812147777082,
      "grad_norm": 0.6306407451629639,
      "learning_rate": 0.00013299120234604105,
      "loss": 0.2428,
      "step": 7345
    },
    {
      "epoch": 1.1499686912961804,
      "grad_norm": 0.6337870955467224,
      "learning_rate": 0.0001329667644183773,
      "loss": 0.315,
      "step": 7346
    },
    {
      "epoch": 1.1501252348152786,
      "grad_norm": 0.47224870324134827,
      "learning_rate": 0.00013294232649071358,
      "loss": 0.2152,
      "step": 7347
    },
    {
      "epoch": 1.150281778334377,
      "grad_norm": 0.8735085725784302,
      "learning_rate": 0.00013291788856304983,
      "loss": 0.3225,
      "step": 7348
    },
    {
      "epoch": 1.1504383218534753,
      "grad_norm": 0.5820073485374451,
      "learning_rate": 0.0001328934506353861,
      "loss": 0.1715,
      "step": 7349
    },
    {
      "epoch": 1.1505948653725735,
      "grad_norm": 1.1216922998428345,
      "learning_rate": 0.00013286901270772238,
      "loss": 0.3605,
      "step": 7350
    },
    {
      "epoch": 1.150751408891672,
      "grad_norm": 1.0515015125274658,
      "learning_rate": 0.00013284457478005864,
      "loss": 0.381,
      "step": 7351
    },
    {
      "epoch": 1.1509079524107702,
      "grad_norm": 0.727737545967102,
      "learning_rate": 0.0001328201368523949,
      "loss": 0.3203,
      "step": 7352
    },
    {
      "epoch": 1.1510644959298686,
      "grad_norm": 1.1194623708724976,
      "learning_rate": 0.00013279569892473117,
      "loss": 0.3155,
      "step": 7353
    },
    {
      "epoch": 1.1512210394489668,
      "grad_norm": 0.822720468044281,
      "learning_rate": 0.00013277126099706744,
      "loss": 0.3207,
      "step": 7354
    },
    {
      "epoch": 1.151377582968065,
      "grad_norm": 1.256633996963501,
      "learning_rate": 0.0001327468230694037,
      "loss": 0.4093,
      "step": 7355
    },
    {
      "epoch": 1.1515341264871635,
      "grad_norm": 1.5676660537719727,
      "learning_rate": 0.00013272238514173997,
      "loss": 0.4382,
      "step": 7356
    },
    {
      "epoch": 1.1516906700062617,
      "grad_norm": 1.2648556232452393,
      "learning_rate": 0.00013269794721407622,
      "loss": 0.5514,
      "step": 7357
    },
    {
      "epoch": 1.1518472135253601,
      "grad_norm": 1.1596871614456177,
      "learning_rate": 0.0001326735092864125,
      "loss": 0.5204,
      "step": 7358
    },
    {
      "epoch": 1.1520037570444583,
      "grad_norm": 1.1163748502731323,
      "learning_rate": 0.00013264907135874878,
      "loss": 0.3169,
      "step": 7359
    },
    {
      "epoch": 1.1521603005635566,
      "grad_norm": 1.4104461669921875,
      "learning_rate": 0.00013262463343108503,
      "loss": 0.5897,
      "step": 7360
    },
    {
      "epoch": 1.152316844082655,
      "grad_norm": 0.9608981013298035,
      "learning_rate": 0.00013260019550342128,
      "loss": 0.5129,
      "step": 7361
    },
    {
      "epoch": 1.1524733876017532,
      "grad_norm": 1.5088484287261963,
      "learning_rate": 0.00013257575757575756,
      "loss": 0.6277,
      "step": 7362
    },
    {
      "epoch": 1.1526299311208517,
      "grad_norm": 1.8139734268188477,
      "learning_rate": 0.00013255131964809384,
      "loss": 0.6778,
      "step": 7363
    },
    {
      "epoch": 1.1527864746399499,
      "grad_norm": 1.7767802476882935,
      "learning_rate": 0.0001325268817204301,
      "loss": 0.6887,
      "step": 7364
    },
    {
      "epoch": 1.152943018159048,
      "grad_norm": 1.6628837585449219,
      "learning_rate": 0.00013250244379276637,
      "loss": 0.3635,
      "step": 7365
    },
    {
      "epoch": 1.1530995616781465,
      "grad_norm": 1.664116621017456,
      "learning_rate": 0.00013247800586510262,
      "loss": 0.4019,
      "step": 7366
    },
    {
      "epoch": 1.1532561051972448,
      "grad_norm": 1.989309549331665,
      "learning_rate": 0.0001324535679374389,
      "loss": 0.9246,
      "step": 7367
    },
    {
      "epoch": 1.1534126487163432,
      "grad_norm": 9.230467796325684,
      "learning_rate": 0.00013242913000977515,
      "loss": 0.8514,
      "step": 7368
    },
    {
      "epoch": 1.1535691922354414,
      "grad_norm": 1.3691836595535278,
      "learning_rate": 0.00013240469208211143,
      "loss": 0.4943,
      "step": 7369
    },
    {
      "epoch": 1.1537257357545398,
      "grad_norm": 1.0850152969360352,
      "learning_rate": 0.00013238025415444768,
      "loss": 0.5152,
      "step": 7370
    },
    {
      "epoch": 1.153882279273638,
      "grad_norm": 2.6206274032592773,
      "learning_rate": 0.00013235581622678396,
      "loss": 0.5313,
      "step": 7371
    },
    {
      "epoch": 1.1540388227927363,
      "grad_norm": 1.9368460178375244,
      "learning_rate": 0.0001323313782991202,
      "loss": 0.5789,
      "step": 7372
    },
    {
      "epoch": 1.1541953663118347,
      "grad_norm": 3.7106120586395264,
      "learning_rate": 0.0001323069403714565,
      "loss": 1.0362,
      "step": 7373
    },
    {
      "epoch": 1.154351909830933,
      "grad_norm": 1.4057819843292236,
      "learning_rate": 0.00013228250244379277,
      "loss": 0.5185,
      "step": 7374
    },
    {
      "epoch": 1.1545084533500314,
      "grad_norm": 2.9499166011810303,
      "learning_rate": 0.00013225806451612902,
      "loss": 1.1545,
      "step": 7375
    },
    {
      "epoch": 1.1546649968691296,
      "grad_norm": 2.847531318664551,
      "learning_rate": 0.00013223362658846527,
      "loss": 0.9092,
      "step": 7376
    },
    {
      "epoch": 1.154821540388228,
      "grad_norm": 2.4017632007598877,
      "learning_rate": 0.00013220918866080155,
      "loss": 1.0395,
      "step": 7377
    },
    {
      "epoch": 1.1549780839073263,
      "grad_norm": 3.083125591278076,
      "learning_rate": 0.00013218475073313783,
      "loss": 0.9163,
      "step": 7378
    },
    {
      "epoch": 1.1551346274264245,
      "grad_norm": 1.6642545461654663,
      "learning_rate": 0.00013216031280547408,
      "loss": 0.7357,
      "step": 7379
    },
    {
      "epoch": 1.155291170945523,
      "grad_norm": 2.6173577308654785,
      "learning_rate": 0.00013213587487781036,
      "loss": 1.3048,
      "step": 7380
    },
    {
      "epoch": 1.1554477144646211,
      "grad_norm": 4.377566337585449,
      "learning_rate": 0.0001321114369501466,
      "loss": 1.3639,
      "step": 7381
    },
    {
      "epoch": 1.1556042579837196,
      "grad_norm": 2.4152894020080566,
      "learning_rate": 0.00013208699902248289,
      "loss": 1.1259,
      "step": 7382
    },
    {
      "epoch": 1.1557608015028178,
      "grad_norm": 2.7679460048675537,
      "learning_rate": 0.00013206256109481916,
      "loss": 0.9502,
      "step": 7383
    },
    {
      "epoch": 1.155917345021916,
      "grad_norm": 1.6410295963287354,
      "learning_rate": 0.00013203812316715541,
      "loss": 0.563,
      "step": 7384
    },
    {
      "epoch": 1.1560738885410144,
      "grad_norm": 1.5366709232330322,
      "learning_rate": 0.00013201368523949167,
      "loss": 0.3714,
      "step": 7385
    },
    {
      "epoch": 1.1562304320601127,
      "grad_norm": 2.130624771118164,
      "learning_rate": 0.00013198924731182794,
      "loss": 0.5333,
      "step": 7386
    },
    {
      "epoch": 1.156386975579211,
      "grad_norm": 0.982833743095398,
      "learning_rate": 0.00013196480938416422,
      "loss": 0.1698,
      "step": 7387
    },
    {
      "epoch": 1.1565435190983093,
      "grad_norm": 3.6173794269561768,
      "learning_rate": 0.00013194037145650047,
      "loss": 1.0673,
      "step": 7388
    },
    {
      "epoch": 1.1567000626174075,
      "grad_norm": 0.45585983991622925,
      "learning_rate": 0.00013191593352883675,
      "loss": 0.2432,
      "step": 7389
    },
    {
      "epoch": 1.156856606136506,
      "grad_norm": 0.5601482391357422,
      "learning_rate": 0.000131891495601173,
      "loss": 0.2334,
      "step": 7390
    },
    {
      "epoch": 1.1570131496556042,
      "grad_norm": 0.5899141430854797,
      "learning_rate": 0.00013186705767350928,
      "loss": 0.2739,
      "step": 7391
    },
    {
      "epoch": 1.1571696931747026,
      "grad_norm": 0.6862410306930542,
      "learning_rate": 0.00013184261974584553,
      "loss": 0.1941,
      "step": 7392
    },
    {
      "epoch": 1.1573262366938009,
      "grad_norm": 0.8341695070266724,
      "learning_rate": 0.0001318181818181818,
      "loss": 0.2567,
      "step": 7393
    },
    {
      "epoch": 1.157482780212899,
      "grad_norm": 0.7754533290863037,
      "learning_rate": 0.00013179374389051806,
      "loss": 0.3397,
      "step": 7394
    },
    {
      "epoch": 1.1576393237319975,
      "grad_norm": 1.1983238458633423,
      "learning_rate": 0.00013176930596285434,
      "loss": 0.3212,
      "step": 7395
    },
    {
      "epoch": 1.1577958672510957,
      "grad_norm": 0.5611344575881958,
      "learning_rate": 0.0001317448680351906,
      "loss": 0.2555,
      "step": 7396
    },
    {
      "epoch": 1.1579524107701942,
      "grad_norm": 0.4834452271461487,
      "learning_rate": 0.00013172043010752687,
      "loss": 0.2091,
      "step": 7397
    },
    {
      "epoch": 1.1581089542892924,
      "grad_norm": 1.169364333152771,
      "learning_rate": 0.00013169599217986315,
      "loss": 0.3318,
      "step": 7398
    },
    {
      "epoch": 1.1582654978083906,
      "grad_norm": 0.8401133418083191,
      "learning_rate": 0.0001316715542521994,
      "loss": 0.1914,
      "step": 7399
    },
    {
      "epoch": 1.158422041327489,
      "grad_norm": 1.0948725938796997,
      "learning_rate": 0.00013164711632453565,
      "loss": 0.4167,
      "step": 7400
    },
    {
      "epoch": 1.1585785848465873,
      "grad_norm": 0.6502231955528259,
      "learning_rate": 0.00013162267839687193,
      "loss": 0.3204,
      "step": 7401
    },
    {
      "epoch": 1.1587351283656857,
      "grad_norm": 0.8095880746841431,
      "learning_rate": 0.0001315982404692082,
      "loss": 0.2374,
      "step": 7402
    },
    {
      "epoch": 1.158891671884784,
      "grad_norm": 1.0704963207244873,
      "learning_rate": 0.00013157380254154446,
      "loss": 0.3654,
      "step": 7403
    },
    {
      "epoch": 1.1590482154038824,
      "grad_norm": 1.0541001558303833,
      "learning_rate": 0.00013154936461388074,
      "loss": 0.3836,
      "step": 7404
    },
    {
      "epoch": 1.1592047589229806,
      "grad_norm": 0.9816923141479492,
      "learning_rate": 0.000131524926686217,
      "loss": 0.4202,
      "step": 7405
    },
    {
      "epoch": 1.159361302442079,
      "grad_norm": 1.2711502313613892,
      "learning_rate": 0.00013150048875855327,
      "loss": 0.2639,
      "step": 7406
    },
    {
      "epoch": 1.1595178459611772,
      "grad_norm": 1.2869318723678589,
      "learning_rate": 0.00013147605083088955,
      "loss": 0.3089,
      "step": 7407
    },
    {
      "epoch": 1.1596743894802755,
      "grad_norm": 1.1554820537567139,
      "learning_rate": 0.0001314516129032258,
      "loss": 0.3307,
      "step": 7408
    },
    {
      "epoch": 1.159830932999374,
      "grad_norm": 1.7759408950805664,
      "learning_rate": 0.00013142717497556205,
      "loss": 0.8516,
      "step": 7409
    },
    {
      "epoch": 1.1599874765184721,
      "grad_norm": 1.0223851203918457,
      "learning_rate": 0.00013140273704789833,
      "loss": 0.3851,
      "step": 7410
    },
    {
      "epoch": 1.1601440200375706,
      "grad_norm": 1.761975646018982,
      "learning_rate": 0.0001313782991202346,
      "loss": 0.4526,
      "step": 7411
    },
    {
      "epoch": 1.1603005635566688,
      "grad_norm": 0.7343161702156067,
      "learning_rate": 0.00013135386119257086,
      "loss": 0.43,
      "step": 7412
    },
    {
      "epoch": 1.160457107075767,
      "grad_norm": 1.6459717750549316,
      "learning_rate": 0.00013132942326490713,
      "loss": 0.7205,
      "step": 7413
    },
    {
      "epoch": 1.1606136505948654,
      "grad_norm": 1.6329326629638672,
      "learning_rate": 0.00013130498533724339,
      "loss": 0.5562,
      "step": 7414
    },
    {
      "epoch": 1.1607701941139636,
      "grad_norm": 2.7363922595977783,
      "learning_rate": 0.00013128054740957964,
      "loss": 0.7984,
      "step": 7415
    },
    {
      "epoch": 1.160926737633062,
      "grad_norm": 1.381791114807129,
      "learning_rate": 0.00013125610948191592,
      "loss": 0.5795,
      "step": 7416
    },
    {
      "epoch": 1.1610832811521603,
      "grad_norm": 1.932962417602539,
      "learning_rate": 0.0001312316715542522,
      "loss": 0.3651,
      "step": 7417
    },
    {
      "epoch": 1.1612398246712585,
      "grad_norm": 1.733339786529541,
      "learning_rate": 0.00013120723362658845,
      "loss": 0.4857,
      "step": 7418
    },
    {
      "epoch": 1.161396368190357,
      "grad_norm": 1.2550915479660034,
      "learning_rate": 0.00013118279569892472,
      "loss": 0.5785,
      "step": 7419
    },
    {
      "epoch": 1.1615529117094552,
      "grad_norm": 1.603405475616455,
      "learning_rate": 0.00013115835777126097,
      "loss": 0.4028,
      "step": 7420
    },
    {
      "epoch": 1.1617094552285536,
      "grad_norm": 0.9606602191925049,
      "learning_rate": 0.00013113391984359725,
      "loss": 0.4517,
      "step": 7421
    },
    {
      "epoch": 1.1618659987476518,
      "grad_norm": 2.063321590423584,
      "learning_rate": 0.00013110948191593353,
      "loss": 0.6706,
      "step": 7422
    },
    {
      "epoch": 1.16202254226675,
      "grad_norm": 2.597672462463379,
      "learning_rate": 0.00013108504398826978,
      "loss": 0.7549,
      "step": 7423
    },
    {
      "epoch": 1.1621790857858485,
      "grad_norm": 3.000110149383545,
      "learning_rate": 0.00013106060606060603,
      "loss": 0.9992,
      "step": 7424
    },
    {
      "epoch": 1.1623356293049467,
      "grad_norm": 2.6195647716522217,
      "learning_rate": 0.0001310361681329423,
      "loss": 1.199,
      "step": 7425
    },
    {
      "epoch": 1.1624921728240452,
      "grad_norm": 3.0307538509368896,
      "learning_rate": 0.0001310117302052786,
      "loss": 0.9195,
      "step": 7426
    },
    {
      "epoch": 1.1626487163431434,
      "grad_norm": 3.3168554306030273,
      "learning_rate": 0.00013098729227761484,
      "loss": 1.1316,
      "step": 7427
    },
    {
      "epoch": 1.1628052598622416,
      "grad_norm": 2.105064630508423,
      "learning_rate": 0.00013096285434995112,
      "loss": 1.2571,
      "step": 7428
    },
    {
      "epoch": 1.16296180338134,
      "grad_norm": 3.6035208702087402,
      "learning_rate": 0.00013093841642228737,
      "loss": 0.9023,
      "step": 7429
    },
    {
      "epoch": 1.1631183469004382,
      "grad_norm": 2.2905807495117188,
      "learning_rate": 0.00013091397849462365,
      "loss": 1.6656,
      "step": 7430
    },
    {
      "epoch": 1.1632748904195367,
      "grad_norm": 2.3140320777893066,
      "learning_rate": 0.00013088954056695993,
      "loss": 1.6509,
      "step": 7431
    },
    {
      "epoch": 1.163431433938635,
      "grad_norm": 4.755015850067139,
      "learning_rate": 0.00013086510263929618,
      "loss": 1.0378,
      "step": 7432
    },
    {
      "epoch": 1.1635879774577333,
      "grad_norm": 1.8031630516052246,
      "learning_rate": 0.00013084066471163243,
      "loss": 0.926,
      "step": 7433
    },
    {
      "epoch": 1.1637445209768316,
      "grad_norm": 1.6042096614837646,
      "learning_rate": 0.0001308162267839687,
      "loss": 0.3366,
      "step": 7434
    },
    {
      "epoch": 1.1639010644959298,
      "grad_norm": 1.8584673404693604,
      "learning_rate": 0.000130791788856305,
      "loss": 0.8145,
      "step": 7435
    },
    {
      "epoch": 1.1640576080150282,
      "grad_norm": 2.789856433868408,
      "learning_rate": 0.00013076735092864124,
      "loss": 0.6222,
      "step": 7436
    },
    {
      "epoch": 1.1642141515341264,
      "grad_norm": 2.3871476650238037,
      "learning_rate": 0.00013074291300097752,
      "loss": 0.7207,
      "step": 7437
    },
    {
      "epoch": 1.1643706950532249,
      "grad_norm": 2.0751960277557373,
      "learning_rate": 0.00013071847507331377,
      "loss": 0.6828,
      "step": 7438
    },
    {
      "epoch": 1.164527238572323,
      "grad_norm": 0.985443115234375,
      "learning_rate": 0.00013069403714565002,
      "loss": 0.183,
      "step": 7439
    },
    {
      "epoch": 1.1646837820914215,
      "grad_norm": 0.46885955333709717,
      "learning_rate": 0.0001306695992179863,
      "loss": 0.2147,
      "step": 7440
    },
    {
      "epoch": 1.1648403256105198,
      "grad_norm": 0.5241273641586304,
      "learning_rate": 0.00013064516129032258,
      "loss": 0.2076,
      "step": 7441
    },
    {
      "epoch": 1.164996869129618,
      "grad_norm": 0.5375959873199463,
      "learning_rate": 0.00013062072336265883,
      "loss": 0.1665,
      "step": 7442
    },
    {
      "epoch": 1.1651534126487164,
      "grad_norm": 0.7745464444160461,
      "learning_rate": 0.0001305962854349951,
      "loss": 0.2586,
      "step": 7443
    },
    {
      "epoch": 1.1653099561678146,
      "grad_norm": 0.511017382144928,
      "learning_rate": 0.00013057184750733136,
      "loss": 0.2376,
      "step": 7444
    },
    {
      "epoch": 1.165466499686913,
      "grad_norm": 0.8513729572296143,
      "learning_rate": 0.00013054740957966764,
      "loss": 0.3418,
      "step": 7445
    },
    {
      "epoch": 1.1656230432060113,
      "grad_norm": 0.9605493545532227,
      "learning_rate": 0.00013052297165200391,
      "loss": 0.3008,
      "step": 7446
    },
    {
      "epoch": 1.1657795867251095,
      "grad_norm": 0.7531507015228271,
      "learning_rate": 0.00013049853372434016,
      "loss": 0.2765,
      "step": 7447
    },
    {
      "epoch": 1.165936130244208,
      "grad_norm": 0.8994906544685364,
      "learning_rate": 0.00013047409579667642,
      "loss": 0.3855,
      "step": 7448
    },
    {
      "epoch": 1.1660926737633062,
      "grad_norm": 0.48256492614746094,
      "learning_rate": 0.0001304496578690127,
      "loss": 0.2462,
      "step": 7449
    },
    {
      "epoch": 1.1662492172824046,
      "grad_norm": 1.1383583545684814,
      "learning_rate": 0.00013042521994134897,
      "loss": 0.3641,
      "step": 7450
    },
    {
      "epoch": 1.1664057608015028,
      "grad_norm": 1.2914742231369019,
      "learning_rate": 0.00013040078201368522,
      "loss": 0.2236,
      "step": 7451
    },
    {
      "epoch": 1.166562304320601,
      "grad_norm": 0.5207766890525818,
      "learning_rate": 0.0001303763440860215,
      "loss": 0.2485,
      "step": 7452
    },
    {
      "epoch": 1.1667188478396995,
      "grad_norm": 1.1940803527832031,
      "learning_rate": 0.00013035190615835775,
      "loss": 0.3618,
      "step": 7453
    },
    {
      "epoch": 1.1668753913587977,
      "grad_norm": 2.0108392238616943,
      "learning_rate": 0.00013032746823069403,
      "loss": 0.7458,
      "step": 7454
    },
    {
      "epoch": 1.1670319348778961,
      "grad_norm": 1.687880277633667,
      "learning_rate": 0.0001303030303030303,
      "loss": 0.22,
      "step": 7455
    },
    {
      "epoch": 1.1671884783969944,
      "grad_norm": 2.061474084854126,
      "learning_rate": 0.00013027859237536656,
      "loss": 0.3396,
      "step": 7456
    },
    {
      "epoch": 1.1673450219160926,
      "grad_norm": 0.789428174495697,
      "learning_rate": 0.0001302541544477028,
      "loss": 0.3342,
      "step": 7457
    },
    {
      "epoch": 1.167501565435191,
      "grad_norm": 1.0013548135757446,
      "learning_rate": 0.0001302297165200391,
      "loss": 0.4941,
      "step": 7458
    },
    {
      "epoch": 1.1676581089542892,
      "grad_norm": 1.2764099836349487,
      "learning_rate": 0.00013020527859237534,
      "loss": 0.4534,
      "step": 7459
    },
    {
      "epoch": 1.1678146524733877,
      "grad_norm": 1.8446581363677979,
      "learning_rate": 0.00013018084066471162,
      "loss": 0.9874,
      "step": 7460
    },
    {
      "epoch": 1.1679711959924859,
      "grad_norm": 1.0535941123962402,
      "learning_rate": 0.0001301564027370479,
      "loss": 0.2257,
      "step": 7461
    },
    {
      "epoch": 1.168127739511584,
      "grad_norm": 2.2027220726013184,
      "learning_rate": 0.00013013196480938415,
      "loss": 0.9646,
      "step": 7462
    },
    {
      "epoch": 1.1682842830306825,
      "grad_norm": 1.3150688409805298,
      "learning_rate": 0.0001301075268817204,
      "loss": 0.5224,
      "step": 7463
    },
    {
      "epoch": 1.1684408265497808,
      "grad_norm": 1.6204612255096436,
      "learning_rate": 0.00013008308895405668,
      "loss": 0.5533,
      "step": 7464
    },
    {
      "epoch": 1.1685973700688792,
      "grad_norm": 7.107356548309326,
      "learning_rate": 0.00013005865102639296,
      "loss": 1.438,
      "step": 7465
    },
    {
      "epoch": 1.1687539135879774,
      "grad_norm": 2.1435375213623047,
      "learning_rate": 0.0001300342130987292,
      "loss": 0.5204,
      "step": 7466
    },
    {
      "epoch": 1.1689104571070759,
      "grad_norm": 2.206997871398926,
      "learning_rate": 0.0001300097751710655,
      "loss": 0.8244,
      "step": 7467
    },
    {
      "epoch": 1.169067000626174,
      "grad_norm": 2.2694783210754395,
      "learning_rate": 0.00012998533724340174,
      "loss": 0.5618,
      "step": 7468
    },
    {
      "epoch": 1.1692235441452723,
      "grad_norm": 2.4387333393096924,
      "learning_rate": 0.00012996089931573802,
      "loss": 1.1019,
      "step": 7469
    },
    {
      "epoch": 1.1693800876643707,
      "grad_norm": 1.3861206769943237,
      "learning_rate": 0.0001299364613880743,
      "loss": 0.4208,
      "step": 7470
    },
    {
      "epoch": 1.169536631183469,
      "grad_norm": 1.4845746755599976,
      "learning_rate": 0.00012991202346041055,
      "loss": 0.6122,
      "step": 7471
    },
    {
      "epoch": 1.1696931747025674,
      "grad_norm": 3.262490749359131,
      "learning_rate": 0.0001298875855327468,
      "loss": 0.7805,
      "step": 7472
    },
    {
      "epoch": 1.1698497182216656,
      "grad_norm": 2.262986421585083,
      "learning_rate": 0.00012986314760508308,
      "loss": 0.5199,
      "step": 7473
    },
    {
      "epoch": 1.170006261740764,
      "grad_norm": 1.8733512163162231,
      "learning_rate": 0.00012983870967741936,
      "loss": 0.8374,
      "step": 7474
    },
    {
      "epoch": 1.1701628052598623,
      "grad_norm": 1.8209384679794312,
      "learning_rate": 0.0001298142717497556,
      "loss": 0.7151,
      "step": 7475
    },
    {
      "epoch": 1.1703193487789605,
      "grad_norm": 1.9471930265426636,
      "learning_rate": 0.00012978983382209188,
      "loss": 0.9015,
      "step": 7476
    },
    {
      "epoch": 1.170475892298059,
      "grad_norm": 2.2029271125793457,
      "learning_rate": 0.00012976539589442814,
      "loss": 0.5801,
      "step": 7477
    },
    {
      "epoch": 1.1706324358171571,
      "grad_norm": 2.9882936477661133,
      "learning_rate": 0.00012974095796676441,
      "loss": 1.2584,
      "step": 7478
    },
    {
      "epoch": 1.1707889793362556,
      "grad_norm": 2.2298552989959717,
      "learning_rate": 0.0001297165200391007,
      "loss": 0.8488,
      "step": 7479
    },
    {
      "epoch": 1.1709455228553538,
      "grad_norm": 2.1507680416107178,
      "learning_rate": 0.00012969208211143694,
      "loss": 0.9486,
      "step": 7480
    },
    {
      "epoch": 1.171102066374452,
      "grad_norm": 2.448796033859253,
      "learning_rate": 0.0001296676441837732,
      "loss": 1.1718,
      "step": 7481
    },
    {
      "epoch": 1.1712586098935505,
      "grad_norm": 1.807196021080017,
      "learning_rate": 0.00012964320625610947,
      "loss": 1.2157,
      "step": 7482
    },
    {
      "epoch": 1.1714151534126487,
      "grad_norm": 1.9326443672180176,
      "learning_rate": 0.00012961876832844572,
      "loss": 1.6794,
      "step": 7483
    },
    {
      "epoch": 1.1715716969317471,
      "grad_norm": 2.4775540828704834,
      "learning_rate": 0.000129594330400782,
      "loss": 1.0942,
      "step": 7484
    },
    {
      "epoch": 1.1717282404508453,
      "grad_norm": 3.0577571392059326,
      "learning_rate": 0.00012956989247311828,
      "loss": 0.9137,
      "step": 7485
    },
    {
      "epoch": 1.1718847839699436,
      "grad_norm": 1.4351956844329834,
      "learning_rate": 0.00012954545454545453,
      "loss": 1.0002,
      "step": 7486
    },
    {
      "epoch": 1.172041327489042,
      "grad_norm": 2.7734296321868896,
      "learning_rate": 0.00012952101661779078,
      "loss": 0.6121,
      "step": 7487
    },
    {
      "epoch": 1.1721978710081402,
      "grad_norm": 3.1361334323883057,
      "learning_rate": 0.00012949657869012706,
      "loss": 0.6366,
      "step": 7488
    },
    {
      "epoch": 1.1723544145272387,
      "grad_norm": 0.458670049905777,
      "learning_rate": 0.00012947214076246334,
      "loss": 0.2406,
      "step": 7489
    },
    {
      "epoch": 1.1725109580463369,
      "grad_norm": 0.4541315734386444,
      "learning_rate": 0.0001294477028347996,
      "loss": 0.1887,
      "step": 7490
    },
    {
      "epoch": 1.172667501565435,
      "grad_norm": 1.0763859748840332,
      "learning_rate": 0.00012942326490713587,
      "loss": 0.6726,
      "step": 7491
    },
    {
      "epoch": 1.1728240450845335,
      "grad_norm": 0.6392644643783569,
      "learning_rate": 0.00012939882697947212,
      "loss": 0.2505,
      "step": 7492
    },
    {
      "epoch": 1.1729805886036317,
      "grad_norm": 0.6221238374710083,
      "learning_rate": 0.0001293743890518084,
      "loss": 0.2844,
      "step": 7493
    },
    {
      "epoch": 1.1731371321227302,
      "grad_norm": 0.6850442886352539,
      "learning_rate": 0.00012934995112414468,
      "loss": 0.2483,
      "step": 7494
    },
    {
      "epoch": 1.1732936756418284,
      "grad_norm": 0.808659553527832,
      "learning_rate": 0.00012932551319648093,
      "loss": 0.2442,
      "step": 7495
    },
    {
      "epoch": 1.1734502191609266,
      "grad_norm": 0.7983716130256653,
      "learning_rate": 0.00012930107526881718,
      "loss": 0.3581,
      "step": 7496
    },
    {
      "epoch": 1.173606762680025,
      "grad_norm": 0.5840036273002625,
      "learning_rate": 0.00012927663734115346,
      "loss": 0.4542,
      "step": 7497
    },
    {
      "epoch": 1.1737633061991233,
      "grad_norm": 0.5525220036506653,
      "learning_rate": 0.00012925219941348974,
      "loss": 0.238,
      "step": 7498
    },
    {
      "epoch": 1.1739198497182217,
      "grad_norm": 1.0121021270751953,
      "learning_rate": 0.000129227761485826,
      "loss": 0.4005,
      "step": 7499
    },
    {
      "epoch": 1.17407639323732,
      "grad_norm": 2.506730556488037,
      "learning_rate": 0.00012920332355816227,
      "loss": 0.3244,
      "step": 7500
    },
    {
      "epoch": 1.1742329367564184,
      "grad_norm": 0.6981834173202515,
      "learning_rate": 0.00012917888563049852,
      "loss": 0.2825,
      "step": 7501
    },
    {
      "epoch": 1.1743894802755166,
      "grad_norm": 1.0318002700805664,
      "learning_rate": 0.0001291544477028348,
      "loss": 0.2646,
      "step": 7502
    },
    {
      "epoch": 1.1745460237946148,
      "grad_norm": 1.0581806898117065,
      "learning_rate": 0.00012913000977517108,
      "loss": 0.3837,
      "step": 7503
    },
    {
      "epoch": 1.1747025673137133,
      "grad_norm": 1.221384048461914,
      "learning_rate": 0.00012910557184750733,
      "loss": 0.3138,
      "step": 7504
    },
    {
      "epoch": 1.1748591108328115,
      "grad_norm": 0.9835034608840942,
      "learning_rate": 0.00012908113391984358,
      "loss": 0.241,
      "step": 7505
    },
    {
      "epoch": 1.17501565435191,
      "grad_norm": 0.9940537810325623,
      "learning_rate": 0.00012905669599217986,
      "loss": 0.2297,
      "step": 7506
    },
    {
      "epoch": 1.1751721978710081,
      "grad_norm": 0.7861664891242981,
      "learning_rate": 0.0001290322580645161,
      "loss": 0.4561,
      "step": 7507
    },
    {
      "epoch": 1.1753287413901066,
      "grad_norm": 1.2852720022201538,
      "learning_rate": 0.00012900782013685239,
      "loss": 0.6284,
      "step": 7508
    },
    {
      "epoch": 1.1754852849092048,
      "grad_norm": 1.645680546760559,
      "learning_rate": 0.00012898338220918866,
      "loss": 0.5136,
      "step": 7509
    },
    {
      "epoch": 1.175641828428303,
      "grad_norm": 1.9410187005996704,
      "learning_rate": 0.00012895894428152492,
      "loss": 0.4755,
      "step": 7510
    },
    {
      "epoch": 1.1757983719474014,
      "grad_norm": 0.807033360004425,
      "learning_rate": 0.00012893450635386117,
      "loss": 0.3838,
      "step": 7511
    },
    {
      "epoch": 1.1759549154664997,
      "grad_norm": 3.5870234966278076,
      "learning_rate": 0.00012891006842619744,
      "loss": 0.3806,
      "step": 7512
    },
    {
      "epoch": 1.176111458985598,
      "grad_norm": 1.55286705493927,
      "learning_rate": 0.00012888563049853372,
      "loss": 0.443,
      "step": 7513
    },
    {
      "epoch": 1.1762680025046963,
      "grad_norm": 1.928011417388916,
      "learning_rate": 0.00012886119257086997,
      "loss": 0.8595,
      "step": 7514
    },
    {
      "epoch": 1.1764245460237945,
      "grad_norm": 1.085798978805542,
      "learning_rate": 0.00012883675464320625,
      "loss": 0.503,
      "step": 7515
    },
    {
      "epoch": 1.176581089542893,
      "grad_norm": 1.993884801864624,
      "learning_rate": 0.0001288123167155425,
      "loss": 0.4457,
      "step": 7516
    },
    {
      "epoch": 1.1767376330619912,
      "grad_norm": 1.3057135343551636,
      "learning_rate": 0.00012878787878787878,
      "loss": 0.6351,
      "step": 7517
    },
    {
      "epoch": 1.1768941765810896,
      "grad_norm": 2.9822328090667725,
      "learning_rate": 0.00012876344086021506,
      "loss": 0.8021,
      "step": 7518
    },
    {
      "epoch": 1.1770507201001879,
      "grad_norm": 1.463084101676941,
      "learning_rate": 0.0001287390029325513,
      "loss": 0.5327,
      "step": 7519
    },
    {
      "epoch": 1.177207263619286,
      "grad_norm": 1.026001214981079,
      "learning_rate": 0.00012871456500488756,
      "loss": 0.4669,
      "step": 7520
    },
    {
      "epoch": 1.1773638071383845,
      "grad_norm": 0.9538169503211975,
      "learning_rate": 0.00012869012707722384,
      "loss": 0.4046,
      "step": 7521
    },
    {
      "epoch": 1.1775203506574827,
      "grad_norm": 1.1642636060714722,
      "learning_rate": 0.00012866568914956012,
      "loss": 0.5993,
      "step": 7522
    },
    {
      "epoch": 1.1776768941765812,
      "grad_norm": 1.9186804294586182,
      "learning_rate": 0.00012864125122189637,
      "loss": 0.7352,
      "step": 7523
    },
    {
      "epoch": 1.1778334376956794,
      "grad_norm": 2.1181206703186035,
      "learning_rate": 0.00012861681329423265,
      "loss": 0.8638,
      "step": 7524
    },
    {
      "epoch": 1.1779899812147776,
      "grad_norm": 2.4149346351623535,
      "learning_rate": 0.0001285923753665689,
      "loss": 0.9885,
      "step": 7525
    },
    {
      "epoch": 1.178146524733876,
      "grad_norm": 1.8316506147384644,
      "learning_rate": 0.00012856793743890518,
      "loss": 0.8638,
      "step": 7526
    },
    {
      "epoch": 1.1783030682529743,
      "grad_norm": 1.9527243375778198,
      "learning_rate": 0.00012854349951124143,
      "loss": 0.8473,
      "step": 7527
    },
    {
      "epoch": 1.1784596117720727,
      "grad_norm": 2.2546331882476807,
      "learning_rate": 0.0001285190615835777,
      "loss": 1.134,
      "step": 7528
    },
    {
      "epoch": 1.178616155291171,
      "grad_norm": 1.679229497909546,
      "learning_rate": 0.00012849462365591396,
      "loss": 1.0515,
      "step": 7529
    },
    {
      "epoch": 1.1787726988102691,
      "grad_norm": 7.857295036315918,
      "learning_rate": 0.00012847018572825024,
      "loss": 1.3671,
      "step": 7530
    },
    {
      "epoch": 1.1789292423293676,
      "grad_norm": 2.6357522010803223,
      "learning_rate": 0.0001284457478005865,
      "loss": 1.3619,
      "step": 7531
    },
    {
      "epoch": 1.1790857858484658,
      "grad_norm": 4.2092084884643555,
      "learning_rate": 0.00012842130987292277,
      "loss": 1.5904,
      "step": 7532
    },
    {
      "epoch": 1.1792423293675642,
      "grad_norm": 2.7293174266815186,
      "learning_rate": 0.00012839687194525905,
      "loss": 1.5906,
      "step": 7533
    },
    {
      "epoch": 1.1793988728866625,
      "grad_norm": 1.7189364433288574,
      "learning_rate": 0.0001283724340175953,
      "loss": 1.0352,
      "step": 7534
    },
    {
      "epoch": 1.179555416405761,
      "grad_norm": 3.9607129096984863,
      "learning_rate": 0.00012834799608993155,
      "loss": 0.4995,
      "step": 7535
    },
    {
      "epoch": 1.179711959924859,
      "grad_norm": 3.564509391784668,
      "learning_rate": 0.00012832355816226783,
      "loss": 0.8067,
      "step": 7536
    },
    {
      "epoch": 1.1798685034439573,
      "grad_norm": 2.912269115447998,
      "learning_rate": 0.0001282991202346041,
      "loss": 0.8116,
      "step": 7537
    },
    {
      "epoch": 1.1800250469630558,
      "grad_norm": 1.9242192506790161,
      "learning_rate": 0.00012827468230694036,
      "loss": 0.7333,
      "step": 7538
    },
    {
      "epoch": 1.180181590482154,
      "grad_norm": 0.3993604779243469,
      "learning_rate": 0.00012825024437927664,
      "loss": 0.1944,
      "step": 7539
    },
    {
      "epoch": 1.1803381340012524,
      "grad_norm": 0.4738137722015381,
      "learning_rate": 0.00012822580645161289,
      "loss": 0.1609,
      "step": 7540
    },
    {
      "epoch": 1.1804946775203506,
      "grad_norm": 0.41736552119255066,
      "learning_rate": 0.00012820136852394916,
      "loss": 0.2162,
      "step": 7541
    },
    {
      "epoch": 1.180651221039449,
      "grad_norm": 0.5973852276802063,
      "learning_rate": 0.00012817693059628544,
      "loss": 0.221,
      "step": 7542
    },
    {
      "epoch": 1.1808077645585473,
      "grad_norm": 0.42182105779647827,
      "learning_rate": 0.0001281524926686217,
      "loss": 0.1776,
      "step": 7543
    },
    {
      "epoch": 1.1809643080776455,
      "grad_norm": 1.2317596673965454,
      "learning_rate": 0.00012812805474095795,
      "loss": 0.2,
      "step": 7544
    },
    {
      "epoch": 1.181120851596744,
      "grad_norm": 0.7421743869781494,
      "learning_rate": 0.00012810361681329422,
      "loss": 0.2841,
      "step": 7545
    },
    {
      "epoch": 1.1812773951158422,
      "grad_norm": 0.7382256388664246,
      "learning_rate": 0.0001280791788856305,
      "loss": 0.2966,
      "step": 7546
    },
    {
      "epoch": 1.1814339386349406,
      "grad_norm": 0.6662998199462891,
      "learning_rate": 0.00012805474095796675,
      "loss": 0.2194,
      "step": 7547
    },
    {
      "epoch": 1.1815904821540388,
      "grad_norm": 0.5631325840950012,
      "learning_rate": 0.000128030303030303,
      "loss": 0.3308,
      "step": 7548
    },
    {
      "epoch": 1.181747025673137,
      "grad_norm": 0.8521823883056641,
      "learning_rate": 0.00012800586510263928,
      "loss": 0.2426,
      "step": 7549
    },
    {
      "epoch": 1.1819035691922355,
      "grad_norm": 0.6082648634910583,
      "learning_rate": 0.00012798142717497556,
      "loss": 0.2999,
      "step": 7550
    },
    {
      "epoch": 1.1820601127113337,
      "grad_norm": 1.2765244245529175,
      "learning_rate": 0.0001279569892473118,
      "loss": 0.5316,
      "step": 7551
    },
    {
      "epoch": 1.1822166562304322,
      "grad_norm": 1.1634889841079712,
      "learning_rate": 0.0001279325513196481,
      "loss": 0.5345,
      "step": 7552
    },
    {
      "epoch": 1.1823731997495304,
      "grad_norm": 1.5110682249069214,
      "learning_rate": 0.00012790811339198434,
      "loss": 0.3129,
      "step": 7553
    },
    {
      "epoch": 1.1825297432686286,
      "grad_norm": 0.8883544206619263,
      "learning_rate": 0.0001278836754643206,
      "loss": 0.2753,
      "step": 7554
    },
    {
      "epoch": 1.182686286787727,
      "grad_norm": 1.2819592952728271,
      "learning_rate": 0.00012785923753665687,
      "loss": 0.312,
      "step": 7555
    },
    {
      "epoch": 1.1828428303068252,
      "grad_norm": 1.9334875345230103,
      "learning_rate": 0.00012783479960899315,
      "loss": 0.5772,
      "step": 7556
    },
    {
      "epoch": 1.1829993738259237,
      "grad_norm": 1.7190932035446167,
      "learning_rate": 0.0001278103616813294,
      "loss": 0.4345,
      "step": 7557
    },
    {
      "epoch": 1.183155917345022,
      "grad_norm": 1.4696842432022095,
      "learning_rate": 0.00012778592375366568,
      "loss": 0.4088,
      "step": 7558
    },
    {
      "epoch": 1.1833124608641201,
      "grad_norm": 0.6554337739944458,
      "learning_rate": 0.00012776148582600193,
      "loss": 0.3475,
      "step": 7559
    },
    {
      "epoch": 1.1834690043832186,
      "grad_norm": 0.9007388353347778,
      "learning_rate": 0.0001277370478983382,
      "loss": 0.3618,
      "step": 7560
    },
    {
      "epoch": 1.1836255479023168,
      "grad_norm": 2.4028732776641846,
      "learning_rate": 0.0001277126099706745,
      "loss": 0.511,
      "step": 7561
    },
    {
      "epoch": 1.1837820914214152,
      "grad_norm": 1.1445221900939941,
      "learning_rate": 0.00012768817204301074,
      "loss": 0.4563,
      "step": 7562
    },
    {
      "epoch": 1.1839386349405134,
      "grad_norm": 1.3807297945022583,
      "learning_rate": 0.000127663734115347,
      "loss": 0.3692,
      "step": 7563
    },
    {
      "epoch": 1.1840951784596117,
      "grad_norm": 1.8694292306900024,
      "learning_rate": 0.00012763929618768327,
      "loss": 0.7251,
      "step": 7564
    },
    {
      "epoch": 1.18425172197871,
      "grad_norm": 2.2322914600372314,
      "learning_rate": 0.00012761485826001955,
      "loss": 0.6297,
      "step": 7565
    },
    {
      "epoch": 1.1844082654978083,
      "grad_norm": 1.5895439386367798,
      "learning_rate": 0.0001275904203323558,
      "loss": 0.4646,
      "step": 7566
    },
    {
      "epoch": 1.1845648090169068,
      "grad_norm": 1.2254077196121216,
      "learning_rate": 0.00012756598240469208,
      "loss": 0.4732,
      "step": 7567
    },
    {
      "epoch": 1.184721352536005,
      "grad_norm": 2.8192131519317627,
      "learning_rate": 0.00012754154447702833,
      "loss": 0.4586,
      "step": 7568
    },
    {
      "epoch": 1.1848778960551034,
      "grad_norm": 1.3876596689224243,
      "learning_rate": 0.0001275171065493646,
      "loss": 0.5003,
      "step": 7569
    },
    {
      "epoch": 1.1850344395742016,
      "grad_norm": 3.00423264503479,
      "learning_rate": 0.00012749266862170088,
      "loss": 1.1688,
      "step": 7570
    },
    {
      "epoch": 1.1851909830932998,
      "grad_norm": 2.323437213897705,
      "learning_rate": 0.00012746823069403714,
      "loss": 0.6379,
      "step": 7571
    },
    {
      "epoch": 1.1853475266123983,
      "grad_norm": 1.7432076930999756,
      "learning_rate": 0.0001274437927663734,
      "loss": 0.8426,
      "step": 7572
    },
    {
      "epoch": 1.1855040701314965,
      "grad_norm": 3.104215621948242,
      "learning_rate": 0.00012741935483870967,
      "loss": 0.9046,
      "step": 7573
    },
    {
      "epoch": 1.185660613650595,
      "grad_norm": 1.629520058631897,
      "learning_rate": 0.00012739491691104592,
      "loss": 0.7857,
      "step": 7574
    },
    {
      "epoch": 1.1858171571696932,
      "grad_norm": 1.4630589485168457,
      "learning_rate": 0.0001273704789833822,
      "loss": 0.7239,
      "step": 7575
    },
    {
      "epoch": 1.1859737006887916,
      "grad_norm": 1.9049055576324463,
      "learning_rate": 0.00012734604105571847,
      "loss": 0.8086,
      "step": 7576
    },
    {
      "epoch": 1.1861302442078898,
      "grad_norm": 2.1848504543304443,
      "learning_rate": 0.00012732160312805472,
      "loss": 1.3258,
      "step": 7577
    },
    {
      "epoch": 1.186286787726988,
      "grad_norm": 3.515286922454834,
      "learning_rate": 0.00012729716520039098,
      "loss": 0.7483,
      "step": 7578
    },
    {
      "epoch": 1.1864433312460865,
      "grad_norm": 2.0993247032165527,
      "learning_rate": 0.00012727272727272725,
      "loss": 0.9094,
      "step": 7579
    },
    {
      "epoch": 1.1865998747651847,
      "grad_norm": 2.1793994903564453,
      "learning_rate": 0.00012724828934506353,
      "loss": 0.8514,
      "step": 7580
    },
    {
      "epoch": 1.1867564182842831,
      "grad_norm": 2.337986469268799,
      "learning_rate": 0.00012722385141739978,
      "loss": 0.89,
      "step": 7581
    },
    {
      "epoch": 1.1869129618033814,
      "grad_norm": 1.6112979650497437,
      "learning_rate": 0.00012719941348973606,
      "loss": 0.8589,
      "step": 7582
    },
    {
      "epoch": 1.1870695053224796,
      "grad_norm": 3.4834136962890625,
      "learning_rate": 0.0001271749755620723,
      "loss": 1.0901,
      "step": 7583
    },
    {
      "epoch": 1.187226048841578,
      "grad_norm": 2.0947000980377197,
      "learning_rate": 0.0001271505376344086,
      "loss": 1.0381,
      "step": 7584
    },
    {
      "epoch": 1.1873825923606762,
      "grad_norm": 3.770942211151123,
      "learning_rate": 0.00012712609970674487,
      "loss": 0.812,
      "step": 7585
    },
    {
      "epoch": 1.1875391358797747,
      "grad_norm": 3.0660061836242676,
      "learning_rate": 0.00012710166177908112,
      "loss": 0.4694,
      "step": 7586
    },
    {
      "epoch": 1.1876956793988729,
      "grad_norm": 2.4887964725494385,
      "learning_rate": 0.00012707722385141737,
      "loss": 0.5581,
      "step": 7587
    },
    {
      "epoch": 1.187852222917971,
      "grad_norm": 1.3472387790679932,
      "learning_rate": 0.00012705278592375365,
      "loss": 0.3082,
      "step": 7588
    },
    {
      "epoch": 1.1880087664370695,
      "grad_norm": 0.712585985660553,
      "learning_rate": 0.00012702834799608993,
      "loss": 0.2746,
      "step": 7589
    },
    {
      "epoch": 1.1881653099561678,
      "grad_norm": 0.5410864353179932,
      "learning_rate": 0.00012700391006842618,
      "loss": 0.2139,
      "step": 7590
    },
    {
      "epoch": 1.1883218534752662,
      "grad_norm": 0.34346261620521545,
      "learning_rate": 0.00012697947214076246,
      "loss": 0.1462,
      "step": 7591
    },
    {
      "epoch": 1.1884783969943644,
      "grad_norm": 0.6002309918403625,
      "learning_rate": 0.0001269550342130987,
      "loss": 0.2067,
      "step": 7592
    },
    {
      "epoch": 1.1886349405134626,
      "grad_norm": 0.6561062335968018,
      "learning_rate": 0.000126930596285435,
      "loss": 0.186,
      "step": 7593
    },
    {
      "epoch": 1.188791484032561,
      "grad_norm": 0.6932880878448486,
      "learning_rate": 0.00012690615835777127,
      "loss": 0.233,
      "step": 7594
    },
    {
      "epoch": 1.1889480275516593,
      "grad_norm": 1.0030970573425293,
      "learning_rate": 0.00012688172043010752,
      "loss": 0.409,
      "step": 7595
    },
    {
      "epoch": 1.1891045710707577,
      "grad_norm": 0.5043971538543701,
      "learning_rate": 0.00012685728250244377,
      "loss": 0.1951,
      "step": 7596
    },
    {
      "epoch": 1.189261114589856,
      "grad_norm": 0.5543043613433838,
      "learning_rate": 0.00012683284457478005,
      "loss": 0.2255,
      "step": 7597
    },
    {
      "epoch": 1.1894176581089542,
      "grad_norm": 0.7690470814704895,
      "learning_rate": 0.0001268084066471163,
      "loss": 0.324,
      "step": 7598
    },
    {
      "epoch": 1.1895742016280526,
      "grad_norm": 0.9051692485809326,
      "learning_rate": 0.00012678396871945258,
      "loss": 0.3889,
      "step": 7599
    },
    {
      "epoch": 1.1897307451471508,
      "grad_norm": 1.1886712312698364,
      "learning_rate": 0.00012675953079178886,
      "loss": 0.279,
      "step": 7600
    },
    {
      "epoch": 1.1898872886662493,
      "grad_norm": 0.8935725688934326,
      "learning_rate": 0.0001267350928641251,
      "loss": 0.2286,
      "step": 7601
    },
    {
      "epoch": 1.1900438321853475,
      "grad_norm": 0.8272036910057068,
      "learning_rate": 0.00012671065493646136,
      "loss": 0.2867,
      "step": 7602
    },
    {
      "epoch": 1.190200375704446,
      "grad_norm": 0.7367708086967468,
      "learning_rate": 0.00012668621700879764,
      "loss": 0.1458,
      "step": 7603
    },
    {
      "epoch": 1.1903569192235441,
      "grad_norm": 0.7323951721191406,
      "learning_rate": 0.00012666177908113391,
      "loss": 0.2471,
      "step": 7604
    },
    {
      "epoch": 1.1905134627426424,
      "grad_norm": 2.0071024894714355,
      "learning_rate": 0.00012663734115347017,
      "loss": 0.8892,
      "step": 7605
    },
    {
      "epoch": 1.1906700062617408,
      "grad_norm": 2.2433035373687744,
      "learning_rate": 0.00012661290322580644,
      "loss": 0.5891,
      "step": 7606
    },
    {
      "epoch": 1.190826549780839,
      "grad_norm": 2.180143356323242,
      "learning_rate": 0.0001265884652981427,
      "loss": 0.2786,
      "step": 7607
    },
    {
      "epoch": 1.1909830932999375,
      "grad_norm": 1.3844753503799438,
      "learning_rate": 0.00012656402737047897,
      "loss": 0.4072,
      "step": 7608
    },
    {
      "epoch": 1.1911396368190357,
      "grad_norm": 1.1671643257141113,
      "learning_rate": 0.00012653958944281525,
      "loss": 0.3149,
      "step": 7609
    },
    {
      "epoch": 1.1912961803381341,
      "grad_norm": 1.9654929637908936,
      "learning_rate": 0.0001265151515151515,
      "loss": 0.4545,
      "step": 7610
    },
    {
      "epoch": 1.1914527238572323,
      "grad_norm": 1.6887761354446411,
      "learning_rate": 0.00012649071358748775,
      "loss": 0.6904,
      "step": 7611
    },
    {
      "epoch": 1.1916092673763305,
      "grad_norm": 1.5617834329605103,
      "learning_rate": 0.00012646627565982403,
      "loss": 0.4628,
      "step": 7612
    },
    {
      "epoch": 1.191765810895429,
      "grad_norm": 4.053790092468262,
      "learning_rate": 0.0001264418377321603,
      "loss": 0.9632,
      "step": 7613
    },
    {
      "epoch": 1.1919223544145272,
      "grad_norm": 1.2538058757781982,
      "learning_rate": 0.00012641739980449656,
      "loss": 0.3495,
      "step": 7614
    },
    {
      "epoch": 1.1920788979336256,
      "grad_norm": 2.1369194984436035,
      "learning_rate": 0.00012639296187683284,
      "loss": 0.8079,
      "step": 7615
    },
    {
      "epoch": 1.1922354414527239,
      "grad_norm": 5.438715934753418,
      "learning_rate": 0.0001263685239491691,
      "loss": 0.7607,
      "step": 7616
    },
    {
      "epoch": 1.192391984971822,
      "grad_norm": 4.414690971374512,
      "learning_rate": 0.00012634408602150537,
      "loss": 0.5004,
      "step": 7617
    },
    {
      "epoch": 1.1925485284909205,
      "grad_norm": 3.980987310409546,
      "learning_rate": 0.00012631964809384162,
      "loss": 0.7119,
      "step": 7618
    },
    {
      "epoch": 1.1927050720100187,
      "grad_norm": 2.606283664703369,
      "learning_rate": 0.0001262952101661779,
      "loss": 0.5477,
      "step": 7619
    },
    {
      "epoch": 1.1928616155291172,
      "grad_norm": 1.2772033214569092,
      "learning_rate": 0.00012627077223851415,
      "loss": 0.4294,
      "step": 7620
    },
    {
      "epoch": 1.1930181590482154,
      "grad_norm": 2.65388822555542,
      "learning_rate": 0.00012624633431085043,
      "loss": 0.9099,
      "step": 7621
    },
    {
      "epoch": 1.1931747025673136,
      "grad_norm": 2.885216236114502,
      "learning_rate": 0.00012622189638318668,
      "loss": 1.0008,
      "step": 7622
    },
    {
      "epoch": 1.193331246086412,
      "grad_norm": 1.257573127746582,
      "learning_rate": 0.00012619745845552296,
      "loss": 0.4982,
      "step": 7623
    },
    {
      "epoch": 1.1934877896055103,
      "grad_norm": 6.712513446807861,
      "learning_rate": 0.00012617302052785924,
      "loss": 1.2771,
      "step": 7624
    },
    {
      "epoch": 1.1936443331246087,
      "grad_norm": 2.6554524898529053,
      "learning_rate": 0.0001261485826001955,
      "loss": 0.7371,
      "step": 7625
    },
    {
      "epoch": 1.193800876643707,
      "grad_norm": 2.5295774936676025,
      "learning_rate": 0.00012612414467253174,
      "loss": 1.43,
      "step": 7626
    },
    {
      "epoch": 1.1939574201628051,
      "grad_norm": 3.8119919300079346,
      "learning_rate": 0.00012609970674486802,
      "loss": 1.3552,
      "step": 7627
    },
    {
      "epoch": 1.1941139636819036,
      "grad_norm": 3.1920182704925537,
      "learning_rate": 0.0001260752688172043,
      "loss": 1.7003,
      "step": 7628
    },
    {
      "epoch": 1.1942705072010018,
      "grad_norm": 3.4198391437530518,
      "learning_rate": 0.00012605083088954055,
      "loss": 0.8041,
      "step": 7629
    },
    {
      "epoch": 1.1944270507201002,
      "grad_norm": 2.1979782581329346,
      "learning_rate": 0.00012602639296187683,
      "loss": 0.9902,
      "step": 7630
    },
    {
      "epoch": 1.1945835942391985,
      "grad_norm": 2.4883174896240234,
      "learning_rate": 0.00012600195503421308,
      "loss": 1.122,
      "step": 7631
    },
    {
      "epoch": 1.1947401377582967,
      "grad_norm": 2.1847050189971924,
      "learning_rate": 0.00012597751710654936,
      "loss": 0.9887,
      "step": 7632
    },
    {
      "epoch": 1.1948966812773951,
      "grad_norm": 2.9857428073883057,
      "learning_rate": 0.00012595307917888563,
      "loss": 1.2614,
      "step": 7633
    },
    {
      "epoch": 1.1950532247964933,
      "grad_norm": 3.8710601329803467,
      "learning_rate": 0.00012592864125122189,
      "loss": 1.1563,
      "step": 7634
    },
    {
      "epoch": 1.1952097683155918,
      "grad_norm": 1.628251314163208,
      "learning_rate": 0.00012590420332355814,
      "loss": 0.281,
      "step": 7635
    },
    {
      "epoch": 1.19536631183469,
      "grad_norm": 2.4681708812713623,
      "learning_rate": 0.00012587976539589442,
      "loss": 0.8694,
      "step": 7636
    },
    {
      "epoch": 1.1955228553537884,
      "grad_norm": 1.361156940460205,
      "learning_rate": 0.0001258553274682307,
      "loss": 0.6151,
      "step": 7637
    },
    {
      "epoch": 1.1956793988728867,
      "grad_norm": 2.5139424800872803,
      "learning_rate": 0.00012583088954056695,
      "loss": 0.4191,
      "step": 7638
    },
    {
      "epoch": 1.195835942391985,
      "grad_norm": 0.4481685161590576,
      "learning_rate": 0.00012580645161290322,
      "loss": 0.266,
      "step": 7639
    },
    {
      "epoch": 1.1959924859110833,
      "grad_norm": 0.5807831287384033,
      "learning_rate": 0.00012578201368523947,
      "loss": 0.3183,
      "step": 7640
    },
    {
      "epoch": 1.1961490294301815,
      "grad_norm": 0.433891236782074,
      "learning_rate": 0.00012575757575757575,
      "loss": 0.203,
      "step": 7641
    },
    {
      "epoch": 1.19630557294928,
      "grad_norm": 0.3798646628856659,
      "learning_rate": 0.000125733137829912,
      "loss": 0.2185,
      "step": 7642
    },
    {
      "epoch": 1.1964621164683782,
      "grad_norm": 0.5597085952758789,
      "learning_rate": 0.00012570869990224828,
      "loss": 0.2084,
      "step": 7643
    },
    {
      "epoch": 1.1966186599874766,
      "grad_norm": 0.8188731670379639,
      "learning_rate": 0.00012568426197458453,
      "loss": 0.3176,
      "step": 7644
    },
    {
      "epoch": 1.1967752035065748,
      "grad_norm": 0.44461318850517273,
      "learning_rate": 0.0001256598240469208,
      "loss": 0.2583,
      "step": 7645
    },
    {
      "epoch": 1.196931747025673,
      "grad_norm": 0.6256333589553833,
      "learning_rate": 0.00012563538611925706,
      "loss": 0.2882,
      "step": 7646
    },
    {
      "epoch": 1.1970882905447715,
      "grad_norm": 0.5847500562667847,
      "learning_rate": 0.00012561094819159334,
      "loss": 0.2987,
      "step": 7647
    },
    {
      "epoch": 1.1972448340638697,
      "grad_norm": 0.4308290183544159,
      "learning_rate": 0.00012558651026392962,
      "loss": 0.2004,
      "step": 7648
    },
    {
      "epoch": 1.1974013775829682,
      "grad_norm": 0.6642153859138489,
      "learning_rate": 0.00012556207233626587,
      "loss": 0.2055,
      "step": 7649
    },
    {
      "epoch": 1.1975579211020664,
      "grad_norm": 1.537541151046753,
      "learning_rate": 0.00012553763440860212,
      "loss": 0.5965,
      "step": 7650
    },
    {
      "epoch": 1.1977144646211646,
      "grad_norm": 0.6113720536231995,
      "learning_rate": 0.0001255131964809384,
      "loss": 0.2567,
      "step": 7651
    },
    {
      "epoch": 1.197871008140263,
      "grad_norm": 1.1813939809799194,
      "learning_rate": 0.00012548875855327468,
      "loss": 0.3958,
      "step": 7652
    },
    {
      "epoch": 1.1980275516593613,
      "grad_norm": 1.415676474571228,
      "learning_rate": 0.00012546432062561093,
      "loss": 0.4031,
      "step": 7653
    },
    {
      "epoch": 1.1981840951784597,
      "grad_norm": 2.827056884765625,
      "learning_rate": 0.0001254398826979472,
      "loss": 0.39,
      "step": 7654
    },
    {
      "epoch": 1.198340638697558,
      "grad_norm": 1.0784145593643188,
      "learning_rate": 0.00012541544477028346,
      "loss": 0.4073,
      "step": 7655
    },
    {
      "epoch": 1.1984971822166561,
      "grad_norm": 1.3074734210968018,
      "learning_rate": 0.00012539100684261974,
      "loss": 0.3409,
      "step": 7656
    },
    {
      "epoch": 1.1986537257357546,
      "grad_norm": 5.250173091888428,
      "learning_rate": 0.00012536656891495602,
      "loss": 0.7839,
      "step": 7657
    },
    {
      "epoch": 1.1988102692548528,
      "grad_norm": 0.9771109223365784,
      "learning_rate": 0.00012534213098729227,
      "loss": 0.4488,
      "step": 7658
    },
    {
      "epoch": 1.1989668127739512,
      "grad_norm": 2.0896804332733154,
      "learning_rate": 0.00012531769305962852,
      "loss": 0.5873,
      "step": 7659
    },
    {
      "epoch": 1.1991233562930494,
      "grad_norm": 1.340659260749817,
      "learning_rate": 0.0001252932551319648,
      "loss": 0.3258,
      "step": 7660
    },
    {
      "epoch": 1.1992798998121477,
      "grad_norm": 2.5258636474609375,
      "learning_rate": 0.00012526881720430108,
      "loss": 0.434,
      "step": 7661
    },
    {
      "epoch": 1.199436443331246,
      "grad_norm": 0.8065431714057922,
      "learning_rate": 0.00012524437927663733,
      "loss": 0.4309,
      "step": 7662
    },
    {
      "epoch": 1.1995929868503443,
      "grad_norm": 2.020216464996338,
      "learning_rate": 0.0001252199413489736,
      "loss": 0.8626,
      "step": 7663
    },
    {
      "epoch": 1.1997495303694428,
      "grad_norm": 1.528834581375122,
      "learning_rate": 0.00012519550342130986,
      "loss": 0.4851,
      "step": 7664
    },
    {
      "epoch": 1.199906073888541,
      "grad_norm": 1.1906168460845947,
      "learning_rate": 0.00012517106549364614,
      "loss": 0.4344,
      "step": 7665
    },
    {
      "epoch": 1.2000626174076394,
      "grad_norm": 1.7082940340042114,
      "learning_rate": 0.0001251466275659824,
      "loss": 0.9714,
      "step": 7666
    },
    {
      "epoch": 1.2002191609267376,
      "grad_norm": 1.5816830396652222,
      "learning_rate": 0.00012512218963831867,
      "loss": 0.4113,
      "step": 7667
    },
    {
      "epoch": 1.2003757044458359,
      "grad_norm": 1.9569361209869385,
      "learning_rate": 0.00012509775171065492,
      "loss": 0.9074,
      "step": 7668
    },
    {
      "epoch": 1.2005322479649343,
      "grad_norm": 2.3706414699554443,
      "learning_rate": 0.0001250733137829912,
      "loss": 0.6514,
      "step": 7669
    },
    {
      "epoch": 1.2006887914840325,
      "grad_norm": 2.0549864768981934,
      "learning_rate": 0.00012504887585532745,
      "loss": 0.6741,
      "step": 7670
    },
    {
      "epoch": 1.200845335003131,
      "grad_norm": 2.067312002182007,
      "learning_rate": 0.00012502443792766372,
      "loss": 0.7198,
      "step": 7671
    },
    {
      "epoch": 1.2010018785222292,
      "grad_norm": 2.430595874786377,
      "learning_rate": 0.000125,
      "loss": 0.9955,
      "step": 7672
    },
    {
      "epoch": 1.2011584220413276,
      "grad_norm": 1.7246347665786743,
      "learning_rate": 0.00012497556207233625,
      "loss": 0.8613,
      "step": 7673
    },
    {
      "epoch": 1.2013149655604258,
      "grad_norm": 3.665044069290161,
      "learning_rate": 0.0001249511241446725,
      "loss": 0.8444,
      "step": 7674
    },
    {
      "epoch": 1.201471509079524,
      "grad_norm": 3.734281539916992,
      "learning_rate": 0.00012492668621700878,
      "loss": 0.8613,
      "step": 7675
    },
    {
      "epoch": 1.2016280525986225,
      "grad_norm": 2.5038020610809326,
      "learning_rate": 0.00012490224828934506,
      "loss": 1.4506,
      "step": 7676
    },
    {
      "epoch": 1.2017845961177207,
      "grad_norm": 2.5142405033111572,
      "learning_rate": 0.0001248778103616813,
      "loss": 0.982,
      "step": 7677
    },
    {
      "epoch": 1.2019411396368191,
      "grad_norm": 4.540086269378662,
      "learning_rate": 0.0001248533724340176,
      "loss": 0.9359,
      "step": 7678
    },
    {
      "epoch": 1.2020976831559174,
      "grad_norm": 1.838870644569397,
      "learning_rate": 0.00012482893450635384,
      "loss": 0.6585,
      "step": 7679
    },
    {
      "epoch": 1.2022542266750156,
      "grad_norm": 3.0553319454193115,
      "learning_rate": 0.00012480449657869012,
      "loss": 1.318,
      "step": 7680
    },
    {
      "epoch": 1.202410770194114,
      "grad_norm": 3.3202075958251953,
      "learning_rate": 0.0001247800586510264,
      "loss": 1.5615,
      "step": 7681
    },
    {
      "epoch": 1.2025673137132122,
      "grad_norm": 5.532286643981934,
      "learning_rate": 0.00012475562072336265,
      "loss": 1.1045,
      "step": 7682
    },
    {
      "epoch": 1.2027238572323107,
      "grad_norm": 2.526785135269165,
      "learning_rate": 0.0001247311827956989,
      "loss": 0.7781,
      "step": 7683
    },
    {
      "epoch": 1.202880400751409,
      "grad_norm": 5.095958709716797,
      "learning_rate": 0.00012470674486803518,
      "loss": 0.9252,
      "step": 7684
    },
    {
      "epoch": 1.2030369442705071,
      "grad_norm": 2.2262609004974365,
      "learning_rate": 0.00012468230694037146,
      "loss": 0.793,
      "step": 7685
    },
    {
      "epoch": 1.2031934877896056,
      "grad_norm": 4.54476261138916,
      "learning_rate": 0.0001246578690127077,
      "loss": 1.4444,
      "step": 7686
    },
    {
      "epoch": 1.2033500313087038,
      "grad_norm": 3.232273578643799,
      "learning_rate": 0.000124633431085044,
      "loss": 1.1172,
      "step": 7687
    },
    {
      "epoch": 1.2035065748278022,
      "grad_norm": 2.0127787590026855,
      "learning_rate": 0.00012460899315738024,
      "loss": 0.6069,
      "step": 7688
    },
    {
      "epoch": 1.2036631183469004,
      "grad_norm": 0.4151703417301178,
      "learning_rate": 0.0001245845552297165,
      "loss": 0.2224,
      "step": 7689
    },
    {
      "epoch": 1.2038196618659986,
      "grad_norm": 0.4115196168422699,
      "learning_rate": 0.00012456011730205277,
      "loss": 0.2122,
      "step": 7690
    },
    {
      "epoch": 1.203976205385097,
      "grad_norm": 0.49626508355140686,
      "learning_rate": 0.00012453567937438905,
      "loss": 0.2164,
      "step": 7691
    },
    {
      "epoch": 1.2041327489041953,
      "grad_norm": 0.6475088000297546,
      "learning_rate": 0.0001245112414467253,
      "loss": 0.1638,
      "step": 7692
    },
    {
      "epoch": 1.2042892924232937,
      "grad_norm": 0.5242400765419006,
      "learning_rate": 0.00012448680351906158,
      "loss": 0.1996,
      "step": 7693
    },
    {
      "epoch": 1.204445835942392,
      "grad_norm": 0.5632511973381042,
      "learning_rate": 0.00012446236559139783,
      "loss": 0.2789,
      "step": 7694
    },
    {
      "epoch": 1.2046023794614902,
      "grad_norm": 0.5035088658332825,
      "learning_rate": 0.0001244379276637341,
      "loss": 0.251,
      "step": 7695
    },
    {
      "epoch": 1.2047589229805886,
      "grad_norm": 0.42443281412124634,
      "learning_rate": 0.00012441348973607038,
      "loss": 0.2104,
      "step": 7696
    },
    {
      "epoch": 1.2049154664996868,
      "grad_norm": 4.329174518585205,
      "learning_rate": 0.00012438905180840664,
      "loss": 0.4027,
      "step": 7697
    },
    {
      "epoch": 1.2050720100187853,
      "grad_norm": 0.7891638875007629,
      "learning_rate": 0.0001243646138807429,
      "loss": 0.3523,
      "step": 7698
    },
    {
      "epoch": 1.2052285535378835,
      "grad_norm": 0.9422610998153687,
      "learning_rate": 0.00012434017595307917,
      "loss": 0.3149,
      "step": 7699
    },
    {
      "epoch": 1.205385097056982,
      "grad_norm": 1.680687427520752,
      "learning_rate": 0.00012431573802541544,
      "loss": 0.3606,
      "step": 7700
    },
    {
      "epoch": 1.2055416405760802,
      "grad_norm": 1.3018105030059814,
      "learning_rate": 0.0001242913000977517,
      "loss": 0.2848,
      "step": 7701
    },
    {
      "epoch": 1.2056981840951784,
      "grad_norm": 2.048955202102661,
      "learning_rate": 0.00012426686217008797,
      "loss": 0.2945,
      "step": 7702
    },
    {
      "epoch": 1.2058547276142768,
      "grad_norm": 1.392334222793579,
      "learning_rate": 0.00012424242424242422,
      "loss": 0.4641,
      "step": 7703
    },
    {
      "epoch": 1.206011271133375,
      "grad_norm": 1.0475993156433105,
      "learning_rate": 0.0001242179863147605,
      "loss": 0.4634,
      "step": 7704
    },
    {
      "epoch": 1.2061678146524735,
      "grad_norm": 1.2346774339675903,
      "learning_rate": 0.00012419354838709678,
      "loss": 0.5416,
      "step": 7705
    },
    {
      "epoch": 1.2063243581715717,
      "grad_norm": 1.159487009048462,
      "learning_rate": 0.00012416911045943303,
      "loss": 0.2803,
      "step": 7706
    },
    {
      "epoch": 1.2064809016906701,
      "grad_norm": 1.400073528289795,
      "learning_rate": 0.00012414467253176928,
      "loss": 0.2959,
      "step": 7707
    },
    {
      "epoch": 1.2066374452097683,
      "grad_norm": 0.6044692397117615,
      "learning_rate": 0.00012412023460410556,
      "loss": 0.2158,
      "step": 7708
    },
    {
      "epoch": 1.2067939887288666,
      "grad_norm": 1.8610830307006836,
      "learning_rate": 0.00012409579667644184,
      "loss": 0.6542,
      "step": 7709
    },
    {
      "epoch": 1.206950532247965,
      "grad_norm": 1.094446063041687,
      "learning_rate": 0.0001240713587487781,
      "loss": 0.4033,
      "step": 7710
    },
    {
      "epoch": 1.2071070757670632,
      "grad_norm": 4.96532678604126,
      "learning_rate": 0.00012404692082111437,
      "loss": 0.4836,
      "step": 7711
    },
    {
      "epoch": 1.2072636192861617,
      "grad_norm": 1.7114042043685913,
      "learning_rate": 0.00012402248289345062,
      "loss": 0.4183,
      "step": 7712
    },
    {
      "epoch": 1.2074201628052599,
      "grad_norm": 1.576427936553955,
      "learning_rate": 0.00012399804496578687,
      "loss": 0.4931,
      "step": 7713
    },
    {
      "epoch": 1.207576706324358,
      "grad_norm": 1.6242084503173828,
      "learning_rate": 0.00012397360703812315,
      "loss": 0.3988,
      "step": 7714
    },
    {
      "epoch": 1.2077332498434565,
      "grad_norm": 6.9016499519348145,
      "learning_rate": 0.00012394916911045943,
      "loss": 1.0594,
      "step": 7715
    },
    {
      "epoch": 1.2078897933625548,
      "grad_norm": 1.4505754709243774,
      "learning_rate": 0.00012392473118279568,
      "loss": 0.5999,
      "step": 7716
    },
    {
      "epoch": 1.2080463368816532,
      "grad_norm": 1.9598108530044556,
      "learning_rate": 0.00012390029325513196,
      "loss": 0.4765,
      "step": 7717
    },
    {
      "epoch": 1.2082028804007514,
      "grad_norm": 1.3153584003448486,
      "learning_rate": 0.0001238758553274682,
      "loss": 0.5586,
      "step": 7718
    },
    {
      "epoch": 1.2083594239198496,
      "grad_norm": 1.5149145126342773,
      "learning_rate": 0.0001238514173998045,
      "loss": 0.4036,
      "step": 7719
    },
    {
      "epoch": 1.208515967438948,
      "grad_norm": 2.801050901412964,
      "learning_rate": 0.00012382697947214077,
      "loss": 0.8044,
      "step": 7720
    },
    {
      "epoch": 1.2086725109580463,
      "grad_norm": 2.656320095062256,
      "learning_rate": 0.00012380254154447702,
      "loss": 0.7702,
      "step": 7721
    },
    {
      "epoch": 1.2088290544771447,
      "grad_norm": 3.1893441677093506,
      "learning_rate": 0.00012377810361681327,
      "loss": 0.8827,
      "step": 7722
    },
    {
      "epoch": 1.208985597996243,
      "grad_norm": 4.1985063552856445,
      "learning_rate": 0.00012375366568914955,
      "loss": 1.0204,
      "step": 7723
    },
    {
      "epoch": 1.2091421415153412,
      "grad_norm": 2.559026002883911,
      "learning_rate": 0.00012372922776148583,
      "loss": 0.9891,
      "step": 7724
    },
    {
      "epoch": 1.2092986850344396,
      "grad_norm": 1.904913067817688,
      "learning_rate": 0.00012370478983382208,
      "loss": 1.0455,
      "step": 7725
    },
    {
      "epoch": 1.2094552285535378,
      "grad_norm": 3.1764469146728516,
      "learning_rate": 0.00012368035190615836,
      "loss": 1.2292,
      "step": 7726
    },
    {
      "epoch": 1.2096117720726363,
      "grad_norm": 1.6916859149932861,
      "learning_rate": 0.0001236559139784946,
      "loss": 0.7425,
      "step": 7727
    },
    {
      "epoch": 1.2097683155917345,
      "grad_norm": 1.8501296043395996,
      "learning_rate": 0.00012363147605083089,
      "loss": 0.9403,
      "step": 7728
    },
    {
      "epoch": 1.2099248591108327,
      "grad_norm": 2.4505608081817627,
      "learning_rate": 0.00012360703812316716,
      "loss": 0.6932,
      "step": 7729
    },
    {
      "epoch": 1.2100814026299311,
      "grad_norm": 2.6362080574035645,
      "learning_rate": 0.00012358260019550342,
      "loss": 0.9876,
      "step": 7730
    },
    {
      "epoch": 1.2102379461490294,
      "grad_norm": 3.168712854385376,
      "learning_rate": 0.00012355816226783967,
      "loss": 1.1326,
      "step": 7731
    },
    {
      "epoch": 1.2103944896681278,
      "grad_norm": 1.8203378915786743,
      "learning_rate": 0.00012353372434017594,
      "loss": 0.9867,
      "step": 7732
    },
    {
      "epoch": 1.210551033187226,
      "grad_norm": 1.6555746793746948,
      "learning_rate": 0.0001235092864125122,
      "loss": 1.2667,
      "step": 7733
    },
    {
      "epoch": 1.2107075767063245,
      "grad_norm": 1.094726324081421,
      "learning_rate": 0.00012348484848484847,
      "loss": 0.3442,
      "step": 7734
    },
    {
      "epoch": 1.2108641202254227,
      "grad_norm": 2.5407097339630127,
      "learning_rate": 0.00012346041055718475,
      "loss": 0.5379,
      "step": 7735
    },
    {
      "epoch": 1.2110206637445209,
      "grad_norm": 1.3805251121520996,
      "learning_rate": 0.000123435972629521,
      "loss": 0.1715,
      "step": 7736
    },
    {
      "epoch": 1.2111772072636193,
      "grad_norm": 3.8366167545318604,
      "learning_rate": 0.00012341153470185726,
      "loss": 0.7459,
      "step": 7737
    },
    {
      "epoch": 1.2113337507827175,
      "grad_norm": 2.2765462398529053,
      "learning_rate": 0.00012338709677419353,
      "loss": 0.9848,
      "step": 7738
    },
    {
      "epoch": 1.211490294301816,
      "grad_norm": 0.41554513573646545,
      "learning_rate": 0.0001233626588465298,
      "loss": 0.2307,
      "step": 7739
    },
    {
      "epoch": 1.2116468378209142,
      "grad_norm": 0.5446311235427856,
      "learning_rate": 0.00012333822091886606,
      "loss": 0.1887,
      "step": 7740
    },
    {
      "epoch": 1.2118033813400126,
      "grad_norm": 0.42583855986595154,
      "learning_rate": 0.00012331378299120234,
      "loss": 0.2161,
      "step": 7741
    },
    {
      "epoch": 1.2119599248591109,
      "grad_norm": 0.40142467617988586,
      "learning_rate": 0.0001232893450635386,
      "loss": 0.2454,
      "step": 7742
    },
    {
      "epoch": 1.212116468378209,
      "grad_norm": 0.6083351969718933,
      "learning_rate": 0.00012326490713587487,
      "loss": 0.2714,
      "step": 7743
    },
    {
      "epoch": 1.2122730118973075,
      "grad_norm": 0.5035018920898438,
      "learning_rate": 0.00012324046920821115,
      "loss": 0.3219,
      "step": 7744
    },
    {
      "epoch": 1.2124295554164057,
      "grad_norm": 0.8749467730522156,
      "learning_rate": 0.0001232160312805474,
      "loss": 0.2845,
      "step": 7745
    },
    {
      "epoch": 1.2125860989355042,
      "grad_norm": 0.8860273957252502,
      "learning_rate": 0.00012319159335288365,
      "loss": 0.2443,
      "step": 7746
    },
    {
      "epoch": 1.2127426424546024,
      "grad_norm": 0.9518566727638245,
      "learning_rate": 0.00012316715542521993,
      "loss": 0.3225,
      "step": 7747
    },
    {
      "epoch": 1.2128991859737006,
      "grad_norm": 0.7895660996437073,
      "learning_rate": 0.0001231427174975562,
      "loss": 0.2441,
      "step": 7748
    },
    {
      "epoch": 1.213055729492799,
      "grad_norm": 0.9704890847206116,
      "learning_rate": 0.00012311827956989246,
      "loss": 0.3121,
      "step": 7749
    },
    {
      "epoch": 1.2132122730118973,
      "grad_norm": 0.8872887492179871,
      "learning_rate": 0.00012309384164222874,
      "loss": 0.302,
      "step": 7750
    },
    {
      "epoch": 1.2133688165309957,
      "grad_norm": 0.5776874423027039,
      "learning_rate": 0.000123069403714565,
      "loss": 0.2676,
      "step": 7751
    },
    {
      "epoch": 1.213525360050094,
      "grad_norm": 0.9891446232795715,
      "learning_rate": 0.00012304496578690127,
      "loss": 0.3352,
      "step": 7752
    },
    {
      "epoch": 1.2136819035691921,
      "grad_norm": 1.1194006204605103,
      "learning_rate": 0.00012302052785923755,
      "loss": 0.3109,
      "step": 7753
    },
    {
      "epoch": 1.2138384470882906,
      "grad_norm": 1.1497498750686646,
      "learning_rate": 0.0001229960899315738,
      "loss": 0.3955,
      "step": 7754
    },
    {
      "epoch": 1.2139949906073888,
      "grad_norm": 2.879631519317627,
      "learning_rate": 0.00012297165200391005,
      "loss": 0.8381,
      "step": 7755
    },
    {
      "epoch": 1.2141515341264872,
      "grad_norm": 2.2817280292510986,
      "learning_rate": 0.00012294721407624633,
      "loss": 0.5012,
      "step": 7756
    },
    {
      "epoch": 1.2143080776455855,
      "grad_norm": 2.1993658542633057,
      "learning_rate": 0.00012292277614858258,
      "loss": 0.4131,
      "step": 7757
    },
    {
      "epoch": 1.2144646211646837,
      "grad_norm": 0.96397864818573,
      "learning_rate": 0.00012289833822091886,
      "loss": 0.3529,
      "step": 7758
    },
    {
      "epoch": 1.2146211646837821,
      "grad_norm": 1.8400605916976929,
      "learning_rate": 0.00012287390029325514,
      "loss": 0.3834,
      "step": 7759
    },
    {
      "epoch": 1.2147777082028803,
      "grad_norm": 2.0213780403137207,
      "learning_rate": 0.00012284946236559139,
      "loss": 0.6518,
      "step": 7760
    },
    {
      "epoch": 1.2149342517219788,
      "grad_norm": 1.1717514991760254,
      "learning_rate": 0.00012282502443792764,
      "loss": 0.5768,
      "step": 7761
    },
    {
      "epoch": 1.215090795241077,
      "grad_norm": 1.5490142107009888,
      "learning_rate": 0.00012280058651026392,
      "loss": 0.6203,
      "step": 7762
    },
    {
      "epoch": 1.2152473387601752,
      "grad_norm": 2.197330951690674,
      "learning_rate": 0.0001227761485826002,
      "loss": 0.8325,
      "step": 7763
    },
    {
      "epoch": 1.2154038822792737,
      "grad_norm": 1.72406804561615,
      "learning_rate": 0.00012275171065493645,
      "loss": 0.449,
      "step": 7764
    },
    {
      "epoch": 1.2155604257983719,
      "grad_norm": 2.904944658279419,
      "learning_rate": 0.00012272727272727272,
      "loss": 0.7901,
      "step": 7765
    },
    {
      "epoch": 1.2157169693174703,
      "grad_norm": 1.9773554801940918,
      "learning_rate": 0.00012270283479960898,
      "loss": 0.8409,
      "step": 7766
    },
    {
      "epoch": 1.2158735128365685,
      "grad_norm": 6.0343337059021,
      "learning_rate": 0.00012267839687194525,
      "loss": 1.4507,
      "step": 7767
    },
    {
      "epoch": 1.216030056355667,
      "grad_norm": 2.157370090484619,
      "learning_rate": 0.00012265395894428153,
      "loss": 0.9013,
      "step": 7768
    },
    {
      "epoch": 1.2161865998747652,
      "grad_norm": 2.05031418800354,
      "learning_rate": 0.00012262952101661778,
      "loss": 0.5475,
      "step": 7769
    },
    {
      "epoch": 1.2163431433938634,
      "grad_norm": 2.67240834236145,
      "learning_rate": 0.00012260508308895403,
      "loss": 1.015,
      "step": 7770
    },
    {
      "epoch": 1.2164996869129618,
      "grad_norm": 4.982744216918945,
      "learning_rate": 0.0001225806451612903,
      "loss": 1.03,
      "step": 7771
    },
    {
      "epoch": 1.21665623043206,
      "grad_norm": 2.2275023460388184,
      "learning_rate": 0.0001225562072336266,
      "loss": 0.9372,
      "step": 7772
    },
    {
      "epoch": 1.2168127739511585,
      "grad_norm": 1.2302268743515015,
      "learning_rate": 0.00012253176930596284,
      "loss": 0.4652,
      "step": 7773
    },
    {
      "epoch": 1.2169693174702567,
      "grad_norm": 2.873462200164795,
      "learning_rate": 0.00012250733137829912,
      "loss": 0.8191,
      "step": 7774
    },
    {
      "epoch": 1.2171258609893552,
      "grad_norm": 2.2558248043060303,
      "learning_rate": 0.00012248289345063537,
      "loss": 0.885,
      "step": 7775
    },
    {
      "epoch": 1.2172824045084534,
      "grad_norm": 2.1223719120025635,
      "learning_rate": 0.00012245845552297165,
      "loss": 0.7907,
      "step": 7776
    },
    {
      "epoch": 1.2174389480275516,
      "grad_norm": 2.519007682800293,
      "learning_rate": 0.0001224340175953079,
      "loss": 0.9959,
      "step": 7777
    },
    {
      "epoch": 1.21759549154665,
      "grad_norm": 2.0962483882904053,
      "learning_rate": 0.00012240957966764418,
      "loss": 0.7831,
      "step": 7778
    },
    {
      "epoch": 1.2177520350657483,
      "grad_norm": 2.0779409408569336,
      "learning_rate": 0.00012238514173998043,
      "loss": 0.5034,
      "step": 7779
    },
    {
      "epoch": 1.2179085785848467,
      "grad_norm": 1.5359145402908325,
      "learning_rate": 0.0001223607038123167,
      "loss": 0.5842,
      "step": 7780
    },
    {
      "epoch": 1.218065122103945,
      "grad_norm": 4.660392761230469,
      "learning_rate": 0.00012233626588465296,
      "loss": 1.1284,
      "step": 7781
    },
    {
      "epoch": 1.2182216656230431,
      "grad_norm": 2.8180553913116455,
      "learning_rate": 0.00012231182795698924,
      "loss": 0.5375,
      "step": 7782
    },
    {
      "epoch": 1.2183782091421416,
      "grad_norm": 2.4774529933929443,
      "learning_rate": 0.00012228739002932552,
      "loss": 1.0843,
      "step": 7783
    },
    {
      "epoch": 1.2185347526612398,
      "grad_norm": 4.466963768005371,
      "learning_rate": 0.00012226295210166177,
      "loss": 1.9087,
      "step": 7784
    },
    {
      "epoch": 1.2186912961803382,
      "grad_norm": 3.7716944217681885,
      "learning_rate": 0.00012223851417399802,
      "loss": 0.9215,
      "step": 7785
    },
    {
      "epoch": 1.2188478396994364,
      "grad_norm": 1.5880441665649414,
      "learning_rate": 0.0001222140762463343,
      "loss": 0.4391,
      "step": 7786
    },
    {
      "epoch": 1.2190043832185347,
      "grad_norm": 1.5129741430282593,
      "learning_rate": 0.00012218963831867058,
      "loss": 0.2122,
      "step": 7787
    },
    {
      "epoch": 1.219160926737633,
      "grad_norm": 2.9316365718841553,
      "learning_rate": 0.00012216520039100683,
      "loss": 0.9607,
      "step": 7788
    },
    {
      "epoch": 1.2193174702567313,
      "grad_norm": 0.47794780135154724,
      "learning_rate": 0.0001221407624633431,
      "loss": 0.2167,
      "step": 7789
    },
    {
      "epoch": 1.2194740137758298,
      "grad_norm": 0.9191654324531555,
      "learning_rate": 0.00012211632453567936,
      "loss": 0.2389,
      "step": 7790
    },
    {
      "epoch": 1.219630557294928,
      "grad_norm": 0.43953564763069153,
      "learning_rate": 0.00012209188660801564,
      "loss": 0.1648,
      "step": 7791
    },
    {
      "epoch": 1.2197871008140262,
      "grad_norm": 0.49689897894859314,
      "learning_rate": 0.0001220674486803519,
      "loss": 0.2088,
      "step": 7792
    },
    {
      "epoch": 1.2199436443331246,
      "grad_norm": 0.5962256193161011,
      "learning_rate": 0.00012204301075268817,
      "loss": 0.2505,
      "step": 7793
    },
    {
      "epoch": 1.2201001878522229,
      "grad_norm": 0.6489814519882202,
      "learning_rate": 0.00012201857282502443,
      "loss": 0.2664,
      "step": 7794
    },
    {
      "epoch": 1.2202567313713213,
      "grad_norm": 0.46907347440719604,
      "learning_rate": 0.0001219941348973607,
      "loss": 0.2782,
      "step": 7795
    },
    {
      "epoch": 1.2204132748904195,
      "grad_norm": 0.5828919410705566,
      "learning_rate": 0.00012196969696969696,
      "loss": 0.1701,
      "step": 7796
    },
    {
      "epoch": 1.2205698184095177,
      "grad_norm": 0.8440871238708496,
      "learning_rate": 0.00012194525904203322,
      "loss": 0.2443,
      "step": 7797
    },
    {
      "epoch": 1.2207263619286162,
      "grad_norm": 0.6275129914283752,
      "learning_rate": 0.0001219208211143695,
      "loss": 0.193,
      "step": 7798
    },
    {
      "epoch": 1.2208829054477144,
      "grad_norm": 0.6567816138267517,
      "learning_rate": 0.00012189638318670575,
      "loss": 0.2657,
      "step": 7799
    },
    {
      "epoch": 1.2210394489668128,
      "grad_norm": 0.7370538711547852,
      "learning_rate": 0.00012187194525904202,
      "loss": 0.3339,
      "step": 7800
    },
    {
      "epoch": 1.221195992485911,
      "grad_norm": 0.6013770699501038,
      "learning_rate": 0.00012184750733137828,
      "loss": 0.2557,
      "step": 7801
    },
    {
      "epoch": 1.2213525360050095,
      "grad_norm": 2.4509990215301514,
      "learning_rate": 0.00012182306940371456,
      "loss": 0.7065,
      "step": 7802
    },
    {
      "epoch": 1.2215090795241077,
      "grad_norm": 0.5421009659767151,
      "learning_rate": 0.00012179863147605081,
      "loss": 0.1996,
      "step": 7803
    },
    {
      "epoch": 1.221665623043206,
      "grad_norm": 1.131841778755188,
      "learning_rate": 0.00012177419354838708,
      "loss": 0.2315,
      "step": 7804
    },
    {
      "epoch": 1.2218221665623044,
      "grad_norm": 1.3324615955352783,
      "learning_rate": 0.00012174975562072336,
      "loss": 0.4474,
      "step": 7805
    },
    {
      "epoch": 1.2219787100814026,
      "grad_norm": 0.9257820248603821,
      "learning_rate": 0.00012172531769305962,
      "loss": 0.4004,
      "step": 7806
    },
    {
      "epoch": 1.222135253600501,
      "grad_norm": 2.541618585586548,
      "learning_rate": 0.00012170087976539587,
      "loss": 0.8599,
      "step": 7807
    },
    {
      "epoch": 1.2222917971195992,
      "grad_norm": 1.269856333732605,
      "learning_rate": 0.00012167644183773215,
      "loss": 0.3127,
      "step": 7808
    },
    {
      "epoch": 1.2224483406386977,
      "grad_norm": 1.0655205249786377,
      "learning_rate": 0.00012165200391006842,
      "loss": 0.3841,
      "step": 7809
    },
    {
      "epoch": 1.222604884157796,
      "grad_norm": 1.21709406375885,
      "learning_rate": 0.00012162756598240467,
      "loss": 0.472,
      "step": 7810
    },
    {
      "epoch": 1.222761427676894,
      "grad_norm": 1.1578439474105835,
      "learning_rate": 0.00012160312805474095,
      "loss": 0.4834,
      "step": 7811
    },
    {
      "epoch": 1.2229179711959925,
      "grad_norm": 1.1873836517333984,
      "learning_rate": 0.00012157869012707721,
      "loss": 0.462,
      "step": 7812
    },
    {
      "epoch": 1.2230745147150908,
      "grad_norm": 1.573827862739563,
      "learning_rate": 0.00012155425219941347,
      "loss": 0.7357,
      "step": 7813
    },
    {
      "epoch": 1.2232310582341892,
      "grad_norm": 2.879338502883911,
      "learning_rate": 0.00012152981427174975,
      "loss": 0.8596,
      "step": 7814
    },
    {
      "epoch": 1.2233876017532874,
      "grad_norm": 1.6422525644302368,
      "learning_rate": 0.000121505376344086,
      "loss": 0.2993,
      "step": 7815
    },
    {
      "epoch": 1.2235441452723856,
      "grad_norm": 1.4364542961120605,
      "learning_rate": 0.00012148093841642227,
      "loss": 0.3512,
      "step": 7816
    },
    {
      "epoch": 1.223700688791484,
      "grad_norm": 2.8140549659729004,
      "learning_rate": 0.00012145650048875855,
      "loss": 0.4175,
      "step": 7817
    },
    {
      "epoch": 1.2238572323105823,
      "grad_norm": 1.8453247547149658,
      "learning_rate": 0.00012143206256109481,
      "loss": 0.3645,
      "step": 7818
    },
    {
      "epoch": 1.2240137758296807,
      "grad_norm": 5.20889139175415,
      "learning_rate": 0.00012140762463343106,
      "loss": 0.8623,
      "step": 7819
    },
    {
      "epoch": 1.224170319348779,
      "grad_norm": 2.917222261428833,
      "learning_rate": 0.00012138318670576734,
      "loss": 0.7932,
      "step": 7820
    },
    {
      "epoch": 1.2243268628678772,
      "grad_norm": 3.251565933227539,
      "learning_rate": 0.00012135874877810361,
      "loss": 1.2039,
      "step": 7821
    },
    {
      "epoch": 1.2244834063869756,
      "grad_norm": 2.457594871520996,
      "learning_rate": 0.00012133431085043986,
      "loss": 0.6937,
      "step": 7822
    },
    {
      "epoch": 1.2246399499060738,
      "grad_norm": 1.1662980318069458,
      "learning_rate": 0.00012130987292277614,
      "loss": 0.5691,
      "step": 7823
    },
    {
      "epoch": 1.2247964934251723,
      "grad_norm": 4.571430206298828,
      "learning_rate": 0.0001212854349951124,
      "loss": 1.1899,
      "step": 7824
    },
    {
      "epoch": 1.2249530369442705,
      "grad_norm": 3.9596447944641113,
      "learning_rate": 0.00012126099706744867,
      "loss": 0.8965,
      "step": 7825
    },
    {
      "epoch": 1.2251095804633687,
      "grad_norm": 3.642796039581299,
      "learning_rate": 0.00012123655913978494,
      "loss": 1.2272,
      "step": 7826
    },
    {
      "epoch": 1.2252661239824671,
      "grad_norm": 2.4847941398620605,
      "learning_rate": 0.0001212121212121212,
      "loss": 1.4294,
      "step": 7827
    },
    {
      "epoch": 1.2254226675015654,
      "grad_norm": 5.294003963470459,
      "learning_rate": 0.00012118768328445746,
      "loss": 1.0586,
      "step": 7828
    },
    {
      "epoch": 1.2255792110206638,
      "grad_norm": 4.286756992340088,
      "learning_rate": 0.00012116324535679374,
      "loss": 1.0392,
      "step": 7829
    },
    {
      "epoch": 1.225735754539762,
      "grad_norm": 3.1308891773223877,
      "learning_rate": 0.00012113880742913,
      "loss": 0.8807,
      "step": 7830
    },
    {
      "epoch": 1.2258922980588602,
      "grad_norm": 4.852659702301025,
      "learning_rate": 0.00012111436950146625,
      "loss": 1.1662,
      "step": 7831
    },
    {
      "epoch": 1.2260488415779587,
      "grad_norm": 4.063992500305176,
      "learning_rate": 0.00012108993157380253,
      "loss": 1.4974,
      "step": 7832
    },
    {
      "epoch": 1.226205385097057,
      "grad_norm": 3.3738880157470703,
      "learning_rate": 0.0001210654936461388,
      "loss": 1.2586,
      "step": 7833
    },
    {
      "epoch": 1.2263619286161553,
      "grad_norm": 2.590223550796509,
      "learning_rate": 0.00012104105571847505,
      "loss": 0.4073,
      "step": 7834
    },
    {
      "epoch": 1.2265184721352536,
      "grad_norm": 2.9616923332214355,
      "learning_rate": 0.00012101661779081133,
      "loss": 0.8075,
      "step": 7835
    },
    {
      "epoch": 1.226675015654352,
      "grad_norm": 3.8399484157562256,
      "learning_rate": 0.00012099217986314759,
      "loss": 0.8251,
      "step": 7836
    },
    {
      "epoch": 1.2268315591734502,
      "grad_norm": 2.232974052429199,
      "learning_rate": 0.00012096774193548386,
      "loss": 0.8066,
      "step": 7837
    },
    {
      "epoch": 1.2269881026925484,
      "grad_norm": 2.5134973526000977,
      "learning_rate": 0.00012094330400782014,
      "loss": 0.7184,
      "step": 7838
    },
    {
      "epoch": 1.2271446462116469,
      "grad_norm": 0.7725585699081421,
      "learning_rate": 0.00012091886608015639,
      "loss": 0.2581,
      "step": 7839
    },
    {
      "epoch": 1.227301189730745,
      "grad_norm": 0.4113602936267853,
      "learning_rate": 0.00012089442815249265,
      "loss": 0.21,
      "step": 7840
    },
    {
      "epoch": 1.2274577332498435,
      "grad_norm": 0.737152636051178,
      "learning_rate": 0.00012086999022482893,
      "loss": 0.2286,
      "step": 7841
    },
    {
      "epoch": 1.2276142767689417,
      "grad_norm": 0.8976015448570251,
      "learning_rate": 0.00012084555229716518,
      "loss": 0.197,
      "step": 7842
    },
    {
      "epoch": 1.2277708202880402,
      "grad_norm": 0.47844988107681274,
      "learning_rate": 0.00012082111436950145,
      "loss": 0.2104,
      "step": 7843
    },
    {
      "epoch": 1.2279273638071384,
      "grad_norm": 0.49797824025154114,
      "learning_rate": 0.00012079667644183772,
      "loss": 0.2007,
      "step": 7844
    },
    {
      "epoch": 1.2280839073262366,
      "grad_norm": 0.6173357963562012,
      "learning_rate": 0.00012077223851417399,
      "loss": 0.2887,
      "step": 7845
    },
    {
      "epoch": 1.228240450845335,
      "grad_norm": 0.406225323677063,
      "learning_rate": 0.00012074780058651024,
      "loss": 0.1982,
      "step": 7846
    },
    {
      "epoch": 1.2283969943644333,
      "grad_norm": 0.7921119332313538,
      "learning_rate": 0.00012072336265884652,
      "loss": 0.3479,
      "step": 7847
    },
    {
      "epoch": 1.2285535378835317,
      "grad_norm": 1.66916024684906,
      "learning_rate": 0.00012069892473118278,
      "loss": 0.3086,
      "step": 7848
    },
    {
      "epoch": 1.22871008140263,
      "grad_norm": 0.6646075248718262,
      "learning_rate": 0.00012067448680351905,
      "loss": 0.3681,
      "step": 7849
    },
    {
      "epoch": 1.2288666249217282,
      "grad_norm": 1.726301670074463,
      "learning_rate": 0.00012065004887585533,
      "loss": 0.6128,
      "step": 7850
    },
    {
      "epoch": 1.2290231684408266,
      "grad_norm": 0.8818532824516296,
      "learning_rate": 0.00012062561094819158,
      "loss": 0.3387,
      "step": 7851
    },
    {
      "epoch": 1.2291797119599248,
      "grad_norm": 0.7736045718193054,
      "learning_rate": 0.00012060117302052784,
      "loss": 0.2012,
      "step": 7852
    },
    {
      "epoch": 1.2293362554790233,
      "grad_norm": 0.7957243919372559,
      "learning_rate": 0.00012057673509286412,
      "loss": 0.3497,
      "step": 7853
    },
    {
      "epoch": 1.2294927989981215,
      "grad_norm": 1.0294607877731323,
      "learning_rate": 0.00012055229716520037,
      "loss": 0.3375,
      "step": 7854
    },
    {
      "epoch": 1.2296493425172197,
      "grad_norm": 3.3992855548858643,
      "learning_rate": 0.00012052785923753664,
      "loss": 0.4712,
      "step": 7855
    },
    {
      "epoch": 1.2298058860363181,
      "grad_norm": 2.0325255393981934,
      "learning_rate": 0.00012050342130987292,
      "loss": 0.208,
      "step": 7856
    },
    {
      "epoch": 1.2299624295554163,
      "grad_norm": 1.33167564868927,
      "learning_rate": 0.00012047898338220918,
      "loss": 0.3885,
      "step": 7857
    },
    {
      "epoch": 1.2301189730745148,
      "grad_norm": 2.1962788105010986,
      "learning_rate": 0.00012045454545454543,
      "loss": 0.3926,
      "step": 7858
    },
    {
      "epoch": 1.230275516593613,
      "grad_norm": 1.1187773942947388,
      "learning_rate": 0.00012043010752688171,
      "loss": 0.5303,
      "step": 7859
    },
    {
      "epoch": 1.2304320601127112,
      "grad_norm": 1.26564621925354,
      "learning_rate": 0.00012040566959921797,
      "loss": 0.373,
      "step": 7860
    },
    {
      "epoch": 1.2305886036318097,
      "grad_norm": 1.3402997255325317,
      "learning_rate": 0.00012038123167155424,
      "loss": 0.36,
      "step": 7861
    },
    {
      "epoch": 1.2307451471509079,
      "grad_norm": 1.738292932510376,
      "learning_rate": 0.00012035679374389052,
      "loss": 0.5391,
      "step": 7862
    },
    {
      "epoch": 1.2309016906700063,
      "grad_norm": 2.344097137451172,
      "learning_rate": 0.00012033235581622677,
      "loss": 0.619,
      "step": 7863
    },
    {
      "epoch": 1.2310582341891045,
      "grad_norm": 1.4428097009658813,
      "learning_rate": 0.00012030791788856303,
      "loss": 0.3952,
      "step": 7864
    },
    {
      "epoch": 1.2312147777082028,
      "grad_norm": 2.1929662227630615,
      "learning_rate": 0.00012028347996089931,
      "loss": 0.578,
      "step": 7865
    },
    {
      "epoch": 1.2313713212273012,
      "grad_norm": 1.8685600757598877,
      "learning_rate": 0.00012025904203323556,
      "loss": 0.8856,
      "step": 7866
    },
    {
      "epoch": 1.2315278647463994,
      "grad_norm": 2.1994404792785645,
      "learning_rate": 0.00012023460410557183,
      "loss": 0.4865,
      "step": 7867
    },
    {
      "epoch": 1.2316844082654979,
      "grad_norm": 8.819302558898926,
      "learning_rate": 0.00012021016617790811,
      "loss": 1.14,
      "step": 7868
    },
    {
      "epoch": 1.231840951784596,
      "grad_norm": 2.020968437194824,
      "learning_rate": 0.00012018572825024437,
      "loss": 0.8263,
      "step": 7869
    },
    {
      "epoch": 1.2319974953036945,
      "grad_norm": 1.5616576671600342,
      "learning_rate": 0.00012016129032258062,
      "loss": 0.5087,
      "step": 7870
    },
    {
      "epoch": 1.2321540388227927,
      "grad_norm": 1.2136770486831665,
      "learning_rate": 0.0001201368523949169,
      "loss": 0.6024,
      "step": 7871
    },
    {
      "epoch": 1.2323105823418912,
      "grad_norm": 1.3263853788375854,
      "learning_rate": 0.00012011241446725317,
      "loss": 0.585,
      "step": 7872
    },
    {
      "epoch": 1.2324671258609894,
      "grad_norm": 3.187013864517212,
      "learning_rate": 0.00012008797653958943,
      "loss": 0.9524,
      "step": 7873
    },
    {
      "epoch": 1.2326236693800876,
      "grad_norm": 4.589766979217529,
      "learning_rate": 0.00012006353861192571,
      "loss": 0.8727,
      "step": 7874
    },
    {
      "epoch": 1.232780212899186,
      "grad_norm": 1.6198341846466064,
      "learning_rate": 0.00012003910068426196,
      "loss": 0.5955,
      "step": 7875
    },
    {
      "epoch": 1.2329367564182843,
      "grad_norm": 2.6360957622528076,
      "learning_rate": 0.00012001466275659823,
      "loss": 0.9213,
      "step": 7876
    },
    {
      "epoch": 1.2330932999373827,
      "grad_norm": 2.5727317333221436,
      "learning_rate": 0.0001199902248289345,
      "loss": 0.9638,
      "step": 7877
    },
    {
      "epoch": 1.233249843456481,
      "grad_norm": 3.1968390941619873,
      "learning_rate": 0.00011996578690127075,
      "loss": 1.1984,
      "step": 7878
    },
    {
      "epoch": 1.2334063869755791,
      "grad_norm": 5.276712417602539,
      "learning_rate": 0.00011994134897360702,
      "loss": 0.7227,
      "step": 7879
    },
    {
      "epoch": 1.2335629304946776,
      "grad_norm": 2.5714964866638184,
      "learning_rate": 0.0001199169110459433,
      "loss": 1.0927,
      "step": 7880
    },
    {
      "epoch": 1.2337194740137758,
      "grad_norm": 2.9321186542510986,
      "learning_rate": 0.00011989247311827956,
      "loss": 0.8852,
      "step": 7881
    },
    {
      "epoch": 1.2338760175328742,
      "grad_norm": 3.1245834827423096,
      "learning_rate": 0.00011986803519061581,
      "loss": 1.2498,
      "step": 7882
    },
    {
      "epoch": 1.2340325610519725,
      "grad_norm": 3.5027828216552734,
      "learning_rate": 0.00011984359726295209,
      "loss": 1.1943,
      "step": 7883
    },
    {
      "epoch": 1.2341891045710707,
      "grad_norm": 2.125910520553589,
      "learning_rate": 0.00011981915933528836,
      "loss": 1.0542,
      "step": 7884
    },
    {
      "epoch": 1.2343456480901691,
      "grad_norm": 1.8235660791397095,
      "learning_rate": 0.00011979472140762462,
      "loss": 0.5816,
      "step": 7885
    },
    {
      "epoch": 1.2345021916092673,
      "grad_norm": 3.679835796356201,
      "learning_rate": 0.00011977028347996089,
      "loss": 1.4135,
      "step": 7886
    },
    {
      "epoch": 1.2346587351283658,
      "grad_norm": 6.588844299316406,
      "learning_rate": 0.00011974584555229715,
      "loss": 1.178,
      "step": 7887
    },
    {
      "epoch": 1.234815278647464,
      "grad_norm": 2.3670153617858887,
      "learning_rate": 0.00011972140762463342,
      "loss": 1.2308,
      "step": 7888
    },
    {
      "epoch": 1.2349718221665622,
      "grad_norm": 0.49170297384262085,
      "learning_rate": 0.0001196969696969697,
      "loss": 0.282,
      "step": 7889
    },
    {
      "epoch": 1.2351283656856606,
      "grad_norm": 0.5877807140350342,
      "learning_rate": 0.00011967253176930595,
      "loss": 0.2686,
      "step": 7890
    },
    {
      "epoch": 1.2352849092047589,
      "grad_norm": 0.6519326567649841,
      "learning_rate": 0.00011964809384164221,
      "loss": 0.2191,
      "step": 7891
    },
    {
      "epoch": 1.2354414527238573,
      "grad_norm": 0.7324579954147339,
      "learning_rate": 0.00011962365591397849,
      "loss": 0.2397,
      "step": 7892
    },
    {
      "epoch": 1.2355979962429555,
      "grad_norm": 0.5463114976882935,
      "learning_rate": 0.00011959921798631475,
      "loss": 0.159,
      "step": 7893
    },
    {
      "epoch": 1.2357545397620537,
      "grad_norm": 0.6586346626281738,
      "learning_rate": 0.000119574780058651,
      "loss": 0.2038,
      "step": 7894
    },
    {
      "epoch": 1.2359110832811522,
      "grad_norm": 0.4430273175239563,
      "learning_rate": 0.00011955034213098728,
      "loss": 0.2114,
      "step": 7895
    },
    {
      "epoch": 1.2360676268002504,
      "grad_norm": 0.7554447054862976,
      "learning_rate": 0.00011952590420332355,
      "loss": 0.2591,
      "step": 7896
    },
    {
      "epoch": 1.2362241703193488,
      "grad_norm": 0.531594455242157,
      "learning_rate": 0.00011950146627565981,
      "loss": 0.2403,
      "step": 7897
    },
    {
      "epoch": 1.236380713838447,
      "grad_norm": 0.8028531670570374,
      "learning_rate": 0.00011947702834799608,
      "loss": 0.2436,
      "step": 7898
    },
    {
      "epoch": 1.2365372573575455,
      "grad_norm": 0.8752454519271851,
      "learning_rate": 0.00011945259042033234,
      "loss": 0.3462,
      "step": 7899
    },
    {
      "epoch": 1.2366938008766437,
      "grad_norm": 1.4788440465927124,
      "learning_rate": 0.00011942815249266861,
      "loss": 0.621,
      "step": 7900
    },
    {
      "epoch": 1.236850344395742,
      "grad_norm": 1.0307644605636597,
      "learning_rate": 0.00011940371456500489,
      "loss": 0.6322,
      "step": 7901
    },
    {
      "epoch": 1.2370068879148404,
      "grad_norm": 0.6076186299324036,
      "learning_rate": 0.00011937927663734114,
      "loss": 0.3076,
      "step": 7902
    },
    {
      "epoch": 1.2371634314339386,
      "grad_norm": 1.0006450414657593,
      "learning_rate": 0.0001193548387096774,
      "loss": 0.2975,
      "step": 7903
    },
    {
      "epoch": 1.237319974953037,
      "grad_norm": 1.008905053138733,
      "learning_rate": 0.00011933040078201368,
      "loss": 0.3799,
      "step": 7904
    },
    {
      "epoch": 1.2374765184721352,
      "grad_norm": 3.882817506790161,
      "learning_rate": 0.00011930596285434995,
      "loss": 0.9816,
      "step": 7905
    },
    {
      "epoch": 1.2376330619912337,
      "grad_norm": 1.3156336545944214,
      "learning_rate": 0.0001192815249266862,
      "loss": 0.4128,
      "step": 7906
    },
    {
      "epoch": 1.237789605510332,
      "grad_norm": 1.1643537282943726,
      "learning_rate": 0.00011925708699902247,
      "loss": 0.5488,
      "step": 7907
    },
    {
      "epoch": 1.2379461490294301,
      "grad_norm": 1.2683813571929932,
      "learning_rate": 0.00011923264907135874,
      "loss": 0.595,
      "step": 7908
    },
    {
      "epoch": 1.2381026925485286,
      "grad_norm": 1.6245311498641968,
      "learning_rate": 0.000119208211143695,
      "loss": 0.3329,
      "step": 7909
    },
    {
      "epoch": 1.2382592360676268,
      "grad_norm": 1.5488073825836182,
      "learning_rate": 0.00011918377321603127,
      "loss": 0.6928,
      "step": 7910
    },
    {
      "epoch": 1.2384157795867252,
      "grad_norm": 1.1169195175170898,
      "learning_rate": 0.00011915933528836753,
      "loss": 0.3563,
      "step": 7911
    },
    {
      "epoch": 1.2385723231058234,
      "grad_norm": 1.5224010944366455,
      "learning_rate": 0.0001191348973607038,
      "loss": 0.7176,
      "step": 7912
    },
    {
      "epoch": 1.2387288666249217,
      "grad_norm": 0.7106965780258179,
      "learning_rate": 0.00011911045943304008,
      "loss": 0.3416,
      "step": 7913
    },
    {
      "epoch": 1.23888541014402,
      "grad_norm": 1.0206400156021118,
      "learning_rate": 0.00011908602150537633,
      "loss": 0.3449,
      "step": 7914
    },
    {
      "epoch": 1.2390419536631183,
      "grad_norm": 1.643692135810852,
      "learning_rate": 0.00011906158357771259,
      "loss": 0.744,
      "step": 7915
    },
    {
      "epoch": 1.2391984971822168,
      "grad_norm": 2.5953471660614014,
      "learning_rate": 0.00011903714565004887,
      "loss": 0.9874,
      "step": 7916
    },
    {
      "epoch": 1.239355040701315,
      "grad_norm": 2.029824733734131,
      "learning_rate": 0.00011901270772238514,
      "loss": 0.7063,
      "step": 7917
    },
    {
      "epoch": 1.2395115842204132,
      "grad_norm": 2.1811952590942383,
      "learning_rate": 0.00011898826979472139,
      "loss": 0.6537,
      "step": 7918
    },
    {
      "epoch": 1.2396681277395116,
      "grad_norm": 2.453786611557007,
      "learning_rate": 0.00011896383186705767,
      "loss": 0.9151,
      "step": 7919
    },
    {
      "epoch": 1.2398246712586098,
      "grad_norm": 2.0153915882110596,
      "learning_rate": 0.00011893939393939393,
      "loss": 1.1769,
      "step": 7920
    },
    {
      "epoch": 1.2399812147777083,
      "grad_norm": 2.056471824645996,
      "learning_rate": 0.0001189149560117302,
      "loss": 0.6084,
      "step": 7921
    },
    {
      "epoch": 1.2401377582968065,
      "grad_norm": 1.4912649393081665,
      "learning_rate": 0.00011889051808406646,
      "loss": 1.0018,
      "step": 7922
    },
    {
      "epoch": 1.2402943018159047,
      "grad_norm": 1.68226158618927,
      "learning_rate": 0.00011886608015640273,
      "loss": 1.2432,
      "step": 7923
    },
    {
      "epoch": 1.2404508453350032,
      "grad_norm": 1.0560473203659058,
      "learning_rate": 0.00011884164222873899,
      "loss": 0.5132,
      "step": 7924
    },
    {
      "epoch": 1.2406073888541014,
      "grad_norm": 1.9787427186965942,
      "learning_rate": 0.00011881720430107527,
      "loss": 1.3016,
      "step": 7925
    },
    {
      "epoch": 1.2407639323731998,
      "grad_norm": 1.528519868850708,
      "learning_rate": 0.00011879276637341152,
      "loss": 0.4644,
      "step": 7926
    },
    {
      "epoch": 1.240920475892298,
      "grad_norm": 1.8350005149841309,
      "learning_rate": 0.00011876832844574778,
      "loss": 1.06,
      "step": 7927
    },
    {
      "epoch": 1.2410770194113963,
      "grad_norm": 2.0662553310394287,
      "learning_rate": 0.00011874389051808406,
      "loss": 0.6492,
      "step": 7928
    },
    {
      "epoch": 1.2412335629304947,
      "grad_norm": 2.514481544494629,
      "learning_rate": 0.00011871945259042033,
      "loss": 1.7245,
      "step": 7929
    },
    {
      "epoch": 1.241390106449593,
      "grad_norm": 1.8698464632034302,
      "learning_rate": 0.00011869501466275658,
      "loss": 1.2266,
      "step": 7930
    },
    {
      "epoch": 1.2415466499686914,
      "grad_norm": 1.368855595588684,
      "learning_rate": 0.00011867057673509286,
      "loss": 0.7572,
      "step": 7931
    },
    {
      "epoch": 1.2417031934877896,
      "grad_norm": 2.3699071407318115,
      "learning_rate": 0.00011864613880742912,
      "loss": 0.8298,
      "step": 7932
    },
    {
      "epoch": 1.241859737006888,
      "grad_norm": 1.3850302696228027,
      "learning_rate": 0.00011862170087976539,
      "loss": 0.4536,
      "step": 7933
    },
    {
      "epoch": 1.2420162805259862,
      "grad_norm": 4.977390289306641,
      "learning_rate": 0.00011859726295210165,
      "loss": 1.0689,
      "step": 7934
    },
    {
      "epoch": 1.2421728240450844,
      "grad_norm": 2.6507201194763184,
      "learning_rate": 0.00011857282502443792,
      "loss": 0.8295,
      "step": 7935
    },
    {
      "epoch": 1.2423293675641829,
      "grad_norm": 1.9919297695159912,
      "learning_rate": 0.00011854838709677418,
      "loss": 0.767,
      "step": 7936
    },
    {
      "epoch": 1.242485911083281,
      "grad_norm": 2.1812164783477783,
      "learning_rate": 0.00011852394916911046,
      "loss": 0.7606,
      "step": 7937
    },
    {
      "epoch": 1.2426424546023795,
      "grad_norm": 4.220646381378174,
      "learning_rate": 0.00011849951124144671,
      "loss": 1.381,
      "step": 7938
    },
    {
      "epoch": 1.2427989981214778,
      "grad_norm": 0.48588910698890686,
      "learning_rate": 0.00011847507331378298,
      "loss": 0.2814,
      "step": 7939
    },
    {
      "epoch": 1.2429555416405762,
      "grad_norm": 0.5703171491622925,
      "learning_rate": 0.00011845063538611925,
      "loss": 0.2328,
      "step": 7940
    },
    {
      "epoch": 1.2431120851596744,
      "grad_norm": 0.4970242977142334,
      "learning_rate": 0.00011842619745845552,
      "loss": 0.1882,
      "step": 7941
    },
    {
      "epoch": 1.2432686286787726,
      "grad_norm": 0.5462607741355896,
      "learning_rate": 0.00011840175953079177,
      "loss": 0.1734,
      "step": 7942
    },
    {
      "epoch": 1.243425172197871,
      "grad_norm": 0.7907721400260925,
      "learning_rate": 0.00011837732160312805,
      "loss": 0.2375,
      "step": 7943
    },
    {
      "epoch": 1.2435817157169693,
      "grad_norm": 0.4387291669845581,
      "learning_rate": 0.00011835288367546431,
      "loss": 0.1791,
      "step": 7944
    },
    {
      "epoch": 1.2437382592360677,
      "grad_norm": 0.5500348210334778,
      "learning_rate": 0.00011832844574780058,
      "loss": 0.2238,
      "step": 7945
    },
    {
      "epoch": 1.243894802755166,
      "grad_norm": 0.6439167261123657,
      "learning_rate": 0.00011830400782013684,
      "loss": 0.2178,
      "step": 7946
    },
    {
      "epoch": 1.2440513462742642,
      "grad_norm": 0.8734773397445679,
      "learning_rate": 0.00011827956989247311,
      "loss": 0.2993,
      "step": 7947
    },
    {
      "epoch": 1.2442078897933626,
      "grad_norm": 1.8835079669952393,
      "learning_rate": 0.00011825513196480937,
      "loss": 0.3273,
      "step": 7948
    },
    {
      "epoch": 1.2443644333124608,
      "grad_norm": 0.723751425743103,
      "learning_rate": 0.00011823069403714565,
      "loss": 0.1828,
      "step": 7949
    },
    {
      "epoch": 1.2445209768315593,
      "grad_norm": 1.6862190961837769,
      "learning_rate": 0.0001182062561094819,
      "loss": 0.3214,
      "step": 7950
    },
    {
      "epoch": 1.2446775203506575,
      "grad_norm": 0.7363371849060059,
      "learning_rate": 0.00011818181818181817,
      "loss": 0.2662,
      "step": 7951
    },
    {
      "epoch": 1.2448340638697557,
      "grad_norm": 1.2656689882278442,
      "learning_rate": 0.00011815738025415444,
      "loss": 0.3986,
      "step": 7952
    },
    {
      "epoch": 1.2449906073888541,
      "grad_norm": 0.9024664759635925,
      "learning_rate": 0.00011813294232649071,
      "loss": 0.405,
      "step": 7953
    },
    {
      "epoch": 1.2451471509079524,
      "grad_norm": 0.8185746669769287,
      "learning_rate": 0.00011810850439882696,
      "loss": 0.4005,
      "step": 7954
    },
    {
      "epoch": 1.2453036944270508,
      "grad_norm": 0.8885353207588196,
      "learning_rate": 0.00011808406647116324,
      "loss": 0.1425,
      "step": 7955
    },
    {
      "epoch": 1.245460237946149,
      "grad_norm": 6.204992294311523,
      "learning_rate": 0.0001180596285434995,
      "loss": 0.3497,
      "step": 7956
    },
    {
      "epoch": 1.2456167814652472,
      "grad_norm": 1.7547203302383423,
      "learning_rate": 0.00011803519061583576,
      "loss": 0.3987,
      "step": 7957
    },
    {
      "epoch": 1.2457733249843457,
      "grad_norm": 1.6890156269073486,
      "learning_rate": 0.00011801075268817203,
      "loss": 0.2755,
      "step": 7958
    },
    {
      "epoch": 1.245929868503444,
      "grad_norm": 1.954634189605713,
      "learning_rate": 0.0001179863147605083,
      "loss": 0.6202,
      "step": 7959
    },
    {
      "epoch": 1.2460864120225423,
      "grad_norm": 1.5542676448822021,
      "learning_rate": 0.00011796187683284456,
      "loss": 0.2909,
      "step": 7960
    },
    {
      "epoch": 1.2462429555416406,
      "grad_norm": 2.495922565460205,
      "learning_rate": 0.00011793743890518084,
      "loss": 0.6644,
      "step": 7961
    },
    {
      "epoch": 1.2463994990607388,
      "grad_norm": 1.6161011457443237,
      "learning_rate": 0.00011791300097751709,
      "loss": 0.3718,
      "step": 7962
    },
    {
      "epoch": 1.2465560425798372,
      "grad_norm": 1.5443997383117676,
      "learning_rate": 0.00011788856304985336,
      "loss": 0.397,
      "step": 7963
    },
    {
      "epoch": 1.2467125860989354,
      "grad_norm": 1.664036750793457,
      "learning_rate": 0.00011786412512218964,
      "loss": 0.3962,
      "step": 7964
    },
    {
      "epoch": 1.2468691296180339,
      "grad_norm": 1.7987858057022095,
      "learning_rate": 0.0001178396871945259,
      "loss": 0.4728,
      "step": 7965
    },
    {
      "epoch": 1.247025673137132,
      "grad_norm": 2.9872450828552246,
      "learning_rate": 0.00011781524926686215,
      "loss": 0.5981,
      "step": 7966
    },
    {
      "epoch": 1.2471822166562305,
      "grad_norm": 1.4099814891815186,
      "learning_rate": 0.00011779081133919843,
      "loss": 0.8175,
      "step": 7967
    },
    {
      "epoch": 1.2473387601753287,
      "grad_norm": 2.3960330486297607,
      "learning_rate": 0.0001177663734115347,
      "loss": 0.937,
      "step": 7968
    },
    {
      "epoch": 1.247495303694427,
      "grad_norm": 1.7411472797393799,
      "learning_rate": 0.00011774193548387095,
      "loss": 0.8408,
      "step": 7969
    },
    {
      "epoch": 1.2476518472135254,
      "grad_norm": 1.6002928018569946,
      "learning_rate": 0.00011771749755620722,
      "loss": 0.6916,
      "step": 7970
    },
    {
      "epoch": 1.2478083907326236,
      "grad_norm": 2.1395416259765625,
      "learning_rate": 0.00011769305962854349,
      "loss": 0.6554,
      "step": 7971
    },
    {
      "epoch": 1.247964934251722,
      "grad_norm": 2.8727903366088867,
      "learning_rate": 0.00011766862170087975,
      "loss": 0.5139,
      "step": 7972
    },
    {
      "epoch": 1.2481214777708203,
      "grad_norm": 1.354569435119629,
      "learning_rate": 0.00011764418377321603,
      "loss": 0.5492,
      "step": 7973
    },
    {
      "epoch": 1.2482780212899187,
      "grad_norm": 4.999713897705078,
      "learning_rate": 0.00011761974584555228,
      "loss": 0.7196,
      "step": 7974
    },
    {
      "epoch": 1.248434564809017,
      "grad_norm": 3.1970131397247314,
      "learning_rate": 0.00011759530791788855,
      "loss": 1.0452,
      "step": 7975
    },
    {
      "epoch": 1.2485911083281152,
      "grad_norm": 3.8125860691070557,
      "learning_rate": 0.00011757086999022483,
      "loss": 1.3657,
      "step": 7976
    },
    {
      "epoch": 1.2487476518472136,
      "grad_norm": 2.1403331756591797,
      "learning_rate": 0.00011754643206256109,
      "loss": 0.9489,
      "step": 7977
    },
    {
      "epoch": 1.2489041953663118,
      "grad_norm": 1.500003457069397,
      "learning_rate": 0.00011752199413489734,
      "loss": 0.7548,
      "step": 7978
    },
    {
      "epoch": 1.2490607388854102,
      "grad_norm": 5.040395259857178,
      "learning_rate": 0.00011749755620723362,
      "loss": 0.5617,
      "step": 7979
    },
    {
      "epoch": 1.2492172824045085,
      "grad_norm": 2.7591681480407715,
      "learning_rate": 0.00011747311827956989,
      "loss": 1.0901,
      "step": 7980
    },
    {
      "epoch": 1.2493738259236067,
      "grad_norm": 1.9304205179214478,
      "learning_rate": 0.00011744868035190614,
      "loss": 0.8664,
      "step": 7981
    },
    {
      "epoch": 1.2495303694427051,
      "grad_norm": 3.9833483695983887,
      "learning_rate": 0.00011742424242424242,
      "loss": 0.7079,
      "step": 7982
    },
    {
      "epoch": 1.2496869129618033,
      "grad_norm": 2.4207115173339844,
      "learning_rate": 0.00011739980449657868,
      "loss": 1.1423,
      "step": 7983
    },
    {
      "epoch": 1.2498434564809018,
      "grad_norm": 2.918473958969116,
      "learning_rate": 0.00011737536656891495,
      "loss": 0.9182,
      "step": 7984
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.54889714717865,
      "learning_rate": 0.00011735092864125122,
      "loss": 0.3946,
      "step": 7985
    },
    {
      "epoch": 1.2501565435190982,
      "grad_norm": 2.440833806991577,
      "learning_rate": 0.00011732649071358748,
      "loss": 0.7028,
      "step": 7986
    },
    {
      "epoch": 1.2503130870381967,
      "grad_norm": 3.6599295139312744,
      "learning_rate": 0.00011730205278592374,
      "loss": 0.6834,
      "step": 7987
    },
    {
      "epoch": 1.2504696305572949,
      "grad_norm": 2.046987533569336,
      "learning_rate": 0.00011727761485826002,
      "loss": 0.59,
      "step": 7988
    },
    {
      "epoch": 1.2506261740763933,
      "grad_norm": 0.6683407425880432,
      "learning_rate": 0.00011725317693059628,
      "loss": 0.1788,
      "step": 7989
    },
    {
      "epoch": 1.2507827175954915,
      "grad_norm": 0.6008917093276978,
      "learning_rate": 0.00011722873900293253,
      "loss": 0.1947,
      "step": 7990
    },
    {
      "epoch": 1.2509392611145898,
      "grad_norm": 0.8003541827201843,
      "learning_rate": 0.00011720430107526881,
      "loss": 0.2844,
      "step": 7991
    },
    {
      "epoch": 1.2510958046336882,
      "grad_norm": 0.4843752980232239,
      "learning_rate": 0.00011717986314760508,
      "loss": 0.3327,
      "step": 7992
    },
    {
      "epoch": 1.2512523481527864,
      "grad_norm": 0.6065008640289307,
      "learning_rate": 0.00011715542521994133,
      "loss": 0.2476,
      "step": 7993
    },
    {
      "epoch": 1.2514088916718848,
      "grad_norm": 0.4822901785373688,
      "learning_rate": 0.00011713098729227761,
      "loss": 0.2134,
      "step": 7994
    },
    {
      "epoch": 1.251565435190983,
      "grad_norm": 0.6061071753501892,
      "learning_rate": 0.00011710654936461387,
      "loss": 0.198,
      "step": 7995
    },
    {
      "epoch": 1.2517219787100813,
      "grad_norm": 0.7697739005088806,
      "learning_rate": 0.00011708211143695014,
      "loss": 0.3114,
      "step": 7996
    },
    {
      "epoch": 1.2518785222291797,
      "grad_norm": 0.9554542899131775,
      "learning_rate": 0.00011705767350928642,
      "loss": 0.5542,
      "step": 7997
    },
    {
      "epoch": 1.252035065748278,
      "grad_norm": 0.5352123379707336,
      "learning_rate": 0.00011703323558162267,
      "loss": 0.2086,
      "step": 7998
    },
    {
      "epoch": 1.2521916092673764,
      "grad_norm": 1.2916967868804932,
      "learning_rate": 0.00011700879765395893,
      "loss": 0.1956,
      "step": 7999
    },
    {
      "epoch": 1.2523481527864746,
      "grad_norm": 0.7141245603561401,
      "learning_rate": 0.00011698435972629521,
      "loss": 0.2089,
      "step": 8000
    },
    {
      "epoch": 1.2523481527864746,
      "eval_loss": 0.5047087669372559,
      "eval_runtime": 206.5621,
      "eval_samples_per_second": 59.948,
      "eval_steps_per_second": 3.747,
      "eval_wer": 0.3239156767324734,
      "step": 8000
    },
    {
      "epoch": 1.2525046963055728,
      "grad_norm": 0.8334164023399353,
      "learning_rate": 0.00011695992179863146,
      "loss": 0.4006,
      "step": 8001
    },
    {
      "epoch": 1.2526612398246713,
      "grad_norm": 0.8113241195678711,
      "learning_rate": 0.00011693548387096773,
      "loss": 0.3454,
      "step": 8002
    },
    {
      "epoch": 1.2528177833437697,
      "grad_norm": 0.6484245657920837,
      "learning_rate": 0.000116911045943304,
      "loss": 0.2453,
      "step": 8003
    },
    {
      "epoch": 1.252974326862868,
      "grad_norm": 0.7791106700897217,
      "learning_rate": 0.00011688660801564027,
      "loss": 0.2726,
      "step": 8004
    },
    {
      "epoch": 1.2531308703819661,
      "grad_norm": 1.1766725778579712,
      "learning_rate": 0.00011686217008797652,
      "loss": 0.4168,
      "step": 8005
    },
    {
      "epoch": 1.2532874139010646,
      "grad_norm": 1.342408537864685,
      "learning_rate": 0.0001168377321603128,
      "loss": 0.3907,
      "step": 8006
    },
    {
      "epoch": 1.2534439574201628,
      "grad_norm": 1.233473300933838,
      "learning_rate": 0.00011681329423264906,
      "loss": 0.3619,
      "step": 8007
    },
    {
      "epoch": 1.2536005009392612,
      "grad_norm": 1.4168598651885986,
      "learning_rate": 0.00011678885630498533,
      "loss": 0.4103,
      "step": 8008
    },
    {
      "epoch": 1.2537570444583594,
      "grad_norm": 1.5982956886291504,
      "learning_rate": 0.0001167644183773216,
      "loss": 0.433,
      "step": 8009
    },
    {
      "epoch": 1.2539135879774577,
      "grad_norm": 1.1311919689178467,
      "learning_rate": 0.00011673998044965786,
      "loss": 0.4682,
      "step": 8010
    },
    {
      "epoch": 1.254070131496556,
      "grad_norm": 1.4813308715820312,
      "learning_rate": 0.00011671554252199412,
      "loss": 0.3652,
      "step": 8011
    },
    {
      "epoch": 1.2542266750156543,
      "grad_norm": 1.0279951095581055,
      "learning_rate": 0.0001166911045943304,
      "loss": 0.5578,
      "step": 8012
    },
    {
      "epoch": 1.2543832185347528,
      "grad_norm": 1.014884352684021,
      "learning_rate": 0.00011666666666666665,
      "loss": 0.3986,
      "step": 8013
    },
    {
      "epoch": 1.254539762053851,
      "grad_norm": 2.5004680156707764,
      "learning_rate": 0.00011664222873900292,
      "loss": 0.5565,
      "step": 8014
    },
    {
      "epoch": 1.2546963055729492,
      "grad_norm": 1.6078928709030151,
      "learning_rate": 0.0001166177908113392,
      "loss": 0.5358,
      "step": 8015
    },
    {
      "epoch": 1.2548528490920476,
      "grad_norm": 1.8332239389419556,
      "learning_rate": 0.00011659335288367546,
      "loss": 0.2927,
      "step": 8016
    },
    {
      "epoch": 1.2550093926111459,
      "grad_norm": 1.8943898677825928,
      "learning_rate": 0.00011656891495601171,
      "loss": 0.5876,
      "step": 8017
    },
    {
      "epoch": 1.2551659361302443,
      "grad_norm": 1.8188652992248535,
      "learning_rate": 0.00011654447702834799,
      "loss": 0.4997,
      "step": 8018
    },
    {
      "epoch": 1.2553224796493425,
      "grad_norm": 2.3401331901550293,
      "learning_rate": 0.00011652003910068425,
      "loss": 0.3398,
      "step": 8019
    },
    {
      "epoch": 1.2554790231684407,
      "grad_norm": 1.9932831525802612,
      "learning_rate": 0.00011649560117302052,
      "loss": 0.7564,
      "step": 8020
    },
    {
      "epoch": 1.2556355666875392,
      "grad_norm": 2.9455413818359375,
      "learning_rate": 0.0001164711632453568,
      "loss": 0.6872,
      "step": 8021
    },
    {
      "epoch": 1.2557921102066374,
      "grad_norm": 1.4435310363769531,
      "learning_rate": 0.00011644672531769305,
      "loss": 0.6771,
      "step": 8022
    },
    {
      "epoch": 1.2559486537257358,
      "grad_norm": 3.489488363265991,
      "learning_rate": 0.00011642228739002931,
      "loss": 0.7648,
      "step": 8023
    },
    {
      "epoch": 1.256105197244834,
      "grad_norm": 1.4743152856826782,
      "learning_rate": 0.00011639784946236559,
      "loss": 0.4852,
      "step": 8024
    },
    {
      "epoch": 1.2562617407639323,
      "grad_norm": 3.376526355743408,
      "learning_rate": 0.00011637341153470184,
      "loss": 0.6416,
      "step": 8025
    },
    {
      "epoch": 1.2564182842830307,
      "grad_norm": 1.8703612089157104,
      "learning_rate": 0.00011634897360703811,
      "loss": 1.1278,
      "step": 8026
    },
    {
      "epoch": 1.256574827802129,
      "grad_norm": 2.682764768600464,
      "learning_rate": 0.00011632453567937439,
      "loss": 1.1146,
      "step": 8027
    },
    {
      "epoch": 1.2567313713212274,
      "grad_norm": 2.28507399559021,
      "learning_rate": 0.00011630009775171065,
      "loss": 1.0425,
      "step": 8028
    },
    {
      "epoch": 1.2568879148403256,
      "grad_norm": 4.47053861618042,
      "learning_rate": 0.0001162756598240469,
      "loss": 0.8179,
      "step": 8029
    },
    {
      "epoch": 1.2570444583594238,
      "grad_norm": 3.5144686698913574,
      "learning_rate": 0.00011625122189638318,
      "loss": 1.1578,
      "step": 8030
    },
    {
      "epoch": 1.2572010018785222,
      "grad_norm": 3.6762499809265137,
      "learning_rate": 0.00011622678396871945,
      "loss": 1.1386,
      "step": 8031
    },
    {
      "epoch": 1.2573575453976205,
      "grad_norm": 2.404191017150879,
      "learning_rate": 0.00011620234604105571,
      "loss": 0.9114,
      "step": 8032
    },
    {
      "epoch": 1.257514088916719,
      "grad_norm": 2.5697121620178223,
      "learning_rate": 0.00011617790811339199,
      "loss": 0.5878,
      "step": 8033
    },
    {
      "epoch": 1.2576706324358171,
      "grad_norm": 2.2942793369293213,
      "learning_rate": 0.00011615347018572824,
      "loss": 0.6517,
      "step": 8034
    },
    {
      "epoch": 1.2578271759549153,
      "grad_norm": 1.6695375442504883,
      "learning_rate": 0.0001161290322580645,
      "loss": 0.5267,
      "step": 8035
    },
    {
      "epoch": 1.2579837194740138,
      "grad_norm": 1.7396913766860962,
      "learning_rate": 0.00011610459433040078,
      "loss": 0.5404,
      "step": 8036
    },
    {
      "epoch": 1.2581402629931122,
      "grad_norm": 2.4069020748138428,
      "learning_rate": 0.00011608015640273703,
      "loss": 0.7826,
      "step": 8037
    },
    {
      "epoch": 1.2582968065122104,
      "grad_norm": 2.687602996826172,
      "learning_rate": 0.0001160557184750733,
      "loss": 0.7537,
      "step": 8038
    },
    {
      "epoch": 1.2584533500313086,
      "grad_norm": 0.9946150779724121,
      "learning_rate": 0.00011603128054740958,
      "loss": 0.3249,
      "step": 8039
    },
    {
      "epoch": 1.258609893550407,
      "grad_norm": 0.7107114791870117,
      "learning_rate": 0.00011600684261974584,
      "loss": 0.364,
      "step": 8040
    },
    {
      "epoch": 1.2587664370695053,
      "grad_norm": 1.1093772649765015,
      "learning_rate": 0.0001159824046920821,
      "loss": 0.247,
      "step": 8041
    },
    {
      "epoch": 1.2589229805886037,
      "grad_norm": 0.6720324754714966,
      "learning_rate": 0.00011595796676441837,
      "loss": 0.209,
      "step": 8042
    },
    {
      "epoch": 1.259079524107702,
      "grad_norm": 0.6645235419273376,
      "learning_rate": 0.00011593352883675464,
      "loss": 0.2063,
      "step": 8043
    },
    {
      "epoch": 1.2592360676268002,
      "grad_norm": 0.49136972427368164,
      "learning_rate": 0.0001159090909090909,
      "loss": 0.2311,
      "step": 8044
    },
    {
      "epoch": 1.2593926111458986,
      "grad_norm": 1.026901364326477,
      "learning_rate": 0.00011588465298142717,
      "loss": 0.2226,
      "step": 8045
    },
    {
      "epoch": 1.2595491546649968,
      "grad_norm": 0.6816409826278687,
      "learning_rate": 0.00011586021505376343,
      "loss": 0.2241,
      "step": 8046
    },
    {
      "epoch": 1.2597056981840953,
      "grad_norm": 0.4918442666530609,
      "learning_rate": 0.0001158357771260997,
      "loss": 0.2395,
      "step": 8047
    },
    {
      "epoch": 1.2598622417031935,
      "grad_norm": 0.6494612097740173,
      "learning_rate": 0.00011581133919843597,
      "loss": 0.2683,
      "step": 8048
    },
    {
      "epoch": 1.2600187852222917,
      "grad_norm": 0.875194787979126,
      "learning_rate": 0.00011578690127077223,
      "loss": 0.363,
      "step": 8049
    },
    {
      "epoch": 1.2601753287413902,
      "grad_norm": 0.8091587424278259,
      "learning_rate": 0.00011576246334310849,
      "loss": 0.2729,
      "step": 8050
    },
    {
      "epoch": 1.2603318722604884,
      "grad_norm": 0.5551338791847229,
      "learning_rate": 0.00011573802541544477,
      "loss": 0.2604,
      "step": 8051
    },
    {
      "epoch": 1.2604884157795868,
      "grad_norm": 0.5007133483886719,
      "learning_rate": 0.00011571358748778103,
      "loss": 0.2483,
      "step": 8052
    },
    {
      "epoch": 1.260644959298685,
      "grad_norm": 0.773539662361145,
      "learning_rate": 0.00011568914956011728,
      "loss": 0.175,
      "step": 8053
    },
    {
      "epoch": 1.2608015028177832,
      "grad_norm": 1.5341578722000122,
      "learning_rate": 0.00011566471163245356,
      "loss": 0.6749,
      "step": 8054
    },
    {
      "epoch": 1.2609580463368817,
      "grad_norm": 0.6845079660415649,
      "learning_rate": 0.00011564027370478983,
      "loss": 0.2875,
      "step": 8055
    },
    {
      "epoch": 1.26111458985598,
      "grad_norm": 1.1050750017166138,
      "learning_rate": 0.00011561583577712609,
      "loss": 0.2767,
      "step": 8056
    },
    {
      "epoch": 1.2612711333750783,
      "grad_norm": 0.9473204612731934,
      "learning_rate": 0.00011559139784946234,
      "loss": 0.3132,
      "step": 8057
    },
    {
      "epoch": 1.2614276768941766,
      "grad_norm": 2.2619035243988037,
      "learning_rate": 0.00011556695992179862,
      "loss": 0.4922,
      "step": 8058
    },
    {
      "epoch": 1.2615842204132748,
      "grad_norm": 2.7712318897247314,
      "learning_rate": 0.00011554252199413489,
      "loss": 0.9514,
      "step": 8059
    },
    {
      "epoch": 1.2617407639323732,
      "grad_norm": 1.5762767791748047,
      "learning_rate": 0.00011551808406647114,
      "loss": 0.6567,
      "step": 8060
    },
    {
      "epoch": 1.2618973074514714,
      "grad_norm": 0.8980954885482788,
      "learning_rate": 0.00011549364613880742,
      "loss": 0.3136,
      "step": 8061
    },
    {
      "epoch": 1.2620538509705699,
      "grad_norm": 5.263506889343262,
      "learning_rate": 0.00011546920821114368,
      "loss": 0.8192,
      "step": 8062
    },
    {
      "epoch": 1.262210394489668,
      "grad_norm": 1.3415788412094116,
      "learning_rate": 0.00011544477028347995,
      "loss": 0.3902,
      "step": 8063
    },
    {
      "epoch": 1.2623669380087663,
      "grad_norm": 1.4701817035675049,
      "learning_rate": 0.00011542033235581622,
      "loss": 0.7186,
      "step": 8064
    },
    {
      "epoch": 1.2625234815278648,
      "grad_norm": 1.1412750482559204,
      "learning_rate": 0.00011539589442815248,
      "loss": 0.5271,
      "step": 8065
    },
    {
      "epoch": 1.262680025046963,
      "grad_norm": 1.3679181337356567,
      "learning_rate": 0.00011537145650048874,
      "loss": 0.7291,
      "step": 8066
    },
    {
      "epoch": 1.2628365685660614,
      "grad_norm": 2.3653368949890137,
      "learning_rate": 0.00011534701857282502,
      "loss": 0.9745,
      "step": 8067
    },
    {
      "epoch": 1.2629931120851596,
      "grad_norm": 3.5362980365753174,
      "learning_rate": 0.00011532258064516128,
      "loss": 0.8932,
      "step": 8068
    },
    {
      "epoch": 1.2631496556042578,
      "grad_norm": 1.438018560409546,
      "learning_rate": 0.00011529814271749753,
      "loss": 0.5735,
      "step": 8069
    },
    {
      "epoch": 1.2633061991233563,
      "grad_norm": 1.334162950515747,
      "learning_rate": 0.00011527370478983381,
      "loss": 0.4908,
      "step": 8070
    },
    {
      "epoch": 1.2634627426424547,
      "grad_norm": 1.1316155195236206,
      "learning_rate": 0.00011524926686217008,
      "loss": 0.3825,
      "step": 8071
    },
    {
      "epoch": 1.263619286161553,
      "grad_norm": 3.4603209495544434,
      "learning_rate": 0.00011522482893450633,
      "loss": 1.0631,
      "step": 8072
    },
    {
      "epoch": 1.2637758296806512,
      "grad_norm": 3.4349160194396973,
      "learning_rate": 0.00011520039100684261,
      "loss": 0.6785,
      "step": 8073
    },
    {
      "epoch": 1.2639323731997496,
      "grad_norm": 2.098545789718628,
      "learning_rate": 0.00011517595307917887,
      "loss": 0.6231,
      "step": 8074
    },
    {
      "epoch": 1.2640889167188478,
      "grad_norm": 1.599894642829895,
      "learning_rate": 0.00011515151515151514,
      "loss": 0.4584,
      "step": 8075
    },
    {
      "epoch": 1.2642454602379463,
      "grad_norm": 2.5117294788360596,
      "learning_rate": 0.00011512707722385142,
      "loss": 0.7141,
      "step": 8076
    },
    {
      "epoch": 1.2644020037570445,
      "grad_norm": 3.738943576812744,
      "learning_rate": 0.00011510263929618767,
      "loss": 1.4734,
      "step": 8077
    },
    {
      "epoch": 1.2645585472761427,
      "grad_norm": 2.5964126586914062,
      "learning_rate": 0.00011507820136852393,
      "loss": 0.6161,
      "step": 8078
    },
    {
      "epoch": 1.2647150907952411,
      "grad_norm": 3.135176420211792,
      "learning_rate": 0.00011505376344086021,
      "loss": 1.5048,
      "step": 8079
    },
    {
      "epoch": 1.2648716343143394,
      "grad_norm": 2.939114570617676,
      "learning_rate": 0.00011502932551319647,
      "loss": 1.0791,
      "step": 8080
    },
    {
      "epoch": 1.2650281778334378,
      "grad_norm": 2.810537815093994,
      "learning_rate": 0.00011500488758553273,
      "loss": 2.0609,
      "step": 8081
    },
    {
      "epoch": 1.265184721352536,
      "grad_norm": 3.116837978363037,
      "learning_rate": 0.000114980449657869,
      "loss": 0.956,
      "step": 8082
    },
    {
      "epoch": 1.2653412648716342,
      "grad_norm": 2.0274899005889893,
      "learning_rate": 0.00011495601173020527,
      "loss": 1.196,
      "step": 8083
    },
    {
      "epoch": 1.2654978083907327,
      "grad_norm": 2.4018070697784424,
      "learning_rate": 0.00011493157380254152,
      "loss": 1.0615,
      "step": 8084
    },
    {
      "epoch": 1.2656543519098309,
      "grad_norm": 3.216063976287842,
      "learning_rate": 0.0001149071358748778,
      "loss": 0.4063,
      "step": 8085
    },
    {
      "epoch": 1.2658108954289293,
      "grad_norm": 1.749650001525879,
      "learning_rate": 0.00011488269794721406,
      "loss": 0.4202,
      "step": 8086
    },
    {
      "epoch": 1.2659674389480275,
      "grad_norm": 3.246669292449951,
      "learning_rate": 0.00011485826001955033,
      "loss": 0.8565,
      "step": 8087
    },
    {
      "epoch": 1.2661239824671258,
      "grad_norm": 1.4922568798065186,
      "learning_rate": 0.00011483382209188661,
      "loss": 0.634,
      "step": 8088
    },
    {
      "epoch": 1.2662805259862242,
      "grad_norm": 0.5845685601234436,
      "learning_rate": 0.00011480938416422286,
      "loss": 0.2766,
      "step": 8089
    },
    {
      "epoch": 1.2664370695053224,
      "grad_norm": 0.3810208737850189,
      "learning_rate": 0.00011478494623655912,
      "loss": 0.2173,
      "step": 8090
    },
    {
      "epoch": 1.2665936130244209,
      "grad_norm": 0.5567788481712341,
      "learning_rate": 0.0001147605083088954,
      "loss": 0.2469,
      "step": 8091
    },
    {
      "epoch": 1.266750156543519,
      "grad_norm": 0.5596190690994263,
      "learning_rate": 0.00011473607038123167,
      "loss": 0.2548,
      "step": 8092
    },
    {
      "epoch": 1.2669067000626173,
      "grad_norm": 0.5146242380142212,
      "learning_rate": 0.00011471163245356792,
      "loss": 0.2395,
      "step": 8093
    },
    {
      "epoch": 1.2670632435817157,
      "grad_norm": 0.6236430406570435,
      "learning_rate": 0.0001146871945259042,
      "loss": 0.3531,
      "step": 8094
    },
    {
      "epoch": 1.267219787100814,
      "grad_norm": 0.6425842046737671,
      "learning_rate": 0.00011466275659824046,
      "loss": 0.3415,
      "step": 8095
    },
    {
      "epoch": 1.2673763306199124,
      "grad_norm": 0.39682844281196594,
      "learning_rate": 0.00011463831867057671,
      "loss": 0.2194,
      "step": 8096
    },
    {
      "epoch": 1.2675328741390106,
      "grad_norm": 2.4693057537078857,
      "learning_rate": 0.00011461388074291299,
      "loss": 0.3667,
      "step": 8097
    },
    {
      "epoch": 1.2676894176581088,
      "grad_norm": 0.5568252801895142,
      "learning_rate": 0.00011458944281524925,
      "loss": 0.3198,
      "step": 8098
    },
    {
      "epoch": 1.2678459611772073,
      "grad_norm": 0.7685470581054688,
      "learning_rate": 0.00011456500488758552,
      "loss": 0.3658,
      "step": 8099
    },
    {
      "epoch": 1.2680025046963057,
      "grad_norm": 1.0343455076217651,
      "learning_rate": 0.0001145405669599218,
      "loss": 0.2678,
      "step": 8100
    },
    {
      "epoch": 1.268159048215404,
      "grad_norm": 2.833596706390381,
      "learning_rate": 0.00011451612903225805,
      "loss": 0.6284,
      "step": 8101
    },
    {
      "epoch": 1.2683155917345021,
      "grad_norm": 0.7539536952972412,
      "learning_rate": 0.00011449169110459431,
      "loss": 0.2395,
      "step": 8102
    },
    {
      "epoch": 1.2684721352536004,
      "grad_norm": 1.2098023891448975,
      "learning_rate": 0.00011446725317693059,
      "loss": 0.2926,
      "step": 8103
    },
    {
      "epoch": 1.2686286787726988,
      "grad_norm": 1.0733698606491089,
      "learning_rate": 0.00011444281524926686,
      "loss": 0.4996,
      "step": 8104
    },
    {
      "epoch": 1.2687852222917972,
      "grad_norm": 1.2377598285675049,
      "learning_rate": 0.00011441837732160311,
      "loss": 0.5157,
      "step": 8105
    },
    {
      "epoch": 1.2689417658108955,
      "grad_norm": 0.8723770976066589,
      "learning_rate": 0.00011439393939393939,
      "loss": 0.3019,
      "step": 8106
    },
    {
      "epoch": 1.2690983093299937,
      "grad_norm": 1.6629414558410645,
      "learning_rate": 0.00011436950146627565,
      "loss": 0.5106,
      "step": 8107
    },
    {
      "epoch": 1.2692548528490921,
      "grad_norm": 1.0047154426574707,
      "learning_rate": 0.0001143450635386119,
      "loss": 0.4231,
      "step": 8108
    },
    {
      "epoch": 1.2694113963681903,
      "grad_norm": 1.7095948457717896,
      "learning_rate": 0.00011432062561094818,
      "loss": 0.6328,
      "step": 8109
    },
    {
      "epoch": 1.2695679398872888,
      "grad_norm": 1.4480046033859253,
      "learning_rate": 0.00011429618768328445,
      "loss": 0.535,
      "step": 8110
    },
    {
      "epoch": 1.269724483406387,
      "grad_norm": 0.7682520151138306,
      "learning_rate": 0.00011427174975562071,
      "loss": 0.2345,
      "step": 8111
    },
    {
      "epoch": 1.2698810269254852,
      "grad_norm": 1.8508092164993286,
      "learning_rate": 0.00011424731182795699,
      "loss": 0.7018,
      "step": 8112
    },
    {
      "epoch": 1.2700375704445837,
      "grad_norm": 1.5471041202545166,
      "learning_rate": 0.00011422287390029324,
      "loss": 0.5689,
      "step": 8113
    },
    {
      "epoch": 1.2701941139636819,
      "grad_norm": 1.868023157119751,
      "learning_rate": 0.0001141984359726295,
      "loss": 0.5738,
      "step": 8114
    },
    {
      "epoch": 1.2703506574827803,
      "grad_norm": 5.055905818939209,
      "learning_rate": 0.00011417399804496578,
      "loss": 0.5171,
      "step": 8115
    },
    {
      "epoch": 1.2705072010018785,
      "grad_norm": 1.255279779434204,
      "learning_rate": 0.00011414956011730203,
      "loss": 0.6595,
      "step": 8116
    },
    {
      "epoch": 1.2706637445209767,
      "grad_norm": 2.3022544384002686,
      "learning_rate": 0.0001141251221896383,
      "loss": 0.5065,
      "step": 8117
    },
    {
      "epoch": 1.2708202880400752,
      "grad_norm": 2.671542167663574,
      "learning_rate": 0.00011410068426197458,
      "loss": 1.1003,
      "step": 8118
    },
    {
      "epoch": 1.2709768315591734,
      "grad_norm": 2.1620357036590576,
      "learning_rate": 0.00011407624633431084,
      "loss": 0.6959,
      "step": 8119
    },
    {
      "epoch": 1.2711333750782718,
      "grad_norm": 2.031935214996338,
      "learning_rate": 0.0001140518084066471,
      "loss": 0.9013,
      "step": 8120
    },
    {
      "epoch": 1.27128991859737,
      "grad_norm": 2.194514751434326,
      "learning_rate": 0.00011402737047898337,
      "loss": 0.5662,
      "step": 8121
    },
    {
      "epoch": 1.2714464621164683,
      "grad_norm": 1.3475483655929565,
      "learning_rate": 0.00011400293255131964,
      "loss": 0.5597,
      "step": 8122
    },
    {
      "epoch": 1.2716030056355667,
      "grad_norm": 4.436418533325195,
      "learning_rate": 0.0001139784946236559,
      "loss": 0.9418,
      "step": 8123
    },
    {
      "epoch": 1.271759549154665,
      "grad_norm": 5.266567230224609,
      "learning_rate": 0.00011395405669599218,
      "loss": 1.1798,
      "step": 8124
    },
    {
      "epoch": 1.2719160926737634,
      "grad_norm": 3.8170948028564453,
      "learning_rate": 0.00011392961876832843,
      "loss": 0.9634,
      "step": 8125
    },
    {
      "epoch": 1.2720726361928616,
      "grad_norm": 2.858017921447754,
      "learning_rate": 0.0001139051808406647,
      "loss": 1.2201,
      "step": 8126
    },
    {
      "epoch": 1.2722291797119598,
      "grad_norm": 3.335280418395996,
      "learning_rate": 0.00011388074291300097,
      "loss": 0.9677,
      "step": 8127
    },
    {
      "epoch": 1.2723857232310583,
      "grad_norm": 2.3671767711639404,
      "learning_rate": 0.00011385630498533723,
      "loss": 1.3408,
      "step": 8128
    },
    {
      "epoch": 1.2725422667501565,
      "grad_norm": 1.9784801006317139,
      "learning_rate": 0.00011383186705767349,
      "loss": 0.6781,
      "step": 8129
    },
    {
      "epoch": 1.272698810269255,
      "grad_norm": 2.1394803524017334,
      "learning_rate": 0.00011380742913000977,
      "loss": 1.1412,
      "step": 8130
    },
    {
      "epoch": 1.2728553537883531,
      "grad_norm": 4.030261039733887,
      "learning_rate": 0.00011378299120234603,
      "loss": 1.4167,
      "step": 8131
    },
    {
      "epoch": 1.2730118973074513,
      "grad_norm": 2.028381824493408,
      "learning_rate": 0.00011375855327468229,
      "loss": 0.9049,
      "step": 8132
    },
    {
      "epoch": 1.2731684408265498,
      "grad_norm": 3.869459629058838,
      "learning_rate": 0.00011373411534701856,
      "loss": 1.1694,
      "step": 8133
    },
    {
      "epoch": 1.2733249843456482,
      "grad_norm": 2.090665340423584,
      "learning_rate": 0.00011370967741935483,
      "loss": 0.9433,
      "step": 8134
    },
    {
      "epoch": 1.2734815278647464,
      "grad_norm": 3.8290891647338867,
      "learning_rate": 0.00011368523949169109,
      "loss": 1.0372,
      "step": 8135
    },
    {
      "epoch": 1.2736380713838447,
      "grad_norm": 1.362321138381958,
      "learning_rate": 0.00011366080156402737,
      "loss": 0.7292,
      "step": 8136
    },
    {
      "epoch": 1.273794614902943,
      "grad_norm": 2.492807149887085,
      "learning_rate": 0.00011363636363636362,
      "loss": 1.0427,
      "step": 8137
    },
    {
      "epoch": 1.2739511584220413,
      "grad_norm": 2.186556339263916,
      "learning_rate": 0.00011361192570869989,
      "loss": 0.6872,
      "step": 8138
    },
    {
      "epoch": 1.2741077019411398,
      "grad_norm": 0.4808526933193207,
      "learning_rate": 0.00011358748778103617,
      "loss": 0.2192,
      "step": 8139
    },
    {
      "epoch": 1.274264245460238,
      "grad_norm": 0.7778594493865967,
      "learning_rate": 0.00011356304985337242,
      "loss": 0.4221,
      "step": 8140
    },
    {
      "epoch": 1.2744207889793362,
      "grad_norm": 0.5027886629104614,
      "learning_rate": 0.00011353861192570868,
      "loss": 0.2751,
      "step": 8141
    },
    {
      "epoch": 1.2745773324984346,
      "grad_norm": 0.5371960997581482,
      "learning_rate": 0.00011351417399804496,
      "loss": 0.2394,
      "step": 8142
    },
    {
      "epoch": 1.2747338760175329,
      "grad_norm": 0.42323970794677734,
      "learning_rate": 0.00011348973607038123,
      "loss": 0.2188,
      "step": 8143
    },
    {
      "epoch": 1.2748904195366313,
      "grad_norm": 0.8892855644226074,
      "learning_rate": 0.00011346529814271748,
      "loss": 0.2642,
      "step": 8144
    },
    {
      "epoch": 1.2750469630557295,
      "grad_norm": 0.6918807625770569,
      "learning_rate": 0.00011344086021505375,
      "loss": 0.204,
      "step": 8145
    },
    {
      "epoch": 1.2752035065748277,
      "grad_norm": 0.6264479756355286,
      "learning_rate": 0.00011341642228739002,
      "loss": 0.2568,
      "step": 8146
    },
    {
      "epoch": 1.2753600500939262,
      "grad_norm": 0.6950996518135071,
      "learning_rate": 0.00011339198435972628,
      "loss": 0.244,
      "step": 8147
    },
    {
      "epoch": 1.2755165936130244,
      "grad_norm": 0.6539076566696167,
      "learning_rate": 0.00011336754643206256,
      "loss": 0.2628,
      "step": 8148
    },
    {
      "epoch": 1.2756731371321228,
      "grad_norm": 0.7629345655441284,
      "learning_rate": 0.00011334310850439881,
      "loss": 0.2214,
      "step": 8149
    },
    {
      "epoch": 1.275829680651221,
      "grad_norm": 0.7511631846427917,
      "learning_rate": 0.00011331867057673508,
      "loss": 0.2655,
      "step": 8150
    },
    {
      "epoch": 1.2759862241703193,
      "grad_norm": 3.1061666011810303,
      "learning_rate": 0.00011329423264907136,
      "loss": 0.3632,
      "step": 8151
    },
    {
      "epoch": 1.2761427676894177,
      "grad_norm": 0.8534873127937317,
      "learning_rate": 0.00011326979472140761,
      "loss": 0.2821,
      "step": 8152
    },
    {
      "epoch": 1.276299311208516,
      "grad_norm": 0.879711389541626,
      "learning_rate": 0.00011324535679374387,
      "loss": 0.2113,
      "step": 8153
    },
    {
      "epoch": 1.2764558547276144,
      "grad_norm": 1.0535318851470947,
      "learning_rate": 0.00011322091886608015,
      "loss": 0.471,
      "step": 8154
    },
    {
      "epoch": 1.2766123982467126,
      "grad_norm": 3.926427125930786,
      "learning_rate": 0.00011319648093841642,
      "loss": 0.7819,
      "step": 8155
    },
    {
      "epoch": 1.2767689417658108,
      "grad_norm": 0.8711987137794495,
      "learning_rate": 0.00011317204301075267,
      "loss": 0.2567,
      "step": 8156
    },
    {
      "epoch": 1.2769254852849092,
      "grad_norm": 1.3189749717712402,
      "learning_rate": 0.00011314760508308895,
      "loss": 0.5032,
      "step": 8157
    },
    {
      "epoch": 1.2770820288040075,
      "grad_norm": 2.5916378498077393,
      "learning_rate": 0.00011312316715542521,
      "loss": 0.4017,
      "step": 8158
    },
    {
      "epoch": 1.277238572323106,
      "grad_norm": 2.361572504043579,
      "learning_rate": 0.00011309872922776148,
      "loss": 0.5681,
      "step": 8159
    },
    {
      "epoch": 1.277395115842204,
      "grad_norm": 3.2034995555877686,
      "learning_rate": 0.00011307429130009774,
      "loss": 0.8003,
      "step": 8160
    },
    {
      "epoch": 1.2775516593613023,
      "grad_norm": 0.9315476417541504,
      "learning_rate": 0.000113049853372434,
      "loss": 0.3008,
      "step": 8161
    },
    {
      "epoch": 1.2777082028804008,
      "grad_norm": 1.9534274339675903,
      "learning_rate": 0.00011302541544477027,
      "loss": 0.6728,
      "step": 8162
    },
    {
      "epoch": 1.277864746399499,
      "grad_norm": 1.1545753479003906,
      "learning_rate": 0.00011300097751710655,
      "loss": 0.4329,
      "step": 8163
    },
    {
      "epoch": 1.2780212899185974,
      "grad_norm": 3.3171403408050537,
      "learning_rate": 0.0001129765395894428,
      "loss": 0.7543,
      "step": 8164
    },
    {
      "epoch": 1.2781778334376956,
      "grad_norm": 1.9916654825210571,
      "learning_rate": 0.00011295210166177906,
      "loss": 0.7833,
      "step": 8165
    },
    {
      "epoch": 1.2783343769567939,
      "grad_norm": 1.3301039934158325,
      "learning_rate": 0.00011292766373411534,
      "loss": 0.5321,
      "step": 8166
    },
    {
      "epoch": 1.2784909204758923,
      "grad_norm": 1.8553197383880615,
      "learning_rate": 0.00011290322580645161,
      "loss": 0.7389,
      "step": 8167
    },
    {
      "epoch": 1.2786474639949907,
      "grad_norm": 3.052807092666626,
      "learning_rate": 0.00011287878787878786,
      "loss": 0.7611,
      "step": 8168
    },
    {
      "epoch": 1.278804007514089,
      "grad_norm": 3.283430814743042,
      "learning_rate": 0.00011285434995112414,
      "loss": 1.0834,
      "step": 8169
    },
    {
      "epoch": 1.2789605510331872,
      "grad_norm": 1.6933223009109497,
      "learning_rate": 0.0001128299120234604,
      "loss": 0.747,
      "step": 8170
    },
    {
      "epoch": 1.2791170945522856,
      "grad_norm": 2.0889194011688232,
      "learning_rate": 0.00011280547409579667,
      "loss": 0.7694,
      "step": 8171
    },
    {
      "epoch": 1.2792736380713838,
      "grad_norm": 1.8516852855682373,
      "learning_rate": 0.00011278103616813293,
      "loss": 0.7245,
      "step": 8172
    },
    {
      "epoch": 1.2794301815904823,
      "grad_norm": 2.9009687900543213,
      "learning_rate": 0.0001127565982404692,
      "loss": 1.0895,
      "step": 8173
    },
    {
      "epoch": 1.2795867251095805,
      "grad_norm": 2.16554856300354,
      "learning_rate": 0.00011273216031280546,
      "loss": 0.9004,
      "step": 8174
    },
    {
      "epoch": 1.2797432686286787,
      "grad_norm": 1.9977914094924927,
      "learning_rate": 0.00011270772238514174,
      "loss": 0.9605,
      "step": 8175
    },
    {
      "epoch": 1.2798998121477771,
      "grad_norm": 2.656644344329834,
      "learning_rate": 0.00011268328445747799,
      "loss": 0.9607,
      "step": 8176
    },
    {
      "epoch": 1.2800563556668754,
      "grad_norm": 2.0683369636535645,
      "learning_rate": 0.00011265884652981426,
      "loss": 0.866,
      "step": 8177
    },
    {
      "epoch": 1.2802128991859738,
      "grad_norm": 3.113068103790283,
      "learning_rate": 0.00011263440860215053,
      "loss": 1.2594,
      "step": 8178
    },
    {
      "epoch": 1.280369442705072,
      "grad_norm": 2.857494831085205,
      "learning_rate": 0.0001126099706744868,
      "loss": 1.3643,
      "step": 8179
    },
    {
      "epoch": 1.2805259862241702,
      "grad_norm": 2.787754535675049,
      "learning_rate": 0.00011258553274682305,
      "loss": 1.0681,
      "step": 8180
    },
    {
      "epoch": 1.2806825297432687,
      "grad_norm": 2.6079978942871094,
      "learning_rate": 0.00011256109481915933,
      "loss": 0.9953,
      "step": 8181
    },
    {
      "epoch": 1.280839073262367,
      "grad_norm": 1.9221597909927368,
      "learning_rate": 0.00011253665689149559,
      "loss": 0.7956,
      "step": 8182
    },
    {
      "epoch": 1.2809956167814653,
      "grad_norm": 3.6456334590911865,
      "learning_rate": 0.00011251221896383186,
      "loss": 0.6788,
      "step": 8183
    },
    {
      "epoch": 1.2811521603005636,
      "grad_norm": 4.128708839416504,
      "learning_rate": 0.00011248778103616812,
      "loss": 0.808,
      "step": 8184
    },
    {
      "epoch": 1.2813087038196618,
      "grad_norm": 1.7657461166381836,
      "learning_rate": 0.00011246334310850439,
      "loss": 0.5298,
      "step": 8185
    },
    {
      "epoch": 1.2814652473387602,
      "grad_norm": 2.3742294311523438,
      "learning_rate": 0.00011243890518084065,
      "loss": 0.5269,
      "step": 8186
    },
    {
      "epoch": 1.2816217908578584,
      "grad_norm": 2.828449249267578,
      "learning_rate": 0.00011241446725317693,
      "loss": 0.8815,
      "step": 8187
    },
    {
      "epoch": 1.2817783343769569,
      "grad_norm": 1.5230042934417725,
      "learning_rate": 0.00011239002932551318,
      "loss": 0.553,
      "step": 8188
    },
    {
      "epoch": 1.281934877896055,
      "grad_norm": 0.6481024622917175,
      "learning_rate": 0.00011236559139784945,
      "loss": 0.2289,
      "step": 8189
    },
    {
      "epoch": 1.2820914214151533,
      "grad_norm": 2.089646577835083,
      "learning_rate": 0.00011234115347018572,
      "loss": 0.8718,
      "step": 8190
    },
    {
      "epoch": 1.2822479649342517,
      "grad_norm": 0.42590638995170593,
      "learning_rate": 0.00011231671554252199,
      "loss": 0.1615,
      "step": 8191
    },
    {
      "epoch": 1.28240450845335,
      "grad_norm": 0.3601533770561218,
      "learning_rate": 0.00011229227761485824,
      "loss": 0.164,
      "step": 8192
    },
    {
      "epoch": 1.2825610519724484,
      "grad_norm": 0.6453453898429871,
      "learning_rate": 0.00011226783968719452,
      "loss": 0.3217,
      "step": 8193
    },
    {
      "epoch": 1.2827175954915466,
      "grad_norm": 0.7831723093986511,
      "learning_rate": 0.00011224340175953078,
      "loss": 0.2657,
      "step": 8194
    },
    {
      "epoch": 1.2828741390106448,
      "grad_norm": 0.6272794008255005,
      "learning_rate": 0.00011221896383186705,
      "loss": 0.278,
      "step": 8195
    },
    {
      "epoch": 1.2830306825297433,
      "grad_norm": 0.6463831067085266,
      "learning_rate": 0.00011219452590420331,
      "loss": 0.3111,
      "step": 8196
    },
    {
      "epoch": 1.2831872260488415,
      "grad_norm": 0.7817267775535583,
      "learning_rate": 0.00011217008797653958,
      "loss": 0.2155,
      "step": 8197
    },
    {
      "epoch": 1.28334376956794,
      "grad_norm": 0.6414055824279785,
      "learning_rate": 0.00011214565004887584,
      "loss": 0.2571,
      "step": 8198
    },
    {
      "epoch": 1.2835003130870382,
      "grad_norm": 1.242760181427002,
      "learning_rate": 0.00011212121212121212,
      "loss": 0.2441,
      "step": 8199
    },
    {
      "epoch": 1.2836568566061364,
      "grad_norm": 0.43037867546081543,
      "learning_rate": 0.00011209677419354837,
      "loss": 0.206,
      "step": 8200
    },
    {
      "epoch": 1.2838134001252348,
      "grad_norm": 0.5707399845123291,
      "learning_rate": 0.00011207233626588464,
      "loss": 0.2067,
      "step": 8201
    },
    {
      "epoch": 1.2839699436443333,
      "grad_norm": 1.2979316711425781,
      "learning_rate": 0.00011204789833822092,
      "loss": 0.3798,
      "step": 8202
    },
    {
      "epoch": 1.2841264871634315,
      "grad_norm": 0.5706567764282227,
      "learning_rate": 0.00011202346041055718,
      "loss": 0.162,
      "step": 8203
    },
    {
      "epoch": 1.2842830306825297,
      "grad_norm": 1.7582261562347412,
      "learning_rate": 0.00011199902248289343,
      "loss": 0.5603,
      "step": 8204
    },
    {
      "epoch": 1.2844395742016281,
      "grad_norm": 2.591653823852539,
      "learning_rate": 0.00011197458455522971,
      "loss": 0.4609,
      "step": 8205
    },
    {
      "epoch": 1.2845961177207263,
      "grad_norm": 0.9483798146247864,
      "learning_rate": 0.00011195014662756598,
      "loss": 0.3353,
      "step": 8206
    },
    {
      "epoch": 1.2847526612398248,
      "grad_norm": 1.1728459596633911,
      "learning_rate": 0.00011192570869990224,
      "loss": 0.3398,
      "step": 8207
    },
    {
      "epoch": 1.284909204758923,
      "grad_norm": 0.952708899974823,
      "learning_rate": 0.0001119012707722385,
      "loss": 0.3597,
      "step": 8208
    },
    {
      "epoch": 1.2850657482780212,
      "grad_norm": 1.8662762641906738,
      "learning_rate": 0.00011187683284457477,
      "loss": 0.6664,
      "step": 8209
    },
    {
      "epoch": 1.2852222917971197,
      "grad_norm": 1.1493111848831177,
      "learning_rate": 0.00011185239491691103,
      "loss": 0.8306,
      "step": 8210
    },
    {
      "epoch": 1.2853788353162179,
      "grad_norm": 1.7580465078353882,
      "learning_rate": 0.00011182795698924731,
      "loss": 0.4859,
      "step": 8211
    },
    {
      "epoch": 1.2855353788353163,
      "grad_norm": 0.9127596020698547,
      "learning_rate": 0.00011180351906158356,
      "loss": 0.2457,
      "step": 8212
    },
    {
      "epoch": 1.2856919223544145,
      "grad_norm": 1.9063149690628052,
      "learning_rate": 0.00011177908113391983,
      "loss": 0.4663,
      "step": 8213
    },
    {
      "epoch": 1.2858484658735128,
      "grad_norm": 1.1802544593811035,
      "learning_rate": 0.00011175464320625611,
      "loss": 0.3457,
      "step": 8214
    },
    {
      "epoch": 1.2860050093926112,
      "grad_norm": 2.804399013519287,
      "learning_rate": 0.00011173020527859237,
      "loss": 0.6875,
      "step": 8215
    },
    {
      "epoch": 1.2861615529117094,
      "grad_norm": 1.3575066328048706,
      "learning_rate": 0.00011170576735092862,
      "loss": 0.5672,
      "step": 8216
    },
    {
      "epoch": 1.2863180964308079,
      "grad_norm": 1.3274577856063843,
      "learning_rate": 0.0001116813294232649,
      "loss": 0.731,
      "step": 8217
    },
    {
      "epoch": 1.286474639949906,
      "grad_norm": 1.7587782144546509,
      "learning_rate": 0.00011165689149560117,
      "loss": 0.6624,
      "step": 8218
    },
    {
      "epoch": 1.2866311834690043,
      "grad_norm": 1.3552849292755127,
      "learning_rate": 0.00011163245356793742,
      "loss": 0.5267,
      "step": 8219
    },
    {
      "epoch": 1.2867877269881027,
      "grad_norm": 1.7056719064712524,
      "learning_rate": 0.0001116080156402737,
      "loss": 0.8283,
      "step": 8220
    },
    {
      "epoch": 1.286944270507201,
      "grad_norm": 2.418220281600952,
      "learning_rate": 0.00011158357771260996,
      "loss": 0.7403,
      "step": 8221
    },
    {
      "epoch": 1.2871008140262994,
      "grad_norm": 1.3776272535324097,
      "learning_rate": 0.00011155913978494623,
      "loss": 0.4315,
      "step": 8222
    },
    {
      "epoch": 1.2872573575453976,
      "grad_norm": 1.3834933042526245,
      "learning_rate": 0.0001115347018572825,
      "loss": 0.4908,
      "step": 8223
    },
    {
      "epoch": 1.2874139010644958,
      "grad_norm": 1.5205992460250854,
      "learning_rate": 0.00011151026392961876,
      "loss": 0.8942,
      "step": 8224
    },
    {
      "epoch": 1.2875704445835943,
      "grad_norm": 3.1065549850463867,
      "learning_rate": 0.00011148582600195502,
      "loss": 1.2169,
      "step": 8225
    },
    {
      "epoch": 1.2877269881026925,
      "grad_norm": 3.65362811088562,
      "learning_rate": 0.0001114613880742913,
      "loss": 1.0125,
      "step": 8226
    },
    {
      "epoch": 1.287883531621791,
      "grad_norm": 1.9371777772903442,
      "learning_rate": 0.00011143695014662756,
      "loss": 1.1199,
      "step": 8227
    },
    {
      "epoch": 1.2880400751408891,
      "grad_norm": 3.518394947052002,
      "learning_rate": 0.00011141251221896381,
      "loss": 0.8579,
      "step": 8228
    },
    {
      "epoch": 1.2881966186599874,
      "grad_norm": 2.363788604736328,
      "learning_rate": 0.00011138807429130009,
      "loss": 0.6714,
      "step": 8229
    },
    {
      "epoch": 1.2883531621790858,
      "grad_norm": 2.800300359725952,
      "learning_rate": 0.00011136363636363636,
      "loss": 0.9438,
      "step": 8230
    },
    {
      "epoch": 1.288509705698184,
      "grad_norm": 2.6638309955596924,
      "learning_rate": 0.00011133919843597261,
      "loss": 1.5732,
      "step": 8231
    },
    {
      "epoch": 1.2886662492172825,
      "grad_norm": 1.8096959590911865,
      "learning_rate": 0.00011131476050830889,
      "loss": 0.8746,
      "step": 8232
    },
    {
      "epoch": 1.2888227927363807,
      "grad_norm": 2.5477139949798584,
      "learning_rate": 0.00011129032258064515,
      "loss": 0.7608,
      "step": 8233
    },
    {
      "epoch": 1.288979336255479,
      "grad_norm": 5.085765361785889,
      "learning_rate": 0.00011126588465298142,
      "loss": 0.746,
      "step": 8234
    },
    {
      "epoch": 1.2891358797745773,
      "grad_norm": 2.019724130630493,
      "learning_rate": 0.0001112414467253177,
      "loss": 0.3566,
      "step": 8235
    },
    {
      "epoch": 1.2892924232936758,
      "grad_norm": 6.809519290924072,
      "learning_rate": 0.00011121700879765395,
      "loss": 0.8774,
      "step": 8236
    },
    {
      "epoch": 1.289448966812774,
      "grad_norm": 4.453648090362549,
      "learning_rate": 0.00011119257086999021,
      "loss": 0.3884,
      "step": 8237
    },
    {
      "epoch": 1.2896055103318722,
      "grad_norm": 2.842747449874878,
      "learning_rate": 0.00011116813294232649,
      "loss": 1.12,
      "step": 8238
    },
    {
      "epoch": 1.2897620538509706,
      "grad_norm": 0.7022001147270203,
      "learning_rate": 0.00011114369501466275,
      "loss": 0.2758,
      "step": 8239
    },
    {
      "epoch": 1.2899185973700689,
      "grad_norm": 0.49633824825286865,
      "learning_rate": 0.000111119257086999,
      "loss": 0.2534,
      "step": 8240
    },
    {
      "epoch": 1.2900751408891673,
      "grad_norm": 0.6422197818756104,
      "learning_rate": 0.00011109481915933528,
      "loss": 0.2632,
      "step": 8241
    },
    {
      "epoch": 1.2902316844082655,
      "grad_norm": 1.4231148958206177,
      "learning_rate": 0.00011107038123167155,
      "loss": 0.4564,
      "step": 8242
    },
    {
      "epoch": 1.2903882279273637,
      "grad_norm": 0.8207333087921143,
      "learning_rate": 0.0001110459433040078,
      "loss": 0.2579,
      "step": 8243
    },
    {
      "epoch": 1.2905447714464622,
      "grad_norm": 0.8601287603378296,
      "learning_rate": 0.00011102150537634408,
      "loss": 0.3219,
      "step": 8244
    },
    {
      "epoch": 1.2907013149655604,
      "grad_norm": 0.8240009546279907,
      "learning_rate": 0.00011099706744868034,
      "loss": 0.3404,
      "step": 8245
    },
    {
      "epoch": 1.2908578584846588,
      "grad_norm": 0.8633431792259216,
      "learning_rate": 0.00011097262952101661,
      "loss": 0.2921,
      "step": 8246
    },
    {
      "epoch": 1.291014402003757,
      "grad_norm": 1.7753422260284424,
      "learning_rate": 0.00011094819159335289,
      "loss": 0.4439,
      "step": 8247
    },
    {
      "epoch": 1.2911709455228553,
      "grad_norm": 1.0025125741958618,
      "learning_rate": 0.00011092375366568914,
      "loss": 0.317,
      "step": 8248
    },
    {
      "epoch": 1.2913274890419537,
      "grad_norm": 1.528899073600769,
      "learning_rate": 0.0001108993157380254,
      "loss": 0.3887,
      "step": 8249
    },
    {
      "epoch": 1.291484032561052,
      "grad_norm": 0.7538073062896729,
      "learning_rate": 0.00011087487781036168,
      "loss": 0.3759,
      "step": 8250
    },
    {
      "epoch": 1.2916405760801504,
      "grad_norm": 0.7805300951004028,
      "learning_rate": 0.00011085043988269795,
      "loss": 0.2921,
      "step": 8251
    },
    {
      "epoch": 1.2917971195992486,
      "grad_norm": 1.1637154817581177,
      "learning_rate": 0.0001108260019550342,
      "loss": 0.4664,
      "step": 8252
    },
    {
      "epoch": 1.2919536631183468,
      "grad_norm": 0.8898148536682129,
      "learning_rate": 0.00011080156402737048,
      "loss": 0.3022,
      "step": 8253
    },
    {
      "epoch": 1.2921102066374452,
      "grad_norm": 1.7075831890106201,
      "learning_rate": 0.00011077712609970674,
      "loss": 0.4828,
      "step": 8254
    },
    {
      "epoch": 1.2922667501565435,
      "grad_norm": 1.2227857112884521,
      "learning_rate": 0.00011075268817204299,
      "loss": 0.3136,
      "step": 8255
    },
    {
      "epoch": 1.292423293675642,
      "grad_norm": 1.10588800907135,
      "learning_rate": 0.00011072825024437927,
      "loss": 0.209,
      "step": 8256
    },
    {
      "epoch": 1.2925798371947401,
      "grad_norm": 1.42483389377594,
      "learning_rate": 0.00011070381231671553,
      "loss": 0.6039,
      "step": 8257
    },
    {
      "epoch": 1.2927363807138383,
      "grad_norm": 4.606034755706787,
      "learning_rate": 0.0001106793743890518,
      "loss": 1.3905,
      "step": 8258
    },
    {
      "epoch": 1.2928929242329368,
      "grad_norm": 1.765781283378601,
      "learning_rate": 0.00011065493646138808,
      "loss": 0.5845,
      "step": 8259
    },
    {
      "epoch": 1.293049467752035,
      "grad_norm": 1.1531269550323486,
      "learning_rate": 0.00011063049853372433,
      "loss": 0.339,
      "step": 8260
    },
    {
      "epoch": 1.2932060112711334,
      "grad_norm": 2.018428087234497,
      "learning_rate": 0.0001106060606060606,
      "loss": 0.5406,
      "step": 8261
    },
    {
      "epoch": 1.2933625547902317,
      "grad_norm": 1.3222421407699585,
      "learning_rate": 0.00011058162267839687,
      "loss": 0.3733,
      "step": 8262
    },
    {
      "epoch": 1.2935190983093299,
      "grad_norm": 1.9479771852493286,
      "learning_rate": 0.00011055718475073312,
      "loss": 0.5531,
      "step": 8263
    },
    {
      "epoch": 1.2936756418284283,
      "grad_norm": 0.9404929876327515,
      "learning_rate": 0.00011053274682306939,
      "loss": 0.456,
      "step": 8264
    },
    {
      "epoch": 1.2938321853475265,
      "grad_norm": 2.357956886291504,
      "learning_rate": 0.00011050830889540567,
      "loss": 0.6238,
      "step": 8265
    },
    {
      "epoch": 1.293988728866625,
      "grad_norm": 1.7230677604675293,
      "learning_rate": 0.00011048387096774193,
      "loss": 0.565,
      "step": 8266
    },
    {
      "epoch": 1.2941452723857232,
      "grad_norm": 4.3762526512146,
      "learning_rate": 0.00011045943304007818,
      "loss": 0.9608,
      "step": 8267
    },
    {
      "epoch": 1.2943018159048214,
      "grad_norm": 3.0639162063598633,
      "learning_rate": 0.00011043499511241446,
      "loss": 0.7233,
      "step": 8268
    },
    {
      "epoch": 1.2944583594239198,
      "grad_norm": 1.96674644947052,
      "learning_rate": 0.00011041055718475073,
      "loss": 0.7643,
      "step": 8269
    },
    {
      "epoch": 1.2946149029430183,
      "grad_norm": 4.862483978271484,
      "learning_rate": 0.00011038611925708699,
      "loss": 0.6065,
      "step": 8270
    },
    {
      "epoch": 1.2947714464621165,
      "grad_norm": 2.0144357681274414,
      "learning_rate": 0.00011036168132942327,
      "loss": 0.8087,
      "step": 8271
    },
    {
      "epoch": 1.2949279899812147,
      "grad_norm": 3.6097030639648438,
      "learning_rate": 0.00011033724340175952,
      "loss": 1.2482,
      "step": 8272
    },
    {
      "epoch": 1.2950845335003132,
      "grad_norm": 1.5838818550109863,
      "learning_rate": 0.00011031280547409578,
      "loss": 0.5976,
      "step": 8273
    },
    {
      "epoch": 1.2952410770194114,
      "grad_norm": 2.3606481552124023,
      "learning_rate": 0.00011028836754643206,
      "loss": 0.7551,
      "step": 8274
    },
    {
      "epoch": 1.2953976205385098,
      "grad_norm": 1.6453297138214111,
      "learning_rate": 0.00011026392961876831,
      "loss": 0.6356,
      "step": 8275
    },
    {
      "epoch": 1.295554164057608,
      "grad_norm": 2.8421103954315186,
      "learning_rate": 0.00011023949169110458,
      "loss": 0.8155,
      "step": 8276
    },
    {
      "epoch": 1.2957107075767063,
      "grad_norm": 2.4521589279174805,
      "learning_rate": 0.00011021505376344086,
      "loss": 0.813,
      "step": 8277
    },
    {
      "epoch": 1.2958672510958047,
      "grad_norm": 2.380380868911743,
      "learning_rate": 0.00011019061583577712,
      "loss": 0.981,
      "step": 8278
    },
    {
      "epoch": 1.296023794614903,
      "grad_norm": 2.3057072162628174,
      "learning_rate": 0.00011016617790811337,
      "loss": 1.051,
      "step": 8279
    },
    {
      "epoch": 1.2961803381340014,
      "grad_norm": 2.9905688762664795,
      "learning_rate": 0.00011014173998044965,
      "loss": 1.2693,
      "step": 8280
    },
    {
      "epoch": 1.2963368816530996,
      "grad_norm": 9.889979362487793,
      "learning_rate": 0.00011011730205278592,
      "loss": 1.336,
      "step": 8281
    },
    {
      "epoch": 1.2964934251721978,
      "grad_norm": 2.7968623638153076,
      "learning_rate": 0.00011009286412512218,
      "loss": 1.5213,
      "step": 8282
    },
    {
      "epoch": 1.2966499686912962,
      "grad_norm": 2.403857469558716,
      "learning_rate": 0.00011006842619745846,
      "loss": 0.7344,
      "step": 8283
    },
    {
      "epoch": 1.2968065122103944,
      "grad_norm": 1.8447998762130737,
      "learning_rate": 0.00011004398826979471,
      "loss": 0.5804,
      "step": 8284
    },
    {
      "epoch": 1.2969630557294929,
      "grad_norm": 1.6647576093673706,
      "learning_rate": 0.00011001955034213098,
      "loss": 0.6722,
      "step": 8285
    },
    {
      "epoch": 1.297119599248591,
      "grad_norm": 3.1341192722320557,
      "learning_rate": 0.00010999511241446725,
      "loss": 0.8471,
      "step": 8286
    },
    {
      "epoch": 1.2972761427676893,
      "grad_norm": 2.227797746658325,
      "learning_rate": 0.0001099706744868035,
      "loss": 0.7715,
      "step": 8287
    },
    {
      "epoch": 1.2974326862867878,
      "grad_norm": 1.991193413734436,
      "learning_rate": 0.00010994623655913977,
      "loss": 0.7457,
      "step": 8288
    },
    {
      "epoch": 1.297589229805886,
      "grad_norm": 0.5508354902267456,
      "learning_rate": 0.00010992179863147605,
      "loss": 0.2327,
      "step": 8289
    },
    {
      "epoch": 1.2977457733249844,
      "grad_norm": 0.48610156774520874,
      "learning_rate": 0.00010989736070381231,
      "loss": 0.1992,
      "step": 8290
    },
    {
      "epoch": 1.2979023168440826,
      "grad_norm": 0.6914536356925964,
      "learning_rate": 0.00010987292277614856,
      "loss": 0.2157,
      "step": 8291
    },
    {
      "epoch": 1.2980588603631809,
      "grad_norm": 0.6526252627372742,
      "learning_rate": 0.00010984848484848484,
      "loss": 0.2804,
      "step": 8292
    },
    {
      "epoch": 1.2982154038822793,
      "grad_norm": 0.7797197103500366,
      "learning_rate": 0.00010982404692082111,
      "loss": 0.2843,
      "step": 8293
    },
    {
      "epoch": 1.2983719474013775,
      "grad_norm": 1.6517852544784546,
      "learning_rate": 0.00010979960899315737,
      "loss": 0.4454,
      "step": 8294
    },
    {
      "epoch": 1.298528490920476,
      "grad_norm": 0.5789437294006348,
      "learning_rate": 0.00010977517106549365,
      "loss": 0.1868,
      "step": 8295
    },
    {
      "epoch": 1.2986850344395742,
      "grad_norm": 1.5858449935913086,
      "learning_rate": 0.0001097507331378299,
      "loss": 0.3895,
      "step": 8296
    },
    {
      "epoch": 1.2988415779586724,
      "grad_norm": 0.5253259539604187,
      "learning_rate": 0.00010972629521016617,
      "loss": 0.2817,
      "step": 8297
    },
    {
      "epoch": 1.2989981214777708,
      "grad_norm": 0.7501121163368225,
      "learning_rate": 0.00010970185728250245,
      "loss": 0.279,
      "step": 8298
    },
    {
      "epoch": 1.2991546649968693,
      "grad_norm": 0.6583108305931091,
      "learning_rate": 0.0001096774193548387,
      "loss": 0.2974,
      "step": 8299
    },
    {
      "epoch": 1.2993112085159675,
      "grad_norm": 0.7893651723861694,
      "learning_rate": 0.00010965298142717496,
      "loss": 0.1879,
      "step": 8300
    },
    {
      "epoch": 1.2994677520350657,
      "grad_norm": 0.6830184459686279,
      "learning_rate": 0.00010962854349951124,
      "loss": 0.2392,
      "step": 8301
    },
    {
      "epoch": 1.299624295554164,
      "grad_norm": 0.9842543601989746,
      "learning_rate": 0.0001096041055718475,
      "loss": 0.3198,
      "step": 8302
    },
    {
      "epoch": 1.2997808390732624,
      "grad_norm": 1.0459553003311157,
      "learning_rate": 0.00010957966764418376,
      "loss": 0.3716,
      "step": 8303
    },
    {
      "epoch": 1.2999373825923608,
      "grad_norm": 0.7785957455635071,
      "learning_rate": 0.00010955522971652003,
      "loss": 0.2841,
      "step": 8304
    },
    {
      "epoch": 1.300093926111459,
      "grad_norm": 1.5206092596054077,
      "learning_rate": 0.0001095307917888563,
      "loss": 0.2479,
      "step": 8305
    },
    {
      "epoch": 1.3002504696305572,
      "grad_norm": 1.0379140377044678,
      "learning_rate": 0.00010950635386119256,
      "loss": 0.2994,
      "step": 8306
    },
    {
      "epoch": 1.3004070131496557,
      "grad_norm": 2.3565833568573,
      "learning_rate": 0.00010948191593352884,
      "loss": 0.6375,
      "step": 8307
    },
    {
      "epoch": 1.300563556668754,
      "grad_norm": 1.7786818742752075,
      "learning_rate": 0.0001094574780058651,
      "loss": 0.653,
      "step": 8308
    },
    {
      "epoch": 1.3007201001878523,
      "grad_norm": 1.0531212091445923,
      "learning_rate": 0.00010943304007820136,
      "loss": 0.3824,
      "step": 8309
    },
    {
      "epoch": 1.3008766437069506,
      "grad_norm": 1.700603723526001,
      "learning_rate": 0.00010940860215053764,
      "loss": 0.6215,
      "step": 8310
    },
    {
      "epoch": 1.3010331872260488,
      "grad_norm": 1.701421856880188,
      "learning_rate": 0.00010938416422287389,
      "loss": 0.4508,
      "step": 8311
    },
    {
      "epoch": 1.3011897307451472,
      "grad_norm": 1.6104190349578857,
      "learning_rate": 0.00010935972629521015,
      "loss": 0.5921,
      "step": 8312
    },
    {
      "epoch": 1.3013462742642454,
      "grad_norm": 0.7885456681251526,
      "learning_rate": 0.00010933528836754642,
      "loss": 0.3565,
      "step": 8313
    },
    {
      "epoch": 1.3015028177833439,
      "grad_norm": 1.718679666519165,
      "learning_rate": 0.0001093108504398827,
      "loss": 0.6124,
      "step": 8314
    },
    {
      "epoch": 1.301659361302442,
      "grad_norm": 4.118200302124023,
      "learning_rate": 0.00010928641251221895,
      "loss": 0.8574,
      "step": 8315
    },
    {
      "epoch": 1.3018159048215403,
      "grad_norm": 1.17429518699646,
      "learning_rate": 0.00010926197458455521,
      "loss": 0.5231,
      "step": 8316
    },
    {
      "epoch": 1.3019724483406387,
      "grad_norm": 3.1723527908325195,
      "learning_rate": 0.00010923753665689149,
      "loss": 0.3513,
      "step": 8317
    },
    {
      "epoch": 1.302128991859737,
      "grad_norm": 2.3258113861083984,
      "learning_rate": 0.00010921309872922775,
      "loss": 0.5292,
      "step": 8318
    },
    {
      "epoch": 1.3022855353788354,
      "grad_norm": 3.5120906829833984,
      "learning_rate": 0.000109188660801564,
      "loss": 0.2251,
      "step": 8319
    },
    {
      "epoch": 1.3024420788979336,
      "grad_norm": 1.4291130304336548,
      "learning_rate": 0.00010916422287390028,
      "loss": 0.4773,
      "step": 8320
    },
    {
      "epoch": 1.3025986224170318,
      "grad_norm": 3.090890884399414,
      "learning_rate": 0.00010913978494623655,
      "loss": 0.7817,
      "step": 8321
    },
    {
      "epoch": 1.3027551659361303,
      "grad_norm": 3.119511842727661,
      "learning_rate": 0.00010911534701857281,
      "loss": 0.8859,
      "step": 8322
    },
    {
      "epoch": 1.3029117094552285,
      "grad_norm": 1.6605695486068726,
      "learning_rate": 0.00010909090909090908,
      "loss": 0.2614,
      "step": 8323
    },
    {
      "epoch": 1.303068252974327,
      "grad_norm": 3.2734925746917725,
      "learning_rate": 0.00010906647116324534,
      "loss": 0.7882,
      "step": 8324
    },
    {
      "epoch": 1.3032247964934252,
      "grad_norm": 1.581061601638794,
      "learning_rate": 0.00010904203323558161,
      "loss": 0.5519,
      "step": 8325
    },
    {
      "epoch": 1.3033813400125234,
      "grad_norm": 1.6887787580490112,
      "learning_rate": 0.00010901759530791789,
      "loss": 1.1315,
      "step": 8326
    },
    {
      "epoch": 1.3035378835316218,
      "grad_norm": 5.0255937576293945,
      "learning_rate": 0.00010899315738025414,
      "loss": 1.2345,
      "step": 8327
    },
    {
      "epoch": 1.30369442705072,
      "grad_norm": 3.7134711742401123,
      "learning_rate": 0.0001089687194525904,
      "loss": 1.2284,
      "step": 8328
    },
    {
      "epoch": 1.3038509705698185,
      "grad_norm": 2.470165252685547,
      "learning_rate": 0.00010894428152492668,
      "loss": 0.588,
      "step": 8329
    },
    {
      "epoch": 1.3040075140889167,
      "grad_norm": 2.310760021209717,
      "learning_rate": 0.00010891984359726295,
      "loss": 0.8248,
      "step": 8330
    },
    {
      "epoch": 1.304164057608015,
      "grad_norm": 4.917720794677734,
      "learning_rate": 0.0001088954056695992,
      "loss": 1.843,
      "step": 8331
    },
    {
      "epoch": 1.3043206011271133,
      "grad_norm": 3.521848440170288,
      "learning_rate": 0.00010887096774193548,
      "loss": 1.4091,
      "step": 8332
    },
    {
      "epoch": 1.3044771446462118,
      "grad_norm": 1.4144970178604126,
      "learning_rate": 0.00010884652981427174,
      "loss": 1.0417,
      "step": 8333
    },
    {
      "epoch": 1.30463368816531,
      "grad_norm": 2.08368182182312,
      "learning_rate": 0.00010882209188660799,
      "loss": 0.7129,
      "step": 8334
    },
    {
      "epoch": 1.3047902316844082,
      "grad_norm": 2.0993666648864746,
      "learning_rate": 0.00010879765395894427,
      "loss": 1.2135,
      "step": 8335
    },
    {
      "epoch": 1.3049467752035064,
      "grad_norm": 4.826048374176025,
      "learning_rate": 0.00010877321603128053,
      "loss": 0.8379,
      "step": 8336
    },
    {
      "epoch": 1.3051033187226049,
      "grad_norm": 2.809587001800537,
      "learning_rate": 0.0001087487781036168,
      "loss": 0.603,
      "step": 8337
    },
    {
      "epoch": 1.3052598622417033,
      "grad_norm": 5.488523483276367,
      "learning_rate": 0.00010872434017595308,
      "loss": 1.2882,
      "step": 8338
    },
    {
      "epoch": 1.3054164057608015,
      "grad_norm": 1.1666176319122314,
      "learning_rate": 0.00010869990224828933,
      "loss": 0.3117,
      "step": 8339
    },
    {
      "epoch": 1.3055729492798998,
      "grad_norm": 0.6575944423675537,
      "learning_rate": 0.0001086754643206256,
      "loss": 0.2904,
      "step": 8340
    },
    {
      "epoch": 1.3057294927989982,
      "grad_norm": 0.6943413615226746,
      "learning_rate": 0.00010865102639296187,
      "loss": 0.202,
      "step": 8341
    },
    {
      "epoch": 1.3058860363180964,
      "grad_norm": 0.5365055799484253,
      "learning_rate": 0.00010862658846529814,
      "loss": 0.2215,
      "step": 8342
    },
    {
      "epoch": 1.3060425798371949,
      "grad_norm": 0.6395205855369568,
      "learning_rate": 0.00010860215053763439,
      "loss": 0.2398,
      "step": 8343
    },
    {
      "epoch": 1.306199123356293,
      "grad_norm": 0.6453742980957031,
      "learning_rate": 0.00010857771260997067,
      "loss": 0.2191,
      "step": 8344
    },
    {
      "epoch": 1.3063556668753913,
      "grad_norm": 0.4960818290710449,
      "learning_rate": 0.00010855327468230693,
      "loss": 0.1847,
      "step": 8345
    },
    {
      "epoch": 1.3065122103944897,
      "grad_norm": 0.8823752403259277,
      "learning_rate": 0.00010852883675464318,
      "loss": 0.2645,
      "step": 8346
    },
    {
      "epoch": 1.306668753913588,
      "grad_norm": 0.6216477751731873,
      "learning_rate": 0.00010850439882697946,
      "loss": 0.1919,
      "step": 8347
    },
    {
      "epoch": 1.3068252974326864,
      "grad_norm": 0.8468354344367981,
      "learning_rate": 0.00010847996089931573,
      "loss": 0.2694,
      "step": 8348
    },
    {
      "epoch": 1.3069818409517846,
      "grad_norm": 1.344497799873352,
      "learning_rate": 0.00010845552297165199,
      "loss": 0.3727,
      "step": 8349
    },
    {
      "epoch": 1.3071383844708828,
      "grad_norm": 1.8707456588745117,
      "learning_rate": 0.00010843108504398827,
      "loss": 0.3696,
      "step": 8350
    },
    {
      "epoch": 1.3072949279899813,
      "grad_norm": 1.084493637084961,
      "learning_rate": 0.00010840664711632452,
      "loss": 0.2772,
      "step": 8351
    },
    {
      "epoch": 1.3074514715090795,
      "grad_norm": 0.9700431823730469,
      "learning_rate": 0.00010838220918866079,
      "loss": 0.2514,
      "step": 8352
    },
    {
      "epoch": 1.307608015028178,
      "grad_norm": 1.6982876062393188,
      "learning_rate": 0.00010835777126099706,
      "loss": 0.2902,
      "step": 8353
    },
    {
      "epoch": 1.3077645585472761,
      "grad_norm": 1.004989504814148,
      "learning_rate": 0.00010833333333333333,
      "loss": 0.3883,
      "step": 8354
    },
    {
      "epoch": 1.3079211020663744,
      "grad_norm": 1.6661001443862915,
      "learning_rate": 0.00010830889540566958,
      "loss": 0.5643,
      "step": 8355
    },
    {
      "epoch": 1.3080776455854728,
      "grad_norm": 1.4825246334075928,
      "learning_rate": 0.00010828445747800586,
      "loss": 0.5259,
      "step": 8356
    },
    {
      "epoch": 1.308234189104571,
      "grad_norm": 1.141077995300293,
      "learning_rate": 0.00010826001955034212,
      "loss": 0.2891,
      "step": 8357
    },
    {
      "epoch": 1.3083907326236695,
      "grad_norm": 2.092963695526123,
      "learning_rate": 0.00010823558162267837,
      "loss": 0.7122,
      "step": 8358
    },
    {
      "epoch": 1.3085472761427677,
      "grad_norm": 0.9098227024078369,
      "learning_rate": 0.00010821114369501465,
      "loss": 0.3163,
      "step": 8359
    },
    {
      "epoch": 1.3087038196618659,
      "grad_norm": 2.1590492725372314,
      "learning_rate": 0.00010818670576735092,
      "loss": 0.6714,
      "step": 8360
    },
    {
      "epoch": 1.3088603631809643,
      "grad_norm": 1.1368892192840576,
      "learning_rate": 0.00010816226783968718,
      "loss": 0.5353,
      "step": 8361
    },
    {
      "epoch": 1.3090169067000625,
      "grad_norm": 1.050818681716919,
      "learning_rate": 0.00010813782991202346,
      "loss": 0.4366,
      "step": 8362
    },
    {
      "epoch": 1.309173450219161,
      "grad_norm": 0.9604326486587524,
      "learning_rate": 0.00010811339198435971,
      "loss": 0.377,
      "step": 8363
    },
    {
      "epoch": 1.3093299937382592,
      "grad_norm": 2.909843683242798,
      "learning_rate": 0.00010808895405669598,
      "loss": 0.709,
      "step": 8364
    },
    {
      "epoch": 1.3094865372573574,
      "grad_norm": 2.03232741355896,
      "learning_rate": 0.00010806451612903225,
      "loss": 0.5225,
      "step": 8365
    },
    {
      "epoch": 1.3096430807764559,
      "grad_norm": 1.5411139726638794,
      "learning_rate": 0.00010804007820136852,
      "loss": 0.7947,
      "step": 8366
    },
    {
      "epoch": 1.3097996242955543,
      "grad_norm": 1.4965941905975342,
      "learning_rate": 0.00010801564027370477,
      "loss": 0.5839,
      "step": 8367
    },
    {
      "epoch": 1.3099561678146525,
      "grad_norm": 2.01226806640625,
      "learning_rate": 0.00010799120234604105,
      "loss": 0.6892,
      "step": 8368
    },
    {
      "epoch": 1.3101127113337507,
      "grad_norm": 4.478280067443848,
      "learning_rate": 0.00010796676441837731,
      "loss": 0.437,
      "step": 8369
    },
    {
      "epoch": 1.3102692548528492,
      "grad_norm": 2.566091299057007,
      "learning_rate": 0.00010794232649071357,
      "loss": 0.5926,
      "step": 8370
    },
    {
      "epoch": 1.3104257983719474,
      "grad_norm": 2.0691118240356445,
      "learning_rate": 0.00010791788856304984,
      "loss": 0.6444,
      "step": 8371
    },
    {
      "epoch": 1.3105823418910458,
      "grad_norm": 2.3462331295013428,
      "learning_rate": 0.00010789345063538611,
      "loss": 0.8449,
      "step": 8372
    },
    {
      "epoch": 1.310738885410144,
      "grad_norm": 1.7684667110443115,
      "learning_rate": 0.00010786901270772237,
      "loss": 0.6608,
      "step": 8373
    },
    {
      "epoch": 1.3108954289292423,
      "grad_norm": 2.054476022720337,
      "learning_rate": 0.00010784457478005865,
      "loss": 1.0546,
      "step": 8374
    },
    {
      "epoch": 1.3110519724483407,
      "grad_norm": 2.8694870471954346,
      "learning_rate": 0.0001078201368523949,
      "loss": 0.9144,
      "step": 8375
    },
    {
      "epoch": 1.311208515967439,
      "grad_norm": 1.2306618690490723,
      "learning_rate": 0.00010779569892473117,
      "loss": 0.5151,
      "step": 8376
    },
    {
      "epoch": 1.3113650594865374,
      "grad_norm": 1.6525115966796875,
      "learning_rate": 0.00010777126099706745,
      "loss": 0.8821,
      "step": 8377
    },
    {
      "epoch": 1.3115216030056356,
      "grad_norm": 3.0999720096588135,
      "learning_rate": 0.0001077468230694037,
      "loss": 1.6076,
      "step": 8378
    },
    {
      "epoch": 1.3116781465247338,
      "grad_norm": 1.6865127086639404,
      "learning_rate": 0.00010772238514173996,
      "loss": 0.8059,
      "step": 8379
    },
    {
      "epoch": 1.3118346900438322,
      "grad_norm": 2.4270031452178955,
      "learning_rate": 0.00010769794721407624,
      "loss": 0.5467,
      "step": 8380
    },
    {
      "epoch": 1.3119912335629305,
      "grad_norm": 2.8954577445983887,
      "learning_rate": 0.0001076735092864125,
      "loss": 0.532,
      "step": 8381
    },
    {
      "epoch": 1.312147777082029,
      "grad_norm": 4.275869846343994,
      "learning_rate": 0.00010764907135874876,
      "loss": 1.7656,
      "step": 8382
    },
    {
      "epoch": 1.3123043206011271,
      "grad_norm": 1.7701374292373657,
      "learning_rate": 0.00010762463343108503,
      "loss": 1.1036,
      "step": 8383
    },
    {
      "epoch": 1.3124608641202253,
      "grad_norm": 2.1359615325927734,
      "learning_rate": 0.0001076001955034213,
      "loss": 0.4755,
      "step": 8384
    },
    {
      "epoch": 1.3126174076393238,
      "grad_norm": 1.1913822889328003,
      "learning_rate": 0.00010757575757575756,
      "loss": 0.349,
      "step": 8385
    },
    {
      "epoch": 1.312773951158422,
      "grad_norm": 1.3649957180023193,
      "learning_rate": 0.00010755131964809384,
      "loss": 0.4667,
      "step": 8386
    },
    {
      "epoch": 1.3129304946775204,
      "grad_norm": 2.1878700256347656,
      "learning_rate": 0.0001075268817204301,
      "loss": 0.7366,
      "step": 8387
    },
    {
      "epoch": 1.3130870381966186,
      "grad_norm": 2.3514275550842285,
      "learning_rate": 0.00010750244379276636,
      "loss": 1.0322,
      "step": 8388
    },
    {
      "epoch": 1.3132435817157169,
      "grad_norm": 0.49473220109939575,
      "learning_rate": 0.00010747800586510264,
      "loss": 0.211,
      "step": 8389
    },
    {
      "epoch": 1.3134001252348153,
      "grad_norm": 1.2258360385894775,
      "learning_rate": 0.00010745356793743889,
      "loss": 0.2018,
      "step": 8390
    },
    {
      "epoch": 1.3135566687539135,
      "grad_norm": 0.7705320119857788,
      "learning_rate": 0.00010742913000977515,
      "loss": 0.2365,
      "step": 8391
    },
    {
      "epoch": 1.313713212273012,
      "grad_norm": 0.46400439739227295,
      "learning_rate": 0.00010740469208211143,
      "loss": 0.168,
      "step": 8392
    },
    {
      "epoch": 1.3138697557921102,
      "grad_norm": 0.7878824472427368,
      "learning_rate": 0.0001073802541544477,
      "loss": 0.3099,
      "step": 8393
    },
    {
      "epoch": 1.3140262993112084,
      "grad_norm": 0.9498413801193237,
      "learning_rate": 0.00010735581622678395,
      "loss": 0.1739,
      "step": 8394
    },
    {
      "epoch": 1.3141828428303068,
      "grad_norm": 0.9091798663139343,
      "learning_rate": 0.00010733137829912023,
      "loss": 0.2746,
      "step": 8395
    },
    {
      "epoch": 1.314339386349405,
      "grad_norm": 0.8814572691917419,
      "learning_rate": 0.00010730694037145649,
      "loss": 0.2913,
      "step": 8396
    },
    {
      "epoch": 1.3144959298685035,
      "grad_norm": 0.8799420595169067,
      "learning_rate": 0.00010728250244379276,
      "loss": 0.3373,
      "step": 8397
    },
    {
      "epoch": 1.3146524733876017,
      "grad_norm": 0.6307176351547241,
      "learning_rate": 0.00010725806451612903,
      "loss": 0.3035,
      "step": 8398
    },
    {
      "epoch": 1.3148090169067,
      "grad_norm": 0.962165355682373,
      "learning_rate": 0.00010723362658846529,
      "loss": 0.2801,
      "step": 8399
    },
    {
      "epoch": 1.3149655604257984,
      "grad_norm": 0.4829339385032654,
      "learning_rate": 0.00010720918866080155,
      "loss": 0.1575,
      "step": 8400
    },
    {
      "epoch": 1.3151221039448968,
      "grad_norm": 0.9093053936958313,
      "learning_rate": 0.00010718475073313783,
      "loss": 0.3151,
      "step": 8401
    },
    {
      "epoch": 1.315278647463995,
      "grad_norm": 1.592719316482544,
      "learning_rate": 0.00010716031280547408,
      "loss": 0.3769,
      "step": 8402
    },
    {
      "epoch": 1.3154351909830932,
      "grad_norm": 1.357019066810608,
      "learning_rate": 0.00010713587487781034,
      "loss": 0.3612,
      "step": 8403
    },
    {
      "epoch": 1.3155917345021917,
      "grad_norm": 4.118210315704346,
      "learning_rate": 0.00010711143695014662,
      "loss": 0.3585,
      "step": 8404
    },
    {
      "epoch": 1.31574827802129,
      "grad_norm": 1.2734813690185547,
      "learning_rate": 0.00010708699902248289,
      "loss": 0.4046,
      "step": 8405
    },
    {
      "epoch": 1.3159048215403883,
      "grad_norm": 0.6813839077949524,
      "learning_rate": 0.00010706256109481914,
      "loss": 0.2412,
      "step": 8406
    },
    {
      "epoch": 1.3160613650594866,
      "grad_norm": 1.974395751953125,
      "learning_rate": 0.00010703812316715542,
      "loss": 0.3266,
      "step": 8407
    },
    {
      "epoch": 1.3162179085785848,
      "grad_norm": 1.261323094367981,
      "learning_rate": 0.00010701368523949168,
      "loss": 0.3412,
      "step": 8408
    },
    {
      "epoch": 1.3163744520976832,
      "grad_norm": 1.1508007049560547,
      "learning_rate": 0.00010698924731182795,
      "loss": 0.3782,
      "step": 8409
    },
    {
      "epoch": 1.3165309956167814,
      "grad_norm": 1.8937280178070068,
      "learning_rate": 0.00010696480938416422,
      "loss": 0.5881,
      "step": 8410
    },
    {
      "epoch": 1.3166875391358799,
      "grad_norm": 1.3329272270202637,
      "learning_rate": 0.00010694037145650048,
      "loss": 0.3392,
      "step": 8411
    },
    {
      "epoch": 1.316844082654978,
      "grad_norm": 0.9232223033905029,
      "learning_rate": 0.00010691593352883674,
      "loss": 0.5811,
      "step": 8412
    },
    {
      "epoch": 1.3170006261740763,
      "grad_norm": 1.0369586944580078,
      "learning_rate": 0.00010689149560117302,
      "loss": 0.4383,
      "step": 8413
    },
    {
      "epoch": 1.3171571696931748,
      "grad_norm": 2.1007254123687744,
      "learning_rate": 0.00010686705767350927,
      "loss": 0.554,
      "step": 8414
    },
    {
      "epoch": 1.317313713212273,
      "grad_norm": 1.1402606964111328,
      "learning_rate": 0.00010684261974584554,
      "loss": 0.4968,
      "step": 8415
    },
    {
      "epoch": 1.3174702567313714,
      "grad_norm": 1.1033568382263184,
      "learning_rate": 0.00010681818181818181,
      "loss": 0.3661,
      "step": 8416
    },
    {
      "epoch": 1.3176268002504696,
      "grad_norm": 1.5588769912719727,
      "learning_rate": 0.00010679374389051808,
      "loss": 0.3178,
      "step": 8417
    },
    {
      "epoch": 1.3177833437695678,
      "grad_norm": 1.2382311820983887,
      "learning_rate": 0.00010676930596285433,
      "loss": 0.5225,
      "step": 8418
    },
    {
      "epoch": 1.3179398872886663,
      "grad_norm": 2.5015268325805664,
      "learning_rate": 0.00010674486803519061,
      "loss": 0.7478,
      "step": 8419
    },
    {
      "epoch": 1.3180964308077645,
      "grad_norm": 3.2753753662109375,
      "learning_rate": 0.00010672043010752687,
      "loss": 0.5885,
      "step": 8420
    },
    {
      "epoch": 1.318252974326863,
      "grad_norm": 2.250828742980957,
      "learning_rate": 0.00010669599217986314,
      "loss": 0.6413,
      "step": 8421
    },
    {
      "epoch": 1.3184095178459612,
      "grad_norm": 1.6455849409103394,
      "learning_rate": 0.0001066715542521994,
      "loss": 0.8944,
      "step": 8422
    },
    {
      "epoch": 1.3185660613650594,
      "grad_norm": 1.4225467443466187,
      "learning_rate": 0.00010664711632453567,
      "loss": 1.0937,
      "step": 8423
    },
    {
      "epoch": 1.3187226048841578,
      "grad_norm": 3.475102663040161,
      "learning_rate": 0.00010662267839687193,
      "loss": 1.155,
      "step": 8424
    },
    {
      "epoch": 1.318879148403256,
      "grad_norm": 1.6069629192352295,
      "learning_rate": 0.00010659824046920821,
      "loss": 1.1532,
      "step": 8425
    },
    {
      "epoch": 1.3190356919223545,
      "grad_norm": 2.740570306777954,
      "learning_rate": 0.00010657380254154446,
      "loss": 0.9993,
      "step": 8426
    },
    {
      "epoch": 1.3191922354414527,
      "grad_norm": 1.2884479761123657,
      "learning_rate": 0.00010654936461388073,
      "loss": 0.342,
      "step": 8427
    },
    {
      "epoch": 1.319348778960551,
      "grad_norm": 5.032875061035156,
      "learning_rate": 0.000106524926686217,
      "loss": 0.8971,
      "step": 8428
    },
    {
      "epoch": 1.3195053224796494,
      "grad_norm": 3.1008434295654297,
      "learning_rate": 0.00010650048875855327,
      "loss": 1.1503,
      "step": 8429
    },
    {
      "epoch": 1.3196618659987476,
      "grad_norm": 3.7066233158111572,
      "learning_rate": 0.00010647605083088952,
      "loss": 1.0131,
      "step": 8430
    },
    {
      "epoch": 1.319818409517846,
      "grad_norm": 2.721331834793091,
      "learning_rate": 0.0001064516129032258,
      "loss": 1.4015,
      "step": 8431
    },
    {
      "epoch": 1.3199749530369442,
      "grad_norm": 2.37866473197937,
      "learning_rate": 0.00010642717497556206,
      "loss": 0.7506,
      "step": 8432
    },
    {
      "epoch": 1.3201314965560424,
      "grad_norm": 1.654268503189087,
      "learning_rate": 0.00010640273704789833,
      "loss": 1.0727,
      "step": 8433
    },
    {
      "epoch": 1.320288040075141,
      "grad_norm": 0.7855532765388489,
      "learning_rate": 0.0001063782991202346,
      "loss": 0.427,
      "step": 8434
    },
    {
      "epoch": 1.3204445835942393,
      "grad_norm": 2.7681164741516113,
      "learning_rate": 0.00010635386119257086,
      "loss": 0.6105,
      "step": 8435
    },
    {
      "epoch": 1.3206011271133375,
      "grad_norm": 4.213181495666504,
      "learning_rate": 0.00010632942326490712,
      "loss": 0.7922,
      "step": 8436
    },
    {
      "epoch": 1.3207576706324358,
      "grad_norm": 4.588217258453369,
      "learning_rate": 0.0001063049853372434,
      "loss": 1.3575,
      "step": 8437
    },
    {
      "epoch": 1.3209142141515342,
      "grad_norm": 1.3220864534378052,
      "learning_rate": 0.00010628054740957965,
      "loss": 0.9862,
      "step": 8438
    },
    {
      "epoch": 1.3210707576706324,
      "grad_norm": 0.4366258680820465,
      "learning_rate": 0.00010625610948191592,
      "loss": 0.2151,
      "step": 8439
    },
    {
      "epoch": 1.3212273011897309,
      "grad_norm": 0.3557334244251251,
      "learning_rate": 0.0001062316715542522,
      "loss": 0.1961,
      "step": 8440
    },
    {
      "epoch": 1.321383844708829,
      "grad_norm": 0.6270286440849304,
      "learning_rate": 0.00010620723362658846,
      "loss": 0.3202,
      "step": 8441
    },
    {
      "epoch": 1.3215403882279273,
      "grad_norm": 0.5191538333892822,
      "learning_rate": 0.00010618279569892471,
      "loss": 0.1946,
      "step": 8442
    },
    {
      "epoch": 1.3216969317470257,
      "grad_norm": 0.35942304134368896,
      "learning_rate": 0.00010615835777126099,
      "loss": 0.2023,
      "step": 8443
    },
    {
      "epoch": 1.321853475266124,
      "grad_norm": 0.5088504552841187,
      "learning_rate": 0.00010613391984359726,
      "loss": 0.2216,
      "step": 8444
    },
    {
      "epoch": 1.3220100187852224,
      "grad_norm": 0.6313852667808533,
      "learning_rate": 0.00010610948191593352,
      "loss": 0.2147,
      "step": 8445
    },
    {
      "epoch": 1.3221665623043206,
      "grad_norm": 1.02388596534729,
      "learning_rate": 0.00010608504398826978,
      "loss": 0.3264,
      "step": 8446
    },
    {
      "epoch": 1.3223231058234188,
      "grad_norm": 0.7219918370246887,
      "learning_rate": 0.00010606060606060605,
      "loss": 0.3412,
      "step": 8447
    },
    {
      "epoch": 1.3224796493425173,
      "grad_norm": 1.3611302375793457,
      "learning_rate": 0.00010603616813294231,
      "loss": 0.2682,
      "step": 8448
    },
    {
      "epoch": 1.3226361928616155,
      "grad_norm": 0.6783427596092224,
      "learning_rate": 0.00010601173020527859,
      "loss": 0.2816,
      "step": 8449
    },
    {
      "epoch": 1.322792736380714,
      "grad_norm": 1.3119559288024902,
      "learning_rate": 0.00010598729227761484,
      "loss": 0.5017,
      "step": 8450
    },
    {
      "epoch": 1.3229492798998121,
      "grad_norm": 2.026618719100952,
      "learning_rate": 0.00010596285434995111,
      "loss": 0.6098,
      "step": 8451
    },
    {
      "epoch": 1.3231058234189104,
      "grad_norm": 1.1035137176513672,
      "learning_rate": 0.00010593841642228739,
      "loss": 0.3054,
      "step": 8452
    },
    {
      "epoch": 1.3232623669380088,
      "grad_norm": 1.0470503568649292,
      "learning_rate": 0.00010591397849462365,
      "loss": 0.322,
      "step": 8453
    },
    {
      "epoch": 1.323418910457107,
      "grad_norm": 1.4040778875350952,
      "learning_rate": 0.0001058895405669599,
      "loss": 0.4344,
      "step": 8454
    },
    {
      "epoch": 1.3235754539762055,
      "grad_norm": 1.1632486581802368,
      "learning_rate": 0.00010586510263929618,
      "loss": 0.4741,
      "step": 8455
    },
    {
      "epoch": 1.3237319974953037,
      "grad_norm": 2.275049924850464,
      "learning_rate": 0.00010584066471163245,
      "loss": 0.4954,
      "step": 8456
    },
    {
      "epoch": 1.323888541014402,
      "grad_norm": 0.912585973739624,
      "learning_rate": 0.00010581622678396871,
      "loss": 0.2482,
      "step": 8457
    },
    {
      "epoch": 1.3240450845335003,
      "grad_norm": 2.720503568649292,
      "learning_rate": 0.00010579178885630498,
      "loss": 0.563,
      "step": 8458
    },
    {
      "epoch": 1.3242016280525986,
      "grad_norm": 1.2635301351547241,
      "learning_rate": 0.00010576735092864124,
      "loss": 0.3343,
      "step": 8459
    },
    {
      "epoch": 1.324358171571697,
      "grad_norm": 2.1451432704925537,
      "learning_rate": 0.0001057429130009775,
      "loss": 0.4257,
      "step": 8460
    },
    {
      "epoch": 1.3245147150907952,
      "grad_norm": 0.8080310225486755,
      "learning_rate": 0.00010571847507331378,
      "loss": 0.3141,
      "step": 8461
    },
    {
      "epoch": 1.3246712586098934,
      "grad_norm": 2.107248544692993,
      "learning_rate": 0.00010569403714565004,
      "loss": 0.3324,
      "step": 8462
    },
    {
      "epoch": 1.3248278021289919,
      "grad_norm": 1.0356974601745605,
      "learning_rate": 0.0001056695992179863,
      "loss": 0.3303,
      "step": 8463
    },
    {
      "epoch": 1.32498434564809,
      "grad_norm": 1.8852458000183105,
      "learning_rate": 0.00010564516129032258,
      "loss": 0.4119,
      "step": 8464
    },
    {
      "epoch": 1.3251408891671885,
      "grad_norm": 1.1586437225341797,
      "learning_rate": 0.00010562072336265884,
      "loss": 0.4388,
      "step": 8465
    },
    {
      "epoch": 1.3252974326862867,
      "grad_norm": 2.8293962478637695,
      "learning_rate": 0.0001055962854349951,
      "loss": 0.7354,
      "step": 8466
    },
    {
      "epoch": 1.325453976205385,
      "grad_norm": 3.1717278957366943,
      "learning_rate": 0.00010557184750733137,
      "loss": 0.6096,
      "step": 8467
    },
    {
      "epoch": 1.3256105197244834,
      "grad_norm": 3.8848044872283936,
      "learning_rate": 0.00010554740957966764,
      "loss": 1.0067,
      "step": 8468
    },
    {
      "epoch": 1.3257670632435818,
      "grad_norm": 1.52760648727417,
      "learning_rate": 0.0001055229716520039,
      "loss": 0.5702,
      "step": 8469
    },
    {
      "epoch": 1.32592360676268,
      "grad_norm": 1.8875513076782227,
      "learning_rate": 0.00010549853372434017,
      "loss": 0.8501,
      "step": 8470
    },
    {
      "epoch": 1.3260801502817783,
      "grad_norm": 4.692626953125,
      "learning_rate": 0.00010547409579667643,
      "loss": 1.1365,
      "step": 8471
    },
    {
      "epoch": 1.3262366938008767,
      "grad_norm": 3.195230484008789,
      "learning_rate": 0.0001054496578690127,
      "loss": 1.165,
      "step": 8472
    },
    {
      "epoch": 1.326393237319975,
      "grad_norm": 2.567884683609009,
      "learning_rate": 0.00010542521994134898,
      "loss": 1.016,
      "step": 8473
    },
    {
      "epoch": 1.3265497808390734,
      "grad_norm": 2.327993392944336,
      "learning_rate": 0.00010540078201368523,
      "loss": 1.1238,
      "step": 8474
    },
    {
      "epoch": 1.3267063243581716,
      "grad_norm": 4.008900165557861,
      "learning_rate": 0.00010537634408602149,
      "loss": 1.3329,
      "step": 8475
    },
    {
      "epoch": 1.3268628678772698,
      "grad_norm": 3.4523065090179443,
      "learning_rate": 0.00010535190615835777,
      "loss": 1.2717,
      "step": 8476
    },
    {
      "epoch": 1.3270194113963683,
      "grad_norm": 2.4509525299072266,
      "learning_rate": 0.00010532746823069403,
      "loss": 1.3055,
      "step": 8477
    },
    {
      "epoch": 1.3271759549154665,
      "grad_norm": 2.2657418251037598,
      "learning_rate": 0.00010530303030303029,
      "loss": 1.0867,
      "step": 8478
    },
    {
      "epoch": 1.327332498434565,
      "grad_norm": 10.114274024963379,
      "learning_rate": 0.00010527859237536656,
      "loss": 1.2424,
      "step": 8479
    },
    {
      "epoch": 1.3274890419536631,
      "grad_norm": 2.6086230278015137,
      "learning_rate": 0.00010525415444770283,
      "loss": 1.5289,
      "step": 8480
    },
    {
      "epoch": 1.3276455854727613,
      "grad_norm": 7.500784397125244,
      "learning_rate": 0.0001052297165200391,
      "loss": 0.8191,
      "step": 8481
    },
    {
      "epoch": 1.3278021289918598,
      "grad_norm": 2.1050162315368652,
      "learning_rate": 0.00010520527859237536,
      "loss": 0.9757,
      "step": 8482
    },
    {
      "epoch": 1.327958672510958,
      "grad_norm": 1.964867353439331,
      "learning_rate": 0.00010518084066471162,
      "loss": 1.1642,
      "step": 8483
    },
    {
      "epoch": 1.3281152160300564,
      "grad_norm": 2.479874849319458,
      "learning_rate": 0.00010515640273704789,
      "loss": 0.9598,
      "step": 8484
    },
    {
      "epoch": 1.3282717595491547,
      "grad_norm": 1.8766109943389893,
      "learning_rate": 0.00010513196480938417,
      "loss": 0.3375,
      "step": 8485
    },
    {
      "epoch": 1.3284283030682529,
      "grad_norm": 2.781697988510132,
      "learning_rate": 0.00010510752688172042,
      "loss": 1.2596,
      "step": 8486
    },
    {
      "epoch": 1.3285848465873513,
      "grad_norm": 4.228986740112305,
      "learning_rate": 0.00010508308895405668,
      "loss": 1.0103,
      "step": 8487
    },
    {
      "epoch": 1.3287413901064495,
      "grad_norm": 1.7900171279907227,
      "learning_rate": 0.00010505865102639296,
      "loss": 0.83,
      "step": 8488
    },
    {
      "epoch": 1.328897933625548,
      "grad_norm": 0.6766242384910583,
      "learning_rate": 0.00010503421309872923,
      "loss": 0.2403,
      "step": 8489
    },
    {
      "epoch": 1.3290544771446462,
      "grad_norm": 0.5489671230316162,
      "learning_rate": 0.00010500977517106548,
      "loss": 0.3599,
      "step": 8490
    },
    {
      "epoch": 1.3292110206637444,
      "grad_norm": 0.7693419456481934,
      "learning_rate": 0.00010498533724340176,
      "loss": 0.2344,
      "step": 8491
    },
    {
      "epoch": 1.3293675641828429,
      "grad_norm": 0.5848720073699951,
      "learning_rate": 0.00010496089931573802,
      "loss": 0.2692,
      "step": 8492
    },
    {
      "epoch": 1.329524107701941,
      "grad_norm": 0.4316113591194153,
      "learning_rate": 0.00010493646138807427,
      "loss": 0.1891,
      "step": 8493
    },
    {
      "epoch": 1.3296806512210395,
      "grad_norm": 0.5180829167366028,
      "learning_rate": 0.00010491202346041055,
      "loss": 0.2097,
      "step": 8494
    },
    {
      "epoch": 1.3298371947401377,
      "grad_norm": 0.31930309534072876,
      "learning_rate": 0.00010488758553274681,
      "loss": 0.1346,
      "step": 8495
    },
    {
      "epoch": 1.329993738259236,
      "grad_norm": 0.5279285907745361,
      "learning_rate": 0.00010486314760508308,
      "loss": 0.3029,
      "step": 8496
    },
    {
      "epoch": 1.3301502817783344,
      "grad_norm": 0.6613864898681641,
      "learning_rate": 0.00010483870967741936,
      "loss": 0.319,
      "step": 8497
    },
    {
      "epoch": 1.3303068252974326,
      "grad_norm": 0.5798127055168152,
      "learning_rate": 0.00010481427174975561,
      "loss": 0.2005,
      "step": 8498
    },
    {
      "epoch": 1.330463368816531,
      "grad_norm": 0.7351884245872498,
      "learning_rate": 0.00010478983382209187,
      "loss": 0.2422,
      "step": 8499
    },
    {
      "epoch": 1.3306199123356293,
      "grad_norm": 0.8078888058662415,
      "learning_rate": 0.00010476539589442815,
      "loss": 0.3467,
      "step": 8500
    },
    {
      "epoch": 1.3307764558547275,
      "grad_norm": 0.8279627561569214,
      "learning_rate": 0.00010474095796676442,
      "loss": 0.2722,
      "step": 8501
    },
    {
      "epoch": 1.330932999373826,
      "grad_norm": 1.2416898012161255,
      "learning_rate": 0.00010471652003910067,
      "loss": 0.4529,
      "step": 8502
    },
    {
      "epoch": 1.3310895428929244,
      "grad_norm": 1.2886146306991577,
      "learning_rate": 0.00010469208211143695,
      "loss": 0.4516,
      "step": 8503
    },
    {
      "epoch": 1.3312460864120226,
      "grad_norm": 0.4999758303165436,
      "learning_rate": 0.00010466764418377321,
      "loss": 0.1717,
      "step": 8504
    },
    {
      "epoch": 1.3314026299311208,
      "grad_norm": 1.2985860109329224,
      "learning_rate": 0.00010464320625610946,
      "loss": 0.3384,
      "step": 8505
    },
    {
      "epoch": 1.3315591734502192,
      "grad_norm": 2.1126205921173096,
      "learning_rate": 0.00010461876832844574,
      "loss": 0.4438,
      "step": 8506
    },
    {
      "epoch": 1.3317157169693175,
      "grad_norm": 2.8582603931427,
      "learning_rate": 0.000104594330400782,
      "loss": 0.2766,
      "step": 8507
    },
    {
      "epoch": 1.331872260488416,
      "grad_norm": 1.146053433418274,
      "learning_rate": 0.00010456989247311827,
      "loss": 0.3095,
      "step": 8508
    },
    {
      "epoch": 1.3320288040075141,
      "grad_norm": 1.1559828519821167,
      "learning_rate": 0.00010454545454545455,
      "loss": 0.5287,
      "step": 8509
    },
    {
      "epoch": 1.3321853475266123,
      "grad_norm": 1.217333436012268,
      "learning_rate": 0.0001045210166177908,
      "loss": 0.3,
      "step": 8510
    },
    {
      "epoch": 1.3323418910457108,
      "grad_norm": 1.9483610391616821,
      "learning_rate": 0.00010449657869012706,
      "loss": 0.6234,
      "step": 8511
    },
    {
      "epoch": 1.332498434564809,
      "grad_norm": 1.2724167108535767,
      "learning_rate": 0.00010447214076246334,
      "loss": 0.6622,
      "step": 8512
    },
    {
      "epoch": 1.3326549780839074,
      "grad_norm": 3.2737345695495605,
      "learning_rate": 0.00010444770283479961,
      "loss": 0.8264,
      "step": 8513
    },
    {
      "epoch": 1.3328115216030056,
      "grad_norm": 2.594571352005005,
      "learning_rate": 0.00010442326490713586,
      "loss": 0.649,
      "step": 8514
    },
    {
      "epoch": 1.3329680651221039,
      "grad_norm": 2.1215741634368896,
      "learning_rate": 0.00010439882697947214,
      "loss": 0.4745,
      "step": 8515
    },
    {
      "epoch": 1.3331246086412023,
      "grad_norm": 2.0820720195770264,
      "learning_rate": 0.0001043743890518084,
      "loss": 0.8717,
      "step": 8516
    },
    {
      "epoch": 1.3332811521603005,
      "grad_norm": 1.5317957401275635,
      "learning_rate": 0.00010434995112414465,
      "loss": 0.6137,
      "step": 8517
    },
    {
      "epoch": 1.333437695679399,
      "grad_norm": 2.0376477241516113,
      "learning_rate": 0.00010432551319648093,
      "loss": 0.9649,
      "step": 8518
    },
    {
      "epoch": 1.3335942391984972,
      "grad_norm": 2.7941699028015137,
      "learning_rate": 0.0001043010752688172,
      "loss": 0.963,
      "step": 8519
    },
    {
      "epoch": 1.3337507827175954,
      "grad_norm": 2.3799190521240234,
      "learning_rate": 0.00010427663734115346,
      "loss": 0.5505,
      "step": 8520
    },
    {
      "epoch": 1.3339073262366938,
      "grad_norm": 1.9923593997955322,
      "learning_rate": 0.00010425219941348974,
      "loss": 0.4311,
      "step": 8521
    },
    {
      "epoch": 1.334063869755792,
      "grad_norm": 1.5143426656723022,
      "learning_rate": 0.00010422776148582599,
      "loss": 0.8483,
      "step": 8522
    },
    {
      "epoch": 1.3342204132748905,
      "grad_norm": 1.211293339729309,
      "learning_rate": 0.00010420332355816226,
      "loss": 0.7053,
      "step": 8523
    },
    {
      "epoch": 1.3343769567939887,
      "grad_norm": 2.1427128314971924,
      "learning_rate": 0.00010417888563049853,
      "loss": 0.7195,
      "step": 8524
    },
    {
      "epoch": 1.334533500313087,
      "grad_norm": 2.3068511486053467,
      "learning_rate": 0.0001041544477028348,
      "loss": 1.0497,
      "step": 8525
    },
    {
      "epoch": 1.3346900438321854,
      "grad_norm": 1.6837023496627808,
      "learning_rate": 0.00010413000977517105,
      "loss": 0.8406,
      "step": 8526
    },
    {
      "epoch": 1.3348465873512836,
      "grad_norm": 2.0553340911865234,
      "learning_rate": 0.00010410557184750733,
      "loss": 1.1097,
      "step": 8527
    },
    {
      "epoch": 1.335003130870382,
      "grad_norm": 2.150421619415283,
      "learning_rate": 0.0001040811339198436,
      "loss": 1.5506,
      "step": 8528
    },
    {
      "epoch": 1.3351596743894802,
      "grad_norm": 1.8495726585388184,
      "learning_rate": 0.00010405669599217984,
      "loss": 1.0281,
      "step": 8529
    },
    {
      "epoch": 1.3353162179085785,
      "grad_norm": 5.749996662139893,
      "learning_rate": 0.00010403225806451612,
      "loss": 1.2466,
      "step": 8530
    },
    {
      "epoch": 1.335472761427677,
      "grad_norm": 3.338156223297119,
      "learning_rate": 0.00010400782013685239,
      "loss": 0.9421,
      "step": 8531
    },
    {
      "epoch": 1.3356293049467753,
      "grad_norm": 1.7294635772705078,
      "learning_rate": 0.00010398338220918865,
      "loss": 0.594,
      "step": 8532
    },
    {
      "epoch": 1.3357858484658736,
      "grad_norm": 1.9416723251342773,
      "learning_rate": 0.00010395894428152493,
      "loss": 0.5398,
      "step": 8533
    },
    {
      "epoch": 1.3359423919849718,
      "grad_norm": 4.1962785720825195,
      "learning_rate": 0.00010393450635386118,
      "loss": 0.8419,
      "step": 8534
    },
    {
      "epoch": 1.33609893550407,
      "grad_norm": 1.9734654426574707,
      "learning_rate": 0.00010391006842619745,
      "loss": 0.3254,
      "step": 8535
    },
    {
      "epoch": 1.3362554790231684,
      "grad_norm": 2.3827269077301025,
      "learning_rate": 0.00010388563049853373,
      "loss": 0.8437,
      "step": 8536
    },
    {
      "epoch": 1.3364120225422669,
      "grad_norm": 1.408588171005249,
      "learning_rate": 0.00010386119257086998,
      "loss": 0.6051,
      "step": 8537
    },
    {
      "epoch": 1.336568566061365,
      "grad_norm": 2.1959314346313477,
      "learning_rate": 0.00010383675464320624,
      "loss": 0.855,
      "step": 8538
    },
    {
      "epoch": 1.3367251095804633,
      "grad_norm": 0.5544465780258179,
      "learning_rate": 0.00010381231671554252,
      "loss": 0.2188,
      "step": 8539
    },
    {
      "epoch": 1.3368816530995618,
      "grad_norm": 1.3485573530197144,
      "learning_rate": 0.00010378787878787878,
      "loss": 0.2731,
      "step": 8540
    },
    {
      "epoch": 1.33703819661866,
      "grad_norm": 0.4418382942676544,
      "learning_rate": 0.00010376344086021504,
      "loss": 0.2697,
      "step": 8541
    },
    {
      "epoch": 1.3371947401377584,
      "grad_norm": 0.8457216024398804,
      "learning_rate": 0.00010373900293255131,
      "loss": 0.3345,
      "step": 8542
    },
    {
      "epoch": 1.3373512836568566,
      "grad_norm": 0.6509329080581665,
      "learning_rate": 0.00010371456500488758,
      "loss": 0.2579,
      "step": 8543
    },
    {
      "epoch": 1.3375078271759548,
      "grad_norm": 0.7329632043838501,
      "learning_rate": 0.00010369012707722384,
      "loss": 0.3396,
      "step": 8544
    },
    {
      "epoch": 1.3376643706950533,
      "grad_norm": 0.5292937755584717,
      "learning_rate": 0.00010366568914956012,
      "loss": 0.1596,
      "step": 8545
    },
    {
      "epoch": 1.3378209142141515,
      "grad_norm": 0.620823085308075,
      "learning_rate": 0.00010364125122189637,
      "loss": 0.3351,
      "step": 8546
    },
    {
      "epoch": 1.33797745773325,
      "grad_norm": 0.5701507329940796,
      "learning_rate": 0.00010361681329423264,
      "loss": 0.2396,
      "step": 8547
    },
    {
      "epoch": 1.3381340012523482,
      "grad_norm": 0.6084917783737183,
      "learning_rate": 0.00010359237536656892,
      "loss": 0.2507,
      "step": 8548
    },
    {
      "epoch": 1.3382905447714464,
      "grad_norm": 0.8983578085899353,
      "learning_rate": 0.00010356793743890517,
      "loss": 0.305,
      "step": 8549
    },
    {
      "epoch": 1.3384470882905448,
      "grad_norm": 0.9507225751876831,
      "learning_rate": 0.00010354349951124143,
      "loss": 0.3077,
      "step": 8550
    },
    {
      "epoch": 1.338603631809643,
      "grad_norm": 1.0135985612869263,
      "learning_rate": 0.00010351906158357771,
      "loss": 0.2864,
      "step": 8551
    },
    {
      "epoch": 1.3387601753287415,
      "grad_norm": 1.2849130630493164,
      "learning_rate": 0.00010349462365591398,
      "loss": 0.2869,
      "step": 8552
    },
    {
      "epoch": 1.3389167188478397,
      "grad_norm": 1.036739706993103,
      "learning_rate": 0.00010347018572825023,
      "loss": 0.2438,
      "step": 8553
    },
    {
      "epoch": 1.339073262366938,
      "grad_norm": 0.8827303051948547,
      "learning_rate": 0.0001034457478005865,
      "loss": 0.4037,
      "step": 8554
    },
    {
      "epoch": 1.3392298058860364,
      "grad_norm": 0.9790602326393127,
      "learning_rate": 0.00010342130987292277,
      "loss": 0.4922,
      "step": 8555
    },
    {
      "epoch": 1.3393863494051346,
      "grad_norm": 1.958410382270813,
      "learning_rate": 0.00010339687194525903,
      "loss": 0.4532,
      "step": 8556
    },
    {
      "epoch": 1.339542892924233,
      "grad_norm": 1.1825295686721802,
      "learning_rate": 0.00010337243401759531,
      "loss": 0.3544,
      "step": 8557
    },
    {
      "epoch": 1.3396994364433312,
      "grad_norm": 0.9426168203353882,
      "learning_rate": 0.00010334799608993156,
      "loss": 0.1968,
      "step": 8558
    },
    {
      "epoch": 1.3398559799624294,
      "grad_norm": 0.9069138765335083,
      "learning_rate": 0.00010332355816226783,
      "loss": 0.3628,
      "step": 8559
    },
    {
      "epoch": 1.3400125234815279,
      "grad_norm": 1.00847589969635,
      "learning_rate": 0.00010329912023460411,
      "loss": 0.1958,
      "step": 8560
    },
    {
      "epoch": 1.340169067000626,
      "grad_norm": 2.047116994857788,
      "learning_rate": 0.00010327468230694036,
      "loss": 0.3903,
      "step": 8561
    },
    {
      "epoch": 1.3403256105197245,
      "grad_norm": 2.307471513748169,
      "learning_rate": 0.00010325024437927662,
      "loss": 0.6619,
      "step": 8562
    },
    {
      "epoch": 1.3404821540388228,
      "grad_norm": 2.9408271312713623,
      "learning_rate": 0.0001032258064516129,
      "loss": 0.7681,
      "step": 8563
    },
    {
      "epoch": 1.340638697557921,
      "grad_norm": 1.2898457050323486,
      "learning_rate": 0.00010320136852394917,
      "loss": 0.5818,
      "step": 8564
    },
    {
      "epoch": 1.3407952410770194,
      "grad_norm": 1.0917787551879883,
      "learning_rate": 0.00010317693059628542,
      "loss": 0.2375,
      "step": 8565
    },
    {
      "epoch": 1.3409517845961179,
      "grad_norm": 5.130285739898682,
      "learning_rate": 0.0001031524926686217,
      "loss": 0.9421,
      "step": 8566
    },
    {
      "epoch": 1.341108328115216,
      "grad_norm": 3.367882013320923,
      "learning_rate": 0.00010312805474095796,
      "loss": 0.9799,
      "step": 8567
    },
    {
      "epoch": 1.3412648716343143,
      "grad_norm": 1.5562324523925781,
      "learning_rate": 0.00010310361681329423,
      "loss": 0.4556,
      "step": 8568
    },
    {
      "epoch": 1.3414214151534125,
      "grad_norm": 2.622732400894165,
      "learning_rate": 0.00010307917888563048,
      "loss": 0.7766,
      "step": 8569
    },
    {
      "epoch": 1.341577958672511,
      "grad_norm": 6.2125630378723145,
      "learning_rate": 0.00010305474095796676,
      "loss": 0.9861,
      "step": 8570
    },
    {
      "epoch": 1.3417345021916094,
      "grad_norm": 2.0387191772460938,
      "learning_rate": 0.00010303030303030302,
      "loss": 1.042,
      "step": 8571
    },
    {
      "epoch": 1.3418910457107076,
      "grad_norm": 2.0245776176452637,
      "learning_rate": 0.00010300586510263929,
      "loss": 0.9906,
      "step": 8572
    },
    {
      "epoch": 1.3420475892298058,
      "grad_norm": 1.5084283351898193,
      "learning_rate": 0.00010298142717497555,
      "loss": 0.3743,
      "step": 8573
    },
    {
      "epoch": 1.3422041327489043,
      "grad_norm": 2.843535900115967,
      "learning_rate": 0.00010295698924731181,
      "loss": 0.8406,
      "step": 8574
    },
    {
      "epoch": 1.3423606762680025,
      "grad_norm": 3.4681601524353027,
      "learning_rate": 0.00010293255131964808,
      "loss": 0.788,
      "step": 8575
    },
    {
      "epoch": 1.342517219787101,
      "grad_norm": 2.2604522705078125,
      "learning_rate": 0.00010290811339198436,
      "loss": 0.7895,
      "step": 8576
    },
    {
      "epoch": 1.3426737633061991,
      "grad_norm": 3.683725357055664,
      "learning_rate": 0.00010288367546432061,
      "loss": 0.685,
      "step": 8577
    },
    {
      "epoch": 1.3428303068252974,
      "grad_norm": 2.207420825958252,
      "learning_rate": 0.00010285923753665687,
      "loss": 0.9342,
      "step": 8578
    },
    {
      "epoch": 1.3429868503443958,
      "grad_norm": 2.529273748397827,
      "learning_rate": 0.00010283479960899315,
      "loss": 1.3186,
      "step": 8579
    },
    {
      "epoch": 1.343143393863494,
      "grad_norm": 2.481981039047241,
      "learning_rate": 0.00010281036168132942,
      "loss": 1.0946,
      "step": 8580
    },
    {
      "epoch": 1.3432999373825925,
      "grad_norm": 1.3697443008422852,
      "learning_rate": 0.00010278592375366567,
      "loss": 0.6584,
      "step": 8581
    },
    {
      "epoch": 1.3434564809016907,
      "grad_norm": 2.0801591873168945,
      "learning_rate": 0.00010276148582600195,
      "loss": 0.5824,
      "step": 8582
    },
    {
      "epoch": 1.343613024420789,
      "grad_norm": 2.740255832672119,
      "learning_rate": 0.00010273704789833821,
      "loss": 0.7734,
      "step": 8583
    },
    {
      "epoch": 1.3437695679398873,
      "grad_norm": 2.066756010055542,
      "learning_rate": 0.00010271260997067448,
      "loss": 0.8381,
      "step": 8584
    },
    {
      "epoch": 1.3439261114589856,
      "grad_norm": 2.6519453525543213,
      "learning_rate": 0.00010268817204301074,
      "loss": 0.7212,
      "step": 8585
    },
    {
      "epoch": 1.344082654978084,
      "grad_norm": 2.464167833328247,
      "learning_rate": 0.000102663734115347,
      "loss": 0.7086,
      "step": 8586
    },
    {
      "epoch": 1.3442391984971822,
      "grad_norm": 2.8775999546051025,
      "learning_rate": 0.00010263929618768327,
      "loss": 0.8188,
      "step": 8587
    },
    {
      "epoch": 1.3443957420162804,
      "grad_norm": 2.8111610412597656,
      "learning_rate": 0.00010261485826001955,
      "loss": 0.6777,
      "step": 8588
    },
    {
      "epoch": 1.3445522855353789,
      "grad_norm": 0.4312553405761719,
      "learning_rate": 0.0001025904203323558,
      "loss": 0.1694,
      "step": 8589
    },
    {
      "epoch": 1.344708829054477,
      "grad_norm": 0.5284144282341003,
      "learning_rate": 0.00010256598240469207,
      "loss": 0.1899,
      "step": 8590
    },
    {
      "epoch": 1.3448653725735755,
      "grad_norm": 0.9777967929840088,
      "learning_rate": 0.00010254154447702834,
      "loss": 0.3319,
      "step": 8591
    },
    {
      "epoch": 1.3450219160926737,
      "grad_norm": 1.211936116218567,
      "learning_rate": 0.00010251710654936461,
      "loss": 0.1933,
      "step": 8592
    },
    {
      "epoch": 1.345178459611772,
      "grad_norm": 0.5879170894622803,
      "learning_rate": 0.00010249266862170086,
      "loss": 0.2897,
      "step": 8593
    },
    {
      "epoch": 1.3453350031308704,
      "grad_norm": 1.0071542263031006,
      "learning_rate": 0.00010246823069403714,
      "loss": 0.2997,
      "step": 8594
    },
    {
      "epoch": 1.3454915466499686,
      "grad_norm": 0.840803325176239,
      "learning_rate": 0.0001024437927663734,
      "loss": 0.2348,
      "step": 8595
    },
    {
      "epoch": 1.345648090169067,
      "grad_norm": 0.6389973163604736,
      "learning_rate": 0.00010241935483870965,
      "loss": 0.1762,
      "step": 8596
    },
    {
      "epoch": 1.3458046336881653,
      "grad_norm": 0.771003007888794,
      "learning_rate": 0.00010239491691104593,
      "loss": 0.278,
      "step": 8597
    },
    {
      "epoch": 1.3459611772072635,
      "grad_norm": 1.348070740699768,
      "learning_rate": 0.0001023704789833822,
      "loss": 0.3955,
      "step": 8598
    },
    {
      "epoch": 1.346117720726362,
      "grad_norm": 0.9808512926101685,
      "learning_rate": 0.00010234604105571846,
      "loss": 0.3047,
      "step": 8599
    },
    {
      "epoch": 1.3462742642454604,
      "grad_norm": 1.0971183776855469,
      "learning_rate": 0.00010232160312805474,
      "loss": 0.4669,
      "step": 8600
    },
    {
      "epoch": 1.3464308077645586,
      "grad_norm": 2.088040351867676,
      "learning_rate": 0.00010229716520039099,
      "loss": 0.5127,
      "step": 8601
    },
    {
      "epoch": 1.3465873512836568,
      "grad_norm": 0.7336698770523071,
      "learning_rate": 0.00010227272727272726,
      "loss": 0.3575,
      "step": 8602
    },
    {
      "epoch": 1.3467438948027552,
      "grad_norm": 0.7726986408233643,
      "learning_rate": 0.00010224828934506353,
      "loss": 0.2192,
      "step": 8603
    },
    {
      "epoch": 1.3469004383218535,
      "grad_norm": 1.5100910663604736,
      "learning_rate": 0.0001022238514173998,
      "loss": 0.3309,
      "step": 8604
    },
    {
      "epoch": 1.347056981840952,
      "grad_norm": 1.1917885541915894,
      "learning_rate": 0.00010219941348973605,
      "loss": 0.4132,
      "step": 8605
    },
    {
      "epoch": 1.3472135253600501,
      "grad_norm": 1.4236356019973755,
      "learning_rate": 0.00010217497556207233,
      "loss": 0.455,
      "step": 8606
    },
    {
      "epoch": 1.3473700688791483,
      "grad_norm": 1.214121699333191,
      "learning_rate": 0.0001021505376344086,
      "loss": 0.4558,
      "step": 8607
    },
    {
      "epoch": 1.3475266123982468,
      "grad_norm": 1.2391715049743652,
      "learning_rate": 0.00010212609970674485,
      "loss": 0.3848,
      "step": 8608
    },
    {
      "epoch": 1.347683155917345,
      "grad_norm": 1.3483924865722656,
      "learning_rate": 0.00010210166177908112,
      "loss": 0.5066,
      "step": 8609
    },
    {
      "epoch": 1.3478396994364434,
      "grad_norm": 1.2172237634658813,
      "learning_rate": 0.00010207722385141739,
      "loss": 0.373,
      "step": 8610
    },
    {
      "epoch": 1.3479962429555417,
      "grad_norm": 0.9454227685928345,
      "learning_rate": 0.00010205278592375365,
      "loss": 0.3206,
      "step": 8611
    },
    {
      "epoch": 1.3481527864746399,
      "grad_norm": 2.5474867820739746,
      "learning_rate": 0.00010202834799608993,
      "loss": 0.4893,
      "step": 8612
    },
    {
      "epoch": 1.3483093299937383,
      "grad_norm": 0.9093128442764282,
      "learning_rate": 0.00010200391006842618,
      "loss": 0.404,
      "step": 8613
    },
    {
      "epoch": 1.3484658735128365,
      "grad_norm": 2.522016763687134,
      "learning_rate": 0.00010197947214076245,
      "loss": 0.8356,
      "step": 8614
    },
    {
      "epoch": 1.348622417031935,
      "grad_norm": 1.7662311792373657,
      "learning_rate": 0.00010195503421309873,
      "loss": 0.395,
      "step": 8615
    },
    {
      "epoch": 1.3487789605510332,
      "grad_norm": 1.433812141418457,
      "learning_rate": 0.00010193059628543499,
      "loss": 0.5021,
      "step": 8616
    },
    {
      "epoch": 1.3489355040701314,
      "grad_norm": 1.4101475477218628,
      "learning_rate": 0.00010190615835777124,
      "loss": 0.5901,
      "step": 8617
    },
    {
      "epoch": 1.3490920475892298,
      "grad_norm": 4.594160556793213,
      "learning_rate": 0.00010188172043010752,
      "loss": 0.8687,
      "step": 8618
    },
    {
      "epoch": 1.349248591108328,
      "grad_norm": 1.1630065441131592,
      "learning_rate": 0.00010185728250244379,
      "loss": 0.4597,
      "step": 8619
    },
    {
      "epoch": 1.3494051346274265,
      "grad_norm": 2.0793771743774414,
      "learning_rate": 0.00010183284457478004,
      "loss": 0.6061,
      "step": 8620
    },
    {
      "epoch": 1.3495616781465247,
      "grad_norm": 2.358942985534668,
      "learning_rate": 0.00010180840664711631,
      "loss": 0.8025,
      "step": 8621
    },
    {
      "epoch": 1.349718221665623,
      "grad_norm": 2.1486661434173584,
      "learning_rate": 0.00010178396871945258,
      "loss": 0.6873,
      "step": 8622
    },
    {
      "epoch": 1.3498747651847214,
      "grad_norm": 4.260776519775391,
      "learning_rate": 0.00010175953079178884,
      "loss": 1.0033,
      "step": 8623
    },
    {
      "epoch": 1.3500313087038196,
      "grad_norm": 1.1536997556686401,
      "learning_rate": 0.00010173509286412512,
      "loss": 0.3625,
      "step": 8624
    },
    {
      "epoch": 1.350187852222918,
      "grad_norm": 2.115208625793457,
      "learning_rate": 0.00010171065493646137,
      "loss": 0.7172,
      "step": 8625
    },
    {
      "epoch": 1.3503443957420163,
      "grad_norm": 2.6683943271636963,
      "learning_rate": 0.00010168621700879764,
      "loss": 1.299,
      "step": 8626
    },
    {
      "epoch": 1.3505009392611145,
      "grad_norm": 1.838832139968872,
      "learning_rate": 0.00010166177908113392,
      "loss": 1.0832,
      "step": 8627
    },
    {
      "epoch": 1.350657482780213,
      "grad_norm": 3.921957492828369,
      "learning_rate": 0.00010163734115347018,
      "loss": 1.1096,
      "step": 8628
    },
    {
      "epoch": 1.3508140262993111,
      "grad_norm": 2.774120569229126,
      "learning_rate": 0.00010161290322580643,
      "loss": 1.1012,
      "step": 8629
    },
    {
      "epoch": 1.3509705698184096,
      "grad_norm": 2.269116163253784,
      "learning_rate": 0.00010158846529814271,
      "loss": 0.8941,
      "step": 8630
    },
    {
      "epoch": 1.3511271133375078,
      "grad_norm": 2.243638038635254,
      "learning_rate": 0.00010156402737047898,
      "loss": 1.0635,
      "step": 8631
    },
    {
      "epoch": 1.351283656856606,
      "grad_norm": 1.6823785305023193,
      "learning_rate": 0.00010153958944281523,
      "loss": 0.4038,
      "step": 8632
    },
    {
      "epoch": 1.3514402003757044,
      "grad_norm": 3.375223159790039,
      "learning_rate": 0.0001015151515151515,
      "loss": 0.9487,
      "step": 8633
    },
    {
      "epoch": 1.3515967438948029,
      "grad_norm": 2.887037754058838,
      "learning_rate": 0.00010149071358748777,
      "loss": 1.2298,
      "step": 8634
    },
    {
      "epoch": 1.351753287413901,
      "grad_norm": 2.0357108116149902,
      "learning_rate": 0.00010146627565982404,
      "loss": 0.7477,
      "step": 8635
    },
    {
      "epoch": 1.3519098309329993,
      "grad_norm": 2.5809435844421387,
      "learning_rate": 0.00010144183773216031,
      "loss": 1.3611,
      "step": 8636
    },
    {
      "epoch": 1.3520663744520978,
      "grad_norm": 1.5694135427474976,
      "learning_rate": 0.00010141739980449657,
      "loss": 0.5658,
      "step": 8637
    },
    {
      "epoch": 1.352222917971196,
      "grad_norm": 2.64839243888855,
      "learning_rate": 0.00010139296187683283,
      "loss": 0.9227,
      "step": 8638
    },
    {
      "epoch": 1.3523794614902944,
      "grad_norm": 0.7386291027069092,
      "learning_rate": 0.00010136852394916911,
      "loss": 0.2149,
      "step": 8639
    },
    {
      "epoch": 1.3525360050093926,
      "grad_norm": 0.43771645426750183,
      "learning_rate": 0.00010134408602150537,
      "loss": 0.2091,
      "step": 8640
    },
    {
      "epoch": 1.3526925485284909,
      "grad_norm": 0.7924376726150513,
      "learning_rate": 0.00010131964809384162,
      "loss": 0.3522,
      "step": 8641
    },
    {
      "epoch": 1.3528490920475893,
      "grad_norm": 0.5517721176147461,
      "learning_rate": 0.0001012952101661779,
      "loss": 0.264,
      "step": 8642
    },
    {
      "epoch": 1.3530056355666875,
      "grad_norm": 0.5062223672866821,
      "learning_rate": 0.00010127077223851417,
      "loss": 0.1477,
      "step": 8643
    },
    {
      "epoch": 1.353162179085786,
      "grad_norm": 0.5685814619064331,
      "learning_rate": 0.00010124633431085042,
      "loss": 0.3525,
      "step": 8644
    },
    {
      "epoch": 1.3533187226048842,
      "grad_norm": 0.5402782559394836,
      "learning_rate": 0.0001012218963831867,
      "loss": 0.2294,
      "step": 8645
    },
    {
      "epoch": 1.3534752661239824,
      "grad_norm": 0.5032253861427307,
      "learning_rate": 0.00010119745845552296,
      "loss": 0.2081,
      "step": 8646
    },
    {
      "epoch": 1.3536318096430808,
      "grad_norm": 1.4722563028335571,
      "learning_rate": 0.00010117302052785923,
      "loss": 0.3869,
      "step": 8647
    },
    {
      "epoch": 1.353788353162179,
      "grad_norm": 0.9200348854064941,
      "learning_rate": 0.0001011485826001955,
      "loss": 0.3021,
      "step": 8648
    },
    {
      "epoch": 1.3539448966812775,
      "grad_norm": 0.6962469816207886,
      "learning_rate": 0.00010112414467253176,
      "loss": 0.1811,
      "step": 8649
    },
    {
      "epoch": 1.3541014402003757,
      "grad_norm": 0.8069591522216797,
      "learning_rate": 0.00010109970674486802,
      "loss": 0.3023,
      "step": 8650
    },
    {
      "epoch": 1.354257983719474,
      "grad_norm": 0.669252336025238,
      "learning_rate": 0.0001010752688172043,
      "loss": 0.1662,
      "step": 8651
    },
    {
      "epoch": 1.3544145272385724,
      "grad_norm": 0.6303135752677917,
      "learning_rate": 0.00010105083088954055,
      "loss": 0.3415,
      "step": 8652
    },
    {
      "epoch": 1.3545710707576706,
      "grad_norm": 0.9292386770248413,
      "learning_rate": 0.00010102639296187682,
      "loss": 0.284,
      "step": 8653
    },
    {
      "epoch": 1.354727614276769,
      "grad_norm": 1.1860474348068237,
      "learning_rate": 0.0001010019550342131,
      "loss": 0.2997,
      "step": 8654
    },
    {
      "epoch": 1.3548841577958672,
      "grad_norm": 1.6328331232070923,
      "learning_rate": 0.00010097751710654936,
      "loss": 0.5064,
      "step": 8655
    },
    {
      "epoch": 1.3550407013149655,
      "grad_norm": 1.2533385753631592,
      "learning_rate": 0.00010095307917888561,
      "loss": 0.4723,
      "step": 8656
    },
    {
      "epoch": 1.355197244834064,
      "grad_norm": 1.2162647247314453,
      "learning_rate": 0.00010092864125122189,
      "loss": 0.3477,
      "step": 8657
    },
    {
      "epoch": 1.3553537883531621,
      "grad_norm": 1.2495274543762207,
      "learning_rate": 0.00010090420332355815,
      "loss": 0.2559,
      "step": 8658
    },
    {
      "epoch": 1.3555103318722606,
      "grad_norm": 1.11675226688385,
      "learning_rate": 0.00010087976539589442,
      "loss": 0.3749,
      "step": 8659
    },
    {
      "epoch": 1.3556668753913588,
      "grad_norm": 1.302672028541565,
      "learning_rate": 0.0001008553274682307,
      "loss": 0.2769,
      "step": 8660
    },
    {
      "epoch": 1.355823418910457,
      "grad_norm": 1.504712462425232,
      "learning_rate": 0.00010083088954056695,
      "loss": 0.7427,
      "step": 8661
    },
    {
      "epoch": 1.3559799624295554,
      "grad_norm": 1.3615977764129639,
      "learning_rate": 0.00010080645161290321,
      "loss": 0.824,
      "step": 8662
    },
    {
      "epoch": 1.3561365059486536,
      "grad_norm": 1.5800176858901978,
      "learning_rate": 0.00010078201368523949,
      "loss": 0.4904,
      "step": 8663
    },
    {
      "epoch": 1.356293049467752,
      "grad_norm": 1.6747968196868896,
      "learning_rate": 0.00010075757575757574,
      "loss": 0.4018,
      "step": 8664
    },
    {
      "epoch": 1.3564495929868503,
      "grad_norm": 1.6093438863754272,
      "learning_rate": 0.000100733137829912,
      "loss": 0.4234,
      "step": 8665
    },
    {
      "epoch": 1.3566061365059485,
      "grad_norm": 1.9799742698669434,
      "learning_rate": 0.00010070869990224828,
      "loss": 0.6212,
      "step": 8666
    },
    {
      "epoch": 1.356762680025047,
      "grad_norm": 1.8439103364944458,
      "learning_rate": 0.00010068426197458455,
      "loss": 0.5132,
      "step": 8667
    },
    {
      "epoch": 1.3569192235441454,
      "grad_norm": 2.6742753982543945,
      "learning_rate": 0.0001006598240469208,
      "loss": 0.6544,
      "step": 8668
    },
    {
      "epoch": 1.3570757670632436,
      "grad_norm": 2.446819543838501,
      "learning_rate": 0.00010063538611925708,
      "loss": 0.9793,
      "step": 8669
    },
    {
      "epoch": 1.3572323105823418,
      "grad_norm": 1.334675669670105,
      "learning_rate": 0.00010061094819159334,
      "loss": 0.5525,
      "step": 8670
    },
    {
      "epoch": 1.3573888541014403,
      "grad_norm": 1.5871644020080566,
      "learning_rate": 0.00010058651026392961,
      "loss": 0.4494,
      "step": 8671
    },
    {
      "epoch": 1.3575453976205385,
      "grad_norm": 1.5029113292694092,
      "learning_rate": 0.00010056207233626589,
      "loss": 0.8743,
      "step": 8672
    },
    {
      "epoch": 1.357701941139637,
      "grad_norm": 2.237409830093384,
      "learning_rate": 0.00010053763440860214,
      "loss": 0.806,
      "step": 8673
    },
    {
      "epoch": 1.3578584846587352,
      "grad_norm": 2.4241881370544434,
      "learning_rate": 0.0001005131964809384,
      "loss": 0.8528,
      "step": 8674
    },
    {
      "epoch": 1.3580150281778334,
      "grad_norm": 2.1619865894317627,
      "learning_rate": 0.00010048875855327468,
      "loss": 0.5835,
      "step": 8675
    },
    {
      "epoch": 1.3581715716969318,
      "grad_norm": 2.61045503616333,
      "learning_rate": 0.00010046432062561093,
      "loss": 1.6964,
      "step": 8676
    },
    {
      "epoch": 1.35832811521603,
      "grad_norm": 1.533729910850525,
      "learning_rate": 0.0001004398826979472,
      "loss": 0.8914,
      "step": 8677
    },
    {
      "epoch": 1.3584846587351285,
      "grad_norm": 2.500321388244629,
      "learning_rate": 0.00010041544477028348,
      "loss": 0.9521,
      "step": 8678
    },
    {
      "epoch": 1.3586412022542267,
      "grad_norm": 5.0488080978393555,
      "learning_rate": 0.00010039100684261974,
      "loss": 1.4087,
      "step": 8679
    },
    {
      "epoch": 1.358797745773325,
      "grad_norm": 2.802337408065796,
      "learning_rate": 0.00010036656891495599,
      "loss": 0.6997,
      "step": 8680
    },
    {
      "epoch": 1.3589542892924233,
      "grad_norm": 1.9060883522033691,
      "learning_rate": 0.00010034213098729227,
      "loss": 0.6674,
      "step": 8681
    },
    {
      "epoch": 1.3591108328115216,
      "grad_norm": 6.120393753051758,
      "learning_rate": 0.00010031769305962854,
      "loss": 1.335,
      "step": 8682
    },
    {
      "epoch": 1.35926737633062,
      "grad_norm": 1.4589465856552124,
      "learning_rate": 0.0001002932551319648,
      "loss": 0.6943,
      "step": 8683
    },
    {
      "epoch": 1.3594239198497182,
      "grad_norm": 3.765908718109131,
      "learning_rate": 0.00010026881720430108,
      "loss": 0.6131,
      "step": 8684
    },
    {
      "epoch": 1.3595804633688164,
      "grad_norm": 2.0184710025787354,
      "learning_rate": 0.00010024437927663733,
      "loss": 0.577,
      "step": 8685
    },
    {
      "epoch": 1.3597370068879149,
      "grad_norm": 1.7017927169799805,
      "learning_rate": 0.0001002199413489736,
      "loss": 0.6143,
      "step": 8686
    },
    {
      "epoch": 1.359893550407013,
      "grad_norm": 3.1961417198181152,
      "learning_rate": 0.00010019550342130987,
      "loss": 0.6772,
      "step": 8687
    },
    {
      "epoch": 1.3600500939261115,
      "grad_norm": 4.469157695770264,
      "learning_rate": 0.00010017106549364612,
      "loss": 1.4173,
      "step": 8688
    },
    {
      "epoch": 1.3602066374452098,
      "grad_norm": 0.6359954476356506,
      "learning_rate": 0.00010014662756598239,
      "loss": 0.316,
      "step": 8689
    },
    {
      "epoch": 1.360363180964308,
      "grad_norm": 1.053896427154541,
      "learning_rate": 0.00010012218963831867,
      "loss": 0.1854,
      "step": 8690
    },
    {
      "epoch": 1.3605197244834064,
      "grad_norm": 0.57310950756073,
      "learning_rate": 0.00010009775171065493,
      "loss": 0.2186,
      "step": 8691
    },
    {
      "epoch": 1.3606762680025046,
      "grad_norm": 0.5779302716255188,
      "learning_rate": 0.00010007331378299118,
      "loss": 0.2643,
      "step": 8692
    },
    {
      "epoch": 1.360832811521603,
      "grad_norm": 0.4520057439804077,
      "learning_rate": 0.00010004887585532746,
      "loss": 0.2138,
      "step": 8693
    },
    {
      "epoch": 1.3609893550407013,
      "grad_norm": 1.1857097148895264,
      "learning_rate": 0.00010002443792766373,
      "loss": 0.2805,
      "step": 8694
    },
    {
      "epoch": 1.3611458985597995,
      "grad_norm": 0.5992613434791565,
      "learning_rate": 9.999999999999999e-05,
      "loss": 0.3539,
      "step": 8695
    },
    {
      "epoch": 1.361302442078898,
      "grad_norm": 0.8540019392967224,
      "learning_rate": 9.997556207233626e-05,
      "loss": 0.2893,
      "step": 8696
    },
    {
      "epoch": 1.3614589855979962,
      "grad_norm": 0.5429804921150208,
      "learning_rate": 9.995112414467252e-05,
      "loss": 0.2457,
      "step": 8697
    },
    {
      "epoch": 1.3616155291170946,
      "grad_norm": 1.0506290197372437,
      "learning_rate": 9.992668621700879e-05,
      "loss": 0.2567,
      "step": 8698
    },
    {
      "epoch": 1.3617720726361928,
      "grad_norm": 0.7274707555770874,
      "learning_rate": 9.990224828934506e-05,
      "loss": 0.1966,
      "step": 8699
    },
    {
      "epoch": 1.361928616155291,
      "grad_norm": 1.6726579666137695,
      "learning_rate": 9.987781036168132e-05,
      "loss": 0.3843,
      "step": 8700
    },
    {
      "epoch": 1.3620851596743895,
      "grad_norm": 1.3097379207611084,
      "learning_rate": 9.985337243401758e-05,
      "loss": 0.304,
      "step": 8701
    },
    {
      "epoch": 1.362241703193488,
      "grad_norm": 1.2650363445281982,
      "learning_rate": 9.982893450635386e-05,
      "loss": 0.453,
      "step": 8702
    },
    {
      "epoch": 1.3623982467125861,
      "grad_norm": 0.9016621112823486,
      "learning_rate": 9.980449657869012e-05,
      "loss": 0.3309,
      "step": 8703
    },
    {
      "epoch": 1.3625547902316844,
      "grad_norm": 1.0004808902740479,
      "learning_rate": 9.978005865102637e-05,
      "loss": 0.3763,
      "step": 8704
    },
    {
      "epoch": 1.3627113337507828,
      "grad_norm": 1.2892861366271973,
      "learning_rate": 9.975562072336265e-05,
      "loss": 0.4991,
      "step": 8705
    },
    {
      "epoch": 1.362867877269881,
      "grad_norm": 1.6277682781219482,
      "learning_rate": 9.973118279569892e-05,
      "loss": 0.3803,
      "step": 8706
    },
    {
      "epoch": 1.3630244207889795,
      "grad_norm": 1.4325026273727417,
      "learning_rate": 9.970674486803518e-05,
      "loss": 0.4247,
      "step": 8707
    },
    {
      "epoch": 1.3631809643080777,
      "grad_norm": 1.4070262908935547,
      "learning_rate": 9.968230694037145e-05,
      "loss": 0.4286,
      "step": 8708
    },
    {
      "epoch": 1.3633375078271759,
      "grad_norm": 2.1237456798553467,
      "learning_rate": 9.965786901270771e-05,
      "loss": 0.65,
      "step": 8709
    },
    {
      "epoch": 1.3634940513462743,
      "grad_norm": 2.7738819122314453,
      "learning_rate": 9.963343108504398e-05,
      "loss": 0.858,
      "step": 8710
    },
    {
      "epoch": 1.3636505948653725,
      "grad_norm": 1.125198483467102,
      "learning_rate": 9.960899315738026e-05,
      "loss": 0.3916,
      "step": 8711
    },
    {
      "epoch": 1.363807138384471,
      "grad_norm": 1.3788397312164307,
      "learning_rate": 9.95845552297165e-05,
      "loss": 0.3602,
      "step": 8712
    },
    {
      "epoch": 1.3639636819035692,
      "grad_norm": 1.1802959442138672,
      "learning_rate": 9.956011730205277e-05,
      "loss": 0.3027,
      "step": 8713
    },
    {
      "epoch": 1.3641202254226674,
      "grad_norm": 1.0281505584716797,
      "learning_rate": 9.953567937438905e-05,
      "loss": 0.3809,
      "step": 8714
    },
    {
      "epoch": 1.3642767689417659,
      "grad_norm": 2.2165725231170654,
      "learning_rate": 9.951124144672531e-05,
      "loss": 0.8278,
      "step": 8715
    },
    {
      "epoch": 1.364433312460864,
      "grad_norm": 2.219921350479126,
      "learning_rate": 9.948680351906157e-05,
      "loss": 0.944,
      "step": 8716
    },
    {
      "epoch": 1.3645898559799625,
      "grad_norm": 2.5214898586273193,
      "learning_rate": 9.946236559139784e-05,
      "loss": 0.9238,
      "step": 8717
    },
    {
      "epoch": 1.3647463994990607,
      "grad_norm": 2.641808032989502,
      "learning_rate": 9.943792766373411e-05,
      "loss": 0.6352,
      "step": 8718
    },
    {
      "epoch": 1.364902943018159,
      "grad_norm": 1.4690858125686646,
      "learning_rate": 9.941348973607037e-05,
      "loss": 0.7535,
      "step": 8719
    },
    {
      "epoch": 1.3650594865372574,
      "grad_norm": 1.5977320671081543,
      "learning_rate": 9.938905180840664e-05,
      "loss": 0.6958,
      "step": 8720
    },
    {
      "epoch": 1.3652160300563556,
      "grad_norm": 1.7913562059402466,
      "learning_rate": 9.93646138807429e-05,
      "loss": 0.846,
      "step": 8721
    },
    {
      "epoch": 1.365372573575454,
      "grad_norm": 2.0344135761260986,
      "learning_rate": 9.934017595307917e-05,
      "loss": 0.7296,
      "step": 8722
    },
    {
      "epoch": 1.3655291170945523,
      "grad_norm": 3.8352715969085693,
      "learning_rate": 9.931573802541545e-05,
      "loss": 0.7046,
      "step": 8723
    },
    {
      "epoch": 1.3656856606136505,
      "grad_norm": 1.1829397678375244,
      "learning_rate": 9.92913000977517e-05,
      "loss": 0.5948,
      "step": 8724
    },
    {
      "epoch": 1.365842204132749,
      "grad_norm": 3.249732732772827,
      "learning_rate": 9.926686217008796e-05,
      "loss": 1.1317,
      "step": 8725
    },
    {
      "epoch": 1.3659987476518471,
      "grad_norm": 2.0440311431884766,
      "learning_rate": 9.924242424242424e-05,
      "loss": 0.8155,
      "step": 8726
    },
    {
      "epoch": 1.3661552911709456,
      "grad_norm": 2.4997398853302,
      "learning_rate": 9.92179863147605e-05,
      "loss": 0.8181,
      "step": 8727
    },
    {
      "epoch": 1.3663118346900438,
      "grad_norm": 4.196403503417969,
      "learning_rate": 9.919354838709676e-05,
      "loss": 0.573,
      "step": 8728
    },
    {
      "epoch": 1.366468378209142,
      "grad_norm": 1.9175196886062622,
      "learning_rate": 9.916911045943304e-05,
      "loss": 0.5787,
      "step": 8729
    },
    {
      "epoch": 1.3666249217282405,
      "grad_norm": 1.8270318508148193,
      "learning_rate": 9.91446725317693e-05,
      "loss": 0.7122,
      "step": 8730
    },
    {
      "epoch": 1.3667814652473387,
      "grad_norm": 2.314758539199829,
      "learning_rate": 9.912023460410556e-05,
      "loss": 0.9239,
      "step": 8731
    },
    {
      "epoch": 1.3669380087664371,
      "grad_norm": 1.4689182043075562,
      "learning_rate": 9.909579667644183e-05,
      "loss": 0.8547,
      "step": 8732
    },
    {
      "epoch": 1.3670945522855353,
      "grad_norm": 2.735502004623413,
      "learning_rate": 9.90713587487781e-05,
      "loss": 0.6393,
      "step": 8733
    },
    {
      "epoch": 1.3672510958046336,
      "grad_norm": 4.347554683685303,
      "learning_rate": 9.904692082111436e-05,
      "loss": 1.0392,
      "step": 8734
    },
    {
      "epoch": 1.367407639323732,
      "grad_norm": 1.6307176351547241,
      "learning_rate": 9.902248289345064e-05,
      "loss": 0.525,
      "step": 8735
    },
    {
      "epoch": 1.3675641828428304,
      "grad_norm": 2.497677803039551,
      "learning_rate": 9.899804496578689e-05,
      "loss": 0.6347,
      "step": 8736
    },
    {
      "epoch": 1.3677207263619287,
      "grad_norm": 2.2993502616882324,
      "learning_rate": 9.897360703812315e-05,
      "loss": 0.6741,
      "step": 8737
    },
    {
      "epoch": 1.3678772698810269,
      "grad_norm": 1.1915621757507324,
      "learning_rate": 9.894916911045943e-05,
      "loss": 0.4096,
      "step": 8738
    },
    {
      "epoch": 1.3680338134001253,
      "grad_norm": 0.5528977513313293,
      "learning_rate": 9.89247311827957e-05,
      "loss": 0.2614,
      "step": 8739
    },
    {
      "epoch": 1.3681903569192235,
      "grad_norm": 0.7073773145675659,
      "learning_rate": 9.890029325513195e-05,
      "loss": 0.2172,
      "step": 8740
    },
    {
      "epoch": 1.368346900438322,
      "grad_norm": 0.5945557355880737,
      "learning_rate": 9.887585532746823e-05,
      "loss": 0.2084,
      "step": 8741
    },
    {
      "epoch": 1.3685034439574202,
      "grad_norm": 0.41675809025764465,
      "learning_rate": 9.885141739980449e-05,
      "loss": 0.1236,
      "step": 8742
    },
    {
      "epoch": 1.3686599874765184,
      "grad_norm": 0.6224748492240906,
      "learning_rate": 9.882697947214076e-05,
      "loss": 0.2195,
      "step": 8743
    },
    {
      "epoch": 1.3688165309956168,
      "grad_norm": 1.0293275117874146,
      "learning_rate": 9.880254154447702e-05,
      "loss": 0.5121,
      "step": 8744
    },
    {
      "epoch": 1.368973074514715,
      "grad_norm": 0.8710348606109619,
      "learning_rate": 9.877810361681329e-05,
      "loss": 0.2933,
      "step": 8745
    },
    {
      "epoch": 1.3691296180338135,
      "grad_norm": 0.6916895508766174,
      "learning_rate": 9.875366568914955e-05,
      "loss": 0.1909,
      "step": 8746
    },
    {
      "epoch": 1.3692861615529117,
      "grad_norm": 0.8933185338973999,
      "learning_rate": 9.872922776148583e-05,
      "loss": 0.2403,
      "step": 8747
    },
    {
      "epoch": 1.36944270507201,
      "grad_norm": 1.0398600101470947,
      "learning_rate": 9.870478983382208e-05,
      "loss": 0.3148,
      "step": 8748
    },
    {
      "epoch": 1.3695992485911084,
      "grad_norm": 1.0829379558563232,
      "learning_rate": 9.868035190615834e-05,
      "loss": 0.3194,
      "step": 8749
    },
    {
      "epoch": 1.3697557921102066,
      "grad_norm": 0.7273138761520386,
      "learning_rate": 9.865591397849462e-05,
      "loss": 0.2394,
      "step": 8750
    },
    {
      "epoch": 1.369912335629305,
      "grad_norm": 2.8995516300201416,
      "learning_rate": 9.863147605083089e-05,
      "loss": 0.7665,
      "step": 8751
    },
    {
      "epoch": 1.3700688791484033,
      "grad_norm": 1.2653355598449707,
      "learning_rate": 9.860703812316714e-05,
      "loss": 0.3913,
      "step": 8752
    },
    {
      "epoch": 1.3702254226675015,
      "grad_norm": 4.484793663024902,
      "learning_rate": 9.858260019550342e-05,
      "loss": 0.4381,
      "step": 8753
    },
    {
      "epoch": 1.3703819661866,
      "grad_norm": 0.5917729735374451,
      "learning_rate": 9.855816226783968e-05,
      "loss": 0.1318,
      "step": 8754
    },
    {
      "epoch": 1.3705385097056981,
      "grad_norm": 1.4675835371017456,
      "learning_rate": 9.853372434017593e-05,
      "loss": 0.3232,
      "step": 8755
    },
    {
      "epoch": 1.3706950532247966,
      "grad_norm": 1.0259408950805664,
      "learning_rate": 9.850928641251221e-05,
      "loss": 0.3359,
      "step": 8756
    },
    {
      "epoch": 1.3708515967438948,
      "grad_norm": 1.2516374588012695,
      "learning_rate": 9.848484848484848e-05,
      "loss": 0.5625,
      "step": 8757
    },
    {
      "epoch": 1.371008140262993,
      "grad_norm": 3.212458372116089,
      "learning_rate": 9.846041055718474e-05,
      "loss": 0.378,
      "step": 8758
    },
    {
      "epoch": 1.3711646837820914,
      "grad_norm": 2.0374770164489746,
      "learning_rate": 9.843597262952102e-05,
      "loss": 0.4768,
      "step": 8759
    },
    {
      "epoch": 1.3713212273011897,
      "grad_norm": 2.4900288581848145,
      "learning_rate": 9.841153470185727e-05,
      "loss": 0.4226,
      "step": 8760
    },
    {
      "epoch": 1.371477770820288,
      "grad_norm": 1.8129522800445557,
      "learning_rate": 9.838709677419354e-05,
      "loss": 0.5211,
      "step": 8761
    },
    {
      "epoch": 1.3716343143393863,
      "grad_norm": 1.4478105306625366,
      "learning_rate": 9.836265884652981e-05,
      "loss": 0.7839,
      "step": 8762
    },
    {
      "epoch": 1.3717908578584845,
      "grad_norm": 1.799777626991272,
      "learning_rate": 9.833822091886608e-05,
      "loss": 0.4166,
      "step": 8763
    },
    {
      "epoch": 1.371947401377583,
      "grad_norm": 3.246655225753784,
      "learning_rate": 9.831378299120233e-05,
      "loss": 0.8867,
      "step": 8764
    },
    {
      "epoch": 1.3721039448966814,
      "grad_norm": 1.721030354499817,
      "learning_rate": 9.828934506353861e-05,
      "loss": 0.5577,
      "step": 8765
    },
    {
      "epoch": 1.3722604884157796,
      "grad_norm": 1.6854079961776733,
      "learning_rate": 9.826490713587487e-05,
      "loss": 0.573,
      "step": 8766
    },
    {
      "epoch": 1.3724170319348779,
      "grad_norm": 2.2403340339660645,
      "learning_rate": 9.824046920821112e-05,
      "loss": 0.7653,
      "step": 8767
    },
    {
      "epoch": 1.372573575453976,
      "grad_norm": 1.6862674951553345,
      "learning_rate": 9.82160312805474e-05,
      "loss": 0.5029,
      "step": 8768
    },
    {
      "epoch": 1.3727301189730745,
      "grad_norm": 2.492246627807617,
      "learning_rate": 9.819159335288367e-05,
      "loss": 0.772,
      "step": 8769
    },
    {
      "epoch": 1.372886662492173,
      "grad_norm": 3.606316328048706,
      "learning_rate": 9.816715542521993e-05,
      "loss": 1.0772,
      "step": 8770
    },
    {
      "epoch": 1.3730432060112712,
      "grad_norm": 1.919001579284668,
      "learning_rate": 9.814271749755621e-05,
      "loss": 0.3547,
      "step": 8771
    },
    {
      "epoch": 1.3731997495303694,
      "grad_norm": 1.8840322494506836,
      "learning_rate": 9.811827956989246e-05,
      "loss": 0.7453,
      "step": 8772
    },
    {
      "epoch": 1.3733562930494678,
      "grad_norm": 1.7198199033737183,
      "learning_rate": 9.809384164222873e-05,
      "loss": 0.516,
      "step": 8773
    },
    {
      "epoch": 1.373512836568566,
      "grad_norm": 3.6644818782806396,
      "learning_rate": 9.8069403714565e-05,
      "loss": 1.4235,
      "step": 8774
    },
    {
      "epoch": 1.3736693800876645,
      "grad_norm": 2.1743552684783936,
      "learning_rate": 9.804496578690127e-05,
      "loss": 1.0742,
      "step": 8775
    },
    {
      "epoch": 1.3738259236067627,
      "grad_norm": 4.740699291229248,
      "learning_rate": 9.802052785923752e-05,
      "loss": 1.6689,
      "step": 8776
    },
    {
      "epoch": 1.373982467125861,
      "grad_norm": 3.387509822845459,
      "learning_rate": 9.79960899315738e-05,
      "loss": 1.5587,
      "step": 8777
    },
    {
      "epoch": 1.3741390106449594,
      "grad_norm": 1.966965675354004,
      "learning_rate": 9.797165200391006e-05,
      "loss": 0.8358,
      "step": 8778
    },
    {
      "epoch": 1.3742955541640576,
      "grad_norm": 2.8340470790863037,
      "learning_rate": 9.794721407624632e-05,
      "loss": 1.0338,
      "step": 8779
    },
    {
      "epoch": 1.374452097683156,
      "grad_norm": 3.0556154251098633,
      "learning_rate": 9.79227761485826e-05,
      "loss": 1.4119,
      "step": 8780
    },
    {
      "epoch": 1.3746086412022542,
      "grad_norm": 3.300295829772949,
      "learning_rate": 9.789833822091886e-05,
      "loss": 1.1295,
      "step": 8781
    },
    {
      "epoch": 1.3747651847213525,
      "grad_norm": 2.788999557495117,
      "learning_rate": 9.787390029325512e-05,
      "loss": 0.8346,
      "step": 8782
    },
    {
      "epoch": 1.374921728240451,
      "grad_norm": 1.3150262832641602,
      "learning_rate": 9.78494623655914e-05,
      "loss": 0.6381,
      "step": 8783
    },
    {
      "epoch": 1.375078271759549,
      "grad_norm": 3.5385401248931885,
      "learning_rate": 9.782502443792765e-05,
      "loss": 0.5895,
      "step": 8784
    },
    {
      "epoch": 1.3752348152786475,
      "grad_norm": 1.0753779411315918,
      "learning_rate": 9.780058651026392e-05,
      "loss": 0.2589,
      "step": 8785
    },
    {
      "epoch": 1.3753913587977458,
      "grad_norm": 2.1634578704833984,
      "learning_rate": 9.77761485826002e-05,
      "loss": 0.4002,
      "step": 8786
    },
    {
      "epoch": 1.375547902316844,
      "grad_norm": 2.8193857669830322,
      "learning_rate": 9.775171065493646e-05,
      "loss": 1.104,
      "step": 8787
    },
    {
      "epoch": 1.3757044458359424,
      "grad_norm": 2.0298500061035156,
      "learning_rate": 9.772727272727271e-05,
      "loss": 0.8568,
      "step": 8788
    },
    {
      "epoch": 1.3758609893550406,
      "grad_norm": 0.7730556130409241,
      "learning_rate": 9.770283479960899e-05,
      "loss": 0.3035,
      "step": 8789
    },
    {
      "epoch": 1.376017532874139,
      "grad_norm": 0.873766303062439,
      "learning_rate": 9.767839687194526e-05,
      "loss": 0.2784,
      "step": 8790
    },
    {
      "epoch": 1.3761740763932373,
      "grad_norm": 0.5062800049781799,
      "learning_rate": 9.765395894428151e-05,
      "loss": 0.242,
      "step": 8791
    },
    {
      "epoch": 1.3763306199123355,
      "grad_norm": 0.6650285124778748,
      "learning_rate": 9.762952101661779e-05,
      "loss": 0.2121,
      "step": 8792
    },
    {
      "epoch": 1.376487163431434,
      "grad_norm": 0.42068031430244446,
      "learning_rate": 9.760508308895405e-05,
      "loss": 0.2907,
      "step": 8793
    },
    {
      "epoch": 1.3766437069505322,
      "grad_norm": 0.5812735557556152,
      "learning_rate": 9.758064516129031e-05,
      "loss": 0.1929,
      "step": 8794
    },
    {
      "epoch": 1.3768002504696306,
      "grad_norm": 0.9358962178230286,
      "learning_rate": 9.75562072336266e-05,
      "loss": 0.1967,
      "step": 8795
    },
    {
      "epoch": 1.3769567939887288,
      "grad_norm": 0.5094256401062012,
      "learning_rate": 9.753176930596284e-05,
      "loss": 0.1991,
      "step": 8796
    },
    {
      "epoch": 1.377113337507827,
      "grad_norm": 4.9580302238464355,
      "learning_rate": 9.750733137829911e-05,
      "loss": 0.9328,
      "step": 8797
    },
    {
      "epoch": 1.3772698810269255,
      "grad_norm": 0.7062557935714722,
      "learning_rate": 9.748289345063539e-05,
      "loss": 0.2952,
      "step": 8798
    },
    {
      "epoch": 1.377426424546024,
      "grad_norm": 1.7369906902313232,
      "learning_rate": 9.745845552297165e-05,
      "loss": 0.3241,
      "step": 8799
    },
    {
      "epoch": 1.3775829680651221,
      "grad_norm": 0.8181784152984619,
      "learning_rate": 9.74340175953079e-05,
      "loss": 0.3769,
      "step": 8800
    },
    {
      "epoch": 1.3777395115842204,
      "grad_norm": 0.6165406107902527,
      "learning_rate": 9.740957966764418e-05,
      "loss": 0.2526,
      "step": 8801
    },
    {
      "epoch": 1.3778960551033186,
      "grad_norm": 0.9449729919433594,
      "learning_rate": 9.738514173998045e-05,
      "loss": 0.4735,
      "step": 8802
    },
    {
      "epoch": 1.378052598622417,
      "grad_norm": 1.4198623895645142,
      "learning_rate": 9.73607038123167e-05,
      "loss": 0.3983,
      "step": 8803
    },
    {
      "epoch": 1.3782091421415155,
      "grad_norm": 1.2503142356872559,
      "learning_rate": 9.733626588465298e-05,
      "loss": 0.3594,
      "step": 8804
    },
    {
      "epoch": 1.3783656856606137,
      "grad_norm": 1.537210464477539,
      "learning_rate": 9.731182795698924e-05,
      "loss": 0.3658,
      "step": 8805
    },
    {
      "epoch": 1.378522229179712,
      "grad_norm": 0.8660850524902344,
      "learning_rate": 9.72873900293255e-05,
      "loss": 0.2598,
      "step": 8806
    },
    {
      "epoch": 1.3786787726988103,
      "grad_norm": 0.7644106149673462,
      "learning_rate": 9.726295210166178e-05,
      "loss": 0.3427,
      "step": 8807
    },
    {
      "epoch": 1.3788353162179086,
      "grad_norm": 2.4255013465881348,
      "learning_rate": 9.723851417399804e-05,
      "loss": 0.5187,
      "step": 8808
    },
    {
      "epoch": 1.378991859737007,
      "grad_norm": 0.9475809931755066,
      "learning_rate": 9.72140762463343e-05,
      "loss": 0.172,
      "step": 8809
    },
    {
      "epoch": 1.3791484032561052,
      "grad_norm": 1.3600800037384033,
      "learning_rate": 9.718963831867058e-05,
      "loss": 0.6254,
      "step": 8810
    },
    {
      "epoch": 1.3793049467752034,
      "grad_norm": 1.693412184715271,
      "learning_rate": 9.716520039100683e-05,
      "loss": 0.3988,
      "step": 8811
    },
    {
      "epoch": 1.3794614902943019,
      "grad_norm": 1.2597637176513672,
      "learning_rate": 9.71407624633431e-05,
      "loss": 0.3481,
      "step": 8812
    },
    {
      "epoch": 1.3796180338134,
      "grad_norm": 1.089922308921814,
      "learning_rate": 9.711632453567937e-05,
      "loss": 0.3203,
      "step": 8813
    },
    {
      "epoch": 1.3797745773324985,
      "grad_norm": 1.767945647239685,
      "learning_rate": 9.709188660801564e-05,
      "loss": 0.5356,
      "step": 8814
    },
    {
      "epoch": 1.3799311208515967,
      "grad_norm": 1.7489302158355713,
      "learning_rate": 9.706744868035189e-05,
      "loss": 0.8449,
      "step": 8815
    },
    {
      "epoch": 1.380087664370695,
      "grad_norm": 2.965064525604248,
      "learning_rate": 9.704301075268817e-05,
      "loss": 0.4853,
      "step": 8816
    },
    {
      "epoch": 1.3802442078897934,
      "grad_norm": 3.0038321018218994,
      "learning_rate": 9.701857282502443e-05,
      "loss": 0.993,
      "step": 8817
    },
    {
      "epoch": 1.3804007514088916,
      "grad_norm": 3.7745330333709717,
      "learning_rate": 9.69941348973607e-05,
      "loss": 0.7505,
      "step": 8818
    },
    {
      "epoch": 1.38055729492799,
      "grad_norm": 1.8408567905426025,
      "learning_rate": 9.696969696969698e-05,
      "loss": 0.5765,
      "step": 8819
    },
    {
      "epoch": 1.3807138384470883,
      "grad_norm": 1.4488403797149658,
      "learning_rate": 9.694525904203323e-05,
      "loss": 0.7841,
      "step": 8820
    },
    {
      "epoch": 1.3808703819661865,
      "grad_norm": 2.437166690826416,
      "learning_rate": 9.692082111436949e-05,
      "loss": 0.9474,
      "step": 8821
    },
    {
      "epoch": 1.381026925485285,
      "grad_norm": 1.6625169515609741,
      "learning_rate": 9.689638318670577e-05,
      "loss": 0.6246,
      "step": 8822
    },
    {
      "epoch": 1.3811834690043832,
      "grad_norm": 2.3555285930633545,
      "learning_rate": 9.687194525904202e-05,
      "loss": 1.0428,
      "step": 8823
    },
    {
      "epoch": 1.3813400125234816,
      "grad_norm": 2.821054220199585,
      "learning_rate": 9.684750733137829e-05,
      "loss": 0.5075,
      "step": 8824
    },
    {
      "epoch": 1.3814965560425798,
      "grad_norm": 1.759642243385315,
      "learning_rate": 9.682306940371455e-05,
      "loss": 0.6551,
      "step": 8825
    },
    {
      "epoch": 1.381653099561678,
      "grad_norm": 1.687735676765442,
      "learning_rate": 9.679863147605083e-05,
      "loss": 1.2204,
      "step": 8826
    },
    {
      "epoch": 1.3818096430807765,
      "grad_norm": 1.8050616979599,
      "learning_rate": 9.677419354838708e-05,
      "loss": 1.0335,
      "step": 8827
    },
    {
      "epoch": 1.3819661865998747,
      "grad_norm": 7.936594009399414,
      "learning_rate": 9.674975562072335e-05,
      "loss": 0.9699,
      "step": 8828
    },
    {
      "epoch": 1.3821227301189731,
      "grad_norm": 2.0747787952423096,
      "learning_rate": 9.672531769305962e-05,
      "loss": 0.9377,
      "step": 8829
    },
    {
      "epoch": 1.3822792736380713,
      "grad_norm": 2.84826922416687,
      "learning_rate": 9.670087976539589e-05,
      "loss": 1.1424,
      "step": 8830
    },
    {
      "epoch": 1.3824358171571696,
      "grad_norm": 1.3028104305267334,
      "learning_rate": 9.667644183773214e-05,
      "loss": 0.5916,
      "step": 8831
    },
    {
      "epoch": 1.382592360676268,
      "grad_norm": 2.0030343532562256,
      "learning_rate": 9.665200391006842e-05,
      "loss": 1.2034,
      "step": 8832
    },
    {
      "epoch": 1.3827489041953664,
      "grad_norm": 3.1390602588653564,
      "learning_rate": 9.662756598240468e-05,
      "loss": 0.9957,
      "step": 8833
    },
    {
      "epoch": 1.3829054477144647,
      "grad_norm": 1.9023643732070923,
      "learning_rate": 9.660312805474095e-05,
      "loss": 0.4849,
      "step": 8834
    },
    {
      "epoch": 1.3830619912335629,
      "grad_norm": 5.415816307067871,
      "learning_rate": 9.657869012707721e-05,
      "loss": 1.2315,
      "step": 8835
    },
    {
      "epoch": 1.3832185347526613,
      "grad_norm": 2.7929747104644775,
      "learning_rate": 9.655425219941348e-05,
      "loss": 1.195,
      "step": 8836
    },
    {
      "epoch": 1.3833750782717595,
      "grad_norm": 2.5922513008117676,
      "learning_rate": 9.652981427174974e-05,
      "loss": 0.97,
      "step": 8837
    },
    {
      "epoch": 1.383531621790858,
      "grad_norm": 2.4513044357299805,
      "learning_rate": 9.650537634408602e-05,
      "loss": 0.9625,
      "step": 8838
    },
    {
      "epoch": 1.3836881653099562,
      "grad_norm": 0.5424944758415222,
      "learning_rate": 9.648093841642227e-05,
      "loss": 0.2183,
      "step": 8839
    },
    {
      "epoch": 1.3838447088290544,
      "grad_norm": 1.09577214717865,
      "learning_rate": 9.645650048875854e-05,
      "loss": 0.2732,
      "step": 8840
    },
    {
      "epoch": 1.3840012523481529,
      "grad_norm": 0.6740018129348755,
      "learning_rate": 9.643206256109481e-05,
      "loss": 0.1935,
      "step": 8841
    },
    {
      "epoch": 1.384157795867251,
      "grad_norm": 0.7105699181556702,
      "learning_rate": 9.640762463343108e-05,
      "loss": 0.2475,
      "step": 8842
    },
    {
      "epoch": 1.3843143393863495,
      "grad_norm": 0.4238963723182678,
      "learning_rate": 9.638318670576733e-05,
      "loss": 0.273,
      "step": 8843
    },
    {
      "epoch": 1.3844708829054477,
      "grad_norm": 0.680479884147644,
      "learning_rate": 9.635874877810361e-05,
      "loss": 0.2303,
      "step": 8844
    },
    {
      "epoch": 1.384627426424546,
      "grad_norm": 0.9009299874305725,
      "learning_rate": 9.633431085043987e-05,
      "loss": 0.3086,
      "step": 8845
    },
    {
      "epoch": 1.3847839699436444,
      "grad_norm": 0.7667379379272461,
      "learning_rate": 9.630987292277614e-05,
      "loss": 0.2803,
      "step": 8846
    },
    {
      "epoch": 1.3849405134627426,
      "grad_norm": 0.854888379573822,
      "learning_rate": 9.62854349951124e-05,
      "loss": 0.3391,
      "step": 8847
    },
    {
      "epoch": 1.385097056981841,
      "grad_norm": 0.6643977761268616,
      "learning_rate": 9.626099706744867e-05,
      "loss": 0.2158,
      "step": 8848
    },
    {
      "epoch": 1.3852536005009393,
      "grad_norm": 0.7794937491416931,
      "learning_rate": 9.623655913978493e-05,
      "loss": 0.257,
      "step": 8849
    },
    {
      "epoch": 1.3854101440200375,
      "grad_norm": 0.7044182419776917,
      "learning_rate": 9.621212121212121e-05,
      "loss": 0.2123,
      "step": 8850
    },
    {
      "epoch": 1.385566687539136,
      "grad_norm": 1.4797452688217163,
      "learning_rate": 9.618768328445746e-05,
      "loss": 0.5418,
      "step": 8851
    },
    {
      "epoch": 1.3857232310582341,
      "grad_norm": 0.6202751398086548,
      "learning_rate": 9.616324535679373e-05,
      "loss": 0.2076,
      "step": 8852
    },
    {
      "epoch": 1.3858797745773326,
      "grad_norm": 0.7685756087303162,
      "learning_rate": 9.613880742913e-05,
      "loss": 0.2865,
      "step": 8853
    },
    {
      "epoch": 1.3860363180964308,
      "grad_norm": 0.6632737517356873,
      "learning_rate": 9.611436950146627e-05,
      "loss": 0.3327,
      "step": 8854
    },
    {
      "epoch": 1.386192861615529,
      "grad_norm": 1.1949931383132935,
      "learning_rate": 9.608993157380252e-05,
      "loss": 0.4345,
      "step": 8855
    },
    {
      "epoch": 1.3863494051346275,
      "grad_norm": 0.9757672548294067,
      "learning_rate": 9.60654936461388e-05,
      "loss": 0.477,
      "step": 8856
    },
    {
      "epoch": 1.3865059486537257,
      "grad_norm": 1.91054105758667,
      "learning_rate": 9.604105571847507e-05,
      "loss": 0.5074,
      "step": 8857
    },
    {
      "epoch": 1.3866624921728241,
      "grad_norm": 1.4399696588516235,
      "learning_rate": 9.601661779081133e-05,
      "loss": 0.3122,
      "step": 8858
    },
    {
      "epoch": 1.3868190356919223,
      "grad_norm": 2.84247088432312,
      "learning_rate": 9.59921798631476e-05,
      "loss": 0.7512,
      "step": 8859
    },
    {
      "epoch": 1.3869755792110205,
      "grad_norm": 1.674396276473999,
      "learning_rate": 9.596774193548386e-05,
      "loss": 0.4991,
      "step": 8860
    },
    {
      "epoch": 1.387132122730119,
      "grad_norm": 2.1071906089782715,
      "learning_rate": 9.594330400782012e-05,
      "loss": 0.8064,
      "step": 8861
    },
    {
      "epoch": 1.3872886662492172,
      "grad_norm": 1.3319578170776367,
      "learning_rate": 9.59188660801564e-05,
      "loss": 0.3922,
      "step": 8862
    },
    {
      "epoch": 1.3874452097683156,
      "grad_norm": 1.4761604070663452,
      "learning_rate": 9.589442815249265e-05,
      "loss": 0.5648,
      "step": 8863
    },
    {
      "epoch": 1.3876017532874139,
      "grad_norm": 2.2004549503326416,
      "learning_rate": 9.586999022482892e-05,
      "loss": 0.5432,
      "step": 8864
    },
    {
      "epoch": 1.387758296806512,
      "grad_norm": 1.39911949634552,
      "learning_rate": 9.58455522971652e-05,
      "loss": 0.576,
      "step": 8865
    },
    {
      "epoch": 1.3879148403256105,
      "grad_norm": 1.3971905708312988,
      "learning_rate": 9.582111436950146e-05,
      "loss": 0.6516,
      "step": 8866
    },
    {
      "epoch": 1.388071383844709,
      "grad_norm": 1.2392330169677734,
      "learning_rate": 9.579667644183771e-05,
      "loss": 0.457,
      "step": 8867
    },
    {
      "epoch": 1.3882279273638072,
      "grad_norm": 3.725346088409424,
      "learning_rate": 9.577223851417399e-05,
      "loss": 0.6318,
      "step": 8868
    },
    {
      "epoch": 1.3883844708829054,
      "grad_norm": 4.512824058532715,
      "learning_rate": 9.574780058651026e-05,
      "loss": 0.9208,
      "step": 8869
    },
    {
      "epoch": 1.3885410144020038,
      "grad_norm": 2.2564964294433594,
      "learning_rate": 9.572336265884651e-05,
      "loss": 0.852,
      "step": 8870
    },
    {
      "epoch": 1.388697557921102,
      "grad_norm": 1.9278607368469238,
      "learning_rate": 9.569892473118279e-05,
      "loss": 0.7657,
      "step": 8871
    },
    {
      "epoch": 1.3888541014402005,
      "grad_norm": 2.015615940093994,
      "learning_rate": 9.567448680351905e-05,
      "loss": 0.4866,
      "step": 8872
    },
    {
      "epoch": 1.3890106449592987,
      "grad_norm": 2.1916158199310303,
      "learning_rate": 9.565004887585532e-05,
      "loss": 0.7778,
      "step": 8873
    },
    {
      "epoch": 1.389167188478397,
      "grad_norm": 3.4717800617218018,
      "learning_rate": 9.56256109481916e-05,
      "loss": 0.7985,
      "step": 8874
    },
    {
      "epoch": 1.3893237319974954,
      "grad_norm": 1.8821927309036255,
      "learning_rate": 9.560117302052785e-05,
      "loss": 0.7681,
      "step": 8875
    },
    {
      "epoch": 1.3894802755165936,
      "grad_norm": 1.9250900745391846,
      "learning_rate": 9.557673509286411e-05,
      "loss": 0.9574,
      "step": 8876
    },
    {
      "epoch": 1.389636819035692,
      "grad_norm": 2.0036773681640625,
      "learning_rate": 9.555229716520039e-05,
      "loss": 0.8076,
      "step": 8877
    },
    {
      "epoch": 1.3897933625547902,
      "grad_norm": 1.968286395072937,
      "learning_rate": 9.552785923753665e-05,
      "loss": 0.7549,
      "step": 8878
    },
    {
      "epoch": 1.3899499060738885,
      "grad_norm": 2.4910616874694824,
      "learning_rate": 9.55034213098729e-05,
      "loss": 0.7469,
      "step": 8879
    },
    {
      "epoch": 1.390106449592987,
      "grad_norm": 2.2749216556549072,
      "learning_rate": 9.547898338220918e-05,
      "loss": 1.0438,
      "step": 8880
    },
    {
      "epoch": 1.3902629931120851,
      "grad_norm": 1.8309866189956665,
      "learning_rate": 9.545454545454545e-05,
      "loss": 0.8261,
      "step": 8881
    },
    {
      "epoch": 1.3904195366311836,
      "grad_norm": 3.1002039909362793,
      "learning_rate": 9.54301075268817e-05,
      "loss": 0.6295,
      "step": 8882
    },
    {
      "epoch": 1.3905760801502818,
      "grad_norm": 3.578477144241333,
      "learning_rate": 9.540566959921798e-05,
      "loss": 1.5787,
      "step": 8883
    },
    {
      "epoch": 1.39073262366938,
      "grad_norm": 0.985260009765625,
      "learning_rate": 9.538123167155424e-05,
      "loss": 0.5352,
      "step": 8884
    },
    {
      "epoch": 1.3908891671884784,
      "grad_norm": 1.0070950984954834,
      "learning_rate": 9.53567937438905e-05,
      "loss": 0.5997,
      "step": 8885
    },
    {
      "epoch": 1.3910457107075767,
      "grad_norm": 3.7554800510406494,
      "learning_rate": 9.533235581622678e-05,
      "loss": 0.7481,
      "step": 8886
    },
    {
      "epoch": 1.391202254226675,
      "grad_norm": 1.8878397941589355,
      "learning_rate": 9.530791788856304e-05,
      "loss": 1.1695,
      "step": 8887
    },
    {
      "epoch": 1.3913587977457733,
      "grad_norm": 3.70890474319458,
      "learning_rate": 9.52834799608993e-05,
      "loss": 1.3039,
      "step": 8888
    },
    {
      "epoch": 1.3915153412648715,
      "grad_norm": 0.3560074269771576,
      "learning_rate": 9.525904203323558e-05,
      "loss": 0.1757,
      "step": 8889
    },
    {
      "epoch": 1.39167188478397,
      "grad_norm": 0.8983970880508423,
      "learning_rate": 9.523460410557184e-05,
      "loss": 0.2835,
      "step": 8890
    },
    {
      "epoch": 1.3918284283030682,
      "grad_norm": 0.8294395804405212,
      "learning_rate": 9.52101661779081e-05,
      "loss": 0.1884,
      "step": 8891
    },
    {
      "epoch": 1.3919849718221666,
      "grad_norm": 1.0664957761764526,
      "learning_rate": 9.518572825024437e-05,
      "loss": 0.2084,
      "step": 8892
    },
    {
      "epoch": 1.3921415153412648,
      "grad_norm": 1.1653019189834595,
      "learning_rate": 9.516129032258064e-05,
      "loss": 0.2246,
      "step": 8893
    },
    {
      "epoch": 1.392298058860363,
      "grad_norm": 0.8518573641777039,
      "learning_rate": 9.513685239491689e-05,
      "loss": 0.6462,
      "step": 8894
    },
    {
      "epoch": 1.3924546023794615,
      "grad_norm": 0.7155981063842773,
      "learning_rate": 9.511241446725317e-05,
      "loss": 0.278,
      "step": 8895
    },
    {
      "epoch": 1.3926111458985597,
      "grad_norm": 0.8205424547195435,
      "learning_rate": 9.508797653958943e-05,
      "loss": 0.2204,
      "step": 8896
    },
    {
      "epoch": 1.3927676894176582,
      "grad_norm": 0.9535171389579773,
      "learning_rate": 9.50635386119257e-05,
      "loss": 0.1775,
      "step": 8897
    },
    {
      "epoch": 1.3929242329367564,
      "grad_norm": 0.6753984689712524,
      "learning_rate": 9.503910068426198e-05,
      "loss": 0.2603,
      "step": 8898
    },
    {
      "epoch": 1.3930807764558546,
      "grad_norm": 0.8725956082344055,
      "learning_rate": 9.501466275659823e-05,
      "loss": 0.3896,
      "step": 8899
    },
    {
      "epoch": 1.393237319974953,
      "grad_norm": 0.7949926257133484,
      "learning_rate": 9.499022482893449e-05,
      "loss": 0.3452,
      "step": 8900
    },
    {
      "epoch": 1.3933938634940515,
      "grad_norm": 5.083495140075684,
      "learning_rate": 9.496578690127077e-05,
      "loss": 0.6093,
      "step": 8901
    },
    {
      "epoch": 1.3935504070131497,
      "grad_norm": 0.8620098233222961,
      "learning_rate": 9.494134897360704e-05,
      "loss": 0.1909,
      "step": 8902
    },
    {
      "epoch": 1.393706950532248,
      "grad_norm": 1.0833512544631958,
      "learning_rate": 9.491691104594329e-05,
      "loss": 0.4185,
      "step": 8903
    },
    {
      "epoch": 1.3938634940513464,
      "grad_norm": 1.1511584520339966,
      "learning_rate": 9.489247311827956e-05,
      "loss": 0.3882,
      "step": 8904
    },
    {
      "epoch": 1.3940200375704446,
      "grad_norm": 1.5453732013702393,
      "learning_rate": 9.486803519061583e-05,
      "loss": 0.4503,
      "step": 8905
    },
    {
      "epoch": 1.394176581089543,
      "grad_norm": 1.2637474536895752,
      "learning_rate": 9.484359726295208e-05,
      "loss": 0.422,
      "step": 8906
    },
    {
      "epoch": 1.3943331246086412,
      "grad_norm": 1.7331559658050537,
      "learning_rate": 9.481915933528836e-05,
      "loss": 0.349,
      "step": 8907
    },
    {
      "epoch": 1.3944896681277394,
      "grad_norm": 1.1923069953918457,
      "learning_rate": 9.479472140762462e-05,
      "loss": 0.3536,
      "step": 8908
    },
    {
      "epoch": 1.3946462116468379,
      "grad_norm": 3.23207950592041,
      "learning_rate": 9.477028347996089e-05,
      "loss": 0.3895,
      "step": 8909
    },
    {
      "epoch": 1.394802755165936,
      "grad_norm": 1.1913764476776123,
      "learning_rate": 9.474584555229717e-05,
      "loss": 0.6068,
      "step": 8910
    },
    {
      "epoch": 1.3949592986850345,
      "grad_norm": 1.4866349697113037,
      "learning_rate": 9.472140762463342e-05,
      "loss": 0.3515,
      "step": 8911
    },
    {
      "epoch": 1.3951158422041328,
      "grad_norm": 1.2634825706481934,
      "learning_rate": 9.469696969696968e-05,
      "loss": 0.5826,
      "step": 8912
    },
    {
      "epoch": 1.395272385723231,
      "grad_norm": 1.2857826948165894,
      "learning_rate": 9.467253176930596e-05,
      "loss": 0.54,
      "step": 8913
    },
    {
      "epoch": 1.3954289292423294,
      "grad_norm": 2.951294422149658,
      "learning_rate": 9.464809384164221e-05,
      "loss": 0.4376,
      "step": 8914
    },
    {
      "epoch": 1.3955854727614276,
      "grad_norm": 2.0184364318847656,
      "learning_rate": 9.462365591397848e-05,
      "loss": 0.5928,
      "step": 8915
    },
    {
      "epoch": 1.395742016280526,
      "grad_norm": 4.42757511138916,
      "learning_rate": 9.459921798631476e-05,
      "loss": 0.8246,
      "step": 8916
    },
    {
      "epoch": 1.3958985597996243,
      "grad_norm": 1.6247820854187012,
      "learning_rate": 9.457478005865102e-05,
      "loss": 0.658,
      "step": 8917
    },
    {
      "epoch": 1.3960551033187225,
      "grad_norm": 1.0989845991134644,
      "learning_rate": 9.455034213098727e-05,
      "loss": 0.5058,
      "step": 8918
    },
    {
      "epoch": 1.396211646837821,
      "grad_norm": 3.0606327056884766,
      "learning_rate": 9.452590420332355e-05,
      "loss": 0.722,
      "step": 8919
    },
    {
      "epoch": 1.3963681903569192,
      "grad_norm": 2.1190545558929443,
      "learning_rate": 9.450146627565982e-05,
      "loss": 0.8267,
      "step": 8920
    },
    {
      "epoch": 1.3965247338760176,
      "grad_norm": 2.202465057373047,
      "learning_rate": 9.447702834799608e-05,
      "loss": 0.8312,
      "step": 8921
    },
    {
      "epoch": 1.3966812773951158,
      "grad_norm": 1.6959621906280518,
      "learning_rate": 9.445259042033236e-05,
      "loss": 0.8328,
      "step": 8922
    },
    {
      "epoch": 1.396837820914214,
      "grad_norm": 2.8653564453125,
      "learning_rate": 9.442815249266861e-05,
      "loss": 0.6202,
      "step": 8923
    },
    {
      "epoch": 1.3969943644333125,
      "grad_norm": 1.8655325174331665,
      "learning_rate": 9.440371456500487e-05,
      "loss": 0.7007,
      "step": 8924
    },
    {
      "epoch": 1.3971509079524107,
      "grad_norm": 1.3610215187072754,
      "learning_rate": 9.437927663734115e-05,
      "loss": 0.42,
      "step": 8925
    },
    {
      "epoch": 1.3973074514715091,
      "grad_norm": 2.2759883403778076,
      "learning_rate": 9.43548387096774e-05,
      "loss": 0.9368,
      "step": 8926
    },
    {
      "epoch": 1.3974639949906074,
      "grad_norm": 2.666337728500366,
      "learning_rate": 9.433040078201367e-05,
      "loss": 1.1112,
      "step": 8927
    },
    {
      "epoch": 1.3976205385097056,
      "grad_norm": 3.074618101119995,
      "learning_rate": 9.430596285434995e-05,
      "loss": 1.377,
      "step": 8928
    },
    {
      "epoch": 1.397777082028804,
      "grad_norm": 1.411201000213623,
      "learning_rate": 9.428152492668621e-05,
      "loss": 0.9021,
      "step": 8929
    },
    {
      "epoch": 1.3979336255479022,
      "grad_norm": 2.165524482727051,
      "learning_rate": 9.425708699902246e-05,
      "loss": 1.0022,
      "step": 8930
    },
    {
      "epoch": 1.3980901690670007,
      "grad_norm": 1.9300085306167603,
      "learning_rate": 9.423264907135874e-05,
      "loss": 1.6909,
      "step": 8931
    },
    {
      "epoch": 1.398246712586099,
      "grad_norm": 2.606879711151123,
      "learning_rate": 9.4208211143695e-05,
      "loss": 0.7395,
      "step": 8932
    },
    {
      "epoch": 1.3984032561051971,
      "grad_norm": 2.456814765930176,
      "learning_rate": 9.418377321603127e-05,
      "loss": 1.3278,
      "step": 8933
    },
    {
      "epoch": 1.3985597996242956,
      "grad_norm": 1.9723178148269653,
      "learning_rate": 9.415933528836755e-05,
      "loss": 0.3545,
      "step": 8934
    },
    {
      "epoch": 1.398716343143394,
      "grad_norm": 1.6386150121688843,
      "learning_rate": 9.41348973607038e-05,
      "loss": 0.4407,
      "step": 8935
    },
    {
      "epoch": 1.3988728866624922,
      "grad_norm": 1.4594956636428833,
      "learning_rate": 9.411045943304007e-05,
      "loss": 0.6822,
      "step": 8936
    },
    {
      "epoch": 1.3990294301815904,
      "grad_norm": 2.2683069705963135,
      "learning_rate": 9.408602150537634e-05,
      "loss": 0.9911,
      "step": 8937
    },
    {
      "epoch": 1.3991859737006889,
      "grad_norm": 3.4247801303863525,
      "learning_rate": 9.40615835777126e-05,
      "loss": 1.4219,
      "step": 8938
    },
    {
      "epoch": 1.399342517219787,
      "grad_norm": 0.4289674758911133,
      "learning_rate": 9.403714565004886e-05,
      "loss": 0.2495,
      "step": 8939
    },
    {
      "epoch": 1.3994990607388855,
      "grad_norm": 0.4746815860271454,
      "learning_rate": 9.401270772238514e-05,
      "loss": 0.2006,
      "step": 8940
    },
    {
      "epoch": 1.3996556042579837,
      "grad_norm": 0.42531949281692505,
      "learning_rate": 9.39882697947214e-05,
      "loss": 0.1976,
      "step": 8941
    },
    {
      "epoch": 1.399812147777082,
      "grad_norm": 0.8540092706680298,
      "learning_rate": 9.396383186705765e-05,
      "loss": 0.2068,
      "step": 8942
    },
    {
      "epoch": 1.3999686912961804,
      "grad_norm": 0.6693518757820129,
      "learning_rate": 9.393939393939393e-05,
      "loss": 0.242,
      "step": 8943
    },
    {
      "epoch": 1.4001252348152786,
      "grad_norm": 0.9643524289131165,
      "learning_rate": 9.39149560117302e-05,
      "loss": 0.335,
      "step": 8944
    },
    {
      "epoch": 1.400281778334377,
      "grad_norm": 0.5744693279266357,
      "learning_rate": 9.389051808406646e-05,
      "loss": 0.222,
      "step": 8945
    },
    {
      "epoch": 1.4004383218534753,
      "grad_norm": 1.0014822483062744,
      "learning_rate": 9.386608015640274e-05,
      "loss": 0.3107,
      "step": 8946
    },
    {
      "epoch": 1.4005948653725735,
      "grad_norm": 0.8832296133041382,
      "learning_rate": 9.384164222873899e-05,
      "loss": 0.3322,
      "step": 8947
    },
    {
      "epoch": 1.400751408891672,
      "grad_norm": 0.46709567308425903,
      "learning_rate": 9.381720430107526e-05,
      "loss": 0.254,
      "step": 8948
    },
    {
      "epoch": 1.4009079524107702,
      "grad_norm": 1.0916463136672974,
      "learning_rate": 9.379276637341154e-05,
      "loss": 0.3261,
      "step": 8949
    },
    {
      "epoch": 1.4010644959298686,
      "grad_norm": 1.3483251333236694,
      "learning_rate": 9.376832844574779e-05,
      "loss": 0.4645,
      "step": 8950
    },
    {
      "epoch": 1.4012210394489668,
      "grad_norm": 1.411783218383789,
      "learning_rate": 9.374389051808405e-05,
      "loss": 0.3502,
      "step": 8951
    },
    {
      "epoch": 1.401377582968065,
      "grad_norm": 0.6709024310112,
      "learning_rate": 9.371945259042033e-05,
      "loss": 0.2547,
      "step": 8952
    },
    {
      "epoch": 1.4015341264871635,
      "grad_norm": 0.7980188727378845,
      "learning_rate": 9.36950146627566e-05,
      "loss": 0.2641,
      "step": 8953
    },
    {
      "epoch": 1.4016906700062617,
      "grad_norm": 1.5928226709365845,
      "learning_rate": 9.367057673509285e-05,
      "loss": 0.5429,
      "step": 8954
    },
    {
      "epoch": 1.4018472135253601,
      "grad_norm": 2.243115186691284,
      "learning_rate": 9.364613880742912e-05,
      "loss": 1.0404,
      "step": 8955
    },
    {
      "epoch": 1.4020037570444583,
      "grad_norm": 2.163041591644287,
      "learning_rate": 9.362170087976539e-05,
      "loss": 0.5766,
      "step": 8956
    },
    {
      "epoch": 1.4021603005635566,
      "grad_norm": 1.891203761100769,
      "learning_rate": 9.359726295210165e-05,
      "loss": 0.5016,
      "step": 8957
    },
    {
      "epoch": 1.402316844082655,
      "grad_norm": 0.8651189208030701,
      "learning_rate": 9.357282502443793e-05,
      "loss": 0.3276,
      "step": 8958
    },
    {
      "epoch": 1.4024733876017532,
      "grad_norm": 2.448225736618042,
      "learning_rate": 9.354838709677418e-05,
      "loss": 0.6193,
      "step": 8959
    },
    {
      "epoch": 1.4026299311208517,
      "grad_norm": 2.657435178756714,
      "learning_rate": 9.352394916911045e-05,
      "loss": 0.4568,
      "step": 8960
    },
    {
      "epoch": 1.4027864746399499,
      "grad_norm": 3.460101842880249,
      "learning_rate": 9.349951124144673e-05,
      "loss": 1.0414,
      "step": 8961
    },
    {
      "epoch": 1.402943018159048,
      "grad_norm": 1.3931571245193481,
      "learning_rate": 9.347507331378298e-05,
      "loss": 0.6641,
      "step": 8962
    },
    {
      "epoch": 1.4030995616781465,
      "grad_norm": 1.7879291772842407,
      "learning_rate": 9.345063538611924e-05,
      "loss": 0.8282,
      "step": 8963
    },
    {
      "epoch": 1.4032561051972448,
      "grad_norm": 1.4838361740112305,
      "learning_rate": 9.342619745845552e-05,
      "loss": 0.7365,
      "step": 8964
    },
    {
      "epoch": 1.4034126487163432,
      "grad_norm": 1.6955534219741821,
      "learning_rate": 9.340175953079179e-05,
      "loss": 0.24,
      "step": 8965
    },
    {
      "epoch": 1.4035691922354414,
      "grad_norm": 2.0134198665618896,
      "learning_rate": 9.337732160312804e-05,
      "loss": 0.7978,
      "step": 8966
    },
    {
      "epoch": 1.4037257357545396,
      "grad_norm": 1.6322916746139526,
      "learning_rate": 9.335288367546432e-05,
      "loss": 0.6308,
      "step": 8967
    },
    {
      "epoch": 1.403882279273638,
      "grad_norm": 1.2652108669281006,
      "learning_rate": 9.332844574780058e-05,
      "loss": 0.504,
      "step": 8968
    },
    {
      "epoch": 1.4040388227927365,
      "grad_norm": 1.0733211040496826,
      "learning_rate": 9.330400782013684e-05,
      "loss": 0.3422,
      "step": 8969
    },
    {
      "epoch": 1.4041953663118347,
      "grad_norm": 1.4503707885742188,
      "learning_rate": 9.327956989247311e-05,
      "loss": 0.3383,
      "step": 8970
    },
    {
      "epoch": 1.404351909830933,
      "grad_norm": 1.633813738822937,
      "learning_rate": 9.325513196480937e-05,
      "loss": 0.7733,
      "step": 8971
    },
    {
      "epoch": 1.4045084533500314,
      "grad_norm": 1.3477745056152344,
      "learning_rate": 9.323069403714564e-05,
      "loss": 0.3886,
      "step": 8972
    },
    {
      "epoch": 1.4046649968691296,
      "grad_norm": 2.0735232830047607,
      "learning_rate": 9.320625610948192e-05,
      "loss": 0.7784,
      "step": 8973
    },
    {
      "epoch": 1.404821540388228,
      "grad_norm": 2.603829860687256,
      "learning_rate": 9.318181818181817e-05,
      "loss": 0.88,
      "step": 8974
    },
    {
      "epoch": 1.4049780839073263,
      "grad_norm": 1.9070336818695068,
      "learning_rate": 9.315738025415443e-05,
      "loss": 0.9435,
      "step": 8975
    },
    {
      "epoch": 1.4051346274264245,
      "grad_norm": 1.9831008911132812,
      "learning_rate": 9.313294232649071e-05,
      "loss": 0.9925,
      "step": 8976
    },
    {
      "epoch": 1.405291170945523,
      "grad_norm": 2.742797374725342,
      "learning_rate": 9.310850439882698e-05,
      "loss": 1.2224,
      "step": 8977
    },
    {
      "epoch": 1.4054477144646211,
      "grad_norm": 2.2004201412200928,
      "learning_rate": 9.308406647116323e-05,
      "loss": 0.9814,
      "step": 8978
    },
    {
      "epoch": 1.4056042579837196,
      "grad_norm": 2.0755558013916016,
      "learning_rate": 9.30596285434995e-05,
      "loss": 1.2202,
      "step": 8979
    },
    {
      "epoch": 1.4057608015028178,
      "grad_norm": 2.546623706817627,
      "learning_rate": 9.303519061583577e-05,
      "loss": 0.8416,
      "step": 8980
    },
    {
      "epoch": 1.405917345021916,
      "grad_norm": 4.5670485496521,
      "learning_rate": 9.301075268817204e-05,
      "loss": 1.7262,
      "step": 8981
    },
    {
      "epoch": 1.4060738885410144,
      "grad_norm": 1.8530974388122559,
      "learning_rate": 9.29863147605083e-05,
      "loss": 1.209,
      "step": 8982
    },
    {
      "epoch": 1.4062304320601127,
      "grad_norm": 1.714417815208435,
      "learning_rate": 9.296187683284457e-05,
      "loss": 1.0305,
      "step": 8983
    },
    {
      "epoch": 1.406386975579211,
      "grad_norm": 3.0805559158325195,
      "learning_rate": 9.293743890518083e-05,
      "loss": 0.7601,
      "step": 8984
    },
    {
      "epoch": 1.4065435190983093,
      "grad_norm": 1.3468403816223145,
      "learning_rate": 9.291300097751711e-05,
      "loss": 0.1582,
      "step": 8985
    },
    {
      "epoch": 1.4067000626174075,
      "grad_norm": 1.245559573173523,
      "learning_rate": 9.288856304985336e-05,
      "loss": 0.3851,
      "step": 8986
    },
    {
      "epoch": 1.406856606136506,
      "grad_norm": 2.107041358947754,
      "learning_rate": 9.286412512218962e-05,
      "loss": 0.5471,
      "step": 8987
    },
    {
      "epoch": 1.4070131496556042,
      "grad_norm": 1.9285342693328857,
      "learning_rate": 9.28396871945259e-05,
      "loss": 0.5434,
      "step": 8988
    },
    {
      "epoch": 1.4071696931747026,
      "grad_norm": 0.39765217900276184,
      "learning_rate": 9.281524926686217e-05,
      "loss": 0.222,
      "step": 8989
    },
    {
      "epoch": 1.4073262366938009,
      "grad_norm": 0.596780002117157,
      "learning_rate": 9.279081133919842e-05,
      "loss": 0.2276,
      "step": 8990
    },
    {
      "epoch": 1.407482780212899,
      "grad_norm": 0.4721631705760956,
      "learning_rate": 9.27663734115347e-05,
      "loss": 0.196,
      "step": 8991
    },
    {
      "epoch": 1.4076393237319975,
      "grad_norm": 2.310553550720215,
      "learning_rate": 9.274193548387096e-05,
      "loss": 0.4844,
      "step": 8992
    },
    {
      "epoch": 1.4077958672510957,
      "grad_norm": 0.47364363074302673,
      "learning_rate": 9.271749755620723e-05,
      "loss": 0.2752,
      "step": 8993
    },
    {
      "epoch": 1.4079524107701942,
      "grad_norm": 1.2700831890106201,
      "learning_rate": 9.269305962854349e-05,
      "loss": 0.4011,
      "step": 8994
    },
    {
      "epoch": 1.4081089542892924,
      "grad_norm": 0.6444479823112488,
      "learning_rate": 9.266862170087976e-05,
      "loss": 0.2654,
      "step": 8995
    },
    {
      "epoch": 1.4082654978083906,
      "grad_norm": 0.6023876667022705,
      "learning_rate": 9.264418377321602e-05,
      "loss": 0.3546,
      "step": 8996
    },
    {
      "epoch": 1.408422041327489,
      "grad_norm": 0.7151503562927246,
      "learning_rate": 9.26197458455523e-05,
      "loss": 0.249,
      "step": 8997
    },
    {
      "epoch": 1.4085785848465875,
      "grad_norm": 0.5194156765937805,
      "learning_rate": 9.259530791788855e-05,
      "loss": 0.2368,
      "step": 8998
    },
    {
      "epoch": 1.4087351283656857,
      "grad_norm": 0.761722207069397,
      "learning_rate": 9.257086999022482e-05,
      "loss": 0.2769,
      "step": 8999
    },
    {
      "epoch": 1.408891671884784,
      "grad_norm": 0.7770941853523254,
      "learning_rate": 9.25464320625611e-05,
      "loss": 0.2916,
      "step": 9000
    },
    {
      "epoch": 1.408891671884784,
      "eval_loss": 0.49010393023490906,
      "eval_runtime": 205.5553,
      "eval_samples_per_second": 60.242,
      "eval_steps_per_second": 3.765,
      "eval_wer": 0.3170729115753235,
      "step": 9000
    },
    {
      "epoch": 1.4090482154038821,
      "grad_norm": 1.107610821723938,
      "learning_rate": 9.252199413489736e-05,
      "loss": 0.4199,
      "step": 9001
    },
    {
      "epoch": 1.4092047589229806,
      "grad_norm": 1.4401216506958008,
      "learning_rate": 9.249755620723361e-05,
      "loss": 0.4704,
      "step": 9002
    },
    {
      "epoch": 1.409361302442079,
      "grad_norm": 0.8990141749382019,
      "learning_rate": 9.247311827956989e-05,
      "loss": 0.325,
      "step": 9003
    },
    {
      "epoch": 1.4095178459611772,
      "grad_norm": 1.2181795835494995,
      "learning_rate": 9.244868035190615e-05,
      "loss": 0.5246,
      "step": 9004
    },
    {
      "epoch": 1.4096743894802755,
      "grad_norm": 0.9682489037513733,
      "learning_rate": 9.242424242424242e-05,
      "loss": 0.3604,
      "step": 9005
    },
    {
      "epoch": 1.409830932999374,
      "grad_norm": 1.354854702949524,
      "learning_rate": 9.239980449657868e-05,
      "loss": 0.5106,
      "step": 9006
    },
    {
      "epoch": 1.4099874765184721,
      "grad_norm": 0.8772504925727844,
      "learning_rate": 9.237536656891495e-05,
      "loss": 0.3421,
      "step": 9007
    },
    {
      "epoch": 1.4101440200375706,
      "grad_norm": 0.9204466938972473,
      "learning_rate": 9.235092864125121e-05,
      "loss": 0.3441,
      "step": 9008
    },
    {
      "epoch": 1.4103005635566688,
      "grad_norm": 1.3267632722854614,
      "learning_rate": 9.232649071358749e-05,
      "loss": 0.5436,
      "step": 9009
    },
    {
      "epoch": 1.410457107075767,
      "grad_norm": 1.8505562543869019,
      "learning_rate": 9.230205278592374e-05,
      "loss": 0.5114,
      "step": 9010
    },
    {
      "epoch": 1.4106136505948654,
      "grad_norm": 2.1749753952026367,
      "learning_rate": 9.227761485826001e-05,
      "loss": 0.4848,
      "step": 9011
    },
    {
      "epoch": 1.4107701941139636,
      "grad_norm": 1.2745190858840942,
      "learning_rate": 9.225317693059629e-05,
      "loss": 0.5289,
      "step": 9012
    },
    {
      "epoch": 1.410926737633062,
      "grad_norm": 0.8753221035003662,
      "learning_rate": 9.222873900293255e-05,
      "loss": 0.2345,
      "step": 9013
    },
    {
      "epoch": 1.4110832811521603,
      "grad_norm": 1.8300694227218628,
      "learning_rate": 9.22043010752688e-05,
      "loss": 0.7186,
      "step": 9014
    },
    {
      "epoch": 1.4112398246712585,
      "grad_norm": 2.5237860679626465,
      "learning_rate": 9.217986314760508e-05,
      "loss": 0.7047,
      "step": 9015
    },
    {
      "epoch": 1.411396368190357,
      "grad_norm": 1.3644801378250122,
      "learning_rate": 9.215542521994134e-05,
      "loss": 0.5634,
      "step": 9016
    },
    {
      "epoch": 1.4115529117094552,
      "grad_norm": 3.8021445274353027,
      "learning_rate": 9.213098729227761e-05,
      "loss": 0.3977,
      "step": 9017
    },
    {
      "epoch": 1.4117094552285536,
      "grad_norm": 2.972228527069092,
      "learning_rate": 9.210654936461387e-05,
      "loss": 1.2041,
      "step": 9018
    },
    {
      "epoch": 1.4118659987476518,
      "grad_norm": 2.1545019149780273,
      "learning_rate": 9.208211143695014e-05,
      "loss": 0.5613,
      "step": 9019
    },
    {
      "epoch": 1.41202254226675,
      "grad_norm": 2.832612991333008,
      "learning_rate": 9.20576735092864e-05,
      "loss": 0.5319,
      "step": 9020
    },
    {
      "epoch": 1.4121790857858485,
      "grad_norm": 2.0338563919067383,
      "learning_rate": 9.203323558162268e-05,
      "loss": 0.7196,
      "step": 9021
    },
    {
      "epoch": 1.4123356293049467,
      "grad_norm": 3.1499016284942627,
      "learning_rate": 9.200879765395893e-05,
      "loss": 1.0724,
      "step": 9022
    },
    {
      "epoch": 1.4124921728240452,
      "grad_norm": 3.166935443878174,
      "learning_rate": 9.19843597262952e-05,
      "loss": 0.8717,
      "step": 9023
    },
    {
      "epoch": 1.4126487163431434,
      "grad_norm": 4.040381908416748,
      "learning_rate": 9.195992179863148e-05,
      "loss": 1.0294,
      "step": 9024
    },
    {
      "epoch": 1.4128052598622416,
      "grad_norm": 3.9483299255371094,
      "learning_rate": 9.193548387096774e-05,
      "loss": 0.8536,
      "step": 9025
    },
    {
      "epoch": 1.41296180338134,
      "grad_norm": 7.188598155975342,
      "learning_rate": 9.191104594330399e-05,
      "loss": 0.9649,
      "step": 9026
    },
    {
      "epoch": 1.4131183469004382,
      "grad_norm": 2.007357597351074,
      "learning_rate": 9.188660801564027e-05,
      "loss": 1.0681,
      "step": 9027
    },
    {
      "epoch": 1.4132748904195367,
      "grad_norm": 3.2346351146698,
      "learning_rate": 9.186217008797654e-05,
      "loss": 0.9114,
      "step": 9028
    },
    {
      "epoch": 1.413431433938635,
      "grad_norm": 3.834306240081787,
      "learning_rate": 9.183773216031279e-05,
      "loss": 0.883,
      "step": 9029
    },
    {
      "epoch": 1.4135879774577331,
      "grad_norm": 4.527556419372559,
      "learning_rate": 9.181329423264907e-05,
      "loss": 0.8806,
      "step": 9030
    },
    {
      "epoch": 1.4137445209768316,
      "grad_norm": 3.6661274433135986,
      "learning_rate": 9.178885630498533e-05,
      "loss": 1.2201,
      "step": 9031
    },
    {
      "epoch": 1.41390106449593,
      "grad_norm": 1.937548279762268,
      "learning_rate": 9.17644183773216e-05,
      "loss": 0.7024,
      "step": 9032
    },
    {
      "epoch": 1.4140576080150282,
      "grad_norm": 2.842707872390747,
      "learning_rate": 9.173998044965787e-05,
      "loss": 0.8705,
      "step": 9033
    },
    {
      "epoch": 1.4142141515341264,
      "grad_norm": 1.4602514505386353,
      "learning_rate": 9.171554252199412e-05,
      "loss": 0.4965,
      "step": 9034
    },
    {
      "epoch": 1.4143706950532247,
      "grad_norm": 2.891000747680664,
      "learning_rate": 9.169110459433039e-05,
      "loss": 1.2006,
      "step": 9035
    },
    {
      "epoch": 1.414527238572323,
      "grad_norm": 1.664858102798462,
      "learning_rate": 9.166666666666667e-05,
      "loss": 0.4995,
      "step": 9036
    },
    {
      "epoch": 1.4146837820914215,
      "grad_norm": 3.580497980117798,
      "learning_rate": 9.164222873900293e-05,
      "loss": 1.5147,
      "step": 9037
    },
    {
      "epoch": 1.4148403256105198,
      "grad_norm": 2.788262367248535,
      "learning_rate": 9.161779081133918e-05,
      "loss": 1.297,
      "step": 9038
    },
    {
      "epoch": 1.414996869129618,
      "grad_norm": 0.46237197518348694,
      "learning_rate": 9.159335288367546e-05,
      "loss": 0.2044,
      "step": 9039
    },
    {
      "epoch": 1.4151534126487164,
      "grad_norm": 0.7125651836395264,
      "learning_rate": 9.156891495601173e-05,
      "loss": 0.2271,
      "step": 9040
    },
    {
      "epoch": 1.4153099561678146,
      "grad_norm": 0.6166035532951355,
      "learning_rate": 9.154447702834798e-05,
      "loss": 0.2674,
      "step": 9041
    },
    {
      "epoch": 1.415466499686913,
      "grad_norm": 0.3849860727787018,
      "learning_rate": 9.152003910068426e-05,
      "loss": 0.2302,
      "step": 9042
    },
    {
      "epoch": 1.4156230432060113,
      "grad_norm": 0.6451732516288757,
      "learning_rate": 9.149560117302052e-05,
      "loss": 0.2397,
      "step": 9043
    },
    {
      "epoch": 1.4157795867251095,
      "grad_norm": 0.41967400908470154,
      "learning_rate": 9.147116324535679e-05,
      "loss": 0.229,
      "step": 9044
    },
    {
      "epoch": 1.415936130244208,
      "grad_norm": 0.9496431946754456,
      "learning_rate": 9.144672531769306e-05,
      "loss": 0.3821,
      "step": 9045
    },
    {
      "epoch": 1.4160926737633062,
      "grad_norm": 0.634460985660553,
      "learning_rate": 9.142228739002932e-05,
      "loss": 0.1948,
      "step": 9046
    },
    {
      "epoch": 1.4162492172824046,
      "grad_norm": 0.9624339938163757,
      "learning_rate": 9.139784946236558e-05,
      "loss": 0.2305,
      "step": 9047
    },
    {
      "epoch": 1.4164057608015028,
      "grad_norm": 1.082798719406128,
      "learning_rate": 9.137341153470186e-05,
      "loss": 0.2501,
      "step": 9048
    },
    {
      "epoch": 1.416562304320601,
      "grad_norm": 0.5590087175369263,
      "learning_rate": 9.134897360703812e-05,
      "loss": 0.1449,
      "step": 9049
    },
    {
      "epoch": 1.4167188478396995,
      "grad_norm": 1.2364987134933472,
      "learning_rate": 9.132453567937437e-05,
      "loss": 0.1856,
      "step": 9050
    },
    {
      "epoch": 1.4168753913587977,
      "grad_norm": 0.5167357325553894,
      "learning_rate": 9.130009775171065e-05,
      "loss": 0.2644,
      "step": 9051
    },
    {
      "epoch": 1.4170319348778961,
      "grad_norm": 2.4510698318481445,
      "learning_rate": 9.127565982404692e-05,
      "loss": 0.4774,
      "step": 9052
    },
    {
      "epoch": 1.4171884783969944,
      "grad_norm": 1.2611892223358154,
      "learning_rate": 9.125122189638317e-05,
      "loss": 0.5563,
      "step": 9053
    },
    {
      "epoch": 1.4173450219160926,
      "grad_norm": 1.279375672340393,
      "learning_rate": 9.122678396871945e-05,
      "loss": 0.4933,
      "step": 9054
    },
    {
      "epoch": 1.417501565435191,
      "grad_norm": 1.3161873817443848,
      "learning_rate": 9.120234604105571e-05,
      "loss": 0.5997,
      "step": 9055
    },
    {
      "epoch": 1.4176581089542892,
      "grad_norm": 0.7085786461830139,
      "learning_rate": 9.117790811339198e-05,
      "loss": 0.2048,
      "step": 9056
    },
    {
      "epoch": 1.4178146524733877,
      "grad_norm": 1.8899716138839722,
      "learning_rate": 9.115347018572826e-05,
      "loss": 0.3395,
      "step": 9057
    },
    {
      "epoch": 1.4179711959924859,
      "grad_norm": 2.1545331478118896,
      "learning_rate": 9.112903225806451e-05,
      "loss": 0.5075,
      "step": 9058
    },
    {
      "epoch": 1.418127739511584,
      "grad_norm": 1.0476807355880737,
      "learning_rate": 9.110459433040077e-05,
      "loss": 0.2844,
      "step": 9059
    },
    {
      "epoch": 1.4182842830306825,
      "grad_norm": 1.534098505973816,
      "learning_rate": 9.108015640273705e-05,
      "loss": 0.481,
      "step": 9060
    },
    {
      "epoch": 1.4184408265497808,
      "grad_norm": 1.6640297174453735,
      "learning_rate": 9.105571847507331e-05,
      "loss": 0.3992,
      "step": 9061
    },
    {
      "epoch": 1.4185973700688792,
      "grad_norm": 1.0398399829864502,
      "learning_rate": 9.103128054740957e-05,
      "loss": 0.484,
      "step": 9062
    },
    {
      "epoch": 1.4187539135879774,
      "grad_norm": 1.172825574874878,
      "learning_rate": 9.100684261974584e-05,
      "loss": 0.5006,
      "step": 9063
    },
    {
      "epoch": 1.4189104571070756,
      "grad_norm": 1.707702875137329,
      "learning_rate": 9.098240469208211e-05,
      "loss": 0.3896,
      "step": 9064
    },
    {
      "epoch": 1.419067000626174,
      "grad_norm": 1.5226316452026367,
      "learning_rate": 9.095796676441836e-05,
      "loss": 0.689,
      "step": 9065
    },
    {
      "epoch": 1.4192235441452725,
      "grad_norm": 1.6208080053329468,
      "learning_rate": 9.093352883675464e-05,
      "loss": 0.4796,
      "step": 9066
    },
    {
      "epoch": 1.4193800876643707,
      "grad_norm": 2.61185622215271,
      "learning_rate": 9.09090909090909e-05,
      "loss": 0.5755,
      "step": 9067
    },
    {
      "epoch": 1.419536631183469,
      "grad_norm": 2.0593795776367188,
      "learning_rate": 9.088465298142717e-05,
      "loss": 0.7231,
      "step": 9068
    },
    {
      "epoch": 1.4196931747025674,
      "grad_norm": 1.8883817195892334,
      "learning_rate": 9.086021505376345e-05,
      "loss": 0.7263,
      "step": 9069
    },
    {
      "epoch": 1.4198497182216656,
      "grad_norm": 3.10384464263916,
      "learning_rate": 9.08357771260997e-05,
      "loss": 0.7273,
      "step": 9070
    },
    {
      "epoch": 1.420006261740764,
      "grad_norm": 1.468448281288147,
      "learning_rate": 9.081133919843596e-05,
      "loss": 0.7773,
      "step": 9071
    },
    {
      "epoch": 1.4201628052598623,
      "grad_norm": 1.079703450202942,
      "learning_rate": 9.078690127077224e-05,
      "loss": 0.4677,
      "step": 9072
    },
    {
      "epoch": 1.4203193487789605,
      "grad_norm": 2.0892856121063232,
      "learning_rate": 9.076246334310849e-05,
      "loss": 0.5249,
      "step": 9073
    },
    {
      "epoch": 1.420475892298059,
      "grad_norm": 2.025650978088379,
      "learning_rate": 9.073802541544476e-05,
      "loss": 0.4734,
      "step": 9074
    },
    {
      "epoch": 1.4206324358171571,
      "grad_norm": 2.4783313274383545,
      "learning_rate": 9.071358748778104e-05,
      "loss": 0.776,
      "step": 9075
    },
    {
      "epoch": 1.4207889793362556,
      "grad_norm": 2.0602166652679443,
      "learning_rate": 9.06891495601173e-05,
      "loss": 0.772,
      "step": 9076
    },
    {
      "epoch": 1.4209455228553538,
      "grad_norm": 2.0721070766448975,
      "learning_rate": 9.066471163245355e-05,
      "loss": 0.5761,
      "step": 9077
    },
    {
      "epoch": 1.421102066374452,
      "grad_norm": 2.9403793811798096,
      "learning_rate": 9.064027370478983e-05,
      "loss": 0.7641,
      "step": 9078
    },
    {
      "epoch": 1.4212586098935505,
      "grad_norm": 2.754152536392212,
      "learning_rate": 9.06158357771261e-05,
      "loss": 1.1597,
      "step": 9079
    },
    {
      "epoch": 1.4214151534126487,
      "grad_norm": 3.737293243408203,
      "learning_rate": 9.059139784946236e-05,
      "loss": 1.0165,
      "step": 9080
    },
    {
      "epoch": 1.4215716969317471,
      "grad_norm": 4.649268627166748,
      "learning_rate": 9.056695992179861e-05,
      "loss": 0.8176,
      "step": 9081
    },
    {
      "epoch": 1.4217282404508453,
      "grad_norm": 4.204807758331299,
      "learning_rate": 9.054252199413489e-05,
      "loss": 1.1525,
      "step": 9082
    },
    {
      "epoch": 1.4218847839699436,
      "grad_norm": 1.6178537607192993,
      "learning_rate": 9.051808406647115e-05,
      "loss": 1.4486,
      "step": 9083
    },
    {
      "epoch": 1.422041327489042,
      "grad_norm": 1.8884669542312622,
      "learning_rate": 9.049364613880742e-05,
      "loss": 0.4542,
      "step": 9084
    },
    {
      "epoch": 1.4221978710081402,
      "grad_norm": 2.976574182510376,
      "learning_rate": 9.046920821114368e-05,
      "loss": 0.7788,
      "step": 9085
    },
    {
      "epoch": 1.4223544145272387,
      "grad_norm": 2.2067227363586426,
      "learning_rate": 9.044477028347995e-05,
      "loss": 0.642,
      "step": 9086
    },
    {
      "epoch": 1.4225109580463369,
      "grad_norm": 1.5230540037155151,
      "learning_rate": 9.042033235581621e-05,
      "loss": 0.4552,
      "step": 9087
    },
    {
      "epoch": 1.422667501565435,
      "grad_norm": 1.6714625358581543,
      "learning_rate": 9.039589442815249e-05,
      "loss": 0.8108,
      "step": 9088
    },
    {
      "epoch": 1.4228240450845335,
      "grad_norm": 0.5419172644615173,
      "learning_rate": 9.037145650048874e-05,
      "loss": 0.2396,
      "step": 9089
    },
    {
      "epoch": 1.4229805886036317,
      "grad_norm": 0.5144709944725037,
      "learning_rate": 9.034701857282501e-05,
      "loss": 0.2611,
      "step": 9090
    },
    {
      "epoch": 1.4231371321227302,
      "grad_norm": 0.7084149718284607,
      "learning_rate": 9.032258064516129e-05,
      "loss": 0.3145,
      "step": 9091
    },
    {
      "epoch": 1.4232936756418284,
      "grad_norm": 0.7008088827133179,
      "learning_rate": 9.029814271749755e-05,
      "loss": 0.265,
      "step": 9092
    },
    {
      "epoch": 1.4234502191609266,
      "grad_norm": 2.3117635250091553,
      "learning_rate": 9.02737047898338e-05,
      "loss": 0.4497,
      "step": 9093
    },
    {
      "epoch": 1.423606762680025,
      "grad_norm": 0.9773268103599548,
      "learning_rate": 9.024926686217008e-05,
      "loss": 0.3003,
      "step": 9094
    },
    {
      "epoch": 1.4237633061991233,
      "grad_norm": 0.8467937111854553,
      "learning_rate": 9.022482893450635e-05,
      "loss": 0.2749,
      "step": 9095
    },
    {
      "epoch": 1.4239198497182217,
      "grad_norm": 0.7267017960548401,
      "learning_rate": 9.020039100684261e-05,
      "loss": 0.2498,
      "step": 9096
    },
    {
      "epoch": 1.42407639323732,
      "grad_norm": 0.7348589301109314,
      "learning_rate": 9.017595307917887e-05,
      "loss": 0.2563,
      "step": 9097
    },
    {
      "epoch": 1.4242329367564182,
      "grad_norm": 1.283211350440979,
      "learning_rate": 9.015151515151514e-05,
      "loss": 0.3757,
      "step": 9098
    },
    {
      "epoch": 1.4243894802755166,
      "grad_norm": 0.4818941652774811,
      "learning_rate": 9.01270772238514e-05,
      "loss": 0.177,
      "step": 9099
    },
    {
      "epoch": 1.424546023794615,
      "grad_norm": 0.830980122089386,
      "learning_rate": 9.010263929618768e-05,
      "loss": 0.2567,
      "step": 9100
    },
    {
      "epoch": 1.4247025673137133,
      "grad_norm": 1.4470046758651733,
      "learning_rate": 9.007820136852393e-05,
      "loss": 0.4034,
      "step": 9101
    },
    {
      "epoch": 1.4248591108328115,
      "grad_norm": 1.1172550916671753,
      "learning_rate": 9.00537634408602e-05,
      "loss": 0.377,
      "step": 9102
    },
    {
      "epoch": 1.42501565435191,
      "grad_norm": 1.1227866411209106,
      "learning_rate": 9.002932551319648e-05,
      "loss": 0.2965,
      "step": 9103
    },
    {
      "epoch": 1.4251721978710081,
      "grad_norm": 0.9577900767326355,
      "learning_rate": 9.000488758553274e-05,
      "loss": 0.3678,
      "step": 9104
    },
    {
      "epoch": 1.4253287413901066,
      "grad_norm": 1.0543158054351807,
      "learning_rate": 8.998044965786899e-05,
      "loss": 0.2968,
      "step": 9105
    },
    {
      "epoch": 1.4254852849092048,
      "grad_norm": 0.9723239541053772,
      "learning_rate": 8.995601173020527e-05,
      "loss": 0.4421,
      "step": 9106
    },
    {
      "epoch": 1.425641828428303,
      "grad_norm": 1.2686231136322021,
      "learning_rate": 8.993157380254154e-05,
      "loss": 0.5202,
      "step": 9107
    },
    {
      "epoch": 1.4257983719474014,
      "grad_norm": 1.9442496299743652,
      "learning_rate": 8.99071358748778e-05,
      "loss": 0.4751,
      "step": 9108
    },
    {
      "epoch": 1.4259549154664997,
      "grad_norm": 2.063786745071411,
      "learning_rate": 8.988269794721407e-05,
      "loss": 0.517,
      "step": 9109
    },
    {
      "epoch": 1.426111458985598,
      "grad_norm": 0.9845725893974304,
      "learning_rate": 8.985826001955033e-05,
      "loss": 0.2919,
      "step": 9110
    },
    {
      "epoch": 1.4262680025046963,
      "grad_norm": 2.66135311126709,
      "learning_rate": 8.98338220918866e-05,
      "loss": 0.9835,
      "step": 9111
    },
    {
      "epoch": 1.4264245460237945,
      "grad_norm": 2.083529233932495,
      "learning_rate": 8.980938416422287e-05,
      "loss": 0.7316,
      "step": 9112
    },
    {
      "epoch": 1.426581089542893,
      "grad_norm": 1.5067460536956787,
      "learning_rate": 8.978494623655913e-05,
      "loss": 0.5985,
      "step": 9113
    },
    {
      "epoch": 1.4267376330619912,
      "grad_norm": 2.6055350303649902,
      "learning_rate": 8.976050830889539e-05,
      "loss": 0.4876,
      "step": 9114
    },
    {
      "epoch": 1.4268941765810896,
      "grad_norm": 1.7742613554000854,
      "learning_rate": 8.973607038123167e-05,
      "loss": 0.4211,
      "step": 9115
    },
    {
      "epoch": 1.4270507201001879,
      "grad_norm": 0.9075111746788025,
      "learning_rate": 8.971163245356793e-05,
      "loss": 0.2719,
      "step": 9116
    },
    {
      "epoch": 1.427207263619286,
      "grad_norm": 1.751970648765564,
      "learning_rate": 8.968719452590418e-05,
      "loss": 0.6691,
      "step": 9117
    },
    {
      "epoch": 1.4273638071383845,
      "grad_norm": 1.8061598539352417,
      "learning_rate": 8.966275659824046e-05,
      "loss": 0.4919,
      "step": 9118
    },
    {
      "epoch": 1.4275203506574827,
      "grad_norm": 5.433608055114746,
      "learning_rate": 8.963831867057673e-05,
      "loss": 1.2975,
      "step": 9119
    },
    {
      "epoch": 1.4276768941765812,
      "grad_norm": 1.150429129600525,
      "learning_rate": 8.961388074291299e-05,
      "loss": 0.3322,
      "step": 9120
    },
    {
      "epoch": 1.4278334376956794,
      "grad_norm": 1.8451645374298096,
      "learning_rate": 8.958944281524926e-05,
      "loss": 0.5617,
      "step": 9121
    },
    {
      "epoch": 1.4279899812147776,
      "grad_norm": 2.325820207595825,
      "learning_rate": 8.956500488758552e-05,
      "loss": 0.9685,
      "step": 9122
    },
    {
      "epoch": 1.428146524733876,
      "grad_norm": 2.732015609741211,
      "learning_rate": 8.954056695992179e-05,
      "loss": 0.6398,
      "step": 9123
    },
    {
      "epoch": 1.4283030682529743,
      "grad_norm": 2.809544086456299,
      "learning_rate": 8.951612903225806e-05,
      "loss": 0.8187,
      "step": 9124
    },
    {
      "epoch": 1.4284596117720727,
      "grad_norm": 2.3526062965393066,
      "learning_rate": 8.949169110459432e-05,
      "loss": 0.9776,
      "step": 9125
    },
    {
      "epoch": 1.428616155291171,
      "grad_norm": 3.0225517749786377,
      "learning_rate": 8.946725317693058e-05,
      "loss": 1.0524,
      "step": 9126
    },
    {
      "epoch": 1.4287726988102691,
      "grad_norm": 9.465794563293457,
      "learning_rate": 8.944281524926686e-05,
      "loss": 0.8393,
      "step": 9127
    },
    {
      "epoch": 1.4289292423293676,
      "grad_norm": 2.5339739322662354,
      "learning_rate": 8.941837732160312e-05,
      "loss": 0.9948,
      "step": 9128
    },
    {
      "epoch": 1.4290857858484658,
      "grad_norm": 1.795351266860962,
      "learning_rate": 8.939393939393938e-05,
      "loss": 0.988,
      "step": 9129
    },
    {
      "epoch": 1.4292423293675642,
      "grad_norm": 2.1344945430755615,
      "learning_rate": 8.936950146627565e-05,
      "loss": 1.1098,
      "step": 9130
    },
    {
      "epoch": 1.4293988728866625,
      "grad_norm": 2.079683542251587,
      "learning_rate": 8.934506353861192e-05,
      "loss": 0.7107,
      "step": 9131
    },
    {
      "epoch": 1.4295554164057607,
      "grad_norm": 2.6614019870758057,
      "learning_rate": 8.932062561094817e-05,
      "loss": 1.8157,
      "step": 9132
    },
    {
      "epoch": 1.429711959924859,
      "grad_norm": 2.139328718185425,
      "learning_rate": 8.929618768328445e-05,
      "loss": 1.1497,
      "step": 9133
    },
    {
      "epoch": 1.4298685034439576,
      "grad_norm": 1.177253246307373,
      "learning_rate": 8.927174975562071e-05,
      "loss": 0.3898,
      "step": 9134
    },
    {
      "epoch": 1.4300250469630558,
      "grad_norm": 1.5603402853012085,
      "learning_rate": 8.924731182795698e-05,
      "loss": 0.8206,
      "step": 9135
    },
    {
      "epoch": 1.430181590482154,
      "grad_norm": 3.4352011680603027,
      "learning_rate": 8.922287390029326e-05,
      "loss": 0.4282,
      "step": 9136
    },
    {
      "epoch": 1.4303381340012524,
      "grad_norm": 1.563423752784729,
      "learning_rate": 8.919843597262951e-05,
      "loss": 0.8339,
      "step": 9137
    },
    {
      "epoch": 1.4304946775203506,
      "grad_norm": 3.0659782886505127,
      "learning_rate": 8.917399804496577e-05,
      "loss": 1.0363,
      "step": 9138
    },
    {
      "epoch": 1.430651221039449,
      "grad_norm": 0.4259863793849945,
      "learning_rate": 8.914956011730205e-05,
      "loss": 0.1943,
      "step": 9139
    },
    {
      "epoch": 1.4308077645585473,
      "grad_norm": 0.8139044046401978,
      "learning_rate": 8.912512218963832e-05,
      "loss": 0.3264,
      "step": 9140
    },
    {
      "epoch": 1.4309643080776455,
      "grad_norm": 0.38787421584129333,
      "learning_rate": 8.910068426197457e-05,
      "loss": 0.2079,
      "step": 9141
    },
    {
      "epoch": 1.431120851596744,
      "grad_norm": 1.820959210395813,
      "learning_rate": 8.907624633431084e-05,
      "loss": 0.2371,
      "step": 9142
    },
    {
      "epoch": 1.4312773951158422,
      "grad_norm": 0.6413818597793579,
      "learning_rate": 8.905180840664711e-05,
      "loss": 0.2075,
      "step": 9143
    },
    {
      "epoch": 1.4314339386349406,
      "grad_norm": 0.7421493530273438,
      "learning_rate": 8.902737047898336e-05,
      "loss": 0.3897,
      "step": 9144
    },
    {
      "epoch": 1.4315904821540388,
      "grad_norm": 1.2470545768737793,
      "learning_rate": 8.900293255131964e-05,
      "loss": 0.2849,
      "step": 9145
    },
    {
      "epoch": 1.431747025673137,
      "grad_norm": 0.5670220255851746,
      "learning_rate": 8.89784946236559e-05,
      "loss": 0.2254,
      "step": 9146
    },
    {
      "epoch": 1.4319035691922355,
      "grad_norm": 0.7760573625564575,
      "learning_rate": 8.895405669599217e-05,
      "loss": 0.2496,
      "step": 9147
    },
    {
      "epoch": 1.4320601127113337,
      "grad_norm": 1.6587246656417847,
      "learning_rate": 8.892961876832845e-05,
      "loss": 0.5412,
      "step": 9148
    },
    {
      "epoch": 1.4322166562304322,
      "grad_norm": 0.6467307806015015,
      "learning_rate": 8.89051808406647e-05,
      "loss": 0.1799,
      "step": 9149
    },
    {
      "epoch": 1.4323731997495304,
      "grad_norm": 1.049241542816162,
      "learning_rate": 8.888074291300096e-05,
      "loss": 0.2824,
      "step": 9150
    },
    {
      "epoch": 1.4325297432686286,
      "grad_norm": 1.2838311195373535,
      "learning_rate": 8.885630498533724e-05,
      "loss": 0.4649,
      "step": 9151
    },
    {
      "epoch": 1.432686286787727,
      "grad_norm": 0.9091600775718689,
      "learning_rate": 8.88318670576735e-05,
      "loss": 0.392,
      "step": 9152
    },
    {
      "epoch": 1.4328428303068252,
      "grad_norm": 1.10474693775177,
      "learning_rate": 8.880742913000976e-05,
      "loss": 0.4008,
      "step": 9153
    },
    {
      "epoch": 1.4329993738259237,
      "grad_norm": 1.1482359170913696,
      "learning_rate": 8.878299120234604e-05,
      "loss": 0.3551,
      "step": 9154
    },
    {
      "epoch": 1.433155917345022,
      "grad_norm": 0.6834793090820312,
      "learning_rate": 8.87585532746823e-05,
      "loss": 0.2164,
      "step": 9155
    },
    {
      "epoch": 1.4333124608641201,
      "grad_norm": 0.9541566371917725,
      "learning_rate": 8.873411534701855e-05,
      "loss": 0.3163,
      "step": 9156
    },
    {
      "epoch": 1.4334690043832186,
      "grad_norm": 2.6311590671539307,
      "learning_rate": 8.870967741935483e-05,
      "loss": 0.3442,
      "step": 9157
    },
    {
      "epoch": 1.4336255479023168,
      "grad_norm": 1.168099045753479,
      "learning_rate": 8.86852394916911e-05,
      "loss": 0.3939,
      "step": 9158
    },
    {
      "epoch": 1.4337820914214152,
      "grad_norm": 1.946874737739563,
      "learning_rate": 8.866080156402736e-05,
      "loss": 0.4121,
      "step": 9159
    },
    {
      "epoch": 1.4339386349405134,
      "grad_norm": 1.7437493801116943,
      "learning_rate": 8.863636363636364e-05,
      "loss": 0.5132,
      "step": 9160
    },
    {
      "epoch": 1.4340951784596117,
      "grad_norm": 1.0139257907867432,
      "learning_rate": 8.861192570869989e-05,
      "loss": 0.2326,
      "step": 9161
    },
    {
      "epoch": 1.43425172197871,
      "grad_norm": 1.9671473503112793,
      "learning_rate": 8.858748778103615e-05,
      "loss": 0.5857,
      "step": 9162
    },
    {
      "epoch": 1.4344082654978083,
      "grad_norm": 7.848803520202637,
      "learning_rate": 8.856304985337243e-05,
      "loss": 0.4004,
      "step": 9163
    },
    {
      "epoch": 1.4345648090169068,
      "grad_norm": 1.3776273727416992,
      "learning_rate": 8.85386119257087e-05,
      "loss": 0.4384,
      "step": 9164
    },
    {
      "epoch": 1.434721352536005,
      "grad_norm": 1.2949081659317017,
      "learning_rate": 8.851417399804495e-05,
      "loss": 0.4875,
      "step": 9165
    },
    {
      "epoch": 1.4348778960551032,
      "grad_norm": 3.020263433456421,
      "learning_rate": 8.848973607038123e-05,
      "loss": 0.7237,
      "step": 9166
    },
    {
      "epoch": 1.4350344395742016,
      "grad_norm": 2.61968994140625,
      "learning_rate": 8.846529814271749e-05,
      "loss": 0.6773,
      "step": 9167
    },
    {
      "epoch": 1.4351909830933,
      "grad_norm": 1.7759416103363037,
      "learning_rate": 8.844086021505374e-05,
      "loss": 0.903,
      "step": 9168
    },
    {
      "epoch": 1.4353475266123983,
      "grad_norm": 1.5969222784042358,
      "learning_rate": 8.841642228739002e-05,
      "loss": 0.4451,
      "step": 9169
    },
    {
      "epoch": 1.4355040701314965,
      "grad_norm": 1.7365254163742065,
      "learning_rate": 8.839198435972629e-05,
      "loss": 0.4141,
      "step": 9170
    },
    {
      "epoch": 1.435660613650595,
      "grad_norm": 3.0730156898498535,
      "learning_rate": 8.836754643206255e-05,
      "loss": 1.1615,
      "step": 9171
    },
    {
      "epoch": 1.4358171571696932,
      "grad_norm": 2.1355032920837402,
      "learning_rate": 8.834310850439883e-05,
      "loss": 0.6218,
      "step": 9172
    },
    {
      "epoch": 1.4359737006887916,
      "grad_norm": 2.897294759750366,
      "learning_rate": 8.831867057673508e-05,
      "loss": 1.368,
      "step": 9173
    },
    {
      "epoch": 1.4361302442078898,
      "grad_norm": 1.462666630744934,
      "learning_rate": 8.829423264907135e-05,
      "loss": 0.4797,
      "step": 9174
    },
    {
      "epoch": 1.436286787726988,
      "grad_norm": 3.690739631652832,
      "learning_rate": 8.826979472140762e-05,
      "loss": 0.8913,
      "step": 9175
    },
    {
      "epoch": 1.4364433312460865,
      "grad_norm": 3.024446964263916,
      "learning_rate": 8.824535679374389e-05,
      "loss": 0.573,
      "step": 9176
    },
    {
      "epoch": 1.4365998747651847,
      "grad_norm": 1.713948130607605,
      "learning_rate": 8.822091886608014e-05,
      "loss": 0.5313,
      "step": 9177
    },
    {
      "epoch": 1.4367564182842831,
      "grad_norm": 3.0209193229675293,
      "learning_rate": 8.819648093841642e-05,
      "loss": 1.3517,
      "step": 9178
    },
    {
      "epoch": 1.4369129618033814,
      "grad_norm": 4.99892520904541,
      "learning_rate": 8.817204301075268e-05,
      "loss": 1.1816,
      "step": 9179
    },
    {
      "epoch": 1.4370695053224796,
      "grad_norm": 2.933112144470215,
      "learning_rate": 8.814760508308893e-05,
      "loss": 0.6138,
      "step": 9180
    },
    {
      "epoch": 1.437226048841578,
      "grad_norm": 1.887498378753662,
      "learning_rate": 8.812316715542521e-05,
      "loss": 1.1627,
      "step": 9181
    },
    {
      "epoch": 1.4373825923606762,
      "grad_norm": 2.869887351989746,
      "learning_rate": 8.809872922776148e-05,
      "loss": 0.8495,
      "step": 9182
    },
    {
      "epoch": 1.4375391358797747,
      "grad_norm": 2.865903854370117,
      "learning_rate": 8.807429130009774e-05,
      "loss": 0.8784,
      "step": 9183
    },
    {
      "epoch": 1.4376956793988729,
      "grad_norm": 1.4047309160232544,
      "learning_rate": 8.804985337243402e-05,
      "loss": 0.5029,
      "step": 9184
    },
    {
      "epoch": 1.437852222917971,
      "grad_norm": 1.5280976295471191,
      "learning_rate": 8.802541544477027e-05,
      "loss": 0.5816,
      "step": 9185
    },
    {
      "epoch": 1.4380087664370695,
      "grad_norm": 2.952573776245117,
      "learning_rate": 8.800097751710654e-05,
      "loss": 0.6115,
      "step": 9186
    },
    {
      "epoch": 1.4381653099561678,
      "grad_norm": 0.848155677318573,
      "learning_rate": 8.797653958944282e-05,
      "loss": 0.3277,
      "step": 9187
    },
    {
      "epoch": 1.4383218534752662,
      "grad_norm": 4.988794326782227,
      "learning_rate": 8.795210166177907e-05,
      "loss": 0.8794,
      "step": 9188
    },
    {
      "epoch": 1.4384783969943644,
      "grad_norm": 0.3419736325740814,
      "learning_rate": 8.792766373411533e-05,
      "loss": 0.2078,
      "step": 9189
    },
    {
      "epoch": 1.4386349405134626,
      "grad_norm": 0.519394040107727,
      "learning_rate": 8.790322580645161e-05,
      "loss": 0.2801,
      "step": 9190
    },
    {
      "epoch": 1.438791484032561,
      "grad_norm": 0.9486970901489258,
      "learning_rate": 8.787878787878787e-05,
      "loss": 0.3893,
      "step": 9191
    },
    {
      "epoch": 1.4389480275516593,
      "grad_norm": 0.5619981288909912,
      "learning_rate": 8.785434995112413e-05,
      "loss": 0.2653,
      "step": 9192
    },
    {
      "epoch": 1.4391045710707577,
      "grad_norm": 0.7176588773727417,
      "learning_rate": 8.78299120234604e-05,
      "loss": 0.3212,
      "step": 9193
    },
    {
      "epoch": 1.439261114589856,
      "grad_norm": 0.6982567310333252,
      "learning_rate": 8.780547409579667e-05,
      "loss": 0.1653,
      "step": 9194
    },
    {
      "epoch": 1.4394176581089542,
      "grad_norm": 0.5017626881599426,
      "learning_rate": 8.778103616813293e-05,
      "loss": 0.2405,
      "step": 9195
    },
    {
      "epoch": 1.4395742016280526,
      "grad_norm": 0.9920409917831421,
      "learning_rate": 8.775659824046921e-05,
      "loss": 0.3396,
      "step": 9196
    },
    {
      "epoch": 1.4397307451471508,
      "grad_norm": 0.9404519200325012,
      "learning_rate": 8.773216031280546e-05,
      "loss": 0.1906,
      "step": 9197
    },
    {
      "epoch": 1.4398872886662493,
      "grad_norm": 0.6449622511863708,
      "learning_rate": 8.770772238514173e-05,
      "loss": 0.2486,
      "step": 9198
    },
    {
      "epoch": 1.4400438321853475,
      "grad_norm": 0.9190917611122131,
      "learning_rate": 8.7683284457478e-05,
      "loss": 0.3537,
      "step": 9199
    },
    {
      "epoch": 1.4402003757044457,
      "grad_norm": 0.9570406079292297,
      "learning_rate": 8.765884652981426e-05,
      "loss": 0.2703,
      "step": 9200
    },
    {
      "epoch": 1.4403569192235441,
      "grad_norm": 1.3468104600906372,
      "learning_rate": 8.763440860215052e-05,
      "loss": 0.4503,
      "step": 9201
    },
    {
      "epoch": 1.4405134627426426,
      "grad_norm": 0.8499281406402588,
      "learning_rate": 8.76099706744868e-05,
      "loss": 0.3656,
      "step": 9202
    },
    {
      "epoch": 1.4406700062617408,
      "grad_norm": 1.0467718839645386,
      "learning_rate": 8.758553274682307e-05,
      "loss": 0.2962,
      "step": 9203
    },
    {
      "epoch": 1.440826549780839,
      "grad_norm": 0.8589369654655457,
      "learning_rate": 8.756109481915932e-05,
      "loss": 0.3295,
      "step": 9204
    },
    {
      "epoch": 1.4409830932999375,
      "grad_norm": 1.2273787260055542,
      "learning_rate": 8.75366568914956e-05,
      "loss": 0.4164,
      "step": 9205
    },
    {
      "epoch": 1.4411396368190357,
      "grad_norm": 1.0226304531097412,
      "learning_rate": 8.751221896383186e-05,
      "loss": 0.5016,
      "step": 9206
    },
    {
      "epoch": 1.4412961803381341,
      "grad_norm": 1.2805366516113281,
      "learning_rate": 8.748778103616812e-05,
      "loss": 0.482,
      "step": 9207
    },
    {
      "epoch": 1.4414527238572323,
      "grad_norm": 1.0432116985321045,
      "learning_rate": 8.74633431085044e-05,
      "loss": 0.3297,
      "step": 9208
    },
    {
      "epoch": 1.4416092673763305,
      "grad_norm": 3.9404122829437256,
      "learning_rate": 8.743890518084065e-05,
      "loss": 0.52,
      "step": 9209
    },
    {
      "epoch": 1.441765810895429,
      "grad_norm": 5.121649742126465,
      "learning_rate": 8.741446725317692e-05,
      "loss": 0.6131,
      "step": 9210
    },
    {
      "epoch": 1.4419223544145272,
      "grad_norm": 1.2252410650253296,
      "learning_rate": 8.73900293255132e-05,
      "loss": 0.4629,
      "step": 9211
    },
    {
      "epoch": 1.4420788979336256,
      "grad_norm": 1.7832077741622925,
      "learning_rate": 8.736559139784945e-05,
      "loss": 0.6523,
      "step": 9212
    },
    {
      "epoch": 1.4422354414527239,
      "grad_norm": 3.1595590114593506,
      "learning_rate": 8.734115347018571e-05,
      "loss": 0.3546,
      "step": 9213
    },
    {
      "epoch": 1.442391984971822,
      "grad_norm": 1.8645236492156982,
      "learning_rate": 8.731671554252199e-05,
      "loss": 0.9762,
      "step": 9214
    },
    {
      "epoch": 1.4425485284909205,
      "grad_norm": 2.1606290340423584,
      "learning_rate": 8.729227761485826e-05,
      "loss": 0.6703,
      "step": 9215
    },
    {
      "epoch": 1.4427050720100187,
      "grad_norm": 3.289787530899048,
      "learning_rate": 8.726783968719451e-05,
      "loss": 0.5665,
      "step": 9216
    },
    {
      "epoch": 1.4428616155291172,
      "grad_norm": 2.60479736328125,
      "learning_rate": 8.724340175953079e-05,
      "loss": 0.864,
      "step": 9217
    },
    {
      "epoch": 1.4430181590482154,
      "grad_norm": 2.7393736839294434,
      "learning_rate": 8.721896383186705e-05,
      "loss": 0.4926,
      "step": 9218
    },
    {
      "epoch": 1.4431747025673136,
      "grad_norm": 2.50663161277771,
      "learning_rate": 8.719452590420332e-05,
      "loss": 0.9697,
      "step": 9219
    },
    {
      "epoch": 1.443331246086412,
      "grad_norm": 1.2371330261230469,
      "learning_rate": 8.71700879765396e-05,
      "loss": 0.3924,
      "step": 9220
    },
    {
      "epoch": 1.4434877896055103,
      "grad_norm": 2.1638879776000977,
      "learning_rate": 8.714565004887585e-05,
      "loss": 0.6051,
      "step": 9221
    },
    {
      "epoch": 1.4436443331246087,
      "grad_norm": 1.8525105714797974,
      "learning_rate": 8.712121212121211e-05,
      "loss": 0.8727,
      "step": 9222
    },
    {
      "epoch": 1.443800876643707,
      "grad_norm": 4.00303840637207,
      "learning_rate": 8.709677419354839e-05,
      "loss": 0.9543,
      "step": 9223
    },
    {
      "epoch": 1.4439574201628051,
      "grad_norm": 1.3205608129501343,
      "learning_rate": 8.707233626588464e-05,
      "loss": 0.4322,
      "step": 9224
    },
    {
      "epoch": 1.4441139636819036,
      "grad_norm": 2.571690559387207,
      "learning_rate": 8.70478983382209e-05,
      "loss": 0.7875,
      "step": 9225
    },
    {
      "epoch": 1.4442705072010018,
      "grad_norm": 3.207707166671753,
      "learning_rate": 8.702346041055718e-05,
      "loss": 1.0932,
      "step": 9226
    },
    {
      "epoch": 1.4444270507201002,
      "grad_norm": 1.89205002784729,
      "learning_rate": 8.699902248289345e-05,
      "loss": 0.7563,
      "step": 9227
    },
    {
      "epoch": 1.4445835942391985,
      "grad_norm": 1.761757493019104,
      "learning_rate": 8.69745845552297e-05,
      "loss": 0.8936,
      "step": 9228
    },
    {
      "epoch": 1.4447401377582967,
      "grad_norm": 4.034841060638428,
      "learning_rate": 8.695014662756598e-05,
      "loss": 1.1292,
      "step": 9229
    },
    {
      "epoch": 1.4448966812773951,
      "grad_norm": 5.492170333862305,
      "learning_rate": 8.692570869990224e-05,
      "loss": 2.0323,
      "step": 9230
    },
    {
      "epoch": 1.4450532247964936,
      "grad_norm": 3.997959613800049,
      "learning_rate": 8.690127077223851e-05,
      "loss": 0.9711,
      "step": 9231
    },
    {
      "epoch": 1.4452097683155918,
      "grad_norm": 3.610450267791748,
      "learning_rate": 8.687683284457477e-05,
      "loss": 1.0842,
      "step": 9232
    },
    {
      "epoch": 1.44536631183469,
      "grad_norm": 2.2445812225341797,
      "learning_rate": 8.685239491691104e-05,
      "loss": 1.499,
      "step": 9233
    },
    {
      "epoch": 1.4455228553537882,
      "grad_norm": 4.479135513305664,
      "learning_rate": 8.68279569892473e-05,
      "loss": 0.4075,
      "step": 9234
    },
    {
      "epoch": 1.4456793988728867,
      "grad_norm": 1.9963476657867432,
      "learning_rate": 8.680351906158358e-05,
      "loss": 0.9783,
      "step": 9235
    },
    {
      "epoch": 1.445835942391985,
      "grad_norm": 1.3088605403900146,
      "learning_rate": 8.677908113391983e-05,
      "loss": 0.4627,
      "step": 9236
    },
    {
      "epoch": 1.4459924859110833,
      "grad_norm": 3.202507257461548,
      "learning_rate": 8.67546432062561e-05,
      "loss": 1.1461,
      "step": 9237
    },
    {
      "epoch": 1.4461490294301815,
      "grad_norm": 2.1418967247009277,
      "learning_rate": 8.673020527859237e-05,
      "loss": 0.9054,
      "step": 9238
    },
    {
      "epoch": 1.44630557294928,
      "grad_norm": 0.5720778703689575,
      "learning_rate": 8.670576735092864e-05,
      "loss": 0.27,
      "step": 9239
    },
    {
      "epoch": 1.4464621164683782,
      "grad_norm": 0.45177343487739563,
      "learning_rate": 8.668132942326489e-05,
      "loss": 0.1783,
      "step": 9240
    },
    {
      "epoch": 1.4466186599874766,
      "grad_norm": 1.081150770187378,
      "learning_rate": 8.665689149560117e-05,
      "loss": 0.4301,
      "step": 9241
    },
    {
      "epoch": 1.4467752035065748,
      "grad_norm": 0.7842183113098145,
      "learning_rate": 8.663245356793743e-05,
      "loss": 0.2019,
      "step": 9242
    },
    {
      "epoch": 1.446931747025673,
      "grad_norm": 0.7052162885665894,
      "learning_rate": 8.66080156402737e-05,
      "loss": 0.2977,
      "step": 9243
    },
    {
      "epoch": 1.4470882905447715,
      "grad_norm": 0.5708596110343933,
      "learning_rate": 8.658357771260996e-05,
      "loss": 0.1986,
      "step": 9244
    },
    {
      "epoch": 1.4472448340638697,
      "grad_norm": 0.4889955222606659,
      "learning_rate": 8.655913978494623e-05,
      "loss": 0.1634,
      "step": 9245
    },
    {
      "epoch": 1.4474013775829682,
      "grad_norm": 0.5861791372299194,
      "learning_rate": 8.653470185728249e-05,
      "loss": 0.156,
      "step": 9246
    },
    {
      "epoch": 1.4475579211020664,
      "grad_norm": 0.8749104142189026,
      "learning_rate": 8.651026392961877e-05,
      "loss": 0.2483,
      "step": 9247
    },
    {
      "epoch": 1.4477144646211646,
      "grad_norm": 0.7886466383934021,
      "learning_rate": 8.648582600195502e-05,
      "loss": 0.339,
      "step": 9248
    },
    {
      "epoch": 1.447871008140263,
      "grad_norm": 1.1284881830215454,
      "learning_rate": 8.646138807429129e-05,
      "loss": 0.3371,
      "step": 9249
    },
    {
      "epoch": 1.4480275516593613,
      "grad_norm": 11.110546112060547,
      "learning_rate": 8.643695014662757e-05,
      "loss": 0.2101,
      "step": 9250
    },
    {
      "epoch": 1.4481840951784597,
      "grad_norm": 0.8709673881530762,
      "learning_rate": 8.641251221896383e-05,
      "loss": 0.3903,
      "step": 9251
    },
    {
      "epoch": 1.448340638697558,
      "grad_norm": 7.446573257446289,
      "learning_rate": 8.638807429130008e-05,
      "loss": 0.495,
      "step": 9252
    },
    {
      "epoch": 1.4484971822166561,
      "grad_norm": 1.0395238399505615,
      "learning_rate": 8.636363636363636e-05,
      "loss": 0.3358,
      "step": 9253
    },
    {
      "epoch": 1.4486537257357546,
      "grad_norm": 1.0077224969863892,
      "learning_rate": 8.633919843597262e-05,
      "loss": 0.3986,
      "step": 9254
    },
    {
      "epoch": 1.4488102692548528,
      "grad_norm": 1.1264681816101074,
      "learning_rate": 8.631476050830889e-05,
      "loss": 0.2916,
      "step": 9255
    },
    {
      "epoch": 1.4489668127739512,
      "grad_norm": 1.2706413269042969,
      "learning_rate": 8.629032258064515e-05,
      "loss": 0.3287,
      "step": 9256
    },
    {
      "epoch": 1.4491233562930494,
      "grad_norm": 1.2557499408721924,
      "learning_rate": 8.626588465298142e-05,
      "loss": 0.3033,
      "step": 9257
    },
    {
      "epoch": 1.4492798998121477,
      "grad_norm": 0.9736500382423401,
      "learning_rate": 8.624144672531768e-05,
      "loss": 0.5071,
      "step": 9258
    },
    {
      "epoch": 1.449436443331246,
      "grad_norm": 1.106518030166626,
      "learning_rate": 8.621700879765396e-05,
      "loss": 0.3498,
      "step": 9259
    },
    {
      "epoch": 1.4495929868503443,
      "grad_norm": 1.3853764533996582,
      "learning_rate": 8.619257086999021e-05,
      "loss": 0.462,
      "step": 9260
    },
    {
      "epoch": 1.4497495303694428,
      "grad_norm": 1.618614912033081,
      "learning_rate": 8.616813294232648e-05,
      "loss": 0.5758,
      "step": 9261
    },
    {
      "epoch": 1.449906073888541,
      "grad_norm": 1.4383054971694946,
      "learning_rate": 8.614369501466276e-05,
      "loss": 0.4159,
      "step": 9262
    },
    {
      "epoch": 1.4500626174076392,
      "grad_norm": 1.0663392543792725,
      "learning_rate": 8.611925708699902e-05,
      "loss": 0.4319,
      "step": 9263
    },
    {
      "epoch": 1.4502191609267376,
      "grad_norm": 2.9339816570281982,
      "learning_rate": 8.609481915933527e-05,
      "loss": 0.6587,
      "step": 9264
    },
    {
      "epoch": 1.450375704445836,
      "grad_norm": 2.176449775695801,
      "learning_rate": 8.607038123167155e-05,
      "loss": 0.6003,
      "step": 9265
    },
    {
      "epoch": 1.4505322479649343,
      "grad_norm": 2.8206968307495117,
      "learning_rate": 8.604594330400782e-05,
      "loss": 0.6367,
      "step": 9266
    },
    {
      "epoch": 1.4506887914840325,
      "grad_norm": 1.6528189182281494,
      "learning_rate": 8.602150537634408e-05,
      "loss": 0.4951,
      "step": 9267
    },
    {
      "epoch": 1.4508453350031307,
      "grad_norm": 2.3281638622283936,
      "learning_rate": 8.599706744868035e-05,
      "loss": 0.7283,
      "step": 9268
    },
    {
      "epoch": 1.4510018785222292,
      "grad_norm": 1.1274622678756714,
      "learning_rate": 8.597262952101661e-05,
      "loss": 0.3834,
      "step": 9269
    },
    {
      "epoch": 1.4511584220413276,
      "grad_norm": 2.182931423187256,
      "learning_rate": 8.594819159335287e-05,
      "loss": 0.7657,
      "step": 9270
    },
    {
      "epoch": 1.4513149655604258,
      "grad_norm": 2.473872423171997,
      "learning_rate": 8.592375366568915e-05,
      "loss": 0.8096,
      "step": 9271
    },
    {
      "epoch": 1.451471509079524,
      "grad_norm": 1.8091752529144287,
      "learning_rate": 8.58993157380254e-05,
      "loss": 0.6692,
      "step": 9272
    },
    {
      "epoch": 1.4516280525986225,
      "grad_norm": 3.2494728565216064,
      "learning_rate": 8.587487781036167e-05,
      "loss": 0.5206,
      "step": 9273
    },
    {
      "epoch": 1.4517845961177207,
      "grad_norm": 6.569478511810303,
      "learning_rate": 8.585043988269795e-05,
      "loss": 0.8174,
      "step": 9274
    },
    {
      "epoch": 1.4519411396368191,
      "grad_norm": 3.5665347576141357,
      "learning_rate": 8.582600195503421e-05,
      "loss": 1.2162,
      "step": 9275
    },
    {
      "epoch": 1.4520976831559174,
      "grad_norm": 3.3499755859375,
      "learning_rate": 8.580156402737046e-05,
      "loss": 1.2148,
      "step": 9276
    },
    {
      "epoch": 1.4522542266750156,
      "grad_norm": 1.8471848964691162,
      "learning_rate": 8.577712609970674e-05,
      "loss": 0.6484,
      "step": 9277
    },
    {
      "epoch": 1.452410770194114,
      "grad_norm": 3.943901777267456,
      "learning_rate": 8.575268817204301e-05,
      "loss": 0.8956,
      "step": 9278
    },
    {
      "epoch": 1.4525673137132122,
      "grad_norm": 3.428492784500122,
      "learning_rate": 8.572825024437927e-05,
      "loss": 1.4366,
      "step": 9279
    },
    {
      "epoch": 1.4527238572323107,
      "grad_norm": 2.345398426055908,
      "learning_rate": 8.570381231671554e-05,
      "loss": 0.7291,
      "step": 9280
    },
    {
      "epoch": 1.452880400751409,
      "grad_norm": 2.542097806930542,
      "learning_rate": 8.56793743890518e-05,
      "loss": 0.9401,
      "step": 9281
    },
    {
      "epoch": 1.4530369442705071,
      "grad_norm": 1.3730517625808716,
      "learning_rate": 8.565493646138807e-05,
      "loss": 1.1182,
      "step": 9282
    },
    {
      "epoch": 1.4531934877896056,
      "grad_norm": 3.189399003982544,
      "learning_rate": 8.563049853372434e-05,
      "loss": 1.2957,
      "step": 9283
    },
    {
      "epoch": 1.4533500313087038,
      "grad_norm": 2.0311431884765625,
      "learning_rate": 8.56060606060606e-05,
      "loss": 0.4811,
      "step": 9284
    },
    {
      "epoch": 1.4535065748278022,
      "grad_norm": 1.2809655666351318,
      "learning_rate": 8.558162267839686e-05,
      "loss": 0.7376,
      "step": 9285
    },
    {
      "epoch": 1.4536631183469004,
      "grad_norm": 1.898149013519287,
      "learning_rate": 8.555718475073314e-05,
      "loss": 0.5024,
      "step": 9286
    },
    {
      "epoch": 1.4538196618659986,
      "grad_norm": 1.5057889223098755,
      "learning_rate": 8.55327468230694e-05,
      "loss": 0.6526,
      "step": 9287
    },
    {
      "epoch": 1.453976205385097,
      "grad_norm": 6.357172012329102,
      "learning_rate": 8.550830889540565e-05,
      "loss": 1.1135,
      "step": 9288
    },
    {
      "epoch": 1.4541327489041953,
      "grad_norm": 0.4889274537563324,
      "learning_rate": 8.548387096774193e-05,
      "loss": 0.1894,
      "step": 9289
    },
    {
      "epoch": 1.4542892924232937,
      "grad_norm": 0.653469443321228,
      "learning_rate": 8.54594330400782e-05,
      "loss": 0.1703,
      "step": 9290
    },
    {
      "epoch": 1.454445835942392,
      "grad_norm": 0.6301447153091431,
      "learning_rate": 8.543499511241445e-05,
      "loss": 0.2752,
      "step": 9291
    },
    {
      "epoch": 1.4546023794614902,
      "grad_norm": 0.5737555027008057,
      "learning_rate": 8.541055718475073e-05,
      "loss": 0.2306,
      "step": 9292
    },
    {
      "epoch": 1.4547589229805886,
      "grad_norm": 0.48317751288414,
      "learning_rate": 8.538611925708699e-05,
      "loss": 0.2238,
      "step": 9293
    },
    {
      "epoch": 1.4549154664996868,
      "grad_norm": 0.660159170627594,
      "learning_rate": 8.536168132942326e-05,
      "loss": 0.25,
      "step": 9294
    },
    {
      "epoch": 1.4550720100187853,
      "grad_norm": 0.5787570476531982,
      "learning_rate": 8.533724340175954e-05,
      "loss": 0.1323,
      "step": 9295
    },
    {
      "epoch": 1.4552285535378835,
      "grad_norm": 0.7087740898132324,
      "learning_rate": 8.531280547409579e-05,
      "loss": 0.2394,
      "step": 9296
    },
    {
      "epoch": 1.4553850970569817,
      "grad_norm": 1.1337648630142212,
      "learning_rate": 8.528836754643205e-05,
      "loss": 0.2902,
      "step": 9297
    },
    {
      "epoch": 1.4555416405760802,
      "grad_norm": 0.6737481951713562,
      "learning_rate": 8.526392961876833e-05,
      "loss": 0.2308,
      "step": 9298
    },
    {
      "epoch": 1.4556981840951786,
      "grad_norm": 0.6452515721321106,
      "learning_rate": 8.52394916911046e-05,
      "loss": 0.1904,
      "step": 9299
    },
    {
      "epoch": 1.4558547276142768,
      "grad_norm": 0.5262974500656128,
      "learning_rate": 8.521505376344085e-05,
      "loss": 0.1995,
      "step": 9300
    },
    {
      "epoch": 1.456011271133375,
      "grad_norm": 1.1509872674942017,
      "learning_rate": 8.519061583577712e-05,
      "loss": 0.4979,
      "step": 9301
    },
    {
      "epoch": 1.4561678146524735,
      "grad_norm": 0.8436789512634277,
      "learning_rate": 8.516617790811339e-05,
      "loss": 0.2421,
      "step": 9302
    },
    {
      "epoch": 1.4563243581715717,
      "grad_norm": 1.1247527599334717,
      "learning_rate": 8.514173998044964e-05,
      "loss": 0.2879,
      "step": 9303
    },
    {
      "epoch": 1.4564809016906701,
      "grad_norm": 1.1355880498886108,
      "learning_rate": 8.511730205278592e-05,
      "loss": 0.328,
      "step": 9304
    },
    {
      "epoch": 1.4566374452097683,
      "grad_norm": 4.993597984313965,
      "learning_rate": 8.509286412512218e-05,
      "loss": 0.7011,
      "step": 9305
    },
    {
      "epoch": 1.4567939887288666,
      "grad_norm": 0.7840385437011719,
      "learning_rate": 8.506842619745845e-05,
      "loss": 0.1718,
      "step": 9306
    },
    {
      "epoch": 1.456950532247965,
      "grad_norm": 1.4984678030014038,
      "learning_rate": 8.504398826979473e-05,
      "loss": 0.5557,
      "step": 9307
    },
    {
      "epoch": 1.4571070757670632,
      "grad_norm": 1.5178111791610718,
      "learning_rate": 8.501955034213098e-05,
      "loss": 0.548,
      "step": 9308
    },
    {
      "epoch": 1.4572636192861617,
      "grad_norm": 2.9656150341033936,
      "learning_rate": 8.499511241446724e-05,
      "loss": 0.5248,
      "step": 9309
    },
    {
      "epoch": 1.4574201628052599,
      "grad_norm": 1.327226161956787,
      "learning_rate": 8.497067448680352e-05,
      "loss": 0.7294,
      "step": 9310
    },
    {
      "epoch": 1.457576706324358,
      "grad_norm": 1.0923349857330322,
      "learning_rate": 8.494623655913979e-05,
      "loss": 0.4204,
      "step": 9311
    },
    {
      "epoch": 1.4577332498434565,
      "grad_norm": 2.7213759422302246,
      "learning_rate": 8.492179863147604e-05,
      "loss": 0.7072,
      "step": 9312
    },
    {
      "epoch": 1.4578897933625548,
      "grad_norm": 1.255927324295044,
      "learning_rate": 8.489736070381232e-05,
      "loss": 0.7492,
      "step": 9313
    },
    {
      "epoch": 1.4580463368816532,
      "grad_norm": 2.039738416671753,
      "learning_rate": 8.487292277614858e-05,
      "loss": 0.5131,
      "step": 9314
    },
    {
      "epoch": 1.4582028804007514,
      "grad_norm": 2.313304901123047,
      "learning_rate": 8.484848484848483e-05,
      "loss": 0.4055,
      "step": 9315
    },
    {
      "epoch": 1.4583594239198496,
      "grad_norm": 1.3143826723098755,
      "learning_rate": 8.482404692082111e-05,
      "loss": 0.5852,
      "step": 9316
    },
    {
      "epoch": 1.458515967438948,
      "grad_norm": 2.0478479862213135,
      "learning_rate": 8.479960899315737e-05,
      "loss": 0.8262,
      "step": 9317
    },
    {
      "epoch": 1.4586725109580463,
      "grad_norm": 3.0443527698516846,
      "learning_rate": 8.477517106549364e-05,
      "loss": 0.9234,
      "step": 9318
    },
    {
      "epoch": 1.4588290544771447,
      "grad_norm": 5.6164727210998535,
      "learning_rate": 8.475073313782992e-05,
      "loss": 0.708,
      "step": 9319
    },
    {
      "epoch": 1.458985597996243,
      "grad_norm": 1.4889097213745117,
      "learning_rate": 8.472629521016617e-05,
      "loss": 0.9464,
      "step": 9320
    },
    {
      "epoch": 1.4591421415153412,
      "grad_norm": 2.5142250061035156,
      "learning_rate": 8.470185728250243e-05,
      "loss": 1.1254,
      "step": 9321
    },
    {
      "epoch": 1.4592986850344396,
      "grad_norm": 3.2503488063812256,
      "learning_rate": 8.467741935483871e-05,
      "loss": 1.037,
      "step": 9322
    },
    {
      "epoch": 1.4594552285535378,
      "grad_norm": 2.1030449867248535,
      "learning_rate": 8.465298142717498e-05,
      "loss": 0.8223,
      "step": 9323
    },
    {
      "epoch": 1.4596117720726363,
      "grad_norm": 1.6335690021514893,
      "learning_rate": 8.462854349951123e-05,
      "loss": 0.5398,
      "step": 9324
    },
    {
      "epoch": 1.4597683155917345,
      "grad_norm": 2.1362364292144775,
      "learning_rate": 8.46041055718475e-05,
      "loss": 0.3341,
      "step": 9325
    },
    {
      "epoch": 1.4599248591108327,
      "grad_norm": 2.837796449661255,
      "learning_rate": 8.457966764418377e-05,
      "loss": 1.4995,
      "step": 9326
    },
    {
      "epoch": 1.4600814026299311,
      "grad_norm": 4.29352331161499,
      "learning_rate": 8.455522971652002e-05,
      "loss": 1.3477,
      "step": 9327
    },
    {
      "epoch": 1.4602379461490294,
      "grad_norm": 1.990387201309204,
      "learning_rate": 8.45307917888563e-05,
      "loss": 0.8274,
      "step": 9328
    },
    {
      "epoch": 1.4603944896681278,
      "grad_norm": 1.3266398906707764,
      "learning_rate": 8.450635386119257e-05,
      "loss": 0.6768,
      "step": 9329
    },
    {
      "epoch": 1.460551033187226,
      "grad_norm": 4.102444171905518,
      "learning_rate": 8.448191593352883e-05,
      "loss": 0.9489,
      "step": 9330
    },
    {
      "epoch": 1.4607075767063242,
      "grad_norm": 2.765338897705078,
      "learning_rate": 8.445747800586511e-05,
      "loss": 0.8909,
      "step": 9331
    },
    {
      "epoch": 1.4608641202254227,
      "grad_norm": 2.984898805618286,
      "learning_rate": 8.443304007820136e-05,
      "loss": 1.2414,
      "step": 9332
    },
    {
      "epoch": 1.461020663744521,
      "grad_norm": 4.269297122955322,
      "learning_rate": 8.440860215053763e-05,
      "loss": 1.6595,
      "step": 9333
    },
    {
      "epoch": 1.4611772072636193,
      "grad_norm": 1.1461013555526733,
      "learning_rate": 8.43841642228739e-05,
      "loss": 0.1954,
      "step": 9334
    },
    {
      "epoch": 1.4613337507827175,
      "grad_norm": 2.7744386196136475,
      "learning_rate": 8.435972629521017e-05,
      "loss": 0.7527,
      "step": 9335
    },
    {
      "epoch": 1.461490294301816,
      "grad_norm": 1.470504641532898,
      "learning_rate": 8.433528836754642e-05,
      "loss": 0.7368,
      "step": 9336
    },
    {
      "epoch": 1.4616468378209142,
      "grad_norm": 2.8413960933685303,
      "learning_rate": 8.431085043988268e-05,
      "loss": 0.3352,
      "step": 9337
    },
    {
      "epoch": 1.4618033813400126,
      "grad_norm": 1.2316049337387085,
      "learning_rate": 8.428641251221896e-05,
      "loss": 0.5112,
      "step": 9338
    },
    {
      "epoch": 1.4619599248591109,
      "grad_norm": 0.4800587594509125,
      "learning_rate": 8.426197458455521e-05,
      "loss": 0.2659,
      "step": 9339
    },
    {
      "epoch": 1.462116468378209,
      "grad_norm": 0.5300424695014954,
      "learning_rate": 8.423753665689148e-05,
      "loss": 0.2848,
      "step": 9340
    },
    {
      "epoch": 1.4622730118973075,
      "grad_norm": 0.4099821448326111,
      "learning_rate": 8.421309872922776e-05,
      "loss": 0.1509,
      "step": 9341
    },
    {
      "epoch": 1.4624295554164057,
      "grad_norm": 0.5445888042449951,
      "learning_rate": 8.418866080156402e-05,
      "loss": 0.2769,
      "step": 9342
    },
    {
      "epoch": 1.4625860989355042,
      "grad_norm": 0.7307244539260864,
      "learning_rate": 8.416422287390027e-05,
      "loss": 0.24,
      "step": 9343
    },
    {
      "epoch": 1.4627426424546024,
      "grad_norm": 0.590489387512207,
      "learning_rate": 8.413978494623655e-05,
      "loss": 0.174,
      "step": 9344
    },
    {
      "epoch": 1.4628991859737006,
      "grad_norm": 0.8075463175773621,
      "learning_rate": 8.411534701857282e-05,
      "loss": 0.2945,
      "step": 9345
    },
    {
      "epoch": 1.463055729492799,
      "grad_norm": 0.6793466806411743,
      "learning_rate": 8.409090909090908e-05,
      "loss": 0.2182,
      "step": 9346
    },
    {
      "epoch": 1.4632122730118973,
      "grad_norm": 0.9996579885482788,
      "learning_rate": 8.406647116324535e-05,
      "loss": 0.347,
      "step": 9347
    },
    {
      "epoch": 1.4633688165309957,
      "grad_norm": 0.5806265473365784,
      "learning_rate": 8.404203323558161e-05,
      "loss": 0.2452,
      "step": 9348
    },
    {
      "epoch": 1.463525360050094,
      "grad_norm": 0.9214386343955994,
      "learning_rate": 8.401759530791788e-05,
      "loss": 0.3277,
      "step": 9349
    },
    {
      "epoch": 1.4636819035691921,
      "grad_norm": 2.849910020828247,
      "learning_rate": 8.399315738025415e-05,
      "loss": 0.7389,
      "step": 9350
    },
    {
      "epoch": 1.4638384470882906,
      "grad_norm": 0.5131305456161499,
      "learning_rate": 8.39687194525904e-05,
      "loss": 0.2119,
      "step": 9351
    },
    {
      "epoch": 1.4639949906073888,
      "grad_norm": 2.853576898574829,
      "learning_rate": 8.394428152492667e-05,
      "loss": 0.3357,
      "step": 9352
    },
    {
      "epoch": 1.4641515341264872,
      "grad_norm": 1.252537488937378,
      "learning_rate": 8.391984359726295e-05,
      "loss": 0.2544,
      "step": 9353
    },
    {
      "epoch": 1.4643080776455855,
      "grad_norm": 1.4162060022354126,
      "learning_rate": 8.389540566959921e-05,
      "loss": 0.4257,
      "step": 9354
    },
    {
      "epoch": 1.4644646211646837,
      "grad_norm": 1.189596176147461,
      "learning_rate": 8.387096774193546e-05,
      "loss": 0.4705,
      "step": 9355
    },
    {
      "epoch": 1.4646211646837821,
      "grad_norm": 0.7394349575042725,
      "learning_rate": 8.384652981427174e-05,
      "loss": 0.411,
      "step": 9356
    },
    {
      "epoch": 1.4647777082028803,
      "grad_norm": 1.2858307361602783,
      "learning_rate": 8.382209188660801e-05,
      "loss": 0.5221,
      "step": 9357
    },
    {
      "epoch": 1.4649342517219788,
      "grad_norm": 1.020027756690979,
      "learning_rate": 8.379765395894427e-05,
      "loss": 0.407,
      "step": 9358
    },
    {
      "epoch": 1.465090795241077,
      "grad_norm": 1.9283796548843384,
      "learning_rate": 8.377321603128054e-05,
      "loss": 0.4782,
      "step": 9359
    },
    {
      "epoch": 1.4652473387601752,
      "grad_norm": 1.3848600387573242,
      "learning_rate": 8.37487781036168e-05,
      "loss": 0.4688,
      "step": 9360
    },
    {
      "epoch": 1.4654038822792737,
      "grad_norm": 1.769733190536499,
      "learning_rate": 8.372434017595307e-05,
      "loss": 0.8337,
      "step": 9361
    },
    {
      "epoch": 1.4655604257983719,
      "grad_norm": 1.7718392610549927,
      "learning_rate": 8.369990224828934e-05,
      "loss": 0.1812,
      "step": 9362
    },
    {
      "epoch": 1.4657169693174703,
      "grad_norm": 2.2144126892089844,
      "learning_rate": 8.36754643206256e-05,
      "loss": 0.5035,
      "step": 9363
    },
    {
      "epoch": 1.4658735128365685,
      "grad_norm": 0.7767239212989807,
      "learning_rate": 8.365102639296186e-05,
      "loss": 0.3048,
      "step": 9364
    },
    {
      "epoch": 1.4660300563556667,
      "grad_norm": 3.264582872390747,
      "learning_rate": 8.362658846529814e-05,
      "loss": 0.7307,
      "step": 9365
    },
    {
      "epoch": 1.4661865998747652,
      "grad_norm": 1.8177357912063599,
      "learning_rate": 8.36021505376344e-05,
      "loss": 0.6826,
      "step": 9366
    },
    {
      "epoch": 1.4663431433938636,
      "grad_norm": 3.0217385292053223,
      "learning_rate": 8.357771260997066e-05,
      "loss": 0.4975,
      "step": 9367
    },
    {
      "epoch": 1.4664996869129618,
      "grad_norm": 1.1318516731262207,
      "learning_rate": 8.355327468230693e-05,
      "loss": 0.3554,
      "step": 9368
    },
    {
      "epoch": 1.46665623043206,
      "grad_norm": 1.6963642835617065,
      "learning_rate": 8.35288367546432e-05,
      "loss": 0.579,
      "step": 9369
    },
    {
      "epoch": 1.4668127739511585,
      "grad_norm": 2.232435941696167,
      "learning_rate": 8.350439882697946e-05,
      "loss": 0.6147,
      "step": 9370
    },
    {
      "epoch": 1.4669693174702567,
      "grad_norm": 3.5765178203582764,
      "learning_rate": 8.347996089931573e-05,
      "loss": 1.0373,
      "step": 9371
    },
    {
      "epoch": 1.4671258609893552,
      "grad_norm": 1.6168148517608643,
      "learning_rate": 8.345552297165199e-05,
      "loss": 0.3683,
      "step": 9372
    },
    {
      "epoch": 1.4672824045084534,
      "grad_norm": 1.8201894760131836,
      "learning_rate": 8.343108504398826e-05,
      "loss": 0.5754,
      "step": 9373
    },
    {
      "epoch": 1.4674389480275516,
      "grad_norm": 2.91982364654541,
      "learning_rate": 8.340664711632454e-05,
      "loss": 0.6643,
      "step": 9374
    },
    {
      "epoch": 1.46759549154665,
      "grad_norm": 2.993435859680176,
      "learning_rate": 8.338220918866079e-05,
      "loss": 0.624,
      "step": 9375
    },
    {
      "epoch": 1.4677520350657483,
      "grad_norm": 4.31162691116333,
      "learning_rate": 8.335777126099705e-05,
      "loss": 0.7056,
      "step": 9376
    },
    {
      "epoch": 1.4679085785848467,
      "grad_norm": 3.5542516708374023,
      "learning_rate": 8.333333333333333e-05,
      "loss": 1.2868,
      "step": 9377
    },
    {
      "epoch": 1.468065122103945,
      "grad_norm": 2.0428109169006348,
      "learning_rate": 8.33088954056696e-05,
      "loss": 0.9045,
      "step": 9378
    },
    {
      "epoch": 1.4682216656230431,
      "grad_norm": 2.306626796722412,
      "learning_rate": 8.328445747800585e-05,
      "loss": 0.8775,
      "step": 9379
    },
    {
      "epoch": 1.4683782091421416,
      "grad_norm": 2.654895544052124,
      "learning_rate": 8.326001955034212e-05,
      "loss": 1.4198,
      "step": 9380
    },
    {
      "epoch": 1.4685347526612398,
      "grad_norm": 2.5380771160125732,
      "learning_rate": 8.323558162267839e-05,
      "loss": 0.9698,
      "step": 9381
    },
    {
      "epoch": 1.4686912961803382,
      "grad_norm": 3.0304949283599854,
      "learning_rate": 8.321114369501465e-05,
      "loss": 1.164,
      "step": 9382
    },
    {
      "epoch": 1.4688478396994364,
      "grad_norm": 2.77571177482605,
      "learning_rate": 8.318670576735092e-05,
      "loss": 1.1842,
      "step": 9383
    },
    {
      "epoch": 1.4690043832185347,
      "grad_norm": 2.4832112789154053,
      "learning_rate": 8.316226783968718e-05,
      "loss": 1.0286,
      "step": 9384
    },
    {
      "epoch": 1.469160926737633,
      "grad_norm": 4.166812419891357,
      "learning_rate": 8.313782991202345e-05,
      "loss": 1.3073,
      "step": 9385
    },
    {
      "epoch": 1.4693174702567313,
      "grad_norm": 1.6507247686386108,
      "learning_rate": 8.311339198435973e-05,
      "loss": 0.5474,
      "step": 9386
    },
    {
      "epoch": 1.4694740137758298,
      "grad_norm": 4.904731273651123,
      "learning_rate": 8.308895405669598e-05,
      "loss": 0.8368,
      "step": 9387
    },
    {
      "epoch": 1.469630557294928,
      "grad_norm": 1.261414647102356,
      "learning_rate": 8.306451612903224e-05,
      "loss": 0.5796,
      "step": 9388
    },
    {
      "epoch": 1.4697871008140262,
      "grad_norm": 0.3774799704551697,
      "learning_rate": 8.304007820136852e-05,
      "loss": 0.2298,
      "step": 9389
    },
    {
      "epoch": 1.4699436443331246,
      "grad_norm": 0.3805764615535736,
      "learning_rate": 8.301564027370479e-05,
      "loss": 0.1553,
      "step": 9390
    },
    {
      "epoch": 1.4701001878522229,
      "grad_norm": 0.48448646068573,
      "learning_rate": 8.299120234604104e-05,
      "loss": 0.2439,
      "step": 9391
    },
    {
      "epoch": 1.4702567313713213,
      "grad_norm": 0.5146775841712952,
      "learning_rate": 8.296676441837732e-05,
      "loss": 0.2297,
      "step": 9392
    },
    {
      "epoch": 1.4704132748904195,
      "grad_norm": 0.5531075596809387,
      "learning_rate": 8.294232649071358e-05,
      "loss": 0.2025,
      "step": 9393
    },
    {
      "epoch": 1.4705698184095177,
      "grad_norm": 0.606508195400238,
      "learning_rate": 8.291788856304985e-05,
      "loss": 0.2208,
      "step": 9394
    },
    {
      "epoch": 1.4707263619286162,
      "grad_norm": 0.7570547461509705,
      "learning_rate": 8.289345063538611e-05,
      "loss": 0.253,
      "step": 9395
    },
    {
      "epoch": 1.4708829054477144,
      "grad_norm": 0.5803053379058838,
      "learning_rate": 8.286901270772238e-05,
      "loss": 0.2676,
      "step": 9396
    },
    {
      "epoch": 1.4710394489668128,
      "grad_norm": 0.5528594255447388,
      "learning_rate": 8.284457478005864e-05,
      "loss": 0.1975,
      "step": 9397
    },
    {
      "epoch": 1.471195992485911,
      "grad_norm": 0.7934523820877075,
      "learning_rate": 8.282013685239492e-05,
      "loss": 0.2068,
      "step": 9398
    },
    {
      "epoch": 1.4713525360050093,
      "grad_norm": 0.7662394046783447,
      "learning_rate": 8.279569892473117e-05,
      "loss": 0.2537,
      "step": 9399
    },
    {
      "epoch": 1.4715090795241077,
      "grad_norm": 1.507385492324829,
      "learning_rate": 8.277126099706743e-05,
      "loss": 0.247,
      "step": 9400
    },
    {
      "epoch": 1.4716656230432061,
      "grad_norm": 0.87679523229599,
      "learning_rate": 8.274682306940371e-05,
      "loss": 0.2944,
      "step": 9401
    },
    {
      "epoch": 1.4718221665623044,
      "grad_norm": 1.0118069648742676,
      "learning_rate": 8.272238514173998e-05,
      "loss": 0.3211,
      "step": 9402
    },
    {
      "epoch": 1.4719787100814026,
      "grad_norm": 0.34879270195961,
      "learning_rate": 8.269794721407623e-05,
      "loss": 0.1825,
      "step": 9403
    },
    {
      "epoch": 1.472135253600501,
      "grad_norm": 0.8371734619140625,
      "learning_rate": 8.267350928641251e-05,
      "loss": 0.3664,
      "step": 9404
    },
    {
      "epoch": 1.4722917971195992,
      "grad_norm": 1.5805662870407104,
      "learning_rate": 8.264907135874877e-05,
      "loss": 0.3631,
      "step": 9405
    },
    {
      "epoch": 1.4724483406386977,
      "grad_norm": 0.7317862510681152,
      "learning_rate": 8.262463343108502e-05,
      "loss": 0.309,
      "step": 9406
    },
    {
      "epoch": 1.472604884157796,
      "grad_norm": 1.2464789152145386,
      "learning_rate": 8.26001955034213e-05,
      "loss": 0.3384,
      "step": 9407
    },
    {
      "epoch": 1.472761427676894,
      "grad_norm": 0.7848882079124451,
      "learning_rate": 8.257575757575757e-05,
      "loss": 0.4214,
      "step": 9408
    },
    {
      "epoch": 1.4729179711959925,
      "grad_norm": 3.3078129291534424,
      "learning_rate": 8.255131964809383e-05,
      "loss": 0.4749,
      "step": 9409
    },
    {
      "epoch": 1.4730745147150908,
      "grad_norm": 6.444918155670166,
      "learning_rate": 8.252688172043011e-05,
      "loss": 0.5161,
      "step": 9410
    },
    {
      "epoch": 1.4732310582341892,
      "grad_norm": 1.534546136856079,
      "learning_rate": 8.250244379276636e-05,
      "loss": 0.5021,
      "step": 9411
    },
    {
      "epoch": 1.4733876017532874,
      "grad_norm": 1.4558238983154297,
      "learning_rate": 8.247800586510263e-05,
      "loss": 0.4859,
      "step": 9412
    },
    {
      "epoch": 1.4735441452723856,
      "grad_norm": 0.8588922023773193,
      "learning_rate": 8.24535679374389e-05,
      "loss": 0.262,
      "step": 9413
    },
    {
      "epoch": 1.473700688791484,
      "grad_norm": 1.7501471042633057,
      "learning_rate": 8.242913000977517e-05,
      "loss": 0.7627,
      "step": 9414
    },
    {
      "epoch": 1.4738572323105823,
      "grad_norm": 4.139252185821533,
      "learning_rate": 8.240469208211142e-05,
      "loss": 0.3825,
      "step": 9415
    },
    {
      "epoch": 1.4740137758296807,
      "grad_norm": 3.472306489944458,
      "learning_rate": 8.23802541544477e-05,
      "loss": 0.9586,
      "step": 9416
    },
    {
      "epoch": 1.474170319348779,
      "grad_norm": 1.744738221168518,
      "learning_rate": 8.235581622678396e-05,
      "loss": 0.6018,
      "step": 9417
    },
    {
      "epoch": 1.4743268628678772,
      "grad_norm": 2.3839869499206543,
      "learning_rate": 8.233137829912021e-05,
      "loss": 0.5255,
      "step": 9418
    },
    {
      "epoch": 1.4744834063869756,
      "grad_norm": 1.5559438467025757,
      "learning_rate": 8.230694037145649e-05,
      "loss": 0.3971,
      "step": 9419
    },
    {
      "epoch": 1.4746399499060738,
      "grad_norm": 7.917959690093994,
      "learning_rate": 8.228250244379276e-05,
      "loss": 0.783,
      "step": 9420
    },
    {
      "epoch": 1.4747964934251723,
      "grad_norm": 2.138312816619873,
      "learning_rate": 8.225806451612902e-05,
      "loss": 0.6701,
      "step": 9421
    },
    {
      "epoch": 1.4749530369442705,
      "grad_norm": 3.2708122730255127,
      "learning_rate": 8.22336265884653e-05,
      "loss": 0.8102,
      "step": 9422
    },
    {
      "epoch": 1.4751095804633687,
      "grad_norm": 2.629610300064087,
      "learning_rate": 8.220918866080155e-05,
      "loss": 0.9962,
      "step": 9423
    },
    {
      "epoch": 1.4752661239824671,
      "grad_norm": 2.9563984870910645,
      "learning_rate": 8.218475073313782e-05,
      "loss": 0.6238,
      "step": 9424
    },
    {
      "epoch": 1.4754226675015654,
      "grad_norm": 3.6667652130126953,
      "learning_rate": 8.21603128054741e-05,
      "loss": 0.965,
      "step": 9425
    },
    {
      "epoch": 1.4755792110206638,
      "grad_norm": 2.6509387493133545,
      "learning_rate": 8.213587487781036e-05,
      "loss": 0.9697,
      "step": 9426
    },
    {
      "epoch": 1.475735754539762,
      "grad_norm": 2.9789719581604004,
      "learning_rate": 8.211143695014661e-05,
      "loss": 0.9831,
      "step": 9427
    },
    {
      "epoch": 1.4758922980588602,
      "grad_norm": 3.1920371055603027,
      "learning_rate": 8.208699902248289e-05,
      "loss": 1.0215,
      "step": 9428
    },
    {
      "epoch": 1.4760488415779587,
      "grad_norm": 2.6322896480560303,
      "learning_rate": 8.206256109481915e-05,
      "loss": 0.941,
      "step": 9429
    },
    {
      "epoch": 1.476205385097057,
      "grad_norm": 3.8062565326690674,
      "learning_rate": 8.20381231671554e-05,
      "loss": 1.4716,
      "step": 9430
    },
    {
      "epoch": 1.4763619286161553,
      "grad_norm": 1.7152727842330933,
      "learning_rate": 8.201368523949168e-05,
      "loss": 1.0996,
      "step": 9431
    },
    {
      "epoch": 1.4765184721352536,
      "grad_norm": 3.0590524673461914,
      "learning_rate": 8.198924731182795e-05,
      "loss": 1.2918,
      "step": 9432
    },
    {
      "epoch": 1.4766750156543518,
      "grad_norm": 2.010657548904419,
      "learning_rate": 8.196480938416421e-05,
      "loss": 0.8202,
      "step": 9433
    },
    {
      "epoch": 1.4768315591734502,
      "grad_norm": 1.9447054862976074,
      "learning_rate": 8.194037145650049e-05,
      "loss": 0.5172,
      "step": 9434
    },
    {
      "epoch": 1.4769881026925487,
      "grad_norm": 1.379649043083191,
      "learning_rate": 8.191593352883674e-05,
      "loss": 0.4932,
      "step": 9435
    },
    {
      "epoch": 1.4771446462116469,
      "grad_norm": 2.9022107124328613,
      "learning_rate": 8.189149560117301e-05,
      "loss": 0.8522,
      "step": 9436
    },
    {
      "epoch": 1.477301189730745,
      "grad_norm": 2.831977128982544,
      "learning_rate": 8.186705767350929e-05,
      "loss": 1.4347,
      "step": 9437
    },
    {
      "epoch": 1.4774577332498435,
      "grad_norm": 1.9703986644744873,
      "learning_rate": 8.184261974584555e-05,
      "loss": 1.0414,
      "step": 9438
    },
    {
      "epoch": 1.4776142767689417,
      "grad_norm": 0.4396578371524811,
      "learning_rate": 8.18181818181818e-05,
      "loss": 0.2257,
      "step": 9439
    },
    {
      "epoch": 1.4777708202880402,
      "grad_norm": 0.6509549617767334,
      "learning_rate": 8.179374389051808e-05,
      "loss": 0.3489,
      "step": 9440
    },
    {
      "epoch": 1.4779273638071384,
      "grad_norm": 0.587127685546875,
      "learning_rate": 8.176930596285435e-05,
      "loss": 0.1851,
      "step": 9441
    },
    {
      "epoch": 1.4780839073262366,
      "grad_norm": 0.4554985463619232,
      "learning_rate": 8.17448680351906e-05,
      "loss": 0.2927,
      "step": 9442
    },
    {
      "epoch": 1.478240450845335,
      "grad_norm": 0.3806717097759247,
      "learning_rate": 8.172043010752688e-05,
      "loss": 0.154,
      "step": 9443
    },
    {
      "epoch": 1.4783969943644333,
      "grad_norm": 0.5629563331604004,
      "learning_rate": 8.169599217986314e-05,
      "loss": 0.2549,
      "step": 9444
    },
    {
      "epoch": 1.4785535378835317,
      "grad_norm": 0.7409935593605042,
      "learning_rate": 8.16715542521994e-05,
      "loss": 0.2599,
      "step": 9445
    },
    {
      "epoch": 1.47871008140263,
      "grad_norm": 1.7930724620819092,
      "learning_rate": 8.164711632453568e-05,
      "loss": 0.3104,
      "step": 9446
    },
    {
      "epoch": 1.4788666249217282,
      "grad_norm": 0.7437885999679565,
      "learning_rate": 8.162267839687193e-05,
      "loss": 0.3755,
      "step": 9447
    },
    {
      "epoch": 1.4790231684408266,
      "grad_norm": 0.7157233357429504,
      "learning_rate": 8.15982404692082e-05,
      "loss": 0.2964,
      "step": 9448
    },
    {
      "epoch": 1.4791797119599248,
      "grad_norm": 0.8939926028251648,
      "learning_rate": 8.157380254154448e-05,
      "loss": 0.1982,
      "step": 9449
    },
    {
      "epoch": 1.4793362554790233,
      "grad_norm": 1.4006277322769165,
      "learning_rate": 8.154936461388073e-05,
      "loss": 0.4599,
      "step": 9450
    },
    {
      "epoch": 1.4794927989981215,
      "grad_norm": 1.421652913093567,
      "learning_rate": 8.1524926686217e-05,
      "loss": 0.4006,
      "step": 9451
    },
    {
      "epoch": 1.4796493425172197,
      "grad_norm": 1.1265709400177002,
      "learning_rate": 8.150048875855327e-05,
      "loss": 0.229,
      "step": 9452
    },
    {
      "epoch": 1.4798058860363181,
      "grad_norm": 1.248371958732605,
      "learning_rate": 8.147605083088954e-05,
      "loss": 0.3313,
      "step": 9453
    },
    {
      "epoch": 1.4799624295554163,
      "grad_norm": 1.0506163835525513,
      "learning_rate": 8.145161290322579e-05,
      "loss": 0.526,
      "step": 9454
    },
    {
      "epoch": 1.4801189730745148,
      "grad_norm": 0.9769994616508484,
      "learning_rate": 8.142717497556207e-05,
      "loss": 0.3684,
      "step": 9455
    },
    {
      "epoch": 1.480275516593613,
      "grad_norm": 1.5080924034118652,
      "learning_rate": 8.140273704789833e-05,
      "loss": 0.4004,
      "step": 9456
    },
    {
      "epoch": 1.4804320601127112,
      "grad_norm": 2.587855339050293,
      "learning_rate": 8.13782991202346e-05,
      "loss": 0.5374,
      "step": 9457
    },
    {
      "epoch": 1.4805886036318097,
      "grad_norm": 1.3442978858947754,
      "learning_rate": 8.135386119257087e-05,
      "loss": 0.4843,
      "step": 9458
    },
    {
      "epoch": 1.4807451471509079,
      "grad_norm": 1.6153483390808105,
      "learning_rate": 8.132942326490713e-05,
      "loss": 0.3719,
      "step": 9459
    },
    {
      "epoch": 1.4809016906700063,
      "grad_norm": 2.865086317062378,
      "learning_rate": 8.130498533724339e-05,
      "loss": 1.0182,
      "step": 9460
    },
    {
      "epoch": 1.4810582341891045,
      "grad_norm": 1.1315138339996338,
      "learning_rate": 8.128054740957967e-05,
      "loss": 0.4096,
      "step": 9461
    },
    {
      "epoch": 1.4812147777082028,
      "grad_norm": 1.3342171907424927,
      "learning_rate": 8.125610948191592e-05,
      "loss": 0.5152,
      "step": 9462
    },
    {
      "epoch": 1.4813713212273012,
      "grad_norm": 1.2070798873901367,
      "learning_rate": 8.123167155425218e-05,
      "loss": 0.4946,
      "step": 9463
    },
    {
      "epoch": 1.4815278647463996,
      "grad_norm": 1.9415396451950073,
      "learning_rate": 8.120723362658846e-05,
      "loss": 0.4527,
      "step": 9464
    },
    {
      "epoch": 1.4816844082654979,
      "grad_norm": 1.080844759941101,
      "learning_rate": 8.118279569892473e-05,
      "loss": 0.2465,
      "step": 9465
    },
    {
      "epoch": 1.481840951784596,
      "grad_norm": 1.6477303504943848,
      "learning_rate": 8.115835777126098e-05,
      "loss": 0.6034,
      "step": 9466
    },
    {
      "epoch": 1.4819974953036943,
      "grad_norm": 2.1239099502563477,
      "learning_rate": 8.113391984359726e-05,
      "loss": 0.6707,
      "step": 9467
    },
    {
      "epoch": 1.4821540388227927,
      "grad_norm": 1.421504259109497,
      "learning_rate": 8.110948191593352e-05,
      "loss": 0.5054,
      "step": 9468
    },
    {
      "epoch": 1.4823105823418912,
      "grad_norm": 3.8904707431793213,
      "learning_rate": 8.108504398826979e-05,
      "loss": 0.4603,
      "step": 9469
    },
    {
      "epoch": 1.4824671258609894,
      "grad_norm": 1.8012278079986572,
      "learning_rate": 8.106060606060607e-05,
      "loss": 0.5213,
      "step": 9470
    },
    {
      "epoch": 1.4826236693800876,
      "grad_norm": 3.616649627685547,
      "learning_rate": 8.103616813294232e-05,
      "loss": 1.315,
      "step": 9471
    },
    {
      "epoch": 1.482780212899186,
      "grad_norm": 2.4311721324920654,
      "learning_rate": 8.101173020527858e-05,
      "loss": 0.9852,
      "step": 9472
    },
    {
      "epoch": 1.4829367564182843,
      "grad_norm": 1.9556763172149658,
      "learning_rate": 8.098729227761486e-05,
      "loss": 0.8551,
      "step": 9473
    },
    {
      "epoch": 1.4830932999373827,
      "grad_norm": 2.610199213027954,
      "learning_rate": 8.096285434995111e-05,
      "loss": 0.4723,
      "step": 9474
    },
    {
      "epoch": 1.483249843456481,
      "grad_norm": 1.7846730947494507,
      "learning_rate": 8.093841642228738e-05,
      "loss": 1.0211,
      "step": 9475
    },
    {
      "epoch": 1.4834063869755791,
      "grad_norm": 2.926453113555908,
      "learning_rate": 8.091397849462365e-05,
      "loss": 1.0827,
      "step": 9476
    },
    {
      "epoch": 1.4835629304946776,
      "grad_norm": 2.9660816192626953,
      "learning_rate": 8.088954056695992e-05,
      "loss": 1.0163,
      "step": 9477
    },
    {
      "epoch": 1.4837194740137758,
      "grad_norm": 2.5968966484069824,
      "learning_rate": 8.086510263929617e-05,
      "loss": 0.6645,
      "step": 9478
    },
    {
      "epoch": 1.4838760175328742,
      "grad_norm": 4.448347091674805,
      "learning_rate": 8.084066471163245e-05,
      "loss": 1.0715,
      "step": 9479
    },
    {
      "epoch": 1.4840325610519725,
      "grad_norm": 2.39528226852417,
      "learning_rate": 8.081622678396871e-05,
      "loss": 1.1425,
      "step": 9480
    },
    {
      "epoch": 1.4841891045710707,
      "grad_norm": 2.856279134750366,
      "learning_rate": 8.079178885630498e-05,
      "loss": 0.9069,
      "step": 9481
    },
    {
      "epoch": 1.4843456480901691,
      "grad_norm": 2.2662487030029297,
      "learning_rate": 8.076735092864126e-05,
      "loss": 0.7697,
      "step": 9482
    },
    {
      "epoch": 1.4845021916092673,
      "grad_norm": 2.555561065673828,
      "learning_rate": 8.074291300097751e-05,
      "loss": 1.2625,
      "step": 9483
    },
    {
      "epoch": 1.4846587351283658,
      "grad_norm": 1.9698431491851807,
      "learning_rate": 8.071847507331377e-05,
      "loss": 1.1531,
      "step": 9484
    },
    {
      "epoch": 1.484815278647464,
      "grad_norm": 2.155409336090088,
      "learning_rate": 8.069403714565005e-05,
      "loss": 0.41,
      "step": 9485
    },
    {
      "epoch": 1.4849718221665622,
      "grad_norm": 2.2174770832061768,
      "learning_rate": 8.06695992179863e-05,
      "loss": 0.8686,
      "step": 9486
    },
    {
      "epoch": 1.4851283656856606,
      "grad_norm": 3.0642926692962646,
      "learning_rate": 8.064516129032257e-05,
      "loss": 1.2679,
      "step": 9487
    },
    {
      "epoch": 1.4852849092047589,
      "grad_norm": 3.7173335552215576,
      "learning_rate": 8.062072336265885e-05,
      "loss": 0.8235,
      "step": 9488
    },
    {
      "epoch": 1.4854414527238573,
      "grad_norm": 0.439689964056015,
      "learning_rate": 8.059628543499511e-05,
      "loss": 0.2204,
      "step": 9489
    },
    {
      "epoch": 1.4855979962429555,
      "grad_norm": 0.3609512448310852,
      "learning_rate": 8.057184750733136e-05,
      "loss": 0.2014,
      "step": 9490
    },
    {
      "epoch": 1.4857545397620537,
      "grad_norm": 0.530227541923523,
      "learning_rate": 8.054740957966764e-05,
      "loss": 0.1972,
      "step": 9491
    },
    {
      "epoch": 1.4859110832811522,
      "grad_norm": 0.6065565943717957,
      "learning_rate": 8.05229716520039e-05,
      "loss": 0.2325,
      "step": 9492
    },
    {
      "epoch": 1.4860676268002504,
      "grad_norm": 0.9413346648216248,
      "learning_rate": 8.049853372434017e-05,
      "loss": 0.3771,
      "step": 9493
    },
    {
      "epoch": 1.4862241703193488,
      "grad_norm": 0.57166588306427,
      "learning_rate": 8.047409579667645e-05,
      "loss": 0.2029,
      "step": 9494
    },
    {
      "epoch": 1.486380713838447,
      "grad_norm": 0.8557118773460388,
      "learning_rate": 8.04496578690127e-05,
      "loss": 0.2793,
      "step": 9495
    },
    {
      "epoch": 1.4865372573575453,
      "grad_norm": 3.641554117202759,
      "learning_rate": 8.042521994134896e-05,
      "loss": 0.764,
      "step": 9496
    },
    {
      "epoch": 1.4866938008766437,
      "grad_norm": 0.5618619918823242,
      "learning_rate": 8.040078201368524e-05,
      "loss": 0.1924,
      "step": 9497
    },
    {
      "epoch": 1.4868503443957422,
      "grad_norm": 1.0576132535934448,
      "learning_rate": 8.03763440860215e-05,
      "loss": 0.3731,
      "step": 9498
    },
    {
      "epoch": 1.4870068879148404,
      "grad_norm": 0.5164195895195007,
      "learning_rate": 8.035190615835776e-05,
      "loss": 0.2424,
      "step": 9499
    },
    {
      "epoch": 1.4871634314339386,
      "grad_norm": 0.5563727617263794,
      "learning_rate": 8.032746823069404e-05,
      "loss": 0.1442,
      "step": 9500
    },
    {
      "epoch": 1.487319974953037,
      "grad_norm": 0.9167290329933167,
      "learning_rate": 8.03030303030303e-05,
      "loss": 0.3146,
      "step": 9501
    },
    {
      "epoch": 1.4874765184721352,
      "grad_norm": 0.6889357566833496,
      "learning_rate": 8.027859237536655e-05,
      "loss": 0.259,
      "step": 9502
    },
    {
      "epoch": 1.4876330619912337,
      "grad_norm": NaN,
      "learning_rate": 8.027859237536655e-05,
      "loss": 0.0,
      "step": 9503
    },
    {
      "epoch": 1.487789605510332,
      "grad_norm": 0.6606603860855103,
      "learning_rate": 8.025415444770283e-05,
      "loss": 0.3823,
      "step": 9504
    },
    {
      "epoch": 1.4879461490294301,
      "grad_norm": 1.8341822624206543,
      "learning_rate": 8.02297165200391e-05,
      "loss": 0.4172,
      "step": 9505
    },
    {
      "epoch": 1.4881026925485286,
      "grad_norm": 1.2621349096298218,
      "learning_rate": 8.020527859237536e-05,
      "loss": 0.3575,
      "step": 9506
    },
    {
      "epoch": 1.4882592360676268,
      "grad_norm": 0.7620284557342529,
      "learning_rate": 8.018084066471163e-05,
      "loss": 0.3535,
      "step": 9507
    },
    {
      "epoch": 1.4884157795867252,
      "grad_norm": 1.7038427591323853,
      "learning_rate": 8.015640273704789e-05,
      "loss": 0.6989,
      "step": 9508
    },
    {
      "epoch": 1.4885723231058234,
      "grad_norm": 1.006224274635315,
      "learning_rate": 8.013196480938415e-05,
      "loss": 0.4143,
      "step": 9509
    },
    {
      "epoch": 1.4887288666249217,
      "grad_norm": 1.3613799810409546,
      "learning_rate": 8.010752688172043e-05,
      "loss": 0.3454,
      "step": 9510
    },
    {
      "epoch": 1.48888541014402,
      "grad_norm": 2.3779265880584717,
      "learning_rate": 8.008308895405668e-05,
      "loss": 0.3727,
      "step": 9511
    },
    {
      "epoch": 1.4890419536631183,
      "grad_norm": 1.3316359519958496,
      "learning_rate": 8.005865102639295e-05,
      "loss": 0.4661,
      "step": 9512
    },
    {
      "epoch": 1.4891984971822168,
      "grad_norm": 2.452564239501953,
      "learning_rate": 8.003421309872923e-05,
      "loss": 0.6581,
      "step": 9513
    },
    {
      "epoch": 1.489355040701315,
      "grad_norm": 2.7504520416259766,
      "learning_rate": 8.000977517106549e-05,
      "loss": 0.7966,
      "step": 9514
    },
    {
      "epoch": 1.4895115842204132,
      "grad_norm": 2.999328851699829,
      "learning_rate": 7.998533724340174e-05,
      "loss": 0.9771,
      "step": 9515
    },
    {
      "epoch": 1.4896681277395116,
      "grad_norm": 2.237630605697632,
      "learning_rate": 7.996089931573802e-05,
      "loss": 0.6239,
      "step": 9516
    },
    {
      "epoch": 1.4898246712586098,
      "grad_norm": 1.3302862644195557,
      "learning_rate": 7.993646138807429e-05,
      "loss": 0.421,
      "step": 9517
    },
    {
      "epoch": 1.4899812147777083,
      "grad_norm": 1.9129620790481567,
      "learning_rate": 7.991202346041055e-05,
      "loss": 0.6495,
      "step": 9518
    },
    {
      "epoch": 1.4901377582968065,
      "grad_norm": 1.6199758052825928,
      "learning_rate": 7.988758553274682e-05,
      "loss": 0.607,
      "step": 9519
    },
    {
      "epoch": 1.4902943018159047,
      "grad_norm": 2.3590195178985596,
      "learning_rate": 7.986314760508308e-05,
      "loss": 1.0782,
      "step": 9520
    },
    {
      "epoch": 1.4904508453350032,
      "grad_norm": 2.1550941467285156,
      "learning_rate": 7.983870967741935e-05,
      "loss": 0.9664,
      "step": 9521
    },
    {
      "epoch": 1.4906073888541014,
      "grad_norm": 4.303733825683594,
      "learning_rate": 7.981427174975562e-05,
      "loss": 0.9356,
      "step": 9522
    },
    {
      "epoch": 1.4907639323731998,
      "grad_norm": 2.1720597743988037,
      "learning_rate": 7.978983382209188e-05,
      "loss": 0.6391,
      "step": 9523
    },
    {
      "epoch": 1.490920475892298,
      "grad_norm": 2.204709053039551,
      "learning_rate": 7.976539589442814e-05,
      "loss": 0.6076,
      "step": 9524
    },
    {
      "epoch": 1.4910770194113963,
      "grad_norm": 2.43810772895813,
      "learning_rate": 7.974095796676442e-05,
      "loss": 1.1823,
      "step": 9525
    },
    {
      "epoch": 1.4912335629304947,
      "grad_norm": 1.9804140329360962,
      "learning_rate": 7.971652003910068e-05,
      "loss": 0.3912,
      "step": 9526
    },
    {
      "epoch": 1.491390106449593,
      "grad_norm": 2.124976396560669,
      "learning_rate": 7.969208211143693e-05,
      "loss": 0.8703,
      "step": 9527
    },
    {
      "epoch": 1.4915466499686914,
      "grad_norm": 1.996584415435791,
      "learning_rate": 7.966764418377321e-05,
      "loss": 0.9466,
      "step": 9528
    },
    {
      "epoch": 1.4917031934877896,
      "grad_norm": 1.8637217283248901,
      "learning_rate": 7.964320625610948e-05,
      "loss": 0.7035,
      "step": 9529
    },
    {
      "epoch": 1.4918597370068878,
      "grad_norm": 4.413573741912842,
      "learning_rate": 7.961876832844574e-05,
      "loss": 0.8354,
      "step": 9530
    },
    {
      "epoch": 1.4920162805259862,
      "grad_norm": 2.3614120483398438,
      "learning_rate": 7.959433040078201e-05,
      "loss": 1.1676,
      "step": 9531
    },
    {
      "epoch": 1.4921728240450847,
      "grad_norm": 2.0499792098999023,
      "learning_rate": 7.956989247311827e-05,
      "loss": 1.3798,
      "step": 9532
    },
    {
      "epoch": 1.4923293675641829,
      "grad_norm": 2.0810813903808594,
      "learning_rate": 7.954545454545454e-05,
      "loss": 0.8426,
      "step": 9533
    },
    {
      "epoch": 1.492485911083281,
      "grad_norm": 2.218729257583618,
      "learning_rate": 7.952101661779082e-05,
      "loss": 1.0149,
      "step": 9534
    },
    {
      "epoch": 1.4926424546023795,
      "grad_norm": 1.0656377077102661,
      "learning_rate": 7.949657869012707e-05,
      "loss": 0.1697,
      "step": 9535
    },
    {
      "epoch": 1.4927989981214778,
      "grad_norm": 2.4055192470550537,
      "learning_rate": 7.947214076246333e-05,
      "loss": 0.6699,
      "step": 9536
    },
    {
      "epoch": 1.4929555416405762,
      "grad_norm": 1.910409927368164,
      "learning_rate": 7.944770283479961e-05,
      "loss": 0.7338,
      "step": 9537
    },
    {
      "epoch": 1.4931120851596744,
      "grad_norm": 1.1563485860824585,
      "learning_rate": 7.942326490713587e-05,
      "loss": 0.3543,
      "step": 9538
    },
    {
      "epoch": 1.4932686286787726,
      "grad_norm": 0.43213382363319397,
      "learning_rate": 7.939882697947213e-05,
      "loss": 0.1907,
      "step": 9539
    },
    {
      "epoch": 1.493425172197871,
      "grad_norm": 0.5200718641281128,
      "learning_rate": 7.93743890518084e-05,
      "loss": 0.2515,
      "step": 9540
    },
    {
      "epoch": 1.4935817157169693,
      "grad_norm": 1.2368117570877075,
      "learning_rate": 7.934995112414467e-05,
      "loss": 0.2597,
      "step": 9541
    },
    {
      "epoch": 1.4937382592360677,
      "grad_norm": 0.9348035454750061,
      "learning_rate": 7.932551319648093e-05,
      "loss": 0.3166,
      "step": 9542
    },
    {
      "epoch": 1.493894802755166,
      "grad_norm": 0.7543662190437317,
      "learning_rate": 7.93010752688172e-05,
      "loss": 0.2521,
      "step": 9543
    },
    {
      "epoch": 1.4940513462742642,
      "grad_norm": 0.44818371534347534,
      "learning_rate": 7.927663734115346e-05,
      "loss": 0.1899,
      "step": 9544
    },
    {
      "epoch": 1.4942078897933626,
      "grad_norm": 0.6756463050842285,
      "learning_rate": 7.925219941348973e-05,
      "loss": 0.2408,
      "step": 9545
    },
    {
      "epoch": 1.4943644333124608,
      "grad_norm": 1.1319537162780762,
      "learning_rate": 7.9227761485826e-05,
      "loss": 0.2446,
      "step": 9546
    },
    {
      "epoch": 1.4945209768315593,
      "grad_norm": 1.639535903930664,
      "learning_rate": 7.920332355816226e-05,
      "loss": 0.2676,
      "step": 9547
    },
    {
      "epoch": 1.4946775203506575,
      "grad_norm": 0.6942873001098633,
      "learning_rate": 7.917888563049852e-05,
      "loss": 0.4021,
      "step": 9548
    },
    {
      "epoch": 1.4948340638697557,
      "grad_norm": 0.61109858751297,
      "learning_rate": 7.91544477028348e-05,
      "loss": 0.2719,
      "step": 9549
    },
    {
      "epoch": 1.4949906073888541,
      "grad_norm": 0.8826609253883362,
      "learning_rate": 7.913000977517107e-05,
      "loss": 0.4091,
      "step": 9550
    },
    {
      "epoch": 1.4951471509079524,
      "grad_norm": 2.2055704593658447,
      "learning_rate": 7.910557184750732e-05,
      "loss": 0.3353,
      "step": 9551
    },
    {
      "epoch": 1.4953036944270508,
      "grad_norm": 4.32928991317749,
      "learning_rate": 7.90811339198436e-05,
      "loss": 0.5081,
      "step": 9552
    },
    {
      "epoch": 1.495460237946149,
      "grad_norm": 0.8008466958999634,
      "learning_rate": 7.905669599217986e-05,
      "loss": 0.2946,
      "step": 9553
    },
    {
      "epoch": 1.4956167814652472,
      "grad_norm": 1.2389719486236572,
      "learning_rate": 7.903225806451613e-05,
      "loss": 0.3579,
      "step": 9554
    },
    {
      "epoch": 1.4957733249843457,
      "grad_norm": 1.683296799659729,
      "learning_rate": 7.900782013685239e-05,
      "loss": 0.4454,
      "step": 9555
    },
    {
      "epoch": 1.495929868503444,
      "grad_norm": 1.3349992036819458,
      "learning_rate": 7.898338220918865e-05,
      "loss": 0.3898,
      "step": 9556
    },
    {
      "epoch": 1.4960864120225423,
      "grad_norm": 0.86898273229599,
      "learning_rate": 7.895894428152492e-05,
      "loss": 0.4127,
      "step": 9557
    },
    {
      "epoch": 1.4962429555416406,
      "grad_norm": 1.2158864736557007,
      "learning_rate": 7.89345063538612e-05,
      "loss": 0.262,
      "step": 9558
    },
    {
      "epoch": 1.4963994990607388,
      "grad_norm": 1.0197027921676636,
      "learning_rate": 7.891006842619745e-05,
      "loss": 0.3522,
      "step": 9559
    },
    {
      "epoch": 1.4965560425798372,
      "grad_norm": 2.6261444091796875,
      "learning_rate": 7.888563049853371e-05,
      "loss": 0.4608,
      "step": 9560
    },
    {
      "epoch": 1.4967125860989354,
      "grad_norm": 1.4743295907974243,
      "learning_rate": 7.886119257086999e-05,
      "loss": 0.4477,
      "step": 9561
    },
    {
      "epoch": 1.4968691296180339,
      "grad_norm": 1.690913438796997,
      "learning_rate": 7.883675464320626e-05,
      "loss": 0.2981,
      "step": 9562
    },
    {
      "epoch": 1.497025673137132,
      "grad_norm": 1.879437804222107,
      "learning_rate": 7.881231671554251e-05,
      "loss": 0.5513,
      "step": 9563
    },
    {
      "epoch": 1.4971822166562303,
      "grad_norm": 2.7271459102630615,
      "learning_rate": 7.878787878787879e-05,
      "loss": 0.677,
      "step": 9564
    },
    {
      "epoch": 1.4973387601753287,
      "grad_norm": 3.2245678901672363,
      "learning_rate": 7.876344086021505e-05,
      "loss": 0.787,
      "step": 9565
    },
    {
      "epoch": 1.4974953036944272,
      "grad_norm": 1.3379689455032349,
      "learning_rate": 7.87390029325513e-05,
      "loss": 0.7679,
      "step": 9566
    },
    {
      "epoch": 1.4976518472135254,
      "grad_norm": 1.8764911890029907,
      "learning_rate": 7.871456500488758e-05,
      "loss": 0.613,
      "step": 9567
    },
    {
      "epoch": 1.4978083907326236,
      "grad_norm": 3.4302899837493896,
      "learning_rate": 7.869012707722385e-05,
      "loss": 0.643,
      "step": 9568
    },
    {
      "epoch": 1.497964934251722,
      "grad_norm": 2.4290006160736084,
      "learning_rate": 7.866568914956011e-05,
      "loss": 0.4512,
      "step": 9569
    },
    {
      "epoch": 1.4981214777708203,
      "grad_norm": 2.3394429683685303,
      "learning_rate": 7.864125122189639e-05,
      "loss": 0.7265,
      "step": 9570
    },
    {
      "epoch": 1.4982780212899187,
      "grad_norm": 3.2177608013153076,
      "learning_rate": 7.861681329423264e-05,
      "loss": 0.7219,
      "step": 9571
    },
    {
      "epoch": 1.498434564809017,
      "grad_norm": 2.652704954147339,
      "learning_rate": 7.85923753665689e-05,
      "loss": 0.921,
      "step": 9572
    },
    {
      "epoch": 1.4985911083281152,
      "grad_norm": 1.5639492273330688,
      "learning_rate": 7.856793743890518e-05,
      "loss": 0.8148,
      "step": 9573
    },
    {
      "epoch": 1.4987476518472136,
      "grad_norm": 1.7422173023223877,
      "learning_rate": 7.854349951124145e-05,
      "loss": 0.429,
      "step": 9574
    },
    {
      "epoch": 1.4989041953663118,
      "grad_norm": 2.2326202392578125,
      "learning_rate": 7.85190615835777e-05,
      "loss": 0.9982,
      "step": 9575
    },
    {
      "epoch": 1.4990607388854102,
      "grad_norm": 2.4067230224609375,
      "learning_rate": 7.849462365591398e-05,
      "loss": 0.8497,
      "step": 9576
    },
    {
      "epoch": 1.4992172824045085,
      "grad_norm": 5.408649444580078,
      "learning_rate": 7.847018572825024e-05,
      "loss": 1.1441,
      "step": 9577
    },
    {
      "epoch": 1.4993738259236067,
      "grad_norm": 2.413092613220215,
      "learning_rate": 7.84457478005865e-05,
      "loss": 1.1049,
      "step": 9578
    },
    {
      "epoch": 1.4995303694427051,
      "grad_norm": 2.305614471435547,
      "learning_rate": 7.842130987292277e-05,
      "loss": 1.061,
      "step": 9579
    },
    {
      "epoch": 1.4996869129618033,
      "grad_norm": 2.732689619064331,
      "learning_rate": 7.839687194525904e-05,
      "loss": 0.5431,
      "step": 9580
    },
    {
      "epoch": 1.4998434564809018,
      "grad_norm": 2.2175774574279785,
      "learning_rate": 7.83724340175953e-05,
      "loss": 1.5274,
      "step": 9581
    },
    {
      "epoch": 1.5,
      "grad_norm": 3.7100279331207275,
      "learning_rate": 7.834799608993158e-05,
      "loss": 1.4016,
      "step": 9582
    },
    {
      "epoch": 1.5001565435190982,
      "grad_norm": 2.078692674636841,
      "learning_rate": 7.832355816226783e-05,
      "loss": 0.6064,
      "step": 9583
    },
    {
      "epoch": 1.5003130870381967,
      "grad_norm": 2.7273752689361572,
      "learning_rate": 7.82991202346041e-05,
      "loss": 0.5273,
      "step": 9584
    },
    {
      "epoch": 1.5004696305572949,
      "grad_norm": 1.3551000356674194,
      "learning_rate": 7.827468230694037e-05,
      "loss": 0.3828,
      "step": 9585
    },
    {
      "epoch": 1.5006261740763933,
      "grad_norm": 3.331238269805908,
      "learning_rate": 7.825024437927664e-05,
      "loss": 1.221,
      "step": 9586
    },
    {
      "epoch": 1.5007827175954915,
      "grad_norm": 2.849250555038452,
      "learning_rate": 7.822580645161289e-05,
      "loss": 0.799,
      "step": 9587
    },
    {
      "epoch": 1.5009392611145898,
      "grad_norm": 2.9601728916168213,
      "learning_rate": 7.820136852394917e-05,
      "loss": 1.6444,
      "step": 9588
    },
    {
      "epoch": 1.5010958046336882,
      "grad_norm": 0.5723337531089783,
      "learning_rate": 7.817693059628543e-05,
      "loss": 0.2622,
      "step": 9589
    },
    {
      "epoch": 1.5012523481527866,
      "grad_norm": 0.8948503732681274,
      "learning_rate": 7.815249266862169e-05,
      "loss": 0.2687,
      "step": 9590
    },
    {
      "epoch": 1.5014088916718848,
      "grad_norm": 0.6991376876831055,
      "learning_rate": 7.812805474095796e-05,
      "loss": 0.2752,
      "step": 9591
    },
    {
      "epoch": 1.501565435190983,
      "grad_norm": 0.6468504667282104,
      "learning_rate": 7.810361681329423e-05,
      "loss": 0.1794,
      "step": 9592
    },
    {
      "epoch": 1.5017219787100813,
      "grad_norm": 0.5929017066955566,
      "learning_rate": 7.807917888563049e-05,
      "loss": 0.2153,
      "step": 9593
    },
    {
      "epoch": 1.5018785222291797,
      "grad_norm": 7.306670188903809,
      "learning_rate": 7.805474095796674e-05,
      "loss": 0.2172,
      "step": 9594
    },
    {
      "epoch": 1.5020350657482782,
      "grad_norm": 0.5348293781280518,
      "learning_rate": 7.803030303030302e-05,
      "loss": 0.2575,
      "step": 9595
    },
    {
      "epoch": 1.5021916092673764,
      "grad_norm": 0.5054227709770203,
      "learning_rate": 7.800586510263929e-05,
      "loss": 0.229,
      "step": 9596
    },
    {
      "epoch": 1.5023481527864746,
      "grad_norm": 0.5968636274337769,
      "learning_rate": 7.798142717497555e-05,
      "loss": 0.1961,
      "step": 9597
    },
    {
      "epoch": 1.5025046963055728,
      "grad_norm": 1.1959178447723389,
      "learning_rate": 7.795698924731183e-05,
      "loss": 0.328,
      "step": 9598
    },
    {
      "epoch": 1.5026612398246713,
      "grad_norm": 1.3676921129226685,
      "learning_rate": 7.793255131964808e-05,
      "loss": 0.294,
      "step": 9599
    },
    {
      "epoch": 1.5028177833437697,
      "grad_norm": 0.5510010123252869,
      "learning_rate": 7.790811339198435e-05,
      "loss": 0.2935,
      "step": 9600
    },
    {
      "epoch": 1.502974326862868,
      "grad_norm": 1.0297011137008667,
      "learning_rate": 7.788367546432063e-05,
      "loss": 0.2587,
      "step": 9601
    },
    {
      "epoch": 1.5031308703819661,
      "grad_norm": 2.1265127658843994,
      "learning_rate": 7.785923753665688e-05,
      "loss": 0.4379,
      "step": 9602
    },
    {
      "epoch": 1.5032874139010644,
      "grad_norm": 1.283608317375183,
      "learning_rate": 7.783479960899314e-05,
      "loss": 0.5179,
      "step": 9603
    },
    {
      "epoch": 1.5034439574201628,
      "grad_norm": 0.9307227730751038,
      "learning_rate": 7.781036168132942e-05,
      "loss": 0.3469,
      "step": 9604
    },
    {
      "epoch": 1.5036005009392612,
      "grad_norm": 1.4115785360336304,
      "learning_rate": 7.778592375366568e-05,
      "loss": 0.3251,
      "step": 9605
    },
    {
      "epoch": 1.5037570444583594,
      "grad_norm": 1.2950360774993896,
      "learning_rate": 7.776148582600194e-05,
      "loss": 0.4412,
      "step": 9606
    },
    {
      "epoch": 1.5039135879774577,
      "grad_norm": 1.8613760471343994,
      "learning_rate": 7.773704789833821e-05,
      "loss": 0.5847,
      "step": 9607
    },
    {
      "epoch": 1.5040701314965559,
      "grad_norm": 1.611189365386963,
      "learning_rate": 7.771260997067448e-05,
      "loss": 0.4761,
      "step": 9608
    },
    {
      "epoch": 1.5042266750156543,
      "grad_norm": 1.2871978282928467,
      "learning_rate": 7.768817204301074e-05,
      "loss": 0.3882,
      "step": 9609
    },
    {
      "epoch": 1.5043832185347528,
      "grad_norm": 0.9030357599258423,
      "learning_rate": 7.766373411534701e-05,
      "loss": 0.3907,
      "step": 9610
    },
    {
      "epoch": 1.504539762053851,
      "grad_norm": 1.4141883850097656,
      "learning_rate": 7.763929618768327e-05,
      "loss": 0.4686,
      "step": 9611
    },
    {
      "epoch": 1.5046963055729492,
      "grad_norm": 2.3717682361602783,
      "learning_rate": 7.761485826001954e-05,
      "loss": 0.5278,
      "step": 9612
    },
    {
      "epoch": 1.5048528490920476,
      "grad_norm": 0.8933207988739014,
      "learning_rate": 7.759042033235582e-05,
      "loss": 0.3828,
      "step": 9613
    },
    {
      "epoch": 1.5050093926111459,
      "grad_norm": 1.469948649406433,
      "learning_rate": 7.756598240469207e-05,
      "loss": 0.6257,
      "step": 9614
    },
    {
      "epoch": 1.5051659361302443,
      "grad_norm": 2.732593297958374,
      "learning_rate": 7.754154447702833e-05,
      "loss": 0.5167,
      "step": 9615
    },
    {
      "epoch": 1.5053224796493425,
      "grad_norm": 2.5369060039520264,
      "learning_rate": 7.751710654936461e-05,
      "loss": 0.4295,
      "step": 9616
    },
    {
      "epoch": 1.5054790231684407,
      "grad_norm": 3.549316883087158,
      "learning_rate": 7.749266862170088e-05,
      "loss": 0.3771,
      "step": 9617
    },
    {
      "epoch": 1.5056355666875392,
      "grad_norm": 1.7720410823822021,
      "learning_rate": 7.746823069403713e-05,
      "loss": 0.6011,
      "step": 9618
    },
    {
      "epoch": 1.5057921102066374,
      "grad_norm": 0.6939887404441833,
      "learning_rate": 7.74437927663734e-05,
      "loss": 0.237,
      "step": 9619
    },
    {
      "epoch": 1.5059486537257358,
      "grad_norm": 2.6376233100891113,
      "learning_rate": 7.741935483870967e-05,
      "loss": 1.1664,
      "step": 9620
    },
    {
      "epoch": 1.506105197244834,
      "grad_norm": 2.496608018875122,
      "learning_rate": 7.739491691104593e-05,
      "loss": 0.7487,
      "step": 9621
    },
    {
      "epoch": 1.5062617407639323,
      "grad_norm": 2.7778053283691406,
      "learning_rate": 7.73704789833822e-05,
      "loss": 0.8468,
      "step": 9622
    },
    {
      "epoch": 1.5064182842830307,
      "grad_norm": 2.1602282524108887,
      "learning_rate": 7.734604105571846e-05,
      "loss": 0.9794,
      "step": 9623
    },
    {
      "epoch": 1.5065748278021291,
      "grad_norm": 2.9261624813079834,
      "learning_rate": 7.732160312805473e-05,
      "loss": 0.953,
      "step": 9624
    },
    {
      "epoch": 1.5067313713212274,
      "grad_norm": 3.67445969581604,
      "learning_rate": 7.729716520039101e-05,
      "loss": 0.8946,
      "step": 9625
    },
    {
      "epoch": 1.5068879148403256,
      "grad_norm": 2.395833730697632,
      "learning_rate": 7.727272727272726e-05,
      "loss": 1.5854,
      "step": 9626
    },
    {
      "epoch": 1.5070444583594238,
      "grad_norm": 2.6899259090423584,
      "learning_rate": 7.724828934506352e-05,
      "loss": 0.5556,
      "step": 9627
    },
    {
      "epoch": 1.5072010018785222,
      "grad_norm": 1.9388134479522705,
      "learning_rate": 7.72238514173998e-05,
      "loss": 1.3441,
      "step": 9628
    },
    {
      "epoch": 1.5073575453976207,
      "grad_norm": 3.582058906555176,
      "learning_rate": 7.719941348973607e-05,
      "loss": 1.3142,
      "step": 9629
    },
    {
      "epoch": 1.507514088916719,
      "grad_norm": 1.9698890447616577,
      "learning_rate": 7.717497556207232e-05,
      "loss": 0.64,
      "step": 9630
    },
    {
      "epoch": 1.5076706324358171,
      "grad_norm": 2.8033182621002197,
      "learning_rate": 7.71505376344086e-05,
      "loss": 1.1611,
      "step": 9631
    },
    {
      "epoch": 1.5078271759549153,
      "grad_norm": 3.156467914581299,
      "learning_rate": 7.712609970674486e-05,
      "loss": 0.9295,
      "step": 9632
    },
    {
      "epoch": 1.5079837194740138,
      "grad_norm": 1.1335583925247192,
      "learning_rate": 7.710166177908113e-05,
      "loss": 0.5743,
      "step": 9633
    },
    {
      "epoch": 1.5081402629931122,
      "grad_norm": 2.90801739692688,
      "learning_rate": 7.707722385141739e-05,
      "loss": 0.6852,
      "step": 9634
    },
    {
      "epoch": 1.5082968065122104,
      "grad_norm": 2.2356724739074707,
      "learning_rate": 7.705278592375366e-05,
      "loss": 0.8531,
      "step": 9635
    },
    {
      "epoch": 1.5084533500313086,
      "grad_norm": 1.6373522281646729,
      "learning_rate": 7.702834799608992e-05,
      "loss": 0.524,
      "step": 9636
    },
    {
      "epoch": 1.5086098935504069,
      "grad_norm": 1.7506439685821533,
      "learning_rate": 7.70039100684262e-05,
      "loss": 0.5397,
      "step": 9637
    },
    {
      "epoch": 1.5087664370695053,
      "grad_norm": 1.3206775188446045,
      "learning_rate": 7.697947214076245e-05,
      "loss": 0.6136,
      "step": 9638
    },
    {
      "epoch": 1.5089229805886037,
      "grad_norm": 0.5027822256088257,
      "learning_rate": 7.695503421309871e-05,
      "loss": 0.2406,
      "step": 9639
    },
    {
      "epoch": 1.509079524107702,
      "grad_norm": 0.5172881484031677,
      "learning_rate": 7.693059628543499e-05,
      "loss": 0.2308,
      "step": 9640
    },
    {
      "epoch": 1.5092360676268002,
      "grad_norm": 0.5118494033813477,
      "learning_rate": 7.690615835777126e-05,
      "loss": 0.2369,
      "step": 9641
    },
    {
      "epoch": 1.5093926111458984,
      "grad_norm": 0.42256832122802734,
      "learning_rate": 7.688172043010751e-05,
      "loss": 0.131,
      "step": 9642
    },
    {
      "epoch": 1.5095491546649968,
      "grad_norm": 0.41296517848968506,
      "learning_rate": 7.685728250244379e-05,
      "loss": 0.1728,
      "step": 9643
    },
    {
      "epoch": 1.5097056981840953,
      "grad_norm": 1.4930837154388428,
      "learning_rate": 7.683284457478005e-05,
      "loss": 0.1832,
      "step": 9644
    },
    {
      "epoch": 1.5098622417031935,
      "grad_norm": 0.46577295660972595,
      "learning_rate": 7.680840664711632e-05,
      "loss": 0.2564,
      "step": 9645
    },
    {
      "epoch": 1.5100187852222917,
      "grad_norm": 0.6412208676338196,
      "learning_rate": 7.678396871945258e-05,
      "loss": 0.3131,
      "step": 9646
    },
    {
      "epoch": 1.5101753287413902,
      "grad_norm": 1.1731911897659302,
      "learning_rate": 7.675953079178885e-05,
      "loss": 0.4227,
      "step": 9647
    },
    {
      "epoch": 1.5103318722604884,
      "grad_norm": 0.7713680267333984,
      "learning_rate": 7.673509286412511e-05,
      "loss": 0.3043,
      "step": 9648
    },
    {
      "epoch": 1.5104884157795868,
      "grad_norm": 0.8555247187614441,
      "learning_rate": 7.671065493646139e-05,
      "loss": 0.3916,
      "step": 9649
    },
    {
      "epoch": 1.510644959298685,
      "grad_norm": 1.737226128578186,
      "learning_rate": 7.668621700879764e-05,
      "loss": 0.2323,
      "step": 9650
    },
    {
      "epoch": 1.5108015028177832,
      "grad_norm": 2.419440746307373,
      "learning_rate": 7.66617790811339e-05,
      "loss": 0.3248,
      "step": 9651
    },
    {
      "epoch": 1.5109580463368817,
      "grad_norm": 1.3268059492111206,
      "learning_rate": 7.663734115347018e-05,
      "loss": 0.37,
      "step": 9652
    },
    {
      "epoch": 1.5111145898559801,
      "grad_norm": 0.7393986582756042,
      "learning_rate": 7.661290322580645e-05,
      "loss": 0.5048,
      "step": 9653
    },
    {
      "epoch": 1.5112711333750783,
      "grad_norm": 1.3429745435714722,
      "learning_rate": 7.65884652981427e-05,
      "loss": 0.3139,
      "step": 9654
    },
    {
      "epoch": 1.5114276768941766,
      "grad_norm": 1.2537013292312622,
      "learning_rate": 7.656402737047898e-05,
      "loss": 0.5404,
      "step": 9655
    },
    {
      "epoch": 1.5115842204132748,
      "grad_norm": 1.820609211921692,
      "learning_rate": 7.653958944281524e-05,
      "loss": 0.4524,
      "step": 9656
    },
    {
      "epoch": 1.5117407639323732,
      "grad_norm": 1.4393458366394043,
      "learning_rate": 7.651515151515151e-05,
      "loss": 0.2998,
      "step": 9657
    },
    {
      "epoch": 1.5118973074514717,
      "grad_norm": 0.7819269895553589,
      "learning_rate": 7.649071358748777e-05,
      "loss": 0.2722,
      "step": 9658
    },
    {
      "epoch": 1.5120538509705699,
      "grad_norm": 1.3116579055786133,
      "learning_rate": 7.646627565982404e-05,
      "loss": 0.3758,
      "step": 9659
    },
    {
      "epoch": 1.512210394489668,
      "grad_norm": 1.4176701307296753,
      "learning_rate": 7.64418377321603e-05,
      "loss": 0.4362,
      "step": 9660
    },
    {
      "epoch": 1.5123669380087663,
      "grad_norm": 1.270081877708435,
      "learning_rate": 7.641739980449658e-05,
      "loss": 0.4902,
      "step": 9661
    },
    {
      "epoch": 1.5125234815278648,
      "grad_norm": 1.4358556270599365,
      "learning_rate": 7.639296187683283e-05,
      "loss": 0.5373,
      "step": 9662
    },
    {
      "epoch": 1.5126800250469632,
      "grad_norm": 2.2279257774353027,
      "learning_rate": 7.63685239491691e-05,
      "loss": 0.6467,
      "step": 9663
    },
    {
      "epoch": 1.5128365685660614,
      "grad_norm": 1.7583202123641968,
      "learning_rate": 7.634408602150538e-05,
      "loss": 0.6748,
      "step": 9664
    },
    {
      "epoch": 1.5129931120851596,
      "grad_norm": 1.953381896018982,
      "learning_rate": 7.631964809384164e-05,
      "loss": 0.7467,
      "step": 9665
    },
    {
      "epoch": 1.5131496556042578,
      "grad_norm": 2.358065605163574,
      "learning_rate": 7.629521016617789e-05,
      "loss": 0.5326,
      "step": 9666
    },
    {
      "epoch": 1.5133061991233563,
      "grad_norm": 2.178169012069702,
      "learning_rate": 7.627077223851417e-05,
      "loss": 0.6585,
      "step": 9667
    },
    {
      "epoch": 1.5134627426424547,
      "grad_norm": 2.588186502456665,
      "learning_rate": 7.624633431085043e-05,
      "loss": 0.7107,
      "step": 9668
    },
    {
      "epoch": 1.513619286161553,
      "grad_norm": 3.952768087387085,
      "learning_rate": 7.62218963831867e-05,
      "loss": 0.578,
      "step": 9669
    },
    {
      "epoch": 1.5137758296806512,
      "grad_norm": 1.8726907968521118,
      "learning_rate": 7.619745845552296e-05,
      "loss": 0.3684,
      "step": 9670
    },
    {
      "epoch": 1.5139323731997494,
      "grad_norm": 11.03661823272705,
      "learning_rate": 7.617302052785923e-05,
      "loss": 0.8064,
      "step": 9671
    },
    {
      "epoch": 1.5140889167188478,
      "grad_norm": 1.7992969751358032,
      "learning_rate": 7.61485826001955e-05,
      "loss": 0.6892,
      "step": 9672
    },
    {
      "epoch": 1.5142454602379463,
      "grad_norm": 1.6284939050674438,
      "learning_rate": 7.612414467253177e-05,
      "loss": 0.4726,
      "step": 9673
    },
    {
      "epoch": 1.5144020037570445,
      "grad_norm": 4.366304874420166,
      "learning_rate": 7.609970674486802e-05,
      "loss": 1.0102,
      "step": 9674
    },
    {
      "epoch": 1.5145585472761427,
      "grad_norm": 2.8104560375213623,
      "learning_rate": 7.607526881720429e-05,
      "loss": 1.1905,
      "step": 9675
    },
    {
      "epoch": 1.514715090795241,
      "grad_norm": 3.4192862510681152,
      "learning_rate": 7.605083088954057e-05,
      "loss": 0.7504,
      "step": 9676
    },
    {
      "epoch": 1.5148716343143394,
      "grad_norm": 3.793952465057373,
      "learning_rate": 7.602639296187683e-05,
      "loss": 0.7833,
      "step": 9677
    },
    {
      "epoch": 1.5150281778334378,
      "grad_norm": 3.155576229095459,
      "learning_rate": 7.600195503421308e-05,
      "loss": 0.9845,
      "step": 9678
    },
    {
      "epoch": 1.515184721352536,
      "grad_norm": 2.9574239253997803,
      "learning_rate": 7.597751710654936e-05,
      "loss": 1.1077,
      "step": 9679
    },
    {
      "epoch": 1.5153412648716342,
      "grad_norm": 2.5134990215301514,
      "learning_rate": 7.595307917888563e-05,
      "loss": 0.9307,
      "step": 9680
    },
    {
      "epoch": 1.5154978083907327,
      "grad_norm": 2.6495211124420166,
      "learning_rate": 7.592864125122188e-05,
      "loss": 1.0685,
      "step": 9681
    },
    {
      "epoch": 1.5156543519098309,
      "grad_norm": 2.684701919555664,
      "learning_rate": 7.590420332355816e-05,
      "loss": 0.8099,
      "step": 9682
    },
    {
      "epoch": 1.5158108954289293,
      "grad_norm": 2.9182863235473633,
      "learning_rate": 7.587976539589442e-05,
      "loss": 1.5933,
      "step": 9683
    },
    {
      "epoch": 1.5159674389480275,
      "grad_norm": 3.108677387237549,
      "learning_rate": 7.585532746823068e-05,
      "loss": 0.9618,
      "step": 9684
    },
    {
      "epoch": 1.5161239824671258,
      "grad_norm": 1.0323936939239502,
      "learning_rate": 7.583088954056696e-05,
      "loss": 0.213,
      "step": 9685
    },
    {
      "epoch": 1.5162805259862242,
      "grad_norm": 2.6111085414886475,
      "learning_rate": 7.580645161290321e-05,
      "loss": 0.8856,
      "step": 9686
    },
    {
      "epoch": 1.5164370695053226,
      "grad_norm": 0.9904088973999023,
      "learning_rate": 7.578201368523948e-05,
      "loss": 0.4987,
      "step": 9687
    },
    {
      "epoch": 1.5165936130244209,
      "grad_norm": 1.8084644079208374,
      "learning_rate": 7.575757575757576e-05,
      "loss": 0.7905,
      "step": 9688
    },
    {
      "epoch": 1.516750156543519,
      "grad_norm": 0.8530622720718384,
      "learning_rate": 7.573313782991202e-05,
      "loss": 0.2475,
      "step": 9689
    },
    {
      "epoch": 1.5169067000626173,
      "grad_norm": 0.5058773756027222,
      "learning_rate": 7.570869990224827e-05,
      "loss": 0.2024,
      "step": 9690
    },
    {
      "epoch": 1.5170632435817157,
      "grad_norm": 0.4346686005592346,
      "learning_rate": 7.568426197458455e-05,
      "loss": 0.185,
      "step": 9691
    },
    {
      "epoch": 1.5172197871008142,
      "grad_norm": 0.6962853074073792,
      "learning_rate": 7.565982404692082e-05,
      "loss": 0.2676,
      "step": 9692
    },
    {
      "epoch": 1.5173763306199124,
      "grad_norm": 0.5338491201400757,
      "learning_rate": 7.563538611925707e-05,
      "loss": 0.243,
      "step": 9693
    },
    {
      "epoch": 1.5175328741390106,
      "grad_norm": 0.5298436284065247,
      "learning_rate": 7.561094819159335e-05,
      "loss": 0.2168,
      "step": 9694
    },
    {
      "epoch": 1.5176894176581088,
      "grad_norm": 0.6006895899772644,
      "learning_rate": 7.558651026392961e-05,
      "loss": 0.2639,
      "step": 9695
    },
    {
      "epoch": 1.5178459611772073,
      "grad_norm": 1.1772099733352661,
      "learning_rate": 7.556207233626588e-05,
      "loss": 0.2938,
      "step": 9696
    },
    {
      "epoch": 1.5180025046963057,
      "grad_norm": 2.0881564617156982,
      "learning_rate": 7.553763440860215e-05,
      "loss": 0.2997,
      "step": 9697
    },
    {
      "epoch": 1.518159048215404,
      "grad_norm": 0.7094396352767944,
      "learning_rate": 7.55131964809384e-05,
      "loss": 0.3687,
      "step": 9698
    },
    {
      "epoch": 1.5183155917345021,
      "grad_norm": 0.6079855561256409,
      "learning_rate": 7.548875855327467e-05,
      "loss": 0.2226,
      "step": 9699
    },
    {
      "epoch": 1.5184721352536004,
      "grad_norm": 0.6672441363334656,
      "learning_rate": 7.546432062561095e-05,
      "loss": 0.2565,
      "step": 9700
    },
    {
      "epoch": 1.5186286787726988,
      "grad_norm": 0.7931745648384094,
      "learning_rate": 7.543988269794721e-05,
      "loss": 0.3589,
      "step": 9701
    },
    {
      "epoch": 1.5187852222917972,
      "grad_norm": 0.5384275913238525,
      "learning_rate": 7.541544477028346e-05,
      "loss": 0.1624,
      "step": 9702
    },
    {
      "epoch": 1.5189417658108955,
      "grad_norm": 0.7884646058082581,
      "learning_rate": 7.539100684261974e-05,
      "loss": 0.2575,
      "step": 9703
    },
    {
      "epoch": 1.5190983093299937,
      "grad_norm": 0.8177517652511597,
      "learning_rate": 7.536656891495601e-05,
      "loss": 0.3014,
      "step": 9704
    },
    {
      "epoch": 1.519254852849092,
      "grad_norm": 1.0651932954788208,
      "learning_rate": 7.534213098729226e-05,
      "loss": 0.1904,
      "step": 9705
    },
    {
      "epoch": 1.5194113963681903,
      "grad_norm": 1.055517554283142,
      "learning_rate": 7.531769305962854e-05,
      "loss": 0.2615,
      "step": 9706
    },
    {
      "epoch": 1.5195679398872888,
      "grad_norm": 1.7985464334487915,
      "learning_rate": 7.52932551319648e-05,
      "loss": 0.5792,
      "step": 9707
    },
    {
      "epoch": 1.519724483406387,
      "grad_norm": 3.7123966217041016,
      "learning_rate": 7.526881720430107e-05,
      "loss": 0.5109,
      "step": 9708
    },
    {
      "epoch": 1.5198810269254852,
      "grad_norm": 1.6406701803207397,
      "learning_rate": 7.524437927663735e-05,
      "loss": 0.4718,
      "step": 9709
    },
    {
      "epoch": 1.5200375704445834,
      "grad_norm": 1.475159764289856,
      "learning_rate": 7.52199413489736e-05,
      "loss": 0.4324,
      "step": 9710
    },
    {
      "epoch": 1.5201941139636819,
      "grad_norm": 3.5248284339904785,
      "learning_rate": 7.519550342130986e-05,
      "loss": 0.663,
      "step": 9711
    },
    {
      "epoch": 1.5203506574827803,
      "grad_norm": 2.4344213008880615,
      "learning_rate": 7.517106549364614e-05,
      "loss": 0.4683,
      "step": 9712
    },
    {
      "epoch": 1.5205072010018785,
      "grad_norm": 1.7357611656188965,
      "learning_rate": 7.51466275659824e-05,
      "loss": 0.6168,
      "step": 9713
    },
    {
      "epoch": 1.5206637445209767,
      "grad_norm": 8.591681480407715,
      "learning_rate": 7.512218963831866e-05,
      "loss": 0.779,
      "step": 9714
    },
    {
      "epoch": 1.5208202880400752,
      "grad_norm": 3.0469913482666016,
      "learning_rate": 7.509775171065493e-05,
      "loss": 0.513,
      "step": 9715
    },
    {
      "epoch": 1.5209768315591734,
      "grad_norm": 4.075806617736816,
      "learning_rate": 7.50733137829912e-05,
      "loss": 0.5417,
      "step": 9716
    },
    {
      "epoch": 1.5211333750782718,
      "grad_norm": 1.6552000045776367,
      "learning_rate": 7.504887585532745e-05,
      "loss": 0.5785,
      "step": 9717
    },
    {
      "epoch": 1.52128991859737,
      "grad_norm": 2.410032033920288,
      "learning_rate": 7.502443792766373e-05,
      "loss": 0.8276,
      "step": 9718
    },
    {
      "epoch": 1.5214464621164683,
      "grad_norm": 1.1762977838516235,
      "learning_rate": 7.5e-05,
      "loss": 0.654,
      "step": 9719
    },
    {
      "epoch": 1.5216030056355667,
      "grad_norm": 5.700626373291016,
      "learning_rate": 7.497556207233626e-05,
      "loss": 0.7655,
      "step": 9720
    },
    {
      "epoch": 1.5217595491546652,
      "grad_norm": 3.678863763809204,
      "learning_rate": 7.495112414467252e-05,
      "loss": 1.2639,
      "step": 9721
    },
    {
      "epoch": 1.5219160926737634,
      "grad_norm": 2.683408737182617,
      "learning_rate": 7.492668621700879e-05,
      "loss": 0.81,
      "step": 9722
    },
    {
      "epoch": 1.5220726361928616,
      "grad_norm": 2.2185420989990234,
      "learning_rate": 7.490224828934507e-05,
      "loss": 0.763,
      "step": 9723
    },
    {
      "epoch": 1.5222291797119598,
      "grad_norm": 2.0188112258911133,
      "learning_rate": 7.487781036168132e-05,
      "loss": 0.7293,
      "step": 9724
    },
    {
      "epoch": 1.5223857232310583,
      "grad_norm": 2.3393964767456055,
      "learning_rate": 7.485337243401758e-05,
      "loss": 1.0011,
      "step": 9725
    },
    {
      "epoch": 1.5225422667501567,
      "grad_norm": 2.7806484699249268,
      "learning_rate": 7.482893450635386e-05,
      "loss": 0.8906,
      "step": 9726
    },
    {
      "epoch": 1.522698810269255,
      "grad_norm": 2.0687015056610107,
      "learning_rate": 7.480449657869011e-05,
      "loss": 0.8827,
      "step": 9727
    },
    {
      "epoch": 1.5228553537883531,
      "grad_norm": 2.678903818130493,
      "learning_rate": 7.478005865102639e-05,
      "loss": 1.36,
      "step": 9728
    },
    {
      "epoch": 1.5230118973074513,
      "grad_norm": 2.200491428375244,
      "learning_rate": 7.475562072336265e-05,
      "loss": 0.89,
      "step": 9729
    },
    {
      "epoch": 1.5231684408265498,
      "grad_norm": 2.768153429031372,
      "learning_rate": 7.473118279569892e-05,
      "loss": 1.4945,
      "step": 9730
    },
    {
      "epoch": 1.5233249843456482,
      "grad_norm": 6.975987434387207,
      "learning_rate": 7.470674486803518e-05,
      "loss": 1.177,
      "step": 9731
    },
    {
      "epoch": 1.5234815278647464,
      "grad_norm": 2.4341022968292236,
      "learning_rate": 7.468230694037145e-05,
      "loss": 1.1726,
      "step": 9732
    },
    {
      "epoch": 1.5236380713838447,
      "grad_norm": 4.012300491333008,
      "learning_rate": 7.465786901270771e-05,
      "loss": 0.6998,
      "step": 9733
    },
    {
      "epoch": 1.5237946149029429,
      "grad_norm": 2.198310136795044,
      "learning_rate": 7.463343108504398e-05,
      "loss": 0.8953,
      "step": 9734
    },
    {
      "epoch": 1.5239511584220413,
      "grad_norm": 1.1857575178146362,
      "learning_rate": 7.460899315738026e-05,
      "loss": 0.5788,
      "step": 9735
    },
    {
      "epoch": 1.5241077019411398,
      "grad_norm": 2.221917152404785,
      "learning_rate": 7.458455522971651e-05,
      "loss": 0.6252,
      "step": 9736
    },
    {
      "epoch": 1.524264245460238,
      "grad_norm": 4.886264801025391,
      "learning_rate": 7.456011730205277e-05,
      "loss": 0.6908,
      "step": 9737
    },
    {
      "epoch": 1.5244207889793362,
      "grad_norm": 2.3509812355041504,
      "learning_rate": 7.453567937438905e-05,
      "loss": 0.8521,
      "step": 9738
    },
    {
      "epoch": 1.5245773324984344,
      "grad_norm": 0.35476207733154297,
      "learning_rate": 7.45112414467253e-05,
      "loss": 0.1878,
      "step": 9739
    },
    {
      "epoch": 1.5247338760175329,
      "grad_norm": 3.2828149795532227,
      "learning_rate": 7.448680351906158e-05,
      "loss": 0.2251,
      "step": 9740
    },
    {
      "epoch": 1.5248904195366313,
      "grad_norm": 0.5039660334587097,
      "learning_rate": 7.446236559139785e-05,
      "loss": 0.19,
      "step": 9741
    },
    {
      "epoch": 1.5250469630557295,
      "grad_norm": 0.5465099215507507,
      "learning_rate": 7.443792766373411e-05,
      "loss": 0.2284,
      "step": 9742
    },
    {
      "epoch": 1.5252035065748277,
      "grad_norm": 1.618552803993225,
      "learning_rate": 7.441348973607038e-05,
      "loss": 0.326,
      "step": 9743
    },
    {
      "epoch": 1.525360050093926,
      "grad_norm": 0.585069477558136,
      "learning_rate": 7.438905180840664e-05,
      "loss": 0.3094,
      "step": 9744
    },
    {
      "epoch": 1.5255165936130244,
      "grad_norm": 1.64920175075531,
      "learning_rate": 7.43646138807429e-05,
      "loss": 0.3384,
      "step": 9745
    },
    {
      "epoch": 1.5256731371321228,
      "grad_norm": 1.8453181982040405,
      "learning_rate": 7.434017595307917e-05,
      "loss": 0.538,
      "step": 9746
    },
    {
      "epoch": 1.525829680651221,
      "grad_norm": 0.8853111267089844,
      "learning_rate": 7.431573802541543e-05,
      "loss": 0.255,
      "step": 9747
    },
    {
      "epoch": 1.5259862241703193,
      "grad_norm": 0.9123733043670654,
      "learning_rate": 7.42913000977517e-05,
      "loss": 0.1581,
      "step": 9748
    },
    {
      "epoch": 1.5261427676894177,
      "grad_norm": 0.7811870574951172,
      "learning_rate": 7.426686217008796e-05,
      "loss": 0.2773,
      "step": 9749
    },
    {
      "epoch": 1.526299311208516,
      "grad_norm": 0.824766993522644,
      "learning_rate": 7.424242424242424e-05,
      "loss": 0.4379,
      "step": 9750
    },
    {
      "epoch": 1.5264558547276144,
      "grad_norm": 1.7148675918579102,
      "learning_rate": 7.42179863147605e-05,
      "loss": 0.4431,
      "step": 9751
    },
    {
      "epoch": 1.5266123982467126,
      "grad_norm": 1.5956602096557617,
      "learning_rate": 7.419354838709677e-05,
      "loss": 0.3849,
      "step": 9752
    },
    {
      "epoch": 1.5267689417658108,
      "grad_norm": 1.4418418407440186,
      "learning_rate": 7.416911045943304e-05,
      "loss": 0.4567,
      "step": 9753
    },
    {
      "epoch": 1.5269254852849092,
      "grad_norm": 1.1421548128128052,
      "learning_rate": 7.41446725317693e-05,
      "loss": 0.5558,
      "step": 9754
    },
    {
      "epoch": 1.5270820288040077,
      "grad_norm": 2.0118582248687744,
      "learning_rate": 7.412023460410557e-05,
      "loss": 0.3335,
      "step": 9755
    },
    {
      "epoch": 1.527238572323106,
      "grad_norm": 1.003547191619873,
      "learning_rate": 7.409579667644183e-05,
      "loss": 0.4836,
      "step": 9756
    },
    {
      "epoch": 1.527395115842204,
      "grad_norm": 1.8095251321792603,
      "learning_rate": 7.40713587487781e-05,
      "loss": 0.4633,
      "step": 9757
    },
    {
      "epoch": 1.5275516593613023,
      "grad_norm": 0.8199507594108582,
      "learning_rate": 7.404692082111436e-05,
      "loss": 0.2844,
      "step": 9758
    },
    {
      "epoch": 1.5277082028804008,
      "grad_norm": 1.0319750308990479,
      "learning_rate": 7.402248289345063e-05,
      "loss": 0.5131,
      "step": 9759
    },
    {
      "epoch": 1.5278647463994992,
      "grad_norm": 1.8906883001327515,
      "learning_rate": 7.399804496578689e-05,
      "loss": 0.5979,
      "step": 9760
    },
    {
      "epoch": 1.5280212899185974,
      "grad_norm": 0.7006099820137024,
      "learning_rate": 7.397360703812316e-05,
      "loss": 0.288,
      "step": 9761
    },
    {
      "epoch": 1.5281778334376956,
      "grad_norm": 2.609555721282959,
      "learning_rate": 7.394916911045943e-05,
      "loss": 0.7689,
      "step": 9762
    },
    {
      "epoch": 1.5283343769567939,
      "grad_norm": 1.5572997331619263,
      "learning_rate": 7.392473118279569e-05,
      "loss": 0.6709,
      "step": 9763
    },
    {
      "epoch": 1.5284909204758923,
      "grad_norm": 1.1742448806762695,
      "learning_rate": 7.390029325513196e-05,
      "loss": 0.4997,
      "step": 9764
    },
    {
      "epoch": 1.5286474639949907,
      "grad_norm": 1.8424162864685059,
      "learning_rate": 7.387585532746823e-05,
      "loss": 0.636,
      "step": 9765
    },
    {
      "epoch": 1.528804007514089,
      "grad_norm": 1.5619405508041382,
      "learning_rate": 7.38514173998045e-05,
      "loss": 0.7927,
      "step": 9766
    },
    {
      "epoch": 1.5289605510331872,
      "grad_norm": 3.71563982963562,
      "learning_rate": 7.382697947214076e-05,
      "loss": 1.1046,
      "step": 9767
    },
    {
      "epoch": 1.5291170945522854,
      "grad_norm": 1.4069294929504395,
      "learning_rate": 7.380254154447702e-05,
      "loss": 0.3665,
      "step": 9768
    },
    {
      "epoch": 1.5292736380713838,
      "grad_norm": 2.09309983253479,
      "learning_rate": 7.377810361681329e-05,
      "loss": 0.6521,
      "step": 9769
    },
    {
      "epoch": 1.5294301815904823,
      "grad_norm": 1.4793744087219238,
      "learning_rate": 7.375366568914955e-05,
      "loss": 0.8215,
      "step": 9770
    },
    {
      "epoch": 1.5295867251095805,
      "grad_norm": 1.229730248451233,
      "learning_rate": 7.372922776148582e-05,
      "loss": 0.4851,
      "step": 9771
    },
    {
      "epoch": 1.5297432686286787,
      "grad_norm": 1.6996830701828003,
      "learning_rate": 7.370478983382208e-05,
      "loss": 0.5851,
      "step": 9772
    },
    {
      "epoch": 1.529899812147777,
      "grad_norm": 1.705766201019287,
      "learning_rate": 7.368035190615835e-05,
      "loss": 0.881,
      "step": 9773
    },
    {
      "epoch": 1.5300563556668754,
      "grad_norm": 1.5375521183013916,
      "learning_rate": 7.365591397849463e-05,
      "loss": 0.3521,
      "step": 9774
    },
    {
      "epoch": 1.5302128991859738,
      "grad_norm": 2.4411001205444336,
      "learning_rate": 7.363147605083088e-05,
      "loss": 0.9904,
      "step": 9775
    },
    {
      "epoch": 1.530369442705072,
      "grad_norm": 2.59539794921875,
      "learning_rate": 7.360703812316715e-05,
      "loss": 1.0197,
      "step": 9776
    },
    {
      "epoch": 1.5305259862241702,
      "grad_norm": 1.2721023559570312,
      "learning_rate": 7.358260019550342e-05,
      "loss": 0.8143,
      "step": 9777
    },
    {
      "epoch": 1.5306825297432687,
      "grad_norm": 3.4059696197509766,
      "learning_rate": 7.355816226783968e-05,
      "loss": 1.3697,
      "step": 9778
    },
    {
      "epoch": 1.530839073262367,
      "grad_norm": 1.2390133142471313,
      "learning_rate": 7.353372434017595e-05,
      "loss": 0.8997,
      "step": 9779
    },
    {
      "epoch": 1.5309956167814653,
      "grad_norm": 2.6192235946655273,
      "learning_rate": 7.350928641251221e-05,
      "loss": 1.2236,
      "step": 9780
    },
    {
      "epoch": 1.5311521603005636,
      "grad_norm": 2.4125397205352783,
      "learning_rate": 7.348484848484848e-05,
      "loss": 0.9419,
      "step": 9781
    },
    {
      "epoch": 1.5313087038196618,
      "grad_norm": 3.147789716720581,
      "learning_rate": 7.346041055718474e-05,
      "loss": 1.8505,
      "step": 9782
    },
    {
      "epoch": 1.5314652473387602,
      "grad_norm": 2.7218902111053467,
      "learning_rate": 7.343597262952101e-05,
      "loss": 0.5833,
      "step": 9783
    },
    {
      "epoch": 1.5316217908578584,
      "grad_norm": 1.0626882314682007,
      "learning_rate": 7.341153470185727e-05,
      "loss": 0.3275,
      "step": 9784
    },
    {
      "epoch": 1.5317783343769569,
      "grad_norm": 2.7859811782836914,
      "learning_rate": 7.338709677419354e-05,
      "loss": 0.8926,
      "step": 9785
    },
    {
      "epoch": 1.531934877896055,
      "grad_norm": 2.831791877746582,
      "learning_rate": 7.33626588465298e-05,
      "loss": 0.5686,
      "step": 9786
    },
    {
      "epoch": 1.5320914214151533,
      "grad_norm": 2.355695962905884,
      "learning_rate": 7.333822091886607e-05,
      "loss": 0.5434,
      "step": 9787
    },
    {
      "epoch": 1.5322479649342517,
      "grad_norm": 1.321400761604309,
      "learning_rate": 7.331378299120235e-05,
      "loss": 0.5117,
      "step": 9788
    },
    {
      "epoch": 1.5324045084533502,
      "grad_norm": 0.5803260207176208,
      "learning_rate": 7.32893450635386e-05,
      "loss": 0.2592,
      "step": 9789
    },
    {
      "epoch": 1.5325610519724484,
      "grad_norm": 0.40602564811706543,
      "learning_rate": 7.326490713587488e-05,
      "loss": 0.1622,
      "step": 9790
    },
    {
      "epoch": 1.5327175954915466,
      "grad_norm": 1.1273137331008911,
      "learning_rate": 7.324046920821114e-05,
      "loss": 0.2955,
      "step": 9791
    },
    {
      "epoch": 1.5328741390106448,
      "grad_norm": 0.4123140573501587,
      "learning_rate": 7.32160312805474e-05,
      "loss": 0.2276,
      "step": 9792
    },
    {
      "epoch": 1.5330306825297433,
      "grad_norm": 0.5335753560066223,
      "learning_rate": 7.319159335288367e-05,
      "loss": 0.1589,
      "step": 9793
    },
    {
      "epoch": 1.5331872260488417,
      "grad_norm": 0.8474439382553101,
      "learning_rate": 7.316715542521993e-05,
      "loss": 0.3083,
      "step": 9794
    },
    {
      "epoch": 1.53334376956794,
      "grad_norm": 0.48397138714790344,
      "learning_rate": 7.31427174975562e-05,
      "loss": 0.1776,
      "step": 9795
    },
    {
      "epoch": 1.5335003130870382,
      "grad_norm": 0.5532123446464539,
      "learning_rate": 7.311827956989246e-05,
      "loss": 0.2416,
      "step": 9796
    },
    {
      "epoch": 1.5336568566061364,
      "grad_norm": 0.656062126159668,
      "learning_rate": 7.309384164222873e-05,
      "loss": 0.2122,
      "step": 9797
    },
    {
      "epoch": 1.5338134001252348,
      "grad_norm": 1.0843610763549805,
      "learning_rate": 7.3069403714565e-05,
      "loss": 0.2524,
      "step": 9798
    },
    {
      "epoch": 1.5339699436443333,
      "grad_norm": 0.891477644443512,
      "learning_rate": 7.304496578690126e-05,
      "loss": 0.437,
      "step": 9799
    },
    {
      "epoch": 1.5341264871634315,
      "grad_norm": 1.0225049257278442,
      "learning_rate": 7.302052785923754e-05,
      "loss": 0.2953,
      "step": 9800
    },
    {
      "epoch": 1.5342830306825297,
      "grad_norm": 1.2247625589370728,
      "learning_rate": 7.299608993157379e-05,
      "loss": 0.492,
      "step": 9801
    },
    {
      "epoch": 1.534439574201628,
      "grad_norm": 0.9728368520736694,
      "learning_rate": 7.297165200391007e-05,
      "loss": 0.3037,
      "step": 9802
    },
    {
      "epoch": 1.5345961177207263,
      "grad_norm": 1.27720046043396,
      "learning_rate": 7.294721407624633e-05,
      "loss": 0.1679,
      "step": 9803
    },
    {
      "epoch": 1.5347526612398248,
      "grad_norm": 0.6740818023681641,
      "learning_rate": 7.29227761485826e-05,
      "loss": 0.1985,
      "step": 9804
    },
    {
      "epoch": 1.534909204758923,
      "grad_norm": 1.0346983671188354,
      "learning_rate": 7.289833822091886e-05,
      "loss": 0.4624,
      "step": 9805
    },
    {
      "epoch": 1.5350657482780212,
      "grad_norm": 1.6507031917572021,
      "learning_rate": 7.287390029325513e-05,
      "loss": 0.1886,
      "step": 9806
    },
    {
      "epoch": 1.5352222917971194,
      "grad_norm": 1.1130856275558472,
      "learning_rate": 7.284946236559139e-05,
      "loss": 0.3239,
      "step": 9807
    },
    {
      "epoch": 1.5353788353162179,
      "grad_norm": 1.4110175371170044,
      "learning_rate": 7.282502443792766e-05,
      "loss": 0.5766,
      "step": 9808
    },
    {
      "epoch": 1.5355353788353163,
      "grad_norm": 15.8782377243042,
      "learning_rate": 7.280058651026392e-05,
      "loss": 0.4304,
      "step": 9809
    },
    {
      "epoch": 1.5356919223544145,
      "grad_norm": 2.349193811416626,
      "learning_rate": 7.277614858260019e-05,
      "loss": 0.4212,
      "step": 9810
    },
    {
      "epoch": 1.5358484658735128,
      "grad_norm": 2.3084702491760254,
      "learning_rate": 7.275171065493645e-05,
      "loss": 0.7621,
      "step": 9811
    },
    {
      "epoch": 1.5360050093926112,
      "grad_norm": 2.6022043228149414,
      "learning_rate": 7.272727272727273e-05,
      "loss": 0.6079,
      "step": 9812
    },
    {
      "epoch": 1.5361615529117094,
      "grad_norm": 1.9148122072219849,
      "learning_rate": 7.270283479960898e-05,
      "loss": 0.505,
      "step": 9813
    },
    {
      "epoch": 1.5363180964308079,
      "grad_norm": 3.0425705909729004,
      "learning_rate": 7.267839687194526e-05,
      "loss": 0.9097,
      "step": 9814
    },
    {
      "epoch": 1.536474639949906,
      "grad_norm": 2.4766037464141846,
      "learning_rate": 7.265395894428152e-05,
      "loss": 0.6503,
      "step": 9815
    },
    {
      "epoch": 1.5366311834690043,
      "grad_norm": 2.7663185596466064,
      "learning_rate": 7.262952101661779e-05,
      "loss": 0.8135,
      "step": 9816
    },
    {
      "epoch": 1.5367877269881027,
      "grad_norm": 1.486636996269226,
      "learning_rate": 7.260508308895405e-05,
      "loss": 0.4896,
      "step": 9817
    },
    {
      "epoch": 1.536944270507201,
      "grad_norm": 1.2840673923492432,
      "learning_rate": 7.258064516129032e-05,
      "loss": 0.3703,
      "step": 9818
    },
    {
      "epoch": 1.5371008140262994,
      "grad_norm": 1.9025120735168457,
      "learning_rate": 7.255620723362658e-05,
      "loss": 0.7038,
      "step": 9819
    },
    {
      "epoch": 1.5372573575453976,
      "grad_norm": 3.2480320930480957,
      "learning_rate": 7.253176930596285e-05,
      "loss": 0.5779,
      "step": 9820
    },
    {
      "epoch": 1.5374139010644958,
      "grad_norm": 2.224262237548828,
      "learning_rate": 7.250733137829911e-05,
      "loss": 1.0415,
      "step": 9821
    },
    {
      "epoch": 1.5375704445835943,
      "grad_norm": 2.9639852046966553,
      "learning_rate": 7.248289345063538e-05,
      "loss": 1.0646,
      "step": 9822
    },
    {
      "epoch": 1.5377269881026927,
      "grad_norm": 2.493807077407837,
      "learning_rate": 7.245845552297164e-05,
      "loss": 0.7018,
      "step": 9823
    },
    {
      "epoch": 1.537883531621791,
      "grad_norm": 2.37644624710083,
      "learning_rate": 7.243401759530792e-05,
      "loss": 0.4942,
      "step": 9824
    },
    {
      "epoch": 1.5380400751408891,
      "grad_norm": 2.5154953002929688,
      "learning_rate": 7.240957966764417e-05,
      "loss": 1.0169,
      "step": 9825
    },
    {
      "epoch": 1.5381966186599874,
      "grad_norm": 1.9995393753051758,
      "learning_rate": 7.238514173998045e-05,
      "loss": 1.0743,
      "step": 9826
    },
    {
      "epoch": 1.5383531621790858,
      "grad_norm": 2.441619396209717,
      "learning_rate": 7.236070381231671e-05,
      "loss": 1.4467,
      "step": 9827
    },
    {
      "epoch": 1.5385097056981842,
      "grad_norm": 2.3574626445770264,
      "learning_rate": 7.233626588465298e-05,
      "loss": 0.5267,
      "step": 9828
    },
    {
      "epoch": 1.5386662492172825,
      "grad_norm": 1.9852875471115112,
      "learning_rate": 7.231182795698924e-05,
      "loss": 0.769,
      "step": 9829
    },
    {
      "epoch": 1.5388227927363807,
      "grad_norm": 5.952073574066162,
      "learning_rate": 7.228739002932551e-05,
      "loss": 1.0293,
      "step": 9830
    },
    {
      "epoch": 1.538979336255479,
      "grad_norm": 3.0914673805236816,
      "learning_rate": 7.226295210166177e-05,
      "loss": 1.2532,
      "step": 9831
    },
    {
      "epoch": 1.5391358797745773,
      "grad_norm": 2.5825436115264893,
      "learning_rate": 7.223851417399804e-05,
      "loss": 0.9199,
      "step": 9832
    },
    {
      "epoch": 1.5392924232936758,
      "grad_norm": 2.3879377841949463,
      "learning_rate": 7.22140762463343e-05,
      "loss": 1.0303,
      "step": 9833
    },
    {
      "epoch": 1.539448966812774,
      "grad_norm": 0.8389413356781006,
      "learning_rate": 7.218963831867057e-05,
      "loss": 0.1337,
      "step": 9834
    },
    {
      "epoch": 1.5396055103318722,
      "grad_norm": 1.7144027948379517,
      "learning_rate": 7.216520039100683e-05,
      "loss": 0.8003,
      "step": 9835
    },
    {
      "epoch": 1.5397620538509704,
      "grad_norm": 3.8174052238464355,
      "learning_rate": 7.214076246334311e-05,
      "loss": 1.1162,
      "step": 9836
    },
    {
      "epoch": 1.5399185973700689,
      "grad_norm": 3.3473262786865234,
      "learning_rate": 7.211632453567936e-05,
      "loss": 1.6102,
      "step": 9837
    },
    {
      "epoch": 1.5400751408891673,
      "grad_norm": 1.6419070959091187,
      "learning_rate": 7.209188660801564e-05,
      "loss": 0.6261,
      "step": 9838
    },
    {
      "epoch": 1.5402316844082655,
      "grad_norm": 0.6053900718688965,
      "learning_rate": 7.20674486803519e-05,
      "loss": 0.198,
      "step": 9839
    },
    {
      "epoch": 1.5403882279273637,
      "grad_norm": 0.49573662877082825,
      "learning_rate": 7.204301075268816e-05,
      "loss": 0.2027,
      "step": 9840
    },
    {
      "epoch": 1.540544771446462,
      "grad_norm": 0.4826977849006653,
      "learning_rate": 7.201857282502443e-05,
      "loss": 0.1637,
      "step": 9841
    },
    {
      "epoch": 1.5407013149655604,
      "grad_norm": 0.7587850689888,
      "learning_rate": 7.19941348973607e-05,
      "loss": 0.1888,
      "step": 9842
    },
    {
      "epoch": 1.5408578584846588,
      "grad_norm": 0.565876841545105,
      "learning_rate": 7.196969696969696e-05,
      "loss": 0.1634,
      "step": 9843
    },
    {
      "epoch": 1.541014402003757,
      "grad_norm": 0.4626244604587555,
      "learning_rate": 7.194525904203323e-05,
      "loss": 0.2195,
      "step": 9844
    },
    {
      "epoch": 1.5411709455228553,
      "grad_norm": 1.5018632411956787,
      "learning_rate": 7.19208211143695e-05,
      "loss": 0.3624,
      "step": 9845
    },
    {
      "epoch": 1.5413274890419537,
      "grad_norm": 0.6011160016059875,
      "learning_rate": 7.189638318670576e-05,
      "loss": 0.2698,
      "step": 9846
    },
    {
      "epoch": 1.541484032561052,
      "grad_norm": 1.015252947807312,
      "learning_rate": 7.187194525904202e-05,
      "loss": 0.3386,
      "step": 9847
    },
    {
      "epoch": 1.5416405760801504,
      "grad_norm": 1.5964446067810059,
      "learning_rate": 7.18475073313783e-05,
      "loss": 0.1781,
      "step": 9848
    },
    {
      "epoch": 1.5417971195992486,
      "grad_norm": 1.3437423706054688,
      "learning_rate": 7.182306940371455e-05,
      "loss": 0.237,
      "step": 9849
    },
    {
      "epoch": 1.5419536631183468,
      "grad_norm": 0.8205925822257996,
      "learning_rate": 7.179863147605083e-05,
      "loss": 0.2588,
      "step": 9850
    },
    {
      "epoch": 1.5421102066374452,
      "grad_norm": 1.1015372276306152,
      "learning_rate": 7.17741935483871e-05,
      "loss": 0.3411,
      "step": 9851
    },
    {
      "epoch": 1.5422667501565435,
      "grad_norm": 3.2911665439605713,
      "learning_rate": 7.174975562072335e-05,
      "loss": 0.2773,
      "step": 9852
    },
    {
      "epoch": 1.542423293675642,
      "grad_norm": 6.5218186378479,
      "learning_rate": 7.172531769305963e-05,
      "loss": 0.4853,
      "step": 9853
    },
    {
      "epoch": 1.5425798371947401,
      "grad_norm": 1.0145297050476074,
      "learning_rate": 7.170087976539589e-05,
      "loss": 0.233,
      "step": 9854
    },
    {
      "epoch": 1.5427363807138383,
      "grad_norm": 1.0070481300354004,
      "learning_rate": 7.167644183773216e-05,
      "loss": 0.3493,
      "step": 9855
    },
    {
      "epoch": 1.5428929242329368,
      "grad_norm": 2.540808916091919,
      "learning_rate": 7.165200391006842e-05,
      "loss": 0.5796,
      "step": 9856
    },
    {
      "epoch": 1.5430494677520352,
      "grad_norm": 2.2027969360351562,
      "learning_rate": 7.162756598240468e-05,
      "loss": 0.4986,
      "step": 9857
    },
    {
      "epoch": 1.5432060112711334,
      "grad_norm": 2.4842472076416016,
      "learning_rate": 7.160312805474095e-05,
      "loss": 0.427,
      "step": 9858
    },
    {
      "epoch": 1.5433625547902317,
      "grad_norm": 2.618431568145752,
      "learning_rate": 7.157869012707721e-05,
      "loss": 0.4044,
      "step": 9859
    },
    {
      "epoch": 1.5435190983093299,
      "grad_norm": 2.222273349761963,
      "learning_rate": 7.155425219941349e-05,
      "loss": 0.5528,
      "step": 9860
    },
    {
      "epoch": 1.5436756418284283,
      "grad_norm": 1.5825259685516357,
      "learning_rate": 7.152981427174974e-05,
      "loss": 0.5638,
      "step": 9861
    },
    {
      "epoch": 1.5438321853475268,
      "grad_norm": 12.67103099822998,
      "learning_rate": 7.150537634408601e-05,
      "loss": 0.4018,
      "step": 9862
    },
    {
      "epoch": 1.543988728866625,
      "grad_norm": 1.7103662490844727,
      "learning_rate": 7.148093841642229e-05,
      "loss": 0.5984,
      "step": 9863
    },
    {
      "epoch": 1.5441452723857232,
      "grad_norm": 1.5723588466644287,
      "learning_rate": 7.145650048875854e-05,
      "loss": 0.6249,
      "step": 9864
    },
    {
      "epoch": 1.5443018159048214,
      "grad_norm": 1.4810919761657715,
      "learning_rate": 7.143206256109482e-05,
      "loss": 0.6729,
      "step": 9865
    },
    {
      "epoch": 1.5444583594239198,
      "grad_norm": 2.4991166591644287,
      "learning_rate": 7.140762463343108e-05,
      "loss": 0.373,
      "step": 9866
    },
    {
      "epoch": 1.5446149029430183,
      "grad_norm": 1.0830026865005493,
      "learning_rate": 7.138318670576735e-05,
      "loss": 0.3133,
      "step": 9867
    },
    {
      "epoch": 1.5447714464621165,
      "grad_norm": 1.9214036464691162,
      "learning_rate": 7.135874877810361e-05,
      "loss": 0.5774,
      "step": 9868
    },
    {
      "epoch": 1.5449279899812147,
      "grad_norm": 1.623348355293274,
      "learning_rate": 7.133431085043988e-05,
      "loss": 0.7693,
      "step": 9869
    },
    {
      "epoch": 1.545084533500313,
      "grad_norm": 2.111285924911499,
      "learning_rate": 7.130987292277614e-05,
      "loss": 0.7588,
      "step": 9870
    },
    {
      "epoch": 1.5452410770194114,
      "grad_norm": 4.0085248947143555,
      "learning_rate": 7.12854349951124e-05,
      "loss": 0.633,
      "step": 9871
    },
    {
      "epoch": 1.5453976205385098,
      "grad_norm": 1.885757327079773,
      "learning_rate": 7.126099706744868e-05,
      "loss": 0.5439,
      "step": 9872
    },
    {
      "epoch": 1.545554164057608,
      "grad_norm": 2.1260712146759033,
      "learning_rate": 7.123655913978494e-05,
      "loss": 0.5721,
      "step": 9873
    },
    {
      "epoch": 1.5457107075767063,
      "grad_norm": 1.1092246770858765,
      "learning_rate": 7.12121212121212e-05,
      "loss": 0.598,
      "step": 9874
    },
    {
      "epoch": 1.5458672510958045,
      "grad_norm": 2.1099350452423096,
      "learning_rate": 7.118768328445748e-05,
      "loss": 1.0518,
      "step": 9875
    },
    {
      "epoch": 1.546023794614903,
      "grad_norm": 2.708494186401367,
      "learning_rate": 7.116324535679373e-05,
      "loss": 0.974,
      "step": 9876
    },
    {
      "epoch": 1.5461803381340014,
      "grad_norm": 2.1030566692352295,
      "learning_rate": 7.113880742913001e-05,
      "loss": 1.064,
      "step": 9877
    },
    {
      "epoch": 1.5463368816530996,
      "grad_norm": 1.5087107419967651,
      "learning_rate": 7.111436950146627e-05,
      "loss": 0.6343,
      "step": 9878
    },
    {
      "epoch": 1.5464934251721978,
      "grad_norm": 2.708134651184082,
      "learning_rate": 7.108993157380254e-05,
      "loss": 1.0731,
      "step": 9879
    },
    {
      "epoch": 1.5466499686912962,
      "grad_norm": 2.4557900428771973,
      "learning_rate": 7.10654936461388e-05,
      "loss": 1.0076,
      "step": 9880
    },
    {
      "epoch": 1.5468065122103944,
      "grad_norm": 4.119977951049805,
      "learning_rate": 7.104105571847507e-05,
      "loss": 0.6316,
      "step": 9881
    },
    {
      "epoch": 1.5469630557294929,
      "grad_norm": 9.010337829589844,
      "learning_rate": 7.101661779081133e-05,
      "loss": 1.0986,
      "step": 9882
    },
    {
      "epoch": 1.547119599248591,
      "grad_norm": 3.3070311546325684,
      "learning_rate": 7.09921798631476e-05,
      "loss": 1.4029,
      "step": 9883
    },
    {
      "epoch": 1.5472761427676893,
      "grad_norm": 1.8002490997314453,
      "learning_rate": 7.096774193548386e-05,
      "loss": 0.467,
      "step": 9884
    },
    {
      "epoch": 1.5474326862867878,
      "grad_norm": 1.5535470247268677,
      "learning_rate": 7.094330400782013e-05,
      "loss": 0.6966,
      "step": 9885
    },
    {
      "epoch": 1.5475892298058862,
      "grad_norm": 1.3492341041564941,
      "learning_rate": 7.091886608015639e-05,
      "loss": 0.4247,
      "step": 9886
    },
    {
      "epoch": 1.5477457733249844,
      "grad_norm": 1.93979811668396,
      "learning_rate": 7.089442815249267e-05,
      "loss": 0.8019,
      "step": 9887
    },
    {
      "epoch": 1.5479023168440826,
      "grad_norm": 1.897113561630249,
      "learning_rate": 7.086999022482892e-05,
      "loss": 0.8177,
      "step": 9888
    },
    {
      "epoch": 1.5480588603631809,
      "grad_norm": 0.4913039207458496,
      "learning_rate": 7.08455522971652e-05,
      "loss": 0.2577,
      "step": 9889
    },
    {
      "epoch": 1.5482154038822793,
      "grad_norm": 0.9754254221916199,
      "learning_rate": 7.082111436950146e-05,
      "loss": 0.2678,
      "step": 9890
    },
    {
      "epoch": 1.5483719474013777,
      "grad_norm": 0.5141871571540833,
      "learning_rate": 7.079667644183773e-05,
      "loss": 0.2229,
      "step": 9891
    },
    {
      "epoch": 1.548528490920476,
      "grad_norm": 0.7318794131278992,
      "learning_rate": 7.0772238514174e-05,
      "loss": 0.2285,
      "step": 9892
    },
    {
      "epoch": 1.5486850344395742,
      "grad_norm": 1.011592984199524,
      "learning_rate": 7.074780058651026e-05,
      "loss": 0.2582,
      "step": 9893
    },
    {
      "epoch": 1.5488415779586724,
      "grad_norm": 0.5550967454910278,
      "learning_rate": 7.072336265884652e-05,
      "loss": 0.3078,
      "step": 9894
    },
    {
      "epoch": 1.5489981214777708,
      "grad_norm": 0.6541218757629395,
      "learning_rate": 7.069892473118279e-05,
      "loss": 0.1527,
      "step": 9895
    },
    {
      "epoch": 1.5491546649968693,
      "grad_norm": 0.6758529543876648,
      "learning_rate": 7.067448680351905e-05,
      "loss": 0.309,
      "step": 9896
    },
    {
      "epoch": 1.5493112085159675,
      "grad_norm": 0.5618512630462646,
      "learning_rate": 7.065004887585532e-05,
      "loss": 0.234,
      "step": 9897
    },
    {
      "epoch": 1.5494677520350657,
      "grad_norm": 0.8279595971107483,
      "learning_rate": 7.062561094819158e-05,
      "loss": 0.3533,
      "step": 9898
    },
    {
      "epoch": 1.549624295554164,
      "grad_norm": 0.7879864573478699,
      "learning_rate": 7.060117302052786e-05,
      "loss": 0.1606,
      "step": 9899
    },
    {
      "epoch": 1.5497808390732624,
      "grad_norm": 0.8414984941482544,
      "learning_rate": 7.057673509286411e-05,
      "loss": 0.33,
      "step": 9900
    },
    {
      "epoch": 1.5499373825923608,
      "grad_norm": 0.6734075546264648,
      "learning_rate": 7.055229716520039e-05,
      "loss": 0.3478,
      "step": 9901
    },
    {
      "epoch": 1.550093926111459,
      "grad_norm": 0.876607358455658,
      "learning_rate": 7.052785923753666e-05,
      "loss": 0.4642,
      "step": 9902
    },
    {
      "epoch": 1.5502504696305572,
      "grad_norm": 1.32351553440094,
      "learning_rate": 7.050342130987292e-05,
      "loss": 0.4359,
      "step": 9903
    },
    {
      "epoch": 1.5504070131496555,
      "grad_norm": 0.7632149457931519,
      "learning_rate": 7.047898338220918e-05,
      "loss": 0.3068,
      "step": 9904
    },
    {
      "epoch": 1.550563556668754,
      "grad_norm": 1.024189829826355,
      "learning_rate": 7.045454545454545e-05,
      "loss": 0.2885,
      "step": 9905
    },
    {
      "epoch": 1.5507201001878523,
      "grad_norm": 1.50118887424469,
      "learning_rate": 7.043010752688171e-05,
      "loss": 0.3361,
      "step": 9906
    },
    {
      "epoch": 1.5508766437069506,
      "grad_norm": 3.2080376148223877,
      "learning_rate": 7.040566959921798e-05,
      "loss": 0.7564,
      "step": 9907
    },
    {
      "epoch": 1.5510331872260488,
      "grad_norm": 1.7170158624649048,
      "learning_rate": 7.038123167155424e-05,
      "loss": 0.471,
      "step": 9908
    },
    {
      "epoch": 1.551189730745147,
      "grad_norm": 2.007455825805664,
      "learning_rate": 7.035679374389051e-05,
      "loss": 0.4193,
      "step": 9909
    },
    {
      "epoch": 1.5513462742642454,
      "grad_norm": 1.438209891319275,
      "learning_rate": 7.033235581622677e-05,
      "loss": 0.4327,
      "step": 9910
    },
    {
      "epoch": 1.5515028177833439,
      "grad_norm": 1.6576300859451294,
      "learning_rate": 7.030791788856304e-05,
      "loss": 0.5447,
      "step": 9911
    },
    {
      "epoch": 1.551659361302442,
      "grad_norm": 1.656591534614563,
      "learning_rate": 7.02834799608993e-05,
      "loss": 0.9359,
      "step": 9912
    },
    {
      "epoch": 1.5518159048215403,
      "grad_norm": 2.055999279022217,
      "learning_rate": 7.025904203323558e-05,
      "loss": 0.7214,
      "step": 9913
    },
    {
      "epoch": 1.5519724483406387,
      "grad_norm": 2.960740089416504,
      "learning_rate": 7.023460410557183e-05,
      "loss": 0.8004,
      "step": 9914
    },
    {
      "epoch": 1.552128991859737,
      "grad_norm": 1.1211111545562744,
      "learning_rate": 7.021016617790811e-05,
      "loss": 0.2811,
      "step": 9915
    },
    {
      "epoch": 1.5522855353788354,
      "grad_norm": 0.9293329119682312,
      "learning_rate": 7.018572825024438e-05,
      "loss": 0.4153,
      "step": 9916
    },
    {
      "epoch": 1.5524420788979336,
      "grad_norm": 1.6393110752105713,
      "learning_rate": 7.016129032258064e-05,
      "loss": 0.6008,
      "step": 9917
    },
    {
      "epoch": 1.5525986224170318,
      "grad_norm": 2.427534341812134,
      "learning_rate": 7.01368523949169e-05,
      "loss": 0.3159,
      "step": 9918
    },
    {
      "epoch": 1.5527551659361303,
      "grad_norm": 2.0107951164245605,
      "learning_rate": 7.011241446725317e-05,
      "loss": 0.677,
      "step": 9919
    },
    {
      "epoch": 1.5529117094552287,
      "grad_norm": 1.3930466175079346,
      "learning_rate": 7.008797653958944e-05,
      "loss": 0.4769,
      "step": 9920
    },
    {
      "epoch": 1.553068252974327,
      "grad_norm": 3.030660390853882,
      "learning_rate": 7.00635386119257e-05,
      "loss": 0.6934,
      "step": 9921
    },
    {
      "epoch": 1.5532247964934252,
      "grad_norm": 1.1675511598587036,
      "learning_rate": 7.003910068426196e-05,
      "loss": 0.3575,
      "step": 9922
    },
    {
      "epoch": 1.5533813400125234,
      "grad_norm": 2.445157527923584,
      "learning_rate": 7.001466275659823e-05,
      "loss": 0.7557,
      "step": 9923
    },
    {
      "epoch": 1.5535378835316218,
      "grad_norm": 2.884352445602417,
      "learning_rate": 6.99902248289345e-05,
      "loss": 0.7685,
      "step": 9924
    },
    {
      "epoch": 1.5536944270507203,
      "grad_norm": 1.9358322620391846,
      "learning_rate": 6.996578690127077e-05,
      "loss": 0.7295,
      "step": 9925
    },
    {
      "epoch": 1.5538509705698185,
      "grad_norm": 2.95656681060791,
      "learning_rate": 6.994134897360702e-05,
      "loss": 0.6324,
      "step": 9926
    },
    {
      "epoch": 1.5540075140889167,
      "grad_norm": 1.8009204864501953,
      "learning_rate": 6.99169110459433e-05,
      "loss": 0.4526,
      "step": 9927
    },
    {
      "epoch": 1.554164057608015,
      "grad_norm": 3.0422096252441406,
      "learning_rate": 6.989247311827957e-05,
      "loss": 0.9872,
      "step": 9928
    },
    {
      "epoch": 1.5543206011271133,
      "grad_norm": 4.271030902862549,
      "learning_rate": 6.986803519061583e-05,
      "loss": 0.8607,
      "step": 9929
    },
    {
      "epoch": 1.5544771446462118,
      "grad_norm": 2.318971633911133,
      "learning_rate": 6.98435972629521e-05,
      "loss": 0.8247,
      "step": 9930
    },
    {
      "epoch": 1.55463368816531,
      "grad_norm": 1.7466161251068115,
      "learning_rate": 6.981915933528836e-05,
      "loss": 0.9417,
      "step": 9931
    },
    {
      "epoch": 1.5547902316844082,
      "grad_norm": 2.927077531814575,
      "learning_rate": 6.979472140762463e-05,
      "loss": 1.8184,
      "step": 9932
    },
    {
      "epoch": 1.5549467752035064,
      "grad_norm": 2.963963747024536,
      "learning_rate": 6.977028347996089e-05,
      "loss": 0.8251,
      "step": 9933
    },
    {
      "epoch": 1.5551033187226049,
      "grad_norm": 2.6687653064727783,
      "learning_rate": 6.974584555229716e-05,
      "loss": 0.8453,
      "step": 9934
    },
    {
      "epoch": 1.5552598622417033,
      "grad_norm": 3.3215959072113037,
      "learning_rate": 6.972140762463342e-05,
      "loss": 0.7546,
      "step": 9935
    },
    {
      "epoch": 1.5554164057608015,
      "grad_norm": 2.8623807430267334,
      "learning_rate": 6.969696969696969e-05,
      "loss": 1.5532,
      "step": 9936
    },
    {
      "epoch": 1.5555729492798998,
      "grad_norm": 2.0836265087127686,
      "learning_rate": 6.967253176930596e-05,
      "loss": 0.8139,
      "step": 9937
    },
    {
      "epoch": 1.555729492798998,
      "grad_norm": 1.388414978981018,
      "learning_rate": 6.964809384164222e-05,
      "loss": 0.8304,
      "step": 9938
    },
    {
      "epoch": 1.5558860363180964,
      "grad_norm": 0.9456003904342651,
      "learning_rate": 6.96236559139785e-05,
      "loss": 0.1613,
      "step": 9939
    },
    {
      "epoch": 1.5560425798371949,
      "grad_norm": 0.5043039917945862,
      "learning_rate": 6.959921798631476e-05,
      "loss": 0.121,
      "step": 9940
    },
    {
      "epoch": 1.556199123356293,
      "grad_norm": 0.7678887248039246,
      "learning_rate": 6.957478005865102e-05,
      "loss": 0.2142,
      "step": 9941
    },
    {
      "epoch": 1.5563556668753913,
      "grad_norm": 0.7166454792022705,
      "learning_rate": 6.955034213098729e-05,
      "loss": 0.2479,
      "step": 9942
    },
    {
      "epoch": 1.5565122103944895,
      "grad_norm": 0.4818023145198822,
      "learning_rate": 6.952590420332355e-05,
      "loss": 0.2025,
      "step": 9943
    },
    {
      "epoch": 1.556668753913588,
      "grad_norm": 0.3814961314201355,
      "learning_rate": 6.950146627565982e-05,
      "loss": 0.1756,
      "step": 9944
    },
    {
      "epoch": 1.5568252974326864,
      "grad_norm": 0.6317948698997498,
      "learning_rate": 6.947702834799608e-05,
      "loss": 0.2258,
      "step": 9945
    },
    {
      "epoch": 1.5569818409517846,
      "grad_norm": 0.6837334036827087,
      "learning_rate": 6.945259042033235e-05,
      "loss": 0.2127,
      "step": 9946
    },
    {
      "epoch": 1.5571383844708828,
      "grad_norm": 0.6503742337226868,
      "learning_rate": 6.942815249266861e-05,
      "loss": 0.2307,
      "step": 9947
    },
    {
      "epoch": 1.5572949279899813,
      "grad_norm": 0.9390753507614136,
      "learning_rate": 6.940371456500488e-05,
      "loss": 0.3621,
      "step": 9948
    },
    {
      "epoch": 1.5574514715090795,
      "grad_norm": 1.152475118637085,
      "learning_rate": 6.937927663734116e-05,
      "loss": 0.2944,
      "step": 9949
    },
    {
      "epoch": 1.557608015028178,
      "grad_norm": 0.7496026754379272,
      "learning_rate": 6.93548387096774e-05,
      "loss": 0.1876,
      "step": 9950
    },
    {
      "epoch": 1.5577645585472761,
      "grad_norm": 0.6168410778045654,
      "learning_rate": 6.933040078201368e-05,
      "loss": 0.2536,
      "step": 9951
    },
    {
      "epoch": 1.5579211020663744,
      "grad_norm": 1.393839716911316,
      "learning_rate": 6.930596285434995e-05,
      "loss": 0.3867,
      "step": 9952
    },
    {
      "epoch": 1.5580776455854728,
      "grad_norm": 0.946195125579834,
      "learning_rate": 6.928152492668621e-05,
      "loss": 0.1936,
      "step": 9953
    },
    {
      "epoch": 1.5582341891045712,
      "grad_norm": 1.4753316640853882,
      "learning_rate": 6.925708699902248e-05,
      "loss": 0.3014,
      "step": 9954
    },
    {
      "epoch": 1.5583907326236695,
      "grad_norm": 1.1015839576721191,
      "learning_rate": 6.923264907135874e-05,
      "loss": 0.4375,
      "step": 9955
    },
    {
      "epoch": 1.5585472761427677,
      "grad_norm": 1.402594804763794,
      "learning_rate": 6.920821114369501e-05,
      "loss": 0.701,
      "step": 9956
    },
    {
      "epoch": 1.5587038196618659,
      "grad_norm": 2.3376095294952393,
      "learning_rate": 6.918377321603127e-05,
      "loss": 0.3449,
      "step": 9957
    },
    {
      "epoch": 1.5588603631809643,
      "grad_norm": 0.5967291593551636,
      "learning_rate": 6.915933528836754e-05,
      "loss": 0.2373,
      "step": 9958
    },
    {
      "epoch": 1.5590169067000628,
      "grad_norm": 2.35986065864563,
      "learning_rate": 6.91348973607038e-05,
      "loss": 0.4976,
      "step": 9959
    },
    {
      "epoch": 1.559173450219161,
      "grad_norm": 1.496276617050171,
      "learning_rate": 6.911045943304007e-05,
      "loss": 0.4235,
      "step": 9960
    },
    {
      "epoch": 1.5593299937382592,
      "grad_norm": 3.5515880584716797,
      "learning_rate": 6.908602150537635e-05,
      "loss": 0.5994,
      "step": 9961
    },
    {
      "epoch": 1.5594865372573574,
      "grad_norm": 1.0382581949234009,
      "learning_rate": 6.90615835777126e-05,
      "loss": 0.3619,
      "step": 9962
    },
    {
      "epoch": 1.5596430807764559,
      "grad_norm": 1.326045274734497,
      "learning_rate": 6.903714565004888e-05,
      "loss": 0.7072,
      "step": 9963
    },
    {
      "epoch": 1.5597996242955543,
      "grad_norm": 3.173861265182495,
      "learning_rate": 6.901270772238514e-05,
      "loss": 0.9478,
      "step": 9964
    },
    {
      "epoch": 1.5599561678146525,
      "grad_norm": 1.280656099319458,
      "learning_rate": 6.898826979472139e-05,
      "loss": 0.5995,
      "step": 9965
    },
    {
      "epoch": 1.5601127113337507,
      "grad_norm": 1.5048918724060059,
      "learning_rate": 6.896383186705767e-05,
      "loss": 0.415,
      "step": 9966
    },
    {
      "epoch": 1.560269254852849,
      "grad_norm": 1.7540141344070435,
      "learning_rate": 6.893939393939393e-05,
      "loss": 0.7435,
      "step": 9967
    },
    {
      "epoch": 1.5604257983719474,
      "grad_norm": 1.4080053567886353,
      "learning_rate": 6.89149560117302e-05,
      "loss": 0.6369,
      "step": 9968
    },
    {
      "epoch": 1.5605823418910458,
      "grad_norm": 1.5890631675720215,
      "learning_rate": 6.889051808406646e-05,
      "loss": 0.4578,
      "step": 9969
    },
    {
      "epoch": 1.560738885410144,
      "grad_norm": 1.6475311517715454,
      "learning_rate": 6.886608015640273e-05,
      "loss": 0.6443,
      "step": 9970
    },
    {
      "epoch": 1.5608954289292423,
      "grad_norm": 1.769564151763916,
      "learning_rate": 6.8841642228739e-05,
      "loss": 0.5241,
      "step": 9971
    },
    {
      "epoch": 1.5610519724483405,
      "grad_norm": 2.2985165119171143,
      "learning_rate": 6.881720430107526e-05,
      "loss": 0.7585,
      "step": 9972
    },
    {
      "epoch": 1.561208515967439,
      "grad_norm": 1.865506649017334,
      "learning_rate": 6.879276637341154e-05,
      "loss": 0.8987,
      "step": 9973
    },
    {
      "epoch": 1.5613650594865374,
      "grad_norm": 2.5502965450286865,
      "learning_rate": 6.876832844574779e-05,
      "loss": 1.1079,
      "step": 9974
    },
    {
      "epoch": 1.5615216030056356,
      "grad_norm": 3.633234977722168,
      "learning_rate": 6.874389051808407e-05,
      "loss": 1.0968,
      "step": 9975
    },
    {
      "epoch": 1.5616781465247338,
      "grad_norm": 2.496279239654541,
      "learning_rate": 6.871945259042033e-05,
      "loss": 0.5893,
      "step": 9976
    },
    {
      "epoch": 1.561834690043832,
      "grad_norm": 2.3446414470672607,
      "learning_rate": 6.869501466275658e-05,
      "loss": 0.8225,
      "step": 9977
    },
    {
      "epoch": 1.5619912335629305,
      "grad_norm": 1.9929825067520142,
      "learning_rate": 6.867057673509286e-05,
      "loss": 0.819,
      "step": 9978
    },
    {
      "epoch": 1.562147777082029,
      "grad_norm": 5.4597086906433105,
      "learning_rate": 6.864613880742913e-05,
      "loss": 1.3806,
      "step": 9979
    },
    {
      "epoch": 1.5623043206011271,
      "grad_norm": 1.8970288038253784,
      "learning_rate": 6.862170087976539e-05,
      "loss": 0.9163,
      "step": 9980
    },
    {
      "epoch": 1.5624608641202253,
      "grad_norm": 2.849325656890869,
      "learning_rate": 6.859726295210166e-05,
      "loss": 0.9537,
      "step": 9981
    },
    {
      "epoch": 1.5626174076393238,
      "grad_norm": 1.5763049125671387,
      "learning_rate": 6.857282502443792e-05,
      "loss": 0.5998,
      "step": 9982
    },
    {
      "epoch": 1.562773951158422,
      "grad_norm": 2.1057538986206055,
      "learning_rate": 6.854838709677419e-05,
      "loss": 1.1581,
      "step": 9983
    },
    {
      "epoch": 1.5629304946775204,
      "grad_norm": 3.0501420497894287,
      "learning_rate": 6.852394916911045e-05,
      "loss": 1.0438,
      "step": 9984
    },
    {
      "epoch": 1.5630870381966186,
      "grad_norm": 1.4863300323486328,
      "learning_rate": 6.849951124144673e-05,
      "loss": 0.4397,
      "step": 9985
    },
    {
      "epoch": 1.5632435817157169,
      "grad_norm": 1.986098289489746,
      "learning_rate": 6.847507331378298e-05,
      "loss": 0.5282,
      "step": 9986
    },
    {
      "epoch": 1.5634001252348153,
      "grad_norm": 3.5367484092712402,
      "learning_rate": 6.845063538611926e-05,
      "loss": 0.4203,
      "step": 9987
    },
    {
      "epoch": 1.5635566687539137,
      "grad_norm": 3.17246413230896,
      "learning_rate": 6.842619745845552e-05,
      "loss": 0.8021,
      "step": 9988
    },
    {
      "epoch": 1.563713212273012,
      "grad_norm": 0.6066386103630066,
      "learning_rate": 6.840175953079177e-05,
      "loss": 0.1862,
      "step": 9989
    },
    {
      "epoch": 1.5638697557921102,
      "grad_norm": 0.8987689018249512,
      "learning_rate": 6.837732160312805e-05,
      "loss": 0.1316,
      "step": 9990
    },
    {
      "epoch": 1.5640262993112084,
      "grad_norm": 0.6625165343284607,
      "learning_rate": 6.835288367546432e-05,
      "loss": 0.2883,
      "step": 9991
    },
    {
      "epoch": 1.5641828428303068,
      "grad_norm": 0.5179951786994934,
      "learning_rate": 6.832844574780058e-05,
      "loss": 0.2005,
      "step": 9992
    },
    {
      "epoch": 1.5643393863494053,
      "grad_norm": 0.7913885116577148,
      "learning_rate": 6.830400782013685e-05,
      "loss": 0.3602,
      "step": 9993
    },
    {
      "epoch": 1.5644959298685035,
      "grad_norm": 0.5309231281280518,
      "learning_rate": 6.827956989247311e-05,
      "loss": 0.3084,
      "step": 9994
    },
    {
      "epoch": 1.5646524733876017,
      "grad_norm": 1.3907285928726196,
      "learning_rate": 6.825513196480938e-05,
      "loss": 0.4261,
      "step": 9995
    },
    {
      "epoch": 1.5648090169067,
      "grad_norm": 0.567585289478302,
      "learning_rate": 6.823069403714564e-05,
      "loss": 0.2173,
      "step": 9996
    },
    {
      "epoch": 1.5649655604257984,
      "grad_norm": 0.8944092392921448,
      "learning_rate": 6.820625610948192e-05,
      "loss": 0.2504,
      "step": 9997
    },
    {
      "epoch": 1.5651221039448968,
      "grad_norm": 0.9722375273704529,
      "learning_rate": 6.818181818181817e-05,
      "loss": 0.3839,
      "step": 9998
    },
    {
      "epoch": 1.565278647463995,
      "grad_norm": 0.5953250527381897,
      "learning_rate": 6.815738025415444e-05,
      "loss": 0.1622,
      "step": 9999
    },
    {
      "epoch": 1.5654351909830932,
      "grad_norm": 0.7407553195953369,
      "learning_rate": 6.813294232649071e-05,
      "loss": 0.1617,
      "step": 10000
    },
    {
      "epoch": 1.5654351909830932,
      "eval_loss": 0.5069952011108398,
      "eval_runtime": 205.9415,
      "eval_samples_per_second": 60.129,
      "eval_steps_per_second": 3.758,
      "eval_wer": 0.3151178358161379,
      "step": 10000
    },
    {
      "epoch": 1.5655917345021915,
      "grad_norm": 0.744092583656311,
      "learning_rate": 6.810850439882697e-05,
      "loss": 0.3372,
      "step": 10001
    },
    {
      "epoch": 1.56574827802129,
      "grad_norm": 2.9647679328918457,
      "learning_rate": 6.808406647116324e-05,
      "loss": 0.3241,
      "step": 10002
    },
    {
      "epoch": 1.5659048215403883,
      "grad_norm": 1.3153102397918701,
      "learning_rate": 6.805962854349951e-05,
      "loss": 0.2338,
      "step": 10003
    },
    {
      "epoch": 1.5660613650594866,
      "grad_norm": 1.2516891956329346,
      "learning_rate": 6.803519061583577e-05,
      "loss": 0.2711,
      "step": 10004
    },
    {
      "epoch": 1.5662179085785848,
      "grad_norm": 0.8349717259407043,
      "learning_rate": 6.801075268817204e-05,
      "loss": 0.2158,
      "step": 10005
    },
    {
      "epoch": 1.566374452097683,
      "grad_norm": 1.2376327514648438,
      "learning_rate": 6.79863147605083e-05,
      "loss": 0.5239,
      "step": 10006
    },
    {
      "epoch": 1.5665309956167814,
      "grad_norm": 1.2116237878799438,
      "learning_rate": 6.796187683284457e-05,
      "loss": 0.3544,
      "step": 10007
    },
    {
      "epoch": 1.5666875391358799,
      "grad_norm": 1.2293322086334229,
      "learning_rate": 6.793743890518083e-05,
      "loss": 0.478,
      "step": 10008
    },
    {
      "epoch": 1.566844082654978,
      "grad_norm": 1.6404024362564087,
      "learning_rate": 6.791300097751711e-05,
      "loss": 0.4453,
      "step": 10009
    },
    {
      "epoch": 1.5670006261740763,
      "grad_norm": 1.6513690948486328,
      "learning_rate": 6.788856304985336e-05,
      "loss": 0.4273,
      "step": 10010
    },
    {
      "epoch": 1.5671571696931748,
      "grad_norm": 1.3088352680206299,
      "learning_rate": 6.786412512218963e-05,
      "loss": 0.3235,
      "step": 10011
    },
    {
      "epoch": 1.567313713212273,
      "grad_norm": 1.2101068496704102,
      "learning_rate": 6.78396871945259e-05,
      "loss": 0.2902,
      "step": 10012
    },
    {
      "epoch": 1.5674702567313714,
      "grad_norm": 2.1694164276123047,
      "learning_rate": 6.781524926686216e-05,
      "loss": 0.6311,
      "step": 10013
    },
    {
      "epoch": 1.5676268002504696,
      "grad_norm": 1.8107420206069946,
      "learning_rate": 6.779081133919843e-05,
      "loss": 0.4426,
      "step": 10014
    },
    {
      "epoch": 1.5677833437695678,
      "grad_norm": 1.5845452547073364,
      "learning_rate": 6.77663734115347e-05,
      "loss": 0.3915,
      "step": 10015
    },
    {
      "epoch": 1.5679398872886663,
      "grad_norm": 4.756518840789795,
      "learning_rate": 6.774193548387096e-05,
      "loss": 0.7383,
      "step": 10016
    },
    {
      "epoch": 1.5680964308077645,
      "grad_norm": 2.210839033126831,
      "learning_rate": 6.771749755620723e-05,
      "loss": 0.7217,
      "step": 10017
    },
    {
      "epoch": 1.568252974326863,
      "grad_norm": 2.380188226699829,
      "learning_rate": 6.76930596285435e-05,
      "loss": 0.7003,
      "step": 10018
    },
    {
      "epoch": 1.5684095178459612,
      "grad_norm": 3.6436707973480225,
      "learning_rate": 6.766862170087976e-05,
      "loss": 1.0092,
      "step": 10019
    },
    {
      "epoch": 1.5685660613650594,
      "grad_norm": 1.1202057600021362,
      "learning_rate": 6.764418377321602e-05,
      "loss": 0.6157,
      "step": 10020
    },
    {
      "epoch": 1.5687226048841578,
      "grad_norm": 3.5373764038085938,
      "learning_rate": 6.761974584555229e-05,
      "loss": 1.0375,
      "step": 10021
    },
    {
      "epoch": 1.5688791484032563,
      "grad_norm": 2.415952205657959,
      "learning_rate": 6.759530791788855e-05,
      "loss": 0.6472,
      "step": 10022
    },
    {
      "epoch": 1.5690356919223545,
      "grad_norm": 2.597670316696167,
      "learning_rate": 6.757086999022482e-05,
      "loss": 0.8034,
      "step": 10023
    },
    {
      "epoch": 1.5691922354414527,
      "grad_norm": 1.0793561935424805,
      "learning_rate": 6.75464320625611e-05,
      "loss": 0.4365,
      "step": 10024
    },
    {
      "epoch": 1.569348778960551,
      "grad_norm": 4.046724796295166,
      "learning_rate": 6.752199413489735e-05,
      "loss": 0.825,
      "step": 10025
    },
    {
      "epoch": 1.5695053224796494,
      "grad_norm": 2.050435781478882,
      "learning_rate": 6.749755620723363e-05,
      "loss": 0.833,
      "step": 10026
    },
    {
      "epoch": 1.5696618659987478,
      "grad_norm": 2.141474723815918,
      "learning_rate": 6.747311827956989e-05,
      "loss": 0.5743,
      "step": 10027
    },
    {
      "epoch": 1.569818409517846,
      "grad_norm": 2.143913507461548,
      "learning_rate": 6.744868035190616e-05,
      "loss": 0.8483,
      "step": 10028
    },
    {
      "epoch": 1.5699749530369442,
      "grad_norm": 3.70343017578125,
      "learning_rate": 6.742424242424242e-05,
      "loss": 1.5895,
      "step": 10029
    },
    {
      "epoch": 1.5701314965560424,
      "grad_norm": 1.8668042421340942,
      "learning_rate": 6.739980449657869e-05,
      "loss": 0.7103,
      "step": 10030
    },
    {
      "epoch": 1.570288040075141,
      "grad_norm": 3.584521770477295,
      "learning_rate": 6.737536656891495e-05,
      "loss": 1.2124,
      "step": 10031
    },
    {
      "epoch": 1.5704445835942393,
      "grad_norm": 1.4927868843078613,
      "learning_rate": 6.735092864125121e-05,
      "loss": 0.7392,
      "step": 10032
    },
    {
      "epoch": 1.5706011271133375,
      "grad_norm": 2.586317300796509,
      "learning_rate": 6.732649071358748e-05,
      "loss": 1.2404,
      "step": 10033
    },
    {
      "epoch": 1.5707576706324358,
      "grad_norm": 1.917419672012329,
      "learning_rate": 6.730205278592374e-05,
      "loss": 1.218,
      "step": 10034
    },
    {
      "epoch": 1.570914214151534,
      "grad_norm": 2.0818874835968018,
      "learning_rate": 6.727761485826001e-05,
      "loss": 1.2963,
      "step": 10035
    },
    {
      "epoch": 1.5710707576706324,
      "grad_norm": 2.2003262042999268,
      "learning_rate": 6.725317693059629e-05,
      "loss": 0.6938,
      "step": 10036
    },
    {
      "epoch": 1.5712273011897309,
      "grad_norm": 2.3637146949768066,
      "learning_rate": 6.722873900293254e-05,
      "loss": 0.6664,
      "step": 10037
    },
    {
      "epoch": 1.571383844708829,
      "grad_norm": 2.3716087341308594,
      "learning_rate": 6.720430107526882e-05,
      "loss": 1.2521,
      "step": 10038
    },
    {
      "epoch": 1.5715403882279273,
      "grad_norm": 0.47229862213134766,
      "learning_rate": 6.717986314760507e-05,
      "loss": 0.1909,
      "step": 10039
    },
    {
      "epoch": 1.5716969317470255,
      "grad_norm": 0.536920428276062,
      "learning_rate": 6.715542521994135e-05,
      "loss": 0.2017,
      "step": 10040
    },
    {
      "epoch": 1.571853475266124,
      "grad_norm": 0.5747900009155273,
      "learning_rate": 6.713098729227761e-05,
      "loss": 0.1635,
      "step": 10041
    },
    {
      "epoch": 1.5720100187852224,
      "grad_norm": 0.4652441740036011,
      "learning_rate": 6.710654936461388e-05,
      "loss": 0.1387,
      "step": 10042
    },
    {
      "epoch": 1.5721665623043206,
      "grad_norm": 0.32238516211509705,
      "learning_rate": 6.708211143695014e-05,
      "loss": 0.1682,
      "step": 10043
    },
    {
      "epoch": 1.5723231058234188,
      "grad_norm": 0.5064279437065125,
      "learning_rate": 6.70576735092864e-05,
      "loss": 0.2486,
      "step": 10044
    },
    {
      "epoch": 1.5724796493425173,
      "grad_norm": 0.5283740758895874,
      "learning_rate": 6.703323558162267e-05,
      "loss": 0.2012,
      "step": 10045
    },
    {
      "epoch": 1.5726361928616155,
      "grad_norm": 0.55204176902771,
      "learning_rate": 6.700879765395894e-05,
      "loss": 0.2433,
      "step": 10046
    },
    {
      "epoch": 1.572792736380714,
      "grad_norm": 0.5996967554092407,
      "learning_rate": 6.69843597262952e-05,
      "loss": 0.2243,
      "step": 10047
    },
    {
      "epoch": 1.5729492798998121,
      "grad_norm": 0.5109484791755676,
      "learning_rate": 6.695992179863147e-05,
      "loss": 0.1715,
      "step": 10048
    },
    {
      "epoch": 1.5731058234189104,
      "grad_norm": 0.9667737483978271,
      "learning_rate": 6.693548387096773e-05,
      "loss": 0.2348,
      "step": 10049
    },
    {
      "epoch": 1.5732623669380088,
      "grad_norm": 1.0587562322616577,
      "learning_rate": 6.691104594330401e-05,
      "loss": 0.2966,
      "step": 10050
    },
    {
      "epoch": 1.573418910457107,
      "grad_norm": 1.2635822296142578,
      "learning_rate": 6.688660801564026e-05,
      "loss": 0.1986,
      "step": 10051
    },
    {
      "epoch": 1.5735754539762055,
      "grad_norm": 0.7891488075256348,
      "learning_rate": 6.686217008797654e-05,
      "loss": 0.2691,
      "step": 10052
    },
    {
      "epoch": 1.5737319974953037,
      "grad_norm": 0.8126229643821716,
      "learning_rate": 6.68377321603128e-05,
      "loss": 0.3324,
      "step": 10053
    },
    {
      "epoch": 1.573888541014402,
      "grad_norm": 1.0948699712753296,
      "learning_rate": 6.681329423264907e-05,
      "loss": 0.3169,
      "step": 10054
    },
    {
      "epoch": 1.5740450845335003,
      "grad_norm": 1.4431370496749878,
      "learning_rate": 6.678885630498533e-05,
      "loss": 0.5378,
      "step": 10055
    },
    {
      "epoch": 1.5742016280525988,
      "grad_norm": 2.1338889598846436,
      "learning_rate": 6.67644183773216e-05,
      "loss": 0.4985,
      "step": 10056
    },
    {
      "epoch": 1.574358171571697,
      "grad_norm": 1.3593388795852661,
      "learning_rate": 6.673998044965786e-05,
      "loss": 0.4749,
      "step": 10057
    },
    {
      "epoch": 1.5745147150907952,
      "grad_norm": 1.260514259338379,
      "learning_rate": 6.671554252199413e-05,
      "loss": 0.431,
      "step": 10058
    },
    {
      "epoch": 1.5746712586098934,
      "grad_norm": 1.087146282196045,
      "learning_rate": 6.669110459433039e-05,
      "loss": 0.4252,
      "step": 10059
    },
    {
      "epoch": 1.5748278021289919,
      "grad_norm": 0.7750207185745239,
      "learning_rate": 6.666666666666666e-05,
      "loss": 0.3251,
      "step": 10060
    },
    {
      "epoch": 1.5749843456480903,
      "grad_norm": 1.3115235567092896,
      "learning_rate": 6.664222873900292e-05,
      "loss": 0.3326,
      "step": 10061
    },
    {
      "epoch": 1.5751408891671885,
      "grad_norm": 1.4149150848388672,
      "learning_rate": 6.66177908113392e-05,
      "loss": 0.5962,
      "step": 10062
    },
    {
      "epoch": 1.5752974326862867,
      "grad_norm": 3.154900550842285,
      "learning_rate": 6.659335288367545e-05,
      "loss": 0.4921,
      "step": 10063
    },
    {
      "epoch": 1.575453976205385,
      "grad_norm": 1.6714290380477905,
      "learning_rate": 6.656891495601173e-05,
      "loss": 0.8742,
      "step": 10064
    },
    {
      "epoch": 1.5756105197244834,
      "grad_norm": 2.3452184200286865,
      "learning_rate": 6.6544477028348e-05,
      "loss": 0.767,
      "step": 10065
    },
    {
      "epoch": 1.5757670632435818,
      "grad_norm": 1.208871603012085,
      "learning_rate": 6.652003910068426e-05,
      "loss": 0.3151,
      "step": 10066
    },
    {
      "epoch": 1.57592360676268,
      "grad_norm": 3.192185878753662,
      "learning_rate": 6.649560117302052e-05,
      "loss": 0.8238,
      "step": 10067
    },
    {
      "epoch": 1.5760801502817783,
      "grad_norm": 3.4253957271575928,
      "learning_rate": 6.647116324535679e-05,
      "loss": 0.8813,
      "step": 10068
    },
    {
      "epoch": 1.5762366938008765,
      "grad_norm": 1.4279910326004028,
      "learning_rate": 6.644672531769305e-05,
      "loss": 0.6148,
      "step": 10069
    },
    {
      "epoch": 1.576393237319975,
      "grad_norm": 3.070469856262207,
      "learning_rate": 6.642228739002932e-05,
      "loss": 0.3901,
      "step": 10070
    },
    {
      "epoch": 1.5765497808390734,
      "grad_norm": 2.3599345684051514,
      "learning_rate": 6.639784946236558e-05,
      "loss": 0.4485,
      "step": 10071
    },
    {
      "epoch": 1.5767063243581716,
      "grad_norm": 1.7810369729995728,
      "learning_rate": 6.637341153470185e-05,
      "loss": 0.7455,
      "step": 10072
    },
    {
      "epoch": 1.5768628678772698,
      "grad_norm": 2.3326375484466553,
      "learning_rate": 6.634897360703811e-05,
      "loss": 0.5618,
      "step": 10073
    },
    {
      "epoch": 1.577019411396368,
      "grad_norm": 2.5533432960510254,
      "learning_rate": 6.632453567937439e-05,
      "loss": 0.9591,
      "step": 10074
    },
    {
      "epoch": 1.5771759549154665,
      "grad_norm": 3.2034530639648438,
      "learning_rate": 6.630009775171064e-05,
      "loss": 0.8351,
      "step": 10075
    },
    {
      "epoch": 1.577332498434565,
      "grad_norm": 2.8962016105651855,
      "learning_rate": 6.627565982404692e-05,
      "loss": 1.478,
      "step": 10076
    },
    {
      "epoch": 1.5774890419536631,
      "grad_norm": 2.253652572631836,
      "learning_rate": 6.625122189638319e-05,
      "loss": 1.1017,
      "step": 10077
    },
    {
      "epoch": 1.5776455854727613,
      "grad_norm": 4.773320198059082,
      "learning_rate": 6.622678396871945e-05,
      "loss": 1.4581,
      "step": 10078
    },
    {
      "epoch": 1.5778021289918598,
      "grad_norm": 4.403872013092041,
      "learning_rate": 6.620234604105571e-05,
      "loss": 0.7407,
      "step": 10079
    },
    {
      "epoch": 1.577958672510958,
      "grad_norm": 2.4393060207366943,
      "learning_rate": 6.617790811339198e-05,
      "loss": 0.9164,
      "step": 10080
    },
    {
      "epoch": 1.5781152160300564,
      "grad_norm": 3.306149959564209,
      "learning_rate": 6.615347018572824e-05,
      "loss": 1.3664,
      "step": 10081
    },
    {
      "epoch": 1.5782717595491547,
      "grad_norm": 2.284524917602539,
      "learning_rate": 6.612903225806451e-05,
      "loss": 0.9943,
      "step": 10082
    },
    {
      "epoch": 1.5784283030682529,
      "grad_norm": 6.304023265838623,
      "learning_rate": 6.610459433040077e-05,
      "loss": 0.3365,
      "step": 10083
    },
    {
      "epoch": 1.5785848465873513,
      "grad_norm": 1.4609630107879639,
      "learning_rate": 6.608015640273704e-05,
      "loss": 0.3204,
      "step": 10084
    },
    {
      "epoch": 1.5787413901064495,
      "grad_norm": 1.2958881855010986,
      "learning_rate": 6.60557184750733e-05,
      "loss": 0.5147,
      "step": 10085
    },
    {
      "epoch": 1.578897933625548,
      "grad_norm": 2.1635255813598633,
      "learning_rate": 6.603128054740958e-05,
      "loss": 0.5687,
      "step": 10086
    },
    {
      "epoch": 1.5790544771446462,
      "grad_norm": 3.585665225982666,
      "learning_rate": 6.600684261974583e-05,
      "loss": 1.3545,
      "step": 10087
    },
    {
      "epoch": 1.5792110206637444,
      "grad_norm": 4.3810834884643555,
      "learning_rate": 6.598240469208211e-05,
      "loss": 1.403,
      "step": 10088
    },
    {
      "epoch": 1.5793675641828429,
      "grad_norm": 1.2125346660614014,
      "learning_rate": 6.595796676441838e-05,
      "loss": 0.195,
      "step": 10089
    },
    {
      "epoch": 1.5795241077019413,
      "grad_norm": 1.367098331451416,
      "learning_rate": 6.593352883675464e-05,
      "loss": 0.303,
      "step": 10090
    },
    {
      "epoch": 1.5796806512210395,
      "grad_norm": 1.04794442653656,
      "learning_rate": 6.59090909090909e-05,
      "loss": 0.3611,
      "step": 10091
    },
    {
      "epoch": 1.5798371947401377,
      "grad_norm": 0.518174946308136,
      "learning_rate": 6.588465298142717e-05,
      "loss": 0.1909,
      "step": 10092
    },
    {
      "epoch": 1.579993738259236,
      "grad_norm": 0.5261414051055908,
      "learning_rate": 6.586021505376344e-05,
      "loss": 0.2269,
      "step": 10093
    },
    {
      "epoch": 1.5801502817783344,
      "grad_norm": 0.4940567910671234,
      "learning_rate": 6.58357771260997e-05,
      "loss": 0.1931,
      "step": 10094
    },
    {
      "epoch": 1.5803068252974328,
      "grad_norm": 0.8084196448326111,
      "learning_rate": 6.581133919843596e-05,
      "loss": 0.3157,
      "step": 10095
    },
    {
      "epoch": 1.580463368816531,
      "grad_norm": 0.5676218867301941,
      "learning_rate": 6.578690127077223e-05,
      "loss": 0.2145,
      "step": 10096
    },
    {
      "epoch": 1.5806199123356293,
      "grad_norm": 0.6355946660041809,
      "learning_rate": 6.57624633431085e-05,
      "loss": 0.1708,
      "step": 10097
    },
    {
      "epoch": 1.5807764558547275,
      "grad_norm": 0.8469789028167725,
      "learning_rate": 6.573802541544477e-05,
      "loss": 0.3097,
      "step": 10098
    },
    {
      "epoch": 1.580932999373826,
      "grad_norm": 2.256359338760376,
      "learning_rate": 6.571358748778102e-05,
      "loss": 0.2384,
      "step": 10099
    },
    {
      "epoch": 1.5810895428929244,
      "grad_norm": 0.8068323731422424,
      "learning_rate": 6.56891495601173e-05,
      "loss": 0.2843,
      "step": 10100
    },
    {
      "epoch": 1.5812460864120226,
      "grad_norm": 0.44950971007347107,
      "learning_rate": 6.566471163245357e-05,
      "loss": 0.1872,
      "step": 10101
    },
    {
      "epoch": 1.5814026299311208,
      "grad_norm": 1.0343793630599976,
      "learning_rate": 6.564027370478982e-05,
      "loss": 0.3186,
      "step": 10102
    },
    {
      "epoch": 1.581559173450219,
      "grad_norm": 2.219566822052002,
      "learning_rate": 6.56158357771261e-05,
      "loss": 0.5353,
      "step": 10103
    },
    {
      "epoch": 1.5817157169693175,
      "grad_norm": 1.2421504259109497,
      "learning_rate": 6.559139784946236e-05,
      "loss": 0.4237,
      "step": 10104
    },
    {
      "epoch": 1.581872260488416,
      "grad_norm": 0.9300140738487244,
      "learning_rate": 6.556695992179863e-05,
      "loss": 0.4712,
      "step": 10105
    },
    {
      "epoch": 1.5820288040075141,
      "grad_norm": 1.3351165056228638,
      "learning_rate": 6.554252199413489e-05,
      "loss": 0.4134,
      "step": 10106
    },
    {
      "epoch": 1.5821853475266123,
      "grad_norm": 1.0536439418792725,
      "learning_rate": 6.551808406647116e-05,
      "loss": 0.337,
      "step": 10107
    },
    {
      "epoch": 1.5823418910457105,
      "grad_norm": 0.8616892695426941,
      "learning_rate": 6.549364613880742e-05,
      "loss": 0.4153,
      "step": 10108
    },
    {
      "epoch": 1.582498434564809,
      "grad_norm": 1.845602035522461,
      "learning_rate": 6.546920821114369e-05,
      "loss": 0.3483,
      "step": 10109
    },
    {
      "epoch": 1.5826549780839074,
      "grad_norm": 4.453863143920898,
      "learning_rate": 6.544477028347996e-05,
      "loss": 0.9881,
      "step": 10110
    },
    {
      "epoch": 1.5828115216030056,
      "grad_norm": 1.3871952295303345,
      "learning_rate": 6.542033235581622e-05,
      "loss": 0.5901,
      "step": 10111
    },
    {
      "epoch": 1.5829680651221039,
      "grad_norm": 1.2875733375549316,
      "learning_rate": 6.53958944281525e-05,
      "loss": 0.5883,
      "step": 10112
    },
    {
      "epoch": 1.5831246086412023,
      "grad_norm": 1.413844108581543,
      "learning_rate": 6.537145650048876e-05,
      "loss": 0.3296,
      "step": 10113
    },
    {
      "epoch": 1.5832811521603005,
      "grad_norm": 1.2688345909118652,
      "learning_rate": 6.534701857282501e-05,
      "loss": 0.4407,
      "step": 10114
    },
    {
      "epoch": 1.583437695679399,
      "grad_norm": 1.8131461143493652,
      "learning_rate": 6.532258064516129e-05,
      "loss": 0.2713,
      "step": 10115
    },
    {
      "epoch": 1.5835942391984972,
      "grad_norm": 1.4916858673095703,
      "learning_rate": 6.529814271749755e-05,
      "loss": 0.5562,
      "step": 10116
    },
    {
      "epoch": 1.5837507827175954,
      "grad_norm": 2.313715696334839,
      "learning_rate": 6.527370478983382e-05,
      "loss": 0.5763,
      "step": 10117
    },
    {
      "epoch": 1.5839073262366938,
      "grad_norm": 1.6103403568267822,
      "learning_rate": 6.524926686217008e-05,
      "loss": 0.562,
      "step": 10118
    },
    {
      "epoch": 1.5840638697557923,
      "grad_norm": 0.9578543901443481,
      "learning_rate": 6.522482893450635e-05,
      "loss": 0.475,
      "step": 10119
    },
    {
      "epoch": 1.5842204132748905,
      "grad_norm": 1.8726333379745483,
      "learning_rate": 6.520039100684261e-05,
      "loss": 0.8251,
      "step": 10120
    },
    {
      "epoch": 1.5843769567939887,
      "grad_norm": 1.8902461528778076,
      "learning_rate": 6.517595307917888e-05,
      "loss": 1.235,
      "step": 10121
    },
    {
      "epoch": 1.584533500313087,
      "grad_norm": 3.1188371181488037,
      "learning_rate": 6.515151515151516e-05,
      "loss": 0.5179,
      "step": 10122
    },
    {
      "epoch": 1.5846900438321854,
      "grad_norm": 4.811651706695557,
      "learning_rate": 6.51270772238514e-05,
      "loss": 0.6082,
      "step": 10123
    },
    {
      "epoch": 1.5848465873512838,
      "grad_norm": 2.434835910797119,
      "learning_rate": 6.510263929618767e-05,
      "loss": 0.6315,
      "step": 10124
    },
    {
      "epoch": 1.585003130870382,
      "grad_norm": 1.2945551872253418,
      "learning_rate": 6.507820136852395e-05,
      "loss": 0.2202,
      "step": 10125
    },
    {
      "epoch": 1.5851596743894802,
      "grad_norm": 1.9833664894104004,
      "learning_rate": 6.50537634408602e-05,
      "loss": 0.7926,
      "step": 10126
    },
    {
      "epoch": 1.5853162179085785,
      "grad_norm": 3.4194037914276123,
      "learning_rate": 6.502932551319648e-05,
      "loss": 1.2268,
      "step": 10127
    },
    {
      "epoch": 1.585472761427677,
      "grad_norm": 2.057821035385132,
      "learning_rate": 6.500488758553274e-05,
      "loss": 1.0213,
      "step": 10128
    },
    {
      "epoch": 1.5856293049467753,
      "grad_norm": 7.126612186431885,
      "learning_rate": 6.498044965786901e-05,
      "loss": 1.0572,
      "step": 10129
    },
    {
      "epoch": 1.5857858484658736,
      "grad_norm": 3.2313385009765625,
      "learning_rate": 6.495601173020527e-05,
      "loss": 1.4866,
      "step": 10130
    },
    {
      "epoch": 1.5859423919849718,
      "grad_norm": 2.4898273944854736,
      "learning_rate": 6.493157380254154e-05,
      "loss": 0.851,
      "step": 10131
    },
    {
      "epoch": 1.58609893550407,
      "grad_norm": 2.793358087539673,
      "learning_rate": 6.49071358748778e-05,
      "loss": 1.1641,
      "step": 10132
    },
    {
      "epoch": 1.5862554790231684,
      "grad_norm": 2.09865665435791,
      "learning_rate": 6.488269794721407e-05,
      "loss": 1.1646,
      "step": 10133
    },
    {
      "epoch": 1.5864120225422669,
      "grad_norm": 1.7419707775115967,
      "learning_rate": 6.485826001955035e-05,
      "loss": 0.6815,
      "step": 10134
    },
    {
      "epoch": 1.586568566061365,
      "grad_norm": 1.2431154251098633,
      "learning_rate": 6.48338220918866e-05,
      "loss": 0.5228,
      "step": 10135
    },
    {
      "epoch": 1.5867251095804633,
      "grad_norm": 0.5367014408111572,
      "learning_rate": 6.480938416422286e-05,
      "loss": 0.1294,
      "step": 10136
    },
    {
      "epoch": 1.5868816530995615,
      "grad_norm": 1.9725390672683716,
      "learning_rate": 6.478494623655914e-05,
      "loss": 0.6302,
      "step": 10137
    },
    {
      "epoch": 1.58703819661866,
      "grad_norm": 2.6568782329559326,
      "learning_rate": 6.476050830889539e-05,
      "loss": 0.6514,
      "step": 10138
    },
    {
      "epoch": 1.5871947401377584,
      "grad_norm": 0.5387908220291138,
      "learning_rate": 6.473607038123167e-05,
      "loss": 0.2569,
      "step": 10139
    },
    {
      "epoch": 1.5873512836568566,
      "grad_norm": 0.6417441964149475,
      "learning_rate": 6.471163245356794e-05,
      "loss": 0.2113,
      "step": 10140
    },
    {
      "epoch": 1.5875078271759548,
      "grad_norm": 0.7953237295150757,
      "learning_rate": 6.46871945259042e-05,
      "loss": 0.2751,
      "step": 10141
    },
    {
      "epoch": 1.587664370695053,
      "grad_norm": 0.5870561003684998,
      "learning_rate": 6.466275659824046e-05,
      "loss": 0.274,
      "step": 10142
    },
    {
      "epoch": 1.5878209142141515,
      "grad_norm": 0.416892409324646,
      "learning_rate": 6.463831867057673e-05,
      "loss": 0.2782,
      "step": 10143
    },
    {
      "epoch": 1.58797745773325,
      "grad_norm": 1.2958848476409912,
      "learning_rate": 6.4613880742913e-05,
      "loss": 0.2186,
      "step": 10144
    },
    {
      "epoch": 1.5881340012523482,
      "grad_norm": 0.7092198133468628,
      "learning_rate": 6.458944281524926e-05,
      "loss": 0.3302,
      "step": 10145
    },
    {
      "epoch": 1.5882905447714464,
      "grad_norm": 1.2143514156341553,
      "learning_rate": 6.456500488758554e-05,
      "loss": 0.4434,
      "step": 10146
    },
    {
      "epoch": 1.5884470882905448,
      "grad_norm": 0.6722570657730103,
      "learning_rate": 6.454056695992179e-05,
      "loss": 0.3993,
      "step": 10147
    },
    {
      "epoch": 1.588603631809643,
      "grad_norm": 0.7443752884864807,
      "learning_rate": 6.451612903225805e-05,
      "loss": 0.4719,
      "step": 10148
    },
    {
      "epoch": 1.5887601753287415,
      "grad_norm": 1.1258270740509033,
      "learning_rate": 6.449169110459433e-05,
      "loss": 0.2445,
      "step": 10149
    },
    {
      "epoch": 1.5889167188478397,
      "grad_norm": 0.5991902351379395,
      "learning_rate": 6.446725317693058e-05,
      "loss": 0.3133,
      "step": 10150
    },
    {
      "epoch": 1.589073262366938,
      "grad_norm": 1.0533815622329712,
      "learning_rate": 6.444281524926686e-05,
      "loss": 0.2713,
      "step": 10151
    },
    {
      "epoch": 1.5892298058860364,
      "grad_norm": 1.2451895475387573,
      "learning_rate": 6.441837732160313e-05,
      "loss": 0.2822,
      "step": 10152
    },
    {
      "epoch": 1.5893863494051348,
      "grad_norm": 1.2398431301116943,
      "learning_rate": 6.439393939393939e-05,
      "loss": 0.4126,
      "step": 10153
    },
    {
      "epoch": 1.589542892924233,
      "grad_norm": 7.360801696777344,
      "learning_rate": 6.436950146627566e-05,
      "loss": 0.4618,
      "step": 10154
    },
    {
      "epoch": 1.5896994364433312,
      "grad_norm": 1.0283269882202148,
      "learning_rate": 6.434506353861192e-05,
      "loss": 0.5885,
      "step": 10155
    },
    {
      "epoch": 1.5898559799624294,
      "grad_norm": 2.743352174758911,
      "learning_rate": 6.432062561094819e-05,
      "loss": 0.4484,
      "step": 10156
    },
    {
      "epoch": 1.5900125234815279,
      "grad_norm": 0.7910766005516052,
      "learning_rate": 6.429618768328445e-05,
      "loss": 0.3717,
      "step": 10157
    },
    {
      "epoch": 1.5901690670006263,
      "grad_norm": 0.8904502987861633,
      "learning_rate": 6.427174975562072e-05,
      "loss": 0.372,
      "step": 10158
    },
    {
      "epoch": 1.5903256105197245,
      "grad_norm": 1.1738675832748413,
      "learning_rate": 6.424731182795698e-05,
      "loss": 0.3387,
      "step": 10159
    },
    {
      "epoch": 1.5904821540388228,
      "grad_norm": 1.5055691003799438,
      "learning_rate": 6.422287390029324e-05,
      "loss": 0.4907,
      "step": 10160
    },
    {
      "epoch": 1.590638697557921,
      "grad_norm": 1.081289291381836,
      "learning_rate": 6.419843597262952e-05,
      "loss": 0.3105,
      "step": 10161
    },
    {
      "epoch": 1.5907952410770194,
      "grad_norm": 1.3567743301391602,
      "learning_rate": 6.417399804496577e-05,
      "loss": 0.3565,
      "step": 10162
    },
    {
      "epoch": 1.5909517845961179,
      "grad_norm": 1.7186039686203003,
      "learning_rate": 6.414956011730205e-05,
      "loss": 0.7406,
      "step": 10163
    },
    {
      "epoch": 1.591108328115216,
      "grad_norm": 1.8988852500915527,
      "learning_rate": 6.412512218963832e-05,
      "loss": 0.3946,
      "step": 10164
    },
    {
      "epoch": 1.5912648716343143,
      "grad_norm": 2.2571020126342773,
      "learning_rate": 6.410068426197458e-05,
      "loss": 0.5579,
      "step": 10165
    },
    {
      "epoch": 1.5914214151534125,
      "grad_norm": 1.9459624290466309,
      "learning_rate": 6.407624633431085e-05,
      "loss": 0.7901,
      "step": 10166
    },
    {
      "epoch": 1.591577958672511,
      "grad_norm": 2.4512932300567627,
      "learning_rate": 6.405180840664711e-05,
      "loss": 1.0994,
      "step": 10167
    },
    {
      "epoch": 1.5917345021916094,
      "grad_norm": 1.6560828685760498,
      "learning_rate": 6.402737047898338e-05,
      "loss": 0.4887,
      "step": 10168
    },
    {
      "epoch": 1.5918910457107076,
      "grad_norm": 1.7962381839752197,
      "learning_rate": 6.400293255131964e-05,
      "loss": 0.649,
      "step": 10169
    },
    {
      "epoch": 1.5920475892298058,
      "grad_norm": 2.8789329528808594,
      "learning_rate": 6.39784946236559e-05,
      "loss": 0.7007,
      "step": 10170
    },
    {
      "epoch": 1.592204132748904,
      "grad_norm": 2.5336639881134033,
      "learning_rate": 6.395405669599217e-05,
      "loss": 1.0201,
      "step": 10171
    },
    {
      "epoch": 1.5923606762680025,
      "grad_norm": 2.5551042556762695,
      "learning_rate": 6.392961876832844e-05,
      "loss": 0.7936,
      "step": 10172
    },
    {
      "epoch": 1.592517219787101,
      "grad_norm": 1.407386064529419,
      "learning_rate": 6.39051808406647e-05,
      "loss": 0.6436,
      "step": 10173
    },
    {
      "epoch": 1.5926737633061991,
      "grad_norm": 2.690272331237793,
      "learning_rate": 6.388074291300097e-05,
      "loss": 0.7574,
      "step": 10174
    },
    {
      "epoch": 1.5928303068252974,
      "grad_norm": 1.8954501152038574,
      "learning_rate": 6.385630498533724e-05,
      "loss": 0.7914,
      "step": 10175
    },
    {
      "epoch": 1.5929868503443956,
      "grad_norm": 2.0154988765716553,
      "learning_rate": 6.38318670576735e-05,
      "loss": 0.6738,
      "step": 10176
    },
    {
      "epoch": 1.593143393863494,
      "grad_norm": 2.260967969894409,
      "learning_rate": 6.380742913000977e-05,
      "loss": 0.5325,
      "step": 10177
    },
    {
      "epoch": 1.5932999373825925,
      "grad_norm": 3.078644037246704,
      "learning_rate": 6.378299120234604e-05,
      "loss": 0.6965,
      "step": 10178
    },
    {
      "epoch": 1.5934564809016907,
      "grad_norm": 2.3561620712280273,
      "learning_rate": 6.37585532746823e-05,
      "loss": 0.9808,
      "step": 10179
    },
    {
      "epoch": 1.593613024420789,
      "grad_norm": 3.07482647895813,
      "learning_rate": 6.373411534701857e-05,
      "loss": 0.9701,
      "step": 10180
    },
    {
      "epoch": 1.5937695679398873,
      "grad_norm": 5.131535530090332,
      "learning_rate": 6.370967741935483e-05,
      "loss": 0.8996,
      "step": 10181
    },
    {
      "epoch": 1.5939261114589856,
      "grad_norm": 2.536735773086548,
      "learning_rate": 6.36852394916911e-05,
      "loss": 1.0249,
      "step": 10182
    },
    {
      "epoch": 1.594082654978084,
      "grad_norm": 1.7344963550567627,
      "learning_rate": 6.366080156402736e-05,
      "loss": 0.9431,
      "step": 10183
    },
    {
      "epoch": 1.5942391984971822,
      "grad_norm": 1.4525513648986816,
      "learning_rate": 6.363636363636363e-05,
      "loss": 0.3811,
      "step": 10184
    },
    {
      "epoch": 1.5943957420162804,
      "grad_norm": 11.553523063659668,
      "learning_rate": 6.361192570869989e-05,
      "loss": 0.7413,
      "step": 10185
    },
    {
      "epoch": 1.5945522855353789,
      "grad_norm": 2.428168535232544,
      "learning_rate": 6.358748778103616e-05,
      "loss": 0.6394,
      "step": 10186
    },
    {
      "epoch": 1.5947088290544773,
      "grad_norm": 5.859124183654785,
      "learning_rate": 6.356304985337244e-05,
      "loss": 0.5436,
      "step": 10187
    },
    {
      "epoch": 1.5948653725735755,
      "grad_norm": 2.319378614425659,
      "learning_rate": 6.353861192570869e-05,
      "loss": 1.0681,
      "step": 10188
    },
    {
      "epoch": 1.5950219160926737,
      "grad_norm": 0.3972637355327606,
      "learning_rate": 6.351417399804496e-05,
      "loss": 0.2068,
      "step": 10189
    },
    {
      "epoch": 1.595178459611772,
      "grad_norm": 0.3770846128463745,
      "learning_rate": 6.348973607038123e-05,
      "loss": 0.1627,
      "step": 10190
    },
    {
      "epoch": 1.5953350031308704,
      "grad_norm": 1.2007943391799927,
      "learning_rate": 6.34652981427175e-05,
      "loss": 0.254,
      "step": 10191
    },
    {
      "epoch": 1.5954915466499688,
      "grad_norm": 2.273597002029419,
      "learning_rate": 6.344086021505376e-05,
      "loss": 0.3272,
      "step": 10192
    },
    {
      "epoch": 1.595648090169067,
      "grad_norm": 0.8018514513969421,
      "learning_rate": 6.341642228739002e-05,
      "loss": 0.2878,
      "step": 10193
    },
    {
      "epoch": 1.5958046336881653,
      "grad_norm": 0.5968719124794006,
      "learning_rate": 6.339198435972629e-05,
      "loss": 0.2637,
      "step": 10194
    },
    {
      "epoch": 1.5959611772072635,
      "grad_norm": 0.6231215000152588,
      "learning_rate": 6.336754643206255e-05,
      "loss": 0.2629,
      "step": 10195
    },
    {
      "epoch": 1.596117720726362,
      "grad_norm": 0.4936104416847229,
      "learning_rate": 6.334310850439882e-05,
      "loss": 0.207,
      "step": 10196
    },
    {
      "epoch": 1.5962742642454604,
      "grad_norm": 0.6326175928115845,
      "learning_rate": 6.331867057673508e-05,
      "loss": 0.4249,
      "step": 10197
    },
    {
      "epoch": 1.5964308077645586,
      "grad_norm": 0.5844799876213074,
      "learning_rate": 6.329423264907135e-05,
      "loss": 0.2245,
      "step": 10198
    },
    {
      "epoch": 1.5965873512836568,
      "grad_norm": 2.547029495239258,
      "learning_rate": 6.326979472140763e-05,
      "loss": 0.4301,
      "step": 10199
    },
    {
      "epoch": 1.596743894802755,
      "grad_norm": 1.4274916648864746,
      "learning_rate": 6.324535679374388e-05,
      "loss": 0.4086,
      "step": 10200
    },
    {
      "epoch": 1.5969004383218535,
      "grad_norm": 1.0012903213500977,
      "learning_rate": 6.322091886608016e-05,
      "loss": 0.2683,
      "step": 10201
    },
    {
      "epoch": 1.597056981840952,
      "grad_norm": 2.6661782264709473,
      "learning_rate": 6.319648093841642e-05,
      "loss": 0.5077,
      "step": 10202
    },
    {
      "epoch": 1.5972135253600501,
      "grad_norm": 0.7351189851760864,
      "learning_rate": 6.317204301075269e-05,
      "loss": 0.3245,
      "step": 10203
    },
    {
      "epoch": 1.5973700688791483,
      "grad_norm": 0.7789361476898193,
      "learning_rate": 6.314760508308895e-05,
      "loss": 0.3436,
      "step": 10204
    },
    {
      "epoch": 1.5975266123982466,
      "grad_norm": 1.3661255836486816,
      "learning_rate": 6.312316715542522e-05,
      "loss": 0.381,
      "step": 10205
    },
    {
      "epoch": 1.597683155917345,
      "grad_norm": 0.8813636898994446,
      "learning_rate": 6.309872922776148e-05,
      "loss": 0.3987,
      "step": 10206
    },
    {
      "epoch": 1.5978396994364434,
      "grad_norm": 0.6111112236976624,
      "learning_rate": 6.307429130009774e-05,
      "loss": 0.1548,
      "step": 10207
    },
    {
      "epoch": 1.5979962429555417,
      "grad_norm": 1.8574801683425903,
      "learning_rate": 6.304985337243401e-05,
      "loss": 0.5964,
      "step": 10208
    },
    {
      "epoch": 1.5981527864746399,
      "grad_norm": 3.280104160308838,
      "learning_rate": 6.302541544477027e-05,
      "loss": 0.4009,
      "step": 10209
    },
    {
      "epoch": 1.5983093299937383,
      "grad_norm": 4.668222427368164,
      "learning_rate": 6.300097751710654e-05,
      "loss": 0.4519,
      "step": 10210
    },
    {
      "epoch": 1.5984658735128365,
      "grad_norm": 2.0994975566864014,
      "learning_rate": 6.297653958944282e-05,
      "loss": 0.5817,
      "step": 10211
    },
    {
      "epoch": 1.598622417031935,
      "grad_norm": 2.334909200668335,
      "learning_rate": 6.295210166177907e-05,
      "loss": 0.5803,
      "step": 10212
    },
    {
      "epoch": 1.5987789605510332,
      "grad_norm": 1.9847476482391357,
      "learning_rate": 6.292766373411535e-05,
      "loss": 0.3882,
      "step": 10213
    },
    {
      "epoch": 1.5989355040701314,
      "grad_norm": 1.3854622840881348,
      "learning_rate": 6.290322580645161e-05,
      "loss": 0.477,
      "step": 10214
    },
    {
      "epoch": 1.5990920475892298,
      "grad_norm": 1.641670823097229,
      "learning_rate": 6.287878787878788e-05,
      "loss": 0.3697,
      "step": 10215
    },
    {
      "epoch": 1.599248591108328,
      "grad_norm": 2.7601287364959717,
      "learning_rate": 6.285434995112414e-05,
      "loss": 0.438,
      "step": 10216
    },
    {
      "epoch": 1.5994051346274265,
      "grad_norm": 2.510101318359375,
      "learning_rate": 6.28299120234604e-05,
      "loss": 0.8074,
      "step": 10217
    },
    {
      "epoch": 1.5995616781465247,
      "grad_norm": 2.204482316970825,
      "learning_rate": 6.280547409579667e-05,
      "loss": 0.5739,
      "step": 10218
    },
    {
      "epoch": 1.599718221665623,
      "grad_norm": 2.250614643096924,
      "learning_rate": 6.278103616813294e-05,
      "loss": 0.9905,
      "step": 10219
    },
    {
      "epoch": 1.5998747651847214,
      "grad_norm": 2.100642204284668,
      "learning_rate": 6.27565982404692e-05,
      "loss": 0.7706,
      "step": 10220
    },
    {
      "epoch": 1.6000313087038198,
      "grad_norm": 2.480836868286133,
      "learning_rate": 6.273216031280547e-05,
      "loss": 0.9495,
      "step": 10221
    },
    {
      "epoch": 1.600187852222918,
      "grad_norm": 2.637479305267334,
      "learning_rate": 6.270772238514173e-05,
      "loss": 0.5965,
      "step": 10222
    },
    {
      "epoch": 1.6003443957420163,
      "grad_norm": 5.914449691772461,
      "learning_rate": 6.268328445747801e-05,
      "loss": 1.0819,
      "step": 10223
    },
    {
      "epoch": 1.6005009392611145,
      "grad_norm": 3.239971876144409,
      "learning_rate": 6.265884652981426e-05,
      "loss": 0.9483,
      "step": 10224
    },
    {
      "epoch": 1.600657482780213,
      "grad_norm": 2.788769483566284,
      "learning_rate": 6.263440860215054e-05,
      "loss": 0.8619,
      "step": 10225
    },
    {
      "epoch": 1.6008140262993114,
      "grad_norm": 3.059751033782959,
      "learning_rate": 6.26099706744868e-05,
      "loss": 0.8397,
      "step": 10226
    },
    {
      "epoch": 1.6009705698184096,
      "grad_norm": 3.003505229949951,
      "learning_rate": 6.258553274682307e-05,
      "loss": 0.8119,
      "step": 10227
    },
    {
      "epoch": 1.6011271133375078,
      "grad_norm": 2.3189423084259033,
      "learning_rate": 6.256109481915933e-05,
      "loss": 0.9308,
      "step": 10228
    },
    {
      "epoch": 1.601283656856606,
      "grad_norm": 5.253555774688721,
      "learning_rate": 6.25366568914956e-05,
      "loss": 0.7128,
      "step": 10229
    },
    {
      "epoch": 1.6014402003757044,
      "grad_norm": 4.00673770904541,
      "learning_rate": 6.251221896383186e-05,
      "loss": 1.3895,
      "step": 10230
    },
    {
      "epoch": 1.6015967438948029,
      "grad_norm": 4.583085060119629,
      "learning_rate": 6.248778103616813e-05,
      "loss": 0.7943,
      "step": 10231
    },
    {
      "epoch": 1.601753287413901,
      "grad_norm": 0.6882913112640381,
      "learning_rate": 6.246334310850439e-05,
      "loss": 0.3133,
      "step": 10232
    },
    {
      "epoch": 1.6019098309329993,
      "grad_norm": 1.9263126850128174,
      "learning_rate": 6.243890518084066e-05,
      "loss": 0.4667,
      "step": 10233
    },
    {
      "epoch": 1.6020663744520975,
      "grad_norm": 13.986974716186523,
      "learning_rate": 6.241446725317692e-05,
      "loss": 0.7877,
      "step": 10234
    },
    {
      "epoch": 1.602222917971196,
      "grad_norm": 0.6357542872428894,
      "learning_rate": 6.23900293255132e-05,
      "loss": 0.2021,
      "step": 10235
    },
    {
      "epoch": 1.6023794614902944,
      "grad_norm": 6.546449661254883,
      "learning_rate": 6.236559139784945e-05,
      "loss": 0.728,
      "step": 10236
    },
    {
      "epoch": 1.6025360050093926,
      "grad_norm": 1.615708351135254,
      "learning_rate": 6.234115347018573e-05,
      "loss": 0.843,
      "step": 10237
    },
    {
      "epoch": 1.6026925485284909,
      "grad_norm": 2.3978271484375,
      "learning_rate": 6.2316715542522e-05,
      "loss": 1.1529,
      "step": 10238
    },
    {
      "epoch": 1.602849092047589,
      "grad_norm": 0.8120617866516113,
      "learning_rate": 6.229227761485825e-05,
      "loss": 0.2389,
      "step": 10239
    },
    {
      "epoch": 1.6030056355666875,
      "grad_norm": 0.506904125213623,
      "learning_rate": 6.226783968719452e-05,
      "loss": 0.2102,
      "step": 10240
    },
    {
      "epoch": 1.603162179085786,
      "grad_norm": 0.43618541955947876,
      "learning_rate": 6.224340175953079e-05,
      "loss": 0.1446,
      "step": 10241
    },
    {
      "epoch": 1.6033187226048842,
      "grad_norm": 0.7896573543548584,
      "learning_rate": 6.221896383186705e-05,
      "loss": 0.3323,
      "step": 10242
    },
    {
      "epoch": 1.6034752661239824,
      "grad_norm": 0.5736374855041504,
      "learning_rate": 6.219452590420332e-05,
      "loss": 0.2047,
      "step": 10243
    },
    {
      "epoch": 1.6036318096430808,
      "grad_norm": 0.6807544231414795,
      "learning_rate": 6.217008797653958e-05,
      "loss": 0.3968,
      "step": 10244
    },
    {
      "epoch": 1.603788353162179,
      "grad_norm": 0.7900474071502686,
      "learning_rate": 6.214565004887585e-05,
      "loss": 0.3243,
      "step": 10245
    },
    {
      "epoch": 1.6039448966812775,
      "grad_norm": 0.6970692873001099,
      "learning_rate": 6.212121212121211e-05,
      "loss": 0.1766,
      "step": 10246
    },
    {
      "epoch": 1.6041014402003757,
      "grad_norm": 2.053403854370117,
      "learning_rate": 6.209677419354839e-05,
      "loss": 0.2549,
      "step": 10247
    },
    {
      "epoch": 1.604257983719474,
      "grad_norm": 2.547222375869751,
      "learning_rate": 6.207233626588464e-05,
      "loss": 0.3394,
      "step": 10248
    },
    {
      "epoch": 1.6044145272385724,
      "grad_norm": 0.9715545773506165,
      "learning_rate": 6.204789833822092e-05,
      "loss": 0.2039,
      "step": 10249
    },
    {
      "epoch": 1.6045710707576706,
      "grad_norm": 0.9468798637390137,
      "learning_rate": 6.202346041055719e-05,
      "loss": 0.3098,
      "step": 10250
    },
    {
      "epoch": 1.604727614276769,
      "grad_norm": 0.8663145303726196,
      "learning_rate": 6.199902248289344e-05,
      "loss": 0.1379,
      "step": 10251
    },
    {
      "epoch": 1.6048841577958672,
      "grad_norm": 0.8937069177627563,
      "learning_rate": 6.197458455522971e-05,
      "loss": 0.3302,
      "step": 10252
    },
    {
      "epoch": 1.6050407013149655,
      "grad_norm": 1.2698081731796265,
      "learning_rate": 6.195014662756598e-05,
      "loss": 0.3953,
      "step": 10253
    },
    {
      "epoch": 1.605197244834064,
      "grad_norm": 0.827766478061676,
      "learning_rate": 6.192570869990224e-05,
      "loss": 0.2489,
      "step": 10254
    },
    {
      "epoch": 1.6053537883531623,
      "grad_norm": 3.880368947982788,
      "learning_rate": 6.190127077223851e-05,
      "loss": 0.6099,
      "step": 10255
    },
    {
      "epoch": 1.6055103318722606,
      "grad_norm": 2.3780124187469482,
      "learning_rate": 6.187683284457477e-05,
      "loss": 0.5125,
      "step": 10256
    },
    {
      "epoch": 1.6056668753913588,
      "grad_norm": 2.895963191986084,
      "learning_rate": 6.185239491691104e-05,
      "loss": 0.2951,
      "step": 10257
    },
    {
      "epoch": 1.605823418910457,
      "grad_norm": 0.9345550537109375,
      "learning_rate": 6.18279569892473e-05,
      "loss": 0.397,
      "step": 10258
    },
    {
      "epoch": 1.6059799624295554,
      "grad_norm": 1.1102604866027832,
      "learning_rate": 6.180351906158358e-05,
      "loss": 0.51,
      "step": 10259
    },
    {
      "epoch": 1.6061365059486539,
      "grad_norm": 0.9499253630638123,
      "learning_rate": 6.177908113391983e-05,
      "loss": 0.2253,
      "step": 10260
    },
    {
      "epoch": 1.606293049467752,
      "grad_norm": 0.8864089250564575,
      "learning_rate": 6.17546432062561e-05,
      "loss": 0.2358,
      "step": 10261
    },
    {
      "epoch": 1.6064495929868503,
      "grad_norm": 1.9460018873214722,
      "learning_rate": 6.173020527859238e-05,
      "loss": 0.5664,
      "step": 10262
    },
    {
      "epoch": 1.6066061365059485,
      "grad_norm": 1.625748634338379,
      "learning_rate": 6.170576735092863e-05,
      "loss": 0.4159,
      "step": 10263
    },
    {
      "epoch": 1.606762680025047,
      "grad_norm": 2.510746479034424,
      "learning_rate": 6.16813294232649e-05,
      "loss": 0.7013,
      "step": 10264
    },
    {
      "epoch": 1.6069192235441454,
      "grad_norm": 3.4259462356567383,
      "learning_rate": 6.165689149560117e-05,
      "loss": 0.326,
      "step": 10265
    },
    {
      "epoch": 1.6070757670632436,
      "grad_norm": 1.4860423803329468,
      "learning_rate": 6.163245356793744e-05,
      "loss": 0.3353,
      "step": 10266
    },
    {
      "epoch": 1.6072323105823418,
      "grad_norm": 1.9358034133911133,
      "learning_rate": 6.16080156402737e-05,
      "loss": 0.6318,
      "step": 10267
    },
    {
      "epoch": 1.60738885410144,
      "grad_norm": 2.1862995624542236,
      "learning_rate": 6.158357771260997e-05,
      "loss": 0.5148,
      "step": 10268
    },
    {
      "epoch": 1.6075453976205385,
      "grad_norm": 1.741811990737915,
      "learning_rate": 6.155913978494623e-05,
      "loss": 0.8787,
      "step": 10269
    },
    {
      "epoch": 1.607701941139637,
      "grad_norm": 1.7040098905563354,
      "learning_rate": 6.15347018572825e-05,
      "loss": 0.4273,
      "step": 10270
    },
    {
      "epoch": 1.6078584846587352,
      "grad_norm": 1.6641523838043213,
      "learning_rate": 6.151026392961877e-05,
      "loss": 0.5667,
      "step": 10271
    },
    {
      "epoch": 1.6080150281778334,
      "grad_norm": 1.784555435180664,
      "learning_rate": 6.148582600195502e-05,
      "loss": 0.69,
      "step": 10272
    },
    {
      "epoch": 1.6081715716969316,
      "grad_norm": 4.743298053741455,
      "learning_rate": 6.146138807429129e-05,
      "loss": 0.6946,
      "step": 10273
    },
    {
      "epoch": 1.60832811521603,
      "grad_norm": 5.469465732574463,
      "learning_rate": 6.143695014662757e-05,
      "loss": 1.1488,
      "step": 10274
    },
    {
      "epoch": 1.6084846587351285,
      "grad_norm": 4.396662712097168,
      "learning_rate": 6.141251221896382e-05,
      "loss": 1.2612,
      "step": 10275
    },
    {
      "epoch": 1.6086412022542267,
      "grad_norm": 2.4530532360076904,
      "learning_rate": 6.13880742913001e-05,
      "loss": 0.768,
      "step": 10276
    },
    {
      "epoch": 1.608797745773325,
      "grad_norm": NaN,
      "learning_rate": 6.13880742913001e-05,
      "loss": 0.0,
      "step": 10277
    },
    {
      "epoch": 1.6089542892924233,
      "grad_norm": 3.0336153507232666,
      "learning_rate": 6.136363636363636e-05,
      "loss": 0.9859,
      "step": 10278
    },
    {
      "epoch": 1.6091108328115216,
      "grad_norm": 2.3858444690704346,
      "learning_rate": 6.133919843597263e-05,
      "loss": 0.8229,
      "step": 10279
    },
    {
      "epoch": 1.60926737633062,
      "grad_norm": 3.8104684352874756,
      "learning_rate": 6.131476050830889e-05,
      "loss": 1.1625,
      "step": 10280
    },
    {
      "epoch": 1.6094239198497182,
      "grad_norm": 1.7261332273483276,
      "learning_rate": 6.129032258064516e-05,
      "loss": 0.8723,
      "step": 10281
    },
    {
      "epoch": 1.6095804633688164,
      "grad_norm": 1.7524932622909546,
      "learning_rate": 6.126588465298142e-05,
      "loss": 0.995,
      "step": 10282
    },
    {
      "epoch": 1.6097370068879149,
      "grad_norm": 2.208996057510376,
      "learning_rate": 6.124144672531769e-05,
      "loss": 1.0345,
      "step": 10283
    },
    {
      "epoch": 1.609893550407013,
      "grad_norm": 9.19688892364502,
      "learning_rate": 6.121700879765395e-05,
      "loss": 0.5331,
      "step": 10284
    },
    {
      "epoch": 1.6100500939261115,
      "grad_norm": 4.053257465362549,
      "learning_rate": 6.119257086999022e-05,
      "loss": 0.705,
      "step": 10285
    },
    {
      "epoch": 1.6102066374452098,
      "grad_norm": 2.6395530700683594,
      "learning_rate": 6.116813294232648e-05,
      "loss": 0.8349,
      "step": 10286
    },
    {
      "epoch": 1.610363180964308,
      "grad_norm": 5.501692771911621,
      "learning_rate": 6.114369501466276e-05,
      "loss": 0.9976,
      "step": 10287
    },
    {
      "epoch": 1.6105197244834064,
      "grad_norm": 2.47517466545105,
      "learning_rate": 6.111925708699901e-05,
      "loss": 1.3165,
      "step": 10288
    },
    {
      "epoch": 1.6106762680025049,
      "grad_norm": 0.6524291634559631,
      "learning_rate": 6.109481915933529e-05,
      "loss": 0.1867,
      "step": 10289
    },
    {
      "epoch": 1.610832811521603,
      "grad_norm": 0.988003134727478,
      "learning_rate": 6.107038123167155e-05,
      "loss": 0.2824,
      "step": 10290
    },
    {
      "epoch": 1.6109893550407013,
      "grad_norm": 0.47198671102523804,
      "learning_rate": 6.104594330400782e-05,
      "loss": 0.1858,
      "step": 10291
    },
    {
      "epoch": 1.6111458985597995,
      "grad_norm": 0.9640408754348755,
      "learning_rate": 6.102150537634408e-05,
      "loss": 0.1501,
      "step": 10292
    },
    {
      "epoch": 1.611302442078898,
      "grad_norm": 1.5908207893371582,
      "learning_rate": 6.099706744868035e-05,
      "loss": 0.2762,
      "step": 10293
    },
    {
      "epoch": 1.6114589855979964,
      "grad_norm": 0.730593740940094,
      "learning_rate": 6.097262952101661e-05,
      "loss": 0.1985,
      "step": 10294
    },
    {
      "epoch": 1.6116155291170946,
      "grad_norm": 0.45398107171058655,
      "learning_rate": 6.094819159335288e-05,
      "loss": 0.2736,
      "step": 10295
    },
    {
      "epoch": 1.6117720726361928,
      "grad_norm": 1.1660237312316895,
      "learning_rate": 6.092375366568914e-05,
      "loss": 0.3241,
      "step": 10296
    },
    {
      "epoch": 1.611928616155291,
      "grad_norm": 0.7459734678268433,
      "learning_rate": 6.089931573802541e-05,
      "loss": 0.2403,
      "step": 10297
    },
    {
      "epoch": 1.6120851596743895,
      "grad_norm": 1.4114768505096436,
      "learning_rate": 6.087487781036168e-05,
      "loss": 0.3583,
      "step": 10298
    },
    {
      "epoch": 1.612241703193488,
      "grad_norm": 0.7342216968536377,
      "learning_rate": 6.0850439882697936e-05,
      "loss": 0.2025,
      "step": 10299
    },
    {
      "epoch": 1.6123982467125861,
      "grad_norm": 0.8763503432273865,
      "learning_rate": 6.082600195503421e-05,
      "loss": 0.2146,
      "step": 10300
    },
    {
      "epoch": 1.6125547902316844,
      "grad_norm": 1.2437353134155273,
      "learning_rate": 6.080156402737047e-05,
      "loss": 0.2597,
      "step": 10301
    },
    {
      "epoch": 1.6127113337507826,
      "grad_norm": 0.5917118191719055,
      "learning_rate": 6.077712609970674e-05,
      "loss": 0.1298,
      "step": 10302
    },
    {
      "epoch": 1.612867877269881,
      "grad_norm": 1.689225196838379,
      "learning_rate": 6.0752688172043e-05,
      "loss": 0.2692,
      "step": 10303
    },
    {
      "epoch": 1.6130244207889795,
      "grad_norm": 1.53922438621521,
      "learning_rate": 6.0728250244379274e-05,
      "loss": 0.3584,
      "step": 10304
    },
    {
      "epoch": 1.6131809643080777,
      "grad_norm": 1.4097189903259277,
      "learning_rate": 6.070381231671553e-05,
      "loss": 0.4676,
      "step": 10305
    },
    {
      "epoch": 1.6133375078271759,
      "grad_norm": 2.491358757019043,
      "learning_rate": 6.0679374389051803e-05,
      "loss": 0.4225,
      "step": 10306
    },
    {
      "epoch": 1.613494051346274,
      "grad_norm": 2.754809617996216,
      "learning_rate": 6.065493646138807e-05,
      "loss": 0.5883,
      "step": 10307
    },
    {
      "epoch": 1.6136505948653725,
      "grad_norm": 1.1778233051300049,
      "learning_rate": 6.063049853372433e-05,
      "loss": 0.6693,
      "step": 10308
    },
    {
      "epoch": 1.613807138384471,
      "grad_norm": 0.8539136052131653,
      "learning_rate": 6.06060606060606e-05,
      "loss": 0.2029,
      "step": 10309
    },
    {
      "epoch": 1.6139636819035692,
      "grad_norm": 2.6600699424743652,
      "learning_rate": 6.058162267839687e-05,
      "loss": 0.3794,
      "step": 10310
    },
    {
      "epoch": 1.6141202254226674,
      "grad_norm": 1.350770354270935,
      "learning_rate": 6.055718475073313e-05,
      "loss": 0.3923,
      "step": 10311
    },
    {
      "epoch": 1.6142767689417659,
      "grad_norm": 2.732351303100586,
      "learning_rate": 6.05327468230694e-05,
      "loss": 0.4274,
      "step": 10312
    },
    {
      "epoch": 1.614433312460864,
      "grad_norm": 1.7970898151397705,
      "learning_rate": 6.0508308895405664e-05,
      "loss": 0.5387,
      "step": 10313
    },
    {
      "epoch": 1.6145898559799625,
      "grad_norm": 2.37106990814209,
      "learning_rate": 6.048387096774193e-05,
      "loss": 0.4457,
      "step": 10314
    },
    {
      "epoch": 1.6147463994990607,
      "grad_norm": 1.6102174520492554,
      "learning_rate": 6.0459433040078193e-05,
      "loss": 0.6682,
      "step": 10315
    },
    {
      "epoch": 1.614902943018159,
      "grad_norm": 1.9735941886901855,
      "learning_rate": 6.0434995112414465e-05,
      "loss": 0.7914,
      "step": 10316
    },
    {
      "epoch": 1.6150594865372574,
      "grad_norm": 1.6379317045211792,
      "learning_rate": 6.041055718475072e-05,
      "loss": 0.6588,
      "step": 10317
    },
    {
      "epoch": 1.6152160300563556,
      "grad_norm": 1.541721224784851,
      "learning_rate": 6.0386119257086995e-05,
      "loss": 0.6006,
      "step": 10318
    },
    {
      "epoch": 1.615372573575454,
      "grad_norm": 5.82126522064209,
      "learning_rate": 6.036168132942326e-05,
      "loss": 0.6219,
      "step": 10319
    },
    {
      "epoch": 1.6155291170945523,
      "grad_norm": 3.0988450050354004,
      "learning_rate": 6.0337243401759524e-05,
      "loss": 0.4162,
      "step": 10320
    },
    {
      "epoch": 1.6156856606136505,
      "grad_norm": 3.262086868286133,
      "learning_rate": 6.031280547409579e-05,
      "loss": 1.2557,
      "step": 10321
    },
    {
      "epoch": 1.615842204132749,
      "grad_norm": 1.9713597297668457,
      "learning_rate": 6.028836754643206e-05,
      "loss": 0.8396,
      "step": 10322
    },
    {
      "epoch": 1.6159987476518474,
      "grad_norm": 2.0286524295806885,
      "learning_rate": 6.026392961876832e-05,
      "loss": 0.5553,
      "step": 10323
    },
    {
      "epoch": 1.6161552911709456,
      "grad_norm": 2.3841817378997803,
      "learning_rate": 6.023949169110459e-05,
      "loss": 0.5284,
      "step": 10324
    },
    {
      "epoch": 1.6163118346900438,
      "grad_norm": 2.634639024734497,
      "learning_rate": 6.0215053763440855e-05,
      "loss": 0.7772,
      "step": 10325
    },
    {
      "epoch": 1.616468378209142,
      "grad_norm": 3.448430299758911,
      "learning_rate": 6.019061583577712e-05,
      "loss": 0.5422,
      "step": 10326
    },
    {
      "epoch": 1.6166249217282405,
      "grad_norm": 1.8632174730300903,
      "learning_rate": 6.0166177908113385e-05,
      "loss": 1.0043,
      "step": 10327
    },
    {
      "epoch": 1.616781465247339,
      "grad_norm": 3.0576517581939697,
      "learning_rate": 6.0141739980449656e-05,
      "loss": 0.906,
      "step": 10328
    },
    {
      "epoch": 1.6169380087664371,
      "grad_norm": 6.012516975402832,
      "learning_rate": 6.0117302052785914e-05,
      "loss": 1.7748,
      "step": 10329
    },
    {
      "epoch": 1.6170945522855353,
      "grad_norm": 1.4931806325912476,
      "learning_rate": 6.0092864125122186e-05,
      "loss": 0.441,
      "step": 10330
    },
    {
      "epoch": 1.6172510958046336,
      "grad_norm": 2.6197221279144287,
      "learning_rate": 6.006842619745845e-05,
      "loss": 1.212,
      "step": 10331
    },
    {
      "epoch": 1.617407639323732,
      "grad_norm": 2.8146133422851562,
      "learning_rate": 6.0043988269794715e-05,
      "loss": 0.9955,
      "step": 10332
    },
    {
      "epoch": 1.6175641828428304,
      "grad_norm": 2.633820056915283,
      "learning_rate": 6.001955034213098e-05,
      "loss": 1.4292,
      "step": 10333
    },
    {
      "epoch": 1.6177207263619287,
      "grad_norm": 5.142744541168213,
      "learning_rate": 5.999511241446725e-05,
      "loss": 1.3371,
      "step": 10334
    },
    {
      "epoch": 1.6178772698810269,
      "grad_norm": 1.9300106763839722,
      "learning_rate": 5.997067448680351e-05,
      "loss": 0.5765,
      "step": 10335
    },
    {
      "epoch": 1.618033813400125,
      "grad_norm": 3.921834945678711,
      "learning_rate": 5.994623655913978e-05,
      "loss": 0.6505,
      "step": 10336
    },
    {
      "epoch": 1.6181903569192235,
      "grad_norm": 2.213167905807495,
      "learning_rate": 5.9921798631476046e-05,
      "loss": 0.8817,
      "step": 10337
    },
    {
      "epoch": 1.618346900438322,
      "grad_norm": 2.0147557258605957,
      "learning_rate": 5.989736070381231e-05,
      "loss": 0.6968,
      "step": 10338
    },
    {
      "epoch": 1.6185034439574202,
      "grad_norm": 0.6969068646430969,
      "learning_rate": 5.9872922776148576e-05,
      "loss": 0.1943,
      "step": 10339
    },
    {
      "epoch": 1.6186599874765184,
      "grad_norm": 0.5584667921066284,
      "learning_rate": 5.984848484848485e-05,
      "loss": 0.18,
      "step": 10340
    },
    {
      "epoch": 1.6188165309956166,
      "grad_norm": 0.6967960596084595,
      "learning_rate": 5.9824046920821105e-05,
      "loss": 0.2321,
      "step": 10341
    },
    {
      "epoch": 1.618973074514715,
      "grad_norm": 0.3680100440979004,
      "learning_rate": 5.979960899315738e-05,
      "loss": 0.135,
      "step": 10342
    },
    {
      "epoch": 1.6191296180338135,
      "grad_norm": 0.755924642086029,
      "learning_rate": 5.977517106549364e-05,
      "loss": 0.2955,
      "step": 10343
    },
    {
      "epoch": 1.6192861615529117,
      "grad_norm": 2.2289376258850098,
      "learning_rate": 5.9750733137829907e-05,
      "loss": 0.2681,
      "step": 10344
    },
    {
      "epoch": 1.61944270507201,
      "grad_norm": 1.0998948812484741,
      "learning_rate": 5.972629521016617e-05,
      "loss": 0.1985,
      "step": 10345
    },
    {
      "epoch": 1.6195992485911084,
      "grad_norm": 0.6350347399711609,
      "learning_rate": 5.970185728250244e-05,
      "loss": 0.266,
      "step": 10346
    },
    {
      "epoch": 1.6197557921102066,
      "grad_norm": 0.7343830466270447,
      "learning_rate": 5.96774193548387e-05,
      "loss": 0.1386,
      "step": 10347
    },
    {
      "epoch": 1.619912335629305,
      "grad_norm": 1.0701186656951904,
      "learning_rate": 5.965298142717497e-05,
      "loss": 0.4067,
      "step": 10348
    },
    {
      "epoch": 1.6200688791484033,
      "grad_norm": 1.645283818244934,
      "learning_rate": 5.962854349951124e-05,
      "loss": 0.5609,
      "step": 10349
    },
    {
      "epoch": 1.6202254226675015,
      "grad_norm": 1.122719168663025,
      "learning_rate": 5.96041055718475e-05,
      "loss": 0.2413,
      "step": 10350
    },
    {
      "epoch": 1.6203819661866,
      "grad_norm": 0.7150799036026001,
      "learning_rate": 5.957966764418377e-05,
      "loss": 0.3426,
      "step": 10351
    },
    {
      "epoch": 1.6205385097056983,
      "grad_norm": 0.598060131072998,
      "learning_rate": 5.955522971652004e-05,
      "loss": 0.1765,
      "step": 10352
    },
    {
      "epoch": 1.6206950532247966,
      "grad_norm": 0.8176664710044861,
      "learning_rate": 5.9530791788856297e-05,
      "loss": 0.29,
      "step": 10353
    },
    {
      "epoch": 1.6208515967438948,
      "grad_norm": 1.5242834091186523,
      "learning_rate": 5.950635386119257e-05,
      "loss": 0.3729,
      "step": 10354
    },
    {
      "epoch": 1.621008140262993,
      "grad_norm": 1.9225590229034424,
      "learning_rate": 5.948191593352883e-05,
      "loss": 0.7854,
      "step": 10355
    },
    {
      "epoch": 1.6211646837820914,
      "grad_norm": 1.2514513731002808,
      "learning_rate": 5.94574780058651e-05,
      "loss": 0.5056,
      "step": 10356
    },
    {
      "epoch": 1.6213212273011899,
      "grad_norm": 1.7647795677185059,
      "learning_rate": 5.943304007820136e-05,
      "loss": 0.4021,
      "step": 10357
    },
    {
      "epoch": 1.621477770820288,
      "grad_norm": 0.9567508697509766,
      "learning_rate": 5.9408602150537634e-05,
      "loss": 0.3619,
      "step": 10358
    },
    {
      "epoch": 1.6216343143393863,
      "grad_norm": 1.5597784519195557,
      "learning_rate": 5.938416422287389e-05,
      "loss": 0.4833,
      "step": 10359
    },
    {
      "epoch": 1.6217908578584845,
      "grad_norm": 2.9853146076202393,
      "learning_rate": 5.9359726295210164e-05,
      "loss": 0.4786,
      "step": 10360
    },
    {
      "epoch": 1.621947401377583,
      "grad_norm": 1.8324689865112305,
      "learning_rate": 5.933528836754643e-05,
      "loss": 0.4453,
      "step": 10361
    },
    {
      "epoch": 1.6221039448966814,
      "grad_norm": 2.3607141971588135,
      "learning_rate": 5.931085043988269e-05,
      "loss": 0.5207,
      "step": 10362
    },
    {
      "epoch": 1.6222604884157796,
      "grad_norm": 1.914950966835022,
      "learning_rate": 5.928641251221896e-05,
      "loss": 0.5193,
      "step": 10363
    },
    {
      "epoch": 1.6224170319348779,
      "grad_norm": 2.1781463623046875,
      "learning_rate": 5.926197458455523e-05,
      "loss": 0.6317,
      "step": 10364
    },
    {
      "epoch": 1.622573575453976,
      "grad_norm": 4.128535270690918,
      "learning_rate": 5.923753665689149e-05,
      "loss": 0.7715,
      "step": 10365
    },
    {
      "epoch": 1.6227301189730745,
      "grad_norm": 3.1875927448272705,
      "learning_rate": 5.921309872922776e-05,
      "loss": 0.7828,
      "step": 10366
    },
    {
      "epoch": 1.622886662492173,
      "grad_norm": 1.828016996383667,
      "learning_rate": 5.9188660801564024e-05,
      "loss": 0.6662,
      "step": 10367
    },
    {
      "epoch": 1.6230432060112712,
      "grad_norm": 2.3392691612243652,
      "learning_rate": 5.916422287390029e-05,
      "loss": 0.647,
      "step": 10368
    },
    {
      "epoch": 1.6231997495303694,
      "grad_norm": 2.4134435653686523,
      "learning_rate": 5.9139784946236554e-05,
      "loss": 0.8108,
      "step": 10369
    },
    {
      "epoch": 1.6233562930494676,
      "grad_norm": 0.9342465996742249,
      "learning_rate": 5.9115347018572825e-05,
      "loss": 0.326,
      "step": 10370
    },
    {
      "epoch": 1.623512836568566,
      "grad_norm": 2.0880491733551025,
      "learning_rate": 5.909090909090908e-05,
      "loss": 0.6187,
      "step": 10371
    },
    {
      "epoch": 1.6236693800876645,
      "grad_norm": 1.288293719291687,
      "learning_rate": 5.9066471163245355e-05,
      "loss": 0.5137,
      "step": 10372
    },
    {
      "epoch": 1.6238259236067627,
      "grad_norm": 2.43259596824646,
      "learning_rate": 5.904203323558162e-05,
      "loss": 0.7306,
      "step": 10373
    },
    {
      "epoch": 1.623982467125861,
      "grad_norm": 1.6936664581298828,
      "learning_rate": 5.901759530791788e-05,
      "loss": 0.7173,
      "step": 10374
    },
    {
      "epoch": 1.6241390106449591,
      "grad_norm": 1.1513222455978394,
      "learning_rate": 5.899315738025415e-05,
      "loss": 0.3325,
      "step": 10375
    },
    {
      "epoch": 1.6242955541640576,
      "grad_norm": 2.4946534633636475,
      "learning_rate": 5.896871945259042e-05,
      "loss": 0.8307,
      "step": 10376
    },
    {
      "epoch": 1.624452097683156,
      "grad_norm": 3.35575270652771,
      "learning_rate": 5.894428152492668e-05,
      "loss": 1.3256,
      "step": 10377
    },
    {
      "epoch": 1.6246086412022542,
      "grad_norm": 8.07103157043457,
      "learning_rate": 5.891984359726295e-05,
      "loss": 0.6945,
      "step": 10378
    },
    {
      "epoch": 1.6247651847213525,
      "grad_norm": 2.7447383403778076,
      "learning_rate": 5.8895405669599215e-05,
      "loss": 0.9647,
      "step": 10379
    },
    {
      "epoch": 1.624921728240451,
      "grad_norm": 2.4406001567840576,
      "learning_rate": 5.887096774193547e-05,
      "loss": 1.3698,
      "step": 10380
    },
    {
      "epoch": 1.625078271759549,
      "grad_norm": 6.9520978927612305,
      "learning_rate": 5.8846529814271745e-05,
      "loss": 0.7408,
      "step": 10381
    },
    {
      "epoch": 1.6252348152786475,
      "grad_norm": 2.9394330978393555,
      "learning_rate": 5.8822091886608016e-05,
      "loss": 1.0021,
      "step": 10382
    },
    {
      "epoch": 1.6253913587977458,
      "grad_norm": 3.4917502403259277,
      "learning_rate": 5.8797653958944274e-05,
      "loss": 1.1361,
      "step": 10383
    },
    {
      "epoch": 1.625547902316844,
      "grad_norm": 1.3461923599243164,
      "learning_rate": 5.8773216031280546e-05,
      "loss": 0.2525,
      "step": 10384
    },
    {
      "epoch": 1.6257044458359424,
      "grad_norm": 1.8824841976165771,
      "learning_rate": 5.874877810361681e-05,
      "loss": 0.7553,
      "step": 10385
    },
    {
      "epoch": 1.6258609893550409,
      "grad_norm": 1.4790881872177124,
      "learning_rate": 5.872434017595307e-05,
      "loss": 0.5558,
      "step": 10386
    },
    {
      "epoch": 1.626017532874139,
      "grad_norm": 1.9646244049072266,
      "learning_rate": 5.869990224828934e-05,
      "loss": 0.4602,
      "step": 10387
    },
    {
      "epoch": 1.6261740763932373,
      "grad_norm": 1.7830438613891602,
      "learning_rate": 5.867546432062561e-05,
      "loss": 0.5251,
      "step": 10388
    },
    {
      "epoch": 1.6263306199123355,
      "grad_norm": 0.5768091678619385,
      "learning_rate": 5.865102639296187e-05,
      "loss": 0.2012,
      "step": 10389
    },
    {
      "epoch": 1.626487163431434,
      "grad_norm": 1.113883137702942,
      "learning_rate": 5.862658846529814e-05,
      "loss": 0.1778,
      "step": 10390
    },
    {
      "epoch": 1.6266437069505324,
      "grad_norm": 0.5060340762138367,
      "learning_rate": 5.8602150537634406e-05,
      "loss": 0.1855,
      "step": 10391
    },
    {
      "epoch": 1.6268002504696306,
      "grad_norm": 0.514045000076294,
      "learning_rate": 5.8577712609970664e-05,
      "loss": 0.1701,
      "step": 10392
    },
    {
      "epoch": 1.6269567939887288,
      "grad_norm": 0.860876202583313,
      "learning_rate": 5.8553274682306936e-05,
      "loss": 0.2594,
      "step": 10393
    },
    {
      "epoch": 1.627113337507827,
      "grad_norm": 1.2746233940124512,
      "learning_rate": 5.852883675464321e-05,
      "loss": 0.1743,
      "step": 10394
    },
    {
      "epoch": 1.6272698810269255,
      "grad_norm": 0.68976891040802,
      "learning_rate": 5.8504398826979466e-05,
      "loss": 0.203,
      "step": 10395
    },
    {
      "epoch": 1.627426424546024,
      "grad_norm": 0.48356834053993225,
      "learning_rate": 5.847996089931573e-05,
      "loss": 0.226,
      "step": 10396
    },
    {
      "epoch": 1.6275829680651221,
      "grad_norm": 1.8807826042175293,
      "learning_rate": 5.8455522971652e-05,
      "loss": 0.1366,
      "step": 10397
    },
    {
      "epoch": 1.6277395115842204,
      "grad_norm": 0.7366307377815247,
      "learning_rate": 5.843108504398826e-05,
      "loss": 0.3315,
      "step": 10398
    },
    {
      "epoch": 1.6278960551033186,
      "grad_norm": 0.8928070068359375,
      "learning_rate": 5.840664711632453e-05,
      "loss": 0.2295,
      "step": 10399
    },
    {
      "epoch": 1.628052598622417,
      "grad_norm": 0.8314988613128662,
      "learning_rate": 5.83822091886608e-05,
      "loss": 0.2468,
      "step": 10400
    },
    {
      "epoch": 1.6282091421415155,
      "grad_norm": 1.0625923871994019,
      "learning_rate": 5.835777126099706e-05,
      "loss": 0.2625,
      "step": 10401
    },
    {
      "epoch": 1.6283656856606137,
      "grad_norm": 3.7548062801361084,
      "learning_rate": 5.8333333333333326e-05,
      "loss": 0.4308,
      "step": 10402
    },
    {
      "epoch": 1.628522229179712,
      "grad_norm": 2.1772689819335938,
      "learning_rate": 5.83088954056696e-05,
      "loss": 0.3888,
      "step": 10403
    },
    {
      "epoch": 1.6286787726988101,
      "grad_norm": 1.233253002166748,
      "learning_rate": 5.8284457478005856e-05,
      "loss": 0.2004,
      "step": 10404
    },
    {
      "epoch": 1.6288353162179086,
      "grad_norm": 1.6462030410766602,
      "learning_rate": 5.826001955034213e-05,
      "loss": 0.3962,
      "step": 10405
    },
    {
      "epoch": 1.628991859737007,
      "grad_norm": 7.115745544433594,
      "learning_rate": 5.82355816226784e-05,
      "loss": 0.3691,
      "step": 10406
    },
    {
      "epoch": 1.6291484032561052,
      "grad_norm": 1.0875389575958252,
      "learning_rate": 5.821114369501466e-05,
      "loss": 0.2721,
      "step": 10407
    },
    {
      "epoch": 1.6293049467752034,
      "grad_norm": 1.3013725280761719,
      "learning_rate": 5.818670576735092e-05,
      "loss": 0.4565,
      "step": 10408
    },
    {
      "epoch": 1.6294614902943017,
      "grad_norm": 2.092606782913208,
      "learning_rate": 5.816226783968719e-05,
      "loss": 0.5847,
      "step": 10409
    },
    {
      "epoch": 1.6296180338134,
      "grad_norm": 1.729488492012024,
      "learning_rate": 5.813782991202345e-05,
      "loss": 0.32,
      "step": 10410
    },
    {
      "epoch": 1.6297745773324985,
      "grad_norm": 1.737029790878296,
      "learning_rate": 5.811339198435972e-05,
      "loss": 0.2449,
      "step": 10411
    },
    {
      "epoch": 1.6299311208515967,
      "grad_norm": 2.1841983795166016,
      "learning_rate": 5.8088954056695994e-05,
      "loss": 0.7863,
      "step": 10412
    },
    {
      "epoch": 1.630087664370695,
      "grad_norm": 0.9990864992141724,
      "learning_rate": 5.806451612903225e-05,
      "loss": 0.2027,
      "step": 10413
    },
    {
      "epoch": 1.6302442078897934,
      "grad_norm": 2.165163040161133,
      "learning_rate": 5.804007820136852e-05,
      "loss": 0.5477,
      "step": 10414
    },
    {
      "epoch": 1.6304007514088916,
      "grad_norm": 5.074351787567139,
      "learning_rate": 5.801564027370479e-05,
      "loss": 0.5501,
      "step": 10415
    },
    {
      "epoch": 1.63055729492799,
      "grad_norm": 1.8900859355926514,
      "learning_rate": 5.799120234604105e-05,
      "loss": 0.4008,
      "step": 10416
    },
    {
      "epoch": 1.6307138384470883,
      "grad_norm": 1.9944469928741455,
      "learning_rate": 5.796676441837732e-05,
      "loss": 0.4468,
      "step": 10417
    },
    {
      "epoch": 1.6308703819661865,
      "grad_norm": 2.274146795272827,
      "learning_rate": 5.794232649071358e-05,
      "loss": 0.7135,
      "step": 10418
    },
    {
      "epoch": 1.631026925485285,
      "grad_norm": 3.8034324645996094,
      "learning_rate": 5.791788856304985e-05,
      "loss": 0.4055,
      "step": 10419
    },
    {
      "epoch": 1.6311834690043834,
      "grad_norm": 2.4638800621032715,
      "learning_rate": 5.789345063538611e-05,
      "loss": 0.8925,
      "step": 10420
    },
    {
      "epoch": 1.6313400125234816,
      "grad_norm": 5.541257858276367,
      "learning_rate": 5.7869012707722384e-05,
      "loss": 0.8966,
      "step": 10421
    },
    {
      "epoch": 1.6314965560425798,
      "grad_norm": 6.215266704559326,
      "learning_rate": 5.784457478005864e-05,
      "loss": 0.6739,
      "step": 10422
    },
    {
      "epoch": 1.631653099561678,
      "grad_norm": 1.26706862449646,
      "learning_rate": 5.7820136852394914e-05,
      "loss": 0.5802,
      "step": 10423
    },
    {
      "epoch": 1.6318096430807765,
      "grad_norm": 2.1300575733184814,
      "learning_rate": 5.779569892473117e-05,
      "loss": 0.6051,
      "step": 10424
    },
    {
      "epoch": 1.631966186599875,
      "grad_norm": 4.257119178771973,
      "learning_rate": 5.7771260997067443e-05,
      "loss": 1.0206,
      "step": 10425
    },
    {
      "epoch": 1.6321227301189731,
      "grad_norm": 2.297468662261963,
      "learning_rate": 5.774682306940371e-05,
      "loss": 0.7068,
      "step": 10426
    },
    {
      "epoch": 1.6322792736380713,
      "grad_norm": 3.372777223587036,
      "learning_rate": 5.772238514173997e-05,
      "loss": 0.7299,
      "step": 10427
    },
    {
      "epoch": 1.6324358171571696,
      "grad_norm": 2.6699585914611816,
      "learning_rate": 5.769794721407624e-05,
      "loss": 1.2506,
      "step": 10428
    },
    {
      "epoch": 1.632592360676268,
      "grad_norm": 2.306000232696533,
      "learning_rate": 5.767350928641251e-05,
      "loss": 0.8118,
      "step": 10429
    },
    {
      "epoch": 1.6327489041953664,
      "grad_norm": 3.001568078994751,
      "learning_rate": 5.764907135874877e-05,
      "loss": 0.8326,
      "step": 10430
    },
    {
      "epoch": 1.6329054477144647,
      "grad_norm": 2.3206942081451416,
      "learning_rate": 5.762463343108504e-05,
      "loss": 1.0572,
      "step": 10431
    },
    {
      "epoch": 1.6330619912335629,
      "grad_norm": 9.964920043945312,
      "learning_rate": 5.7600195503421304e-05,
      "loss": 1.1929,
      "step": 10432
    },
    {
      "epoch": 1.633218534752661,
      "grad_norm": 2.1716456413269043,
      "learning_rate": 5.757575757575757e-05,
      "loss": 0.7854,
      "step": 10433
    },
    {
      "epoch": 1.6333750782717595,
      "grad_norm": 3.3193044662475586,
      "learning_rate": 5.7551319648093833e-05,
      "loss": 0.4832,
      "step": 10434
    },
    {
      "epoch": 1.633531621790858,
      "grad_norm": 3.8049163818359375,
      "learning_rate": 5.7526881720430105e-05,
      "loss": 0.343,
      "step": 10435
    },
    {
      "epoch": 1.6336881653099562,
      "grad_norm": 3.3288490772247314,
      "learning_rate": 5.750244379276636e-05,
      "loss": 0.7935,
      "step": 10436
    },
    {
      "epoch": 1.6338447088290544,
      "grad_norm": 1.9719550609588623,
      "learning_rate": 5.7478005865102635e-05,
      "loss": 1.2361,
      "step": 10437
    },
    {
      "epoch": 1.6340012523481526,
      "grad_norm": 2.7374227046966553,
      "learning_rate": 5.74535679374389e-05,
      "loss": 1.1997,
      "step": 10438
    },
    {
      "epoch": 1.634157795867251,
      "grad_norm": 0.5598339438438416,
      "learning_rate": 5.7429130009775164e-05,
      "loss": 0.1796,
      "step": 10439
    },
    {
      "epoch": 1.6343143393863495,
      "grad_norm": 0.9065269231796265,
      "learning_rate": 5.740469208211143e-05,
      "loss": 0.2172,
      "step": 10440
    },
    {
      "epoch": 1.6344708829054477,
      "grad_norm": 1.6582931280136108,
      "learning_rate": 5.73802541544477e-05,
      "loss": 0.1974,
      "step": 10441
    },
    {
      "epoch": 1.634627426424546,
      "grad_norm": 0.5348918437957764,
      "learning_rate": 5.735581622678396e-05,
      "loss": 0.2214,
      "step": 10442
    },
    {
      "epoch": 1.6347839699436444,
      "grad_norm": 2.297356367111206,
      "learning_rate": 5.733137829912023e-05,
      "loss": 0.1988,
      "step": 10443
    },
    {
      "epoch": 1.6349405134627426,
      "grad_norm": 0.5752388834953308,
      "learning_rate": 5.7306940371456495e-05,
      "loss": 0.1346,
      "step": 10444
    },
    {
      "epoch": 1.635097056981841,
      "grad_norm": 0.8799678683280945,
      "learning_rate": 5.728250244379276e-05,
      "loss": 0.2363,
      "step": 10445
    },
    {
      "epoch": 1.6352536005009393,
      "grad_norm": 0.6415383219718933,
      "learning_rate": 5.7258064516129025e-05,
      "loss": 0.2131,
      "step": 10446
    },
    {
      "epoch": 1.6354101440200375,
      "grad_norm": 17.69931411743164,
      "learning_rate": 5.7233626588465296e-05,
      "loss": 0.344,
      "step": 10447
    },
    {
      "epoch": 1.635566687539136,
      "grad_norm": 1.341117262840271,
      "learning_rate": 5.7209188660801554e-05,
      "loss": 0.2848,
      "step": 10448
    },
    {
      "epoch": 1.6357232310582341,
      "grad_norm": 1.4106292724609375,
      "learning_rate": 5.7184750733137826e-05,
      "loss": 0.238,
      "step": 10449
    },
    {
      "epoch": 1.6358797745773326,
      "grad_norm": 1.0851185321807861,
      "learning_rate": 5.716031280547409e-05,
      "loss": 0.3487,
      "step": 10450
    },
    {
      "epoch": 1.6360363180964308,
      "grad_norm": 1.2558451890945435,
      "learning_rate": 5.7135874877810355e-05,
      "loss": 0.4459,
      "step": 10451
    },
    {
      "epoch": 1.636192861615529,
      "grad_norm": 2.4315550327301025,
      "learning_rate": 5.711143695014662e-05,
      "loss": 0.3275,
      "step": 10452
    },
    {
      "epoch": 1.6363494051346275,
      "grad_norm": 1.1903713941574097,
      "learning_rate": 5.708699902248289e-05,
      "loss": 0.5321,
      "step": 10453
    },
    {
      "epoch": 1.636505948653726,
      "grad_norm": 2.3874330520629883,
      "learning_rate": 5.706256109481915e-05,
      "loss": 0.5209,
      "step": 10454
    },
    {
      "epoch": 1.6366624921728241,
      "grad_norm": 2.2533371448516846,
      "learning_rate": 5.703812316715542e-05,
      "loss": 0.4226,
      "step": 10455
    },
    {
      "epoch": 1.6368190356919223,
      "grad_norm": 1.0902347564697266,
      "learning_rate": 5.7013685239491686e-05,
      "loss": 0.205,
      "step": 10456
    },
    {
      "epoch": 1.6369755792110205,
      "grad_norm": 2.033475875854492,
      "learning_rate": 5.698924731182795e-05,
      "loss": 0.1742,
      "step": 10457
    },
    {
      "epoch": 1.637132122730119,
      "grad_norm": 1.488540530204773,
      "learning_rate": 5.6964809384164216e-05,
      "loss": 0.2724,
      "step": 10458
    },
    {
      "epoch": 1.6372886662492174,
      "grad_norm": 3.625598192214966,
      "learning_rate": 5.694037145650049e-05,
      "loss": 0.6245,
      "step": 10459
    },
    {
      "epoch": 1.6374452097683156,
      "grad_norm": 3.725924015045166,
      "learning_rate": 5.6915933528836745e-05,
      "loss": 0.3794,
      "step": 10460
    },
    {
      "epoch": 1.6376017532874139,
      "grad_norm": 1.230906367301941,
      "learning_rate": 5.689149560117302e-05,
      "loss": 0.2737,
      "step": 10461
    },
    {
      "epoch": 1.637758296806512,
      "grad_norm": 3.53603196144104,
      "learning_rate": 5.686705767350928e-05,
      "loss": 0.5026,
      "step": 10462
    },
    {
      "epoch": 1.6379148403256105,
      "grad_norm": 1.3804330825805664,
      "learning_rate": 5.6842619745845547e-05,
      "loss": 0.2446,
      "step": 10463
    },
    {
      "epoch": 1.638071383844709,
      "grad_norm": 2.4402577877044678,
      "learning_rate": 5.681818181818181e-05,
      "loss": 0.4187,
      "step": 10464
    },
    {
      "epoch": 1.6382279273638072,
      "grad_norm": 1.448628306388855,
      "learning_rate": 5.679374389051808e-05,
      "loss": 0.5199,
      "step": 10465
    },
    {
      "epoch": 1.6383844708829054,
      "grad_norm": 3.0513761043548584,
      "learning_rate": 5.676930596285434e-05,
      "loss": 0.7834,
      "step": 10466
    },
    {
      "epoch": 1.6385410144020036,
      "grad_norm": 2.2523458003997803,
      "learning_rate": 5.674486803519061e-05,
      "loss": 0.8243,
      "step": 10467
    },
    {
      "epoch": 1.638697557921102,
      "grad_norm": 2.0097694396972656,
      "learning_rate": 5.672043010752688e-05,
      "loss": 0.3212,
      "step": 10468
    },
    {
      "epoch": 1.6388541014402005,
      "grad_norm": 4.3316874504089355,
      "learning_rate": 5.669599217986314e-05,
      "loss": 0.6163,
      "step": 10469
    },
    {
      "epoch": 1.6390106449592987,
      "grad_norm": 1.4377896785736084,
      "learning_rate": 5.667155425219941e-05,
      "loss": 0.5379,
      "step": 10470
    },
    {
      "epoch": 1.639167188478397,
      "grad_norm": 3.63924241065979,
      "learning_rate": 5.664711632453568e-05,
      "loss": 0.8601,
      "step": 10471
    },
    {
      "epoch": 1.6393237319974951,
      "grad_norm": 2.5112884044647217,
      "learning_rate": 5.6622678396871937e-05,
      "loss": 0.6937,
      "step": 10472
    },
    {
      "epoch": 1.6394802755165936,
      "grad_norm": 3.881232500076294,
      "learning_rate": 5.659824046920821e-05,
      "loss": 1.0227,
      "step": 10473
    },
    {
      "epoch": 1.639636819035692,
      "grad_norm": 1.790979027748108,
      "learning_rate": 5.657380254154447e-05,
      "loss": 0.9926,
      "step": 10474
    },
    {
      "epoch": 1.6397933625547902,
      "grad_norm": 4.465099811553955,
      "learning_rate": 5.654936461388074e-05,
      "loss": 0.9959,
      "step": 10475
    },
    {
      "epoch": 1.6399499060738885,
      "grad_norm": 1.5608214139938354,
      "learning_rate": 5.6524926686217e-05,
      "loss": 0.4801,
      "step": 10476
    },
    {
      "epoch": 1.640106449592987,
      "grad_norm": 3.728848457336426,
      "learning_rate": 5.6500488758553274e-05,
      "loss": 0.6256,
      "step": 10477
    },
    {
      "epoch": 1.6402629931120851,
      "grad_norm": 1.7102699279785156,
      "learning_rate": 5.647605083088953e-05,
      "loss": 1.2323,
      "step": 10478
    },
    {
      "epoch": 1.6404195366311836,
      "grad_norm": 3.236166477203369,
      "learning_rate": 5.6451612903225804e-05,
      "loss": 1.4178,
      "step": 10479
    },
    {
      "epoch": 1.6405760801502818,
      "grad_norm": 6.140420436859131,
      "learning_rate": 5.642717497556207e-05,
      "loss": 1.1856,
      "step": 10480
    },
    {
      "epoch": 1.64073262366938,
      "grad_norm": 2.2286741733551025,
      "learning_rate": 5.640273704789833e-05,
      "loss": 0.6144,
      "step": 10481
    },
    {
      "epoch": 1.6408891671884784,
      "grad_norm": 2.5856056213378906,
      "learning_rate": 5.63782991202346e-05,
      "loss": 1.2648,
      "step": 10482
    },
    {
      "epoch": 1.6410457107075767,
      "grad_norm": 3.3307247161865234,
      "learning_rate": 5.635386119257087e-05,
      "loss": 0.9104,
      "step": 10483
    },
    {
      "epoch": 1.641202254226675,
      "grad_norm": 2.219688892364502,
      "learning_rate": 5.632942326490713e-05,
      "loss": 0.9191,
      "step": 10484
    },
    {
      "epoch": 1.6413587977457733,
      "grad_norm": 3.4340312480926514,
      "learning_rate": 5.63049853372434e-05,
      "loss": 0.5855,
      "step": 10485
    },
    {
      "epoch": 1.6415153412648715,
      "grad_norm": 1.018791913986206,
      "learning_rate": 5.6280547409579664e-05,
      "loss": 0.3031,
      "step": 10486
    },
    {
      "epoch": 1.64167188478397,
      "grad_norm": 3.0899922847747803,
      "learning_rate": 5.625610948191593e-05,
      "loss": 1.0541,
      "step": 10487
    },
    {
      "epoch": 1.6418284283030684,
      "grad_norm": 2.5007576942443848,
      "learning_rate": 5.6231671554252194e-05,
      "loss": 1.1517,
      "step": 10488
    },
    {
      "epoch": 1.6419849718221666,
      "grad_norm": 0.45454585552215576,
      "learning_rate": 5.6207233626588465e-05,
      "loss": 0.1584,
      "step": 10489
    },
    {
      "epoch": 1.6421415153412648,
      "grad_norm": 0.7792069911956787,
      "learning_rate": 5.618279569892472e-05,
      "loss": 0.2598,
      "step": 10490
    },
    {
      "epoch": 1.642298058860363,
      "grad_norm": 1.4265599250793457,
      "learning_rate": 5.6158357771260995e-05,
      "loss": 0.2881,
      "step": 10491
    },
    {
      "epoch": 1.6424546023794615,
      "grad_norm": 1.610418438911438,
      "learning_rate": 5.613391984359726e-05,
      "loss": 0.2933,
      "step": 10492
    },
    {
      "epoch": 1.64261114589856,
      "grad_norm": 0.5651922225952148,
      "learning_rate": 5.6109481915933524e-05,
      "loss": 0.2019,
      "step": 10493
    },
    {
      "epoch": 1.6427676894176582,
      "grad_norm": 0.757182240486145,
      "learning_rate": 5.608504398826979e-05,
      "loss": 0.1858,
      "step": 10494
    },
    {
      "epoch": 1.6429242329367564,
      "grad_norm": 0.4835294783115387,
      "learning_rate": 5.606060606060606e-05,
      "loss": 0.1283,
      "step": 10495
    },
    {
      "epoch": 1.6430807764558546,
      "grad_norm": 0.6997655034065247,
      "learning_rate": 5.603616813294232e-05,
      "loss": 0.224,
      "step": 10496
    },
    {
      "epoch": 1.643237319974953,
      "grad_norm": 1.8372862339019775,
      "learning_rate": 5.601173020527859e-05,
      "loss": 0.4684,
      "step": 10497
    },
    {
      "epoch": 1.6433938634940515,
      "grad_norm": 1.1879760026931763,
      "learning_rate": 5.5987292277614855e-05,
      "loss": 0.2866,
      "step": 10498
    },
    {
      "epoch": 1.6435504070131497,
      "grad_norm": 0.4589191973209381,
      "learning_rate": 5.596285434995112e-05,
      "loss": 0.2133,
      "step": 10499
    },
    {
      "epoch": 1.643706950532248,
      "grad_norm": 0.8576350212097168,
      "learning_rate": 5.5938416422287385e-05,
      "loss": 0.4608,
      "step": 10500
    },
    {
      "epoch": 1.6438634940513461,
      "grad_norm": 0.8671181201934814,
      "learning_rate": 5.5913978494623656e-05,
      "loss": 0.2109,
      "step": 10501
    },
    {
      "epoch": 1.6440200375704446,
      "grad_norm": 1.038217306137085,
      "learning_rate": 5.5889540566959914e-05,
      "loss": 0.3122,
      "step": 10502
    },
    {
      "epoch": 1.644176581089543,
      "grad_norm": 2.1841461658477783,
      "learning_rate": 5.5865102639296186e-05,
      "loss": 0.5782,
      "step": 10503
    },
    {
      "epoch": 1.6443331246086412,
      "grad_norm": 1.437780737876892,
      "learning_rate": 5.584066471163245e-05,
      "loss": 0.4222,
      "step": 10504
    },
    {
      "epoch": 1.6444896681277394,
      "grad_norm": 1.98343825340271,
      "learning_rate": 5.581622678396871e-05,
      "loss": 0.3952,
      "step": 10505
    },
    {
      "epoch": 1.6446462116468377,
      "grad_norm": 1.1477174758911133,
      "learning_rate": 5.579178885630498e-05,
      "loss": 0.42,
      "step": 10506
    },
    {
      "epoch": 1.644802755165936,
      "grad_norm": 1.4632834196090698,
      "learning_rate": 5.576735092864125e-05,
      "loss": 0.4185,
      "step": 10507
    },
    {
      "epoch": 1.6449592986850345,
      "grad_norm": 1.2096794843673706,
      "learning_rate": 5.574291300097751e-05,
      "loss": 0.3385,
      "step": 10508
    },
    {
      "epoch": 1.6451158422041328,
      "grad_norm": 1.2012349367141724,
      "learning_rate": 5.571847507331378e-05,
      "loss": 0.3774,
      "step": 10509
    },
    {
      "epoch": 1.645272385723231,
      "grad_norm": 1.8027079105377197,
      "learning_rate": 5.5694037145650046e-05,
      "loss": 0.7032,
      "step": 10510
    },
    {
      "epoch": 1.6454289292423294,
      "grad_norm": 2.5102810859680176,
      "learning_rate": 5.5669599217986304e-05,
      "loss": 0.7242,
      "step": 10511
    },
    {
      "epoch": 1.6455854727614276,
      "grad_norm": 2.101835250854492,
      "learning_rate": 5.5645161290322576e-05,
      "loss": 0.5148,
      "step": 10512
    },
    {
      "epoch": 1.645742016280526,
      "grad_norm": 1.309965968132019,
      "learning_rate": 5.562072336265885e-05,
      "loss": 0.3692,
      "step": 10513
    },
    {
      "epoch": 1.6458985597996243,
      "grad_norm": 2.4885432720184326,
      "learning_rate": 5.5596285434995106e-05,
      "loss": 0.6005,
      "step": 10514
    },
    {
      "epoch": 1.6460551033187225,
      "grad_norm": 2.1940715312957764,
      "learning_rate": 5.557184750733138e-05,
      "loss": 0.6397,
      "step": 10515
    },
    {
      "epoch": 1.646211646837821,
      "grad_norm": 4.189990520477295,
      "learning_rate": 5.554740957966764e-05,
      "loss": 0.8497,
      "step": 10516
    },
    {
      "epoch": 1.6463681903569192,
      "grad_norm": 1.0179821252822876,
      "learning_rate": 5.55229716520039e-05,
      "loss": 0.3263,
      "step": 10517
    },
    {
      "epoch": 1.6465247338760176,
      "grad_norm": 2.343964099884033,
      "learning_rate": 5.549853372434017e-05,
      "loss": 0.2546,
      "step": 10518
    },
    {
      "epoch": 1.6466812773951158,
      "grad_norm": 2.5637285709381104,
      "learning_rate": 5.547409579667644e-05,
      "loss": 0.4887,
      "step": 10519
    },
    {
      "epoch": 1.646837820914214,
      "grad_norm": 1.819643497467041,
      "learning_rate": 5.54496578690127e-05,
      "loss": 0.623,
      "step": 10520
    },
    {
      "epoch": 1.6469943644333125,
      "grad_norm": 2.0592472553253174,
      "learning_rate": 5.542521994134897e-05,
      "loss": 0.3924,
      "step": 10521
    },
    {
      "epoch": 1.647150907952411,
      "grad_norm": 1.9924129247665405,
      "learning_rate": 5.540078201368524e-05,
      "loss": 0.4991,
      "step": 10522
    },
    {
      "epoch": 1.6473074514715091,
      "grad_norm": 1.941512942314148,
      "learning_rate": 5.5376344086021496e-05,
      "loss": 0.857,
      "step": 10523
    },
    {
      "epoch": 1.6474639949906074,
      "grad_norm": 1.9502965211868286,
      "learning_rate": 5.535190615835777e-05,
      "loss": 1.019,
      "step": 10524
    },
    {
      "epoch": 1.6476205385097056,
      "grad_norm": 2.0973851680755615,
      "learning_rate": 5.532746823069404e-05,
      "loss": 0.5463,
      "step": 10525
    },
    {
      "epoch": 1.647777082028804,
      "grad_norm": 2.2591187953948975,
      "learning_rate": 5.53030303030303e-05,
      "loss": 0.5572,
      "step": 10526
    },
    {
      "epoch": 1.6479336255479025,
      "grad_norm": 2.8180580139160156,
      "learning_rate": 5.527859237536656e-05,
      "loss": 1.1269,
      "step": 10527
    },
    {
      "epoch": 1.6480901690670007,
      "grad_norm": 3.1859171390533447,
      "learning_rate": 5.525415444770283e-05,
      "loss": 1.1379,
      "step": 10528
    },
    {
      "epoch": 1.648246712586099,
      "grad_norm": 2.108909845352173,
      "learning_rate": 5.522971652003909e-05,
      "loss": 0.3368,
      "step": 10529
    },
    {
      "epoch": 1.6484032561051971,
      "grad_norm": 9.457693099975586,
      "learning_rate": 5.520527859237536e-05,
      "loss": 1.0132,
      "step": 10530
    },
    {
      "epoch": 1.6485597996242956,
      "grad_norm": 4.089739799499512,
      "learning_rate": 5.5180840664711634e-05,
      "loss": 1.6879,
      "step": 10531
    },
    {
      "epoch": 1.648716343143394,
      "grad_norm": 1.7401306629180908,
      "learning_rate": 5.515640273704789e-05,
      "loss": 1.105,
      "step": 10532
    },
    {
      "epoch": 1.6488728866624922,
      "grad_norm": 2.562828540802002,
      "learning_rate": 5.513196480938416e-05,
      "loss": 1.4396,
      "step": 10533
    },
    {
      "epoch": 1.6490294301815904,
      "grad_norm": 3.0045950412750244,
      "learning_rate": 5.510752688172043e-05,
      "loss": 0.8768,
      "step": 10534
    },
    {
      "epoch": 1.6491859737006886,
      "grad_norm": 5.166981220245361,
      "learning_rate": 5.508308895405669e-05,
      "loss": 0.6006,
      "step": 10535
    },
    {
      "epoch": 1.649342517219787,
      "grad_norm": 4.5421929359436035,
      "learning_rate": 5.505865102639296e-05,
      "loss": 0.7751,
      "step": 10536
    },
    {
      "epoch": 1.6494990607388855,
      "grad_norm": 1.448534369468689,
      "learning_rate": 5.503421309872923e-05,
      "loss": 0.5581,
      "step": 10537
    },
    {
      "epoch": 1.6496556042579837,
      "grad_norm": 1.7535744905471802,
      "learning_rate": 5.500977517106549e-05,
      "loss": 0.8802,
      "step": 10538
    },
    {
      "epoch": 1.649812147777082,
      "grad_norm": 0.6873712539672852,
      "learning_rate": 5.498533724340175e-05,
      "loss": 0.1992,
      "step": 10539
    },
    {
      "epoch": 1.6499686912961802,
      "grad_norm": 0.7089383006095886,
      "learning_rate": 5.4960899315738024e-05,
      "loss": 0.195,
      "step": 10540
    },
    {
      "epoch": 1.6501252348152786,
      "grad_norm": 0.3983263671398163,
      "learning_rate": 5.493646138807428e-05,
      "loss": 0.1743,
      "step": 10541
    },
    {
      "epoch": 1.650281778334377,
      "grad_norm": 0.44565579295158386,
      "learning_rate": 5.4912023460410554e-05,
      "loss": 0.1656,
      "step": 10542
    },
    {
      "epoch": 1.6504383218534753,
      "grad_norm": 0.6502916216850281,
      "learning_rate": 5.4887585532746825e-05,
      "loss": 0.245,
      "step": 10543
    },
    {
      "epoch": 1.6505948653725735,
      "grad_norm": 0.8741419315338135,
      "learning_rate": 5.4863147605083083e-05,
      "loss": 0.275,
      "step": 10544
    },
    {
      "epoch": 1.650751408891672,
      "grad_norm": 0.5584045052528381,
      "learning_rate": 5.483870967741935e-05,
      "loss": 0.2486,
      "step": 10545
    },
    {
      "epoch": 1.6509079524107702,
      "grad_norm": 0.7461156845092773,
      "learning_rate": 5.481427174975562e-05,
      "loss": 0.3131,
      "step": 10546
    },
    {
      "epoch": 1.6510644959298686,
      "grad_norm": 0.6761148571968079,
      "learning_rate": 5.478983382209188e-05,
      "loss": 0.1941,
      "step": 10547
    },
    {
      "epoch": 1.6512210394489668,
      "grad_norm": 1.4514672756195068,
      "learning_rate": 5.476539589442815e-05,
      "loss": 0.2663,
      "step": 10548
    },
    {
      "epoch": 1.651377582968065,
      "grad_norm": 0.8751680850982666,
      "learning_rate": 5.474095796676442e-05,
      "loss": 0.2844,
      "step": 10549
    },
    {
      "epoch": 1.6515341264871635,
      "grad_norm": 1.7713919878005981,
      "learning_rate": 5.471652003910068e-05,
      "loss": 0.2972,
      "step": 10550
    },
    {
      "epoch": 1.6516906700062617,
      "grad_norm": 1.0611058473587036,
      "learning_rate": 5.4692082111436944e-05,
      "loss": 0.3436,
      "step": 10551
    },
    {
      "epoch": 1.6518472135253601,
      "grad_norm": 0.7453222274780273,
      "learning_rate": 5.466764418377321e-05,
      "loss": 0.2195,
      "step": 10552
    },
    {
      "epoch": 1.6520037570444583,
      "grad_norm": 2.404334783554077,
      "learning_rate": 5.4643206256109473e-05,
      "loss": 0.3868,
      "step": 10553
    },
    {
      "epoch": 1.6521603005635566,
      "grad_norm": 0.9054138660430908,
      "learning_rate": 5.4618768328445745e-05,
      "loss": 0.21,
      "step": 10554
    },
    {
      "epoch": 1.652316844082655,
      "grad_norm": 2.539930820465088,
      "learning_rate": 5.4594330400782e-05,
      "loss": 0.3469,
      "step": 10555
    },
    {
      "epoch": 1.6524733876017534,
      "grad_norm": 1.4532172679901123,
      "learning_rate": 5.4569892473118275e-05,
      "loss": 0.3711,
      "step": 10556
    },
    {
      "epoch": 1.6526299311208517,
      "grad_norm": 1.0003594160079956,
      "learning_rate": 5.454545454545454e-05,
      "loss": 0.2647,
      "step": 10557
    },
    {
      "epoch": 1.6527864746399499,
      "grad_norm": 1.0202223062515259,
      "learning_rate": 5.4521016617790804e-05,
      "loss": 0.3655,
      "step": 10558
    },
    {
      "epoch": 1.652943018159048,
      "grad_norm": 2.1398379802703857,
      "learning_rate": 5.449657869012707e-05,
      "loss": 0.4682,
      "step": 10559
    },
    {
      "epoch": 1.6530995616781465,
      "grad_norm": 1.0124636888504028,
      "learning_rate": 5.447214076246334e-05,
      "loss": 0.273,
      "step": 10560
    },
    {
      "epoch": 1.653256105197245,
      "grad_norm": 1.2546247243881226,
      "learning_rate": 5.44477028347996e-05,
      "loss": 0.3883,
      "step": 10561
    },
    {
      "epoch": 1.6534126487163432,
      "grad_norm": 1.4561783075332642,
      "learning_rate": 5.442326490713587e-05,
      "loss": 0.397,
      "step": 10562
    },
    {
      "epoch": 1.6535691922354414,
      "grad_norm": 0.9833201169967651,
      "learning_rate": 5.4398826979472135e-05,
      "loss": 0.3122,
      "step": 10563
    },
    {
      "epoch": 1.6537257357545396,
      "grad_norm": 10.254325866699219,
      "learning_rate": 5.43743890518084e-05,
      "loss": 0.4702,
      "step": 10564
    },
    {
      "epoch": 1.653882279273638,
      "grad_norm": 1.9881863594055176,
      "learning_rate": 5.4349951124144665e-05,
      "loss": 0.7521,
      "step": 10565
    },
    {
      "epoch": 1.6540388227927365,
      "grad_norm": 3.919692277908325,
      "learning_rate": 5.4325513196480936e-05,
      "loss": 0.5512,
      "step": 10566
    },
    {
      "epoch": 1.6541953663118347,
      "grad_norm": 2.473768711090088,
      "learning_rate": 5.4301075268817194e-05,
      "loss": 0.4922,
      "step": 10567
    },
    {
      "epoch": 1.654351909830933,
      "grad_norm": 2.2449424266815186,
      "learning_rate": 5.4276637341153466e-05,
      "loss": 0.5504,
      "step": 10568
    },
    {
      "epoch": 1.6545084533500312,
      "grad_norm": 2.9586992263793945,
      "learning_rate": 5.425219941348973e-05,
      "loss": 0.8247,
      "step": 10569
    },
    {
      "epoch": 1.6546649968691296,
      "grad_norm": 1.8599929809570312,
      "learning_rate": 5.4227761485825995e-05,
      "loss": 0.461,
      "step": 10570
    },
    {
      "epoch": 1.654821540388228,
      "grad_norm": 1.9860762357711792,
      "learning_rate": 5.420332355816226e-05,
      "loss": 0.958,
      "step": 10571
    },
    {
      "epoch": 1.6549780839073263,
      "grad_norm": 3.286160707473755,
      "learning_rate": 5.417888563049853e-05,
      "loss": 1.1638,
      "step": 10572
    },
    {
      "epoch": 1.6551346274264245,
      "grad_norm": 1.1726208925247192,
      "learning_rate": 5.415444770283479e-05,
      "loss": 0.4416,
      "step": 10573
    },
    {
      "epoch": 1.6552911709455227,
      "grad_norm": 2.7897346019744873,
      "learning_rate": 5.413000977517106e-05,
      "loss": 0.8697,
      "step": 10574
    },
    {
      "epoch": 1.6554477144646211,
      "grad_norm": 4.233922958374023,
      "learning_rate": 5.4105571847507326e-05,
      "loss": 0.8906,
      "step": 10575
    },
    {
      "epoch": 1.6556042579837196,
      "grad_norm": 1.2990089654922485,
      "learning_rate": 5.408113391984359e-05,
      "loss": 0.6652,
      "step": 10576
    },
    {
      "epoch": 1.6557608015028178,
      "grad_norm": 3.66686749458313,
      "learning_rate": 5.4056695992179856e-05,
      "loss": 1.022,
      "step": 10577
    },
    {
      "epoch": 1.655917345021916,
      "grad_norm": 4.231387615203857,
      "learning_rate": 5.403225806451613e-05,
      "loss": 0.9791,
      "step": 10578
    },
    {
      "epoch": 1.6560738885410144,
      "grad_norm": 2.5381076335906982,
      "learning_rate": 5.4007820136852385e-05,
      "loss": 1.2046,
      "step": 10579
    },
    {
      "epoch": 1.6562304320601127,
      "grad_norm": 3.6664905548095703,
      "learning_rate": 5.398338220918866e-05,
      "loss": 1.5159,
      "step": 10580
    },
    {
      "epoch": 1.656386975579211,
      "grad_norm": 2.639780282974243,
      "learning_rate": 5.395894428152492e-05,
      "loss": 1.4945,
      "step": 10581
    },
    {
      "epoch": 1.6565435190983093,
      "grad_norm": 1.567789912223816,
      "learning_rate": 5.3934506353861187e-05,
      "loss": 0.7653,
      "step": 10582
    },
    {
      "epoch": 1.6567000626174075,
      "grad_norm": 4.89553689956665,
      "learning_rate": 5.391006842619745e-05,
      "loss": 1.0403,
      "step": 10583
    },
    {
      "epoch": 1.656856606136506,
      "grad_norm": NaN,
      "learning_rate": 5.391006842619745e-05,
      "loss": 0.0,
      "step": 10584
    },
    {
      "epoch": 1.6570131496556044,
      "grad_norm": 3.5072648525238037,
      "learning_rate": 5.388563049853372e-05,
      "loss": 0.5282,
      "step": 10585
    },
    {
      "epoch": 1.6571696931747026,
      "grad_norm": 3.3043320178985596,
      "learning_rate": 5.386119257086998e-05,
      "loss": 0.6242,
      "step": 10586
    },
    {
      "epoch": 1.6573262366938009,
      "grad_norm": 2.609529733657837,
      "learning_rate": 5.383675464320625e-05,
      "loss": 1.0082,
      "step": 10587
    },
    {
      "epoch": 1.657482780212899,
      "grad_norm": 1.4475938081741333,
      "learning_rate": 5.381231671554252e-05,
      "loss": 0.9866,
      "step": 10588
    },
    {
      "epoch": 1.6576393237319975,
      "grad_norm": 0.5289618372917175,
      "learning_rate": 5.378787878787878e-05,
      "loss": 0.2334,
      "step": 10589
    },
    {
      "epoch": 1.657795867251096,
      "grad_norm": 0.5513238310813904,
      "learning_rate": 5.376344086021505e-05,
      "loss": 0.2082,
      "step": 10590
    },
    {
      "epoch": 1.6579524107701942,
      "grad_norm": 0.7842631936073303,
      "learning_rate": 5.373900293255132e-05,
      "loss": 0.2386,
      "step": 10591
    },
    {
      "epoch": 1.6581089542892924,
      "grad_norm": 0.7243189811706543,
      "learning_rate": 5.3714565004887577e-05,
      "loss": 0.1653,
      "step": 10592
    },
    {
      "epoch": 1.6582654978083906,
      "grad_norm": 2.165306329727173,
      "learning_rate": 5.369012707722385e-05,
      "loss": 0.208,
      "step": 10593
    },
    {
      "epoch": 1.658422041327489,
      "grad_norm": 0.661629319190979,
      "learning_rate": 5.366568914956011e-05,
      "loss": 0.2278,
      "step": 10594
    },
    {
      "epoch": 1.6585785848465875,
      "grad_norm": 0.931580662727356,
      "learning_rate": 5.364125122189638e-05,
      "loss": 0.3416,
      "step": 10595
    },
    {
      "epoch": 1.6587351283656857,
      "grad_norm": 0.9427894949913025,
      "learning_rate": 5.361681329423264e-05,
      "loss": 0.3323,
      "step": 10596
    },
    {
      "epoch": 1.658891671884784,
      "grad_norm": 0.6458982229232788,
      "learning_rate": 5.3592375366568914e-05,
      "loss": 0.2057,
      "step": 10597
    },
    {
      "epoch": 1.6590482154038821,
      "grad_norm": 1.2524142265319824,
      "learning_rate": 5.356793743890517e-05,
      "loss": 0.3066,
      "step": 10598
    },
    {
      "epoch": 1.6592047589229806,
      "grad_norm": 1.944617509841919,
      "learning_rate": 5.3543499511241444e-05,
      "loss": 0.4275,
      "step": 10599
    },
    {
      "epoch": 1.659361302442079,
      "grad_norm": 3.7578859329223633,
      "learning_rate": 5.351906158357771e-05,
      "loss": 0.1748,
      "step": 10600
    },
    {
      "epoch": 1.6595178459611772,
      "grad_norm": 0.9285878539085388,
      "learning_rate": 5.349462365591397e-05,
      "loss": 0.2722,
      "step": 10601
    },
    {
      "epoch": 1.6596743894802755,
      "grad_norm": 3.483341693878174,
      "learning_rate": 5.347018572825024e-05,
      "loss": 0.295,
      "step": 10602
    },
    {
      "epoch": 1.6598309329993737,
      "grad_norm": 1.6603342294692993,
      "learning_rate": 5.344574780058651e-05,
      "loss": 0.2591,
      "step": 10603
    },
    {
      "epoch": 1.6599874765184721,
      "grad_norm": 2.1465036869049072,
      "learning_rate": 5.342130987292277e-05,
      "loss": 0.3515,
      "step": 10604
    },
    {
      "epoch": 1.6601440200375706,
      "grad_norm": 1.9516934156417847,
      "learning_rate": 5.339687194525904e-05,
      "loss": 0.4868,
      "step": 10605
    },
    {
      "epoch": 1.6603005635566688,
      "grad_norm": 0.8558928966522217,
      "learning_rate": 5.3372434017595304e-05,
      "loss": 0.2049,
      "step": 10606
    },
    {
      "epoch": 1.660457107075767,
      "grad_norm": 1.204277515411377,
      "learning_rate": 5.334799608993157e-05,
      "loss": 0.2129,
      "step": 10607
    },
    {
      "epoch": 1.6606136505948652,
      "grad_norm": 1.400564193725586,
      "learning_rate": 5.3323558162267834e-05,
      "loss": 0.2187,
      "step": 10608
    },
    {
      "epoch": 1.6607701941139636,
      "grad_norm": 1.5953516960144043,
      "learning_rate": 5.3299120234604105e-05,
      "loss": 0.3519,
      "step": 10609
    },
    {
      "epoch": 1.660926737633062,
      "grad_norm": 1.5663790702819824,
      "learning_rate": 5.327468230694036e-05,
      "loss": 0.4418,
      "step": 10610
    },
    {
      "epoch": 1.6610832811521603,
      "grad_norm": 2.3011529445648193,
      "learning_rate": 5.3250244379276635e-05,
      "loss": 0.4294,
      "step": 10611
    },
    {
      "epoch": 1.6612398246712585,
      "grad_norm": 2.8383572101593018,
      "learning_rate": 5.32258064516129e-05,
      "loss": 0.6262,
      "step": 10612
    },
    {
      "epoch": 1.661396368190357,
      "grad_norm": 2.6488254070281982,
      "learning_rate": 5.3201368523949164e-05,
      "loss": 0.5056,
      "step": 10613
    },
    {
      "epoch": 1.6615529117094552,
      "grad_norm": 0.9980300068855286,
      "learning_rate": 5.317693059628543e-05,
      "loss": 0.3669,
      "step": 10614
    },
    {
      "epoch": 1.6617094552285536,
      "grad_norm": 1.887178897857666,
      "learning_rate": 5.31524926686217e-05,
      "loss": 0.5514,
      "step": 10615
    },
    {
      "epoch": 1.6618659987476518,
      "grad_norm": 2.6637489795684814,
      "learning_rate": 5.312805474095796e-05,
      "loss": 0.8874,
      "step": 10616
    },
    {
      "epoch": 1.66202254226675,
      "grad_norm": 2.64481520652771,
      "learning_rate": 5.310361681329423e-05,
      "loss": 0.5448,
      "step": 10617
    },
    {
      "epoch": 1.6621790857858485,
      "grad_norm": 2.7849998474121094,
      "learning_rate": 5.3079178885630495e-05,
      "loss": 0.9028,
      "step": 10618
    },
    {
      "epoch": 1.662335629304947,
      "grad_norm": 3.2674617767333984,
      "learning_rate": 5.305474095796676e-05,
      "loss": 0.9225,
      "step": 10619
    },
    {
      "epoch": 1.6624921728240452,
      "grad_norm": 2.544890880584717,
      "learning_rate": 5.3030303030303025e-05,
      "loss": 0.9553,
      "step": 10620
    },
    {
      "epoch": 1.6626487163431434,
      "grad_norm": 1.3422220945358276,
      "learning_rate": 5.3005865102639296e-05,
      "loss": 0.3589,
      "step": 10621
    },
    {
      "epoch": 1.6628052598622416,
      "grad_norm": 1.9164516925811768,
      "learning_rate": 5.2981427174975554e-05,
      "loss": 0.6147,
      "step": 10622
    },
    {
      "epoch": 1.66296180338134,
      "grad_norm": 5.187346458435059,
      "learning_rate": 5.2956989247311826e-05,
      "loss": 0.6961,
      "step": 10623
    },
    {
      "epoch": 1.6631183469004385,
      "grad_norm": 2.5986785888671875,
      "learning_rate": 5.293255131964809e-05,
      "loss": 0.8906,
      "step": 10624
    },
    {
      "epoch": 1.6632748904195367,
      "grad_norm": 3.034003734588623,
      "learning_rate": 5.2908113391984356e-05,
      "loss": 1.0949,
      "step": 10625
    },
    {
      "epoch": 1.663431433938635,
      "grad_norm": 2.1828017234802246,
      "learning_rate": 5.288367546432062e-05,
      "loss": 0.5176,
      "step": 10626
    },
    {
      "epoch": 1.6635879774577331,
      "grad_norm": 2.277587890625,
      "learning_rate": 5.285923753665689e-05,
      "loss": 0.899,
      "step": 10627
    },
    {
      "epoch": 1.6637445209768316,
      "grad_norm": 3.5474190711975098,
      "learning_rate": 5.283479960899315e-05,
      "loss": 0.6992,
      "step": 10628
    },
    {
      "epoch": 1.66390106449593,
      "grad_norm": 4.0805888175964355,
      "learning_rate": 5.281036168132942e-05,
      "loss": 1.1746,
      "step": 10629
    },
    {
      "epoch": 1.6640576080150282,
      "grad_norm": 3.748154640197754,
      "learning_rate": 5.2785923753665686e-05,
      "loss": 0.922,
      "step": 10630
    },
    {
      "epoch": 1.6642141515341264,
      "grad_norm": 4.198906898498535,
      "learning_rate": 5.276148582600195e-05,
      "loss": 1.751,
      "step": 10631
    },
    {
      "epoch": 1.6643706950532247,
      "grad_norm": 4.466783046722412,
      "learning_rate": 5.2737047898338216e-05,
      "loss": 1.1014,
      "step": 10632
    },
    {
      "epoch": 1.664527238572323,
      "grad_norm": 1.4127699136734009,
      "learning_rate": 5.271260997067449e-05,
      "loss": 0.8603,
      "step": 10633
    },
    {
      "epoch": 1.6646837820914215,
      "grad_norm": 2.8340721130371094,
      "learning_rate": 5.2688172043010746e-05,
      "loss": 1.191,
      "step": 10634
    },
    {
      "epoch": 1.6648403256105198,
      "grad_norm": 2.287118673324585,
      "learning_rate": 5.266373411534702e-05,
      "loss": 0.5875,
      "step": 10635
    },
    {
      "epoch": 1.664996869129618,
      "grad_norm": 2.4103381633758545,
      "learning_rate": 5.263929618768328e-05,
      "loss": 0.7849,
      "step": 10636
    },
    {
      "epoch": 1.6651534126487162,
      "grad_norm": 2.8598053455352783,
      "learning_rate": 5.261485826001955e-05,
      "loss": 0.8766,
      "step": 10637
    },
    {
      "epoch": 1.6653099561678146,
      "grad_norm": 3.413485288619995,
      "learning_rate": 5.259042033235581e-05,
      "loss": 1.239,
      "step": 10638
    },
    {
      "epoch": 1.665466499686913,
      "grad_norm": 1.0679550170898438,
      "learning_rate": 5.256598240469208e-05,
      "loss": 0.6029,
      "step": 10639
    },
    {
      "epoch": 1.6656230432060113,
      "grad_norm": 0.44509389996528625,
      "learning_rate": 5.254154447702834e-05,
      "loss": 0.1905,
      "step": 10640
    },
    {
      "epoch": 1.6657795867251095,
      "grad_norm": 1.0144532918930054,
      "learning_rate": 5.251710654936461e-05,
      "loss": 0.269,
      "step": 10641
    },
    {
      "epoch": 1.6659361302442077,
      "grad_norm": 1.4316686391830444,
      "learning_rate": 5.249266862170088e-05,
      "loss": 0.2504,
      "step": 10642
    },
    {
      "epoch": 1.6660926737633062,
      "grad_norm": 0.9818958044052124,
      "learning_rate": 5.2468230694037136e-05,
      "loss": 0.3228,
      "step": 10643
    },
    {
      "epoch": 1.6662492172824046,
      "grad_norm": 1.124721884727478,
      "learning_rate": 5.244379276637341e-05,
      "loss": 0.3437,
      "step": 10644
    },
    {
      "epoch": 1.6664057608015028,
      "grad_norm": 0.6059370040893555,
      "learning_rate": 5.241935483870968e-05,
      "loss": 0.1469,
      "step": 10645
    },
    {
      "epoch": 1.666562304320601,
      "grad_norm": 1.5121008157730103,
      "learning_rate": 5.239491691104594e-05,
      "loss": 0.2796,
      "step": 10646
    },
    {
      "epoch": 1.6667188478396995,
      "grad_norm": 0.9613409638404846,
      "learning_rate": 5.237047898338221e-05,
      "loss": 0.2226,
      "step": 10647
    },
    {
      "epoch": 1.6668753913587977,
      "grad_norm": 1.1612027883529663,
      "learning_rate": 5.234604105571847e-05,
      "loss": 0.3174,
      "step": 10648
    },
    {
      "epoch": 1.6670319348778961,
      "grad_norm": 2.4976742267608643,
      "learning_rate": 5.232160312805473e-05,
      "loss": 0.3316,
      "step": 10649
    },
    {
      "epoch": 1.6671884783969944,
      "grad_norm": 1.2498027086257935,
      "learning_rate": 5.2297165200391e-05,
      "loss": 0.4277,
      "step": 10650
    },
    {
      "epoch": 1.6673450219160926,
      "grad_norm": 1.24776291847229,
      "learning_rate": 5.2272727272727274e-05,
      "loss": 0.2936,
      "step": 10651
    },
    {
      "epoch": 1.667501565435191,
      "grad_norm": 1.3939722776412964,
      "learning_rate": 5.224828934506353e-05,
      "loss": 0.4268,
      "step": 10652
    },
    {
      "epoch": 1.6676581089542895,
      "grad_norm": 1.675716519355774,
      "learning_rate": 5.2223851417399804e-05,
      "loss": 0.4775,
      "step": 10653
    },
    {
      "epoch": 1.6678146524733877,
      "grad_norm": 1.3907341957092285,
      "learning_rate": 5.219941348973607e-05,
      "loss": 0.3869,
      "step": 10654
    },
    {
      "epoch": 1.6679711959924859,
      "grad_norm": 1.2454923391342163,
      "learning_rate": 5.217497556207233e-05,
      "loss": 0.3018,
      "step": 10655
    },
    {
      "epoch": 1.668127739511584,
      "grad_norm": 3.0145490169525146,
      "learning_rate": 5.21505376344086e-05,
      "loss": 0.3795,
      "step": 10656
    },
    {
      "epoch": 1.6682842830306825,
      "grad_norm": 1.352417230606079,
      "learning_rate": 5.212609970674487e-05,
      "loss": 0.3482,
      "step": 10657
    },
    {
      "epoch": 1.668440826549781,
      "grad_norm": 1.3013248443603516,
      "learning_rate": 5.210166177908113e-05,
      "loss": 0.5697,
      "step": 10658
    },
    {
      "epoch": 1.6685973700688792,
      "grad_norm": 3.581904888153076,
      "learning_rate": 5.20772238514174e-05,
      "loss": 0.8786,
      "step": 10659
    },
    {
      "epoch": 1.6687539135879774,
      "grad_norm": 1.4342955350875854,
      "learning_rate": 5.2052785923753664e-05,
      "loss": 0.4929,
      "step": 10660
    },
    {
      "epoch": 1.6689104571070756,
      "grad_norm": 1.4663060903549194,
      "learning_rate": 5.202834799608992e-05,
      "loss": 0.4167,
      "step": 10661
    },
    {
      "epoch": 1.669067000626174,
      "grad_norm": 1.2736845016479492,
      "learning_rate": 5.2003910068426194e-05,
      "loss": 0.44,
      "step": 10662
    },
    {
      "epoch": 1.6692235441452725,
      "grad_norm": 3.1659364700317383,
      "learning_rate": 5.1979472140762465e-05,
      "loss": 0.406,
      "step": 10663
    },
    {
      "epoch": 1.6693800876643707,
      "grad_norm": 4.501870155334473,
      "learning_rate": 5.1955034213098723e-05,
      "loss": 1.0747,
      "step": 10664
    },
    {
      "epoch": 1.669536631183469,
      "grad_norm": 2.149975299835205,
      "learning_rate": 5.193059628543499e-05,
      "loss": 0.5949,
      "step": 10665
    },
    {
      "epoch": 1.6696931747025672,
      "grad_norm": 1.4252465963363647,
      "learning_rate": 5.190615835777126e-05,
      "loss": 0.4715,
      "step": 10666
    },
    {
      "epoch": 1.6698497182216656,
      "grad_norm": 1.7069036960601807,
      "learning_rate": 5.188172043010752e-05,
      "loss": 0.3242,
      "step": 10667
    },
    {
      "epoch": 1.670006261740764,
      "grad_norm": 2.188441753387451,
      "learning_rate": 5.185728250244379e-05,
      "loss": 0.4015,
      "step": 10668
    },
    {
      "epoch": 1.6701628052598623,
      "grad_norm": 1.5735689401626587,
      "learning_rate": 5.183284457478006e-05,
      "loss": 0.4151,
      "step": 10669
    },
    {
      "epoch": 1.6703193487789605,
      "grad_norm": 4.588582515716553,
      "learning_rate": 5.180840664711632e-05,
      "loss": 0.6561,
      "step": 10670
    },
    {
      "epoch": 1.6704758922980587,
      "grad_norm": 2.366029739379883,
      "learning_rate": 5.1783968719452584e-05,
      "loss": 0.48,
      "step": 10671
    },
    {
      "epoch": 1.6706324358171571,
      "grad_norm": 1.6738721132278442,
      "learning_rate": 5.1759530791788855e-05,
      "loss": 0.5077,
      "step": 10672
    },
    {
      "epoch": 1.6707889793362556,
      "grad_norm": 2.501746892929077,
      "learning_rate": 5.1735092864125113e-05,
      "loss": 0.6517,
      "step": 10673
    },
    {
      "epoch": 1.6709455228553538,
      "grad_norm": 3.7587740421295166,
      "learning_rate": 5.1710654936461385e-05,
      "loss": 1.1346,
      "step": 10674
    },
    {
      "epoch": 1.671102066374452,
      "grad_norm": 15.480533599853516,
      "learning_rate": 5.168621700879766e-05,
      "loss": 0.6876,
      "step": 10675
    },
    {
      "epoch": 1.6712586098935505,
      "grad_norm": 2.9390463829040527,
      "learning_rate": 5.1661779081133915e-05,
      "loss": 0.6573,
      "step": 10676
    },
    {
      "epoch": 1.6714151534126487,
      "grad_norm": 5.4329304695129395,
      "learning_rate": 5.163734115347018e-05,
      "loss": 0.788,
      "step": 10677
    },
    {
      "epoch": 1.6715716969317471,
      "grad_norm": 9.223847389221191,
      "learning_rate": 5.161290322580645e-05,
      "loss": 1.1849,
      "step": 10678
    },
    {
      "epoch": 1.6717282404508453,
      "grad_norm": 2.515810966491699,
      "learning_rate": 5.158846529814271e-05,
      "loss": 0.621,
      "step": 10679
    },
    {
      "epoch": 1.6718847839699436,
      "grad_norm": 3.640105962753296,
      "learning_rate": 5.156402737047898e-05,
      "loss": 1.1932,
      "step": 10680
    },
    {
      "epoch": 1.672041327489042,
      "grad_norm": 5.372037887573242,
      "learning_rate": 5.153958944281524e-05,
      "loss": 0.586,
      "step": 10681
    },
    {
      "epoch": 1.6721978710081402,
      "grad_norm": 1.5914926528930664,
      "learning_rate": 5.151515151515151e-05,
      "loss": 1.0401,
      "step": 10682
    },
    {
      "epoch": 1.6723544145272387,
      "grad_norm": 3.534508466720581,
      "learning_rate": 5.1490713587487775e-05,
      "loss": 1.3516,
      "step": 10683
    },
    {
      "epoch": 1.6725109580463369,
      "grad_norm": 1.6528741121292114,
      "learning_rate": 5.146627565982404e-05,
      "loss": 0.6955,
      "step": 10684
    },
    {
      "epoch": 1.672667501565435,
      "grad_norm": 2.1529715061187744,
      "learning_rate": 5.1441837732160305e-05,
      "loss": 0.8107,
      "step": 10685
    },
    {
      "epoch": 1.6728240450845335,
      "grad_norm": 2.8191683292388916,
      "learning_rate": 5.1417399804496576e-05,
      "loss": 0.3083,
      "step": 10686
    },
    {
      "epoch": 1.672980588603632,
      "grad_norm": 1.5679564476013184,
      "learning_rate": 5.1392961876832834e-05,
      "loss": 0.6602,
      "step": 10687
    },
    {
      "epoch": 1.6731371321227302,
      "grad_norm": 1.8020987510681152,
      "learning_rate": 5.1368523949169106e-05,
      "loss": 0.7978,
      "step": 10688
    },
    {
      "epoch": 1.6732936756418284,
      "grad_norm": 0.6797225475311279,
      "learning_rate": 5.134408602150537e-05,
      "loss": 0.1784,
      "step": 10689
    },
    {
      "epoch": 1.6734502191609266,
      "grad_norm": 1.2861469984054565,
      "learning_rate": 5.1319648093841635e-05,
      "loss": 0.1494,
      "step": 10690
    },
    {
      "epoch": 1.673606762680025,
      "grad_norm": 0.7776259183883667,
      "learning_rate": 5.12952101661779e-05,
      "loss": 0.2241,
      "step": 10691
    },
    {
      "epoch": 1.6737633061991235,
      "grad_norm": 0.7680147290229797,
      "learning_rate": 5.127077223851417e-05,
      "loss": 0.1937,
      "step": 10692
    },
    {
      "epoch": 1.6739198497182217,
      "grad_norm": 0.7081788182258606,
      "learning_rate": 5.124633431085043e-05,
      "loss": 0.1732,
      "step": 10693
    },
    {
      "epoch": 1.67407639323732,
      "grad_norm": 0.6900548338890076,
      "learning_rate": 5.12218963831867e-05,
      "loss": 0.2248,
      "step": 10694
    },
    {
      "epoch": 1.6742329367564182,
      "grad_norm": 1.274192452430725,
      "learning_rate": 5.1197458455522966e-05,
      "loss": 0.3333,
      "step": 10695
    },
    {
      "epoch": 1.6743894802755166,
      "grad_norm": 0.7436234354972839,
      "learning_rate": 5.117302052785923e-05,
      "loss": 0.3009,
      "step": 10696
    },
    {
      "epoch": 1.674546023794615,
      "grad_norm": 1.424438714981079,
      "learning_rate": 5.1148582600195496e-05,
      "loss": 0.3789,
      "step": 10697
    },
    {
      "epoch": 1.6747025673137133,
      "grad_norm": 0.85284423828125,
      "learning_rate": 5.112414467253177e-05,
      "loss": 0.3426,
      "step": 10698
    },
    {
      "epoch": 1.6748591108328115,
      "grad_norm": 1.165490984916687,
      "learning_rate": 5.1099706744868025e-05,
      "loss": 0.2127,
      "step": 10699
    },
    {
      "epoch": 1.6750156543519097,
      "grad_norm": 1.005594253540039,
      "learning_rate": 5.10752688172043e-05,
      "loss": 0.3583,
      "step": 10700
    },
    {
      "epoch": 1.6751721978710081,
      "grad_norm": 2.8059756755828857,
      "learning_rate": 5.105083088954056e-05,
      "loss": 0.3293,
      "step": 10701
    },
    {
      "epoch": 1.6753287413901066,
      "grad_norm": 0.9310165643692017,
      "learning_rate": 5.1026392961876827e-05,
      "loss": 0.2471,
      "step": 10702
    },
    {
      "epoch": 1.6754852849092048,
      "grad_norm": 1.2037992477416992,
      "learning_rate": 5.100195503421309e-05,
      "loss": 0.5662,
      "step": 10703
    },
    {
      "epoch": 1.675641828428303,
      "grad_norm": 2.170868396759033,
      "learning_rate": 5.097751710654936e-05,
      "loss": 0.3375,
      "step": 10704
    },
    {
      "epoch": 1.6757983719474012,
      "grad_norm": 1.1550534963607788,
      "learning_rate": 5.095307917888562e-05,
      "loss": 0.3144,
      "step": 10705
    },
    {
      "epoch": 1.6759549154664997,
      "grad_norm": 2.387791633605957,
      "learning_rate": 5.092864125122189e-05,
      "loss": 0.3797,
      "step": 10706
    },
    {
      "epoch": 1.676111458985598,
      "grad_norm": 1.3175456523895264,
      "learning_rate": 5.090420332355816e-05,
      "loss": 0.2326,
      "step": 10707
    },
    {
      "epoch": 1.6762680025046963,
      "grad_norm": 1.391885757446289,
      "learning_rate": 5.087976539589442e-05,
      "loss": 0.3776,
      "step": 10708
    },
    {
      "epoch": 1.6764245460237945,
      "grad_norm": 3.515990734100342,
      "learning_rate": 5.085532746823069e-05,
      "loss": 0.254,
      "step": 10709
    },
    {
      "epoch": 1.676581089542893,
      "grad_norm": 2.301178455352783,
      "learning_rate": 5.083088954056696e-05,
      "loss": 0.3833,
      "step": 10710
    },
    {
      "epoch": 1.6767376330619912,
      "grad_norm": 0.8507611751556396,
      "learning_rate": 5.0806451612903217e-05,
      "loss": 0.3777,
      "step": 10711
    },
    {
      "epoch": 1.6768941765810896,
      "grad_norm": 1.7067499160766602,
      "learning_rate": 5.078201368523949e-05,
      "loss": 0.4903,
      "step": 10712
    },
    {
      "epoch": 1.6770507201001879,
      "grad_norm": 4.127223014831543,
      "learning_rate": 5.075757575757575e-05,
      "loss": 0.7006,
      "step": 10713
    },
    {
      "epoch": 1.677207263619286,
      "grad_norm": 1.4126865863800049,
      "learning_rate": 5.073313782991202e-05,
      "loss": 0.3373,
      "step": 10714
    },
    {
      "epoch": 1.6773638071383845,
      "grad_norm": 1.455224871635437,
      "learning_rate": 5.070869990224828e-05,
      "loss": 0.9498,
      "step": 10715
    },
    {
      "epoch": 1.6775203506574827,
      "grad_norm": 4.076616287231445,
      "learning_rate": 5.0684261974584554e-05,
      "loss": 0.8621,
      "step": 10716
    },
    {
      "epoch": 1.6776768941765812,
      "grad_norm": 2.2663486003875732,
      "learning_rate": 5.065982404692081e-05,
      "loss": 0.461,
      "step": 10717
    },
    {
      "epoch": 1.6778334376956794,
      "grad_norm": 2.508690118789673,
      "learning_rate": 5.0635386119257084e-05,
      "loss": 0.7204,
      "step": 10718
    },
    {
      "epoch": 1.6779899812147776,
      "grad_norm": 1.673777461051941,
      "learning_rate": 5.061094819159335e-05,
      "loss": 0.6898,
      "step": 10719
    },
    {
      "epoch": 1.678146524733876,
      "grad_norm": 1.676172137260437,
      "learning_rate": 5.058651026392961e-05,
      "loss": 0.5936,
      "step": 10720
    },
    {
      "epoch": 1.6783030682529745,
      "grad_norm": 7.18748664855957,
      "learning_rate": 5.056207233626588e-05,
      "loss": 1.0243,
      "step": 10721
    },
    {
      "epoch": 1.6784596117720727,
      "grad_norm": 2.1801810264587402,
      "learning_rate": 5.053763440860215e-05,
      "loss": 0.5048,
      "step": 10722
    },
    {
      "epoch": 1.678616155291171,
      "grad_norm": 2.0111351013183594,
      "learning_rate": 5.051319648093841e-05,
      "loss": 0.8976,
      "step": 10723
    },
    {
      "epoch": 1.6787726988102691,
      "grad_norm": 5.509698867797852,
      "learning_rate": 5.048875855327468e-05,
      "loss": 1.0209,
      "step": 10724
    },
    {
      "epoch": 1.6789292423293676,
      "grad_norm": 2.151179552078247,
      "learning_rate": 5.0464320625610944e-05,
      "loss": 1.3053,
      "step": 10725
    },
    {
      "epoch": 1.679085785848466,
      "grad_norm": 3.7033746242523193,
      "learning_rate": 5.043988269794721e-05,
      "loss": 1.1283,
      "step": 10726
    },
    {
      "epoch": 1.6792423293675642,
      "grad_norm": 2.599567174911499,
      "learning_rate": 5.0415444770283474e-05,
      "loss": 0.6676,
      "step": 10727
    },
    {
      "epoch": 1.6793988728866625,
      "grad_norm": 2.8762729167938232,
      "learning_rate": 5.0391006842619745e-05,
      "loss": 0.5331,
      "step": 10728
    },
    {
      "epoch": 1.6795554164057607,
      "grad_norm": 4.5754265785217285,
      "learning_rate": 5.0366568914956e-05,
      "loss": 0.6954,
      "step": 10729
    },
    {
      "epoch": 1.679711959924859,
      "grad_norm": 4.394883155822754,
      "learning_rate": 5.0342130987292275e-05,
      "loss": 0.8285,
      "step": 10730
    },
    {
      "epoch": 1.6798685034439576,
      "grad_norm": 3.905444383621216,
      "learning_rate": 5.031769305962854e-05,
      "loss": 1.6435,
      "step": 10731
    },
    {
      "epoch": 1.6800250469630558,
      "grad_norm": 2.222259521484375,
      "learning_rate": 5.0293255131964804e-05,
      "loss": 0.625,
      "step": 10732
    },
    {
      "epoch": 1.680181590482154,
      "grad_norm": 2.369340658187866,
      "learning_rate": 5.026881720430107e-05,
      "loss": 1.2802,
      "step": 10733
    },
    {
      "epoch": 1.6803381340012522,
      "grad_norm": 8.684860229492188,
      "learning_rate": 5.024437927663734e-05,
      "loss": 0.7457,
      "step": 10734
    },
    {
      "epoch": 1.6804946775203506,
      "grad_norm": 0.9671575427055359,
      "learning_rate": 5.02199413489736e-05,
      "loss": 0.3532,
      "step": 10735
    },
    {
      "epoch": 1.680651221039449,
      "grad_norm": 3.1766295433044434,
      "learning_rate": 5.019550342130987e-05,
      "loss": 1.1237,
      "step": 10736
    },
    {
      "epoch": 1.6808077645585473,
      "grad_norm": 4.267437934875488,
      "learning_rate": 5.0171065493646135e-05,
      "loss": 0.5706,
      "step": 10737
    },
    {
      "epoch": 1.6809643080776455,
      "grad_norm": 3.1225411891937256,
      "learning_rate": 5.01466275659824e-05,
      "loss": 0.8392,
      "step": 10738
    },
    {
      "epoch": 1.6811208515967437,
      "grad_norm": 1.9683570861816406,
      "learning_rate": 5.0122189638318665e-05,
      "loss": 0.2465,
      "step": 10739
    },
    {
      "epoch": 1.6812773951158422,
      "grad_norm": 0.6025456190109253,
      "learning_rate": 5.0097751710654936e-05,
      "loss": 0.2046,
      "step": 10740
    },
    {
      "epoch": 1.6814339386349406,
      "grad_norm": 0.7359471321105957,
      "learning_rate": 5.0073313782991194e-05,
      "loss": 0.1829,
      "step": 10741
    },
    {
      "epoch": 1.6815904821540388,
      "grad_norm": 0.568925142288208,
      "learning_rate": 5.0048875855327466e-05,
      "loss": 0.2044,
      "step": 10742
    },
    {
      "epoch": 1.681747025673137,
      "grad_norm": 0.593213677406311,
      "learning_rate": 5.002443792766373e-05,
      "loss": 0.3406,
      "step": 10743
    },
    {
      "epoch": 1.6819035691922355,
      "grad_norm": 0.6833570599555969,
      "learning_rate": 4.9999999999999996e-05,
      "loss": 0.2111,
      "step": 10744
    },
    {
      "epoch": 1.6820601127113337,
      "grad_norm": 1.2374072074890137,
      "learning_rate": 4.997556207233626e-05,
      "loss": 0.3462,
      "step": 10745
    },
    {
      "epoch": 1.6822166562304322,
      "grad_norm": 0.7850791811943054,
      "learning_rate": 4.995112414467253e-05,
      "loss": 0.3209,
      "step": 10746
    },
    {
      "epoch": 1.6823731997495304,
      "grad_norm": 0.6952258944511414,
      "learning_rate": 4.992668621700879e-05,
      "loss": 0.3351,
      "step": 10747
    },
    {
      "epoch": 1.6825297432686286,
      "grad_norm": 1.0042572021484375,
      "learning_rate": 4.990224828934506e-05,
      "loss": 0.2355,
      "step": 10748
    },
    {
      "epoch": 1.682686286787727,
      "grad_norm": 1.752544641494751,
      "learning_rate": 4.9877810361681326e-05,
      "loss": 0.2027,
      "step": 10749
    },
    {
      "epoch": 1.6828428303068252,
      "grad_norm": 1.0626980066299438,
      "learning_rate": 4.985337243401759e-05,
      "loss": 0.2959,
      "step": 10750
    },
    {
      "epoch": 1.6829993738259237,
      "grad_norm": 1.8501675128936768,
      "learning_rate": 4.9828934506353856e-05,
      "loss": 0.3076,
      "step": 10751
    },
    {
      "epoch": 1.683155917345022,
      "grad_norm": 1.9613275527954102,
      "learning_rate": 4.980449657869013e-05,
      "loss": 0.4469,
      "step": 10752
    },
    {
      "epoch": 1.6833124608641201,
      "grad_norm": 0.9938980340957642,
      "learning_rate": 4.9780058651026386e-05,
      "loss": 0.3296,
      "step": 10753
    },
    {
      "epoch": 1.6834690043832186,
      "grad_norm": 3.928609609603882,
      "learning_rate": 4.975562072336266e-05,
      "loss": 0.3944,
      "step": 10754
    },
    {
      "epoch": 1.683625547902317,
      "grad_norm": 1.8537261486053467,
      "learning_rate": 4.973118279569892e-05,
      "loss": 0.4749,
      "step": 10755
    },
    {
      "epoch": 1.6837820914214152,
      "grad_norm": 1.3114187717437744,
      "learning_rate": 4.970674486803519e-05,
      "loss": 0.3248,
      "step": 10756
    },
    {
      "epoch": 1.6839386349405134,
      "grad_norm": 2.2068116664886475,
      "learning_rate": 4.968230694037145e-05,
      "loss": 0.3514,
      "step": 10757
    },
    {
      "epoch": 1.6840951784596117,
      "grad_norm": 1.8806259632110596,
      "learning_rate": 4.965786901270772e-05,
      "loss": 0.4841,
      "step": 10758
    },
    {
      "epoch": 1.68425172197871,
      "grad_norm": 2.068279504776001,
      "learning_rate": 4.963343108504398e-05,
      "loss": 0.7084,
      "step": 10759
    },
    {
      "epoch": 1.6844082654978085,
      "grad_norm": 2.2462806701660156,
      "learning_rate": 4.960899315738025e-05,
      "loss": 0.5013,
      "step": 10760
    },
    {
      "epoch": 1.6845648090169068,
      "grad_norm": 7.449122905731201,
      "learning_rate": 4.958455522971652e-05,
      "loss": 0.4172,
      "step": 10761
    },
    {
      "epoch": 1.684721352536005,
      "grad_norm": 2.3347396850585938,
      "learning_rate": 4.956011730205278e-05,
      "loss": 0.5593,
      "step": 10762
    },
    {
      "epoch": 1.6848778960551032,
      "grad_norm": 1.909669280052185,
      "learning_rate": 4.953567937438905e-05,
      "loss": 0.3513,
      "step": 10763
    },
    {
      "epoch": 1.6850344395742016,
      "grad_norm": 2.685105085372925,
      "learning_rate": 4.951124144672532e-05,
      "loss": 0.641,
      "step": 10764
    },
    {
      "epoch": 1.6851909830933,
      "grad_norm": 1.1838382482528687,
      "learning_rate": 4.948680351906158e-05,
      "loss": 0.5653,
      "step": 10765
    },
    {
      "epoch": 1.6853475266123983,
      "grad_norm": 1.5641990900039673,
      "learning_rate": 4.946236559139785e-05,
      "loss": 0.5987,
      "step": 10766
    },
    {
      "epoch": 1.6855040701314965,
      "grad_norm": 2.0659899711608887,
      "learning_rate": 4.943792766373411e-05,
      "loss": 0.651,
      "step": 10767
    },
    {
      "epoch": 1.6856606136505947,
      "grad_norm": 2.0896155834198,
      "learning_rate": 4.941348973607038e-05,
      "loss": 0.7361,
      "step": 10768
    },
    {
      "epoch": 1.6858171571696932,
      "grad_norm": 4.1929426193237305,
      "learning_rate": 4.938905180840664e-05,
      "loss": 0.5407,
      "step": 10769
    },
    {
      "epoch": 1.6859737006887916,
      "grad_norm": 1.9943493604660034,
      "learning_rate": 4.9364613880742914e-05,
      "loss": 0.6625,
      "step": 10770
    },
    {
      "epoch": 1.6861302442078898,
      "grad_norm": 2.7732839584350586,
      "learning_rate": 4.934017595307917e-05,
      "loss": 1.2895,
      "step": 10771
    },
    {
      "epoch": 1.686286787726988,
      "grad_norm": 2.526482582092285,
      "learning_rate": 4.9315738025415444e-05,
      "loss": 1.0281,
      "step": 10772
    },
    {
      "epoch": 1.6864433312460863,
      "grad_norm": 1.9940377473831177,
      "learning_rate": 4.929130009775171e-05,
      "loss": 1.0371,
      "step": 10773
    },
    {
      "epoch": 1.6865998747651847,
      "grad_norm": 2.744659662246704,
      "learning_rate": 4.926686217008797e-05,
      "loss": 0.6451,
      "step": 10774
    },
    {
      "epoch": 1.6867564182842831,
      "grad_norm": 4.247622013092041,
      "learning_rate": 4.924242424242424e-05,
      "loss": 1.0116,
      "step": 10775
    },
    {
      "epoch": 1.6869129618033814,
      "grad_norm": 2.5660741329193115,
      "learning_rate": 4.921798631476051e-05,
      "loss": 0.9087,
      "step": 10776
    },
    {
      "epoch": 1.6870695053224796,
      "grad_norm": 3.1664459705352783,
      "learning_rate": 4.919354838709677e-05,
      "loss": 0.9292,
      "step": 10777
    },
    {
      "epoch": 1.687226048841578,
      "grad_norm": 1.8890631198883057,
      "learning_rate": 4.916911045943304e-05,
      "loss": 0.9163,
      "step": 10778
    },
    {
      "epoch": 1.6873825923606762,
      "grad_norm": 3.557894706726074,
      "learning_rate": 4.9144672531769304e-05,
      "loss": 0.7879,
      "step": 10779
    },
    {
      "epoch": 1.6875391358797747,
      "grad_norm": 3.2021689414978027,
      "learning_rate": 4.912023460410556e-05,
      "loss": 0.7324,
      "step": 10780
    },
    {
      "epoch": 1.6876956793988729,
      "grad_norm": 3.014575242996216,
      "learning_rate": 4.9095796676441834e-05,
      "loss": 0.7799,
      "step": 10781
    },
    {
      "epoch": 1.687852222917971,
      "grad_norm": 3.13476824760437,
      "learning_rate": 4.9071358748778105e-05,
      "loss": 1.1459,
      "step": 10782
    },
    {
      "epoch": 1.6880087664370695,
      "grad_norm": 2.8570799827575684,
      "learning_rate": 4.9046920821114363e-05,
      "loss": 1.0614,
      "step": 10783
    },
    {
      "epoch": 1.688165309956168,
      "grad_norm": 1.858405351638794,
      "learning_rate": 4.9022482893450635e-05,
      "loss": 0.5261,
      "step": 10784
    },
    {
      "epoch": 1.6883218534752662,
      "grad_norm": 2.642915964126587,
      "learning_rate": 4.89980449657869e-05,
      "loss": 0.8285,
      "step": 10785
    },
    {
      "epoch": 1.6884783969943644,
      "grad_norm": 6.898956775665283,
      "learning_rate": 4.897360703812316e-05,
      "loss": 0.6748,
      "step": 10786
    },
    {
      "epoch": 1.6886349405134626,
      "grad_norm": 2.4896328449249268,
      "learning_rate": 4.894916911045943e-05,
      "loss": 0.7534,
      "step": 10787
    },
    {
      "epoch": 1.688791484032561,
      "grad_norm": 7.32418155670166,
      "learning_rate": 4.89247311827957e-05,
      "loss": 1.3809,
      "step": 10788
    },
    {
      "epoch": 1.6889480275516595,
      "grad_norm": 1.3984144926071167,
      "learning_rate": 4.890029325513196e-05,
      "loss": 0.2564,
      "step": 10789
    },
    {
      "epoch": 1.6891045710707577,
      "grad_norm": 0.826632559299469,
      "learning_rate": 4.887585532746823e-05,
      "loss": 0.2788,
      "step": 10790
    },
    {
      "epoch": 1.689261114589856,
      "grad_norm": 0.395866721868515,
      "learning_rate": 4.8851417399804495e-05,
      "loss": 0.1724,
      "step": 10791
    },
    {
      "epoch": 1.6894176581089542,
      "grad_norm": 0.7708562016487122,
      "learning_rate": 4.8826979472140753e-05,
      "loss": 0.2389,
      "step": 10792
    },
    {
      "epoch": 1.6895742016280526,
      "grad_norm": 0.6404639482498169,
      "learning_rate": 4.8802541544477025e-05,
      "loss": 0.2478,
      "step": 10793
    },
    {
      "epoch": 1.689730745147151,
      "grad_norm": 0.717633068561554,
      "learning_rate": 4.87781036168133e-05,
      "loss": 0.3287,
      "step": 10794
    },
    {
      "epoch": 1.6898872886662493,
      "grad_norm": 1.3549281358718872,
      "learning_rate": 4.8753665689149555e-05,
      "loss": 0.3062,
      "step": 10795
    },
    {
      "epoch": 1.6900438321853475,
      "grad_norm": 1.1032696962356567,
      "learning_rate": 4.8729227761485826e-05,
      "loss": 0.3737,
      "step": 10796
    },
    {
      "epoch": 1.6902003757044457,
      "grad_norm": 1.0608580112457275,
      "learning_rate": 4.870478983382209e-05,
      "loss": 0.3077,
      "step": 10797
    },
    {
      "epoch": 1.6903569192235441,
      "grad_norm": 1.941976547241211,
      "learning_rate": 4.868035190615835e-05,
      "loss": 0.3382,
      "step": 10798
    },
    {
      "epoch": 1.6905134627426426,
      "grad_norm": 1.254552960395813,
      "learning_rate": 4.865591397849462e-05,
      "loss": 0.3143,
      "step": 10799
    },
    {
      "epoch": 1.6906700062617408,
      "grad_norm": 0.9301863312721252,
      "learning_rate": 4.863147605083089e-05,
      "loss": 0.2226,
      "step": 10800
    },
    {
      "epoch": 1.690826549780839,
      "grad_norm": 0.8874600529670715,
      "learning_rate": 4.860703812316715e-05,
      "loss": 0.4288,
      "step": 10801
    },
    {
      "epoch": 1.6909830932999372,
      "grad_norm": 1.8583930730819702,
      "learning_rate": 4.8582600195503415e-05,
      "loss": 0.3001,
      "step": 10802
    },
    {
      "epoch": 1.6911396368190357,
      "grad_norm": 1.5552978515625,
      "learning_rate": 4.8558162267839687e-05,
      "loss": 0.2913,
      "step": 10803
    },
    {
      "epoch": 1.6912961803381341,
      "grad_norm": 0.9442487359046936,
      "learning_rate": 4.8533724340175945e-05,
      "loss": 0.2821,
      "step": 10804
    },
    {
      "epoch": 1.6914527238572323,
      "grad_norm": 1.6942837238311768,
      "learning_rate": 4.8509286412512216e-05,
      "loss": 0.4196,
      "step": 10805
    },
    {
      "epoch": 1.6916092673763305,
      "grad_norm": 1.1128238439559937,
      "learning_rate": 4.848484848484849e-05,
      "loss": 0.3276,
      "step": 10806
    },
    {
      "epoch": 1.6917658108954288,
      "grad_norm": 1.4044594764709473,
      "learning_rate": 4.8460410557184746e-05,
      "loss": 0.3271,
      "step": 10807
    },
    {
      "epoch": 1.6919223544145272,
      "grad_norm": 2.1907525062561035,
      "learning_rate": 4.843597262952101e-05,
      "loss": 0.5816,
      "step": 10808
    },
    {
      "epoch": 1.6920788979336256,
      "grad_norm": 1.5539323091506958,
      "learning_rate": 4.8411534701857275e-05,
      "loss": 0.3358,
      "step": 10809
    },
    {
      "epoch": 1.6922354414527239,
      "grad_norm": 2.5822982788085938,
      "learning_rate": 4.838709677419354e-05,
      "loss": 0.53,
      "step": 10810
    },
    {
      "epoch": 1.692391984971822,
      "grad_norm": 4.984518527984619,
      "learning_rate": 4.836265884652981e-05,
      "loss": 0.8552,
      "step": 10811
    },
    {
      "epoch": 1.6925485284909205,
      "grad_norm": 1.6038222312927246,
      "learning_rate": 4.833822091886607e-05,
      "loss": 0.5728,
      "step": 10812
    },
    {
      "epoch": 1.6927050720100187,
      "grad_norm": 2.4173617362976074,
      "learning_rate": 4.831378299120234e-05,
      "loss": 0.6206,
      "step": 10813
    },
    {
      "epoch": 1.6928616155291172,
      "grad_norm": 3.724741220474243,
      "learning_rate": 4.8289345063538606e-05,
      "loss": 0.6327,
      "step": 10814
    },
    {
      "epoch": 1.6930181590482154,
      "grad_norm": 1.6614710092544556,
      "learning_rate": 4.826490713587487e-05,
      "loss": 0.6322,
      "step": 10815
    },
    {
      "epoch": 1.6931747025673136,
      "grad_norm": 2.336488723754883,
      "learning_rate": 4.8240469208211136e-05,
      "loss": 0.6787,
      "step": 10816
    },
    {
      "epoch": 1.693331246086412,
      "grad_norm": 1.9805657863616943,
      "learning_rate": 4.821603128054741e-05,
      "loss": 0.5463,
      "step": 10817
    },
    {
      "epoch": 1.6934877896055105,
      "grad_norm": 0.8958491086959839,
      "learning_rate": 4.8191593352883665e-05,
      "loss": 0.2454,
      "step": 10818
    },
    {
      "epoch": 1.6936443331246087,
      "grad_norm": 4.5509562492370605,
      "learning_rate": 4.816715542521994e-05,
      "loss": 0.7812,
      "step": 10819
    },
    {
      "epoch": 1.693800876643707,
      "grad_norm": 2.46221923828125,
      "learning_rate": 4.81427174975562e-05,
      "loss": 0.7259,
      "step": 10820
    },
    {
      "epoch": 1.6939574201628051,
      "grad_norm": 2.2861831188201904,
      "learning_rate": 4.8118279569892467e-05,
      "loss": 0.6677,
      "step": 10821
    },
    {
      "epoch": 1.6941139636819036,
      "grad_norm": 2.7029225826263428,
      "learning_rate": 4.809384164222873e-05,
      "loss": 0.4636,
      "step": 10822
    },
    {
      "epoch": 1.694270507201002,
      "grad_norm": 1.2040103673934937,
      "learning_rate": 4.8069403714565e-05,
      "loss": 0.5958,
      "step": 10823
    },
    {
      "epoch": 1.6944270507201002,
      "grad_norm": 5.391923904418945,
      "learning_rate": 4.804496578690126e-05,
      "loss": 0.7894,
      "step": 10824
    },
    {
      "epoch": 1.6945835942391985,
      "grad_norm": 4.4691972732543945,
      "learning_rate": 4.802052785923753e-05,
      "loss": 1.516,
      "step": 10825
    },
    {
      "epoch": 1.6947401377582967,
      "grad_norm": 2.4495632648468018,
      "learning_rate": 4.79960899315738e-05,
      "loss": 1.0442,
      "step": 10826
    },
    {
      "epoch": 1.6948966812773951,
      "grad_norm": 6.339310169219971,
      "learning_rate": 4.797165200391006e-05,
      "loss": 0.9353,
      "step": 10827
    },
    {
      "epoch": 1.6950532247964936,
      "grad_norm": 3.8489506244659424,
      "learning_rate": 4.794721407624633e-05,
      "loss": 1.0026,
      "step": 10828
    },
    {
      "epoch": 1.6952097683155918,
      "grad_norm": 2.871527671813965,
      "learning_rate": 4.79227761485826e-05,
      "loss": 0.6062,
      "step": 10829
    },
    {
      "epoch": 1.69536631183469,
      "grad_norm": 1.528554081916809,
      "learning_rate": 4.7898338220918857e-05,
      "loss": 0.4849,
      "step": 10830
    },
    {
      "epoch": 1.6955228553537882,
      "grad_norm": 4.001394748687744,
      "learning_rate": 4.787390029325513e-05,
      "loss": 1.0743,
      "step": 10831
    },
    {
      "epoch": 1.6956793988728867,
      "grad_norm": 3.781374454498291,
      "learning_rate": 4.784946236559139e-05,
      "loss": 1.147,
      "step": 10832
    },
    {
      "epoch": 1.695835942391985,
      "grad_norm": 4.306796550750732,
      "learning_rate": 4.782502443792766e-05,
      "loss": 0.9814,
      "step": 10833
    },
    {
      "epoch": 1.6959924859110833,
      "grad_norm": 2.652988910675049,
      "learning_rate": 4.780058651026392e-05,
      "loss": 0.9608,
      "step": 10834
    },
    {
      "epoch": 1.6961490294301815,
      "grad_norm": 2.375042676925659,
      "learning_rate": 4.7776148582600194e-05,
      "loss": 0.694,
      "step": 10835
    },
    {
      "epoch": 1.6963055729492797,
      "grad_norm": 1.6284605264663696,
      "learning_rate": 4.775171065493645e-05,
      "loss": 0.4705,
      "step": 10836
    },
    {
      "epoch": 1.6964621164683782,
      "grad_norm": 2.3897979259490967,
      "learning_rate": 4.7727272727272724e-05,
      "loss": 0.7331,
      "step": 10837
    },
    {
      "epoch": 1.6966186599874766,
      "grad_norm": 1.4052478075027466,
      "learning_rate": 4.770283479960899e-05,
      "loss": 0.6741,
      "step": 10838
    },
    {
      "epoch": 1.6967752035065748,
      "grad_norm": 0.5717368125915527,
      "learning_rate": 4.767839687194525e-05,
      "loss": 0.2161,
      "step": 10839
    },
    {
      "epoch": 1.696931747025673,
      "grad_norm": 0.46546775102615356,
      "learning_rate": 4.765395894428152e-05,
      "loss": 0.2037,
      "step": 10840
    },
    {
      "epoch": 1.6970882905447713,
      "grad_norm": 0.6907462477684021,
      "learning_rate": 4.762952101661779e-05,
      "loss": 0.2561,
      "step": 10841
    },
    {
      "epoch": 1.6972448340638697,
      "grad_norm": 0.85224848985672,
      "learning_rate": 4.760508308895405e-05,
      "loss": 0.2443,
      "step": 10842
    },
    {
      "epoch": 1.6974013775829682,
      "grad_norm": 0.8416429162025452,
      "learning_rate": 4.758064516129032e-05,
      "loss": 0.1929,
      "step": 10843
    },
    {
      "epoch": 1.6975579211020664,
      "grad_norm": 0.6139851808547974,
      "learning_rate": 4.7556207233626584e-05,
      "loss": 0.2167,
      "step": 10844
    },
    {
      "epoch": 1.6977144646211646,
      "grad_norm": 0.6043062210083008,
      "learning_rate": 4.753176930596285e-05,
      "loss": 0.2006,
      "step": 10845
    },
    {
      "epoch": 1.697871008140263,
      "grad_norm": 0.43999460339546204,
      "learning_rate": 4.7507331378299114e-05,
      "loss": 0.1848,
      "step": 10846
    },
    {
      "epoch": 1.6980275516593613,
      "grad_norm": 2.4954354763031006,
      "learning_rate": 4.7482893450635385e-05,
      "loss": 0.2923,
      "step": 10847
    },
    {
      "epoch": 1.6981840951784597,
      "grad_norm": 1.6755168437957764,
      "learning_rate": 4.745845552297164e-05,
      "loss": 0.365,
      "step": 10848
    },
    {
      "epoch": 1.698340638697558,
      "grad_norm": 0.7627212405204773,
      "learning_rate": 4.7434017595307915e-05,
      "loss": 0.1334,
      "step": 10849
    },
    {
      "epoch": 1.6984971822166561,
      "grad_norm": 0.7243126034736633,
      "learning_rate": 4.740957966764418e-05,
      "loss": 0.206,
      "step": 10850
    },
    {
      "epoch": 1.6986537257357546,
      "grad_norm": 2.446509599685669,
      "learning_rate": 4.7385141739980444e-05,
      "loss": 0.3525,
      "step": 10851
    },
    {
      "epoch": 1.698810269254853,
      "grad_norm": 2.082062005996704,
      "learning_rate": 4.736070381231671e-05,
      "loss": 0.3959,
      "step": 10852
    },
    {
      "epoch": 1.6989668127739512,
      "grad_norm": 1.239433765411377,
      "learning_rate": 4.733626588465298e-05,
      "loss": 0.36,
      "step": 10853
    },
    {
      "epoch": 1.6991233562930494,
      "grad_norm": 1.6544456481933594,
      "learning_rate": 4.731182795698924e-05,
      "loss": 0.3264,
      "step": 10854
    },
    {
      "epoch": 1.6992798998121477,
      "grad_norm": 16.526798248291016,
      "learning_rate": 4.728739002932551e-05,
      "loss": 0.4697,
      "step": 10855
    },
    {
      "epoch": 1.699436443331246,
      "grad_norm": 1.5879408121109009,
      "learning_rate": 4.7262952101661775e-05,
      "loss": 0.4604,
      "step": 10856
    },
    {
      "epoch": 1.6995929868503445,
      "grad_norm": 1.414926528930664,
      "learning_rate": 4.723851417399804e-05,
      "loss": 0.2591,
      "step": 10857
    },
    {
      "epoch": 1.6997495303694428,
      "grad_norm": 1.0178431272506714,
      "learning_rate": 4.7214076246334305e-05,
      "loss": 0.2829,
      "step": 10858
    },
    {
      "epoch": 1.699906073888541,
      "grad_norm": 1.5842969417572021,
      "learning_rate": 4.7189638318670576e-05,
      "loss": 0.2904,
      "step": 10859
    },
    {
      "epoch": 1.7000626174076392,
      "grad_norm": 1.6290944814682007,
      "learning_rate": 4.7165200391006834e-05,
      "loss": 0.3683,
      "step": 10860
    },
    {
      "epoch": 1.7002191609267376,
      "grad_norm": 1.115478754043579,
      "learning_rate": 4.7140762463343106e-05,
      "loss": 0.4134,
      "step": 10861
    },
    {
      "epoch": 1.700375704445836,
      "grad_norm": 1.0377562046051025,
      "learning_rate": 4.711632453567937e-05,
      "loss": 0.3749,
      "step": 10862
    },
    {
      "epoch": 1.7005322479649343,
      "grad_norm": 3.567615032196045,
      "learning_rate": 4.7091886608015636e-05,
      "loss": 0.6543,
      "step": 10863
    },
    {
      "epoch": 1.7006887914840325,
      "grad_norm": 1.049473524093628,
      "learning_rate": 4.70674486803519e-05,
      "loss": 0.4259,
      "step": 10864
    },
    {
      "epoch": 1.7008453350031307,
      "grad_norm": 0.785376250743866,
      "learning_rate": 4.704301075268817e-05,
      "loss": 0.2459,
      "step": 10865
    },
    {
      "epoch": 1.7010018785222292,
      "grad_norm": 3.4430458545684814,
      "learning_rate": 4.701857282502443e-05,
      "loss": 0.7491,
      "step": 10866
    },
    {
      "epoch": 1.7011584220413276,
      "grad_norm": 0.9417991638183594,
      "learning_rate": 4.69941348973607e-05,
      "loss": 0.4102,
      "step": 10867
    },
    {
      "epoch": 1.7013149655604258,
      "grad_norm": 1.8139441013336182,
      "learning_rate": 4.6969696969696966e-05,
      "loss": 0.6608,
      "step": 10868
    },
    {
      "epoch": 1.701471509079524,
      "grad_norm": 6.10275411605835,
      "learning_rate": 4.694525904203323e-05,
      "loss": 0.6622,
      "step": 10869
    },
    {
      "epoch": 1.7016280525986223,
      "grad_norm": 1.916246771812439,
      "learning_rate": 4.6920821114369496e-05,
      "loss": 0.5263,
      "step": 10870
    },
    {
      "epoch": 1.7017845961177207,
      "grad_norm": 2.4708452224731445,
      "learning_rate": 4.689638318670577e-05,
      "loss": 0.7636,
      "step": 10871
    },
    {
      "epoch": 1.7019411396368191,
      "grad_norm": 1.3258744478225708,
      "learning_rate": 4.6871945259042026e-05,
      "loss": 0.3469,
      "step": 10872
    },
    {
      "epoch": 1.7020976831559174,
      "grad_norm": 1.8512464761734009,
      "learning_rate": 4.68475073313783e-05,
      "loss": 0.7012,
      "step": 10873
    },
    {
      "epoch": 1.7022542266750156,
      "grad_norm": 3.0667710304260254,
      "learning_rate": 4.682306940371456e-05,
      "loss": 1.1852,
      "step": 10874
    },
    {
      "epoch": 1.7024107701941138,
      "grad_norm": 7.509494781494141,
      "learning_rate": 4.679863147605083e-05,
      "loss": 0.9591,
      "step": 10875
    },
    {
      "epoch": 1.7025673137132122,
      "grad_norm": 2.958660364151001,
      "learning_rate": 4.677419354838709e-05,
      "loss": 1.3427,
      "step": 10876
    },
    {
      "epoch": 1.7027238572323107,
      "grad_norm": 2.816685914993286,
      "learning_rate": 4.674975562072336e-05,
      "loss": 0.9676,
      "step": 10877
    },
    {
      "epoch": 1.702880400751409,
      "grad_norm": 3.361295700073242,
      "learning_rate": 4.672531769305962e-05,
      "loss": 0.7556,
      "step": 10878
    },
    {
      "epoch": 1.7030369442705071,
      "grad_norm": 1.7162786722183228,
      "learning_rate": 4.670087976539589e-05,
      "loss": 1.2321,
      "step": 10879
    },
    {
      "epoch": 1.7031934877896056,
      "grad_norm": 9.585728645324707,
      "learning_rate": 4.667644183773216e-05,
      "loss": 0.9282,
      "step": 10880
    },
    {
      "epoch": 1.7033500313087038,
      "grad_norm": 1.799238681793213,
      "learning_rate": 4.665200391006842e-05,
      "loss": 0.8398,
      "step": 10881
    },
    {
      "epoch": 1.7035065748278022,
      "grad_norm": 5.488026142120361,
      "learning_rate": 4.662756598240469e-05,
      "loss": 1.5273,
      "step": 10882
    },
    {
      "epoch": 1.7036631183469004,
      "grad_norm": 3.029956817626953,
      "learning_rate": 4.660312805474096e-05,
      "loss": 0.9112,
      "step": 10883
    },
    {
      "epoch": 1.7038196618659986,
      "grad_norm": 3.336357831954956,
      "learning_rate": 4.657869012707722e-05,
      "loss": 1.0213,
      "step": 10884
    },
    {
      "epoch": 1.703976205385097,
      "grad_norm": 1.5135464668273926,
      "learning_rate": 4.655425219941349e-05,
      "loss": 0.6799,
      "step": 10885
    },
    {
      "epoch": 1.7041327489041955,
      "grad_norm": 2.607713460922241,
      "learning_rate": 4.652981427174975e-05,
      "loss": 0.7304,
      "step": 10886
    },
    {
      "epoch": 1.7042892924232937,
      "grad_norm": 4.478157043457031,
      "learning_rate": 4.650537634408602e-05,
      "loss": 1.1594,
      "step": 10887
    },
    {
      "epoch": 1.704445835942392,
      "grad_norm": 1.5213872194290161,
      "learning_rate": 4.648093841642228e-05,
      "loss": 0.6976,
      "step": 10888
    },
    {
      "epoch": 1.7046023794614902,
      "grad_norm": 1.294400930404663,
      "learning_rate": 4.6456500488758554e-05,
      "loss": 0.3431,
      "step": 10889
    },
    {
      "epoch": 1.7047589229805886,
      "grad_norm": 0.9387235641479492,
      "learning_rate": 4.643206256109481e-05,
      "loss": 0.2044,
      "step": 10890
    },
    {
      "epoch": 1.704915466499687,
      "grad_norm": 0.8334453105926514,
      "learning_rate": 4.6407624633431084e-05,
      "loss": 0.2973,
      "step": 10891
    },
    {
      "epoch": 1.7050720100187853,
      "grad_norm": 0.7536923289299011,
      "learning_rate": 4.638318670576735e-05,
      "loss": 0.3179,
      "step": 10892
    },
    {
      "epoch": 1.7052285535378835,
      "grad_norm": 1.139009714126587,
      "learning_rate": 4.6358748778103614e-05,
      "loss": 0.1864,
      "step": 10893
    },
    {
      "epoch": 1.7053850970569817,
      "grad_norm": 0.825655996799469,
      "learning_rate": 4.633431085043988e-05,
      "loss": 0.1953,
      "step": 10894
    },
    {
      "epoch": 1.7055416405760802,
      "grad_norm": 1.221582055091858,
      "learning_rate": 4.630987292277615e-05,
      "loss": 0.2487,
      "step": 10895
    },
    {
      "epoch": 1.7056981840951786,
      "grad_norm": 0.9275577068328857,
      "learning_rate": 4.628543499511241e-05,
      "loss": 0.2307,
      "step": 10896
    },
    {
      "epoch": 1.7058547276142768,
      "grad_norm": 0.8654758930206299,
      "learning_rate": 4.626099706744868e-05,
      "loss": 0.3036,
      "step": 10897
    },
    {
      "epoch": 1.706011271133375,
      "grad_norm": 5.584901809692383,
      "learning_rate": 4.6236559139784944e-05,
      "loss": 0.2112,
      "step": 10898
    },
    {
      "epoch": 1.7061678146524732,
      "grad_norm": 1.3160961866378784,
      "learning_rate": 4.621212121212121e-05,
      "loss": 0.257,
      "step": 10899
    },
    {
      "epoch": 1.7063243581715717,
      "grad_norm": 0.8160209655761719,
      "learning_rate": 4.6187683284457474e-05,
      "loss": 0.1725,
      "step": 10900
    },
    {
      "epoch": 1.7064809016906701,
      "grad_norm": 1.1387146711349487,
      "learning_rate": 4.6163245356793745e-05,
      "loss": 0.3657,
      "step": 10901
    },
    {
      "epoch": 1.7066374452097683,
      "grad_norm": 2.705634117126465,
      "learning_rate": 4.6138807429130004e-05,
      "loss": 0.5964,
      "step": 10902
    },
    {
      "epoch": 1.7067939887288666,
      "grad_norm": 1.0726594924926758,
      "learning_rate": 4.6114369501466275e-05,
      "loss": 0.3722,
      "step": 10903
    },
    {
      "epoch": 1.7069505322479648,
      "grad_norm": 2.15543794631958,
      "learning_rate": 4.608993157380254e-05,
      "loss": 0.6121,
      "step": 10904
    },
    {
      "epoch": 1.7071070757670632,
      "grad_norm": 2.3426249027252197,
      "learning_rate": 4.6065493646138805e-05,
      "loss": 0.3009,
      "step": 10905
    },
    {
      "epoch": 1.7072636192861617,
      "grad_norm": 1.6663450002670288,
      "learning_rate": 4.604105571847507e-05,
      "loss": 0.366,
      "step": 10906
    },
    {
      "epoch": 1.7074201628052599,
      "grad_norm": 4.950168609619141,
      "learning_rate": 4.601661779081134e-05,
      "loss": 0.392,
      "step": 10907
    },
    {
      "epoch": 1.707576706324358,
      "grad_norm": 2.3819642066955566,
      "learning_rate": 4.59921798631476e-05,
      "loss": 0.3112,
      "step": 10908
    },
    {
      "epoch": 1.7077332498434565,
      "grad_norm": 1.7315618991851807,
      "learning_rate": 4.596774193548387e-05,
      "loss": 0.4806,
      "step": 10909
    },
    {
      "epoch": 1.7078897933625548,
      "grad_norm": 1.356067419052124,
      "learning_rate": 4.5943304007820135e-05,
      "loss": 0.4802,
      "step": 10910
    },
    {
      "epoch": 1.7080463368816532,
      "grad_norm": 1.5837355852127075,
      "learning_rate": 4.5918866080156393e-05,
      "loss": 0.282,
      "step": 10911
    },
    {
      "epoch": 1.7082028804007514,
      "grad_norm": 1.9646797180175781,
      "learning_rate": 4.5894428152492665e-05,
      "loss": 0.3698,
      "step": 10912
    },
    {
      "epoch": 1.7083594239198496,
      "grad_norm": 3.7682149410247803,
      "learning_rate": 4.586999022482894e-05,
      "loss": 0.7863,
      "step": 10913
    },
    {
      "epoch": 1.708515967438948,
      "grad_norm": 2.856004476547241,
      "learning_rate": 4.5845552297165195e-05,
      "loss": 0.5196,
      "step": 10914
    },
    {
      "epoch": 1.7086725109580463,
      "grad_norm": 4.119771957397461,
      "learning_rate": 4.5821114369501466e-05,
      "loss": 0.6147,
      "step": 10915
    },
    {
      "epoch": 1.7088290544771447,
      "grad_norm": 3.8047308921813965,
      "learning_rate": 4.579667644183773e-05,
      "loss": 0.5624,
      "step": 10916
    },
    {
      "epoch": 1.708985597996243,
      "grad_norm": 1.4393452405929565,
      "learning_rate": 4.577223851417399e-05,
      "loss": 1.135,
      "step": 10917
    },
    {
      "epoch": 1.7091421415153412,
      "grad_norm": 2.7042551040649414,
      "learning_rate": 4.574780058651026e-05,
      "loss": 0.7294,
      "step": 10918
    },
    {
      "epoch": 1.7092986850344396,
      "grad_norm": 1.8399097919464111,
      "learning_rate": 4.572336265884653e-05,
      "loss": 0.2943,
      "step": 10919
    },
    {
      "epoch": 1.709455228553538,
      "grad_norm": 3.5511062145233154,
      "learning_rate": 4.569892473118279e-05,
      "loss": 0.5965,
      "step": 10920
    },
    {
      "epoch": 1.7096117720726363,
      "grad_norm": 1.2243882417678833,
      "learning_rate": 4.567448680351906e-05,
      "loss": 0.2229,
      "step": 10921
    },
    {
      "epoch": 1.7097683155917345,
      "grad_norm": 2.0039992332458496,
      "learning_rate": 4.565004887585533e-05,
      "loss": 0.9775,
      "step": 10922
    },
    {
      "epoch": 1.7099248591108327,
      "grad_norm": 2.668609857559204,
      "learning_rate": 4.5625610948191585e-05,
      "loss": 0.749,
      "step": 10923
    },
    {
      "epoch": 1.7100814026299311,
      "grad_norm": 3.119973659515381,
      "learning_rate": 4.5601173020527856e-05,
      "loss": 0.6062,
      "step": 10924
    },
    {
      "epoch": 1.7102379461490296,
      "grad_norm": 4.085299491882324,
      "learning_rate": 4.557673509286413e-05,
      "loss": 1.0235,
      "step": 10925
    },
    {
      "epoch": 1.7103944896681278,
      "grad_norm": 1.9524751901626587,
      "learning_rate": 4.5552297165200386e-05,
      "loss": 0.746,
      "step": 10926
    },
    {
      "epoch": 1.710551033187226,
      "grad_norm": 2.2046310901641846,
      "learning_rate": 4.552785923753666e-05,
      "loss": 0.8188,
      "step": 10927
    },
    {
      "epoch": 1.7107075767063242,
      "grad_norm": 3.634552001953125,
      "learning_rate": 4.550342130987292e-05,
      "loss": 1.0935,
      "step": 10928
    },
    {
      "epoch": 1.7108641202254227,
      "grad_norm": 4.985217571258545,
      "learning_rate": 4.547898338220918e-05,
      "loss": 1.4068,
      "step": 10929
    },
    {
      "epoch": 1.711020663744521,
      "grad_norm": 4.0953288078308105,
      "learning_rate": 4.545454545454545e-05,
      "loss": 0.9341,
      "step": 10930
    },
    {
      "epoch": 1.7111772072636193,
      "grad_norm": 2.7812812328338623,
      "learning_rate": 4.543010752688172e-05,
      "loss": 0.7642,
      "step": 10931
    },
    {
      "epoch": 1.7113337507827175,
      "grad_norm": 2.699666738510132,
      "learning_rate": 4.540566959921798e-05,
      "loss": 1.6983,
      "step": 10932
    },
    {
      "epoch": 1.7114902943018158,
      "grad_norm": 2.4366767406463623,
      "learning_rate": 4.5381231671554246e-05,
      "loss": 1.0511,
      "step": 10933
    },
    {
      "epoch": 1.7116468378209142,
      "grad_norm": 1.5458954572677612,
      "learning_rate": 4.535679374389052e-05,
      "loss": 0.9385,
      "step": 10934
    },
    {
      "epoch": 1.7118033813400126,
      "grad_norm": 0.7985953688621521,
      "learning_rate": 4.5332355816226776e-05,
      "loss": 0.1675,
      "step": 10935
    },
    {
      "epoch": 1.7119599248591109,
      "grad_norm": 2.136017084121704,
      "learning_rate": 4.530791788856305e-05,
      "loss": 0.7345,
      "step": 10936
    },
    {
      "epoch": 1.712116468378209,
      "grad_norm": 1.3035306930541992,
      "learning_rate": 4.5283479960899305e-05,
      "loss": 0.3092,
      "step": 10937
    },
    {
      "epoch": 1.7122730118973073,
      "grad_norm": 2.315929412841797,
      "learning_rate": 4.525904203323558e-05,
      "loss": 1.2992,
      "step": 10938
    },
    {
      "epoch": 1.7124295554164057,
      "grad_norm": 0.8758809566497803,
      "learning_rate": 4.523460410557184e-05,
      "loss": 0.2215,
      "step": 10939
    },
    {
      "epoch": 1.7125860989355042,
      "grad_norm": 0.9055795669555664,
      "learning_rate": 4.5210166177908107e-05,
      "loss": 0.2701,
      "step": 10940
    },
    {
      "epoch": 1.7127426424546024,
      "grad_norm": 0.5900222659111023,
      "learning_rate": 4.518572825024437e-05,
      "loss": 0.2285,
      "step": 10941
    },
    {
      "epoch": 1.7128991859737006,
      "grad_norm": 0.6624120473861694,
      "learning_rate": 4.516129032258064e-05,
      "loss": 0.2342,
      "step": 10942
    },
    {
      "epoch": 1.713055729492799,
      "grad_norm": 0.567440927028656,
      "learning_rate": 4.51368523949169e-05,
      "loss": 0.258,
      "step": 10943
    },
    {
      "epoch": 1.7132122730118973,
      "grad_norm": 0.45192110538482666,
      "learning_rate": 4.511241446725317e-05,
      "loss": 0.1772,
      "step": 10944
    },
    {
      "epoch": 1.7133688165309957,
      "grad_norm": 1.0923038721084595,
      "learning_rate": 4.508797653958944e-05,
      "loss": 0.2843,
      "step": 10945
    },
    {
      "epoch": 1.713525360050094,
      "grad_norm": 0.739136278629303,
      "learning_rate": 4.50635386119257e-05,
      "loss": 0.2207,
      "step": 10946
    },
    {
      "epoch": 1.7136819035691921,
      "grad_norm": 0.9049036502838135,
      "learning_rate": 4.503910068426197e-05,
      "loss": 0.1736,
      "step": 10947
    },
    {
      "epoch": 1.7138384470882906,
      "grad_norm": 1.2949659824371338,
      "learning_rate": 4.501466275659824e-05,
      "loss": 0.2189,
      "step": 10948
    },
    {
      "epoch": 1.7139949906073888,
      "grad_norm": 0.8264017701148987,
      "learning_rate": 4.4990224828934497e-05,
      "loss": 0.2894,
      "step": 10949
    },
    {
      "epoch": 1.7141515341264872,
      "grad_norm": 0.9105465412139893,
      "learning_rate": 4.496578690127077e-05,
      "loss": 0.2483,
      "step": 10950
    },
    {
      "epoch": 1.7143080776455855,
      "grad_norm": 0.6218265891075134,
      "learning_rate": 4.494134897360703e-05,
      "loss": 0.2536,
      "step": 10951
    },
    {
      "epoch": 1.7144646211646837,
      "grad_norm": 1.0677177906036377,
      "learning_rate": 4.49169110459433e-05,
      "loss": 0.464,
      "step": 10952
    },
    {
      "epoch": 1.7146211646837821,
      "grad_norm": 1.5835673809051514,
      "learning_rate": 4.489247311827956e-05,
      "loss": 0.4199,
      "step": 10953
    },
    {
      "epoch": 1.7147777082028806,
      "grad_norm": 3.1192257404327393,
      "learning_rate": 4.4868035190615834e-05,
      "loss": 0.4786,
      "step": 10954
    },
    {
      "epoch": 1.7149342517219788,
      "grad_norm": 2.4409453868865967,
      "learning_rate": 4.484359726295209e-05,
      "loss": 0.3151,
      "step": 10955
    },
    {
      "epoch": 1.715090795241077,
      "grad_norm": 1.3552123308181763,
      "learning_rate": 4.4819159335288364e-05,
      "loss": 0.4947,
      "step": 10956
    },
    {
      "epoch": 1.7152473387601752,
      "grad_norm": 1.606695294380188,
      "learning_rate": 4.479472140762463e-05,
      "loss": 0.6383,
      "step": 10957
    },
    {
      "epoch": 1.7154038822792737,
      "grad_norm": 3.5459699630737305,
      "learning_rate": 4.477028347996089e-05,
      "loss": 0.5953,
      "step": 10958
    },
    {
      "epoch": 1.715560425798372,
      "grad_norm": 1.4051164388656616,
      "learning_rate": 4.474584555229716e-05,
      "loss": 0.5194,
      "step": 10959
    },
    {
      "epoch": 1.7157169693174703,
      "grad_norm": 1.265329122543335,
      "learning_rate": 4.472140762463343e-05,
      "loss": 0.3916,
      "step": 10960
    },
    {
      "epoch": 1.7158735128365685,
      "grad_norm": 2.5221071243286133,
      "learning_rate": 4.469696969696969e-05,
      "loss": 0.5158,
      "step": 10961
    },
    {
      "epoch": 1.7160300563556667,
      "grad_norm": 2.5884549617767334,
      "learning_rate": 4.467253176930596e-05,
      "loss": 0.3095,
      "step": 10962
    },
    {
      "epoch": 1.7161865998747652,
      "grad_norm": 2.0343172550201416,
      "learning_rate": 4.4648093841642224e-05,
      "loss": 0.6869,
      "step": 10963
    },
    {
      "epoch": 1.7163431433938636,
      "grad_norm": 2.2614247798919678,
      "learning_rate": 4.462365591397849e-05,
      "loss": 0.7452,
      "step": 10964
    },
    {
      "epoch": 1.7164996869129618,
      "grad_norm": 1.4859260320663452,
      "learning_rate": 4.4599217986314754e-05,
      "loss": 0.4323,
      "step": 10965
    },
    {
      "epoch": 1.71665623043206,
      "grad_norm": 2.1878414154052734,
      "learning_rate": 4.4574780058651025e-05,
      "loss": 0.6757,
      "step": 10966
    },
    {
      "epoch": 1.7168127739511583,
      "grad_norm": 2.8746695518493652,
      "learning_rate": 4.455034213098728e-05,
      "loss": 0.6629,
      "step": 10967
    },
    {
      "epoch": 1.7169693174702567,
      "grad_norm": 2.4773285388946533,
      "learning_rate": 4.4525904203323555e-05,
      "loss": 0.526,
      "step": 10968
    },
    {
      "epoch": 1.7171258609893552,
      "grad_norm": 4.917586803436279,
      "learning_rate": 4.450146627565982e-05,
      "loss": 0.735,
      "step": 10969
    },
    {
      "epoch": 1.7172824045084534,
      "grad_norm": 3.6827588081359863,
      "learning_rate": 4.4477028347996084e-05,
      "loss": 0.9001,
      "step": 10970
    },
    {
      "epoch": 1.7174389480275516,
      "grad_norm": 3.3840487003326416,
      "learning_rate": 4.445259042033235e-05,
      "loss": 0.8196,
      "step": 10971
    },
    {
      "epoch": 1.7175954915466498,
      "grad_norm": 3.665562391281128,
      "learning_rate": 4.442815249266862e-05,
      "loss": 0.8158,
      "step": 10972
    },
    {
      "epoch": 1.7177520350657483,
      "grad_norm": 1.8126392364501953,
      "learning_rate": 4.440371456500488e-05,
      "loss": 0.8043,
      "step": 10973
    },
    {
      "epoch": 1.7179085785848467,
      "grad_norm": 2.7795164585113525,
      "learning_rate": 4.437927663734115e-05,
      "loss": 0.9601,
      "step": 10974
    },
    {
      "epoch": 1.718065122103945,
      "grad_norm": 3.174985647201538,
      "learning_rate": 4.4354838709677415e-05,
      "loss": 0.6358,
      "step": 10975
    },
    {
      "epoch": 1.7182216656230431,
      "grad_norm": 2.6935548782348633,
      "learning_rate": 4.433040078201368e-05,
      "loss": 0.6868,
      "step": 10976
    },
    {
      "epoch": 1.7183782091421416,
      "grad_norm": 3.725315809249878,
      "learning_rate": 4.4305962854349945e-05,
      "loss": 1.1501,
      "step": 10977
    },
    {
      "epoch": 1.7185347526612398,
      "grad_norm": 3.672032356262207,
      "learning_rate": 4.4281524926686216e-05,
      "loss": 1.0729,
      "step": 10978
    },
    {
      "epoch": 1.7186912961803382,
      "grad_norm": 3.4297823905944824,
      "learning_rate": 4.4257086999022474e-05,
      "loss": 0.9444,
      "step": 10979
    },
    {
      "epoch": 1.7188478396994364,
      "grad_norm": 2.7439005374908447,
      "learning_rate": 4.4232649071358746e-05,
      "loss": 0.7806,
      "step": 10980
    },
    {
      "epoch": 1.7190043832185347,
      "grad_norm": 1.5056949853897095,
      "learning_rate": 4.420821114369501e-05,
      "loss": 0.8998,
      "step": 10981
    },
    {
      "epoch": 1.719160926737633,
      "grad_norm": 4.065816879272461,
      "learning_rate": 4.4183773216031276e-05,
      "loss": 0.5133,
      "step": 10982
    },
    {
      "epoch": 1.7193174702567313,
      "grad_norm": 2.949634552001953,
      "learning_rate": 4.415933528836754e-05,
      "loss": 1.242,
      "step": 10983
    },
    {
      "epoch": 1.7194740137758298,
      "grad_norm": 3.2532920837402344,
      "learning_rate": 4.413489736070381e-05,
      "loss": 0.8277,
      "step": 10984
    },
    {
      "epoch": 1.719630557294928,
      "grad_norm": 2.431811809539795,
      "learning_rate": 4.411045943304007e-05,
      "loss": 0.4927,
      "step": 10985
    },
    {
      "epoch": 1.7197871008140262,
      "grad_norm": 1.749243140220642,
      "learning_rate": 4.408602150537634e-05,
      "loss": 0.6917,
      "step": 10986
    },
    {
      "epoch": 1.7199436443331246,
      "grad_norm": 3.768820285797119,
      "learning_rate": 4.4061583577712606e-05,
      "loss": 0.5891,
      "step": 10987
    },
    {
      "epoch": 1.720100187852223,
      "grad_norm": 1.9826332330703735,
      "learning_rate": 4.403714565004887e-05,
      "loss": 0.6472,
      "step": 10988
    },
    {
      "epoch": 1.7202567313713213,
      "grad_norm": 1.206260323524475,
      "learning_rate": 4.4012707722385136e-05,
      "loss": 0.4265,
      "step": 10989
    },
    {
      "epoch": 1.7204132748904195,
      "grad_norm": 0.634954571723938,
      "learning_rate": 4.398826979472141e-05,
      "loss": 0.1923,
      "step": 10990
    },
    {
      "epoch": 1.7205698184095177,
      "grad_norm": 0.9732500910758972,
      "learning_rate": 4.3963831867057666e-05,
      "loss": 0.1833,
      "step": 10991
    },
    {
      "epoch": 1.7207263619286162,
      "grad_norm": 0.934144139289856,
      "learning_rate": 4.393939393939394e-05,
      "loss": 0.2087,
      "step": 10992
    },
    {
      "epoch": 1.7208829054477146,
      "grad_norm": 1.1597187519073486,
      "learning_rate": 4.39149560117302e-05,
      "loss": 0.3764,
      "step": 10993
    },
    {
      "epoch": 1.7210394489668128,
      "grad_norm": 0.7562986612319946,
      "learning_rate": 4.389051808406647e-05,
      "loss": 0.1977,
      "step": 10994
    },
    {
      "epoch": 1.721195992485911,
      "grad_norm": 0.9375834465026855,
      "learning_rate": 4.386608015640273e-05,
      "loss": 0.3025,
      "step": 10995
    },
    {
      "epoch": 1.7213525360050093,
      "grad_norm": 0.608613908290863,
      "learning_rate": 4.3841642228739e-05,
      "loss": 0.2857,
      "step": 10996
    },
    {
      "epoch": 1.7215090795241077,
      "grad_norm": 0.8662186861038208,
      "learning_rate": 4.381720430107526e-05,
      "loss": 0.2124,
      "step": 10997
    },
    {
      "epoch": 1.7216656230432061,
      "grad_norm": 0.936297595500946,
      "learning_rate": 4.379276637341153e-05,
      "loss": 0.25,
      "step": 10998
    },
    {
      "epoch": 1.7218221665623044,
      "grad_norm": 3.5861268043518066,
      "learning_rate": 4.37683284457478e-05,
      "loss": 0.3277,
      "step": 10999
    },
    {
      "epoch": 1.7219787100814026,
      "grad_norm": 1.0144635438919067,
      "learning_rate": 4.374389051808406e-05,
      "loss": 0.3815,
      "step": 11000
    },
    {
      "epoch": 1.7219787100814026,
      "eval_loss": 0.49481523036956787,
      "eval_runtime": 205.6846,
      "eval_samples_per_second": 60.204,
      "eval_steps_per_second": 3.763,
      "eval_wer": 0.31803982404318165,
      "step": 11000
    },
    {
      "epoch": 1.7221352536005008,
      "grad_norm": 1.0101224184036255,
      "learning_rate": 4.371945259042033e-05,
      "loss": 0.4074,
      "step": 11001
    },
    {
      "epoch": 1.7222917971195992,
      "grad_norm": 1.773930549621582,
      "learning_rate": 4.36950146627566e-05,
      "loss": 0.3936,
      "step": 11002
    },
    {
      "epoch": 1.7224483406386977,
      "grad_norm": 1.4694632291793823,
      "learning_rate": 4.367057673509286e-05,
      "loss": 0.4899,
      "step": 11003
    },
    {
      "epoch": 1.722604884157796,
      "grad_norm": 2.8984897136688232,
      "learning_rate": 4.364613880742913e-05,
      "loss": 0.4263,
      "step": 11004
    },
    {
      "epoch": 1.722761427676894,
      "grad_norm": 1.7322542667388916,
      "learning_rate": 4.362170087976539e-05,
      "loss": 0.4388,
      "step": 11005
    },
    {
      "epoch": 1.7229179711959923,
      "grad_norm": 1.0889065265655518,
      "learning_rate": 4.359726295210166e-05,
      "loss": 0.441,
      "step": 11006
    },
    {
      "epoch": 1.7230745147150908,
      "grad_norm": 0.9822616577148438,
      "learning_rate": 4.357282502443792e-05,
      "loss": 0.3078,
      "step": 11007
    },
    {
      "epoch": 1.7232310582341892,
      "grad_norm": 0.8187744617462158,
      "learning_rate": 4.3548387096774194e-05,
      "loss": 0.3078,
      "step": 11008
    },
    {
      "epoch": 1.7233876017532874,
      "grad_norm": 2.007962465286255,
      "learning_rate": 4.352394916911045e-05,
      "loss": 0.5249,
      "step": 11009
    },
    {
      "epoch": 1.7235441452723856,
      "grad_norm": 1.7840535640716553,
      "learning_rate": 4.3499511241446724e-05,
      "loss": 0.7433,
      "step": 11010
    },
    {
      "epoch": 1.723700688791484,
      "grad_norm": 1.7337597608566284,
      "learning_rate": 4.347507331378299e-05,
      "loss": 0.5195,
      "step": 11011
    },
    {
      "epoch": 1.7238572323105823,
      "grad_norm": 1.91448175907135,
      "learning_rate": 4.3450635386119254e-05,
      "loss": 0.3144,
      "step": 11012
    },
    {
      "epoch": 1.7240137758296807,
      "grad_norm": 2.5847277641296387,
      "learning_rate": 4.342619745845552e-05,
      "loss": 0.5275,
      "step": 11013
    },
    {
      "epoch": 1.724170319348779,
      "grad_norm": 8.203601837158203,
      "learning_rate": 4.340175953079179e-05,
      "loss": 0.6911,
      "step": 11014
    },
    {
      "epoch": 1.7243268628678772,
      "grad_norm": 2.416170835494995,
      "learning_rate": 4.337732160312805e-05,
      "loss": 0.6947,
      "step": 11015
    },
    {
      "epoch": 1.7244834063869756,
      "grad_norm": 2.0026931762695312,
      "learning_rate": 4.335288367546432e-05,
      "loss": 0.4616,
      "step": 11016
    },
    {
      "epoch": 1.724639949906074,
      "grad_norm": 2.050572633743286,
      "learning_rate": 4.3328445747800584e-05,
      "loss": 0.4845,
      "step": 11017
    },
    {
      "epoch": 1.7247964934251723,
      "grad_norm": 2.2032740116119385,
      "learning_rate": 4.330400782013685e-05,
      "loss": 0.469,
      "step": 11018
    },
    {
      "epoch": 1.7249530369442705,
      "grad_norm": 2.604659080505371,
      "learning_rate": 4.3279569892473114e-05,
      "loss": 0.8241,
      "step": 11019
    },
    {
      "epoch": 1.7251095804633687,
      "grad_norm": 1.5613411664962769,
      "learning_rate": 4.3255131964809385e-05,
      "loss": 0.5109,
      "step": 11020
    },
    {
      "epoch": 1.7252661239824671,
      "grad_norm": 1.821577548980713,
      "learning_rate": 4.3230694037145644e-05,
      "loss": 0.5093,
      "step": 11021
    },
    {
      "epoch": 1.7254226675015656,
      "grad_norm": 2.0643904209136963,
      "learning_rate": 4.3206256109481915e-05,
      "loss": 0.5367,
      "step": 11022
    },
    {
      "epoch": 1.7255792110206638,
      "grad_norm": 2.7602663040161133,
      "learning_rate": 4.318181818181818e-05,
      "loss": 1.0387,
      "step": 11023
    },
    {
      "epoch": 1.725735754539762,
      "grad_norm": 5.380650520324707,
      "learning_rate": 4.3157380254154445e-05,
      "loss": 1.1233,
      "step": 11024
    },
    {
      "epoch": 1.7258922980588602,
      "grad_norm": 3.1562740802764893,
      "learning_rate": 4.313294232649071e-05,
      "loss": 1.0676,
      "step": 11025
    },
    {
      "epoch": 1.7260488415779587,
      "grad_norm": 3.140995502471924,
      "learning_rate": 4.310850439882698e-05,
      "loss": 0.8957,
      "step": 11026
    },
    {
      "epoch": 1.7262053850970571,
      "grad_norm": 2.0646467208862305,
      "learning_rate": 4.308406647116324e-05,
      "loss": 0.5783,
      "step": 11027
    },
    {
      "epoch": 1.7263619286161553,
      "grad_norm": 3.1891887187957764,
      "learning_rate": 4.305962854349951e-05,
      "loss": 1.0697,
      "step": 11028
    },
    {
      "epoch": 1.7265184721352536,
      "grad_norm": 4.30599308013916,
      "learning_rate": 4.3035190615835775e-05,
      "loss": 1.599,
      "step": 11029
    },
    {
      "epoch": 1.7266750156543518,
      "grad_norm": 1.5007469654083252,
      "learning_rate": 4.301075268817204e-05,
      "loss": 0.8014,
      "step": 11030
    },
    {
      "epoch": 1.7268315591734502,
      "grad_norm": 3.0806431770324707,
      "learning_rate": 4.2986314760508305e-05,
      "loss": 1.201,
      "step": 11031
    },
    {
      "epoch": 1.7269881026925487,
      "grad_norm": 3.5166988372802734,
      "learning_rate": 4.296187683284458e-05,
      "loss": 0.8214,
      "step": 11032
    },
    {
      "epoch": 1.7271446462116469,
      "grad_norm": 2.085771083831787,
      "learning_rate": 4.2937438905180835e-05,
      "loss": 1.237,
      "step": 11033
    },
    {
      "epoch": 1.727301189730745,
      "grad_norm": 1.5638984441757202,
      "learning_rate": 4.2913000977517106e-05,
      "loss": 0.512,
      "step": 11034
    },
    {
      "epoch": 1.7274577332498433,
      "grad_norm": 1.8894531726837158,
      "learning_rate": 4.288856304985337e-05,
      "loss": 0.7843,
      "step": 11035
    },
    {
      "epoch": 1.7276142767689417,
      "grad_norm": 3.0840094089508057,
      "learning_rate": 4.2864125122189636e-05,
      "loss": 0.777,
      "step": 11036
    },
    {
      "epoch": 1.7277708202880402,
      "grad_norm": 3.5215682983398438,
      "learning_rate": 4.28396871945259e-05,
      "loss": 0.9629,
      "step": 11037
    },
    {
      "epoch": 1.7279273638071384,
      "grad_norm": 2.8310680389404297,
      "learning_rate": 4.281524926686217e-05,
      "loss": 0.923,
      "step": 11038
    },
    {
      "epoch": 1.7280839073262366,
      "grad_norm": 1.4804656505584717,
      "learning_rate": 4.279081133919843e-05,
      "loss": 0.2002,
      "step": 11039
    },
    {
      "epoch": 1.7282404508453348,
      "grad_norm": 0.7842070460319519,
      "learning_rate": 4.27663734115347e-05,
      "loss": 0.2244,
      "step": 11040
    },
    {
      "epoch": 1.7283969943644333,
      "grad_norm": 0.7354947328567505,
      "learning_rate": 4.274193548387097e-05,
      "loss": 0.277,
      "step": 11041
    },
    {
      "epoch": 1.7285535378835317,
      "grad_norm": 0.7232000827789307,
      "learning_rate": 4.2717497556207225e-05,
      "loss": 0.1575,
      "step": 11042
    },
    {
      "epoch": 1.72871008140263,
      "grad_norm": 2.255171537399292,
      "learning_rate": 4.2693059628543496e-05,
      "loss": 0.2009,
      "step": 11043
    },
    {
      "epoch": 1.7288666249217282,
      "grad_norm": 1.2567336559295654,
      "learning_rate": 4.266862170087977e-05,
      "loss": 0.3623,
      "step": 11044
    },
    {
      "epoch": 1.7290231684408266,
      "grad_norm": 0.5677285194396973,
      "learning_rate": 4.2644183773216026e-05,
      "loss": 0.1847,
      "step": 11045
    },
    {
      "epoch": 1.7291797119599248,
      "grad_norm": 4.918628215789795,
      "learning_rate": 4.26197458455523e-05,
      "loss": 0.2151,
      "step": 11046
    },
    {
      "epoch": 1.7293362554790233,
      "grad_norm": 0.9489184617996216,
      "learning_rate": 4.259530791788856e-05,
      "loss": 0.3459,
      "step": 11047
    },
    {
      "epoch": 1.7294927989981215,
      "grad_norm": 3.799226999282837,
      "learning_rate": 4.257086999022482e-05,
      "loss": 0.3196,
      "step": 11048
    },
    {
      "epoch": 1.7296493425172197,
      "grad_norm": 0.8148688673973083,
      "learning_rate": 4.254643206256109e-05,
      "loss": 0.2697,
      "step": 11049
    },
    {
      "epoch": 1.7298058860363181,
      "grad_norm": 0.8470799922943115,
      "learning_rate": 4.252199413489736e-05,
      "loss": 0.3325,
      "step": 11050
    },
    {
      "epoch": 1.7299624295554166,
      "grad_norm": 0.811160147190094,
      "learning_rate": 4.249755620723362e-05,
      "loss": 0.3461,
      "step": 11051
    },
    {
      "epoch": 1.7301189730745148,
      "grad_norm": 1.2158924341201782,
      "learning_rate": 4.247311827956989e-05,
      "loss": 0.5058,
      "step": 11052
    },
    {
      "epoch": 1.730275516593613,
      "grad_norm": 7.7890825271606445,
      "learning_rate": 4.244868035190616e-05,
      "loss": 0.3067,
      "step": 11053
    },
    {
      "epoch": 1.7304320601127112,
      "grad_norm": 2.8912837505340576,
      "learning_rate": 4.2424242424242416e-05,
      "loss": 0.6897,
      "step": 11054
    },
    {
      "epoch": 1.7305886036318097,
      "grad_norm": 1.3283599615097046,
      "learning_rate": 4.239980449657869e-05,
      "loss": 0.533,
      "step": 11055
    },
    {
      "epoch": 1.730745147150908,
      "grad_norm": 0.8418616056442261,
      "learning_rate": 4.237536656891496e-05,
      "loss": 0.2678,
      "step": 11056
    },
    {
      "epoch": 1.7309016906700063,
      "grad_norm": 1.1012041568756104,
      "learning_rate": 4.235092864125122e-05,
      "loss": 0.2165,
      "step": 11057
    },
    {
      "epoch": 1.7310582341891045,
      "grad_norm": 1.7594144344329834,
      "learning_rate": 4.232649071358749e-05,
      "loss": 0.4197,
      "step": 11058
    },
    {
      "epoch": 1.7312147777082028,
      "grad_norm": 2.9779410362243652,
      "learning_rate": 4.230205278592375e-05,
      "loss": 0.7293,
      "step": 11059
    },
    {
      "epoch": 1.7313713212273012,
      "grad_norm": 2.907015085220337,
      "learning_rate": 4.227761485826001e-05,
      "loss": 0.3366,
      "step": 11060
    },
    {
      "epoch": 1.7315278647463996,
      "grad_norm": 1.4411675930023193,
      "learning_rate": 4.225317693059628e-05,
      "loss": 0.3136,
      "step": 11061
    },
    {
      "epoch": 1.7316844082654979,
      "grad_norm": 1.9844144582748413,
      "learning_rate": 4.2228739002932555e-05,
      "loss": 0.6951,
      "step": 11062
    },
    {
      "epoch": 1.731840951784596,
      "grad_norm": 1.4029967784881592,
      "learning_rate": 4.220430107526881e-05,
      "loss": 0.4267,
      "step": 11063
    },
    {
      "epoch": 1.7319974953036943,
      "grad_norm": 2.417787551879883,
      "learning_rate": 4.2179863147605084e-05,
      "loss": 0.7369,
      "step": 11064
    },
    {
      "epoch": 1.7321540388227927,
      "grad_norm": 2.0894277095794678,
      "learning_rate": 4.215542521994134e-05,
      "loss": 0.7088,
      "step": 11065
    },
    {
      "epoch": 1.7323105823418912,
      "grad_norm": 3.367473840713501,
      "learning_rate": 4.213098729227761e-05,
      "loss": 0.5431,
      "step": 11066
    },
    {
      "epoch": 1.7324671258609894,
      "grad_norm": 1.4022886753082275,
      "learning_rate": 4.210654936461388e-05,
      "loss": 0.3418,
      "step": 11067
    },
    {
      "epoch": 1.7326236693800876,
      "grad_norm": 3.337224245071411,
      "learning_rate": 4.2082111436950137e-05,
      "loss": 0.7644,
      "step": 11068
    },
    {
      "epoch": 1.7327802128991858,
      "grad_norm": 3.8841137886047363,
      "learning_rate": 4.205767350928641e-05,
      "loss": 0.7558,
      "step": 11069
    },
    {
      "epoch": 1.7329367564182843,
      "grad_norm": 1.932528018951416,
      "learning_rate": 4.203323558162267e-05,
      "loss": 0.4611,
      "step": 11070
    },
    {
      "epoch": 1.7330932999373827,
      "grad_norm": 2.1921749114990234,
      "learning_rate": 4.200879765395894e-05,
      "loss": 0.3621,
      "step": 11071
    },
    {
      "epoch": 1.733249843456481,
      "grad_norm": 1.8706334829330444,
      "learning_rate": 4.19843597262952e-05,
      "loss": 0.8474,
      "step": 11072
    },
    {
      "epoch": 1.7334063869755791,
      "grad_norm": 2.5191290378570557,
      "learning_rate": 4.1959921798631474e-05,
      "loss": 0.6197,
      "step": 11073
    },
    {
      "epoch": 1.7335629304946774,
      "grad_norm": 2.9892024993896484,
      "learning_rate": 4.193548387096773e-05,
      "loss": 0.5236,
      "step": 11074
    },
    {
      "epoch": 1.7337194740137758,
      "grad_norm": 18.517911911010742,
      "learning_rate": 4.1911045943304004e-05,
      "loss": 1.1508,
      "step": 11075
    },
    {
      "epoch": 1.7338760175328742,
      "grad_norm": 2.092811346054077,
      "learning_rate": 4.188660801564027e-05,
      "loss": 0.9638,
      "step": 11076
    },
    {
      "epoch": 1.7340325610519725,
      "grad_norm": 2.0851621627807617,
      "learning_rate": 4.186217008797653e-05,
      "loss": 0.8178,
      "step": 11077
    },
    {
      "epoch": 1.7341891045710707,
      "grad_norm": 2.2361598014831543,
      "learning_rate": 4.18377321603128e-05,
      "loss": 0.7663,
      "step": 11078
    },
    {
      "epoch": 1.7343456480901691,
      "grad_norm": 2.4493370056152344,
      "learning_rate": 4.181329423264907e-05,
      "loss": 1.1463,
      "step": 11079
    },
    {
      "epoch": 1.7345021916092673,
      "grad_norm": 2.9259376525878906,
      "learning_rate": 4.178885630498533e-05,
      "loss": 1.4037,
      "step": 11080
    },
    {
      "epoch": 1.7346587351283658,
      "grad_norm": 2.476630926132202,
      "learning_rate": 4.17644183773216e-05,
      "loss": 1.0048,
      "step": 11081
    },
    {
      "epoch": 1.734815278647464,
      "grad_norm": 2.6673030853271484,
      "learning_rate": 4.1739980449657864e-05,
      "loss": 1.3103,
      "step": 11082
    },
    {
      "epoch": 1.7349718221665622,
      "grad_norm": 2.157806873321533,
      "learning_rate": 4.171554252199413e-05,
      "loss": 0.9964,
      "step": 11083
    },
    {
      "epoch": 1.7351283656856606,
      "grad_norm": 1.6744898557662964,
      "learning_rate": 4.1691104594330394e-05,
      "loss": 0.4646,
      "step": 11084
    },
    {
      "epoch": 1.735284909204759,
      "grad_norm": 2.1040098667144775,
      "learning_rate": 4.1666666666666665e-05,
      "loss": 0.6487,
      "step": 11085
    },
    {
      "epoch": 1.7354414527238573,
      "grad_norm": 1.4487780332565308,
      "learning_rate": 4.164222873900292e-05,
      "loss": 0.4765,
      "step": 11086
    },
    {
      "epoch": 1.7355979962429555,
      "grad_norm": 1.442169189453125,
      "learning_rate": 4.1617790811339195e-05,
      "loss": 0.5098,
      "step": 11087
    },
    {
      "epoch": 1.7357545397620537,
      "grad_norm": 1.7279036045074463,
      "learning_rate": 4.159335288367546e-05,
      "loss": 1.1097,
      "step": 11088
    },
    {
      "epoch": 1.7359110832811522,
      "grad_norm": 0.6806561350822449,
      "learning_rate": 4.1568914956011724e-05,
      "loss": 0.2139,
      "step": 11089
    },
    {
      "epoch": 1.7360676268002506,
      "grad_norm": 0.7717947959899902,
      "learning_rate": 4.154447702834799e-05,
      "loss": 0.201,
      "step": 11090
    },
    {
      "epoch": 1.7362241703193488,
      "grad_norm": 1.283336877822876,
      "learning_rate": 4.152003910068426e-05,
      "loss": 0.2105,
      "step": 11091
    },
    {
      "epoch": 1.736380713838447,
      "grad_norm": 0.7973412275314331,
      "learning_rate": 4.149560117302052e-05,
      "loss": 0.2479,
      "step": 11092
    },
    {
      "epoch": 1.7365372573575453,
      "grad_norm": 1.4070724248886108,
      "learning_rate": 4.147116324535679e-05,
      "loss": 0.2806,
      "step": 11093
    },
    {
      "epoch": 1.7366938008766437,
      "grad_norm": 0.8974490761756897,
      "learning_rate": 4.1446725317693055e-05,
      "loss": 0.1904,
      "step": 11094
    },
    {
      "epoch": 1.7368503443957422,
      "grad_norm": 1.0805352926254272,
      "learning_rate": 4.142228739002932e-05,
      "loss": 0.3231,
      "step": 11095
    },
    {
      "epoch": 1.7370068879148404,
      "grad_norm": 0.7345172762870789,
      "learning_rate": 4.1397849462365585e-05,
      "loss": 0.2135,
      "step": 11096
    },
    {
      "epoch": 1.7371634314339386,
      "grad_norm": 1.2740095853805542,
      "learning_rate": 4.1373411534701856e-05,
      "loss": 0.3516,
      "step": 11097
    },
    {
      "epoch": 1.7373199749530368,
      "grad_norm": 0.7698444724082947,
      "learning_rate": 4.1348973607038114e-05,
      "loss": 0.362,
      "step": 11098
    },
    {
      "epoch": 1.7374765184721352,
      "grad_norm": 1.8207945823669434,
      "learning_rate": 4.1324535679374386e-05,
      "loss": 0.4615,
      "step": 11099
    },
    {
      "epoch": 1.7376330619912337,
      "grad_norm": 2.351137638092041,
      "learning_rate": 4.130009775171065e-05,
      "loss": 0.3479,
      "step": 11100
    },
    {
      "epoch": 1.737789605510332,
      "grad_norm": 0.9402967095375061,
      "learning_rate": 4.1275659824046916e-05,
      "loss": 0.2351,
      "step": 11101
    },
    {
      "epoch": 1.7379461490294301,
      "grad_norm": 0.9055925011634827,
      "learning_rate": 4.125122189638318e-05,
      "loss": 0.2323,
      "step": 11102
    },
    {
      "epoch": 1.7381026925485283,
      "grad_norm": 1.6312155723571777,
      "learning_rate": 4.122678396871945e-05,
      "loss": 0.3386,
      "step": 11103
    },
    {
      "epoch": 1.7382592360676268,
      "grad_norm": 1.4077454805374146,
      "learning_rate": 4.120234604105571e-05,
      "loss": 0.4023,
      "step": 11104
    },
    {
      "epoch": 1.7384157795867252,
      "grad_norm": 1.1839890480041504,
      "learning_rate": 4.117790811339198e-05,
      "loss": 0.3056,
      "step": 11105
    },
    {
      "epoch": 1.7385723231058234,
      "grad_norm": 1.0270860195159912,
      "learning_rate": 4.1153470185728246e-05,
      "loss": 0.4321,
      "step": 11106
    },
    {
      "epoch": 1.7387288666249217,
      "grad_norm": 1.704751968383789,
      "learning_rate": 4.112903225806451e-05,
      "loss": 0.3378,
      "step": 11107
    },
    {
      "epoch": 1.7388854101440199,
      "grad_norm": 1.4541360139846802,
      "learning_rate": 4.1104594330400776e-05,
      "loss": 0.3107,
      "step": 11108
    },
    {
      "epoch": 1.7390419536631183,
      "grad_norm": 2.142528772354126,
      "learning_rate": 4.108015640273705e-05,
      "loss": 0.5536,
      "step": 11109
    },
    {
      "epoch": 1.7391984971822168,
      "grad_norm": 1.4981656074523926,
      "learning_rate": 4.1055718475073306e-05,
      "loss": 0.5362,
      "step": 11110
    },
    {
      "epoch": 1.739355040701315,
      "grad_norm": 2.808462381362915,
      "learning_rate": 4.103128054740958e-05,
      "loss": 0.5691,
      "step": 11111
    },
    {
      "epoch": 1.7395115842204132,
      "grad_norm": 1.8647723197937012,
      "learning_rate": 4.100684261974584e-05,
      "loss": 0.6461,
      "step": 11112
    },
    {
      "epoch": 1.7396681277395116,
      "grad_norm": 1.9372211694717407,
      "learning_rate": 4.098240469208211e-05,
      "loss": 0.3173,
      "step": 11113
    },
    {
      "epoch": 1.7398246712586098,
      "grad_norm": 1.7311444282531738,
      "learning_rate": 4.095796676441837e-05,
      "loss": 0.3582,
      "step": 11114
    },
    {
      "epoch": 1.7399812147777083,
      "grad_norm": 3.7500319480895996,
      "learning_rate": 4.093352883675464e-05,
      "loss": 0.7594,
      "step": 11115
    },
    {
      "epoch": 1.7401377582968065,
      "grad_norm": 1.9873162508010864,
      "learning_rate": 4.09090909090909e-05,
      "loss": 0.6123,
      "step": 11116
    },
    {
      "epoch": 1.7402943018159047,
      "grad_norm": 15.171030044555664,
      "learning_rate": 4.088465298142717e-05,
      "loss": 0.719,
      "step": 11117
    },
    {
      "epoch": 1.7404508453350032,
      "grad_norm": 6.604526996612549,
      "learning_rate": 4.086021505376344e-05,
      "loss": 0.5115,
      "step": 11118
    },
    {
      "epoch": 1.7406073888541016,
      "grad_norm": 1.5956652164459229,
      "learning_rate": 4.08357771260997e-05,
      "loss": 0.4616,
      "step": 11119
    },
    {
      "epoch": 1.7407639323731998,
      "grad_norm": 2.011253833770752,
      "learning_rate": 4.081133919843597e-05,
      "loss": 0.4544,
      "step": 11120
    },
    {
      "epoch": 1.740920475892298,
      "grad_norm": 1.8588051795959473,
      "learning_rate": 4.078690127077224e-05,
      "loss": 0.6543,
      "step": 11121
    },
    {
      "epoch": 1.7410770194113963,
      "grad_norm": 3.617205858230591,
      "learning_rate": 4.07624633431085e-05,
      "loss": 0.8809,
      "step": 11122
    },
    {
      "epoch": 1.7412335629304947,
      "grad_norm": 5.1503825187683105,
      "learning_rate": 4.073802541544477e-05,
      "loss": 0.3365,
      "step": 11123
    },
    {
      "epoch": 1.7413901064495931,
      "grad_norm": 4.6813154220581055,
      "learning_rate": 4.071358748778103e-05,
      "loss": 0.8173,
      "step": 11124
    },
    {
      "epoch": 1.7415466499686914,
      "grad_norm": 2.9903922080993652,
      "learning_rate": 4.06891495601173e-05,
      "loss": 0.7053,
      "step": 11125
    },
    {
      "epoch": 1.7417031934877896,
      "grad_norm": 3.1063003540039062,
      "learning_rate": 4.066471163245356e-05,
      "loss": 0.7405,
      "step": 11126
    },
    {
      "epoch": 1.7418597370068878,
      "grad_norm": 2.323392868041992,
      "learning_rate": 4.0640273704789834e-05,
      "loss": 0.837,
      "step": 11127
    },
    {
      "epoch": 1.7420162805259862,
      "grad_norm": 4.709715843200684,
      "learning_rate": 4.061583577712609e-05,
      "loss": 1.2962,
      "step": 11128
    },
    {
      "epoch": 1.7421728240450847,
      "grad_norm": 3.221708059310913,
      "learning_rate": 4.0591397849462364e-05,
      "loss": 1.254,
      "step": 11129
    },
    {
      "epoch": 1.7423293675641829,
      "grad_norm": 2.867579936981201,
      "learning_rate": 4.056695992179863e-05,
      "loss": 1.2712,
      "step": 11130
    },
    {
      "epoch": 1.742485911083281,
      "grad_norm": 2.7349681854248047,
      "learning_rate": 4.0542521994134894e-05,
      "loss": 1.0635,
      "step": 11131
    },
    {
      "epoch": 1.7426424546023793,
      "grad_norm": 1.872054100036621,
      "learning_rate": 4.051808406647116e-05,
      "loss": 0.8331,
      "step": 11132
    },
    {
      "epoch": 1.7427989981214778,
      "grad_norm": 4.403456211090088,
      "learning_rate": 4.049364613880743e-05,
      "loss": 1.7437,
      "step": 11133
    },
    {
      "epoch": 1.7429555416405762,
      "grad_norm": 2.841325283050537,
      "learning_rate": 4.046920821114369e-05,
      "loss": 0.5395,
      "step": 11134
    },
    {
      "epoch": 1.7431120851596744,
      "grad_norm": NaN,
      "learning_rate": 4.046920821114369e-05,
      "loss": 0.0,
      "step": 11135
    },
    {
      "epoch": 1.7432686286787726,
      "grad_norm": 1.3489454984664917,
      "learning_rate": 4.044477028347996e-05,
      "loss": 0.3682,
      "step": 11136
    },
    {
      "epoch": 1.7434251721978709,
      "grad_norm": 2.201267719268799,
      "learning_rate": 4.0420332355816224e-05,
      "loss": 0.7724,
      "step": 11137
    },
    {
      "epoch": 1.7435817157169693,
      "grad_norm": 2.0831868648529053,
      "learning_rate": 4.039589442815249e-05,
      "loss": 1.2445,
      "step": 11138
    },
    {
      "epoch": 1.7437382592360677,
      "grad_norm": 1.2250694036483765,
      "learning_rate": 4.0371456500488754e-05,
      "loss": 0.2516,
      "step": 11139
    },
    {
      "epoch": 1.743894802755166,
      "grad_norm": 1.6923754215240479,
      "learning_rate": 4.0347018572825025e-05,
      "loss": 0.3832,
      "step": 11140
    },
    {
      "epoch": 1.7440513462742642,
      "grad_norm": 1.450895071029663,
      "learning_rate": 4.0322580645161284e-05,
      "loss": 0.1989,
      "step": 11141
    },
    {
      "epoch": 1.7442078897933626,
      "grad_norm": 2.482814311981201,
      "learning_rate": 4.0298142717497555e-05,
      "loss": 0.1985,
      "step": 11142
    },
    {
      "epoch": 1.7443644333124608,
      "grad_norm": 0.9089357852935791,
      "learning_rate": 4.027370478983382e-05,
      "loss": 0.2036,
      "step": 11143
    },
    {
      "epoch": 1.7445209768315593,
      "grad_norm": 0.5683241486549377,
      "learning_rate": 4.0249266862170085e-05,
      "loss": 0.2729,
      "step": 11144
    },
    {
      "epoch": 1.7446775203506575,
      "grad_norm": 1.6889972686767578,
      "learning_rate": 4.022482893450635e-05,
      "loss": 0.279,
      "step": 11145
    },
    {
      "epoch": 1.7448340638697557,
      "grad_norm": 0.7691648602485657,
      "learning_rate": 4.020039100684262e-05,
      "loss": 0.2574,
      "step": 11146
    },
    {
      "epoch": 1.7449906073888541,
      "grad_norm": 3.3021767139434814,
      "learning_rate": 4.017595307917888e-05,
      "loss": 0.2245,
      "step": 11147
    },
    {
      "epoch": 1.7451471509079524,
      "grad_norm": 1.7656270265579224,
      "learning_rate": 4.015151515151515e-05,
      "loss": 0.327,
      "step": 11148
    },
    {
      "epoch": 1.7453036944270508,
      "grad_norm": 1.371148705482483,
      "learning_rate": 4.0127077223851415e-05,
      "loss": 0.2726,
      "step": 11149
    },
    {
      "epoch": 1.745460237946149,
      "grad_norm": 0.8461998701095581,
      "learning_rate": 4.010263929618768e-05,
      "loss": 0.2061,
      "step": 11150
    },
    {
      "epoch": 1.7456167814652472,
      "grad_norm": 1.068411111831665,
      "learning_rate": 4.0078201368523945e-05,
      "loss": 0.1667,
      "step": 11151
    },
    {
      "epoch": 1.7457733249843457,
      "grad_norm": 1.8549538850784302,
      "learning_rate": 4.005376344086022e-05,
      "loss": 0.3076,
      "step": 11152
    },
    {
      "epoch": 1.7459298685034441,
      "grad_norm": 1.2925175428390503,
      "learning_rate": 4.0029325513196475e-05,
      "loss": 0.4524,
      "step": 11153
    },
    {
      "epoch": 1.7460864120225423,
      "grad_norm": 1.1557775735855103,
      "learning_rate": 4.0004887585532746e-05,
      "loss": 0.2918,
      "step": 11154
    },
    {
      "epoch": 1.7462429555416406,
      "grad_norm": 1.3997137546539307,
      "learning_rate": 3.998044965786901e-05,
      "loss": 0.4363,
      "step": 11155
    },
    {
      "epoch": 1.7463994990607388,
      "grad_norm": 1.8607534170150757,
      "learning_rate": 3.9956011730205276e-05,
      "loss": 0.6176,
      "step": 11156
    },
    {
      "epoch": 1.7465560425798372,
      "grad_norm": 2.017686605453491,
      "learning_rate": 3.993157380254154e-05,
      "loss": 0.6712,
      "step": 11157
    },
    {
      "epoch": 1.7467125860989356,
      "grad_norm": 2.3117191791534424,
      "learning_rate": 3.990713587487781e-05,
      "loss": 0.553,
      "step": 11158
    },
    {
      "epoch": 1.7468691296180339,
      "grad_norm": 2.349550247192383,
      "learning_rate": 3.988269794721407e-05,
      "loss": 0.3159,
      "step": 11159
    },
    {
      "epoch": 1.747025673137132,
      "grad_norm": 2.04817533493042,
      "learning_rate": 3.985826001955034e-05,
      "loss": 0.5679,
      "step": 11160
    },
    {
      "epoch": 1.7471822166562303,
      "grad_norm": 7.92338228225708,
      "learning_rate": 3.983382209188661e-05,
      "loss": 0.4161,
      "step": 11161
    },
    {
      "epoch": 1.7473387601753287,
      "grad_norm": 2.4229795932769775,
      "learning_rate": 3.980938416422287e-05,
      "loss": 0.4948,
      "step": 11162
    },
    {
      "epoch": 1.7474953036944272,
      "grad_norm": 3.308713436126709,
      "learning_rate": 3.9784946236559136e-05,
      "loss": 0.7086,
      "step": 11163
    },
    {
      "epoch": 1.7476518472135254,
      "grad_norm": 1.2768535614013672,
      "learning_rate": 3.976050830889541e-05,
      "loss": 0.3026,
      "step": 11164
    },
    {
      "epoch": 1.7478083907326236,
      "grad_norm": 2.1575565338134766,
      "learning_rate": 3.9736070381231666e-05,
      "loss": 0.6669,
      "step": 11165
    },
    {
      "epoch": 1.7479649342517218,
      "grad_norm": 2.513392448425293,
      "learning_rate": 3.971163245356794e-05,
      "loss": 0.5661,
      "step": 11166
    },
    {
      "epoch": 1.7481214777708203,
      "grad_norm": 2.0377068519592285,
      "learning_rate": 3.96871945259042e-05,
      "loss": 0.3623,
      "step": 11167
    },
    {
      "epoch": 1.7482780212899187,
      "grad_norm": 3.4843387603759766,
      "learning_rate": 3.966275659824047e-05,
      "loss": 1.0925,
      "step": 11168
    },
    {
      "epoch": 1.748434564809017,
      "grad_norm": 2.729309320449829,
      "learning_rate": 3.963831867057673e-05,
      "loss": 0.7483,
      "step": 11169
    },
    {
      "epoch": 1.7485911083281152,
      "grad_norm": 3.7597618103027344,
      "learning_rate": 3.9613880742913e-05,
      "loss": 0.5983,
      "step": 11170
    },
    {
      "epoch": 1.7487476518472134,
      "grad_norm": 3.6038522720336914,
      "learning_rate": 3.958944281524926e-05,
      "loss": 0.7542,
      "step": 11171
    },
    {
      "epoch": 1.7489041953663118,
      "grad_norm": 2.91158390045166,
      "learning_rate": 3.956500488758553e-05,
      "loss": 0.9845,
      "step": 11172
    },
    {
      "epoch": 1.7490607388854102,
      "grad_norm": 5.084097385406494,
      "learning_rate": 3.95405669599218e-05,
      "loss": 0.6904,
      "step": 11173
    },
    {
      "epoch": 1.7492172824045085,
      "grad_norm": 2.9743316173553467,
      "learning_rate": 3.951612903225806e-05,
      "loss": 0.6184,
      "step": 11174
    },
    {
      "epoch": 1.7493738259236067,
      "grad_norm": 5.7044172286987305,
      "learning_rate": 3.949169110459433e-05,
      "loss": 0.5974,
      "step": 11175
    },
    {
      "epoch": 1.7495303694427051,
      "grad_norm": 2.698913097381592,
      "learning_rate": 3.94672531769306e-05,
      "loss": 1.2483,
      "step": 11176
    },
    {
      "epoch": 1.7496869129618033,
      "grad_norm": 3.227285623550415,
      "learning_rate": 3.944281524926686e-05,
      "loss": 0.9042,
      "step": 11177
    },
    {
      "epoch": 1.7498434564809018,
      "grad_norm": 3.3017385005950928,
      "learning_rate": 3.941837732160313e-05,
      "loss": 0.9025,
      "step": 11178
    },
    {
      "epoch": 1.75,
      "grad_norm": 3.2290198802948,
      "learning_rate": 3.939393939393939e-05,
      "loss": 0.437,
      "step": 11179
    },
    {
      "epoch": 1.7501565435190982,
      "grad_norm": 3.120434522628784,
      "learning_rate": 3.936950146627565e-05,
      "loss": 0.7402,
      "step": 11180
    },
    {
      "epoch": 1.7503130870381967,
      "grad_norm": 2.314493417739868,
      "learning_rate": 3.934506353861192e-05,
      "loss": 0.5906,
      "step": 11181
    },
    {
      "epoch": 1.7504696305572949,
      "grad_norm": 1.895832896232605,
      "learning_rate": 3.9320625610948195e-05,
      "loss": 1.206,
      "step": 11182
    },
    {
      "epoch": 1.7506261740763933,
      "grad_norm": 1.9516853094100952,
      "learning_rate": 3.929618768328445e-05,
      "loss": 1.0012,
      "step": 11183
    },
    {
      "epoch": 1.7507827175954915,
      "grad_norm": 1.7214974164962769,
      "learning_rate": 3.9271749755620724e-05,
      "loss": 0.6269,
      "step": 11184
    },
    {
      "epoch": 1.7509392611145898,
      "grad_norm": 2.6578586101531982,
      "learning_rate": 3.924731182795699e-05,
      "loss": 0.8477,
      "step": 11185
    },
    {
      "epoch": 1.7510958046336882,
      "grad_norm": 2.1132657527923584,
      "learning_rate": 3.922287390029325e-05,
      "loss": 0.7678,
      "step": 11186
    },
    {
      "epoch": 1.7512523481527866,
      "grad_norm": 3.0268445014953613,
      "learning_rate": 3.919843597262952e-05,
      "loss": 0.6208,
      "step": 11187
    },
    {
      "epoch": 1.7514088916718848,
      "grad_norm": 4.13357400894165,
      "learning_rate": 3.917399804496579e-05,
      "loss": 0.8431,
      "step": 11188
    },
    {
      "epoch": 1.751565435190983,
      "grad_norm": 0.46277809143066406,
      "learning_rate": 3.914956011730205e-05,
      "loss": 0.1721,
      "step": 11189
    },
    {
      "epoch": 1.7517219787100813,
      "grad_norm": 0.6381204724311829,
      "learning_rate": 3.912512218963832e-05,
      "loss": 0.271,
      "step": 11190
    },
    {
      "epoch": 1.7518785222291797,
      "grad_norm": 1.2064273357391357,
      "learning_rate": 3.9100684261974585e-05,
      "loss": 0.3409,
      "step": 11191
    },
    {
      "epoch": 1.7520350657482782,
      "grad_norm": 0.6451768279075623,
      "learning_rate": 3.907624633431084e-05,
      "loss": 0.2178,
      "step": 11192
    },
    {
      "epoch": 1.7521916092673764,
      "grad_norm": 0.9377066493034363,
      "learning_rate": 3.9051808406647114e-05,
      "loss": 0.2548,
      "step": 11193
    },
    {
      "epoch": 1.7523481527864746,
      "grad_norm": 1.069899320602417,
      "learning_rate": 3.902737047898337e-05,
      "loss": 0.2966,
      "step": 11194
    },
    {
      "epoch": 1.7525046963055728,
      "grad_norm": 1.6345460414886475,
      "learning_rate": 3.9002932551319644e-05,
      "loss": 0.211,
      "step": 11195
    },
    {
      "epoch": 1.7526612398246713,
      "grad_norm": 1.789792537689209,
      "learning_rate": 3.8978494623655915e-05,
      "loss": 0.2923,
      "step": 11196
    },
    {
      "epoch": 1.7528177833437697,
      "grad_norm": 1.5872666835784912,
      "learning_rate": 3.895405669599217e-05,
      "loss": 0.5062,
      "step": 11197
    },
    {
      "epoch": 1.752974326862868,
      "grad_norm": 2.3181369304656982,
      "learning_rate": 3.892961876832844e-05,
      "loss": 0.2508,
      "step": 11198
    },
    {
      "epoch": 1.7531308703819661,
      "grad_norm": 2.8838610649108887,
      "learning_rate": 3.890518084066471e-05,
      "loss": 0.3586,
      "step": 11199
    },
    {
      "epoch": 1.7532874139010644,
      "grad_norm": 1.250127911567688,
      "learning_rate": 3.888074291300097e-05,
      "loss": 0.2989,
      "step": 11200
    },
    {
      "epoch": 1.7534439574201628,
      "grad_norm": 0.9054256677627563,
      "learning_rate": 3.885630498533724e-05,
      "loss": 0.3618,
      "step": 11201
    },
    {
      "epoch": 1.7536005009392612,
      "grad_norm": 3.5090181827545166,
      "learning_rate": 3.8831867057673504e-05,
      "loss": 0.6206,
      "step": 11202
    },
    {
      "epoch": 1.7537570444583594,
      "grad_norm": 1.8899184465408325,
      "learning_rate": 3.880742913000977e-05,
      "loss": 0.26,
      "step": 11203
    },
    {
      "epoch": 1.7539135879774577,
      "grad_norm": 1.9521454572677612,
      "learning_rate": 3.8782991202346034e-05,
      "loss": 0.4928,
      "step": 11204
    },
    {
      "epoch": 1.7540701314965559,
      "grad_norm": 1.03826904296875,
      "learning_rate": 3.8758553274682305e-05,
      "loss": 0.3366,
      "step": 11205
    },
    {
      "epoch": 1.7542266750156543,
      "grad_norm": 0.876063346862793,
      "learning_rate": 3.873411534701856e-05,
      "loss": 0.3439,
      "step": 11206
    },
    {
      "epoch": 1.7543832185347528,
      "grad_norm": 3.456061363220215,
      "learning_rate": 3.8709677419354835e-05,
      "loss": 0.2939,
      "step": 11207
    },
    {
      "epoch": 1.754539762053851,
      "grad_norm": 1.792235016822815,
      "learning_rate": 3.86852394916911e-05,
      "loss": 0.5492,
      "step": 11208
    },
    {
      "epoch": 1.7546963055729492,
      "grad_norm": 8.215913772583008,
      "learning_rate": 3.8660801564027364e-05,
      "loss": 0.6838,
      "step": 11209
    },
    {
      "epoch": 1.7548528490920476,
      "grad_norm": 8.348455429077148,
      "learning_rate": 3.863636363636363e-05,
      "loss": 0.3709,
      "step": 11210
    },
    {
      "epoch": 1.7550093926111459,
      "grad_norm": 1.567761778831482,
      "learning_rate": 3.86119257086999e-05,
      "loss": 0.3818,
      "step": 11211
    },
    {
      "epoch": 1.7551659361302443,
      "grad_norm": 1.7908146381378174,
      "learning_rate": 3.858748778103616e-05,
      "loss": 0.606,
      "step": 11212
    },
    {
      "epoch": 1.7553224796493425,
      "grad_norm": 2.4007041454315186,
      "learning_rate": 3.856304985337243e-05,
      "loss": 0.5006,
      "step": 11213
    },
    {
      "epoch": 1.7554790231684407,
      "grad_norm": 1.7256523370742798,
      "learning_rate": 3.8538611925708695e-05,
      "loss": 0.5744,
      "step": 11214
    },
    {
      "epoch": 1.7556355666875392,
      "grad_norm": 2.6474239826202393,
      "learning_rate": 3.851417399804496e-05,
      "loss": 0.4488,
      "step": 11215
    },
    {
      "epoch": 1.7557921102066374,
      "grad_norm": 6.710626125335693,
      "learning_rate": 3.8489736070381225e-05,
      "loss": 0.4047,
      "step": 11216
    },
    {
      "epoch": 1.7559486537257358,
      "grad_norm": 2.4664697647094727,
      "learning_rate": 3.8465298142717496e-05,
      "loss": 0.735,
      "step": 11217
    },
    {
      "epoch": 1.756105197244834,
      "grad_norm": 1.39060640335083,
      "learning_rate": 3.8440860215053754e-05,
      "loss": 0.4921,
      "step": 11218
    },
    {
      "epoch": 1.7562617407639323,
      "grad_norm": 3.047032117843628,
      "learning_rate": 3.8416422287390026e-05,
      "loss": 0.5779,
      "step": 11219
    },
    {
      "epoch": 1.7564182842830307,
      "grad_norm": 1.9049490690231323,
      "learning_rate": 3.839198435972629e-05,
      "loss": 0.6209,
      "step": 11220
    },
    {
      "epoch": 1.7565748278021291,
      "grad_norm": 2.976374626159668,
      "learning_rate": 3.8367546432062556e-05,
      "loss": 0.4992,
      "step": 11221
    },
    {
      "epoch": 1.7567313713212274,
      "grad_norm": 2.006279706954956,
      "learning_rate": 3.834310850439882e-05,
      "loss": 0.8037,
      "step": 11222
    },
    {
      "epoch": 1.7568879148403256,
      "grad_norm": 4.789150238037109,
      "learning_rate": 3.831867057673509e-05,
      "loss": 0.9672,
      "step": 11223
    },
    {
      "epoch": 1.7570444583594238,
      "grad_norm": 4.674681186676025,
      "learning_rate": 3.829423264907135e-05,
      "loss": 0.5118,
      "step": 11224
    },
    {
      "epoch": 1.7572010018785222,
      "grad_norm": 8.977636337280273,
      "learning_rate": 3.826979472140762e-05,
      "loss": 0.7725,
      "step": 11225
    },
    {
      "epoch": 1.7573575453976207,
      "grad_norm": 3.718141794204712,
      "learning_rate": 3.8245356793743886e-05,
      "loss": 0.5044,
      "step": 11226
    },
    {
      "epoch": 1.757514088916719,
      "grad_norm": 2.4552061557769775,
      "learning_rate": 3.822091886608015e-05,
      "loss": 0.8362,
      "step": 11227
    },
    {
      "epoch": 1.7576706324358171,
      "grad_norm": 4.80755615234375,
      "learning_rate": 3.8196480938416416e-05,
      "loss": 1.0312,
      "step": 11228
    },
    {
      "epoch": 1.7578271759549153,
      "grad_norm": 5.233382225036621,
      "learning_rate": 3.817204301075269e-05,
      "loss": 1.6852,
      "step": 11229
    },
    {
      "epoch": 1.7579837194740138,
      "grad_norm": 5.555087566375732,
      "learning_rate": 3.8147605083088946e-05,
      "loss": 1.3879,
      "step": 11230
    },
    {
      "epoch": 1.7581402629931122,
      "grad_norm": 5.3640923500061035,
      "learning_rate": 3.812316715542522e-05,
      "loss": 1.2419,
      "step": 11231
    },
    {
      "epoch": 1.7582968065122104,
      "grad_norm": 1.4658650159835815,
      "learning_rate": 3.809872922776148e-05,
      "loss": 0.566,
      "step": 11232
    },
    {
      "epoch": 1.7584533500313086,
      "grad_norm": 10.554597854614258,
      "learning_rate": 3.807429130009775e-05,
      "loss": 1.0166,
      "step": 11233
    },
    {
      "epoch": 1.7586098935504069,
      "grad_norm": 2.0296871662139893,
      "learning_rate": 3.804985337243401e-05,
      "loss": 0.2771,
      "step": 11234
    },
    {
      "epoch": 1.7587664370695053,
      "grad_norm": 2.8788039684295654,
      "learning_rate": 3.802541544477028e-05,
      "loss": 0.7007,
      "step": 11235
    },
    {
      "epoch": 1.7589229805886037,
      "grad_norm": 3.588125228881836,
      "learning_rate": 3.800097751710654e-05,
      "loss": 0.7984,
      "step": 11236
    },
    {
      "epoch": 1.759079524107702,
      "grad_norm": 3.308217763900757,
      "learning_rate": 3.797653958944281e-05,
      "loss": 1.3304,
      "step": 11237
    },
    {
      "epoch": 1.7592360676268002,
      "grad_norm": 2.5939760208129883,
      "learning_rate": 3.795210166177908e-05,
      "loss": 1.4925,
      "step": 11238
    },
    {
      "epoch": 1.7593926111458984,
      "grad_norm": 0.596777617931366,
      "learning_rate": 3.792766373411534e-05,
      "loss": 0.2275,
      "step": 11239
    },
    {
      "epoch": 1.7595491546649968,
      "grad_norm": 1.5214952230453491,
      "learning_rate": 3.790322580645161e-05,
      "loss": 0.2228,
      "step": 11240
    },
    {
      "epoch": 1.7597056981840953,
      "grad_norm": 3.9668772220611572,
      "learning_rate": 3.787878787878788e-05,
      "loss": 0.2029,
      "step": 11241
    },
    {
      "epoch": 1.7598622417031935,
      "grad_norm": 2.6096763610839844,
      "learning_rate": 3.785434995112414e-05,
      "loss": 0.6679,
      "step": 11242
    },
    {
      "epoch": 1.7600187852222917,
      "grad_norm": 0.46722492575645447,
      "learning_rate": 3.782991202346041e-05,
      "loss": 0.2772,
      "step": 11243
    },
    {
      "epoch": 1.7601753287413902,
      "grad_norm": 1.0145384073257446,
      "learning_rate": 3.780547409579667e-05,
      "loss": 0.2544,
      "step": 11244
    },
    {
      "epoch": 1.7603318722604884,
      "grad_norm": 0.5798178315162659,
      "learning_rate": 3.778103616813294e-05,
      "loss": 0.2445,
      "step": 11245
    },
    {
      "epoch": 1.7604884157795868,
      "grad_norm": 0.9099991917610168,
      "learning_rate": 3.77565982404692e-05,
      "loss": 0.2651,
      "step": 11246
    },
    {
      "epoch": 1.760644959298685,
      "grad_norm": 0.7325129508972168,
      "learning_rate": 3.7732160312805474e-05,
      "loss": 0.2738,
      "step": 11247
    },
    {
      "epoch": 1.7608015028177832,
      "grad_norm": 0.7348865866661072,
      "learning_rate": 3.770772238514173e-05,
      "loss": 0.2302,
      "step": 11248
    },
    {
      "epoch": 1.7609580463368817,
      "grad_norm": 0.6817787885665894,
      "learning_rate": 3.7683284457478004e-05,
      "loss": 0.2498,
      "step": 11249
    },
    {
      "epoch": 1.7611145898559801,
      "grad_norm": 5.162353038787842,
      "learning_rate": 3.765884652981427e-05,
      "loss": 0.3367,
      "step": 11250
    },
    {
      "epoch": 1.7612711333750783,
      "grad_norm": 0.7174679636955261,
      "learning_rate": 3.7634408602150534e-05,
      "loss": 0.24,
      "step": 11251
    },
    {
      "epoch": 1.7614276768941766,
      "grad_norm": 1.1641637086868286,
      "learning_rate": 3.76099706744868e-05,
      "loss": 0.3054,
      "step": 11252
    },
    {
      "epoch": 1.7615842204132748,
      "grad_norm": 1.7706217765808105,
      "learning_rate": 3.758553274682307e-05,
      "loss": 0.3698,
      "step": 11253
    },
    {
      "epoch": 1.7617407639323732,
      "grad_norm": 0.8290618658065796,
      "learning_rate": 3.756109481915933e-05,
      "loss": 0.2319,
      "step": 11254
    },
    {
      "epoch": 1.7618973074514717,
      "grad_norm": 1.1751283407211304,
      "learning_rate": 3.75366568914956e-05,
      "loss": 0.3196,
      "step": 11255
    },
    {
      "epoch": 1.7620538509705699,
      "grad_norm": 5.826601982116699,
      "learning_rate": 3.7512218963831864e-05,
      "loss": 1.5936,
      "step": 11256
    },
    {
      "epoch": 1.762210394489668,
      "grad_norm": 1.5065512657165527,
      "learning_rate": 3.748778103616813e-05,
      "loss": 0.3376,
      "step": 11257
    },
    {
      "epoch": 1.7623669380087663,
      "grad_norm": 1.3743765354156494,
      "learning_rate": 3.7463343108504394e-05,
      "loss": 0.3552,
      "step": 11258
    },
    {
      "epoch": 1.7625234815278648,
      "grad_norm": 1.3303420543670654,
      "learning_rate": 3.743890518084066e-05,
      "loss": 0.3617,
      "step": 11259
    },
    {
      "epoch": 1.7626800250469632,
      "grad_norm": 0.8232090473175049,
      "learning_rate": 3.741446725317693e-05,
      "loss": 0.3999,
      "step": 11260
    },
    {
      "epoch": 1.7628365685660614,
      "grad_norm": 1.9334932565689087,
      "learning_rate": 3.7390029325513195e-05,
      "loss": 0.4174,
      "step": 11261
    },
    {
      "epoch": 1.7629931120851596,
      "grad_norm": 1.5762377977371216,
      "learning_rate": 3.736559139784946e-05,
      "loss": 0.3955,
      "step": 11262
    },
    {
      "epoch": 1.7631496556042578,
      "grad_norm": 1.6700406074523926,
      "learning_rate": 3.7341153470185725e-05,
      "loss": 0.5943,
      "step": 11263
    },
    {
      "epoch": 1.7633061991233563,
      "grad_norm": 1.3183702230453491,
      "learning_rate": 3.731671554252199e-05,
      "loss": 0.4755,
      "step": 11264
    },
    {
      "epoch": 1.7634627426424547,
      "grad_norm": 1.5598750114440918,
      "learning_rate": 3.7292277614858254e-05,
      "loss": 0.4606,
      "step": 11265
    },
    {
      "epoch": 1.763619286161553,
      "grad_norm": 1.2456644773483276,
      "learning_rate": 3.7267839687194526e-05,
      "loss": 0.4197,
      "step": 11266
    },
    {
      "epoch": 1.7637758296806512,
      "grad_norm": 2.95483660697937,
      "learning_rate": 3.724340175953079e-05,
      "loss": 0.6294,
      "step": 11267
    },
    {
      "epoch": 1.7639323731997494,
      "grad_norm": 1.7271779775619507,
      "learning_rate": 3.7218963831867055e-05,
      "loss": 0.3793,
      "step": 11268
    },
    {
      "epoch": 1.7640889167188478,
      "grad_norm": 3.223212957382202,
      "learning_rate": 3.719452590420332e-05,
      "loss": 0.7018,
      "step": 11269
    },
    {
      "epoch": 1.7642454602379463,
      "grad_norm": 7.701069355010986,
      "learning_rate": 3.7170087976539585e-05,
      "loss": 0.6636,
      "step": 11270
    },
    {
      "epoch": 1.7644020037570445,
      "grad_norm": 1.763765573501587,
      "learning_rate": 3.714565004887585e-05,
      "loss": 0.8444,
      "step": 11271
    },
    {
      "epoch": 1.7645585472761427,
      "grad_norm": 4.53816032409668,
      "learning_rate": 3.712121212121212e-05,
      "loss": 0.8607,
      "step": 11272
    },
    {
      "epoch": 1.764715090795241,
      "grad_norm": 3.1749961376190186,
      "learning_rate": 3.7096774193548386e-05,
      "loss": 0.611,
      "step": 11273
    },
    {
      "epoch": 1.7648716343143394,
      "grad_norm": 5.895051956176758,
      "learning_rate": 3.707233626588465e-05,
      "loss": 1.0795,
      "step": 11274
    },
    {
      "epoch": 1.7650281778334378,
      "grad_norm": 4.479997634887695,
      "learning_rate": 3.7047898338220916e-05,
      "loss": 0.9543,
      "step": 11275
    },
    {
      "epoch": 1.765184721352536,
      "grad_norm": 4.240951061248779,
      "learning_rate": 3.702346041055718e-05,
      "loss": 0.9561,
      "step": 11276
    },
    {
      "epoch": 1.7653412648716342,
      "grad_norm": 3.595991849899292,
      "learning_rate": 3.6999022482893445e-05,
      "loss": 0.8473,
      "step": 11277
    },
    {
      "epoch": 1.7654978083907327,
      "grad_norm": 3.00730037689209,
      "learning_rate": 3.697458455522972e-05,
      "loss": 1.0942,
      "step": 11278
    },
    {
      "epoch": 1.7656543519098309,
      "grad_norm": 2.618523359298706,
      "learning_rate": 3.695014662756598e-05,
      "loss": 0.9541,
      "step": 11279
    },
    {
      "epoch": 1.7658108954289293,
      "grad_norm": 2.935112237930298,
      "learning_rate": 3.692570869990225e-05,
      "loss": 1.0638,
      "step": 11280
    },
    {
      "epoch": 1.7659674389480275,
      "grad_norm": 3.7480924129486084,
      "learning_rate": 3.690127077223851e-05,
      "loss": 1.0236,
      "step": 11281
    },
    {
      "epoch": 1.7661239824671258,
      "grad_norm": 3.3669211864471436,
      "learning_rate": 3.6876832844574776e-05,
      "loss": 1.3881,
      "step": 11282
    },
    {
      "epoch": 1.7662805259862242,
      "grad_norm": 2.294576406478882,
      "learning_rate": 3.685239491691104e-05,
      "loss": 1.0829,
      "step": 11283
    },
    {
      "epoch": 1.7664370695053226,
      "grad_norm": 2.1476564407348633,
      "learning_rate": 3.682795698924731e-05,
      "loss": 1.1662,
      "step": 11284
    },
    {
      "epoch": 1.7665936130244209,
      "grad_norm": 1.508748173713684,
      "learning_rate": 3.680351906158358e-05,
      "loss": 0.613,
      "step": 11285
    },
    {
      "epoch": 1.766750156543519,
      "grad_norm": 1.9185620546340942,
      "learning_rate": 3.677908113391984e-05,
      "loss": 0.6305,
      "step": 11286
    },
    {
      "epoch": 1.7669067000626173,
      "grad_norm": 2.450026750564575,
      "learning_rate": 3.675464320625611e-05,
      "loss": 0.6495,
      "step": 11287
    },
    {
      "epoch": 1.7670632435817157,
      "grad_norm": 2.664174795150757,
      "learning_rate": 3.673020527859237e-05,
      "loss": 0.5036,
      "step": 11288
    },
    {
      "epoch": 1.7672197871008142,
      "grad_norm": 1.3441131114959717,
      "learning_rate": 3.6705767350928637e-05,
      "loss": 0.2408,
      "step": 11289
    },
    {
      "epoch": 1.7673763306199124,
      "grad_norm": 1.2940044403076172,
      "learning_rate": 3.66813294232649e-05,
      "loss": 0.2059,
      "step": 11290
    },
    {
      "epoch": 1.7675328741390106,
      "grad_norm": 0.5093015432357788,
      "learning_rate": 3.665689149560117e-05,
      "loss": 0.1938,
      "step": 11291
    },
    {
      "epoch": 1.7676894176581088,
      "grad_norm": 0.7823761105537415,
      "learning_rate": 3.663245356793744e-05,
      "loss": 0.213,
      "step": 11292
    },
    {
      "epoch": 1.7678459611772073,
      "grad_norm": 1.0145026445388794,
      "learning_rate": 3.66080156402737e-05,
      "loss": 0.3456,
      "step": 11293
    },
    {
      "epoch": 1.7680025046963057,
      "grad_norm": 1.1629821062088013,
      "learning_rate": 3.658357771260997e-05,
      "loss": 0.1895,
      "step": 11294
    },
    {
      "epoch": 1.768159048215404,
      "grad_norm": 1.3927007913589478,
      "learning_rate": 3.655913978494623e-05,
      "loss": 0.3128,
      "step": 11295
    },
    {
      "epoch": 1.7683155917345021,
      "grad_norm": 1.2417762279510498,
      "learning_rate": 3.65347018572825e-05,
      "loss": 0.4427,
      "step": 11296
    },
    {
      "epoch": 1.7684721352536004,
      "grad_norm": 1.2290184497833252,
      "learning_rate": 3.651026392961877e-05,
      "loss": 0.3086,
      "step": 11297
    },
    {
      "epoch": 1.7686286787726988,
      "grad_norm": 1.231251835823059,
      "learning_rate": 3.648582600195503e-05,
      "loss": 0.3525,
      "step": 11298
    },
    {
      "epoch": 1.7687852222917972,
      "grad_norm": 1.0766758918762207,
      "learning_rate": 3.64613880742913e-05,
      "loss": 0.2957,
      "step": 11299
    },
    {
      "epoch": 1.7689417658108955,
      "grad_norm": 0.900638222694397,
      "learning_rate": 3.643695014662756e-05,
      "loss": 0.2731,
      "step": 11300
    },
    {
      "epoch": 1.7690983093299937,
      "grad_norm": 5.631997108459473,
      "learning_rate": 3.641251221896383e-05,
      "loss": 0.4606,
      "step": 11301
    },
    {
      "epoch": 1.769254852849092,
      "grad_norm": 1.9502328634262085,
      "learning_rate": 3.638807429130009e-05,
      "loss": 0.4096,
      "step": 11302
    },
    {
      "epoch": 1.7694113963681903,
      "grad_norm": 2.7071378231048584,
      "learning_rate": 3.6363636363636364e-05,
      "loss": 0.3263,
      "step": 11303
    },
    {
      "epoch": 1.7695679398872888,
      "grad_norm": 2.8107943534851074,
      "learning_rate": 3.633919843597263e-05,
      "loss": 0.3564,
      "step": 11304
    },
    {
      "epoch": 1.769724483406387,
      "grad_norm": 1.432214617729187,
      "learning_rate": 3.6314760508308894e-05,
      "loss": 0.3068,
      "step": 11305
    },
    {
      "epoch": 1.7698810269254852,
      "grad_norm": 6.236049652099609,
      "learning_rate": 3.629032258064516e-05,
      "loss": 0.6622,
      "step": 11306
    },
    {
      "epoch": 1.7700375704445834,
      "grad_norm": 1.3720581531524658,
      "learning_rate": 3.626588465298142e-05,
      "loss": 0.4399,
      "step": 11307
    },
    {
      "epoch": 1.7701941139636819,
      "grad_norm": 6.22260856628418,
      "learning_rate": 3.624144672531769e-05,
      "loss": 0.6246,
      "step": 11308
    },
    {
      "epoch": 1.7703506574827803,
      "grad_norm": 1.9137059450149536,
      "learning_rate": 3.621700879765396e-05,
      "loss": 0.3889,
      "step": 11309
    },
    {
      "epoch": 1.7705072010018785,
      "grad_norm": 2.993480682373047,
      "learning_rate": 3.6192570869990225e-05,
      "loss": 0.5285,
      "step": 11310
    },
    {
      "epoch": 1.7706637445209767,
      "grad_norm": 4.049930095672607,
      "learning_rate": 3.616813294232649e-05,
      "loss": 0.484,
      "step": 11311
    },
    {
      "epoch": 1.7708202880400752,
      "grad_norm": 3.0351674556732178,
      "learning_rate": 3.6143695014662754e-05,
      "loss": 0.4673,
      "step": 11312
    },
    {
      "epoch": 1.7709768315591734,
      "grad_norm": 3.3933756351470947,
      "learning_rate": 3.611925708699902e-05,
      "loss": 0.42,
      "step": 11313
    },
    {
      "epoch": 1.7711333750782718,
      "grad_norm": 2.0746243000030518,
      "learning_rate": 3.6094819159335284e-05,
      "loss": 0.5587,
      "step": 11314
    },
    {
      "epoch": 1.77128991859737,
      "grad_norm": 2.525590658187866,
      "learning_rate": 3.6070381231671555e-05,
      "loss": 0.4801,
      "step": 11315
    },
    {
      "epoch": 1.7714464621164683,
      "grad_norm": 3.3657305240631104,
      "learning_rate": 3.604594330400782e-05,
      "loss": 0.6689,
      "step": 11316
    },
    {
      "epoch": 1.7716030056355667,
      "grad_norm": 2.2926595211029053,
      "learning_rate": 3.602150537634408e-05,
      "loss": 0.5339,
      "step": 11317
    },
    {
      "epoch": 1.7717595491546652,
      "grad_norm": 2.5583808422088623,
      "learning_rate": 3.599706744868035e-05,
      "loss": 0.4416,
      "step": 11318
    },
    {
      "epoch": 1.7719160926737634,
      "grad_norm": 3.0211122035980225,
      "learning_rate": 3.5972629521016615e-05,
      "loss": 0.4616,
      "step": 11319
    },
    {
      "epoch": 1.7720726361928616,
      "grad_norm": 5.442112445831299,
      "learning_rate": 3.594819159335288e-05,
      "loss": 0.6978,
      "step": 11320
    },
    {
      "epoch": 1.7722291797119598,
      "grad_norm": 2.6855411529541016,
      "learning_rate": 3.592375366568915e-05,
      "loss": 0.6472,
      "step": 11321
    },
    {
      "epoch": 1.7723857232310583,
      "grad_norm": 1.714129090309143,
      "learning_rate": 3.5899315738025416e-05,
      "loss": 0.5912,
      "step": 11322
    },
    {
      "epoch": 1.7725422667501567,
      "grad_norm": 5.199019908905029,
      "learning_rate": 3.5874877810361674e-05,
      "loss": 0.7321,
      "step": 11323
    },
    {
      "epoch": 1.772698810269255,
      "grad_norm": 2.025068759918213,
      "learning_rate": 3.5850439882697945e-05,
      "loss": 0.9314,
      "step": 11324
    },
    {
      "epoch": 1.7728553537883531,
      "grad_norm": 3.1110622882843018,
      "learning_rate": 3.582600195503421e-05,
      "loss": 0.6754,
      "step": 11325
    },
    {
      "epoch": 1.7730118973074513,
      "grad_norm": 1.5338470935821533,
      "learning_rate": 3.5801564027370475e-05,
      "loss": 0.6682,
      "step": 11326
    },
    {
      "epoch": 1.7731684408265498,
      "grad_norm": 1.7983545064926147,
      "learning_rate": 3.5777126099706746e-05,
      "loss": 0.6782,
      "step": 11327
    },
    {
      "epoch": 1.7733249843456482,
      "grad_norm": 8.5779390335083,
      "learning_rate": 3.5752688172043004e-05,
      "loss": 0.6843,
      "step": 11328
    },
    {
      "epoch": 1.7734815278647464,
      "grad_norm": 2.1752936840057373,
      "learning_rate": 3.572825024437927e-05,
      "loss": 0.8621,
      "step": 11329
    },
    {
      "epoch": 1.7736380713838447,
      "grad_norm": 4.906739711761475,
      "learning_rate": 3.570381231671554e-05,
      "loss": 0.4116,
      "step": 11330
    },
    {
      "epoch": 1.7737946149029429,
      "grad_norm": 3.4587199687957764,
      "learning_rate": 3.5679374389051806e-05,
      "loss": 1.0547,
      "step": 11331
    },
    {
      "epoch": 1.7739511584220413,
      "grad_norm": 2.0558602809906006,
      "learning_rate": 3.565493646138807e-05,
      "loss": 0.8674,
      "step": 11332
    },
    {
      "epoch": 1.7741077019411398,
      "grad_norm": 6.779541492462158,
      "learning_rate": 3.563049853372434e-05,
      "loss": 1.3265,
      "step": 11333
    },
    {
      "epoch": 1.774264245460238,
      "grad_norm": 4.761612892150879,
      "learning_rate": 3.56060606060606e-05,
      "loss": 0.8637,
      "step": 11334
    },
    {
      "epoch": 1.7744207889793362,
      "grad_norm": 1.2794733047485352,
      "learning_rate": 3.5581622678396865e-05,
      "loss": 0.3855,
      "step": 11335
    },
    {
      "epoch": 1.7745773324984344,
      "grad_norm": 1.7263725996017456,
      "learning_rate": 3.5557184750733136e-05,
      "loss": 0.5897,
      "step": 11336
    },
    {
      "epoch": 1.7747338760175329,
      "grad_norm": 2.5758883953094482,
      "learning_rate": 3.55327468230694e-05,
      "loss": 1.131,
      "step": 11337
    },
    {
      "epoch": 1.7748904195366313,
      "grad_norm": 3.0026354789733887,
      "learning_rate": 3.5508308895405666e-05,
      "loss": 0.8907,
      "step": 11338
    },
    {
      "epoch": 1.7750469630557295,
      "grad_norm": 0.7817611694335938,
      "learning_rate": 3.548387096774193e-05,
      "loss": 0.3081,
      "step": 11339
    },
    {
      "epoch": 1.7752035065748277,
      "grad_norm": 1.6110485792160034,
      "learning_rate": 3.5459433040078196e-05,
      "loss": 0.2291,
      "step": 11340
    },
    {
      "epoch": 1.775360050093926,
      "grad_norm": 1.1875455379486084,
      "learning_rate": 3.543499511241446e-05,
      "loss": 0.2239,
      "step": 11341
    },
    {
      "epoch": 1.7755165936130244,
      "grad_norm": 0.6927761435508728,
      "learning_rate": 3.541055718475073e-05,
      "loss": 0.2251,
      "step": 11342
    },
    {
      "epoch": 1.7756731371321228,
      "grad_norm": 1.9501267671585083,
      "learning_rate": 3.5386119257087e-05,
      "loss": 0.3533,
      "step": 11343
    },
    {
      "epoch": 1.775829680651221,
      "grad_norm": 1.7915222644805908,
      "learning_rate": 3.536168132942326e-05,
      "loss": 0.4284,
      "step": 11344
    },
    {
      "epoch": 1.7759862241703193,
      "grad_norm": 1.3947383165359497,
      "learning_rate": 3.5337243401759526e-05,
      "loss": 0.3234,
      "step": 11345
    },
    {
      "epoch": 1.7761427676894177,
      "grad_norm": 3.555690288543701,
      "learning_rate": 3.531280547409579e-05,
      "loss": 0.3943,
      "step": 11346
    },
    {
      "epoch": 1.776299311208516,
      "grad_norm": 0.8692519068717957,
      "learning_rate": 3.5288367546432056e-05,
      "loss": 0.2929,
      "step": 11347
    },
    {
      "epoch": 1.7764558547276144,
      "grad_norm": 1.724339246749878,
      "learning_rate": 3.526392961876833e-05,
      "loss": 0.3083,
      "step": 11348
    },
    {
      "epoch": 1.7766123982467126,
      "grad_norm": 1.0251134634017944,
      "learning_rate": 3.523949169110459e-05,
      "loss": 0.3695,
      "step": 11349
    },
    {
      "epoch": 1.7767689417658108,
      "grad_norm": 1.2282941341400146,
      "learning_rate": 3.521505376344086e-05,
      "loss": 0.4185,
      "step": 11350
    },
    {
      "epoch": 1.7769254852849092,
      "grad_norm": 0.9312837719917297,
      "learning_rate": 3.519061583577712e-05,
      "loss": 0.2435,
      "step": 11351
    },
    {
      "epoch": 1.7770820288040077,
      "grad_norm": 1.5906509160995483,
      "learning_rate": 3.516617790811339e-05,
      "loss": 0.3492,
      "step": 11352
    },
    {
      "epoch": 1.777238572323106,
      "grad_norm": 0.7517256140708923,
      "learning_rate": 3.514173998044965e-05,
      "loss": 0.2204,
      "step": 11353
    },
    {
      "epoch": 1.777395115842204,
      "grad_norm": 2.044954776763916,
      "learning_rate": 3.5117302052785916e-05,
      "loss": 0.3857,
      "step": 11354
    },
    {
      "epoch": 1.7775516593613023,
      "grad_norm": 5.702359676361084,
      "learning_rate": 3.509286412512219e-05,
      "loss": 1.0261,
      "step": 11355
    },
    {
      "epoch": 1.7777082028804008,
      "grad_norm": 1.4534990787506104,
      "learning_rate": 3.506842619745845e-05,
      "loss": 0.5519,
      "step": 11356
    },
    {
      "epoch": 1.7778647463994992,
      "grad_norm": 1.8530081510543823,
      "learning_rate": 3.504398826979472e-05,
      "loss": 0.474,
      "step": 11357
    },
    {
      "epoch": 1.7780212899185974,
      "grad_norm": 2.5802159309387207,
      "learning_rate": 3.501955034213098e-05,
      "loss": 0.3142,
      "step": 11358
    },
    {
      "epoch": 1.7781778334376956,
      "grad_norm": 0.7793805599212646,
      "learning_rate": 3.499511241446725e-05,
      "loss": 0.4216,
      "step": 11359
    },
    {
      "epoch": 1.7783343769567939,
      "grad_norm": 2.537649631500244,
      "learning_rate": 3.497067448680351e-05,
      "loss": 0.5939,
      "step": 11360
    },
    {
      "epoch": 1.7784909204758923,
      "grad_norm": 3.798841953277588,
      "learning_rate": 3.4946236559139784e-05,
      "loss": 0.4864,
      "step": 11361
    },
    {
      "epoch": 1.7786474639949907,
      "grad_norm": 2.482792377471924,
      "learning_rate": 3.492179863147605e-05,
      "loss": 0.6817,
      "step": 11362
    },
    {
      "epoch": 1.778804007514089,
      "grad_norm": 0.7896741032600403,
      "learning_rate": 3.489736070381231e-05,
      "loss": 0.185,
      "step": 11363
    },
    {
      "epoch": 1.7789605510331872,
      "grad_norm": 2.387078046798706,
      "learning_rate": 3.487292277614858e-05,
      "loss": 0.5607,
      "step": 11364
    },
    {
      "epoch": 1.7791170945522854,
      "grad_norm": 7.810622215270996,
      "learning_rate": 3.484848484848484e-05,
      "loss": 0.525,
      "step": 11365
    },
    {
      "epoch": 1.7792736380713838,
      "grad_norm": 5.911670207977295,
      "learning_rate": 3.482404692082111e-05,
      "loss": 0.6811,
      "step": 11366
    },
    {
      "epoch": 1.7794301815904823,
      "grad_norm": 1.889855146408081,
      "learning_rate": 3.479960899315738e-05,
      "loss": 0.4714,
      "step": 11367
    },
    {
      "epoch": 1.7795867251095805,
      "grad_norm": 1.3557449579238892,
      "learning_rate": 3.4775171065493644e-05,
      "loss": 0.3659,
      "step": 11368
    },
    {
      "epoch": 1.7797432686286787,
      "grad_norm": 3.040102481842041,
      "learning_rate": 3.475073313782991e-05,
      "loss": 0.6288,
      "step": 11369
    },
    {
      "epoch": 1.779899812147777,
      "grad_norm": 7.0420756340026855,
      "learning_rate": 3.4726295210166174e-05,
      "loss": 0.5326,
      "step": 11370
    },
    {
      "epoch": 1.7800563556668754,
      "grad_norm": 2.125647783279419,
      "learning_rate": 3.470185728250244e-05,
      "loss": 0.6711,
      "step": 11371
    },
    {
      "epoch": 1.7802128991859738,
      "grad_norm": 9.777316093444824,
      "learning_rate": 3.46774193548387e-05,
      "loss": 0.509,
      "step": 11372
    },
    {
      "epoch": 1.780369442705072,
      "grad_norm": 2.167560577392578,
      "learning_rate": 3.4652981427174975e-05,
      "loss": 0.7898,
      "step": 11373
    },
    {
      "epoch": 1.7805259862241702,
      "grad_norm": 5.202997207641602,
      "learning_rate": 3.462854349951124e-05,
      "loss": 0.7889,
      "step": 11374
    },
    {
      "epoch": 1.7806825297432687,
      "grad_norm": 3.4150354862213135,
      "learning_rate": 3.4604105571847504e-05,
      "loss": 1.0625,
      "step": 11375
    },
    {
      "epoch": 1.780839073262367,
      "grad_norm": 2.184953451156616,
      "learning_rate": 3.457966764418377e-05,
      "loss": 0.8714,
      "step": 11376
    },
    {
      "epoch": 1.7809956167814653,
      "grad_norm": 2.1605887413024902,
      "learning_rate": 3.4555229716520034e-05,
      "loss": 1.0218,
      "step": 11377
    },
    {
      "epoch": 1.7811521603005636,
      "grad_norm": 3.93656063079834,
      "learning_rate": 3.45307917888563e-05,
      "loss": 1.2488,
      "step": 11378
    },
    {
      "epoch": 1.7813087038196618,
      "grad_norm": 2.7407288551330566,
      "learning_rate": 3.450635386119257e-05,
      "loss": 0.8522,
      "step": 11379
    },
    {
      "epoch": 1.7814652473387602,
      "grad_norm": 2.3924989700317383,
      "learning_rate": 3.4481915933528835e-05,
      "loss": 1.3899,
      "step": 11380
    },
    {
      "epoch": 1.7816217908578584,
      "grad_norm": 2.2750840187072754,
      "learning_rate": 3.44574780058651e-05,
      "loss": 1.1462,
      "step": 11381
    },
    {
      "epoch": 1.7817783343769569,
      "grad_norm": 1.9890302419662476,
      "learning_rate": 3.4433040078201365e-05,
      "loss": 1.0301,
      "step": 11382
    },
    {
      "epoch": 1.781934877896055,
      "grad_norm": 1.8051940202713013,
      "learning_rate": 3.440860215053763e-05,
      "loss": 0.8329,
      "step": 11383
    },
    {
      "epoch": 1.7820914214151533,
      "grad_norm": 2.2022345066070557,
      "learning_rate": 3.4384164222873894e-05,
      "loss": 0.2328,
      "step": 11384
    },
    {
      "epoch": 1.7822479649342517,
      "grad_norm": 2.2471179962158203,
      "learning_rate": 3.4359726295210166e-05,
      "loss": 0.4919,
      "step": 11385
    },
    {
      "epoch": 1.7824045084533502,
      "grad_norm": 3.4314959049224854,
      "learning_rate": 3.433528836754643e-05,
      "loss": 0.7256,
      "step": 11386
    },
    {
      "epoch": 1.7825610519724484,
      "grad_norm": 2.0382392406463623,
      "learning_rate": 3.4310850439882695e-05,
      "loss": 0.6489,
      "step": 11387
    },
    {
      "epoch": 1.7827175954915466,
      "grad_norm": 12.943007469177246,
      "learning_rate": 3.428641251221896e-05,
      "loss": 0.5647,
      "step": 11388
    },
    {
      "epoch": 1.7828741390106448,
      "grad_norm": 3.573213815689087,
      "learning_rate": 3.4261974584555225e-05,
      "loss": 0.234,
      "step": 11389
    },
    {
      "epoch": 1.7830306825297433,
      "grad_norm": 0.828106164932251,
      "learning_rate": 3.423753665689149e-05,
      "loss": 0.2099,
      "step": 11390
    },
    {
      "epoch": 1.7831872260488417,
      "grad_norm": 0.6829121112823486,
      "learning_rate": 3.421309872922776e-05,
      "loss": 0.2266,
      "step": 11391
    },
    {
      "epoch": 1.78334376956794,
      "grad_norm": 0.8972326517105103,
      "learning_rate": 3.4188660801564026e-05,
      "loss": 0.2118,
      "step": 11392
    },
    {
      "epoch": 1.7835003130870382,
      "grad_norm": 2.1261706352233887,
      "learning_rate": 3.416422287390029e-05,
      "loss": 0.2796,
      "step": 11393
    },
    {
      "epoch": 1.7836568566061364,
      "grad_norm": 0.843867838382721,
      "learning_rate": 3.4139784946236556e-05,
      "loss": 0.257,
      "step": 11394
    },
    {
      "epoch": 1.7838134001252348,
      "grad_norm": 0.617863655090332,
      "learning_rate": 3.411534701857282e-05,
      "loss": 0.3085,
      "step": 11395
    },
    {
      "epoch": 1.7839699436443333,
      "grad_norm": 1.1008343696594238,
      "learning_rate": 3.4090909090909085e-05,
      "loss": 0.3067,
      "step": 11396
    },
    {
      "epoch": 1.7841264871634315,
      "grad_norm": 2.097799777984619,
      "learning_rate": 3.406647116324536e-05,
      "loss": 0.2741,
      "step": 11397
    },
    {
      "epoch": 1.7842830306825297,
      "grad_norm": 5.063704967498779,
      "learning_rate": 3.404203323558162e-05,
      "loss": 0.4028,
      "step": 11398
    },
    {
      "epoch": 1.784439574201628,
      "grad_norm": 0.8035471439361572,
      "learning_rate": 3.401759530791789e-05,
      "loss": 0.2896,
      "step": 11399
    },
    {
      "epoch": 1.7845961177207263,
      "grad_norm": 2.3350865840911865,
      "learning_rate": 3.399315738025415e-05,
      "loss": 0.4598,
      "step": 11400
    },
    {
      "epoch": 1.7847526612398248,
      "grad_norm": 1.9042445421218872,
      "learning_rate": 3.3968719452590416e-05,
      "loss": 0.2244,
      "step": 11401
    },
    {
      "epoch": 1.784909204758923,
      "grad_norm": 3.0151047706604004,
      "learning_rate": 3.394428152492668e-05,
      "loss": 0.4026,
      "step": 11402
    },
    {
      "epoch": 1.7850657482780212,
      "grad_norm": 1.309158205986023,
      "learning_rate": 3.391984359726295e-05,
      "loss": 0.4166,
      "step": 11403
    },
    {
      "epoch": 1.7852222917971194,
      "grad_norm": 0.8744258880615234,
      "learning_rate": 3.389540566959922e-05,
      "loss": 0.4382,
      "step": 11404
    },
    {
      "epoch": 1.7853788353162179,
      "grad_norm": 1.478337049484253,
      "learning_rate": 3.387096774193548e-05,
      "loss": 0.3675,
      "step": 11405
    },
    {
      "epoch": 1.7855353788353163,
      "grad_norm": 1.9362159967422485,
      "learning_rate": 3.384652981427175e-05,
      "loss": 0.663,
      "step": 11406
    },
    {
      "epoch": 1.7856919223544145,
      "grad_norm": 1.9572466611862183,
      "learning_rate": 3.382209188660801e-05,
      "loss": 0.7617,
      "step": 11407
    },
    {
      "epoch": 1.7858484658735128,
      "grad_norm": 0.7849676012992859,
      "learning_rate": 3.379765395894428e-05,
      "loss": 0.3307,
      "step": 11408
    },
    {
      "epoch": 1.7860050093926112,
      "grad_norm": 2.1287667751312256,
      "learning_rate": 3.377321603128055e-05,
      "loss": 0.5584,
      "step": 11409
    },
    {
      "epoch": 1.7861615529117094,
      "grad_norm": 1.399261713027954,
      "learning_rate": 3.374877810361681e-05,
      "loss": 0.3862,
      "step": 11410
    },
    {
      "epoch": 1.7863180964308079,
      "grad_norm": 2.173130750656128,
      "learning_rate": 3.372434017595308e-05,
      "loss": 0.5256,
      "step": 11411
    },
    {
      "epoch": 1.786474639949906,
      "grad_norm": 1.911555290222168,
      "learning_rate": 3.369990224828934e-05,
      "loss": 0.7287,
      "step": 11412
    },
    {
      "epoch": 1.7866311834690043,
      "grad_norm": 3.9470415115356445,
      "learning_rate": 3.367546432062561e-05,
      "loss": 0.5673,
      "step": 11413
    },
    {
      "epoch": 1.7867877269881027,
      "grad_norm": 2.7635679244995117,
      "learning_rate": 3.365102639296187e-05,
      "loss": 0.3453,
      "step": 11414
    },
    {
      "epoch": 1.786944270507201,
      "grad_norm": 2.662376880645752,
      "learning_rate": 3.3626588465298144e-05,
      "loss": 0.6953,
      "step": 11415
    },
    {
      "epoch": 1.7871008140262994,
      "grad_norm": 4.1886444091796875,
      "learning_rate": 3.360215053763441e-05,
      "loss": 0.7028,
      "step": 11416
    },
    {
      "epoch": 1.7872573575453976,
      "grad_norm": 3.9045183658599854,
      "learning_rate": 3.357771260997067e-05,
      "loss": 0.7296,
      "step": 11417
    },
    {
      "epoch": 1.7874139010644958,
      "grad_norm": 2.272017240524292,
      "learning_rate": 3.355327468230694e-05,
      "loss": 0.6113,
      "step": 11418
    },
    {
      "epoch": 1.7875704445835943,
      "grad_norm": 2.5397136211395264,
      "learning_rate": 3.35288367546432e-05,
      "loss": 0.3347,
      "step": 11419
    },
    {
      "epoch": 1.7877269881026927,
      "grad_norm": 4.905436992645264,
      "learning_rate": 3.350439882697947e-05,
      "loss": 0.5801,
      "step": 11420
    },
    {
      "epoch": 1.787883531621791,
      "grad_norm": 1.6117602586746216,
      "learning_rate": 3.347996089931573e-05,
      "loss": 0.8952,
      "step": 11421
    },
    {
      "epoch": 1.7880400751408891,
      "grad_norm": 2.91456937789917,
      "learning_rate": 3.3455522971652004e-05,
      "loss": 0.6686,
      "step": 11422
    },
    {
      "epoch": 1.7881966186599874,
      "grad_norm": 3.0732598304748535,
      "learning_rate": 3.343108504398827e-05,
      "loss": 0.6183,
      "step": 11423
    },
    {
      "epoch": 1.7883531621790858,
      "grad_norm": 1.9497085809707642,
      "learning_rate": 3.3406647116324534e-05,
      "loss": 0.7562,
      "step": 11424
    },
    {
      "epoch": 1.7885097056981842,
      "grad_norm": 5.999476909637451,
      "learning_rate": 3.33822091886608e-05,
      "loss": 0.7507,
      "step": 11425
    },
    {
      "epoch": 1.7886662492172825,
      "grad_norm": 2.643611431121826,
      "learning_rate": 3.335777126099706e-05,
      "loss": 0.8865,
      "step": 11426
    },
    {
      "epoch": 1.7888227927363807,
      "grad_norm": 6.0978803634643555,
      "learning_rate": 3.333333333333333e-05,
      "loss": 1.0517,
      "step": 11427
    },
    {
      "epoch": 1.788979336255479,
      "grad_norm": 1.9144535064697266,
      "learning_rate": 3.33088954056696e-05,
      "loss": 0.9786,
      "step": 11428
    },
    {
      "epoch": 1.7891358797745773,
      "grad_norm": 4.194984436035156,
      "learning_rate": 3.3284457478005865e-05,
      "loss": 1.137,
      "step": 11429
    },
    {
      "epoch": 1.7892924232936758,
      "grad_norm": 2.2141382694244385,
      "learning_rate": 3.326001955034213e-05,
      "loss": 1.0596,
      "step": 11430
    },
    {
      "epoch": 1.789448966812774,
      "grad_norm": 1.324143886566162,
      "learning_rate": 3.3235581622678394e-05,
      "loss": 0.6956,
      "step": 11431
    },
    {
      "epoch": 1.7896055103318722,
      "grad_norm": 2.4526519775390625,
      "learning_rate": 3.321114369501466e-05,
      "loss": 0.6576,
      "step": 11432
    },
    {
      "epoch": 1.7897620538509704,
      "grad_norm": 1.4210792779922485,
      "learning_rate": 3.3186705767350924e-05,
      "loss": 0.6591,
      "step": 11433
    },
    {
      "epoch": 1.7899185973700689,
      "grad_norm": 1.1268730163574219,
      "learning_rate": 3.3162267839687195e-05,
      "loss": 0.3686,
      "step": 11434
    },
    {
      "epoch": 1.7900751408891673,
      "grad_norm": 2.319267988204956,
      "learning_rate": 3.313782991202346e-05,
      "loss": 0.7514,
      "step": 11435
    },
    {
      "epoch": 1.7902316844082655,
      "grad_norm": 2.7974939346313477,
      "learning_rate": 3.3113391984359725e-05,
      "loss": 0.7639,
      "step": 11436
    },
    {
      "epoch": 1.7903882279273637,
      "grad_norm": 4.421706676483154,
      "learning_rate": 3.308895405669599e-05,
      "loss": 1.1673,
      "step": 11437
    },
    {
      "epoch": 1.790544771446462,
      "grad_norm": 2.369821310043335,
      "learning_rate": 3.3064516129032255e-05,
      "loss": 0.8485,
      "step": 11438
    },
    {
      "epoch": 1.7907013149655604,
      "grad_norm": 2.404873847961426,
      "learning_rate": 3.304007820136852e-05,
      "loss": 0.3957,
      "step": 11439
    },
    {
      "epoch": 1.7908578584846588,
      "grad_norm": 0.9003664255142212,
      "learning_rate": 3.301564027370479e-05,
      "loss": 0.2733,
      "step": 11440
    },
    {
      "epoch": 1.791014402003757,
      "grad_norm": 0.6666517853736877,
      "learning_rate": 3.2991202346041056e-05,
      "loss": 0.2546,
      "step": 11441
    },
    {
      "epoch": 1.7911709455228553,
      "grad_norm": 0.587860107421875,
      "learning_rate": 3.296676441837732e-05,
      "loss": 0.2187,
      "step": 11442
    },
    {
      "epoch": 1.7913274890419537,
      "grad_norm": 1.2249459028244019,
      "learning_rate": 3.2942326490713585e-05,
      "loss": 0.2001,
      "step": 11443
    },
    {
      "epoch": 1.791484032561052,
      "grad_norm": 1.0504732131958008,
      "learning_rate": 3.291788856304985e-05,
      "loss": 0.3076,
      "step": 11444
    },
    {
      "epoch": 1.7916405760801504,
      "grad_norm": 0.6395697593688965,
      "learning_rate": 3.2893450635386115e-05,
      "loss": 0.2633,
      "step": 11445
    },
    {
      "epoch": 1.7917971195992486,
      "grad_norm": 1.8298180103302002,
      "learning_rate": 3.2869012707722386e-05,
      "loss": 0.4211,
      "step": 11446
    },
    {
      "epoch": 1.7919536631183468,
      "grad_norm": 1.2210631370544434,
      "learning_rate": 3.284457478005865e-05,
      "loss": 0.3673,
      "step": 11447
    },
    {
      "epoch": 1.7921102066374452,
      "grad_norm": 0.7577955722808838,
      "learning_rate": 3.282013685239491e-05,
      "loss": 0.2212,
      "step": 11448
    },
    {
      "epoch": 1.7922667501565435,
      "grad_norm": 1.5126709938049316,
      "learning_rate": 3.279569892473118e-05,
      "loss": 0.294,
      "step": 11449
    },
    {
      "epoch": 1.792423293675642,
      "grad_norm": 1.66423499584198,
      "learning_rate": 3.2771260997067446e-05,
      "loss": 0.236,
      "step": 11450
    },
    {
      "epoch": 1.7925798371947401,
      "grad_norm": 1.5655282735824585,
      "learning_rate": 3.274682306940371e-05,
      "loss": 0.3446,
      "step": 11451
    },
    {
      "epoch": 1.7927363807138383,
      "grad_norm": 0.9023974537849426,
      "learning_rate": 3.272238514173998e-05,
      "loss": 0.3498,
      "step": 11452
    },
    {
      "epoch": 1.7928929242329368,
      "grad_norm": 1.6287553310394287,
      "learning_rate": 3.269794721407625e-05,
      "loss": 0.3794,
      "step": 11453
    },
    {
      "epoch": 1.7930494677520352,
      "grad_norm": 2.9960827827453613,
      "learning_rate": 3.2673509286412505e-05,
      "loss": 0.6226,
      "step": 11454
    },
    {
      "epoch": 1.7932060112711334,
      "grad_norm": 8.331686019897461,
      "learning_rate": 3.2649071358748776e-05,
      "loss": 0.3432,
      "step": 11455
    },
    {
      "epoch": 1.7933625547902317,
      "grad_norm": 1.214149832725525,
      "learning_rate": 3.262463343108504e-05,
      "loss": 0.3093,
      "step": 11456
    },
    {
      "epoch": 1.7935190983093299,
      "grad_norm": 1.9629710912704468,
      "learning_rate": 3.2600195503421306e-05,
      "loss": 0.424,
      "step": 11457
    },
    {
      "epoch": 1.7936756418284283,
      "grad_norm": 4.693020820617676,
      "learning_rate": 3.257575757575758e-05,
      "loss": 0.8316,
      "step": 11458
    },
    {
      "epoch": 1.7938321853475268,
      "grad_norm": 3.069445848464966,
      "learning_rate": 3.2551319648093836e-05,
      "loss": 0.5424,
      "step": 11459
    },
    {
      "epoch": 1.793988728866625,
      "grad_norm": 1.917677879333496,
      "learning_rate": 3.25268817204301e-05,
      "loss": 0.3951,
      "step": 11460
    },
    {
      "epoch": 1.7941452723857232,
      "grad_norm": 1.1094872951507568,
      "learning_rate": 3.250244379276637e-05,
      "loss": 0.4418,
      "step": 11461
    },
    {
      "epoch": 1.7943018159048214,
      "grad_norm": 1.7542659044265747,
      "learning_rate": 3.247800586510264e-05,
      "loss": 0.6939,
      "step": 11462
    },
    {
      "epoch": 1.7944583594239198,
      "grad_norm": 4.421488285064697,
      "learning_rate": 3.24535679374389e-05,
      "loss": 0.5249,
      "step": 11463
    },
    {
      "epoch": 1.7946149029430183,
      "grad_norm": 3.417511224746704,
      "learning_rate": 3.242913000977517e-05,
      "loss": 0.6136,
      "step": 11464
    },
    {
      "epoch": 1.7947714464621165,
      "grad_norm": 4.0269622802734375,
      "learning_rate": 3.240469208211143e-05,
      "loss": 1.0622,
      "step": 11465
    },
    {
      "epoch": 1.7949279899812147,
      "grad_norm": 1.2010209560394287,
      "learning_rate": 3.2380254154447696e-05,
      "loss": 0.371,
      "step": 11466
    },
    {
      "epoch": 1.795084533500313,
      "grad_norm": 3.1197423934936523,
      "learning_rate": 3.235581622678397e-05,
      "loss": 0.6811,
      "step": 11467
    },
    {
      "epoch": 1.7952410770194114,
      "grad_norm": 2.041327714920044,
      "learning_rate": 3.233137829912023e-05,
      "loss": 0.4798,
      "step": 11468
    },
    {
      "epoch": 1.7953976205385098,
      "grad_norm": 2.050049066543579,
      "learning_rate": 3.23069403714565e-05,
      "loss": 0.4145,
      "step": 11469
    },
    {
      "epoch": 1.795554164057608,
      "grad_norm": 5.6066575050354,
      "learning_rate": 3.228250244379277e-05,
      "loss": 0.7301,
      "step": 11470
    },
    {
      "epoch": 1.7957107075767063,
      "grad_norm": 2.6833720207214355,
      "learning_rate": 3.225806451612903e-05,
      "loss": 0.9764,
      "step": 11471
    },
    {
      "epoch": 1.7958672510958045,
      "grad_norm": 1.4853190183639526,
      "learning_rate": 3.223362658846529e-05,
      "loss": 0.3526,
      "step": 11472
    },
    {
      "epoch": 1.796023794614903,
      "grad_norm": 1.8210937976837158,
      "learning_rate": 3.220918866080156e-05,
      "loss": 0.4752,
      "step": 11473
    },
    {
      "epoch": 1.7961803381340014,
      "grad_norm": 2.770622968673706,
      "learning_rate": 3.218475073313783e-05,
      "loss": 0.9654,
      "step": 11474
    },
    {
      "epoch": 1.7963368816530996,
      "grad_norm": 2.7523603439331055,
      "learning_rate": 3.216031280547409e-05,
      "loss": 0.4661,
      "step": 11475
    },
    {
      "epoch": 1.7964934251721978,
      "grad_norm": 4.842382907867432,
      "learning_rate": 3.213587487781036e-05,
      "loss": 0.7784,
      "step": 11476
    },
    {
      "epoch": 1.7966499686912962,
      "grad_norm": 2.4741690158843994,
      "learning_rate": 3.211143695014662e-05,
      "loss": 1.0925,
      "step": 11477
    },
    {
      "epoch": 1.7968065122103944,
      "grad_norm": 1.1925292015075684,
      "learning_rate": 3.208699902248289e-05,
      "loss": 0.5622,
      "step": 11478
    },
    {
      "epoch": 1.7969630557294929,
      "grad_norm": 3.3205652236938477,
      "learning_rate": 3.206256109481916e-05,
      "loss": 1.3511,
      "step": 11479
    },
    {
      "epoch": 1.797119599248591,
      "grad_norm": 3.456491231918335,
      "learning_rate": 3.2038123167155424e-05,
      "loss": 1.5569,
      "step": 11480
    },
    {
      "epoch": 1.7972761427676893,
      "grad_norm": 3.728654146194458,
      "learning_rate": 3.201368523949169e-05,
      "loss": 0.8711,
      "step": 11481
    },
    {
      "epoch": 1.7974326862867878,
      "grad_norm": 2.7066574096679688,
      "learning_rate": 3.198924731182795e-05,
      "loss": 1.4595,
      "step": 11482
    },
    {
      "epoch": 1.7975892298058862,
      "grad_norm": 2.8487234115600586,
      "learning_rate": 3.196480938416422e-05,
      "loss": 1.2391,
      "step": 11483
    },
    {
      "epoch": 1.7977457733249844,
      "grad_norm": 1.6365246772766113,
      "learning_rate": 3.194037145650048e-05,
      "loss": 0.6185,
      "step": 11484
    },
    {
      "epoch": 1.7979023168440826,
      "grad_norm": 0.9667677879333496,
      "learning_rate": 3.191593352883675e-05,
      "loss": 0.3946,
      "step": 11485
    },
    {
      "epoch": 1.7980588603631809,
      "grad_norm": 1.4002686738967896,
      "learning_rate": 3.189149560117302e-05,
      "loss": 0.4003,
      "step": 11486
    },
    {
      "epoch": 1.7982154038822793,
      "grad_norm": 1.4334136247634888,
      "learning_rate": 3.1867057673509284e-05,
      "loss": 0.3993,
      "step": 11487
    },
    {
      "epoch": 1.7983719474013777,
      "grad_norm": 3.9255292415618896,
      "learning_rate": 3.184261974584555e-05,
      "loss": 0.837,
      "step": 11488
    },
    {
      "epoch": 1.798528490920476,
      "grad_norm": 2.0887675285339355,
      "learning_rate": 3.1818181818181814e-05,
      "loss": 0.2766,
      "step": 11489
    },
    {
      "epoch": 1.7986850344395742,
      "grad_norm": 1.7812479734420776,
      "learning_rate": 3.179374389051808e-05,
      "loss": 0.4492,
      "step": 11490
    },
    {
      "epoch": 1.7988415779586724,
      "grad_norm": 1.9391423463821411,
      "learning_rate": 3.176930596285434e-05,
      "loss": 0.2828,
      "step": 11491
    },
    {
      "epoch": 1.7989981214777708,
      "grad_norm": 1.2495836019515991,
      "learning_rate": 3.1744868035190615e-05,
      "loss": 0.3079,
      "step": 11492
    },
    {
      "epoch": 1.7991546649968693,
      "grad_norm": 1.1032434701919556,
      "learning_rate": 3.172043010752688e-05,
      "loss": 0.2513,
      "step": 11493
    },
    {
      "epoch": 1.7993112085159675,
      "grad_norm": 1.8104841709136963,
      "learning_rate": 3.1695992179863144e-05,
      "loss": 0.2143,
      "step": 11494
    },
    {
      "epoch": 1.7994677520350657,
      "grad_norm": 1.4198352098464966,
      "learning_rate": 3.167155425219941e-05,
      "loss": 0.3067,
      "step": 11495
    },
    {
      "epoch": 1.799624295554164,
      "grad_norm": 0.8170778155326843,
      "learning_rate": 3.1647116324535674e-05,
      "loss": 0.2628,
      "step": 11496
    },
    {
      "epoch": 1.7997808390732624,
      "grad_norm": 2.481227159500122,
      "learning_rate": 3.162267839687194e-05,
      "loss": 0.455,
      "step": 11497
    },
    {
      "epoch": 1.7999373825923608,
      "grad_norm": 1.1111358404159546,
      "learning_rate": 3.159824046920821e-05,
      "loss": 0.3932,
      "step": 11498
    },
    {
      "epoch": 1.800093926111459,
      "grad_norm": 2.602691888809204,
      "learning_rate": 3.1573802541544475e-05,
      "loss": 0.4981,
      "step": 11499
    },
    {
      "epoch": 1.8002504696305572,
      "grad_norm": 0.9818456172943115,
      "learning_rate": 3.154936461388074e-05,
      "loss": 0.5675,
      "step": 11500
    },
    {
      "epoch": 1.8004070131496555,
      "grad_norm": 1.2159143686294556,
      "learning_rate": 3.1524926686217005e-05,
      "loss": 0.3184,
      "step": 11501
    },
    {
      "epoch": 1.800563556668754,
      "grad_norm": 2.042997360229492,
      "learning_rate": 3.150048875855327e-05,
      "loss": 0.2769,
      "step": 11502
    },
    {
      "epoch": 1.8007201001878523,
      "grad_norm": 1.841683268547058,
      "learning_rate": 3.1476050830889534e-05,
      "loss": 0.4187,
      "step": 11503
    },
    {
      "epoch": 1.8008766437069506,
      "grad_norm": 1.4786103963851929,
      "learning_rate": 3.1451612903225806e-05,
      "loss": 0.3185,
      "step": 11504
    },
    {
      "epoch": 1.8010331872260488,
      "grad_norm": 1.3413180112838745,
      "learning_rate": 3.142717497556207e-05,
      "loss": 0.4065,
      "step": 11505
    },
    {
      "epoch": 1.801189730745147,
      "grad_norm": 1.192056655883789,
      "learning_rate": 3.1402737047898335e-05,
      "loss": 0.2588,
      "step": 11506
    },
    {
      "epoch": 1.8013462742642454,
      "grad_norm": 3.1147072315216064,
      "learning_rate": 3.13782991202346e-05,
      "loss": 0.3511,
      "step": 11507
    },
    {
      "epoch": 1.8015028177833439,
      "grad_norm": 0.9415939450263977,
      "learning_rate": 3.1353861192570865e-05,
      "loss": 0.3664,
      "step": 11508
    },
    {
      "epoch": 1.801659361302442,
      "grad_norm": 2.014275074005127,
      "learning_rate": 3.132942326490713e-05,
      "loss": 0.5367,
      "step": 11509
    },
    {
      "epoch": 1.8018159048215403,
      "grad_norm": 2.412937879562378,
      "learning_rate": 3.13049853372434e-05,
      "loss": 0.4566,
      "step": 11510
    },
    {
      "epoch": 1.8019724483406387,
      "grad_norm": 1.8484764099121094,
      "learning_rate": 3.1280547409579666e-05,
      "loss": 0.4879,
      "step": 11511
    },
    {
      "epoch": 1.802128991859737,
      "grad_norm": 2.9329864978790283,
      "learning_rate": 3.125610948191593e-05,
      "loss": 0.4734,
      "step": 11512
    },
    {
      "epoch": 1.8022855353788354,
      "grad_norm": 3.0174472332000732,
      "learning_rate": 3.1231671554252196e-05,
      "loss": 0.6371,
      "step": 11513
    },
    {
      "epoch": 1.8024420788979336,
      "grad_norm": 2.343379020690918,
      "learning_rate": 3.120723362658846e-05,
      "loss": 0.5017,
      "step": 11514
    },
    {
      "epoch": 1.8025986224170318,
      "grad_norm": 2.4087560176849365,
      "learning_rate": 3.1182795698924725e-05,
      "loss": 0.5473,
      "step": 11515
    },
    {
      "epoch": 1.8027551659361303,
      "grad_norm": 5.394991874694824,
      "learning_rate": 3.1158357771261e-05,
      "loss": 0.5428,
      "step": 11516
    },
    {
      "epoch": 1.8029117094552287,
      "grad_norm": 1.5382750034332275,
      "learning_rate": 3.113391984359726e-05,
      "loss": 0.5551,
      "step": 11517
    },
    {
      "epoch": 1.803068252974327,
      "grad_norm": 1.568321704864502,
      "learning_rate": 3.110948191593353e-05,
      "loss": 0.4123,
      "step": 11518
    },
    {
      "epoch": 1.8032247964934252,
      "grad_norm": 2.0196034908294678,
      "learning_rate": 3.108504398826979e-05,
      "loss": 0.6364,
      "step": 11519
    },
    {
      "epoch": 1.8033813400125234,
      "grad_norm": 2.843515634536743,
      "learning_rate": 3.1060606060606056e-05,
      "loss": 0.8347,
      "step": 11520
    },
    {
      "epoch": 1.8035378835316218,
      "grad_norm": 2.5024216175079346,
      "learning_rate": 3.103616813294232e-05,
      "loss": 0.7907,
      "step": 11521
    },
    {
      "epoch": 1.8036944270507203,
      "grad_norm": NaN,
      "learning_rate": 3.103616813294232e-05,
      "loss": 0.0,
      "step": 11522
    },
    {
      "epoch": 1.8038509705698185,
      "grad_norm": 2.9168970584869385,
      "learning_rate": 3.101173020527859e-05,
      "loss": 0.9792,
      "step": 11523
    },
    {
      "epoch": 1.8040075140889167,
      "grad_norm": 3.7329742908477783,
      "learning_rate": 3.098729227761486e-05,
      "loss": 0.7339,
      "step": 11524
    },
    {
      "epoch": 1.804164057608015,
      "grad_norm": 4.0727972984313965,
      "learning_rate": 3.096285434995112e-05,
      "loss": 1.2615,
      "step": 11525
    },
    {
      "epoch": 1.8043206011271133,
      "grad_norm": 2.7813782691955566,
      "learning_rate": 3.093841642228739e-05,
      "loss": 0.8195,
      "step": 11526
    },
    {
      "epoch": 1.8044771446462118,
      "grad_norm": 3.9311771392822266,
      "learning_rate": 3.091397849462365e-05,
      "loss": 1.0503,
      "step": 11527
    },
    {
      "epoch": 1.80463368816531,
      "grad_norm": 2.577922821044922,
      "learning_rate": 3.088954056695992e-05,
      "loss": 0.7532,
      "step": 11528
    },
    {
      "epoch": 1.8047902316844082,
      "grad_norm": 5.859790325164795,
      "learning_rate": 3.086510263929619e-05,
      "loss": 1.3083,
      "step": 11529
    },
    {
      "epoch": 1.8049467752035064,
      "grad_norm": 2.1421098709106445,
      "learning_rate": 3.084066471163245e-05,
      "loss": 0.8507,
      "step": 11530
    },
    {
      "epoch": 1.8051033187226049,
      "grad_norm": 3.9190313816070557,
      "learning_rate": 3.081622678396872e-05,
      "loss": 1.1791,
      "step": 11531
    },
    {
      "epoch": 1.8052598622417033,
      "grad_norm": 1.5476529598236084,
      "learning_rate": 3.079178885630498e-05,
      "loss": 0.9817,
      "step": 11532
    },
    {
      "epoch": 1.8054164057608015,
      "grad_norm": 1.8120455741882324,
      "learning_rate": 3.076735092864125e-05,
      "loss": 0.6504,
      "step": 11533
    },
    {
      "epoch": 1.8055729492798998,
      "grad_norm": 1.9655290842056274,
      "learning_rate": 3.074291300097751e-05,
      "loss": 0.5684,
      "step": 11534
    },
    {
      "epoch": 1.805729492798998,
      "grad_norm": 2.028470993041992,
      "learning_rate": 3.0718475073313784e-05,
      "loss": 0.7158,
      "step": 11535
    },
    {
      "epoch": 1.8058860363180964,
      "grad_norm": 1.6118348836898804,
      "learning_rate": 3.069403714565005e-05,
      "loss": 0.5727,
      "step": 11536
    },
    {
      "epoch": 1.8060425798371949,
      "grad_norm": 1.6711117029190063,
      "learning_rate": 3.066959921798631e-05,
      "loss": 0.5635,
      "step": 11537
    },
    {
      "epoch": 1.806199123356293,
      "grad_norm": 1.72394859790802,
      "learning_rate": 3.064516129032258e-05,
      "loss": 0.7681,
      "step": 11538
    },
    {
      "epoch": 1.8063556668753913,
      "grad_norm": 1.3141530752182007,
      "learning_rate": 3.062072336265884e-05,
      "loss": 0.3143,
      "step": 11539
    },
    {
      "epoch": 1.8065122103944895,
      "grad_norm": 2.4097068309783936,
      "learning_rate": 3.059628543499511e-05,
      "loss": 0.2761,
      "step": 11540
    },
    {
      "epoch": 1.806668753913588,
      "grad_norm": 1.3687645196914673,
      "learning_rate": 3.057184750733138e-05,
      "loss": 0.2951,
      "step": 11541
    },
    {
      "epoch": 1.8068252974326864,
      "grad_norm": 1.9082344770431519,
      "learning_rate": 3.0547409579667644e-05,
      "loss": 0.3424,
      "step": 11542
    },
    {
      "epoch": 1.8069818409517846,
      "grad_norm": 1.1238840818405151,
      "learning_rate": 3.052297165200391e-05,
      "loss": 0.3282,
      "step": 11543
    },
    {
      "epoch": 1.8071383844708828,
      "grad_norm": 14.494555473327637,
      "learning_rate": 3.0498533724340174e-05,
      "loss": 0.3696,
      "step": 11544
    },
    {
      "epoch": 1.8072949279899813,
      "grad_norm": 0.5469344854354858,
      "learning_rate": 3.047409579667644e-05,
      "loss": 0.3021,
      "step": 11545
    },
    {
      "epoch": 1.8074514715090795,
      "grad_norm": 0.5608265995979309,
      "learning_rate": 3.0449657869012703e-05,
      "loss": 0.2567,
      "step": 11546
    },
    {
      "epoch": 1.807608015028178,
      "grad_norm": 0.9688588380813599,
      "learning_rate": 3.0425219941348968e-05,
      "loss": 0.3498,
      "step": 11547
    },
    {
      "epoch": 1.8077645585472761,
      "grad_norm": 1.1317929029464722,
      "learning_rate": 3.0400782013685236e-05,
      "loss": 0.2691,
      "step": 11548
    },
    {
      "epoch": 1.8079211020663744,
      "grad_norm": 0.9738819003105164,
      "learning_rate": 3.03763440860215e-05,
      "loss": 0.2949,
      "step": 11549
    },
    {
      "epoch": 1.8080776455854728,
      "grad_norm": 1.1929585933685303,
      "learning_rate": 3.0351906158357766e-05,
      "loss": 0.4013,
      "step": 11550
    },
    {
      "epoch": 1.8082341891045712,
      "grad_norm": 1.0364540815353394,
      "learning_rate": 3.0327468230694034e-05,
      "loss": 0.4509,
      "step": 11551
    },
    {
      "epoch": 1.8083907326236695,
      "grad_norm": 2.415992259979248,
      "learning_rate": 3.03030303030303e-05,
      "loss": 0.4985,
      "step": 11552
    },
    {
      "epoch": 1.8085472761427677,
      "grad_norm": 1.438122272491455,
      "learning_rate": 3.0278592375366564e-05,
      "loss": 0.443,
      "step": 11553
    },
    {
      "epoch": 1.8087038196618659,
      "grad_norm": 1.7964818477630615,
      "learning_rate": 3.0254154447702832e-05,
      "loss": 0.4412,
      "step": 11554
    },
    {
      "epoch": 1.8088603631809643,
      "grad_norm": 4.328434467315674,
      "learning_rate": 3.0229716520039097e-05,
      "loss": 0.4248,
      "step": 11555
    },
    {
      "epoch": 1.8090169067000628,
      "grad_norm": 2.232316017150879,
      "learning_rate": 3.020527859237536e-05,
      "loss": 0.4388,
      "step": 11556
    },
    {
      "epoch": 1.809173450219161,
      "grad_norm": 1.9502800703048706,
      "learning_rate": 3.018084066471163e-05,
      "loss": 0.3977,
      "step": 11557
    },
    {
      "epoch": 1.8093299937382592,
      "grad_norm": 5.183807373046875,
      "learning_rate": 3.0156402737047895e-05,
      "loss": 0.7417,
      "step": 11558
    },
    {
      "epoch": 1.8094865372573574,
      "grad_norm": 4.440307140350342,
      "learning_rate": 3.013196480938416e-05,
      "loss": 0.6555,
      "step": 11559
    },
    {
      "epoch": 1.8096430807764559,
      "grad_norm": 1.0355379581451416,
      "learning_rate": 3.0107526881720428e-05,
      "loss": 0.4009,
      "step": 11560
    },
    {
      "epoch": 1.8097996242955543,
      "grad_norm": 2.554629325866699,
      "learning_rate": 3.0083088954056692e-05,
      "loss": 0.4825,
      "step": 11561
    },
    {
      "epoch": 1.8099561678146525,
      "grad_norm": 1.1384446620941162,
      "learning_rate": 3.0058651026392957e-05,
      "loss": 0.5349,
      "step": 11562
    },
    {
      "epoch": 1.8101127113337507,
      "grad_norm": 1.610360026359558,
      "learning_rate": 3.0034213098729225e-05,
      "loss": 0.3389,
      "step": 11563
    },
    {
      "epoch": 1.810269254852849,
      "grad_norm": 1.8687621355056763,
      "learning_rate": 3.000977517106549e-05,
      "loss": 0.5675,
      "step": 11564
    },
    {
      "epoch": 1.8104257983719474,
      "grad_norm": 4.0817108154296875,
      "learning_rate": 2.9985337243401755e-05,
      "loss": 0.3691,
      "step": 11565
    },
    {
      "epoch": 1.8105823418910458,
      "grad_norm": 5.202278137207031,
      "learning_rate": 2.9960899315738023e-05,
      "loss": 0.6155,
      "step": 11566
    },
    {
      "epoch": 1.810738885410144,
      "grad_norm": 2.133866548538208,
      "learning_rate": 2.9936461388074288e-05,
      "loss": 0.7993,
      "step": 11567
    },
    {
      "epoch": 1.8108954289292423,
      "grad_norm": 3.848477363586426,
      "learning_rate": 2.9912023460410553e-05,
      "loss": 1.0991,
      "step": 11568
    },
    {
      "epoch": 1.8110519724483405,
      "grad_norm": 1.6100995540618896,
      "learning_rate": 2.988758553274682e-05,
      "loss": 0.4259,
      "step": 11569
    },
    {
      "epoch": 1.811208515967439,
      "grad_norm": 3.559994697570801,
      "learning_rate": 2.9863147605083086e-05,
      "loss": 0.5782,
      "step": 11570
    },
    {
      "epoch": 1.8113650594865374,
      "grad_norm": 3.698235034942627,
      "learning_rate": 2.983870967741935e-05,
      "loss": 0.8529,
      "step": 11571
    },
    {
      "epoch": 1.8115216030056356,
      "grad_norm": 3.099733591079712,
      "learning_rate": 2.981427174975562e-05,
      "loss": 0.5965,
      "step": 11572
    },
    {
      "epoch": 1.8116781465247338,
      "grad_norm": 1.290650725364685,
      "learning_rate": 2.9789833822091883e-05,
      "loss": 0.6342,
      "step": 11573
    },
    {
      "epoch": 1.811834690043832,
      "grad_norm": 2.6807796955108643,
      "learning_rate": 2.9765395894428148e-05,
      "loss": 1.0052,
      "step": 11574
    },
    {
      "epoch": 1.8119912335629305,
      "grad_norm": 1.7438832521438599,
      "learning_rate": 2.9740957966764416e-05,
      "loss": 0.6427,
      "step": 11575
    },
    {
      "epoch": 1.812147777082029,
      "grad_norm": 2.453213691711426,
      "learning_rate": 2.971652003910068e-05,
      "loss": 1.0218,
      "step": 11576
    },
    {
      "epoch": 1.8123043206011271,
      "grad_norm": 2.5815582275390625,
      "learning_rate": 2.9692082111436946e-05,
      "loss": 0.3113,
      "step": 11577
    },
    {
      "epoch": 1.8124608641202253,
      "grad_norm": 1.8507500886917114,
      "learning_rate": 2.9667644183773214e-05,
      "loss": 0.7022,
      "step": 11578
    },
    {
      "epoch": 1.8126174076393238,
      "grad_norm": 5.00587797164917,
      "learning_rate": 2.964320625610948e-05,
      "loss": 1.1861,
      "step": 11579
    },
    {
      "epoch": 1.812773951158422,
      "grad_norm": 4.860931873321533,
      "learning_rate": 2.9618768328445744e-05,
      "loss": 0.8452,
      "step": 11580
    },
    {
      "epoch": 1.8129304946775204,
      "grad_norm": 1.9036601781845093,
      "learning_rate": 2.9594330400782012e-05,
      "loss": 0.8327,
      "step": 11581
    },
    {
      "epoch": 1.8130870381966186,
      "grad_norm": 2.3255369663238525,
      "learning_rate": 2.9569892473118277e-05,
      "loss": 1.3153,
      "step": 11582
    },
    {
      "epoch": 1.8132435817157169,
      "grad_norm": 3.8739614486694336,
      "learning_rate": 2.954545454545454e-05,
      "loss": 0.8975,
      "step": 11583
    },
    {
      "epoch": 1.8134001252348153,
      "grad_norm": 2.4773049354553223,
      "learning_rate": 2.952101661779081e-05,
      "loss": 1.295,
      "step": 11584
    },
    {
      "epoch": 1.8135566687539137,
      "grad_norm": 1.24504816532135,
      "learning_rate": 2.9496578690127075e-05,
      "loss": 0.1969,
      "step": 11585
    },
    {
      "epoch": 1.813713212273012,
      "grad_norm": 3.3709805011749268,
      "learning_rate": 2.947214076246334e-05,
      "loss": 0.9211,
      "step": 11586
    },
    {
      "epoch": 1.8138697557921102,
      "grad_norm": 4.954073905944824,
      "learning_rate": 2.9447702834799608e-05,
      "loss": 0.5325,
      "step": 11587
    },
    {
      "epoch": 1.8140262993112084,
      "grad_norm": 2.0279388427734375,
      "learning_rate": 2.9423264907135872e-05,
      "loss": 0.796,
      "step": 11588
    },
    {
      "epoch": 1.8141828428303068,
      "grad_norm": 0.8926810622215271,
      "learning_rate": 2.9398826979472137e-05,
      "loss": 0.3563,
      "step": 11589
    },
    {
      "epoch": 1.8143393863494053,
      "grad_norm": 0.5909395217895508,
      "learning_rate": 2.9374389051808405e-05,
      "loss": 0.347,
      "step": 11590
    },
    {
      "epoch": 1.8144959298685035,
      "grad_norm": 1.289638876914978,
      "learning_rate": 2.934995112414467e-05,
      "loss": 0.3006,
      "step": 11591
    },
    {
      "epoch": 1.8146524733876017,
      "grad_norm": 0.5214605927467346,
      "learning_rate": 2.9325513196480935e-05,
      "loss": 0.2591,
      "step": 11592
    },
    {
      "epoch": 1.8148090169067,
      "grad_norm": 3.6117069721221924,
      "learning_rate": 2.9301075268817203e-05,
      "loss": 0.2681,
      "step": 11593
    },
    {
      "epoch": 1.8149655604257984,
      "grad_norm": 1.9626871347427368,
      "learning_rate": 2.9276637341153468e-05,
      "loss": 0.371,
      "step": 11594
    },
    {
      "epoch": 1.8151221039448968,
      "grad_norm": 0.8882358074188232,
      "learning_rate": 2.9252199413489733e-05,
      "loss": 0.3304,
      "step": 11595
    },
    {
      "epoch": 1.815278647463995,
      "grad_norm": 2.33246111869812,
      "learning_rate": 2.9227761485826e-05,
      "loss": 0.3598,
      "step": 11596
    },
    {
      "epoch": 1.8154351909830932,
      "grad_norm": 1.491754174232483,
      "learning_rate": 2.9203323558162266e-05,
      "loss": 0.3543,
      "step": 11597
    },
    {
      "epoch": 1.8155917345021915,
      "grad_norm": 1.187076449394226,
      "learning_rate": 2.917888563049853e-05,
      "loss": 0.3254,
      "step": 11598
    },
    {
      "epoch": 1.81574827802129,
      "grad_norm": 0.9200171232223511,
      "learning_rate": 2.91544477028348e-05,
      "loss": 0.3543,
      "step": 11599
    },
    {
      "epoch": 1.8159048215403883,
      "grad_norm": 1.1525431871414185,
      "learning_rate": 2.9130009775171064e-05,
      "loss": 0.3431,
      "step": 11600
    },
    {
      "epoch": 1.8160613650594866,
      "grad_norm": 0.8428003787994385,
      "learning_rate": 2.910557184750733e-05,
      "loss": 0.2869,
      "step": 11601
    },
    {
      "epoch": 1.8162179085785848,
      "grad_norm": 2.701383590698242,
      "learning_rate": 2.9081133919843597e-05,
      "loss": 0.585,
      "step": 11602
    },
    {
      "epoch": 1.816374452097683,
      "grad_norm": 1.2327849864959717,
      "learning_rate": 2.905669599217986e-05,
      "loss": 0.3806,
      "step": 11603
    },
    {
      "epoch": 1.8165309956167814,
      "grad_norm": 1.1938436031341553,
      "learning_rate": 2.9032258064516126e-05,
      "loss": 0.5627,
      "step": 11604
    },
    {
      "epoch": 1.8166875391358799,
      "grad_norm": 1.1973328590393066,
      "learning_rate": 2.9007820136852394e-05,
      "loss": 0.4786,
      "step": 11605
    },
    {
      "epoch": 1.816844082654978,
      "grad_norm": 7.545814514160156,
      "learning_rate": 2.898338220918866e-05,
      "loss": 0.4375,
      "step": 11606
    },
    {
      "epoch": 1.8170006261740763,
      "grad_norm": 3.719249963760376,
      "learning_rate": 2.8958944281524924e-05,
      "loss": 0.607,
      "step": 11607
    },
    {
      "epoch": 1.8171571696931748,
      "grad_norm": 1.4751845598220825,
      "learning_rate": 2.8934506353861192e-05,
      "loss": 0.5607,
      "step": 11608
    },
    {
      "epoch": 1.817313713212273,
      "grad_norm": 9.847101211547852,
      "learning_rate": 2.8910068426197457e-05,
      "loss": 0.5926,
      "step": 11609
    },
    {
      "epoch": 1.8174702567313714,
      "grad_norm": 1.2304680347442627,
      "learning_rate": 2.8885630498533722e-05,
      "loss": 0.4732,
      "step": 11610
    },
    {
      "epoch": 1.8176268002504696,
      "grad_norm": 4.059071063995361,
      "learning_rate": 2.8861192570869987e-05,
      "loss": 0.338,
      "step": 11611
    },
    {
      "epoch": 1.8177833437695678,
      "grad_norm": 1.5331727266311646,
      "learning_rate": 2.8836754643206255e-05,
      "loss": 0.4844,
      "step": 11612
    },
    {
      "epoch": 1.8179398872886663,
      "grad_norm": 2.8758628368377686,
      "learning_rate": 2.881231671554252e-05,
      "loss": 0.5759,
      "step": 11613
    },
    {
      "epoch": 1.8180964308077645,
      "grad_norm": 4.602398872375488,
      "learning_rate": 2.8787878787878784e-05,
      "loss": 0.4847,
      "step": 11614
    },
    {
      "epoch": 1.818252974326863,
      "grad_norm": 5.723301410675049,
      "learning_rate": 2.8763440860215053e-05,
      "loss": 1.2581,
      "step": 11615
    },
    {
      "epoch": 1.8184095178459612,
      "grad_norm": 2.905620813369751,
      "learning_rate": 2.8739002932551317e-05,
      "loss": 0.5796,
      "step": 11616
    },
    {
      "epoch": 1.8185660613650594,
      "grad_norm": 2.289865732192993,
      "learning_rate": 2.8714565004887582e-05,
      "loss": 0.6131,
      "step": 11617
    },
    {
      "epoch": 1.8187226048841578,
      "grad_norm": 1.782109022140503,
      "learning_rate": 2.869012707722385e-05,
      "loss": 0.523,
      "step": 11618
    },
    {
      "epoch": 1.8188791484032563,
      "grad_norm": 8.425079345703125,
      "learning_rate": 2.8665689149560115e-05,
      "loss": 0.9899,
      "step": 11619
    },
    {
      "epoch": 1.8190356919223545,
      "grad_norm": 2.678179979324341,
      "learning_rate": 2.864125122189638e-05,
      "loss": 0.7031,
      "step": 11620
    },
    {
      "epoch": 1.8191922354414527,
      "grad_norm": 9.00810432434082,
      "learning_rate": 2.8616813294232648e-05,
      "loss": 0.5054,
      "step": 11621
    },
    {
      "epoch": 1.819348778960551,
      "grad_norm": 2.7023744583129883,
      "learning_rate": 2.8592375366568913e-05,
      "loss": 0.8168,
      "step": 11622
    },
    {
      "epoch": 1.8195053224796494,
      "grad_norm": 2.590367078781128,
      "learning_rate": 2.8567937438905178e-05,
      "loss": 0.7713,
      "step": 11623
    },
    {
      "epoch": 1.8196618659987478,
      "grad_norm": 5.677480220794678,
      "learning_rate": 2.8543499511241446e-05,
      "loss": 0.5236,
      "step": 11624
    },
    {
      "epoch": 1.819818409517846,
      "grad_norm": 5.0781426429748535,
      "learning_rate": 2.851906158357771e-05,
      "loss": 0.6324,
      "step": 11625
    },
    {
      "epoch": 1.8199749530369442,
      "grad_norm": 3.239406108856201,
      "learning_rate": 2.8494623655913975e-05,
      "loss": 0.4184,
      "step": 11626
    },
    {
      "epoch": 1.8201314965560424,
      "grad_norm": 2.7294325828552246,
      "learning_rate": 2.8470185728250244e-05,
      "loss": 1.0267,
      "step": 11627
    },
    {
      "epoch": 1.820288040075141,
      "grad_norm": 3.9341413974761963,
      "learning_rate": 2.844574780058651e-05,
      "loss": 1.3683,
      "step": 11628
    },
    {
      "epoch": 1.8204445835942393,
      "grad_norm": 4.064838409423828,
      "learning_rate": 2.8421309872922773e-05,
      "loss": 1.1582,
      "step": 11629
    },
    {
      "epoch": 1.8206011271133375,
      "grad_norm": 2.5269718170166016,
      "learning_rate": 2.839687194525904e-05,
      "loss": 0.4504,
      "step": 11630
    },
    {
      "epoch": 1.8207576706324358,
      "grad_norm": 2.9901621341705322,
      "learning_rate": 2.8372434017595306e-05,
      "loss": 1.0721,
      "step": 11631
    },
    {
      "epoch": 1.820914214151534,
      "grad_norm": 2.5948750972747803,
      "learning_rate": 2.834799608993157e-05,
      "loss": 0.8848,
      "step": 11632
    },
    {
      "epoch": 1.8210707576706324,
      "grad_norm": 7.479366779327393,
      "learning_rate": 2.832355816226784e-05,
      "loss": 0.9791,
      "step": 11633
    },
    {
      "epoch": 1.8212273011897309,
      "grad_norm": 6.228923797607422,
      "learning_rate": 2.8299120234604104e-05,
      "loss": 1.3042,
      "step": 11634
    },
    {
      "epoch": 1.821383844708829,
      "grad_norm": 2.04010272026062,
      "learning_rate": 2.827468230694037e-05,
      "loss": 0.7378,
      "step": 11635
    },
    {
      "epoch": 1.8215403882279273,
      "grad_norm": 3.85182523727417,
      "learning_rate": 2.8250244379276637e-05,
      "loss": 0.6142,
      "step": 11636
    },
    {
      "epoch": 1.8216969317470255,
      "grad_norm": 2.9816079139709473,
      "learning_rate": 2.8225806451612902e-05,
      "loss": 0.6967,
      "step": 11637
    },
    {
      "epoch": 1.821853475266124,
      "grad_norm": 2.144773483276367,
      "learning_rate": 2.8201368523949167e-05,
      "loss": 0.9379,
      "step": 11638
    },
    {
      "epoch": 1.8220100187852224,
      "grad_norm": 4.1807661056518555,
      "learning_rate": 2.8176930596285435e-05,
      "loss": 0.3605,
      "step": 11639
    },
    {
      "epoch": 1.8221665623043206,
      "grad_norm": 1.5326061248779297,
      "learning_rate": 2.81524926686217e-05,
      "loss": 0.345,
      "step": 11640
    },
    {
      "epoch": 1.8223231058234188,
      "grad_norm": 0.7716324925422668,
      "learning_rate": 2.8128054740957964e-05,
      "loss": 0.3745,
      "step": 11641
    },
    {
      "epoch": 1.8224796493425173,
      "grad_norm": 0.9222685098648071,
      "learning_rate": 2.8103616813294233e-05,
      "loss": 0.4254,
      "step": 11642
    },
    {
      "epoch": 1.8226361928616155,
      "grad_norm": 0.9720975160598755,
      "learning_rate": 2.8079178885630497e-05,
      "loss": 0.3535,
      "step": 11643
    },
    {
      "epoch": 1.822792736380714,
      "grad_norm": 1.215984582901001,
      "learning_rate": 2.8054740957966762e-05,
      "loss": 0.4628,
      "step": 11644
    },
    {
      "epoch": 1.8229492798998121,
      "grad_norm": 1.166796326637268,
      "learning_rate": 2.803030303030303e-05,
      "loss": 0.4993,
      "step": 11645
    },
    {
      "epoch": 1.8231058234189104,
      "grad_norm": 1.0708175897598267,
      "learning_rate": 2.8005865102639295e-05,
      "loss": 0.4912,
      "step": 11646
    },
    {
      "epoch": 1.8232623669380088,
      "grad_norm": 1.6838656663894653,
      "learning_rate": 2.798142717497556e-05,
      "loss": 0.5915,
      "step": 11647
    },
    {
      "epoch": 1.823418910457107,
      "grad_norm": 1.128220558166504,
      "learning_rate": 2.7956989247311828e-05,
      "loss": 0.4936,
      "step": 11648
    },
    {
      "epoch": 1.8235754539762055,
      "grad_norm": 2.355700969696045,
      "learning_rate": 2.7932551319648093e-05,
      "loss": 0.7338,
      "step": 11649
    },
    {
      "epoch": 1.8237319974953037,
      "grad_norm": 2.227602481842041,
      "learning_rate": 2.7908113391984354e-05,
      "loss": 0.8051,
      "step": 11650
    },
    {
      "epoch": 1.823888541014402,
      "grad_norm": 1.5010913610458374,
      "learning_rate": 2.7883675464320626e-05,
      "loss": 0.9211,
      "step": 11651
    },
    {
      "epoch": 1.8240450845335003,
      "grad_norm": 2.1629629135131836,
      "learning_rate": 2.785923753665689e-05,
      "loss": 0.7823,
      "step": 11652
    },
    {
      "epoch": 1.8242016280525988,
      "grad_norm": 1.6575279235839844,
      "learning_rate": 2.7834799608993152e-05,
      "loss": 0.7279,
      "step": 11653
    },
    {
      "epoch": 1.824358171571697,
      "grad_norm": 1.6518474817276,
      "learning_rate": 2.7810361681329424e-05,
      "loss": 0.768,
      "step": 11654
    },
    {
      "epoch": 1.8245147150907952,
      "grad_norm": 2.1042938232421875,
      "learning_rate": 2.778592375366569e-05,
      "loss": 0.551,
      "step": 11655
    },
    {
      "epoch": 1.8246712586098934,
      "grad_norm": 5.784224987030029,
      "learning_rate": 2.776148582600195e-05,
      "loss": 0.6896,
      "step": 11656
    },
    {
      "epoch": 1.8248278021289919,
      "grad_norm": 1.9954174757003784,
      "learning_rate": 2.773704789833822e-05,
      "loss": 0.5671,
      "step": 11657
    },
    {
      "epoch": 1.8249843456480903,
      "grad_norm": 1.8126837015151978,
      "learning_rate": 2.7712609970674486e-05,
      "loss": 0.5581,
      "step": 11658
    },
    {
      "epoch": 1.8251408891671885,
      "grad_norm": 7.1128106117248535,
      "learning_rate": 2.7688172043010748e-05,
      "loss": 0.4665,
      "step": 11659
    },
    {
      "epoch": 1.8252974326862867,
      "grad_norm": 3.6487762928009033,
      "learning_rate": 2.766373411534702e-05,
      "loss": 0.6816,
      "step": 11660
    },
    {
      "epoch": 1.825453976205385,
      "grad_norm": 5.3055338859558105,
      "learning_rate": 2.763929618768328e-05,
      "loss": 0.7622,
      "step": 11661
    },
    {
      "epoch": 1.8256105197244834,
      "grad_norm": 2.0376229286193848,
      "learning_rate": 2.7614858260019546e-05,
      "loss": 0.4507,
      "step": 11662
    },
    {
      "epoch": 1.8257670632435818,
      "grad_norm": 2.7106873989105225,
      "learning_rate": 2.7590420332355817e-05,
      "loss": 0.4463,
      "step": 11663
    },
    {
      "epoch": 1.82592360676268,
      "grad_norm": 2.749912977218628,
      "learning_rate": 2.756598240469208e-05,
      "loss": 0.3819,
      "step": 11664
    },
    {
      "epoch": 1.8260801502817783,
      "grad_norm": 4.844040393829346,
      "learning_rate": 2.7541544477028343e-05,
      "loss": 0.4986,
      "step": 11665
    },
    {
      "epoch": 1.8262366938008765,
      "grad_norm": 2.121793746948242,
      "learning_rate": 2.7517106549364615e-05,
      "loss": 0.2946,
      "step": 11666
    },
    {
      "epoch": 1.826393237319975,
      "grad_norm": 2.5618605613708496,
      "learning_rate": 2.7492668621700876e-05,
      "loss": 0.6871,
      "step": 11667
    },
    {
      "epoch": 1.8265497808390734,
      "grad_norm": 6.57215690612793,
      "learning_rate": 2.746823069403714e-05,
      "loss": 0.4507,
      "step": 11668
    },
    {
      "epoch": 1.8267063243581716,
      "grad_norm": 1.615665316581726,
      "learning_rate": 2.7443792766373413e-05,
      "loss": 0.4263,
      "step": 11669
    },
    {
      "epoch": 1.8268628678772698,
      "grad_norm": 2.805304765701294,
      "learning_rate": 2.7419354838709674e-05,
      "loss": 0.9011,
      "step": 11670
    },
    {
      "epoch": 1.827019411396368,
      "grad_norm": 2.946972370147705,
      "learning_rate": 2.739491691104594e-05,
      "loss": 0.4498,
      "step": 11671
    },
    {
      "epoch": 1.8271759549154665,
      "grad_norm": 2.347245693206787,
      "learning_rate": 2.737047898338221e-05,
      "loss": 0.9303,
      "step": 11672
    },
    {
      "epoch": 1.827332498434565,
      "grad_norm": 2.6300110816955566,
      "learning_rate": 2.7346041055718472e-05,
      "loss": 1.0643,
      "step": 11673
    },
    {
      "epoch": 1.8274890419536631,
      "grad_norm": 5.476263046264648,
      "learning_rate": 2.7321603128054737e-05,
      "loss": 1.1481,
      "step": 11674
    },
    {
      "epoch": 1.8276455854727613,
      "grad_norm": 4.003941059112549,
      "learning_rate": 2.7297165200391e-05,
      "loss": 0.8726,
      "step": 11675
    },
    {
      "epoch": 1.8278021289918598,
      "grad_norm": 3.651909351348877,
      "learning_rate": 2.727272727272727e-05,
      "loss": 0.7593,
      "step": 11676
    },
    {
      "epoch": 1.827958672510958,
      "grad_norm": 5.829402923583984,
      "learning_rate": 2.7248289345063535e-05,
      "loss": 1.0815,
      "step": 11677
    },
    {
      "epoch": 1.8281152160300564,
      "grad_norm": 4.480926036834717,
      "learning_rate": 2.72238514173998e-05,
      "loss": 0.8428,
      "step": 11678
    },
    {
      "epoch": 1.8282717595491547,
      "grad_norm": 3.8357913494110107,
      "learning_rate": 2.7199413489736068e-05,
      "loss": 1.0317,
      "step": 11679
    },
    {
      "epoch": 1.8284283030682529,
      "grad_norm": 6.306792259216309,
      "learning_rate": 2.7174975562072332e-05,
      "loss": 1.6432,
      "step": 11680
    },
    {
      "epoch": 1.8285848465873513,
      "grad_norm": 3.932875156402588,
      "learning_rate": 2.7150537634408597e-05,
      "loss": 1.1734,
      "step": 11681
    },
    {
      "epoch": 1.8287413901064495,
      "grad_norm": 3.3402152061462402,
      "learning_rate": 2.7126099706744865e-05,
      "loss": 1.1751,
      "step": 11682
    },
    {
      "epoch": 1.828897933625548,
      "grad_norm": 4.163675785064697,
      "learning_rate": 2.710166177908113e-05,
      "loss": 1.3794,
      "step": 11683
    },
    {
      "epoch": 1.8290544771446462,
      "grad_norm": 2.3751349449157715,
      "learning_rate": 2.7077223851417395e-05,
      "loss": 0.7359,
      "step": 11684
    },
    {
      "epoch": 1.8292110206637444,
      "grad_norm": 1.7543164491653442,
      "learning_rate": 2.7052785923753663e-05,
      "loss": 0.4729,
      "step": 11685
    },
    {
      "epoch": 1.8293675641828429,
      "grad_norm": 3.514760732650757,
      "learning_rate": 2.7028347996089928e-05,
      "loss": 0.4018,
      "step": 11686
    },
    {
      "epoch": 1.8295241077019413,
      "grad_norm": 2.0676984786987305,
      "learning_rate": 2.7003910068426193e-05,
      "loss": 0.6138,
      "step": 11687
    },
    {
      "epoch": 1.8296806512210395,
      "grad_norm": 2.1401822566986084,
      "learning_rate": 2.697947214076246e-05,
      "loss": 0.8691,
      "step": 11688
    },
    {
      "epoch": 1.8298371947401377,
      "grad_norm": 1.3781170845031738,
      "learning_rate": 2.6955034213098726e-05,
      "loss": 0.3156,
      "step": 11689
    },
    {
      "epoch": 1.829993738259236,
      "grad_norm": 3.473524808883667,
      "learning_rate": 2.693059628543499e-05,
      "loss": 0.3452,
      "step": 11690
    },
    {
      "epoch": 1.8301502817783344,
      "grad_norm": 2.099724531173706,
      "learning_rate": 2.690615835777126e-05,
      "loss": 0.2908,
      "step": 11691
    },
    {
      "epoch": 1.8303068252974328,
      "grad_norm": 1.1955111026763916,
      "learning_rate": 2.6881720430107523e-05,
      "loss": 0.3801,
      "step": 11692
    },
    {
      "epoch": 1.830463368816531,
      "grad_norm": 1.2508939504623413,
      "learning_rate": 2.6857282502443788e-05,
      "loss": 0.3253,
      "step": 11693
    },
    {
      "epoch": 1.8306199123356293,
      "grad_norm": 1.153827428817749,
      "learning_rate": 2.6832844574780056e-05,
      "loss": 0.3607,
      "step": 11694
    },
    {
      "epoch": 1.8307764558547275,
      "grad_norm": 0.8758578300476074,
      "learning_rate": 2.680840664711632e-05,
      "loss": 0.411,
      "step": 11695
    },
    {
      "epoch": 1.830932999373826,
      "grad_norm": 1.912964105606079,
      "learning_rate": 2.6783968719452586e-05,
      "loss": 0.3108,
      "step": 11696
    },
    {
      "epoch": 1.8310895428929244,
      "grad_norm": 1.8823193311691284,
      "learning_rate": 2.6759530791788854e-05,
      "loss": 0.4525,
      "step": 11697
    },
    {
      "epoch": 1.8312460864120226,
      "grad_norm": 1.5550634860992432,
      "learning_rate": 2.673509286412512e-05,
      "loss": 0.5261,
      "step": 11698
    },
    {
      "epoch": 1.8314026299311208,
      "grad_norm": 1.8203532695770264,
      "learning_rate": 2.6710654936461384e-05,
      "loss": 0.3801,
      "step": 11699
    },
    {
      "epoch": 1.831559173450219,
      "grad_norm": 3.2406249046325684,
      "learning_rate": 2.6686217008797652e-05,
      "loss": 0.4729,
      "step": 11700
    },
    {
      "epoch": 1.8317157169693175,
      "grad_norm": 0.9475135207176208,
      "learning_rate": 2.6661779081133917e-05,
      "loss": 0.4572,
      "step": 11701
    },
    {
      "epoch": 1.831872260488416,
      "grad_norm": 1.6032726764678955,
      "learning_rate": 2.663734115347018e-05,
      "loss": 0.5409,
      "step": 11702
    },
    {
      "epoch": 1.8320288040075141,
      "grad_norm": 1.069624423980713,
      "learning_rate": 2.661290322580645e-05,
      "loss": 0.4271,
      "step": 11703
    },
    {
      "epoch": 1.8321853475266123,
      "grad_norm": 1.8359605073928833,
      "learning_rate": 2.6588465298142715e-05,
      "loss": 0.4135,
      "step": 11704
    },
    {
      "epoch": 1.8323418910457105,
      "grad_norm": 4.056179046630859,
      "learning_rate": 2.656402737047898e-05,
      "loss": 0.6246,
      "step": 11705
    },
    {
      "epoch": 1.832498434564809,
      "grad_norm": 1.7407777309417725,
      "learning_rate": 2.6539589442815248e-05,
      "loss": 0.4708,
      "step": 11706
    },
    {
      "epoch": 1.8326549780839074,
      "grad_norm": 1.6690481901168823,
      "learning_rate": 2.6515151515151512e-05,
      "loss": 0.5088,
      "step": 11707
    },
    {
      "epoch": 1.8328115216030056,
      "grad_norm": 1.289447546005249,
      "learning_rate": 2.6490713587487777e-05,
      "loss": 0.6305,
      "step": 11708
    },
    {
      "epoch": 1.8329680651221039,
      "grad_norm": 2.4667999744415283,
      "learning_rate": 2.6466275659824045e-05,
      "loss": 0.7835,
      "step": 11709
    },
    {
      "epoch": 1.8331246086412023,
      "grad_norm": 1.8310896158218384,
      "learning_rate": 2.644183773216031e-05,
      "loss": 0.6884,
      "step": 11710
    },
    {
      "epoch": 1.8332811521603005,
      "grad_norm": 1.9276304244995117,
      "learning_rate": 2.6417399804496575e-05,
      "loss": 0.7651,
      "step": 11711
    },
    {
      "epoch": 1.833437695679399,
      "grad_norm": 2.878755807876587,
      "learning_rate": 2.6392961876832843e-05,
      "loss": 0.5969,
      "step": 11712
    },
    {
      "epoch": 1.8335942391984972,
      "grad_norm": 1.5785597562789917,
      "learning_rate": 2.6368523949169108e-05,
      "loss": 0.6253,
      "step": 11713
    },
    {
      "epoch": 1.8337507827175954,
      "grad_norm": 1.4301279783248901,
      "learning_rate": 2.6344086021505373e-05,
      "loss": 0.4936,
      "step": 11714
    },
    {
      "epoch": 1.8339073262366938,
      "grad_norm": 3.3966879844665527,
      "learning_rate": 2.631964809384164e-05,
      "loss": 0.7848,
      "step": 11715
    },
    {
      "epoch": 1.8340638697557923,
      "grad_norm": 5.173041820526123,
      "learning_rate": 2.6295210166177906e-05,
      "loss": 0.7803,
      "step": 11716
    },
    {
      "epoch": 1.8342204132748905,
      "grad_norm": 3.9286694526672363,
      "learning_rate": 2.627077223851417e-05,
      "loss": 0.8126,
      "step": 11717
    },
    {
      "epoch": 1.8343769567939887,
      "grad_norm": 6.795763969421387,
      "learning_rate": 2.624633431085044e-05,
      "loss": 0.6296,
      "step": 11718
    },
    {
      "epoch": 1.834533500313087,
      "grad_norm": 6.505904197692871,
      "learning_rate": 2.6221896383186704e-05,
      "loss": 0.9323,
      "step": 11719
    },
    {
      "epoch": 1.8346900438321854,
      "grad_norm": 2.0406224727630615,
      "learning_rate": 2.619745845552297e-05,
      "loss": 0.8992,
      "step": 11720
    },
    {
      "epoch": 1.8348465873512838,
      "grad_norm": 11.124521255493164,
      "learning_rate": 2.6173020527859237e-05,
      "loss": 0.9607,
      "step": 11721
    },
    {
      "epoch": 1.835003130870382,
      "grad_norm": 10.044174194335938,
      "learning_rate": 2.61485826001955e-05,
      "loss": 1.0505,
      "step": 11722
    },
    {
      "epoch": 1.8351596743894802,
      "grad_norm": 5.03192138671875,
      "learning_rate": 2.6124144672531766e-05,
      "loss": 0.7101,
      "step": 11723
    },
    {
      "epoch": 1.8353162179085785,
      "grad_norm": 3.8350234031677246,
      "learning_rate": 2.6099706744868034e-05,
      "loss": 0.6641,
      "step": 11724
    },
    {
      "epoch": 1.835472761427677,
      "grad_norm": 1.9279121160507202,
      "learning_rate": 2.60752688172043e-05,
      "loss": 0.5772,
      "step": 11725
    },
    {
      "epoch": 1.8356293049467753,
      "grad_norm": 2.837632656097412,
      "learning_rate": 2.6050830889540564e-05,
      "loss": 0.5835,
      "step": 11726
    },
    {
      "epoch": 1.8357858484658736,
      "grad_norm": 4.1351237297058105,
      "learning_rate": 2.6026392961876832e-05,
      "loss": 0.9303,
      "step": 11727
    },
    {
      "epoch": 1.8359423919849718,
      "grad_norm": 2.9830517768859863,
      "learning_rate": 2.6001955034213097e-05,
      "loss": 0.7467,
      "step": 11728
    },
    {
      "epoch": 1.83609893550407,
      "grad_norm": 2.7831618785858154,
      "learning_rate": 2.5977517106549362e-05,
      "loss": 0.8099,
      "step": 11729
    },
    {
      "epoch": 1.8362554790231684,
      "grad_norm": 2.7024545669555664,
      "learning_rate": 2.595307917888563e-05,
      "loss": 1.2066,
      "step": 11730
    },
    {
      "epoch": 1.8364120225422669,
      "grad_norm": 7.1745524406433105,
      "learning_rate": 2.5928641251221895e-05,
      "loss": 1.3392,
      "step": 11731
    },
    {
      "epoch": 1.836568566061365,
      "grad_norm": 3.5159523487091064,
      "learning_rate": 2.590420332355816e-05,
      "loss": 1.2869,
      "step": 11732
    },
    {
      "epoch": 1.8367251095804633,
      "grad_norm": 1.527916669845581,
      "learning_rate": 2.5879765395894428e-05,
      "loss": 0.8537,
      "step": 11733
    },
    {
      "epoch": 1.8368816530995615,
      "grad_norm": 1.3133277893066406,
      "learning_rate": 2.5855327468230693e-05,
      "loss": 0.4337,
      "step": 11734
    },
    {
      "epoch": 1.83703819661866,
      "grad_norm": 1.1044042110443115,
      "learning_rate": 2.5830889540566957e-05,
      "loss": 0.5792,
      "step": 11735
    },
    {
      "epoch": 1.8371947401377584,
      "grad_norm": 1.173004388809204,
      "learning_rate": 2.5806451612903226e-05,
      "loss": 0.3835,
      "step": 11736
    },
    {
      "epoch": 1.8373512836568566,
      "grad_norm": 3.5974175930023193,
      "learning_rate": 2.578201368523949e-05,
      "loss": 0.7743,
      "step": 11737
    },
    {
      "epoch": 1.8375078271759548,
      "grad_norm": 1.2620445489883423,
      "learning_rate": 2.5757575757575755e-05,
      "loss": 0.6945,
      "step": 11738
    },
    {
      "epoch": 1.837664370695053,
      "grad_norm": 0.7731130719184875,
      "learning_rate": 2.573313782991202e-05,
      "loss": 0.4413,
      "step": 11739
    },
    {
      "epoch": 1.8378209142141515,
      "grad_norm": 1.389115810394287,
      "learning_rate": 2.5708699902248288e-05,
      "loss": 0.416,
      "step": 11740
    },
    {
      "epoch": 1.83797745773325,
      "grad_norm": 0.8706973195075989,
      "learning_rate": 2.5684261974584553e-05,
      "loss": 0.5073,
      "step": 11741
    },
    {
      "epoch": 1.8381340012523482,
      "grad_norm": 0.5937680602073669,
      "learning_rate": 2.5659824046920818e-05,
      "loss": 0.3761,
      "step": 11742
    },
    {
      "epoch": 1.8382905447714464,
      "grad_norm": 1.8303773403167725,
      "learning_rate": 2.5635386119257086e-05,
      "loss": 0.4664,
      "step": 11743
    },
    {
      "epoch": 1.8384470882905448,
      "grad_norm": 1.462870717048645,
      "learning_rate": 2.561094819159335e-05,
      "loss": 0.5187,
      "step": 11744
    },
    {
      "epoch": 1.838603631809643,
      "grad_norm": 1.8268141746520996,
      "learning_rate": 2.5586510263929615e-05,
      "loss": 0.5145,
      "step": 11745
    },
    {
      "epoch": 1.8387601753287415,
      "grad_norm": 1.6612815856933594,
      "learning_rate": 2.5562072336265884e-05,
      "loss": 0.5543,
      "step": 11746
    },
    {
      "epoch": 1.8389167188478397,
      "grad_norm": 3.4995779991149902,
      "learning_rate": 2.553763440860215e-05,
      "loss": 0.4739,
      "step": 11747
    },
    {
      "epoch": 1.839073262366938,
      "grad_norm": 1.3062471151351929,
      "learning_rate": 2.5513196480938413e-05,
      "loss": 0.4802,
      "step": 11748
    },
    {
      "epoch": 1.8392298058860364,
      "grad_norm": 6.06448221206665,
      "learning_rate": 2.548875855327468e-05,
      "loss": 1.0193,
      "step": 11749
    },
    {
      "epoch": 1.8393863494051348,
      "grad_norm": 2.5236899852752686,
      "learning_rate": 2.5464320625610946e-05,
      "loss": 0.5685,
      "step": 11750
    },
    {
      "epoch": 1.839542892924233,
      "grad_norm": 1.4289917945861816,
      "learning_rate": 2.543988269794721e-05,
      "loss": 0.5858,
      "step": 11751
    },
    {
      "epoch": 1.8396994364433312,
      "grad_norm": 1.3175865411758423,
      "learning_rate": 2.541544477028348e-05,
      "loss": 0.6848,
      "step": 11752
    },
    {
      "epoch": 1.8398559799624294,
      "grad_norm": 3.198878288269043,
      "learning_rate": 2.5391006842619744e-05,
      "loss": 0.62,
      "step": 11753
    },
    {
      "epoch": 1.8400125234815279,
      "grad_norm": 2.0949974060058594,
      "learning_rate": 2.536656891495601e-05,
      "loss": 0.5892,
      "step": 11754
    },
    {
      "epoch": 1.8401690670006263,
      "grad_norm": 2.16085147857666,
      "learning_rate": 2.5342130987292277e-05,
      "loss": 0.8502,
      "step": 11755
    },
    {
      "epoch": 1.8403256105197245,
      "grad_norm": 3.7577946186065674,
      "learning_rate": 2.5317693059628542e-05,
      "loss": 0.7301,
      "step": 11756
    },
    {
      "epoch": 1.8404821540388228,
      "grad_norm": 1.981907606124878,
      "learning_rate": 2.5293255131964807e-05,
      "loss": 0.7686,
      "step": 11757
    },
    {
      "epoch": 1.840638697557921,
      "grad_norm": 4.2499308586120605,
      "learning_rate": 2.5268817204301075e-05,
      "loss": 0.663,
      "step": 11758
    },
    {
      "epoch": 1.8407952410770194,
      "grad_norm": 5.370046138763428,
      "learning_rate": 2.524437927663734e-05,
      "loss": 0.8082,
      "step": 11759
    },
    {
      "epoch": 1.8409517845961179,
      "grad_norm": 1.3637969493865967,
      "learning_rate": 2.5219941348973604e-05,
      "loss": 0.7207,
      "step": 11760
    },
    {
      "epoch": 1.841108328115216,
      "grad_norm": 1.314660668373108,
      "learning_rate": 2.5195503421309873e-05,
      "loss": 0.6059,
      "step": 11761
    },
    {
      "epoch": 1.8412648716343143,
      "grad_norm": 6.1758809089660645,
      "learning_rate": 2.5171065493646137e-05,
      "loss": 0.684,
      "step": 11762
    },
    {
      "epoch": 1.8414214151534125,
      "grad_norm": 2.818993330001831,
      "learning_rate": 2.5146627565982402e-05,
      "loss": 0.7404,
      "step": 11763
    },
    {
      "epoch": 1.841577958672511,
      "grad_norm": 1.828629732131958,
      "learning_rate": 2.512218963831867e-05,
      "loss": 0.8012,
      "step": 11764
    },
    {
      "epoch": 1.8417345021916094,
      "grad_norm": 2.7739205360412598,
      "learning_rate": 2.5097751710654935e-05,
      "loss": 0.7304,
      "step": 11765
    },
    {
      "epoch": 1.8418910457107076,
      "grad_norm": 8.794857025146484,
      "learning_rate": 2.50733137829912e-05,
      "loss": 0.6273,
      "step": 11766
    },
    {
      "epoch": 1.8420475892298058,
      "grad_norm": 1.3181114196777344,
      "learning_rate": 2.5048875855327468e-05,
      "loss": 0.6637,
      "step": 11767
    },
    {
      "epoch": 1.842204132748904,
      "grad_norm": 1.4442148208618164,
      "learning_rate": 2.5024437927663733e-05,
      "loss": 0.5413,
      "step": 11768
    },
    {
      "epoch": 1.8423606762680025,
      "grad_norm": 4.040245532989502,
      "learning_rate": 2.4999999999999998e-05,
      "loss": 0.7349,
      "step": 11769
    },
    {
      "epoch": 1.842517219787101,
      "grad_norm": 3.4803292751312256,
      "learning_rate": 2.4975562072336266e-05,
      "loss": 0.8002,
      "step": 11770
    },
    {
      "epoch": 1.8426737633061991,
      "grad_norm": 27.806154251098633,
      "learning_rate": 2.495112414467253e-05,
      "loss": 0.4308,
      "step": 11771
    },
    {
      "epoch": 1.8428303068252974,
      "grad_norm": 6.975183963775635,
      "learning_rate": 2.4926686217008796e-05,
      "loss": 1.6088,
      "step": 11772
    },
    {
      "epoch": 1.8429868503443956,
      "grad_norm": 3.424659013748169,
      "learning_rate": 2.4902248289345064e-05,
      "loss": 0.9311,
      "step": 11773
    },
    {
      "epoch": 1.843143393863494,
      "grad_norm": 2.6333138942718506,
      "learning_rate": 2.487781036168133e-05,
      "loss": 1.0048,
      "step": 11774
    },
    {
      "epoch": 1.8432999373825925,
      "grad_norm": 3.147855520248413,
      "learning_rate": 2.4853372434017593e-05,
      "loss": 0.8559,
      "step": 11775
    },
    {
      "epoch": 1.8434564809016907,
      "grad_norm": 2.6393229961395264,
      "learning_rate": 2.482893450635386e-05,
      "loss": 1.0206,
      "step": 11776
    },
    {
      "epoch": 1.843613024420789,
      "grad_norm": 2.7040412425994873,
      "learning_rate": 2.4804496578690126e-05,
      "loss": 0.4979,
      "step": 11777
    },
    {
      "epoch": 1.8437695679398873,
      "grad_norm": 3.9352643489837646,
      "learning_rate": 2.478005865102639e-05,
      "loss": 0.7279,
      "step": 11778
    },
    {
      "epoch": 1.8439261114589856,
      "grad_norm": 2.641021966934204,
      "learning_rate": 2.475562072336266e-05,
      "loss": 0.9773,
      "step": 11779
    },
    {
      "epoch": 1.844082654978084,
      "grad_norm": 2.157909631729126,
      "learning_rate": 2.4731182795698924e-05,
      "loss": 0.9536,
      "step": 11780
    },
    {
      "epoch": 1.8442391984971822,
      "grad_norm": 3.1116890907287598,
      "learning_rate": 2.470674486803519e-05,
      "loss": 1.4514,
      "step": 11781
    },
    {
      "epoch": 1.8443957420162804,
      "grad_norm": 3.8484981060028076,
      "learning_rate": 2.4682306940371457e-05,
      "loss": 0.6041,
      "step": 11782
    },
    {
      "epoch": 1.8445522855353789,
      "grad_norm": 2.849015951156616,
      "learning_rate": 2.4657869012707722e-05,
      "loss": 0.9883,
      "step": 11783
    },
    {
      "epoch": 1.8447088290544773,
      "grad_norm": 1.907538652420044,
      "learning_rate": 2.4633431085043983e-05,
      "loss": 0.4752,
      "step": 11784
    },
    {
      "epoch": 1.8448653725735755,
      "grad_norm": 4.029622554779053,
      "learning_rate": 2.4608993157380255e-05,
      "loss": 1.0383,
      "step": 11785
    },
    {
      "epoch": 1.8450219160926737,
      "grad_norm": 4.055169105529785,
      "learning_rate": 2.458455522971652e-05,
      "loss": 0.9394,
      "step": 11786
    },
    {
      "epoch": 1.845178459611772,
      "grad_norm": 1.425485372543335,
      "learning_rate": 2.456011730205278e-05,
      "loss": 0.3284,
      "step": 11787
    },
    {
      "epoch": 1.8453350031308704,
      "grad_norm": 3.0947279930114746,
      "learning_rate": 2.4535679374389053e-05,
      "loss": 0.9571,
      "step": 11788
    },
    {
      "epoch": 1.8454915466499688,
      "grad_norm": 2.302706718444824,
      "learning_rate": 2.4511241446725318e-05,
      "loss": 0.8344,
      "step": 11789
    },
    {
      "epoch": 1.845648090169067,
      "grad_norm": 5.67050838470459,
      "learning_rate": 2.448680351906158e-05,
      "loss": 0.5112,
      "step": 11790
    },
    {
      "epoch": 1.8458046336881653,
      "grad_norm": 1.0470259189605713,
      "learning_rate": 2.446236559139785e-05,
      "loss": 0.4184,
      "step": 11791
    },
    {
      "epoch": 1.8459611772072635,
      "grad_norm": 1.2459172010421753,
      "learning_rate": 2.4437927663734115e-05,
      "loss": 0.5126,
      "step": 11792
    },
    {
      "epoch": 1.846117720726362,
      "grad_norm": 0.8190130591392517,
      "learning_rate": 2.4413489736070377e-05,
      "loss": 0.4512,
      "step": 11793
    },
    {
      "epoch": 1.8462742642454604,
      "grad_norm": 3.6219711303710938,
      "learning_rate": 2.438905180840665e-05,
      "loss": 0.916,
      "step": 11794
    },
    {
      "epoch": 1.8464308077645586,
      "grad_norm": 2.1488611698150635,
      "learning_rate": 2.4364613880742913e-05,
      "loss": 0.6591,
      "step": 11795
    },
    {
      "epoch": 1.8465873512836568,
      "grad_norm": 0.7660203576087952,
      "learning_rate": 2.4340175953079175e-05,
      "loss": 0.3781,
      "step": 11796
    },
    {
      "epoch": 1.846743894802755,
      "grad_norm": 1.8825334310531616,
      "learning_rate": 2.4315738025415446e-05,
      "loss": 0.4816,
      "step": 11797
    },
    {
      "epoch": 1.8469004383218535,
      "grad_norm": 1.4968349933624268,
      "learning_rate": 2.4291300097751708e-05,
      "loss": 0.5589,
      "step": 11798
    },
    {
      "epoch": 1.847056981840952,
      "grad_norm": 3.8094868659973145,
      "learning_rate": 2.4266862170087972e-05,
      "loss": 0.4607,
      "step": 11799
    },
    {
      "epoch": 1.8472135253600501,
      "grad_norm": 2.436556816101074,
      "learning_rate": 2.4242424242424244e-05,
      "loss": 0.5093,
      "step": 11800
    },
    {
      "epoch": 1.8473700688791483,
      "grad_norm": 3.266652822494507,
      "learning_rate": 2.4217986314760505e-05,
      "loss": 0.6517,
      "step": 11801
    },
    {
      "epoch": 1.8475266123982466,
      "grad_norm": 2.8735289573669434,
      "learning_rate": 2.419354838709677e-05,
      "loss": 0.6738,
      "step": 11802
    },
    {
      "epoch": 1.847683155917345,
      "grad_norm": 2.9087893962860107,
      "learning_rate": 2.4169110459433035e-05,
      "loss": 0.9188,
      "step": 11803
    },
    {
      "epoch": 1.8478396994364434,
      "grad_norm": 1.770983338356018,
      "learning_rate": 2.4144672531769303e-05,
      "loss": 0.5283,
      "step": 11804
    },
    {
      "epoch": 1.8479962429555417,
      "grad_norm": 4.079898357391357,
      "learning_rate": 2.4120234604105568e-05,
      "loss": 0.6439,
      "step": 11805
    },
    {
      "epoch": 1.8481527864746399,
      "grad_norm": 2.5677294731140137,
      "learning_rate": 2.4095796676441833e-05,
      "loss": 0.9994,
      "step": 11806
    },
    {
      "epoch": 1.8483093299937383,
      "grad_norm": 1.662412166595459,
      "learning_rate": 2.40713587487781e-05,
      "loss": 0.7825,
      "step": 11807
    },
    {
      "epoch": 1.8484658735128365,
      "grad_norm": 1.9668818712234497,
      "learning_rate": 2.4046920821114366e-05,
      "loss": 0.6303,
      "step": 11808
    },
    {
      "epoch": 1.848622417031935,
      "grad_norm": 2.1464879512786865,
      "learning_rate": 2.402248289345063e-05,
      "loss": 0.709,
      "step": 11809
    },
    {
      "epoch": 1.8487789605510332,
      "grad_norm": 5.349493980407715,
      "learning_rate": 2.39980449657869e-05,
      "loss": 0.6679,
      "step": 11810
    },
    {
      "epoch": 1.8489355040701314,
      "grad_norm": 1.8723517656326294,
      "learning_rate": 2.3973607038123163e-05,
      "loss": 0.8859,
      "step": 11811
    },
    {
      "epoch": 1.8490920475892298,
      "grad_norm": 2.6535778045654297,
      "learning_rate": 2.3949169110459428e-05,
      "loss": 0.5719,
      "step": 11812
    },
    {
      "epoch": 1.849248591108328,
      "grad_norm": 1.8219172954559326,
      "learning_rate": 2.3924731182795696e-05,
      "loss": 0.6213,
      "step": 11813
    },
    {
      "epoch": 1.8494051346274265,
      "grad_norm": 2.3253509998321533,
      "learning_rate": 2.390029325513196e-05,
      "loss": 0.6391,
      "step": 11814
    },
    {
      "epoch": 1.8495616781465247,
      "grad_norm": 5.5953545570373535,
      "learning_rate": 2.3875855327468226e-05,
      "loss": 0.4743,
      "step": 11815
    },
    {
      "epoch": 1.849718221665623,
      "grad_norm": 4.139413356781006,
      "learning_rate": 2.3851417399804494e-05,
      "loss": 0.798,
      "step": 11816
    },
    {
      "epoch": 1.8498747651847214,
      "grad_norm": 10.96532154083252,
      "learning_rate": 2.382697947214076e-05,
      "loss": 0.4565,
      "step": 11817
    },
    {
      "epoch": 1.8500313087038198,
      "grad_norm": 2.8876092433929443,
      "learning_rate": 2.3802541544477024e-05,
      "loss": 0.9685,
      "step": 11818
    },
    {
      "epoch": 1.850187852222918,
      "grad_norm": 3.1957509517669678,
      "learning_rate": 2.3778103616813292e-05,
      "loss": 1.0197,
      "step": 11819
    },
    {
      "epoch": 1.8503443957420163,
      "grad_norm": 3.1257941722869873,
      "learning_rate": 2.3753665689149557e-05,
      "loss": 0.8237,
      "step": 11820
    },
    {
      "epoch": 1.8505009392611145,
      "grad_norm": 3.2138302326202393,
      "learning_rate": 2.372922776148582e-05,
      "loss": 0.8114,
      "step": 11821
    },
    {
      "epoch": 1.850657482780213,
      "grad_norm": 7.504042625427246,
      "learning_rate": 2.370478983382209e-05,
      "loss": 0.7801,
      "step": 11822
    },
    {
      "epoch": 1.8508140262993114,
      "grad_norm": 5.465351104736328,
      "learning_rate": 2.3680351906158355e-05,
      "loss": 1.1196,
      "step": 11823
    },
    {
      "epoch": 1.8509705698184096,
      "grad_norm": 5.04585599899292,
      "learning_rate": 2.365591397849462e-05,
      "loss": 0.9248,
      "step": 11824
    },
    {
      "epoch": 1.8511271133375078,
      "grad_norm": 1.4739549160003662,
      "learning_rate": 2.3631476050830888e-05,
      "loss": 0.6686,
      "step": 11825
    },
    {
      "epoch": 1.851283656856606,
      "grad_norm": 3.8544936180114746,
      "learning_rate": 2.3607038123167152e-05,
      "loss": 0.923,
      "step": 11826
    },
    {
      "epoch": 1.8514402003757044,
      "grad_norm": 5.232646942138672,
      "learning_rate": 2.3582600195503417e-05,
      "loss": 0.8553,
      "step": 11827
    },
    {
      "epoch": 1.8515967438948029,
      "grad_norm": 3.4390921592712402,
      "learning_rate": 2.3558162267839685e-05,
      "loss": 1.4067,
      "step": 11828
    },
    {
      "epoch": 1.851753287413901,
      "grad_norm": 4.184832572937012,
      "learning_rate": 2.353372434017595e-05,
      "loss": 0.9317,
      "step": 11829
    },
    {
      "epoch": 1.8519098309329993,
      "grad_norm": 3.6399333477020264,
      "learning_rate": 2.3509286412512215e-05,
      "loss": 1.2859,
      "step": 11830
    },
    {
      "epoch": 1.8520663744520975,
      "grad_norm": 2.263460636138916,
      "learning_rate": 2.3484848484848483e-05,
      "loss": 0.8343,
      "step": 11831
    },
    {
      "epoch": 1.852222917971196,
      "grad_norm": 1.4903042316436768,
      "learning_rate": 2.3460410557184748e-05,
      "loss": 0.7115,
      "step": 11832
    },
    {
      "epoch": 1.8523794614902944,
      "grad_norm": 4.350191593170166,
      "learning_rate": 2.3435972629521013e-05,
      "loss": 1.8051,
      "step": 11833
    },
    {
      "epoch": 1.8525360050093926,
      "grad_norm": 3.3762741088867188,
      "learning_rate": 2.341153470185728e-05,
      "loss": 0.6415,
      "step": 11834
    },
    {
      "epoch": 1.8526925485284909,
      "grad_norm": 2.8959567546844482,
      "learning_rate": 2.3387096774193546e-05,
      "loss": 0.9455,
      "step": 11835
    },
    {
      "epoch": 1.852849092047589,
      "grad_norm": 4.2399444580078125,
      "learning_rate": 2.336265884652981e-05,
      "loss": 1.3687,
      "step": 11836
    },
    {
      "epoch": 1.8530056355666875,
      "grad_norm": 3.185884952545166,
      "learning_rate": 2.333822091886608e-05,
      "loss": 0.6854,
      "step": 11837
    },
    {
      "epoch": 1.853162179085786,
      "grad_norm": 1.1546235084533691,
      "learning_rate": 2.3313782991202344e-05,
      "loss": 0.4202,
      "step": 11838
    },
    {
      "epoch": 1.8533187226048842,
      "grad_norm": 0.8933812975883484,
      "learning_rate": 2.328934506353861e-05,
      "loss": 0.594,
      "step": 11839
    },
    {
      "epoch": 1.8534752661239824,
      "grad_norm": 1.431520938873291,
      "learning_rate": 2.3264907135874877e-05,
      "loss": 0.5386,
      "step": 11840
    },
    {
      "epoch": 1.8536318096430808,
      "grad_norm": 1.0850852727890015,
      "learning_rate": 2.324046920821114e-05,
      "loss": 0.596,
      "step": 11841
    },
    {
      "epoch": 1.853788353162179,
      "grad_norm": 1.6052504777908325,
      "learning_rate": 2.3216031280547406e-05,
      "loss": 0.6123,
      "step": 11842
    },
    {
      "epoch": 1.8539448966812775,
      "grad_norm": 1.0671995878219604,
      "learning_rate": 2.3191593352883674e-05,
      "loss": 0.5618,
      "step": 11843
    },
    {
      "epoch": 1.8541014402003757,
      "grad_norm": 1.0191351175308228,
      "learning_rate": 2.316715542521994e-05,
      "loss": 0.5923,
      "step": 11844
    },
    {
      "epoch": 1.854257983719474,
      "grad_norm": 1.2922497987747192,
      "learning_rate": 2.3142717497556204e-05,
      "loss": 0.5884,
      "step": 11845
    },
    {
      "epoch": 1.8544145272385724,
      "grad_norm": 0.6985080242156982,
      "learning_rate": 2.3118279569892472e-05,
      "loss": 0.549,
      "step": 11846
    },
    {
      "epoch": 1.8545710707576706,
      "grad_norm": 1.6202582120895386,
      "learning_rate": 2.3093841642228737e-05,
      "loss": 0.7027,
      "step": 11847
    },
    {
      "epoch": 1.854727614276769,
      "grad_norm": 6.858791828155518,
      "learning_rate": 2.3069403714565002e-05,
      "loss": 1.267,
      "step": 11848
    },
    {
      "epoch": 1.8548841577958672,
      "grad_norm": 2.2614457607269287,
      "learning_rate": 2.304496578690127e-05,
      "loss": 0.7434,
      "step": 11849
    },
    {
      "epoch": 1.8550407013149655,
      "grad_norm": 6.331788539886475,
      "learning_rate": 2.3020527859237535e-05,
      "loss": 0.8123,
      "step": 11850
    },
    {
      "epoch": 1.855197244834064,
      "grad_norm": 1.8438414335250854,
      "learning_rate": 2.29960899315738e-05,
      "loss": 0.754,
      "step": 11851
    },
    {
      "epoch": 1.8553537883531623,
      "grad_norm": 1.1426644325256348,
      "learning_rate": 2.2971652003910068e-05,
      "loss": 0.5913,
      "step": 11852
    },
    {
      "epoch": 1.8555103318722606,
      "grad_norm": 1.538586974143982,
      "learning_rate": 2.2947214076246333e-05,
      "loss": 0.5843,
      "step": 11853
    },
    {
      "epoch": 1.8556668753913588,
      "grad_norm": 2.8902087211608887,
      "learning_rate": 2.2922776148582597e-05,
      "loss": 0.8857,
      "step": 11854
    },
    {
      "epoch": 1.855823418910457,
      "grad_norm": 2.553431749343872,
      "learning_rate": 2.2898338220918866e-05,
      "loss": 0.8052,
      "step": 11855
    },
    {
      "epoch": 1.8559799624295554,
      "grad_norm": 3.586069107055664,
      "learning_rate": 2.287390029325513e-05,
      "loss": 0.7178,
      "step": 11856
    },
    {
      "epoch": 1.8561365059486539,
      "grad_norm": 1.4194121360778809,
      "learning_rate": 2.2849462365591395e-05,
      "loss": 0.7,
      "step": 11857
    },
    {
      "epoch": 1.856293049467752,
      "grad_norm": 1.709073781967163,
      "learning_rate": 2.2825024437927663e-05,
      "loss": 0.613,
      "step": 11858
    },
    {
      "epoch": 1.8564495929868503,
      "grad_norm": 1.8721200227737427,
      "learning_rate": 2.2800586510263928e-05,
      "loss": 0.6448,
      "step": 11859
    },
    {
      "epoch": 1.8566061365059485,
      "grad_norm": 3.26497745513916,
      "learning_rate": 2.2776148582600193e-05,
      "loss": 0.7323,
      "step": 11860
    },
    {
      "epoch": 1.856762680025047,
      "grad_norm": 3.8040733337402344,
      "learning_rate": 2.275171065493646e-05,
      "loss": 1.2313,
      "step": 11861
    },
    {
      "epoch": 1.8569192235441454,
      "grad_norm": 3.2131476402282715,
      "learning_rate": 2.2727272727272726e-05,
      "loss": 0.5404,
      "step": 11862
    },
    {
      "epoch": 1.8570757670632436,
      "grad_norm": 2.0257785320281982,
      "learning_rate": 2.270283479960899e-05,
      "loss": 0.5706,
      "step": 11863
    },
    {
      "epoch": 1.8572323105823418,
      "grad_norm": 1.4355109930038452,
      "learning_rate": 2.267839687194526e-05,
      "loss": 0.7594,
      "step": 11864
    },
    {
      "epoch": 1.85738885410144,
      "grad_norm": 2.189570426940918,
      "learning_rate": 2.2653958944281524e-05,
      "loss": 0.892,
      "step": 11865
    },
    {
      "epoch": 1.8575453976205385,
      "grad_norm": 1.5247050523757935,
      "learning_rate": 2.262952101661779e-05,
      "loss": 0.7958,
      "step": 11866
    },
    {
      "epoch": 1.857701941139637,
      "grad_norm": 2.8987576961517334,
      "learning_rate": 2.2605083088954053e-05,
      "loss": 0.8679,
      "step": 11867
    },
    {
      "epoch": 1.8578584846587352,
      "grad_norm": 2.054922580718994,
      "learning_rate": 2.258064516129032e-05,
      "loss": 0.8147,
      "step": 11868
    },
    {
      "epoch": 1.8580150281778334,
      "grad_norm": 2.206418752670288,
      "learning_rate": 2.2556207233626586e-05,
      "loss": 0.4729,
      "step": 11869
    },
    {
      "epoch": 1.8581715716969316,
      "grad_norm": 3.0962157249450684,
      "learning_rate": 2.253176930596285e-05,
      "loss": 1.0916,
      "step": 11870
    },
    {
      "epoch": 1.85832811521603,
      "grad_norm": 3.1131982803344727,
      "learning_rate": 2.250733137829912e-05,
      "loss": 1.1446,
      "step": 11871
    },
    {
      "epoch": 1.8584846587351285,
      "grad_norm": 3.34653639793396,
      "learning_rate": 2.2482893450635384e-05,
      "loss": 1.1253,
      "step": 11872
    },
    {
      "epoch": 1.8586412022542267,
      "grad_norm": 7.987216472625732,
      "learning_rate": 2.245845552297165e-05,
      "loss": 1.4304,
      "step": 11873
    },
    {
      "epoch": 1.858797745773325,
      "grad_norm": 3.8446297645568848,
      "learning_rate": 2.2434017595307917e-05,
      "loss": 0.5307,
      "step": 11874
    },
    {
      "epoch": 1.8589542892924233,
      "grad_norm": 2.1328821182250977,
      "learning_rate": 2.2409579667644182e-05,
      "loss": 0.765,
      "step": 11875
    },
    {
      "epoch": 1.8591108328115216,
      "grad_norm": 4.147828578948975,
      "learning_rate": 2.2385141739980447e-05,
      "loss": 1.1964,
      "step": 11876
    },
    {
      "epoch": 1.85926737633062,
      "grad_norm": 1.5424137115478516,
      "learning_rate": 2.2360703812316715e-05,
      "loss": 0.7498,
      "step": 11877
    },
    {
      "epoch": 1.8594239198497182,
      "grad_norm": 2.9707698822021484,
      "learning_rate": 2.233626588465298e-05,
      "loss": 0.8177,
      "step": 11878
    },
    {
      "epoch": 1.8595804633688164,
      "grad_norm": 8.598952293395996,
      "learning_rate": 2.2311827956989244e-05,
      "loss": 0.8357,
      "step": 11879
    },
    {
      "epoch": 1.8597370068879149,
      "grad_norm": 1.6900843381881714,
      "learning_rate": 2.2287390029325513e-05,
      "loss": 0.6847,
      "step": 11880
    },
    {
      "epoch": 1.859893550407013,
      "grad_norm": 5.230086803436279,
      "learning_rate": 2.2262952101661777e-05,
      "loss": 1.5568,
      "step": 11881
    },
    {
      "epoch": 1.8600500939261115,
      "grad_norm": 3.5463037490844727,
      "learning_rate": 2.2238514173998042e-05,
      "loss": 1.0774,
      "step": 11882
    },
    {
      "epoch": 1.8602066374452098,
      "grad_norm": 3.144085168838501,
      "learning_rate": 2.221407624633431e-05,
      "loss": 0.8156,
      "step": 11883
    },
    {
      "epoch": 1.860363180964308,
      "grad_norm": 2.6396665573120117,
      "learning_rate": 2.2189638318670575e-05,
      "loss": 0.6252,
      "step": 11884
    },
    {
      "epoch": 1.8605197244834064,
      "grad_norm": 0.6238581538200378,
      "learning_rate": 2.216520039100684e-05,
      "loss": 0.1875,
      "step": 11885
    },
    {
      "epoch": 1.8606762680025049,
      "grad_norm": 2.998420238494873,
      "learning_rate": 2.2140762463343108e-05,
      "loss": 0.52,
      "step": 11886
    },
    {
      "epoch": 1.860832811521603,
      "grad_norm": 3.8253471851348877,
      "learning_rate": 2.2116324535679373e-05,
      "loss": 1.4961,
      "step": 11887
    },
    {
      "epoch": 1.8609893550407013,
      "grad_norm": 4.558570861816406,
      "learning_rate": 2.2091886608015638e-05,
      "loss": 0.85,
      "step": 11888
    },
    {
      "epoch": 1.8611458985597995,
      "grad_norm": 1.2218017578125,
      "learning_rate": 2.2067448680351906e-05,
      "loss": 0.634,
      "step": 11889
    },
    {
      "epoch": 1.861302442078898,
      "grad_norm": 2.944087266921997,
      "learning_rate": 2.204301075268817e-05,
      "loss": 0.5668,
      "step": 11890
    },
    {
      "epoch": 1.8614589855979964,
      "grad_norm": 1.4352192878723145,
      "learning_rate": 2.2018572825024436e-05,
      "loss": 0.6317,
      "step": 11891
    },
    {
      "epoch": 1.8616155291170946,
      "grad_norm": 1.2094552516937256,
      "learning_rate": 2.1994134897360704e-05,
      "loss": 0.5786,
      "step": 11892
    },
    {
      "epoch": 1.8617720726361928,
      "grad_norm": 1.98757803440094,
      "learning_rate": 2.196969696969697e-05,
      "loss": 0.6373,
      "step": 11893
    },
    {
      "epoch": 1.861928616155291,
      "grad_norm": 1.0095068216323853,
      "learning_rate": 2.1945259042033233e-05,
      "loss": 0.6269,
      "step": 11894
    },
    {
      "epoch": 1.8620851596743895,
      "grad_norm": 1.0049716234207153,
      "learning_rate": 2.19208211143695e-05,
      "loss": 0.6092,
      "step": 11895
    },
    {
      "epoch": 1.862241703193488,
      "grad_norm": 1.7188302278518677,
      "learning_rate": 2.1896383186705766e-05,
      "loss": 0.5908,
      "step": 11896
    },
    {
      "epoch": 1.8623982467125861,
      "grad_norm": 2.2089591026306152,
      "learning_rate": 2.187194525904203e-05,
      "loss": 0.6305,
      "step": 11897
    },
    {
      "epoch": 1.8625547902316844,
      "grad_norm": 2.382549285888672,
      "learning_rate": 2.18475073313783e-05,
      "loss": 0.7027,
      "step": 11898
    },
    {
      "epoch": 1.8627113337507826,
      "grad_norm": 1.0006448030471802,
      "learning_rate": 2.1823069403714564e-05,
      "loss": 0.6653,
      "step": 11899
    },
    {
      "epoch": 1.862867877269881,
      "grad_norm": 1.5975639820098877,
      "learning_rate": 2.179863147605083e-05,
      "loss": 0.7094,
      "step": 11900
    },
    {
      "epoch": 1.8630244207889795,
      "grad_norm": 1.290367841720581,
      "learning_rate": 2.1774193548387097e-05,
      "loss": 0.7759,
      "step": 11901
    },
    {
      "epoch": 1.8631809643080777,
      "grad_norm": 2.142758846282959,
      "learning_rate": 2.1749755620723362e-05,
      "loss": 0.7417,
      "step": 11902
    },
    {
      "epoch": 1.8633375078271759,
      "grad_norm": 1.4143774509429932,
      "learning_rate": 2.1725317693059627e-05,
      "loss": 0.8493,
      "step": 11903
    },
    {
      "epoch": 1.863494051346274,
      "grad_norm": 1.213179588317871,
      "learning_rate": 2.1700879765395895e-05,
      "loss": 0.7674,
      "step": 11904
    },
    {
      "epoch": 1.8636505948653725,
      "grad_norm": 1.534143328666687,
      "learning_rate": 2.167644183773216e-05,
      "loss": 0.8457,
      "step": 11905
    },
    {
      "epoch": 1.863807138384471,
      "grad_norm": 14.284627914428711,
      "learning_rate": 2.1652003910068425e-05,
      "loss": 0.7507,
      "step": 11906
    },
    {
      "epoch": 1.8639636819035692,
      "grad_norm": 1.4157757759094238,
      "learning_rate": 2.1627565982404693e-05,
      "loss": 0.7684,
      "step": 11907
    },
    {
      "epoch": 1.8641202254226674,
      "grad_norm": 1.829573392868042,
      "learning_rate": 2.1603128054740958e-05,
      "loss": 0.7432,
      "step": 11908
    },
    {
      "epoch": 1.8642767689417659,
      "grad_norm": 1.5792940855026245,
      "learning_rate": 2.1578690127077222e-05,
      "loss": 0.9471,
      "step": 11909
    },
    {
      "epoch": 1.864433312460864,
      "grad_norm": 3.3256168365478516,
      "learning_rate": 2.155425219941349e-05,
      "loss": 1.042,
      "step": 11910
    },
    {
      "epoch": 1.8645898559799625,
      "grad_norm": 2.1688032150268555,
      "learning_rate": 2.1529814271749755e-05,
      "loss": 0.8677,
      "step": 11911
    },
    {
      "epoch": 1.8647463994990607,
      "grad_norm": 2.214085578918457,
      "learning_rate": 2.150537634408602e-05,
      "loss": 1.0286,
      "step": 11912
    },
    {
      "epoch": 1.864902943018159,
      "grad_norm": 2.812800168991089,
      "learning_rate": 2.148093841642229e-05,
      "loss": 0.8419,
      "step": 11913
    },
    {
      "epoch": 1.8650594865372574,
      "grad_norm": 2.8499462604522705,
      "learning_rate": 2.1456500488758553e-05,
      "loss": 0.7129,
      "step": 11914
    },
    {
      "epoch": 1.8652160300563556,
      "grad_norm": 2.5196754932403564,
      "learning_rate": 2.1432062561094818e-05,
      "loss": 0.9942,
      "step": 11915
    },
    {
      "epoch": 1.865372573575454,
      "grad_norm": 1.3992620706558228,
      "learning_rate": 2.1407624633431086e-05,
      "loss": 0.7659,
      "step": 11916
    },
    {
      "epoch": 1.8655291170945523,
      "grad_norm": 3.1790847778320312,
      "learning_rate": 2.138318670576735e-05,
      "loss": 0.7056,
      "step": 11917
    },
    {
      "epoch": 1.8656856606136505,
      "grad_norm": 3.775634527206421,
      "learning_rate": 2.1358748778103612e-05,
      "loss": 1.2447,
      "step": 11918
    },
    {
      "epoch": 1.865842204132749,
      "grad_norm": 2.107165813446045,
      "learning_rate": 2.1334310850439884e-05,
      "loss": 0.5885,
      "step": 11919
    },
    {
      "epoch": 1.8659987476518474,
      "grad_norm": 2.489856243133545,
      "learning_rate": 2.130987292277615e-05,
      "loss": 0.6303,
      "step": 11920
    },
    {
      "epoch": 1.8661552911709456,
      "grad_norm": 6.781438827514648,
      "learning_rate": 2.128543499511241e-05,
      "loss": 0.6447,
      "step": 11921
    },
    {
      "epoch": 1.8663118346900438,
      "grad_norm": 3.0962233543395996,
      "learning_rate": 2.126099706744868e-05,
      "loss": 0.6563,
      "step": 11922
    },
    {
      "epoch": 1.866468378209142,
      "grad_norm": 3.9209532737731934,
      "learning_rate": 2.1236559139784946e-05,
      "loss": 1.2628,
      "step": 11923
    },
    {
      "epoch": 1.8666249217282405,
      "grad_norm": 2.2564404010772705,
      "learning_rate": 2.1212121212121208e-05,
      "loss": 0.7253,
      "step": 11924
    },
    {
      "epoch": 1.866781465247339,
      "grad_norm": 5.064412593841553,
      "learning_rate": 2.118768328445748e-05,
      "loss": 1.0094,
      "step": 11925
    },
    {
      "epoch": 1.8669380087664371,
      "grad_norm": 4.73468542098999,
      "learning_rate": 2.1163245356793744e-05,
      "loss": 1.0255,
      "step": 11926
    },
    {
      "epoch": 1.8670945522855353,
      "grad_norm": 4.486331462860107,
      "learning_rate": 2.1138807429130006e-05,
      "loss": 1.1333,
      "step": 11927
    },
    {
      "epoch": 1.8672510958046336,
      "grad_norm": 4.3150811195373535,
      "learning_rate": 2.1114369501466277e-05,
      "loss": 0.555,
      "step": 11928
    },
    {
      "epoch": 1.867407639323732,
      "grad_norm": 4.391129970550537,
      "learning_rate": 2.1089931573802542e-05,
      "loss": 1.3198,
      "step": 11929
    },
    {
      "epoch": 1.8675641828428304,
      "grad_norm": 3.979997158050537,
      "learning_rate": 2.1065493646138803e-05,
      "loss": 1.0015,
      "step": 11930
    },
    {
      "epoch": 1.8677207263619287,
      "grad_norm": 2.4785280227661133,
      "learning_rate": 2.1041055718475068e-05,
      "loss": 1.9898,
      "step": 11931
    },
    {
      "epoch": 1.8678772698810269,
      "grad_norm": 2.6585922241210938,
      "learning_rate": 2.1016617790811336e-05,
      "loss": 1.0243,
      "step": 11932
    },
    {
      "epoch": 1.868033813400125,
      "grad_norm": NaN,
      "learning_rate": 2.1016617790811336e-05,
      "loss": 0.0,
      "step": 11933
    },
    {
      "epoch": 1.8681903569192235,
      "grad_norm": 3.6180975437164307,
      "learning_rate": 2.09921798631476e-05,
      "loss": 0.9492,
      "step": 11934
    },
    {
      "epoch": 1.868346900438322,
      "grad_norm": 4.195194721221924,
      "learning_rate": 2.0967741935483866e-05,
      "loss": 1.0283,
      "step": 11935
    },
    {
      "epoch": 1.8685034439574202,
      "grad_norm": 1.3655951023101807,
      "learning_rate": 2.0943304007820134e-05,
      "loss": 0.5503,
      "step": 11936
    },
    {
      "epoch": 1.8686599874765184,
      "grad_norm": 2.214198112487793,
      "learning_rate": 2.09188660801564e-05,
      "loss": 0.5951,
      "step": 11937
    },
    {
      "epoch": 1.8688165309956166,
      "grad_norm": 2.850289821624756,
      "learning_rate": 2.0894428152492664e-05,
      "loss": 0.8592,
      "step": 11938
    },
    {
      "epoch": 1.868973074514715,
      "grad_norm": 0.8146681189537048,
      "learning_rate": 2.0869990224828932e-05,
      "loss": 0.8295,
      "step": 11939
    },
    {
      "epoch": 1.8691296180338135,
      "grad_norm": 1.004290223121643,
      "learning_rate": 2.0845552297165197e-05,
      "loss": 0.8397,
      "step": 11940
    },
    {
      "epoch": 1.8692861615529117,
      "grad_norm": 1.3244619369506836,
      "learning_rate": 2.082111436950146e-05,
      "loss": 0.9296,
      "step": 11941
    },
    {
      "epoch": 1.86944270507201,
      "grad_norm": 1.1102343797683716,
      "learning_rate": 2.079667644183773e-05,
      "loss": 0.8361,
      "step": 11942
    },
    {
      "epoch": 1.8695992485911084,
      "grad_norm": 1.2211482524871826,
      "learning_rate": 2.0772238514173995e-05,
      "loss": 1.0738,
      "step": 11943
    },
    {
      "epoch": 1.8697557921102066,
      "grad_norm": 0.993922233581543,
      "learning_rate": 2.074780058651026e-05,
      "loss": 0.876,
      "step": 11944
    },
    {
      "epoch": 1.869912335629305,
      "grad_norm": 1.7707405090332031,
      "learning_rate": 2.0723362658846528e-05,
      "loss": 0.8152,
      "step": 11945
    },
    {
      "epoch": 1.8700688791484033,
      "grad_norm": 0.997743546962738,
      "learning_rate": 2.0698924731182792e-05,
      "loss": 0.8948,
      "step": 11946
    },
    {
      "epoch": 1.8702254226675015,
      "grad_norm": 0.9178709983825684,
      "learning_rate": 2.0674486803519057e-05,
      "loss": 0.8765,
      "step": 11947
    },
    {
      "epoch": 1.8703819661866,
      "grad_norm": 2.155320167541504,
      "learning_rate": 2.0650048875855325e-05,
      "loss": 0.9537,
      "step": 11948
    },
    {
      "epoch": 1.8705385097056983,
      "grad_norm": 1.6522964239120483,
      "learning_rate": 2.062561094819159e-05,
      "loss": 0.9491,
      "step": 11949
    },
    {
      "epoch": 1.8706950532247966,
      "grad_norm": 1.9385637044906616,
      "learning_rate": 2.0601173020527855e-05,
      "loss": 0.9958,
      "step": 11950
    },
    {
      "epoch": 1.8708515967438948,
      "grad_norm": 2.451505422592163,
      "learning_rate": 2.0576735092864123e-05,
      "loss": 0.9823,
      "step": 11951
    },
    {
      "epoch": 1.871008140262993,
      "grad_norm": 9.86997127532959,
      "learning_rate": 2.0552297165200388e-05,
      "loss": 0.9815,
      "step": 11952
    },
    {
      "epoch": 1.8711646837820914,
      "grad_norm": 1.331578016281128,
      "learning_rate": 2.0527859237536653e-05,
      "loss": 0.9844,
      "step": 11953
    },
    {
      "epoch": 1.8713212273011899,
      "grad_norm": 2.020040273666382,
      "learning_rate": 2.050342130987292e-05,
      "loss": 0.8801,
      "step": 11954
    },
    {
      "epoch": 1.871477770820288,
      "grad_norm": 2.5946261882781982,
      "learning_rate": 2.0478983382209186e-05,
      "loss": 1.0083,
      "step": 11955
    },
    {
      "epoch": 1.8716343143393863,
      "grad_norm": 1.8665767908096313,
      "learning_rate": 2.045454545454545e-05,
      "loss": 0.9543,
      "step": 11956
    },
    {
      "epoch": 1.8717908578584845,
      "grad_norm": 2.101576089859009,
      "learning_rate": 2.043010752688172e-05,
      "loss": 0.8922,
      "step": 11957
    },
    {
      "epoch": 1.871947401377583,
      "grad_norm": 1.9192637205123901,
      "learning_rate": 2.0405669599217984e-05,
      "loss": 0.9835,
      "step": 11958
    },
    {
      "epoch": 1.8721039448966814,
      "grad_norm": 6.741157054901123,
      "learning_rate": 2.038123167155425e-05,
      "loss": 0.9142,
      "step": 11959
    },
    {
      "epoch": 1.8722604884157796,
      "grad_norm": 2.8808178901672363,
      "learning_rate": 2.0356793743890517e-05,
      "loss": 0.9528,
      "step": 11960
    },
    {
      "epoch": 1.8724170319348779,
      "grad_norm": 7.245316505432129,
      "learning_rate": 2.033235581622678e-05,
      "loss": 1.0117,
      "step": 11961
    },
    {
      "epoch": 1.872573575453976,
      "grad_norm": 1.8030654191970825,
      "learning_rate": 2.0307917888563046e-05,
      "loss": 0.736,
      "step": 11962
    },
    {
      "epoch": 1.8727301189730745,
      "grad_norm": 19.674142837524414,
      "learning_rate": 2.0283479960899314e-05,
      "loss": 0.9961,
      "step": 11963
    },
    {
      "epoch": 1.872886662492173,
      "grad_norm": 4.334254741668701,
      "learning_rate": 2.025904203323558e-05,
      "loss": 0.7848,
      "step": 11964
    },
    {
      "epoch": 1.8730432060112712,
      "grad_norm": 1.3288064002990723,
      "learning_rate": 2.0234604105571844e-05,
      "loss": 0.7798,
      "step": 11965
    },
    {
      "epoch": 1.8731997495303694,
      "grad_norm": 3.3582398891448975,
      "learning_rate": 2.0210166177908112e-05,
      "loss": 1.0681,
      "step": 11966
    },
    {
      "epoch": 1.8733562930494676,
      "grad_norm": 4.449483394622803,
      "learning_rate": 2.0185728250244377e-05,
      "loss": 0.8853,
      "step": 11967
    },
    {
      "epoch": 1.873512836568566,
      "grad_norm": 2.989022731781006,
      "learning_rate": 2.0161290322580642e-05,
      "loss": 0.6398,
      "step": 11968
    },
    {
      "epoch": 1.8736693800876645,
      "grad_norm": 4.922720432281494,
      "learning_rate": 2.013685239491691e-05,
      "loss": 0.7581,
      "step": 11969
    },
    {
      "epoch": 1.8738259236067627,
      "grad_norm": 4.551708221435547,
      "learning_rate": 2.0112414467253175e-05,
      "loss": 0.9456,
      "step": 11970
    },
    {
      "epoch": 1.873982467125861,
      "grad_norm": 2.336358070373535,
      "learning_rate": 2.008797653958944e-05,
      "loss": 0.8504,
      "step": 11971
    },
    {
      "epoch": 1.8741390106449591,
      "grad_norm": 2.310737371444702,
      "learning_rate": 2.0063538611925708e-05,
      "loss": 0.7855,
      "step": 11972
    },
    {
      "epoch": 1.8742955541640576,
      "grad_norm": 10.708081245422363,
      "learning_rate": 2.0039100684261973e-05,
      "loss": 0.775,
      "step": 11973
    },
    {
      "epoch": 1.874452097683156,
      "grad_norm": 3.0271284580230713,
      "learning_rate": 2.0014662756598237e-05,
      "loss": 0.7938,
      "step": 11974
    },
    {
      "epoch": 1.8746086412022542,
      "grad_norm": 7.497602462768555,
      "learning_rate": 1.9990224828934506e-05,
      "loss": 1.235,
      "step": 11975
    },
    {
      "epoch": 1.8747651847213525,
      "grad_norm": 7.694507598876953,
      "learning_rate": 1.996578690127077e-05,
      "loss": 1.2037,
      "step": 11976
    },
    {
      "epoch": 1.874921728240451,
      "grad_norm": 5.209584712982178,
      "learning_rate": 1.9941348973607035e-05,
      "loss": 0.6108,
      "step": 11977
    },
    {
      "epoch": 1.875078271759549,
      "grad_norm": 5.772160053253174,
      "learning_rate": 1.9916911045943303e-05,
      "loss": 0.9119,
      "step": 11978
    },
    {
      "epoch": 1.8752348152786475,
      "grad_norm": 2.1883044242858887,
      "learning_rate": 1.9892473118279568e-05,
      "loss": 1.2656,
      "step": 11979
    },
    {
      "epoch": 1.8753913587977458,
      "grad_norm": 3.705591917037964,
      "learning_rate": 1.9868035190615833e-05,
      "loss": 0.9636,
      "step": 11980
    },
    {
      "epoch": 1.875547902316844,
      "grad_norm": 2.439570188522339,
      "learning_rate": 1.98435972629521e-05,
      "loss": 1.0948,
      "step": 11981
    },
    {
      "epoch": 1.8757044458359424,
      "grad_norm": 2.082089424133301,
      "learning_rate": 1.9819159335288366e-05,
      "loss": 0.6857,
      "step": 11982
    },
    {
      "epoch": 1.8758609893550409,
      "grad_norm": 1.2534931898117065,
      "learning_rate": 1.979472140762463e-05,
      "loss": 0.3912,
      "step": 11983
    },
    {
      "epoch": 1.876017532874139,
      "grad_norm": 3.958746910095215,
      "learning_rate": 1.97702834799609e-05,
      "loss": 0.6163,
      "step": 11984
    },
    {
      "epoch": 1.8761740763932373,
      "grad_norm": 1.1258634328842163,
      "learning_rate": 1.9745845552297164e-05,
      "loss": 0.4008,
      "step": 11985
    },
    {
      "epoch": 1.8763306199123355,
      "grad_norm": 3.0614774227142334,
      "learning_rate": 1.972140762463343e-05,
      "loss": 0.7214,
      "step": 11986
    },
    {
      "epoch": 1.876487163431434,
      "grad_norm": 3.998211145401001,
      "learning_rate": 1.9696969696969697e-05,
      "loss": 1.0756,
      "step": 11987
    },
    {
      "epoch": 1.8766437069505324,
      "grad_norm": 2.4418106079101562,
      "learning_rate": 1.967253176930596e-05,
      "loss": 1.0684,
      "step": 11988
    },
    {
      "epoch": 1.8768002504696306,
      "grad_norm": 0.8472214937210083,
      "learning_rate": 1.9648093841642226e-05,
      "loss": 1.1258,
      "step": 11989
    },
    {
      "epoch": 1.8769567939887288,
      "grad_norm": 1.6514756679534912,
      "learning_rate": 1.9623655913978494e-05,
      "loss": 1.1072,
      "step": 11990
    },
    {
      "epoch": 1.877113337507827,
      "grad_norm": 1.2808918952941895,
      "learning_rate": 1.959921798631476e-05,
      "loss": 1.2211,
      "step": 11991
    },
    {
      "epoch": 1.8772698810269255,
      "grad_norm": 0.9649450182914734,
      "learning_rate": 1.9574780058651024e-05,
      "loss": 1.1116,
      "step": 11992
    },
    {
      "epoch": 1.877426424546024,
      "grad_norm": 1.2473208904266357,
      "learning_rate": 1.9550342130987292e-05,
      "loss": 1.1229,
      "step": 11993
    },
    {
      "epoch": 1.8775829680651221,
      "grad_norm": 1.2597932815551758,
      "learning_rate": 1.9525904203323557e-05,
      "loss": 1.1491,
      "step": 11994
    },
    {
      "epoch": 1.8777395115842204,
      "grad_norm": 1.4384278059005737,
      "learning_rate": 1.9501466275659822e-05,
      "loss": 1.136,
      "step": 11995
    },
    {
      "epoch": 1.8778960551033186,
      "grad_norm": 1.308000087738037,
      "learning_rate": 1.9477028347996087e-05,
      "loss": 1.0778,
      "step": 11996
    },
    {
      "epoch": 1.878052598622417,
      "grad_norm": 1.234831690788269,
      "learning_rate": 1.9452590420332355e-05,
      "loss": 1.0682,
      "step": 11997
    },
    {
      "epoch": 1.8782091421415155,
      "grad_norm": 0.9568355083465576,
      "learning_rate": 1.942815249266862e-05,
      "loss": 1.0062,
      "step": 11998
    },
    {
      "epoch": 1.8783656856606137,
      "grad_norm": 3.727036952972412,
      "learning_rate": 1.9403714565004884e-05,
      "loss": 1.0344,
      "step": 11999
    },
    {
      "epoch": 1.878522229179712,
      "grad_norm": 1.185569167137146,
      "learning_rate": 1.9379276637341153e-05,
      "loss": 1.0171,
      "step": 12000
    },
    {
      "epoch": 1.878522229179712,
      "eval_loss": 0.9464879631996155,
      "eval_runtime": 205.315,
      "eval_samples_per_second": 60.312,
      "eval_steps_per_second": 3.77,
      "eval_wer": 0.8378987185753448,
      "step": 12000
    },
    {
      "epoch": 1.8786787726988101,
      "grad_norm": 4.092573165893555,
      "learning_rate": 1.9354838709677417e-05,
      "loss": 1.4622,
      "step": 12001
    },
    {
      "epoch": 1.8788353162179086,
      "grad_norm": 2.7687883377075195,
      "learning_rate": 1.9330400782013682e-05,
      "loss": 1.3294,
      "step": 12002
    },
    {
      "epoch": 1.878991859737007,
      "grad_norm": 1.519518494606018,
      "learning_rate": 1.930596285434995e-05,
      "loss": 1.0254,
      "step": 12003
    },
    {
      "epoch": 1.8791484032561052,
      "grad_norm": 1.2389482259750366,
      "learning_rate": 1.9281524926686215e-05,
      "loss": 0.9684,
      "step": 12004
    },
    {
      "epoch": 1.8793049467752034,
      "grad_norm": 11.871675491333008,
      "learning_rate": 1.925708699902248e-05,
      "loss": 1.044,
      "step": 12005
    },
    {
      "epoch": 1.8794614902943017,
      "grad_norm": 2.5590877532958984,
      "learning_rate": 1.9232649071358748e-05,
      "loss": 0.875,
      "step": 12006
    },
    {
      "epoch": 1.8796180338134,
      "grad_norm": 2.1708011627197266,
      "learning_rate": 1.9208211143695013e-05,
      "loss": 0.9461,
      "step": 12007
    },
    {
      "epoch": 1.8797745773324985,
      "grad_norm": 5.7094502449035645,
      "learning_rate": 1.9183773216031278e-05,
      "loss": 0.9795,
      "step": 12008
    },
    {
      "epoch": 1.8799311208515967,
      "grad_norm": 2.582882881164551,
      "learning_rate": 1.9159335288367546e-05,
      "loss": 1.1206,
      "step": 12009
    },
    {
      "epoch": 1.880087664370695,
      "grad_norm": 2.906423330307007,
      "learning_rate": 1.913489736070381e-05,
      "loss": 1.0056,
      "step": 12010
    },
    {
      "epoch": 1.8802442078897934,
      "grad_norm": 4.6472320556640625,
      "learning_rate": 1.9110459433040076e-05,
      "loss": 0.9007,
      "step": 12011
    },
    {
      "epoch": 1.8804007514088916,
      "grad_norm": 3.4553442001342773,
      "learning_rate": 1.9086021505376344e-05,
      "loss": 0.851,
      "step": 12012
    },
    {
      "epoch": 1.88055729492799,
      "grad_norm": 3.4687905311584473,
      "learning_rate": 1.906158357771261e-05,
      "loss": 1.0446,
      "step": 12013
    },
    {
      "epoch": 1.8807138384470883,
      "grad_norm": 1.8787938356399536,
      "learning_rate": 1.9037145650048873e-05,
      "loss": 0.9464,
      "step": 12014
    },
    {
      "epoch": 1.8808703819661865,
      "grad_norm": 2.372514009475708,
      "learning_rate": 1.901270772238514e-05,
      "loss": 0.8582,
      "step": 12015
    },
    {
      "epoch": 1.881026925485285,
      "grad_norm": 3.6794726848602295,
      "learning_rate": 1.8988269794721406e-05,
      "loss": 0.7178,
      "step": 12016
    },
    {
      "epoch": 1.8811834690043834,
      "grad_norm": 1.5233488082885742,
      "learning_rate": 1.896383186705767e-05,
      "loss": 0.8946,
      "step": 12017
    },
    {
      "epoch": 1.8813400125234816,
      "grad_norm": 2.525583505630493,
      "learning_rate": 1.893939393939394e-05,
      "loss": 0.7768,
      "step": 12018
    },
    {
      "epoch": 1.8814965560425798,
      "grad_norm": 6.541940689086914,
      "learning_rate": 1.8914956011730204e-05,
      "loss": 1.2274,
      "step": 12019
    },
    {
      "epoch": 1.881653099561678,
      "grad_norm": 8.669783592224121,
      "learning_rate": 1.889051808406647e-05,
      "loss": 0.8303,
      "step": 12020
    },
    {
      "epoch": 1.8818096430807765,
      "grad_norm": 5.07046365737915,
      "learning_rate": 1.8866080156402737e-05,
      "loss": 1.1194,
      "step": 12021
    },
    {
      "epoch": 1.881966186599875,
      "grad_norm": 2.6451454162597656,
      "learning_rate": 1.8841642228739002e-05,
      "loss": 1.1634,
      "step": 12022
    },
    {
      "epoch": 1.8821227301189731,
      "grad_norm": 11.504127502441406,
      "learning_rate": 1.8817204301075267e-05,
      "loss": 1.2416,
      "step": 12023
    },
    {
      "epoch": 1.8822792736380713,
      "grad_norm": 4.630967617034912,
      "learning_rate": 1.8792766373411535e-05,
      "loss": 0.732,
      "step": 12024
    },
    {
      "epoch": 1.8824358171571696,
      "grad_norm": 5.2356648445129395,
      "learning_rate": 1.87683284457478e-05,
      "loss": 1.012,
      "step": 12025
    },
    {
      "epoch": 1.882592360676268,
      "grad_norm": 6.520271301269531,
      "learning_rate": 1.8743890518084065e-05,
      "loss": 0.7886,
      "step": 12026
    },
    {
      "epoch": 1.8827489041953664,
      "grad_norm": 8.99795913696289,
      "learning_rate": 1.871945259042033e-05,
      "loss": 1.1509,
      "step": 12027
    },
    {
      "epoch": 1.8829054477144647,
      "grad_norm": 2.0889909267425537,
      "learning_rate": 1.8695014662756598e-05,
      "loss": 0.8647,
      "step": 12028
    },
    {
      "epoch": 1.8830619912335629,
      "grad_norm": 9.168989181518555,
      "learning_rate": 1.8670576735092862e-05,
      "loss": 1.0876,
      "step": 12029
    },
    {
      "epoch": 1.883218534752661,
      "grad_norm": 3.345892906188965,
      "learning_rate": 1.8646138807429127e-05,
      "loss": 1.0499,
      "step": 12030
    },
    {
      "epoch": 1.8833750782717595,
      "grad_norm": 3.142423391342163,
      "learning_rate": 1.8621700879765395e-05,
      "loss": 1.5739,
      "step": 12031
    },
    {
      "epoch": 1.883531621790858,
      "grad_norm": 6.219061851501465,
      "learning_rate": 1.859726295210166e-05,
      "loss": 1.2463,
      "step": 12032
    },
    {
      "epoch": 1.8836881653099562,
      "grad_norm": 3.891333818435669,
      "learning_rate": 1.8572825024437925e-05,
      "loss": 0.8387,
      "step": 12033
    },
    {
      "epoch": 1.8838447088290544,
      "grad_norm": 2.0812160968780518,
      "learning_rate": 1.8548387096774193e-05,
      "loss": 0.5832,
      "step": 12034
    },
    {
      "epoch": 1.8840012523481526,
      "grad_norm": 2.1762547492980957,
      "learning_rate": 1.8523949169110458e-05,
      "loss": 0.3952,
      "step": 12035
    },
    {
      "epoch": 1.884157795867251,
      "grad_norm": 1.8617538213729858,
      "learning_rate": 1.8499511241446723e-05,
      "loss": 0.6069,
      "step": 12036
    },
    {
      "epoch": 1.8843143393863495,
      "grad_norm": 2.6567537784576416,
      "learning_rate": 1.847507331378299e-05,
      "loss": 0.8268,
      "step": 12037
    },
    {
      "epoch": 1.8844708829054477,
      "grad_norm": 3.485652208328247,
      "learning_rate": 1.8450635386119256e-05,
      "loss": 0.7337,
      "step": 12038
    },
    {
      "epoch": 1.884627426424546,
      "grad_norm": 1.9477510452270508,
      "learning_rate": 1.842619745845552e-05,
      "loss": 1.2784,
      "step": 12039
    },
    {
      "epoch": 1.8847839699436444,
      "grad_norm": 0.9730574488639832,
      "learning_rate": 1.840175953079179e-05,
      "loss": 1.221,
      "step": 12040
    },
    {
      "epoch": 1.8849405134627426,
      "grad_norm": 1.2898110151290894,
      "learning_rate": 1.8377321603128054e-05,
      "loss": 1.1966,
      "step": 12041
    },
    {
      "epoch": 1.885097056981841,
      "grad_norm": 0.9540336728096008,
      "learning_rate": 1.8352883675464318e-05,
      "loss": 1.1641,
      "step": 12042
    },
    {
      "epoch": 1.8852536005009393,
      "grad_norm": 4.748780727386475,
      "learning_rate": 1.8328445747800586e-05,
      "loss": 1.1545,
      "step": 12043
    },
    {
      "epoch": 1.8854101440200375,
      "grad_norm": 1.0382719039916992,
      "learning_rate": 1.830400782013685e-05,
      "loss": 1.2024,
      "step": 12044
    },
    {
      "epoch": 1.885566687539136,
      "grad_norm": 3.70912766456604,
      "learning_rate": 1.8279569892473116e-05,
      "loss": 1.3272,
      "step": 12045
    },
    {
      "epoch": 1.8857232310582341,
      "grad_norm": 2.094374656677246,
      "learning_rate": 1.8255131964809384e-05,
      "loss": 1.1662,
      "step": 12046
    },
    {
      "epoch": 1.8858797745773326,
      "grad_norm": 1.1259913444519043,
      "learning_rate": 1.823069403714565e-05,
      "loss": 1.0841,
      "step": 12047
    },
    {
      "epoch": 1.8860363180964308,
      "grad_norm": 1.507389783859253,
      "learning_rate": 1.8206256109481914e-05,
      "loss": 1.1885,
      "step": 12048
    },
    {
      "epoch": 1.886192861615529,
      "grad_norm": 1.2091017961502075,
      "learning_rate": 1.8181818181818182e-05,
      "loss": 1.1419,
      "step": 12049
    },
    {
      "epoch": 1.8863494051346275,
      "grad_norm": 2.392782688140869,
      "learning_rate": 1.8157380254154447e-05,
      "loss": 1.1869,
      "step": 12050
    },
    {
      "epoch": 1.886505948653726,
      "grad_norm": 3.752117156982422,
      "learning_rate": 1.813294232649071e-05,
      "loss": 1.1169,
      "step": 12051
    },
    {
      "epoch": 1.8866624921728241,
      "grad_norm": 3.1273210048675537,
      "learning_rate": 1.810850439882698e-05,
      "loss": 1.2533,
      "step": 12052
    },
    {
      "epoch": 1.8868190356919223,
      "grad_norm": 1.2055354118347168,
      "learning_rate": 1.8084066471163245e-05,
      "loss": 1.0157,
      "step": 12053
    },
    {
      "epoch": 1.8869755792110205,
      "grad_norm": 4.132384300231934,
      "learning_rate": 1.805962854349951e-05,
      "loss": 1.0156,
      "step": 12054
    },
    {
      "epoch": 1.887132122730119,
      "grad_norm": 1.6139036417007446,
      "learning_rate": 1.8035190615835778e-05,
      "loss": 1.3168,
      "step": 12055
    },
    {
      "epoch": 1.8872886662492174,
      "grad_norm": 1.806853175163269,
      "learning_rate": 1.801075268817204e-05,
      "loss": 1.38,
      "step": 12056
    },
    {
      "epoch": 1.8874452097683156,
      "grad_norm": 1.3997470140457153,
      "learning_rate": 1.7986314760508307e-05,
      "loss": 1.1021,
      "step": 12057
    },
    {
      "epoch": 1.8876017532874139,
      "grad_norm": 2.818284511566162,
      "learning_rate": 1.7961876832844575e-05,
      "loss": 1.0186,
      "step": 12058
    },
    {
      "epoch": 1.887758296806512,
      "grad_norm": 10.638510704040527,
      "learning_rate": 1.7937438905180837e-05,
      "loss": 1.4395,
      "step": 12059
    },
    {
      "epoch": 1.8879148403256105,
      "grad_norm": 4.576744079589844,
      "learning_rate": 1.7913000977517105e-05,
      "loss": 1.2109,
      "step": 12060
    },
    {
      "epoch": 1.888071383844709,
      "grad_norm": 9.343724250793457,
      "learning_rate": 1.7888563049853373e-05,
      "loss": 1.1902,
      "step": 12061
    },
    {
      "epoch": 1.8882279273638072,
      "grad_norm": 1.4243091344833374,
      "learning_rate": 1.7864125122189635e-05,
      "loss": 1.0696,
      "step": 12062
    },
    {
      "epoch": 1.8883844708829054,
      "grad_norm": 4.524175643920898,
      "learning_rate": 1.7839687194525903e-05,
      "loss": 1.3022,
      "step": 12063
    },
    {
      "epoch": 1.8885410144020036,
      "grad_norm": 2.529660224914551,
      "learning_rate": 1.781524926686217e-05,
      "loss": 1.1638,
      "step": 12064
    },
    {
      "epoch": 1.888697557921102,
      "grad_norm": 2.6119816303253174,
      "learning_rate": 1.7790811339198432e-05,
      "loss": 1.2616,
      "step": 12065
    },
    {
      "epoch": 1.8888541014402005,
      "grad_norm": 4.102001667022705,
      "learning_rate": 1.77663734115347e-05,
      "loss": 0.9599,
      "step": 12066
    },
    {
      "epoch": 1.8890106449592987,
      "grad_norm": 6.444525241851807,
      "learning_rate": 1.7741935483870965e-05,
      "loss": 1.0821,
      "step": 12067
    },
    {
      "epoch": 1.889167188478397,
      "grad_norm": 3.062340021133423,
      "learning_rate": 1.771749755620723e-05,
      "loss": 1.2128,
      "step": 12068
    },
    {
      "epoch": 1.8893237319974951,
      "grad_norm": 2.605454444885254,
      "learning_rate": 1.76930596285435e-05,
      "loss": 1.2174,
      "step": 12069
    },
    {
      "epoch": 1.8894802755165936,
      "grad_norm": 4.152836799621582,
      "learning_rate": 1.7668621700879763e-05,
      "loss": 1.1536,
      "step": 12070
    },
    {
      "epoch": 1.889636819035692,
      "grad_norm": 5.219440460205078,
      "learning_rate": 1.7644183773216028e-05,
      "loss": 1.4407,
      "step": 12071
    },
    {
      "epoch": 1.8897933625547902,
      "grad_norm": 3.174717426300049,
      "learning_rate": 1.7619745845552296e-05,
      "loss": 1.3388,
      "step": 12072
    },
    {
      "epoch": 1.8899499060738885,
      "grad_norm": 3.0966529846191406,
      "learning_rate": 1.759530791788856e-05,
      "loss": 1.0997,
      "step": 12073
    },
    {
      "epoch": 1.890106449592987,
      "grad_norm": 7.947587013244629,
      "learning_rate": 1.7570869990224826e-05,
      "loss": 1.1042,
      "step": 12074
    },
    {
      "epoch": 1.8902629931120851,
      "grad_norm": 5.379951477050781,
      "learning_rate": 1.7546432062561094e-05,
      "loss": 1.5781,
      "step": 12075
    },
    {
      "epoch": 1.8904195366311836,
      "grad_norm": 3.88952898979187,
      "learning_rate": 1.752199413489736e-05,
      "loss": 1.295,
      "step": 12076
    },
    {
      "epoch": 1.8905760801502818,
      "grad_norm": 3.303056001663208,
      "learning_rate": 1.7497556207233624e-05,
      "loss": 1.4006,
      "step": 12077
    },
    {
      "epoch": 1.89073262366938,
      "grad_norm": 3.121760845184326,
      "learning_rate": 1.7473118279569892e-05,
      "loss": 1.1342,
      "step": 12078
    },
    {
      "epoch": 1.8908891671884784,
      "grad_norm": 6.825216293334961,
      "learning_rate": 1.7448680351906157e-05,
      "loss": 1.2744,
      "step": 12079
    },
    {
      "epoch": 1.8910457107075767,
      "grad_norm": 3.57566237449646,
      "learning_rate": 1.742424242424242e-05,
      "loss": 1.141,
      "step": 12080
    },
    {
      "epoch": 1.891202254226675,
      "grad_norm": 6.941944122314453,
      "learning_rate": 1.739980449657869e-05,
      "loss": 1.4476,
      "step": 12081
    },
    {
      "epoch": 1.8913587977457733,
      "grad_norm": 3.8195106983184814,
      "learning_rate": 1.7375366568914954e-05,
      "loss": 1.4218,
      "step": 12082
    },
    {
      "epoch": 1.8915153412648715,
      "grad_norm": 2.770172119140625,
      "learning_rate": 1.735092864125122e-05,
      "loss": 0.7745,
      "step": 12083
    },
    {
      "epoch": 1.89167188478397,
      "grad_norm": 2.7398681640625,
      "learning_rate": 1.7326490713587487e-05,
      "loss": 0.8573,
      "step": 12084
    },
    {
      "epoch": 1.8918284283030684,
      "grad_norm": 5.767883777618408,
      "learning_rate": 1.7302052785923752e-05,
      "loss": 0.7147,
      "step": 12085
    },
    {
      "epoch": 1.8919849718221666,
      "grad_norm": 2.8740317821502686,
      "learning_rate": 1.7277614858260017e-05,
      "loss": 0.7094,
      "step": 12086
    },
    {
      "epoch": 1.8921415153412648,
      "grad_norm": 3.8529751300811768,
      "learning_rate": 1.7253176930596285e-05,
      "loss": 1.3801,
      "step": 12087
    },
    {
      "epoch": 1.892298058860363,
      "grad_norm": 1.6639562845230103,
      "learning_rate": 1.722873900293255e-05,
      "loss": 0.7619,
      "step": 12088
    },
    {
      "epoch": 1.8924546023794615,
      "grad_norm": 1.0102930068969727,
      "learning_rate": 1.7204301075268815e-05,
      "loss": 1.4944,
      "step": 12089
    },
    {
      "epoch": 1.89261114589856,
      "grad_norm": 1.1898506879806519,
      "learning_rate": 1.7179863147605083e-05,
      "loss": 1.4635,
      "step": 12090
    },
    {
      "epoch": 1.8927676894176582,
      "grad_norm": 1.3134690523147583,
      "learning_rate": 1.7155425219941348e-05,
      "loss": 1.491,
      "step": 12091
    },
    {
      "epoch": 1.8929242329367564,
      "grad_norm": 1.2267783880233765,
      "learning_rate": 1.7130987292277613e-05,
      "loss": 1.4765,
      "step": 12092
    },
    {
      "epoch": 1.8930807764558546,
      "grad_norm": 1.480941653251648,
      "learning_rate": 1.710654936461388e-05,
      "loss": 1.3881,
      "step": 12093
    },
    {
      "epoch": 1.893237319974953,
      "grad_norm": 1.0729501247406006,
      "learning_rate": 1.7082111436950146e-05,
      "loss": 1.3982,
      "step": 12094
    },
    {
      "epoch": 1.8933938634940515,
      "grad_norm": 1.248131275177002,
      "learning_rate": 1.705767350928641e-05,
      "loss": 1.5171,
      "step": 12095
    },
    {
      "epoch": 1.8935504070131497,
      "grad_norm": 2.5417959690093994,
      "learning_rate": 1.703323558162268e-05,
      "loss": 1.4509,
      "step": 12096
    },
    {
      "epoch": 1.893706950532248,
      "grad_norm": 1.5075997114181519,
      "learning_rate": 1.7008797653958943e-05,
      "loss": 1.3934,
      "step": 12097
    },
    {
      "epoch": 1.8938634940513461,
      "grad_norm": 5.500137805938721,
      "learning_rate": 1.6984359726295208e-05,
      "loss": 1.4268,
      "step": 12098
    },
    {
      "epoch": 1.8940200375704446,
      "grad_norm": 1.5557372570037842,
      "learning_rate": 1.6959921798631476e-05,
      "loss": 1.368,
      "step": 12099
    },
    {
      "epoch": 1.894176581089543,
      "grad_norm": 2.184826374053955,
      "learning_rate": 1.693548387096774e-05,
      "loss": 1.3948,
      "step": 12100
    },
    {
      "epoch": 1.8943331246086412,
      "grad_norm": 1.7609930038452148,
      "learning_rate": 1.6911045943304006e-05,
      "loss": 1.3585,
      "step": 12101
    },
    {
      "epoch": 1.8944896681277394,
      "grad_norm": 2.9163129329681396,
      "learning_rate": 1.6886608015640274e-05,
      "loss": 1.3926,
      "step": 12102
    },
    {
      "epoch": 1.8946462116468377,
      "grad_norm": 1.8992156982421875,
      "learning_rate": 1.686217008797654e-05,
      "loss": 1.2834,
      "step": 12103
    },
    {
      "epoch": 1.894802755165936,
      "grad_norm": 2.0363168716430664,
      "learning_rate": 1.6837732160312804e-05,
      "loss": 1.4879,
      "step": 12104
    },
    {
      "epoch": 1.8949592986850345,
      "grad_norm": 1.4370715618133545,
      "learning_rate": 1.6813294232649072e-05,
      "loss": 1.119,
      "step": 12105
    },
    {
      "epoch": 1.8951158422041328,
      "grad_norm": 1.8881416320800781,
      "learning_rate": 1.6788856304985337e-05,
      "loss": 1.2856,
      "step": 12106
    },
    {
      "epoch": 1.895272385723231,
      "grad_norm": 1.4255024194717407,
      "learning_rate": 1.67644183773216e-05,
      "loss": 1.3787,
      "step": 12107
    },
    {
      "epoch": 1.8954289292423294,
      "grad_norm": 2.249227285385132,
      "learning_rate": 1.6739980449657866e-05,
      "loss": 1.2557,
      "step": 12108
    },
    {
      "epoch": 1.8955854727614276,
      "grad_norm": 4.423599720001221,
      "learning_rate": 1.6715542521994134e-05,
      "loss": 1.3487,
      "step": 12109
    },
    {
      "epoch": 1.895742016280526,
      "grad_norm": 4.681179523468018,
      "learning_rate": 1.66911045943304e-05,
      "loss": 1.192,
      "step": 12110
    },
    {
      "epoch": 1.8958985597996243,
      "grad_norm": 1.816692590713501,
      "learning_rate": 1.6666666666666664e-05,
      "loss": 1.1183,
      "step": 12111
    },
    {
      "epoch": 1.8960551033187225,
      "grad_norm": 4.783669948577881,
      "learning_rate": 1.6642228739002932e-05,
      "loss": 1.3788,
      "step": 12112
    },
    {
      "epoch": 1.896211646837821,
      "grad_norm": 3.048429250717163,
      "learning_rate": 1.6617790811339197e-05,
      "loss": 1.1074,
      "step": 12113
    },
    {
      "epoch": 1.8963681903569192,
      "grad_norm": 3.0569400787353516,
      "learning_rate": 1.6593352883675462e-05,
      "loss": 1.274,
      "step": 12114
    },
    {
      "epoch": 1.8965247338760176,
      "grad_norm": 2.585090398788452,
      "learning_rate": 1.656891495601173e-05,
      "loss": 1.0185,
      "step": 12115
    },
    {
      "epoch": 1.8966812773951158,
      "grad_norm": 2.0414528846740723,
      "learning_rate": 1.6544477028347995e-05,
      "loss": 1.3508,
      "step": 12116
    },
    {
      "epoch": 1.896837820914214,
      "grad_norm": 2.784245729446411,
      "learning_rate": 1.652003910068426e-05,
      "loss": 1.2198,
      "step": 12117
    },
    {
      "epoch": 1.8969943644333125,
      "grad_norm": 2.629258394241333,
      "learning_rate": 1.6495601173020528e-05,
      "loss": 1.103,
      "step": 12118
    },
    {
      "epoch": 1.897150907952411,
      "grad_norm": 6.373242378234863,
      "learning_rate": 1.6471163245356793e-05,
      "loss": 1.291,
      "step": 12119
    },
    {
      "epoch": 1.8973074514715091,
      "grad_norm": 3.1650185585021973,
      "learning_rate": 1.6446725317693057e-05,
      "loss": 1.0345,
      "step": 12120
    },
    {
      "epoch": 1.8974639949906074,
      "grad_norm": 2.357555389404297,
      "learning_rate": 1.6422287390029326e-05,
      "loss": 0.8684,
      "step": 12121
    },
    {
      "epoch": 1.8976205385097056,
      "grad_norm": 3.3148272037506104,
      "learning_rate": 1.639784946236559e-05,
      "loss": 1.0331,
      "step": 12122
    },
    {
      "epoch": 1.897777082028804,
      "grad_norm": 3.1584115028381348,
      "learning_rate": 1.6373411534701855e-05,
      "loss": 0.9614,
      "step": 12123
    },
    {
      "epoch": 1.8979336255479025,
      "grad_norm": 5.584422588348389,
      "learning_rate": 1.6348973607038123e-05,
      "loss": 1.0514,
      "step": 12124
    },
    {
      "epoch": 1.8980901690670007,
      "grad_norm": 3.8823444843292236,
      "learning_rate": 1.6324535679374388e-05,
      "loss": 1.5685,
      "step": 12125
    },
    {
      "epoch": 1.898246712586099,
      "grad_norm": 3.763765811920166,
      "learning_rate": 1.6300097751710653e-05,
      "loss": 1.2219,
      "step": 12126
    },
    {
      "epoch": 1.8984032561051971,
      "grad_norm": 2.749150037765503,
      "learning_rate": 1.6275659824046918e-05,
      "loss": 1.0447,
      "step": 12127
    },
    {
      "epoch": 1.8985597996242956,
      "grad_norm": 5.038095951080322,
      "learning_rate": 1.6251221896383186e-05,
      "loss": 1.004,
      "step": 12128
    },
    {
      "epoch": 1.898716343143394,
      "grad_norm": 8.337298393249512,
      "learning_rate": 1.622678396871945e-05,
      "loss": 1.027,
      "step": 12129
    },
    {
      "epoch": 1.8988728866624922,
      "grad_norm": 3.5762979984283447,
      "learning_rate": 1.6202346041055716e-05,
      "loss": 1.4483,
      "step": 12130
    },
    {
      "epoch": 1.8990294301815904,
      "grad_norm": 2.6334099769592285,
      "learning_rate": 1.6177908113391984e-05,
      "loss": 1.2446,
      "step": 12131
    },
    {
      "epoch": 1.8991859737006886,
      "grad_norm": 2.6013121604919434,
      "learning_rate": 1.615347018572825e-05,
      "loss": 1.2401,
      "step": 12132
    },
    {
      "epoch": 1.899342517219787,
      "grad_norm": 3.092857599258423,
      "learning_rate": 1.6129032258064513e-05,
      "loss": 1.249,
      "step": 12133
    },
    {
      "epoch": 1.8994990607388855,
      "grad_norm": 4.714329719543457,
      "learning_rate": 1.610459433040078e-05,
      "loss": 0.4669,
      "step": 12134
    },
    {
      "epoch": 1.8996556042579837,
      "grad_norm": 1.6379783153533936,
      "learning_rate": 1.6080156402737046e-05,
      "loss": 0.4258,
      "step": 12135
    },
    {
      "epoch": 1.899812147777082,
      "grad_norm": 3.6598150730133057,
      "learning_rate": 1.605571847507331e-05,
      "loss": 0.6978,
      "step": 12136
    },
    {
      "epoch": 1.8999686912961802,
      "grad_norm": 4.708147048950195,
      "learning_rate": 1.603128054740958e-05,
      "loss": 1.3925,
      "step": 12137
    },
    {
      "epoch": 1.9001252348152786,
      "grad_norm": 1.7821239233016968,
      "learning_rate": 1.6006842619745844e-05,
      "loss": 0.5404,
      "step": 12138
    },
    {
      "epoch": 1.900281778334377,
      "grad_norm": 1.8768203258514404,
      "learning_rate": 1.598240469208211e-05,
      "loss": 1.1651,
      "step": 12139
    },
    {
      "epoch": 1.9004383218534753,
      "grad_norm": 0.9923833012580872,
      "learning_rate": 1.5957966764418374e-05,
      "loss": 1.237,
      "step": 12140
    },
    {
      "epoch": 1.9005948653725735,
      "grad_norm": 0.9963833689689636,
      "learning_rate": 1.5933528836754642e-05,
      "loss": 1.2096,
      "step": 12141
    },
    {
      "epoch": 1.900751408891672,
      "grad_norm": 1.4401146173477173,
      "learning_rate": 1.5909090909090907e-05,
      "loss": 1.2101,
      "step": 12142
    },
    {
      "epoch": 1.9009079524107702,
      "grad_norm": 0.8284012079238892,
      "learning_rate": 1.588465298142717e-05,
      "loss": 1.0725,
      "step": 12143
    },
    {
      "epoch": 1.9010644959298686,
      "grad_norm": 1.0937573909759521,
      "learning_rate": 1.586021505376344e-05,
      "loss": 1.2081,
      "step": 12144
    },
    {
      "epoch": 1.9012210394489668,
      "grad_norm": 0.7755000591278076,
      "learning_rate": 1.5835777126099705e-05,
      "loss": 1.0808,
      "step": 12145
    },
    {
      "epoch": 1.901377582968065,
      "grad_norm": 1.1034554243087769,
      "learning_rate": 1.581133919843597e-05,
      "loss": 1.183,
      "step": 12146
    },
    {
      "epoch": 1.9015341264871635,
      "grad_norm": 1.378918170928955,
      "learning_rate": 1.5786901270772238e-05,
      "loss": 1.0971,
      "step": 12147
    },
    {
      "epoch": 1.9016906700062617,
      "grad_norm": 1.6147481203079224,
      "learning_rate": 1.5762463343108502e-05,
      "loss": 1.2029,
      "step": 12148
    },
    {
      "epoch": 1.9018472135253601,
      "grad_norm": 1.2892570495605469,
      "learning_rate": 1.5738025415444767e-05,
      "loss": 1.002,
      "step": 12149
    },
    {
      "epoch": 1.9020037570444583,
      "grad_norm": 1.0253342390060425,
      "learning_rate": 1.5713587487781035e-05,
      "loss": 1.105,
      "step": 12150
    },
    {
      "epoch": 1.9021603005635566,
      "grad_norm": 1.6563441753387451,
      "learning_rate": 1.56891495601173e-05,
      "loss": 1.1433,
      "step": 12151
    },
    {
      "epoch": 1.902316844082655,
      "grad_norm": 1.512212872505188,
      "learning_rate": 1.5664711632453565e-05,
      "loss": 0.9589,
      "step": 12152
    },
    {
      "epoch": 1.9024733876017534,
      "grad_norm": 0.9118117690086365,
      "learning_rate": 1.5640273704789833e-05,
      "loss": 1.0516,
      "step": 12153
    },
    {
      "epoch": 1.9026299311208517,
      "grad_norm": 2.5056610107421875,
      "learning_rate": 1.5615835777126098e-05,
      "loss": 1.2032,
      "step": 12154
    },
    {
      "epoch": 1.9027864746399499,
      "grad_norm": 3.2829456329345703,
      "learning_rate": 1.5591397849462363e-05,
      "loss": 1.0465,
      "step": 12155
    },
    {
      "epoch": 1.902943018159048,
      "grad_norm": 4.280078411102295,
      "learning_rate": 1.556695992179863e-05,
      "loss": 1.1008,
      "step": 12156
    },
    {
      "epoch": 1.9030995616781465,
      "grad_norm": 1.5060573816299438,
      "learning_rate": 1.5542521994134896e-05,
      "loss": 1.0278,
      "step": 12157
    },
    {
      "epoch": 1.903256105197245,
      "grad_norm": 2.12257981300354,
      "learning_rate": 1.551808406647116e-05,
      "loss": 1.1257,
      "step": 12158
    },
    {
      "epoch": 1.9034126487163432,
      "grad_norm": 2.310483694076538,
      "learning_rate": 1.549364613880743e-05,
      "loss": 1.2765,
      "step": 12159
    },
    {
      "epoch": 1.9035691922354414,
      "grad_norm": 2.682175636291504,
      "learning_rate": 1.5469208211143694e-05,
      "loss": 1.1797,
      "step": 12160
    },
    {
      "epoch": 1.9037257357545396,
      "grad_norm": 3.6471972465515137,
      "learning_rate": 1.544477028347996e-05,
      "loss": 1.0389,
      "step": 12161
    },
    {
      "epoch": 1.903882279273638,
      "grad_norm": 2.102436065673828,
      "learning_rate": 1.5420332355816226e-05,
      "loss": 1.0323,
      "step": 12162
    },
    {
      "epoch": 1.9040388227927365,
      "grad_norm": 4.041779041290283,
      "learning_rate": 1.539589442815249e-05,
      "loss": 1.1483,
      "step": 12163
    },
    {
      "epoch": 1.9041953663118347,
      "grad_norm": 3.722378730773926,
      "learning_rate": 1.5371456500488756e-05,
      "loss": 1.1631,
      "step": 12164
    },
    {
      "epoch": 1.904351909830933,
      "grad_norm": 4.074342727661133,
      "learning_rate": 1.5347018572825024e-05,
      "loss": 1.3785,
      "step": 12165
    },
    {
      "epoch": 1.9045084533500312,
      "grad_norm": 6.567094326019287,
      "learning_rate": 1.532258064516129e-05,
      "loss": 0.945,
      "step": 12166
    },
    {
      "epoch": 1.9046649968691296,
      "grad_norm": 7.025415897369385,
      "learning_rate": 1.5298142717497554e-05,
      "loss": 1.2596,
      "step": 12167
    },
    {
      "epoch": 1.904821540388228,
      "grad_norm": 3.313201904296875,
      "learning_rate": 1.5273704789833822e-05,
      "loss": 1.2277,
      "step": 12168
    },
    {
      "epoch": 1.9049780839073263,
      "grad_norm": 4.0791120529174805,
      "learning_rate": 1.5249266862170087e-05,
      "loss": 1.2767,
      "step": 12169
    },
    {
      "epoch": 1.9051346274264245,
      "grad_norm": 4.89728307723999,
      "learning_rate": 1.5224828934506352e-05,
      "loss": 1.1568,
      "step": 12170
    },
    {
      "epoch": 1.9052911709455227,
      "grad_norm": 3.8333678245544434,
      "learning_rate": 1.5200391006842618e-05,
      "loss": 1.1248,
      "step": 12171
    },
    {
      "epoch": 1.9054477144646211,
      "grad_norm": 4.54195499420166,
      "learning_rate": 1.5175953079178883e-05,
      "loss": 1.0531,
      "step": 12172
    },
    {
      "epoch": 1.9056042579837196,
      "grad_norm": 4.03914737701416,
      "learning_rate": 1.515151515151515e-05,
      "loss": 1.124,
      "step": 12173
    },
    {
      "epoch": 1.9057608015028178,
      "grad_norm": 7.725468635559082,
      "learning_rate": 1.5127077223851416e-05,
      "loss": 1.2547,
      "step": 12174
    },
    {
      "epoch": 1.905917345021916,
      "grad_norm": 4.000185966491699,
      "learning_rate": 1.510263929618768e-05,
      "loss": 1.0578,
      "step": 12175
    },
    {
      "epoch": 1.9060738885410144,
      "grad_norm": 3.3531341552734375,
      "learning_rate": 1.5078201368523947e-05,
      "loss": 1.0488,
      "step": 12176
    },
    {
      "epoch": 1.9062304320601127,
      "grad_norm": 1.9735767841339111,
      "learning_rate": 1.5053763440860214e-05,
      "loss": 1.0028,
      "step": 12177
    },
    {
      "epoch": 1.906386975579211,
      "grad_norm": 6.0913262367248535,
      "learning_rate": 1.5029325513196479e-05,
      "loss": 1.0639,
      "step": 12178
    },
    {
      "epoch": 1.9065435190983093,
      "grad_norm": 1.993403434753418,
      "learning_rate": 1.5004887585532745e-05,
      "loss": 0.9197,
      "step": 12179
    },
    {
      "epoch": 1.9067000626174075,
      "grad_norm": 3.3327815532684326,
      "learning_rate": 1.4980449657869012e-05,
      "loss": 1.3543,
      "step": 12180
    },
    {
      "epoch": 1.906856606136506,
      "grad_norm": 3.145503044128418,
      "learning_rate": 1.4956011730205276e-05,
      "loss": 0.9099,
      "step": 12181
    },
    {
      "epoch": 1.9070131496556044,
      "grad_norm": 2.0428433418273926,
      "learning_rate": 1.4931573802541543e-05,
      "loss": 0.6648,
      "step": 12182
    },
    {
      "epoch": 1.9071696931747026,
      "grad_norm": 1.2480318546295166,
      "learning_rate": 1.490713587487781e-05,
      "loss": 0.533,
      "step": 12183
    },
    {
      "epoch": 1.9073262366938009,
      "grad_norm": 3.0121002197265625,
      "learning_rate": 1.4882697947214074e-05,
      "loss": 1.3884,
      "step": 12184
    },
    {
      "epoch": 1.907482780212899,
      "grad_norm": 1.4688293933868408,
      "learning_rate": 1.485826001955034e-05,
      "loss": 0.562,
      "step": 12185
    },
    {
      "epoch": 1.9076393237319975,
      "grad_norm": 1.736817479133606,
      "learning_rate": 1.4833822091886607e-05,
      "loss": 0.3936,
      "step": 12186
    },
    {
      "epoch": 1.907795867251096,
      "grad_norm": 2.580030918121338,
      "learning_rate": 1.4809384164222872e-05,
      "loss": 0.5264,
      "step": 12187
    },
    {
      "epoch": 1.9079524107701942,
      "grad_norm": 5.177095413208008,
      "learning_rate": 1.4784946236559138e-05,
      "loss": 1.3178,
      "step": 12188
    },
    {
      "epoch": 1.9081089542892924,
      "grad_norm": 0.8959721326828003,
      "learning_rate": 1.4760508308895405e-05,
      "loss": 1.284,
      "step": 12189
    },
    {
      "epoch": 1.9082654978083906,
      "grad_norm": 1.049804449081421,
      "learning_rate": 1.473607038123167e-05,
      "loss": 1.3241,
      "step": 12190
    },
    {
      "epoch": 1.908422041327489,
      "grad_norm": 4.327507495880127,
      "learning_rate": 1.4711632453567936e-05,
      "loss": 1.1876,
      "step": 12191
    },
    {
      "epoch": 1.9085785848465875,
      "grad_norm": 2.699472188949585,
      "learning_rate": 1.4687194525904203e-05,
      "loss": 1.2219,
      "step": 12192
    },
    {
      "epoch": 1.9087351283656857,
      "grad_norm": 1.9174275398254395,
      "learning_rate": 1.4662756598240468e-05,
      "loss": 1.2485,
      "step": 12193
    },
    {
      "epoch": 1.908891671884784,
      "grad_norm": 3.461306095123291,
      "learning_rate": 1.4638318670576734e-05,
      "loss": 1.3244,
      "step": 12194
    },
    {
      "epoch": 1.9090482154038821,
      "grad_norm": 2.7849996089935303,
      "learning_rate": 1.4613880742913e-05,
      "loss": 1.2386,
      "step": 12195
    },
    {
      "epoch": 1.9092047589229806,
      "grad_norm": 3.3640100955963135,
      "learning_rate": 1.4589442815249265e-05,
      "loss": 1.2349,
      "step": 12196
    },
    {
      "epoch": 1.909361302442079,
      "grad_norm": 0.8825156688690186,
      "learning_rate": 1.4565004887585532e-05,
      "loss": 1.253,
      "step": 12197
    },
    {
      "epoch": 1.9095178459611772,
      "grad_norm": 2.1742002964019775,
      "learning_rate": 1.4540566959921798e-05,
      "loss": 1.1779,
      "step": 12198
    },
    {
      "epoch": 1.9096743894802755,
      "grad_norm": 1.2652525901794434,
      "learning_rate": 1.4516129032258063e-05,
      "loss": 1.1674,
      "step": 12199
    },
    {
      "epoch": 1.9098309329993737,
      "grad_norm": 1.1848137378692627,
      "learning_rate": 1.449169110459433e-05,
      "loss": 1.1022,
      "step": 12200
    },
    {
      "epoch": 1.9099874765184721,
      "grad_norm": 4.88668155670166,
      "learning_rate": 1.4467253176930596e-05,
      "loss": 0.9741,
      "step": 12201
    },
    {
      "epoch": 1.9101440200375706,
      "grad_norm": 1.7493770122528076,
      "learning_rate": 1.4442815249266861e-05,
      "loss": 1.3092,
      "step": 12202
    },
    {
      "epoch": 1.9103005635566688,
      "grad_norm": 2.5054683685302734,
      "learning_rate": 1.4418377321603127e-05,
      "loss": 1.1745,
      "step": 12203
    },
    {
      "epoch": 1.910457107075767,
      "grad_norm": 1.4955793619155884,
      "learning_rate": 1.4393939393939392e-05,
      "loss": 1.1197,
      "step": 12204
    },
    {
      "epoch": 1.9106136505948652,
      "grad_norm": 1.2048388719558716,
      "learning_rate": 1.4369501466275659e-05,
      "loss": 1.1417,
      "step": 12205
    },
    {
      "epoch": 1.9107701941139636,
      "grad_norm": 1.501969814300537,
      "learning_rate": 1.4345063538611925e-05,
      "loss": 1.2411,
      "step": 12206
    },
    {
      "epoch": 1.910926737633062,
      "grad_norm": 1.6169402599334717,
      "learning_rate": 1.432062561094819e-05,
      "loss": 1.1179,
      "step": 12207
    },
    {
      "epoch": 1.9110832811521603,
      "grad_norm": 2.1456000804901123,
      "learning_rate": 1.4296187683284456e-05,
      "loss": 1.1619,
      "step": 12208
    },
    {
      "epoch": 1.9112398246712585,
      "grad_norm": 3.0159125328063965,
      "learning_rate": 1.4271749755620723e-05,
      "loss": 1.2492,
      "step": 12209
    },
    {
      "epoch": 1.911396368190357,
      "grad_norm": 1.6835179328918457,
      "learning_rate": 1.4247311827956988e-05,
      "loss": 1.0944,
      "step": 12210
    },
    {
      "epoch": 1.9115529117094552,
      "grad_norm": 2.42319393157959,
      "learning_rate": 1.4222873900293254e-05,
      "loss": 0.9977,
      "step": 12211
    },
    {
      "epoch": 1.9117094552285536,
      "grad_norm": 1.6393407583236694,
      "learning_rate": 1.419843597262952e-05,
      "loss": 1.3491,
      "step": 12212
    },
    {
      "epoch": 1.9118659987476518,
      "grad_norm": 4.471738815307617,
      "learning_rate": 1.4173998044965786e-05,
      "loss": 1.274,
      "step": 12213
    },
    {
      "epoch": 1.91202254226675,
      "grad_norm": 2.172912836074829,
      "learning_rate": 1.4149560117302052e-05,
      "loss": 1.1159,
      "step": 12214
    },
    {
      "epoch": 1.9121790857858485,
      "grad_norm": 2.280036449432373,
      "learning_rate": 1.4125122189638319e-05,
      "loss": 1.0586,
      "step": 12215
    },
    {
      "epoch": 1.912335629304947,
      "grad_norm": 5.947153568267822,
      "learning_rate": 1.4100684261974583e-05,
      "loss": 0.9635,
      "step": 12216
    },
    {
      "epoch": 1.9124921728240452,
      "grad_norm": 5.538031101226807,
      "learning_rate": 1.407624633431085e-05,
      "loss": 1.4723,
      "step": 12217
    },
    {
      "epoch": 1.9126487163431434,
      "grad_norm": 3.303689479827881,
      "learning_rate": 1.4051808406647116e-05,
      "loss": 1.4025,
      "step": 12218
    },
    {
      "epoch": 1.9128052598622416,
      "grad_norm": 2.7196755409240723,
      "learning_rate": 1.4027370478983381e-05,
      "loss": 1.0212,
      "step": 12219
    },
    {
      "epoch": 1.91296180338134,
      "grad_norm": 14.392826080322266,
      "learning_rate": 1.4002932551319648e-05,
      "loss": 1.114,
      "step": 12220
    },
    {
      "epoch": 1.9131183469004385,
      "grad_norm": 5.36333703994751,
      "learning_rate": 1.3978494623655914e-05,
      "loss": 1.1304,
      "step": 12221
    },
    {
      "epoch": 1.9132748904195367,
      "grad_norm": 2.031944513320923,
      "learning_rate": 1.3954056695992177e-05,
      "loss": 0.9993,
      "step": 12222
    },
    {
      "epoch": 1.913431433938635,
      "grad_norm": 7.611963748931885,
      "learning_rate": 1.3929618768328445e-05,
      "loss": 1.1506,
      "step": 12223
    },
    {
      "epoch": 1.9135879774577331,
      "grad_norm": 2.747000217437744,
      "learning_rate": 1.3905180840664712e-05,
      "loss": 0.9725,
      "step": 12224
    },
    {
      "epoch": 1.9137445209768316,
      "grad_norm": 5.841665267944336,
      "learning_rate": 1.3880742913000975e-05,
      "loss": 0.7786,
      "step": 12225
    },
    {
      "epoch": 1.91390106449593,
      "grad_norm": 6.18753719329834,
      "learning_rate": 1.3856304985337243e-05,
      "loss": 1.0598,
      "step": 12226
    },
    {
      "epoch": 1.9140576080150282,
      "grad_norm": 3.2494606971740723,
      "learning_rate": 1.383186705767351e-05,
      "loss": 1.1976,
      "step": 12227
    },
    {
      "epoch": 1.9142141515341264,
      "grad_norm": 7.703588008880615,
      "learning_rate": 1.3807429130009773e-05,
      "loss": 1.413,
      "step": 12228
    },
    {
      "epoch": 1.9143706950532247,
      "grad_norm": 3.169772148132324,
      "learning_rate": 1.378299120234604e-05,
      "loss": 1.4166,
      "step": 12229
    },
    {
      "epoch": 1.914527238572323,
      "grad_norm": 3.6064870357513428,
      "learning_rate": 1.3758553274682307e-05,
      "loss": 1.3009,
      "step": 12230
    },
    {
      "epoch": 1.9146837820914215,
      "grad_norm": 2.5471582412719727,
      "learning_rate": 1.373411534701857e-05,
      "loss": 0.8335,
      "step": 12231
    },
    {
      "epoch": 1.9148403256105198,
      "grad_norm": 3.2321269512176514,
      "learning_rate": 1.3709677419354837e-05,
      "loss": 1.1152,
      "step": 12232
    },
    {
      "epoch": 1.914996869129618,
      "grad_norm": 3.2195959091186523,
      "learning_rate": 1.3685239491691105e-05,
      "loss": 0.8062,
      "step": 12233
    },
    {
      "epoch": 1.9151534126487162,
      "grad_norm": 2.1859567165374756,
      "learning_rate": 1.3660801564027368e-05,
      "loss": 0.541,
      "step": 12234
    },
    {
      "epoch": 1.9153099561678146,
      "grad_norm": 1.7205095291137695,
      "learning_rate": 1.3636363636363635e-05,
      "loss": 0.5953,
      "step": 12235
    },
    {
      "epoch": 1.915466499686913,
      "grad_norm": 4.676856517791748,
      "learning_rate": 1.36119257086999e-05,
      "loss": 0.5505,
      "step": 12236
    },
    {
      "epoch": 1.9156230432060113,
      "grad_norm": 2.9190919399261475,
      "learning_rate": 1.3587487781036166e-05,
      "loss": 0.9582,
      "step": 12237
    },
    {
      "epoch": 1.9157795867251095,
      "grad_norm": 2.5346970558166504,
      "learning_rate": 1.3563049853372433e-05,
      "loss": 1.1218,
      "step": 12238
    },
    {
      "epoch": 1.9159361302442077,
      "grad_norm": 0.8043964505195618,
      "learning_rate": 1.3538611925708697e-05,
      "loss": 1.4646,
      "step": 12239
    },
    {
      "epoch": 1.9160926737633062,
      "grad_norm": 1.0437607765197754,
      "learning_rate": 1.3514173998044964e-05,
      "loss": 1.4516,
      "step": 12240
    },
    {
      "epoch": 1.9162492172824046,
      "grad_norm": 1.5638072490692139,
      "learning_rate": 1.348973607038123e-05,
      "loss": 1.4577,
      "step": 12241
    },
    {
      "epoch": 1.9164057608015028,
      "grad_norm": 1.5612990856170654,
      "learning_rate": 1.3465298142717495e-05,
      "loss": 1.4553,
      "step": 12242
    },
    {
      "epoch": 1.916562304320601,
      "grad_norm": 1.4811925888061523,
      "learning_rate": 1.3440860215053762e-05,
      "loss": 1.5817,
      "step": 12243
    },
    {
      "epoch": 1.9167188478396995,
      "grad_norm": 1.6641279458999634,
      "learning_rate": 1.3416422287390028e-05,
      "loss": 1.4004,
      "step": 12244
    },
    {
      "epoch": 1.9168753913587977,
      "grad_norm": 1.62802255153656,
      "learning_rate": 1.3391984359726293e-05,
      "loss": 1.3879,
      "step": 12245
    },
    {
      "epoch": 1.9170319348778961,
      "grad_norm": 1.1935174465179443,
      "learning_rate": 1.336754643206256e-05,
      "loss": 1.3205,
      "step": 12246
    },
    {
      "epoch": 1.9171884783969944,
      "grad_norm": 2.306370496749878,
      "learning_rate": 1.3343108504398826e-05,
      "loss": 1.2546,
      "step": 12247
    },
    {
      "epoch": 1.9173450219160926,
      "grad_norm": 3.6753339767456055,
      "learning_rate": 1.331867057673509e-05,
      "loss": 1.4271,
      "step": 12248
    },
    {
      "epoch": 1.917501565435191,
      "grad_norm": 2.2451188564300537,
      "learning_rate": 1.3294232649071357e-05,
      "loss": 1.318,
      "step": 12249
    },
    {
      "epoch": 1.9176581089542895,
      "grad_norm": 3.015744209289551,
      "learning_rate": 1.3269794721407624e-05,
      "loss": 1.3066,
      "step": 12250
    },
    {
      "epoch": 1.9178146524733877,
      "grad_norm": 0.8539578318595886,
      "learning_rate": 1.3245356793743889e-05,
      "loss": 1.2688,
      "step": 12251
    },
    {
      "epoch": 1.9179711959924859,
      "grad_norm": 6.493170261383057,
      "learning_rate": 1.3220918866080155e-05,
      "loss": 1.1816,
      "step": 12252
    },
    {
      "epoch": 1.918127739511584,
      "grad_norm": 1.088154673576355,
      "learning_rate": 1.3196480938416422e-05,
      "loss": 1.213,
      "step": 12253
    },
    {
      "epoch": 1.9182842830306825,
      "grad_norm": 2.4062836170196533,
      "learning_rate": 1.3172043010752686e-05,
      "loss": 1.1441,
      "step": 12254
    },
    {
      "epoch": 1.918440826549781,
      "grad_norm": 1.412260890007019,
      "learning_rate": 1.3147605083088953e-05,
      "loss": 1.3002,
      "step": 12255
    },
    {
      "epoch": 1.9185973700688792,
      "grad_norm": 3.7139716148376465,
      "learning_rate": 1.312316715542522e-05,
      "loss": 1.1326,
      "step": 12256
    },
    {
      "epoch": 1.9187539135879774,
      "grad_norm": 4.428962707519531,
      "learning_rate": 1.3098729227761484e-05,
      "loss": 1.2002,
      "step": 12257
    },
    {
      "epoch": 1.9189104571070756,
      "grad_norm": 1.157462239265442,
      "learning_rate": 1.307429130009775e-05,
      "loss": 1.1597,
      "step": 12258
    },
    {
      "epoch": 1.919067000626174,
      "grad_norm": 3.1243197917938232,
      "learning_rate": 1.3049853372434017e-05,
      "loss": 1.1902,
      "step": 12259
    },
    {
      "epoch": 1.9192235441452725,
      "grad_norm": 1.9715791940689087,
      "learning_rate": 1.3025415444770282e-05,
      "loss": 1.0987,
      "step": 12260
    },
    {
      "epoch": 1.9193800876643707,
      "grad_norm": 8.637343406677246,
      "learning_rate": 1.3000977517106548e-05,
      "loss": 1.2952,
      "step": 12261
    },
    {
      "epoch": 1.919536631183469,
      "grad_norm": 6.371841907501221,
      "learning_rate": 1.2976539589442815e-05,
      "loss": 1.6552,
      "step": 12262
    },
    {
      "epoch": 1.9196931747025672,
      "grad_norm": 2.1483147144317627,
      "learning_rate": 1.295210166177908e-05,
      "loss": 1.0953,
      "step": 12263
    },
    {
      "epoch": 1.9198497182216656,
      "grad_norm": 1.9315729141235352,
      "learning_rate": 1.2927663734115346e-05,
      "loss": 1.0004,
      "step": 12264
    },
    {
      "epoch": 1.920006261740764,
      "grad_norm": 3.5723578929901123,
      "learning_rate": 1.2903225806451613e-05,
      "loss": 0.8629,
      "step": 12265
    },
    {
      "epoch": 1.9201628052598623,
      "grad_norm": 2.428359270095825,
      "learning_rate": 1.2878787878787878e-05,
      "loss": 1.2251,
      "step": 12266
    },
    {
      "epoch": 1.9203193487789605,
      "grad_norm": 3.252751350402832,
      "learning_rate": 1.2854349951124144e-05,
      "loss": 1.1408,
      "step": 12267
    },
    {
      "epoch": 1.9204758922980587,
      "grad_norm": 2.705955743789673,
      "learning_rate": 1.2829912023460409e-05,
      "loss": 1.0692,
      "step": 12268
    },
    {
      "epoch": 1.9206324358171571,
      "grad_norm": 4.607133388519287,
      "learning_rate": 1.2805474095796675e-05,
      "loss": 1.0189,
      "step": 12269
    },
    {
      "epoch": 1.9207889793362556,
      "grad_norm": 2.5843207836151123,
      "learning_rate": 1.2781036168132942e-05,
      "loss": 1.3724,
      "step": 12270
    },
    {
      "epoch": 1.9209455228553538,
      "grad_norm": 4.102626323699951,
      "learning_rate": 1.2756598240469207e-05,
      "loss": 0.8495,
      "step": 12271
    },
    {
      "epoch": 1.921102066374452,
      "grad_norm": 3.620480537414551,
      "learning_rate": 1.2732160312805473e-05,
      "loss": 1.047,
      "step": 12272
    },
    {
      "epoch": 1.9212586098935505,
      "grad_norm": 7.478079795837402,
      "learning_rate": 1.270772238514174e-05,
      "loss": 0.7859,
      "step": 12273
    },
    {
      "epoch": 1.9214151534126487,
      "grad_norm": 25.899625778198242,
      "learning_rate": 1.2683284457478004e-05,
      "loss": 0.8097,
      "step": 12274
    },
    {
      "epoch": 1.9215716969317471,
      "grad_norm": 2.6998584270477295,
      "learning_rate": 1.2658846529814271e-05,
      "loss": 1.3919,
      "step": 12275
    },
    {
      "epoch": 1.9217282404508453,
      "grad_norm": 5.945512771606445,
      "learning_rate": 1.2634408602150537e-05,
      "loss": 1.2487,
      "step": 12276
    },
    {
      "epoch": 1.9218847839699436,
      "grad_norm": 6.3635029792785645,
      "learning_rate": 1.2609970674486802e-05,
      "loss": 0.9452,
      "step": 12277
    },
    {
      "epoch": 1.922041327489042,
      "grad_norm": 5.040246963500977,
      "learning_rate": 1.2585532746823069e-05,
      "loss": 1.1909,
      "step": 12278
    },
    {
      "epoch": 1.9221978710081402,
      "grad_norm": 4.729580402374268,
      "learning_rate": 1.2561094819159335e-05,
      "loss": 1.3885,
      "step": 12279
    },
    {
      "epoch": 1.9223544145272387,
      "grad_norm": 4.690464973449707,
      "learning_rate": 1.25366568914956e-05,
      "loss": 1.4546,
      "step": 12280
    },
    {
      "epoch": 1.9225109580463369,
      "grad_norm": 4.445624828338623,
      "learning_rate": 1.2512218963831867e-05,
      "loss": 1.3911,
      "step": 12281
    },
    {
      "epoch": 1.922667501565435,
      "grad_norm": 2.9584579467773438,
      "learning_rate": 1.2487781036168133e-05,
      "loss": 0.9071,
      "step": 12282
    },
    {
      "epoch": 1.9228240450845335,
      "grad_norm": 4.518764495849609,
      "learning_rate": 1.2463343108504398e-05,
      "loss": 0.8461,
      "step": 12283
    },
    {
      "epoch": 1.922980588603632,
      "grad_norm": 3.6593761444091797,
      "learning_rate": 1.2438905180840664e-05,
      "loss": 0.6349,
      "step": 12284
    },
    {
      "epoch": 1.9231371321227302,
      "grad_norm": 3.229717254638672,
      "learning_rate": 1.241446725317693e-05,
      "loss": 0.6635,
      "step": 12285
    },
    {
      "epoch": 1.9232936756418284,
      "grad_norm": 1.8013453483581543,
      "learning_rate": 1.2390029325513196e-05,
      "loss": 0.4047,
      "step": 12286
    },
    {
      "epoch": 1.9234502191609266,
      "grad_norm": 3.6808831691741943,
      "learning_rate": 1.2365591397849462e-05,
      "loss": 0.5197,
      "step": 12287
    },
    {
      "epoch": 1.923606762680025,
      "grad_norm": 2.7701351642608643,
      "learning_rate": 1.2341153470185729e-05,
      "loss": 0.539,
      "step": 12288
    },
    {
      "epoch": 1.9237633061991235,
      "grad_norm": 1.3340402841567993,
      "learning_rate": 1.2316715542521992e-05,
      "loss": 1.382,
      "step": 12289
    },
    {
      "epoch": 1.9239198497182217,
      "grad_norm": 1.5330002307891846,
      "learning_rate": 1.229227761485826e-05,
      "loss": 1.4128,
      "step": 12290
    },
    {
      "epoch": 1.92407639323732,
      "grad_norm": 0.8415979743003845,
      "learning_rate": 1.2267839687194526e-05,
      "loss": 1.3799,
      "step": 12291
    },
    {
      "epoch": 1.9242329367564182,
      "grad_norm": 0.8485933542251587,
      "learning_rate": 1.224340175953079e-05,
      "loss": 1.4151,
      "step": 12292
    },
    {
      "epoch": 1.9243894802755166,
      "grad_norm": 1.0369019508361816,
      "learning_rate": 1.2218963831867058e-05,
      "loss": 1.4682,
      "step": 12293
    },
    {
      "epoch": 1.924546023794615,
      "grad_norm": 1.0517786741256714,
      "learning_rate": 1.2194525904203324e-05,
      "loss": 1.4581,
      "step": 12294
    },
    {
      "epoch": 1.9247025673137133,
      "grad_norm": 1.096622347831726,
      "learning_rate": 1.2170087976539587e-05,
      "loss": 1.3303,
      "step": 12295
    },
    {
      "epoch": 1.9248591108328115,
      "grad_norm": 3.3548057079315186,
      "learning_rate": 1.2145650048875854e-05,
      "loss": 1.2956,
      "step": 12296
    },
    {
      "epoch": 1.9250156543519097,
      "grad_norm": 1.0958143472671509,
      "learning_rate": 1.2121212121212122e-05,
      "loss": 1.3452,
      "step": 12297
    },
    {
      "epoch": 1.9251721978710081,
      "grad_norm": 2.7665908336639404,
      "learning_rate": 1.2096774193548385e-05,
      "loss": 1.322,
      "step": 12298
    },
    {
      "epoch": 1.9253287413901066,
      "grad_norm": 1.590676188468933,
      "learning_rate": 1.2072336265884652e-05,
      "loss": 1.2135,
      "step": 12299
    },
    {
      "epoch": 1.9254852849092048,
      "grad_norm": 4.261856555938721,
      "learning_rate": 1.2047898338220916e-05,
      "loss": 1.506,
      "step": 12300
    },
    {
      "epoch": 1.925641828428303,
      "grad_norm": 5.3091230392456055,
      "learning_rate": 1.2023460410557183e-05,
      "loss": 1.3583,
      "step": 12301
    },
    {
      "epoch": 1.9257983719474012,
      "grad_norm": 2.1772117614746094,
      "learning_rate": 1.199902248289345e-05,
      "loss": 1.3899,
      "step": 12302
    },
    {
      "epoch": 1.9259549154664997,
      "grad_norm": 1.1379402875900269,
      "learning_rate": 1.1974584555229714e-05,
      "loss": 1.2902,
      "step": 12303
    },
    {
      "epoch": 1.926111458985598,
      "grad_norm": 1.190233826637268,
      "learning_rate": 1.195014662756598e-05,
      "loss": 1.4014,
      "step": 12304
    },
    {
      "epoch": 1.9262680025046963,
      "grad_norm": 1.8833681344985962,
      "learning_rate": 1.1925708699902247e-05,
      "loss": 1.2716,
      "step": 12305
    },
    {
      "epoch": 1.9264245460237945,
      "grad_norm": 2.2918057441711426,
      "learning_rate": 1.1901270772238512e-05,
      "loss": 1.2386,
      "step": 12306
    },
    {
      "epoch": 1.926581089542893,
      "grad_norm": 2.4016504287719727,
      "learning_rate": 1.1876832844574778e-05,
      "loss": 1.4535,
      "step": 12307
    },
    {
      "epoch": 1.9267376330619912,
      "grad_norm": 2.2692906856536865,
      "learning_rate": 1.1852394916911045e-05,
      "loss": 1.2666,
      "step": 12308
    },
    {
      "epoch": 1.9268941765810896,
      "grad_norm": 5.164487838745117,
      "learning_rate": 1.182795698924731e-05,
      "loss": 1.1711,
      "step": 12309
    },
    {
      "epoch": 1.9270507201001879,
      "grad_norm": 3.0277092456817627,
      "learning_rate": 1.1803519061583576e-05,
      "loss": 1.2333,
      "step": 12310
    },
    {
      "epoch": 1.927207263619286,
      "grad_norm": 2.941403865814209,
      "learning_rate": 1.1779081133919843e-05,
      "loss": 1.3658,
      "step": 12311
    },
    {
      "epoch": 1.9273638071383845,
      "grad_norm": 4.588296413421631,
      "learning_rate": 1.1754643206256108e-05,
      "loss": 1.2713,
      "step": 12312
    },
    {
      "epoch": 1.9275203506574827,
      "grad_norm": 1.65639328956604,
      "learning_rate": 1.1730205278592374e-05,
      "loss": 1.054,
      "step": 12313
    },
    {
      "epoch": 1.9276768941765812,
      "grad_norm": 2.7353925704956055,
      "learning_rate": 1.170576735092864e-05,
      "loss": 1.3822,
      "step": 12314
    },
    {
      "epoch": 1.9278334376956794,
      "grad_norm": 4.178305149078369,
      "learning_rate": 1.1681329423264905e-05,
      "loss": 1.4245,
      "step": 12315
    },
    {
      "epoch": 1.9279899812147776,
      "grad_norm": 4.869446754455566,
      "learning_rate": 1.1656891495601172e-05,
      "loss": 1.1584,
      "step": 12316
    },
    {
      "epoch": 1.928146524733876,
      "grad_norm": 2.118698835372925,
      "learning_rate": 1.1632453567937438e-05,
      "loss": 1.1799,
      "step": 12317
    },
    {
      "epoch": 1.9283030682529745,
      "grad_norm": 7.572887897491455,
      "learning_rate": 1.1608015640273703e-05,
      "loss": 1.1696,
      "step": 12318
    },
    {
      "epoch": 1.9284596117720727,
      "grad_norm": 2.0993082523345947,
      "learning_rate": 1.158357771260997e-05,
      "loss": 1.0388,
      "step": 12319
    },
    {
      "epoch": 1.928616155291171,
      "grad_norm": 6.564274787902832,
      "learning_rate": 1.1559139784946236e-05,
      "loss": 1.1272,
      "step": 12320
    },
    {
      "epoch": 1.9287726988102691,
      "grad_norm": 4.388607978820801,
      "learning_rate": 1.1534701857282501e-05,
      "loss": 1.0672,
      "step": 12321
    },
    {
      "epoch": 1.9289292423293676,
      "grad_norm": 3.7181320190429688,
      "learning_rate": 1.1510263929618767e-05,
      "loss": 1.1001,
      "step": 12322
    },
    {
      "epoch": 1.929085785848466,
      "grad_norm": 2.3277645111083984,
      "learning_rate": 1.1485826001955034e-05,
      "loss": 1.2546,
      "step": 12323
    },
    {
      "epoch": 1.9292423293675642,
      "grad_norm": 5.6356282234191895,
      "learning_rate": 1.1461388074291299e-05,
      "loss": 1.56,
      "step": 12324
    },
    {
      "epoch": 1.9293988728866625,
      "grad_norm": 2.452418804168701,
      "learning_rate": 1.1436950146627565e-05,
      "loss": 1.4428,
      "step": 12325
    },
    {
      "epoch": 1.9295554164057607,
      "grad_norm": 8.939882278442383,
      "learning_rate": 1.1412512218963832e-05,
      "loss": 1.2936,
      "step": 12326
    },
    {
      "epoch": 1.929711959924859,
      "grad_norm": 2.89434552192688,
      "learning_rate": 1.1388074291300096e-05,
      "loss": 1.0515,
      "step": 12327
    },
    {
      "epoch": 1.9298685034439576,
      "grad_norm": 5.483556270599365,
      "learning_rate": 1.1363636363636363e-05,
      "loss": 1.2122,
      "step": 12328
    },
    {
      "epoch": 1.9300250469630558,
      "grad_norm": 3.8645455837249756,
      "learning_rate": 1.133919843597263e-05,
      "loss": 0.9129,
      "step": 12329
    },
    {
      "epoch": 1.930181590482154,
      "grad_norm": 3.9111804962158203,
      "learning_rate": 1.1314760508308894e-05,
      "loss": 1.2189,
      "step": 12330
    },
    {
      "epoch": 1.9303381340012522,
      "grad_norm": 3.18015456199646,
      "learning_rate": 1.129032258064516e-05,
      "loss": 0.7437,
      "step": 12331
    },
    {
      "epoch": 1.9304946775203506,
      "grad_norm": 8.127317428588867,
      "learning_rate": 1.1265884652981426e-05,
      "loss": 1.3882,
      "step": 12332
    },
    {
      "epoch": 1.930651221039449,
      "grad_norm": 2.8855273723602295,
      "learning_rate": 1.1241446725317692e-05,
      "loss": 1.6402,
      "step": 12333
    },
    {
      "epoch": 1.9308077645585473,
      "grad_norm": 6.404877662658691,
      "learning_rate": 1.1217008797653959e-05,
      "loss": 0.8951,
      "step": 12334
    },
    {
      "epoch": 1.9309643080776455,
      "grad_norm": 12.837389945983887,
      "learning_rate": 1.1192570869990223e-05,
      "loss": 0.9152,
      "step": 12335
    },
    {
      "epoch": 1.9311208515967437,
      "grad_norm": 1.6619367599487305,
      "learning_rate": 1.116813294232649e-05,
      "loss": 0.7218,
      "step": 12336
    },
    {
      "epoch": 1.9312773951158422,
      "grad_norm": 1.492072343826294,
      "learning_rate": 1.1143695014662756e-05,
      "loss": 0.4666,
      "step": 12337
    },
    {
      "epoch": 1.9314339386349406,
      "grad_norm": 2.7796103954315186,
      "learning_rate": 1.1119257086999021e-05,
      "loss": 1.1153,
      "step": 12338
    },
    {
      "epoch": 1.9315904821540388,
      "grad_norm": 6.7155938148498535,
      "learning_rate": 1.1094819159335288e-05,
      "loss": 1.9017,
      "step": 12339
    },
    {
      "epoch": 1.931747025673137,
      "grad_norm": 0.7669269442558289,
      "learning_rate": 1.1070381231671554e-05,
      "loss": 1.356,
      "step": 12340
    },
    {
      "epoch": 1.9319035691922355,
      "grad_norm": 0.7935886383056641,
      "learning_rate": 1.1045943304007819e-05,
      "loss": 1.3973,
      "step": 12341
    },
    {
      "epoch": 1.9320601127113337,
      "grad_norm": 1.2134813070297241,
      "learning_rate": 1.1021505376344085e-05,
      "loss": 1.2953,
      "step": 12342
    },
    {
      "epoch": 1.9322166562304322,
      "grad_norm": 0.8597235083580017,
      "learning_rate": 1.0997067448680352e-05,
      "loss": 1.3599,
      "step": 12343
    },
    {
      "epoch": 1.9323731997495304,
      "grad_norm": 0.8375195860862732,
      "learning_rate": 1.0972629521016617e-05,
      "loss": 1.2758,
      "step": 12344
    },
    {
      "epoch": 1.9325297432686286,
      "grad_norm": 1.4954638481140137,
      "learning_rate": 1.0948191593352883e-05,
      "loss": 1.3688,
      "step": 12345
    },
    {
      "epoch": 1.932686286787727,
      "grad_norm": 1.4607149362564087,
      "learning_rate": 1.092375366568915e-05,
      "loss": 1.3371,
      "step": 12346
    },
    {
      "epoch": 1.9328428303068252,
      "grad_norm": 1.151136040687561,
      "learning_rate": 1.0899315738025414e-05,
      "loss": 1.34,
      "step": 12347
    },
    {
      "epoch": 1.9329993738259237,
      "grad_norm": 1.224950909614563,
      "learning_rate": 1.0874877810361681e-05,
      "loss": 1.3688,
      "step": 12348
    },
    {
      "epoch": 1.933155917345022,
      "grad_norm": 6.187594890594482,
      "learning_rate": 1.0850439882697947e-05,
      "loss": 1.2989,
      "step": 12349
    },
    {
      "epoch": 1.9333124608641201,
      "grad_norm": 1.705400824546814,
      "learning_rate": 1.0826001955034212e-05,
      "loss": 1.6841,
      "step": 12350
    },
    {
      "epoch": 1.9334690043832186,
      "grad_norm": 1.4183433055877686,
      "learning_rate": 1.0801564027370479e-05,
      "loss": 1.2696,
      "step": 12351
    },
    {
      "epoch": 1.933625547902317,
      "grad_norm": 2.378031015396118,
      "learning_rate": 1.0777126099706745e-05,
      "loss": 1.5075,
      "step": 12352
    },
    {
      "epoch": 1.9337820914214152,
      "grad_norm": 1.5742651224136353,
      "learning_rate": 1.075268817204301e-05,
      "loss": 1.2745,
      "step": 12353
    },
    {
      "epoch": 1.9339386349405134,
      "grad_norm": 1.3239703178405762,
      "learning_rate": 1.0728250244379277e-05,
      "loss": 1.3535,
      "step": 12354
    },
    {
      "epoch": 1.9340951784596117,
      "grad_norm": 3.5635673999786377,
      "learning_rate": 1.0703812316715543e-05,
      "loss": 1.6419,
      "step": 12355
    },
    {
      "epoch": 1.93425172197871,
      "grad_norm": 2.4734835624694824,
      "learning_rate": 1.0679374389051806e-05,
      "loss": 1.3285,
      "step": 12356
    },
    {
      "epoch": 1.9344082654978085,
      "grad_norm": 2.6944169998168945,
      "learning_rate": 1.0654936461388074e-05,
      "loss": 1.4164,
      "step": 12357
    },
    {
      "epoch": 1.9345648090169068,
      "grad_norm": 1.283929467201233,
      "learning_rate": 1.063049853372434e-05,
      "loss": 1.3136,
      "step": 12358
    },
    {
      "epoch": 1.934721352536005,
      "grad_norm": 1.476118564605713,
      "learning_rate": 1.0606060606060604e-05,
      "loss": 1.1611,
      "step": 12359
    },
    {
      "epoch": 1.9348778960551032,
      "grad_norm": 2.960695505142212,
      "learning_rate": 1.0581622678396872e-05,
      "loss": 1.3293,
      "step": 12360
    },
    {
      "epoch": 1.9350344395742016,
      "grad_norm": 4.04657506942749,
      "learning_rate": 1.0557184750733139e-05,
      "loss": 1.3852,
      "step": 12361
    },
    {
      "epoch": 1.9351909830933,
      "grad_norm": 1.8370991945266724,
      "learning_rate": 1.0532746823069402e-05,
      "loss": 1.2408,
      "step": 12362
    },
    {
      "epoch": 1.9353475266123983,
      "grad_norm": 3.660257577896118,
      "learning_rate": 1.0508308895405668e-05,
      "loss": 1.395,
      "step": 12363
    },
    {
      "epoch": 1.9355040701314965,
      "grad_norm": 4.005573749542236,
      "learning_rate": 1.0483870967741933e-05,
      "loss": 1.2782,
      "step": 12364
    },
    {
      "epoch": 1.9356606136505947,
      "grad_norm": 8.64341926574707,
      "learning_rate": 1.04594330400782e-05,
      "loss": 1.4618,
      "step": 12365
    },
    {
      "epoch": 1.9358171571696932,
      "grad_norm": 8.12671947479248,
      "learning_rate": 1.0434995112414466e-05,
      "loss": 1.9106,
      "step": 12366
    },
    {
      "epoch": 1.9359737006887916,
      "grad_norm": 2.215592384338379,
      "learning_rate": 1.041055718475073e-05,
      "loss": 1.3265,
      "step": 12367
    },
    {
      "epoch": 1.9361302442078898,
      "grad_norm": 4.294419288635254,
      "learning_rate": 1.0386119257086997e-05,
      "loss": 1.1993,
      "step": 12368
    },
    {
      "epoch": 1.936286787726988,
      "grad_norm": 2.263684034347534,
      "learning_rate": 1.0361681329423264e-05,
      "loss": 1.1445,
      "step": 12369
    },
    {
      "epoch": 1.9364433312460863,
      "grad_norm": 3.5883326530456543,
      "learning_rate": 1.0337243401759529e-05,
      "loss": 1.4243,
      "step": 12370
    },
    {
      "epoch": 1.9365998747651847,
      "grad_norm": 1.9435137510299683,
      "learning_rate": 1.0312805474095795e-05,
      "loss": 0.9561,
      "step": 12371
    },
    {
      "epoch": 1.9367564182842831,
      "grad_norm": 3.0882208347320557,
      "learning_rate": 1.0288367546432062e-05,
      "loss": 0.8797,
      "step": 12372
    },
    {
      "epoch": 1.9369129618033814,
      "grad_norm": 7.617794036865234,
      "learning_rate": 1.0263929618768326e-05,
      "loss": 1.2746,
      "step": 12373
    },
    {
      "epoch": 1.9370695053224796,
      "grad_norm": 4.225636959075928,
      "learning_rate": 1.0239491691104593e-05,
      "loss": 1.4779,
      "step": 12374
    },
    {
      "epoch": 1.937226048841578,
      "grad_norm": 15.488640785217285,
      "learning_rate": 1.021505376344086e-05,
      "loss": 1.4581,
      "step": 12375
    },
    {
      "epoch": 1.9373825923606762,
      "grad_norm": 5.777149200439453,
      "learning_rate": 1.0190615835777124e-05,
      "loss": 1.4015,
      "step": 12376
    },
    {
      "epoch": 1.9375391358797747,
      "grad_norm": 12.551581382751465,
      "learning_rate": 1.016617790811339e-05,
      "loss": 1.2572,
      "step": 12377
    },
    {
      "epoch": 1.9376956793988729,
      "grad_norm": 4.459105491638184,
      "learning_rate": 1.0141739980449657e-05,
      "loss": 1.0849,
      "step": 12378
    },
    {
      "epoch": 1.937852222917971,
      "grad_norm": 3.434291362762451,
      "learning_rate": 1.0117302052785922e-05,
      "loss": 1.3216,
      "step": 12379
    },
    {
      "epoch": 1.9380087664370695,
      "grad_norm": 2.0103023052215576,
      "learning_rate": 1.0092864125122188e-05,
      "loss": 0.7784,
      "step": 12380
    },
    {
      "epoch": 1.938165309956168,
      "grad_norm": 2.6893234252929688,
      "learning_rate": 1.0068426197458455e-05,
      "loss": 0.7125,
      "step": 12381
    },
    {
      "epoch": 1.9383218534752662,
      "grad_norm": 5.581327438354492,
      "learning_rate": 1.004398826979472e-05,
      "loss": 0.998,
      "step": 12382
    },
    {
      "epoch": 1.9384783969943644,
      "grad_norm": 3.5496435165405273,
      "learning_rate": 1.0019550342130986e-05,
      "loss": 1.2393,
      "step": 12383
    },
    {
      "epoch": 1.9386349405134626,
      "grad_norm": 2.4860739707946777,
      "learning_rate": 9.995112414467253e-06,
      "loss": 1.3163,
      "step": 12384
    },
    {
      "epoch": 1.938791484032561,
      "grad_norm": 1.844963550567627,
      "learning_rate": 9.970674486803518e-06,
      "loss": 0.7082,
      "step": 12385
    },
    {
      "epoch": 1.9389480275516595,
      "grad_norm": 2.7974820137023926,
      "learning_rate": 9.946236559139784e-06,
      "loss": 0.6463,
      "step": 12386
    },
    {
      "epoch": 1.9391045710707577,
      "grad_norm": 1.4978660345077515,
      "learning_rate": 9.92179863147605e-06,
      "loss": 0.3869,
      "step": 12387
    },
    {
      "epoch": 1.939261114589856,
      "grad_norm": 2.3386178016662598,
      "learning_rate": 9.897360703812315e-06,
      "loss": 1.1001,
      "step": 12388
    },
    {
      "epoch": 1.9394176581089542,
      "grad_norm": 0.8785395622253418,
      "learning_rate": 9.872922776148582e-06,
      "loss": 1.5033,
      "step": 12389
    },
    {
      "epoch": 1.9395742016280526,
      "grad_norm": 0.8468477725982666,
      "learning_rate": 9.848484848484848e-06,
      "loss": 1.4967,
      "step": 12390
    },
    {
      "epoch": 1.939730745147151,
      "grad_norm": 0.7762795686721802,
      "learning_rate": 9.824046920821113e-06,
      "loss": 1.4881,
      "step": 12391
    },
    {
      "epoch": 1.9398872886662493,
      "grad_norm": 2.385578155517578,
      "learning_rate": 9.79960899315738e-06,
      "loss": 1.3553,
      "step": 12392
    },
    {
      "epoch": 1.9400438321853475,
      "grad_norm": 1.221888542175293,
      "learning_rate": 9.775171065493646e-06,
      "loss": 1.4372,
      "step": 12393
    },
    {
      "epoch": 1.9402003757044457,
      "grad_norm": 3.5840225219726562,
      "learning_rate": 9.750733137829911e-06,
      "loss": 1.4205,
      "step": 12394
    },
    {
      "epoch": 1.9403569192235441,
      "grad_norm": 0.9820275902748108,
      "learning_rate": 9.726295210166177e-06,
      "loss": 1.3966,
      "step": 12395
    },
    {
      "epoch": 1.9405134627426426,
      "grad_norm": 1.1478222608566284,
      "learning_rate": 9.701857282502442e-06,
      "loss": 1.4722,
      "step": 12396
    },
    {
      "epoch": 1.9406700062617408,
      "grad_norm": 2.0976521968841553,
      "learning_rate": 9.677419354838709e-06,
      "loss": 1.4673,
      "step": 12397
    },
    {
      "epoch": 1.940826549780839,
      "grad_norm": 1.5438917875289917,
      "learning_rate": 9.652981427174975e-06,
      "loss": 1.5253,
      "step": 12398
    },
    {
      "epoch": 1.9409830932999372,
      "grad_norm": 1.2291433811187744,
      "learning_rate": 9.62854349951124e-06,
      "loss": 1.4075,
      "step": 12399
    },
    {
      "epoch": 1.9411396368190357,
      "grad_norm": 1.6943989992141724,
      "learning_rate": 9.604105571847507e-06,
      "loss": 1.3334,
      "step": 12400
    },
    {
      "epoch": 1.9412961803381341,
      "grad_norm": 1.9269627332687378,
      "learning_rate": 9.579667644183773e-06,
      "loss": 1.4683,
      "step": 12401
    },
    {
      "epoch": 1.9414527238572323,
      "grad_norm": 1.9164787530899048,
      "learning_rate": 9.555229716520038e-06,
      "loss": 1.5016,
      "step": 12402
    },
    {
      "epoch": 1.9416092673763305,
      "grad_norm": 1.3316177129745483,
      "learning_rate": 9.530791788856304e-06,
      "loss": 1.2375,
      "step": 12403
    },
    {
      "epoch": 1.9417658108954288,
      "grad_norm": 1.476271390914917,
      "learning_rate": 9.50635386119257e-06,
      "loss": 1.2165,
      "step": 12404
    },
    {
      "epoch": 1.9419223544145272,
      "grad_norm": 1.4863686561584473,
      "learning_rate": 9.481915933528836e-06,
      "loss": 1.2883,
      "step": 12405
    },
    {
      "epoch": 1.9420788979336256,
      "grad_norm": 1.1693488359451294,
      "learning_rate": 9.457478005865102e-06,
      "loss": 1.4016,
      "step": 12406
    },
    {
      "epoch": 1.9422354414527239,
      "grad_norm": 2.0392889976501465,
      "learning_rate": 9.433040078201369e-06,
      "loss": 1.2879,
      "step": 12407
    },
    {
      "epoch": 1.942391984971822,
      "grad_norm": 2.9882824420928955,
      "learning_rate": 9.408602150537633e-06,
      "loss": 1.3189,
      "step": 12408
    },
    {
      "epoch": 1.9425485284909205,
      "grad_norm": 1.9439655542373657,
      "learning_rate": 9.3841642228739e-06,
      "loss": 1.2713,
      "step": 12409
    },
    {
      "epoch": 1.9427050720100187,
      "grad_norm": 2.5149528980255127,
      "learning_rate": 9.359726295210165e-06,
      "loss": 1.4599,
      "step": 12410
    },
    {
      "epoch": 1.9428616155291172,
      "grad_norm": 1.612697720527649,
      "learning_rate": 9.335288367546431e-06,
      "loss": 1.2472,
      "step": 12411
    },
    {
      "epoch": 1.9430181590482154,
      "grad_norm": 4.500641345977783,
      "learning_rate": 9.310850439882698e-06,
      "loss": 1.208,
      "step": 12412
    },
    {
      "epoch": 1.9431747025673136,
      "grad_norm": 1.9317268133163452,
      "learning_rate": 9.286412512218962e-06,
      "loss": 1.5878,
      "step": 12413
    },
    {
      "epoch": 1.943331246086412,
      "grad_norm": 6.950214862823486,
      "learning_rate": 9.261974584555229e-06,
      "loss": 1.4057,
      "step": 12414
    },
    {
      "epoch": 1.9434877896055105,
      "grad_norm": 4.08366584777832,
      "learning_rate": 9.237536656891495e-06,
      "loss": 1.3945,
      "step": 12415
    },
    {
      "epoch": 1.9436443331246087,
      "grad_norm": 1.9059251546859741,
      "learning_rate": 9.21309872922776e-06,
      "loss": 1.0823,
      "step": 12416
    },
    {
      "epoch": 1.943800876643707,
      "grad_norm": 4.59588098526001,
      "learning_rate": 9.188660801564027e-06,
      "loss": 1.4755,
      "step": 12417
    },
    {
      "epoch": 1.9439574201628051,
      "grad_norm": 3.578123092651367,
      "learning_rate": 9.164222873900293e-06,
      "loss": 1.1854,
      "step": 12418
    },
    {
      "epoch": 1.9441139636819036,
      "grad_norm": 1.8698164224624634,
      "learning_rate": 9.139784946236558e-06,
      "loss": 1.2789,
      "step": 12419
    },
    {
      "epoch": 1.944270507201002,
      "grad_norm": 2.4767820835113525,
      "learning_rate": 9.115347018572825e-06,
      "loss": 1.3508,
      "step": 12420
    },
    {
      "epoch": 1.9444270507201002,
      "grad_norm": 1.976932406425476,
      "learning_rate": 9.090909090909091e-06,
      "loss": 0.9728,
      "step": 12421
    },
    {
      "epoch": 1.9445835942391985,
      "grad_norm": 2.43166184425354,
      "learning_rate": 9.066471163245356e-06,
      "loss": 1.157,
      "step": 12422
    },
    {
      "epoch": 1.9447401377582967,
      "grad_norm": 1.8392789363861084,
      "learning_rate": 9.042033235581622e-06,
      "loss": 0.9677,
      "step": 12423
    },
    {
      "epoch": 1.9448966812773951,
      "grad_norm": 6.830517768859863,
      "learning_rate": 9.017595307917889e-06,
      "loss": 1.2045,
      "step": 12424
    },
    {
      "epoch": 1.9450532247964936,
      "grad_norm": 4.780092239379883,
      "learning_rate": 8.993157380254154e-06,
      "loss": 1.522,
      "step": 12425
    },
    {
      "epoch": 1.9452097683155918,
      "grad_norm": 6.710925102233887,
      "learning_rate": 8.968719452590418e-06,
      "loss": 1.4294,
      "step": 12426
    },
    {
      "epoch": 1.94536631183469,
      "grad_norm": 4.024580478668213,
      "learning_rate": 8.944281524926687e-06,
      "loss": 1.1853,
      "step": 12427
    },
    {
      "epoch": 1.9455228553537882,
      "grad_norm": 5.138179302215576,
      "learning_rate": 8.919843597262951e-06,
      "loss": 1.195,
      "step": 12428
    },
    {
      "epoch": 1.9456793988728867,
      "grad_norm": 4.0508270263671875,
      "learning_rate": 8.895405669599216e-06,
      "loss": 1.5426,
      "step": 12429
    },
    {
      "epoch": 1.945835942391985,
      "grad_norm": 5.0509538650512695,
      "learning_rate": 8.870967741935483e-06,
      "loss": 1.3934,
      "step": 12430
    },
    {
      "epoch": 1.9459924859110833,
      "grad_norm": 7.239458084106445,
      "learning_rate": 8.84652981427175e-06,
      "loss": 1.361,
      "step": 12431
    },
    {
      "epoch": 1.9461490294301815,
      "grad_norm": 4.996304988861084,
      "learning_rate": 8.822091886608014e-06,
      "loss": 1.1353,
      "step": 12432
    },
    {
      "epoch": 1.9463055729492797,
      "grad_norm": 1.4053064584732056,
      "learning_rate": 8.79765395894428e-06,
      "loss": 1.052,
      "step": 12433
    },
    {
      "epoch": 1.9464621164683782,
      "grad_norm": 4.434309959411621,
      "learning_rate": 8.773216031280547e-06,
      "loss": 1.125,
      "step": 12434
    },
    {
      "epoch": 1.9466186599874766,
      "grad_norm": 2.092318058013916,
      "learning_rate": 8.748778103616812e-06,
      "loss": 0.8836,
      "step": 12435
    },
    {
      "epoch": 1.9467752035065748,
      "grad_norm": 2.4105422496795654,
      "learning_rate": 8.724340175953078e-06,
      "loss": 0.9855,
      "step": 12436
    },
    {
      "epoch": 1.946931747025673,
      "grad_norm": 3.4820504188537598,
      "learning_rate": 8.699902248289345e-06,
      "loss": 1.01,
      "step": 12437
    },
    {
      "epoch": 1.9470882905447713,
      "grad_norm": 2.5442054271698,
      "learning_rate": 8.67546432062561e-06,
      "loss": 0.9834,
      "step": 12438
    },
    {
      "epoch": 1.9472448340638697,
      "grad_norm": 0.9357713460922241,
      "learning_rate": 8.651026392961876e-06,
      "loss": 1.4554,
      "step": 12439
    },
    {
      "epoch": 1.9474013775829682,
      "grad_norm": 0.6884608268737793,
      "learning_rate": 8.626588465298143e-06,
      "loss": 1.4628,
      "step": 12440
    },
    {
      "epoch": 1.9475579211020664,
      "grad_norm": 1.8610217571258545,
      "learning_rate": 8.602150537634407e-06,
      "loss": 1.3942,
      "step": 12441
    },
    {
      "epoch": 1.9477144646211646,
      "grad_norm": 1.9256157875061035,
      "learning_rate": 8.577712609970674e-06,
      "loss": 1.4434,
      "step": 12442
    },
    {
      "epoch": 1.947871008140263,
      "grad_norm": 1.0926450490951538,
      "learning_rate": 8.55327468230694e-06,
      "loss": 1.4785,
      "step": 12443
    },
    {
      "epoch": 1.9480275516593613,
      "grad_norm": 1.835080623626709,
      "learning_rate": 8.528836754643205e-06,
      "loss": 1.4919,
      "step": 12444
    },
    {
      "epoch": 1.9481840951784597,
      "grad_norm": 0.9636733531951904,
      "learning_rate": 8.504398826979472e-06,
      "loss": 1.485,
      "step": 12445
    },
    {
      "epoch": 1.948340638697558,
      "grad_norm": 1.1740062236785889,
      "learning_rate": 8.479960899315738e-06,
      "loss": 1.5098,
      "step": 12446
    },
    {
      "epoch": 1.9484971822166561,
      "grad_norm": 1.6411094665527344,
      "learning_rate": 8.455522971652003e-06,
      "loss": 1.4772,
      "step": 12447
    },
    {
      "epoch": 1.9486537257357546,
      "grad_norm": 1.150183916091919,
      "learning_rate": 8.43108504398827e-06,
      "loss": 1.4486,
      "step": 12448
    },
    {
      "epoch": 1.948810269254853,
      "grad_norm": 2.3827710151672363,
      "learning_rate": 8.406647116324536e-06,
      "loss": 1.4765,
      "step": 12449
    },
    {
      "epoch": 1.9489668127739512,
      "grad_norm": 1.509945034980774,
      "learning_rate": 8.3822091886608e-06,
      "loss": 1.2571,
      "step": 12450
    },
    {
      "epoch": 1.9491233562930494,
      "grad_norm": 1.6995009183883667,
      "learning_rate": 8.357771260997067e-06,
      "loss": 1.3709,
      "step": 12451
    },
    {
      "epoch": 1.9492798998121477,
      "grad_norm": 1.6770323514938354,
      "learning_rate": 8.333333333333332e-06,
      "loss": 1.4892,
      "step": 12452
    },
    {
      "epoch": 1.949436443331246,
      "grad_norm": 1.3949050903320312,
      "learning_rate": 8.308895405669599e-06,
      "loss": 1.4031,
      "step": 12453
    },
    {
      "epoch": 1.9495929868503445,
      "grad_norm": 4.16380500793457,
      "learning_rate": 8.284457478005865e-06,
      "loss": 1.3407,
      "step": 12454
    },
    {
      "epoch": 1.9497495303694428,
      "grad_norm": 1.7018452882766724,
      "learning_rate": 8.26001955034213e-06,
      "loss": 1.6301,
      "step": 12455
    },
    {
      "epoch": 1.949906073888541,
      "grad_norm": 1.3921611309051514,
      "learning_rate": 8.235581622678396e-06,
      "loss": 1.4144,
      "step": 12456
    },
    {
      "epoch": 1.9500626174076392,
      "grad_norm": 2.7200441360473633,
      "learning_rate": 8.211143695014663e-06,
      "loss": 1.4583,
      "step": 12457
    },
    {
      "epoch": 1.9502191609267376,
      "grad_norm": 1.539747714996338,
      "learning_rate": 8.186705767350928e-06,
      "loss": 1.4575,
      "step": 12458
    },
    {
      "epoch": 1.950375704445836,
      "grad_norm": 1.5239930152893066,
      "learning_rate": 8.162267839687194e-06,
      "loss": 1.266,
      "step": 12459
    },
    {
      "epoch": 1.9505322479649343,
      "grad_norm": 2.7010016441345215,
      "learning_rate": 8.137829912023459e-06,
      "loss": 1.2497,
      "step": 12460
    },
    {
      "epoch": 1.9506887914840325,
      "grad_norm": 3.007016658782959,
      "learning_rate": 8.113391984359725e-06,
      "loss": 1.537,
      "step": 12461
    },
    {
      "epoch": 1.9508453350031307,
      "grad_norm": 1.432244896888733,
      "learning_rate": 8.088954056695992e-06,
      "loss": 1.4236,
      "step": 12462
    },
    {
      "epoch": 1.9510018785222292,
      "grad_norm": 1.9789886474609375,
      "learning_rate": 8.064516129032257e-06,
      "loss": 1.5051,
      "step": 12463
    },
    {
      "epoch": 1.9511584220413276,
      "grad_norm": 1.625560998916626,
      "learning_rate": 8.040078201368523e-06,
      "loss": 1.3133,
      "step": 12464
    },
    {
      "epoch": 1.9513149655604258,
      "grad_norm": 5.823700428009033,
      "learning_rate": 8.01564027370479e-06,
      "loss": 1.3131,
      "step": 12465
    },
    {
      "epoch": 1.951471509079524,
      "grad_norm": 1.7935535907745361,
      "learning_rate": 7.991202346041054e-06,
      "loss": 1.2024,
      "step": 12466
    },
    {
      "epoch": 1.9516280525986223,
      "grad_norm": 2.3289988040924072,
      "learning_rate": 7.966764418377321e-06,
      "loss": 1.0846,
      "step": 12467
    },
    {
      "epoch": 1.9517845961177207,
      "grad_norm": 7.128284454345703,
      "learning_rate": 7.942326490713586e-06,
      "loss": 1.53,
      "step": 12468
    },
    {
      "epoch": 1.9519411396368191,
      "grad_norm": 1.9482355117797852,
      "learning_rate": 7.917888563049852e-06,
      "loss": 1.2947,
      "step": 12469
    },
    {
      "epoch": 1.9520976831559174,
      "grad_norm": 1.5053616762161255,
      "learning_rate": 7.893450635386119e-06,
      "loss": 1.2437,
      "step": 12470
    },
    {
      "epoch": 1.9522542266750156,
      "grad_norm": 8.75932502746582,
      "learning_rate": 7.869012707722384e-06,
      "loss": 1.5358,
      "step": 12471
    },
    {
      "epoch": 1.9524107701941138,
      "grad_norm": 2.771195411682129,
      "learning_rate": 7.84457478005865e-06,
      "loss": 0.8694,
      "step": 12472
    },
    {
      "epoch": 1.9525673137132122,
      "grad_norm": 5.4038262367248535,
      "learning_rate": 7.820136852394917e-06,
      "loss": 1.6965,
      "step": 12473
    },
    {
      "epoch": 1.9527238572323107,
      "grad_norm": 2.8601737022399902,
      "learning_rate": 7.795698924731181e-06,
      "loss": 1.3113,
      "step": 12474
    },
    {
      "epoch": 1.952880400751409,
      "grad_norm": 13.03695011138916,
      "learning_rate": 7.771260997067448e-06,
      "loss": 1.778,
      "step": 12475
    },
    {
      "epoch": 1.9530369442705071,
      "grad_norm": 4.585375785827637,
      "learning_rate": 7.746823069403714e-06,
      "loss": 0.9247,
      "step": 12476
    },
    {
      "epoch": 1.9531934877896056,
      "grad_norm": 2.3865814208984375,
      "learning_rate": 7.72238514173998e-06,
      "loss": 1.1538,
      "step": 12477
    },
    {
      "epoch": 1.9533500313087038,
      "grad_norm": 4.630260944366455,
      "learning_rate": 7.697947214076246e-06,
      "loss": 1.0104,
      "step": 12478
    },
    {
      "epoch": 1.9535065748278022,
      "grad_norm": 2.182964563369751,
      "learning_rate": 7.673509286412512e-06,
      "loss": 1.0723,
      "step": 12479
    },
    {
      "epoch": 1.9536631183469004,
      "grad_norm": 4.5911760330200195,
      "learning_rate": 7.649071358748777e-06,
      "loss": 0.6017,
      "step": 12480
    },
    {
      "epoch": 1.9538196618659986,
      "grad_norm": 3.150099754333496,
      "learning_rate": 7.6246334310850434e-06,
      "loss": 0.8234,
      "step": 12481
    },
    {
      "epoch": 1.953976205385097,
      "grad_norm": 1.7521438598632812,
      "learning_rate": 7.600195503421309e-06,
      "loss": 1.0587,
      "step": 12482
    },
    {
      "epoch": 1.9541327489041955,
      "grad_norm": 3.584641933441162,
      "learning_rate": 7.575757575757575e-06,
      "loss": 1.5662,
      "step": 12483
    },
    {
      "epoch": 1.9542892924232937,
      "grad_norm": 3.3541533946990967,
      "learning_rate": 7.55131964809384e-06,
      "loss": 1.109,
      "step": 12484
    },
    {
      "epoch": 1.954445835942392,
      "grad_norm": 7.451817512512207,
      "learning_rate": 7.526881720430107e-06,
      "loss": 1.2002,
      "step": 12485
    },
    {
      "epoch": 1.9546023794614902,
      "grad_norm": 6.928215503692627,
      "learning_rate": 7.5024437927663725e-06,
      "loss": 0.6299,
      "step": 12486
    },
    {
      "epoch": 1.9547589229805886,
      "grad_norm": 3.0479836463928223,
      "learning_rate": 7.478005865102638e-06,
      "loss": 1.0142,
      "step": 12487
    },
    {
      "epoch": 1.954915466499687,
      "grad_norm": 1.0150418281555176,
      "learning_rate": 7.453567937438905e-06,
      "loss": 0.4941,
      "step": 12488
    },
    {
      "epoch": 1.9550720100187853,
      "grad_norm": 1.5538381338119507,
      "learning_rate": 7.42913000977517e-06,
      "loss": 1.4971,
      "step": 12489
    },
    {
      "epoch": 1.9552285535378835,
      "grad_norm": 0.806178867816925,
      "learning_rate": 7.404692082111436e-06,
      "loss": 1.4253,
      "step": 12490
    },
    {
      "epoch": 1.9553850970569817,
      "grad_norm": 1.2289999723434448,
      "learning_rate": 7.3802541544477025e-06,
      "loss": 1.2646,
      "step": 12491
    },
    {
      "epoch": 1.9555416405760802,
      "grad_norm": 1.0183123350143433,
      "learning_rate": 7.355816226783968e-06,
      "loss": 1.4464,
      "step": 12492
    },
    {
      "epoch": 1.9556981840951786,
      "grad_norm": 0.9893926382064819,
      "learning_rate": 7.331378299120234e-06,
      "loss": 1.3698,
      "step": 12493
    },
    {
      "epoch": 1.9558547276142768,
      "grad_norm": 1.2529398202896118,
      "learning_rate": 7.3069403714565e-06,
      "loss": 1.507,
      "step": 12494
    },
    {
      "epoch": 1.956011271133375,
      "grad_norm": 0.9131374955177307,
      "learning_rate": 7.282502443792766e-06,
      "loss": 1.3113,
      "step": 12495
    },
    {
      "epoch": 1.9561678146524732,
      "grad_norm": 0.9663822650909424,
      "learning_rate": 7.2580645161290315e-06,
      "loss": 1.3692,
      "step": 12496
    },
    {
      "epoch": 1.9563243581715717,
      "grad_norm": 4.0027265548706055,
      "learning_rate": 7.233626588465298e-06,
      "loss": 1.4265,
      "step": 12497
    },
    {
      "epoch": 1.9564809016906701,
      "grad_norm": 1.6491836309432983,
      "learning_rate": 7.209188660801564e-06,
      "loss": 1.4167,
      "step": 12498
    },
    {
      "epoch": 1.9566374452097683,
      "grad_norm": 1.1959609985351562,
      "learning_rate": 7.184750733137829e-06,
      "loss": 1.5605,
      "step": 12499
    },
    {
      "epoch": 1.9567939887288666,
      "grad_norm": 2.8737237453460693,
      "learning_rate": 7.160312805474095e-06,
      "loss": 1.417,
      "step": 12500
    },
    {
      "epoch": 1.9569505322479648,
      "grad_norm": 2.0030691623687744,
      "learning_rate": 7.1358748778103615e-06,
      "loss": 1.3524,
      "step": 12501
    },
    {
      "epoch": 1.9571070757670632,
      "grad_norm": 1.234277606010437,
      "learning_rate": 7.111436950146627e-06,
      "loss": 1.3771,
      "step": 12502
    },
    {
      "epoch": 1.9572636192861617,
      "grad_norm": 3.153789520263672,
      "learning_rate": 7.086999022482893e-06,
      "loss": 1.4366,
      "step": 12503
    },
    {
      "epoch": 1.9574201628052599,
      "grad_norm": 2.164249897003174,
      "learning_rate": 7.062561094819159e-06,
      "loss": 1.5128,
      "step": 12504
    },
    {
      "epoch": 1.957576706324358,
      "grad_norm": 2.0796103477478027,
      "learning_rate": 7.038123167155425e-06,
      "loss": 1.3222,
      "step": 12505
    },
    {
      "epoch": 1.9577332498434565,
      "grad_norm": 10.709492683410645,
      "learning_rate": 7.0136852394916906e-06,
      "loss": 1.4278,
      "step": 12506
    },
    {
      "epoch": 1.9578897933625548,
      "grad_norm": 3.223146915435791,
      "learning_rate": 6.989247311827957e-06,
      "loss": 1.3899,
      "step": 12507
    },
    {
      "epoch": 1.9580463368816532,
      "grad_norm": 1.8081053495407104,
      "learning_rate": 6.964809384164223e-06,
      "loss": 1.3115,
      "step": 12508
    },
    {
      "epoch": 1.9582028804007514,
      "grad_norm": 2.1025900840759277,
      "learning_rate": 6.9403714565004875e-06,
      "loss": 1.1652,
      "step": 12509
    },
    {
      "epoch": 1.9583594239198496,
      "grad_norm": 2.087188720703125,
      "learning_rate": 6.915933528836755e-06,
      "loss": 1.33,
      "step": 12510
    },
    {
      "epoch": 1.958515967438948,
      "grad_norm": 1.6247758865356445,
      "learning_rate": 6.89149560117302e-06,
      "loss": 1.3163,
      "step": 12511
    },
    {
      "epoch": 1.9586725109580463,
      "grad_norm": 5.537015914916992,
      "learning_rate": 6.867057673509285e-06,
      "loss": 1.6139,
      "step": 12512
    },
    {
      "epoch": 1.9588290544771447,
      "grad_norm": 5.143603801727295,
      "learning_rate": 6.842619745845553e-06,
      "loss": 1.3413,
      "step": 12513
    },
    {
      "epoch": 1.958985597996243,
      "grad_norm": 6.505813121795654,
      "learning_rate": 6.8181818181818174e-06,
      "loss": 1.2207,
      "step": 12514
    },
    {
      "epoch": 1.9591421415153412,
      "grad_norm": 3.945194721221924,
      "learning_rate": 6.793743890518083e-06,
      "loss": 1.3264,
      "step": 12515
    },
    {
      "epoch": 1.9592986850344396,
      "grad_norm": 3.9010744094848633,
      "learning_rate": 6.769305962854349e-06,
      "loss": 1.1143,
      "step": 12516
    },
    {
      "epoch": 1.959455228553538,
      "grad_norm": 3.1134190559387207,
      "learning_rate": 6.744868035190615e-06,
      "loss": 1.3981,
      "step": 12517
    },
    {
      "epoch": 1.9596117720726363,
      "grad_norm": 1.898162841796875,
      "learning_rate": 6.720430107526881e-06,
      "loss": 0.932,
      "step": 12518
    },
    {
      "epoch": 1.9597683155917345,
      "grad_norm": 4.892297744750977,
      "learning_rate": 6.6959921798631465e-06,
      "loss": 1.4597,
      "step": 12519
    },
    {
      "epoch": 1.9599248591108327,
      "grad_norm": 2.914050340652466,
      "learning_rate": 6.671554252199413e-06,
      "loss": 1.2172,
      "step": 12520
    },
    {
      "epoch": 1.9600814026299311,
      "grad_norm": 2.2120187282562256,
      "learning_rate": 6.647116324535679e-06,
      "loss": 1.0094,
      "step": 12521
    },
    {
      "epoch": 1.9602379461490296,
      "grad_norm": 3.932375907897949,
      "learning_rate": 6.622678396871944e-06,
      "loss": 1.2513,
      "step": 12522
    },
    {
      "epoch": 1.9603944896681278,
      "grad_norm": 2.4831714630126953,
      "learning_rate": 6.598240469208211e-06,
      "loss": 1.0083,
      "step": 12523
    },
    {
      "epoch": 1.960551033187226,
      "grad_norm": 1.6757017374038696,
      "learning_rate": 6.5738025415444764e-06,
      "loss": 1.1801,
      "step": 12524
    },
    {
      "epoch": 1.9607075767063242,
      "grad_norm": 4.2157745361328125,
      "learning_rate": 6.549364613880742e-06,
      "loss": 1.4077,
      "step": 12525
    },
    {
      "epoch": 1.9608641202254227,
      "grad_norm": 7.748180866241455,
      "learning_rate": 6.524926686217009e-06,
      "loss": 1.1976,
      "step": 12526
    },
    {
      "epoch": 1.961020663744521,
      "grad_norm": 2.130333423614502,
      "learning_rate": 6.500488758553274e-06,
      "loss": 0.8903,
      "step": 12527
    },
    {
      "epoch": 1.9611772072636193,
      "grad_norm": 11.016569137573242,
      "learning_rate": 6.47605083088954e-06,
      "loss": 1.4919,
      "step": 12528
    },
    {
      "epoch": 1.9613337507827175,
      "grad_norm": 3.6152400970458984,
      "learning_rate": 6.451612903225806e-06,
      "loss": 1.0693,
      "step": 12529
    },
    {
      "epoch": 1.9614902943018158,
      "grad_norm": 5.328700542449951,
      "learning_rate": 6.427174975562072e-06,
      "loss": 0.9952,
      "step": 12530
    },
    {
      "epoch": 1.9616468378209142,
      "grad_norm": 4.276764869689941,
      "learning_rate": 6.402737047898338e-06,
      "loss": 1.0022,
      "step": 12531
    },
    {
      "epoch": 1.9618033813400126,
      "grad_norm": 5.339593887329102,
      "learning_rate": 6.378299120234603e-06,
      "loss": 1.5189,
      "step": 12532
    },
    {
      "epoch": 1.9619599248591109,
      "grad_norm": 3.1604864597320557,
      "learning_rate": 6.35386119257087e-06,
      "loss": 0.8523,
      "step": 12533
    },
    {
      "epoch": 1.962116468378209,
      "grad_norm": 2.233135461807251,
      "learning_rate": 6.3294232649071355e-06,
      "loss": 1.2364,
      "step": 12534
    },
    {
      "epoch": 1.9622730118973073,
      "grad_norm": 9.808837890625,
      "learning_rate": 6.304985337243401e-06,
      "loss": 0.8228,
      "step": 12535
    },
    {
      "epoch": 1.9624295554164057,
      "grad_norm": 4.273067474365234,
      "learning_rate": 6.280547409579668e-06,
      "loss": 0.731,
      "step": 12536
    },
    {
      "epoch": 1.9625860989355042,
      "grad_norm": 4.046677112579346,
      "learning_rate": 6.256109481915933e-06,
      "loss": 0.725,
      "step": 12537
    },
    {
      "epoch": 1.9627426424546024,
      "grad_norm": 3.113222599029541,
      "learning_rate": 6.231671554252199e-06,
      "loss": 0.5296,
      "step": 12538
    },
    {
      "epoch": 1.9628991859737006,
      "grad_norm": 1.053043007850647,
      "learning_rate": 6.207233626588465e-06,
      "loss": 1.5546,
      "step": 12539
    },
    {
      "epoch": 1.963055729492799,
      "grad_norm": 1.1030488014221191,
      "learning_rate": 6.182795698924731e-06,
      "loss": 1.3777,
      "step": 12540
    },
    {
      "epoch": 1.9632122730118973,
      "grad_norm": 0.8204811215400696,
      "learning_rate": 6.158357771260996e-06,
      "loss": 1.3897,
      "step": 12541
    },
    {
      "epoch": 1.9633688165309957,
      "grad_norm": 0.8840574622154236,
      "learning_rate": 6.133919843597263e-06,
      "loss": 1.4116,
      "step": 12542
    },
    {
      "epoch": 1.963525360050094,
      "grad_norm": 1.045058012008667,
      "learning_rate": 6.109481915933529e-06,
      "loss": 1.4974,
      "step": 12543
    },
    {
      "epoch": 1.9636819035691921,
      "grad_norm": 2.036625623703003,
      "learning_rate": 6.085043988269794e-06,
      "loss": 1.5181,
      "step": 12544
    },
    {
      "epoch": 1.9638384470882906,
      "grad_norm": 1.2316803932189941,
      "learning_rate": 6.060606060606061e-06,
      "loss": 1.4962,
      "step": 12545
    },
    {
      "epoch": 1.9639949906073888,
      "grad_norm": 1.3155983686447144,
      "learning_rate": 6.036168132942326e-06,
      "loss": 1.4232,
      "step": 12546
    },
    {
      "epoch": 1.9641515341264872,
      "grad_norm": 1.7271487712860107,
      "learning_rate": 6.011730205278591e-06,
      "loss": 1.3122,
      "step": 12547
    },
    {
      "epoch": 1.9643080776455855,
      "grad_norm": 2.852642774581909,
      "learning_rate": 5.987292277614857e-06,
      "loss": 1.4496,
      "step": 12548
    },
    {
      "epoch": 1.9644646211646837,
      "grad_norm": 1.5159505605697632,
      "learning_rate": 5.9628543499511236e-06,
      "loss": 1.4374,
      "step": 12549
    },
    {
      "epoch": 1.9646211646837821,
      "grad_norm": 0.9547608494758606,
      "learning_rate": 5.938416422287389e-06,
      "loss": 1.4034,
      "step": 12550
    },
    {
      "epoch": 1.9647777082028806,
      "grad_norm": 2.335413694381714,
      "learning_rate": 5.913978494623655e-06,
      "loss": 1.428,
      "step": 12551
    },
    {
      "epoch": 1.9649342517219788,
      "grad_norm": 1.940859317779541,
      "learning_rate": 5.889540566959921e-06,
      "loss": 1.5437,
      "step": 12552
    },
    {
      "epoch": 1.965090795241077,
      "grad_norm": 1.7635548114776611,
      "learning_rate": 5.865102639296187e-06,
      "loss": 1.5615,
      "step": 12553
    },
    {
      "epoch": 1.9652473387601752,
      "grad_norm": 1.3363462686538696,
      "learning_rate": 5.840664711632453e-06,
      "loss": 1.3193,
      "step": 12554
    },
    {
      "epoch": 1.9654038822792737,
      "grad_norm": 1.3077514171600342,
      "learning_rate": 5.816226783968719e-06,
      "loss": 1.4132,
      "step": 12555
    },
    {
      "epoch": 1.965560425798372,
      "grad_norm": 1.0712963342666626,
      "learning_rate": 5.791788856304985e-06,
      "loss": 1.3992,
      "step": 12556
    },
    {
      "epoch": 1.9657169693174703,
      "grad_norm": 1.7005242109298706,
      "learning_rate": 5.7673509286412504e-06,
      "loss": 1.4594,
      "step": 12557
    },
    {
      "epoch": 1.9658735128365685,
      "grad_norm": 1.507432460784912,
      "learning_rate": 5.742913000977517e-06,
      "loss": 1.3866,
      "step": 12558
    },
    {
      "epoch": 1.9660300563556667,
      "grad_norm": 2.3320140838623047,
      "learning_rate": 5.718475073313783e-06,
      "loss": 1.5171,
      "step": 12559
    },
    {
      "epoch": 1.9661865998747652,
      "grad_norm": 1.6303120851516724,
      "learning_rate": 5.694037145650048e-06,
      "loss": 1.3281,
      "step": 12560
    },
    {
      "epoch": 1.9663431433938636,
      "grad_norm": 9.600627899169922,
      "learning_rate": 5.669599217986315e-06,
      "loss": 1.3929,
      "step": 12561
    },
    {
      "epoch": 1.9664996869129618,
      "grad_norm": 2.142137050628662,
      "learning_rate": 5.64516129032258e-06,
      "loss": 1.2811,
      "step": 12562
    },
    {
      "epoch": 1.96665623043206,
      "grad_norm": 3.0277087688446045,
      "learning_rate": 5.620723362658846e-06,
      "loss": 1.3192,
      "step": 12563
    },
    {
      "epoch": 1.9668127739511583,
      "grad_norm": 7.777707576751709,
      "learning_rate": 5.596285434995112e-06,
      "loss": 1.5502,
      "step": 12564
    },
    {
      "epoch": 1.9669693174702567,
      "grad_norm": 2.6765975952148438,
      "learning_rate": 5.571847507331378e-06,
      "loss": 1.2499,
      "step": 12565
    },
    {
      "epoch": 1.9671258609893552,
      "grad_norm": 2.20479416847229,
      "learning_rate": 5.547409579667644e-06,
      "loss": 1.1884,
      "step": 12566
    },
    {
      "epoch": 1.9672824045084534,
      "grad_norm": 2.699751377105713,
      "learning_rate": 5.5229716520039095e-06,
      "loss": 1.1944,
      "step": 12567
    },
    {
      "epoch": 1.9674389480275516,
      "grad_norm": 10.28531265258789,
      "learning_rate": 5.498533724340176e-06,
      "loss": 1.5012,
      "step": 12568
    },
    {
      "epoch": 1.9675954915466498,
      "grad_norm": 4.221407890319824,
      "learning_rate": 5.474095796676442e-06,
      "loss": 1.1851,
      "step": 12569
    },
    {
      "epoch": 1.9677520350657483,
      "grad_norm": 4.658267498016357,
      "learning_rate": 5.449657869012707e-06,
      "loss": 1.0217,
      "step": 12570
    },
    {
      "epoch": 1.9679085785848467,
      "grad_norm": 7.977578163146973,
      "learning_rate": 5.425219941348974e-06,
      "loss": 1.545,
      "step": 12571
    },
    {
      "epoch": 1.968065122103945,
      "grad_norm": 2.949538469314575,
      "learning_rate": 5.400782013685239e-06,
      "loss": 1.2254,
      "step": 12572
    },
    {
      "epoch": 1.9682216656230431,
      "grad_norm": 5.280106067657471,
      "learning_rate": 5.376344086021505e-06,
      "loss": 1.1792,
      "step": 12573
    },
    {
      "epoch": 1.9683782091421416,
      "grad_norm": 6.7711710929870605,
      "learning_rate": 5.3519061583577715e-06,
      "loss": 1.3272,
      "step": 12574
    },
    {
      "epoch": 1.9685347526612398,
      "grad_norm": 5.601173400878906,
      "learning_rate": 5.327468230694037e-06,
      "loss": 1.4548,
      "step": 12575
    },
    {
      "epoch": 1.9686912961803382,
      "grad_norm": 2.446711540222168,
      "learning_rate": 5.303030303030302e-06,
      "loss": 0.905,
      "step": 12576
    },
    {
      "epoch": 1.9688478396994364,
      "grad_norm": 2.5086371898651123,
      "learning_rate": 5.278592375366569e-06,
      "loss": 1.2203,
      "step": 12577
    },
    {
      "epoch": 1.9690043832185347,
      "grad_norm": 2.3690879344940186,
      "learning_rate": 5.254154447702834e-06,
      "loss": 0.9831,
      "step": 12578
    },
    {
      "epoch": 1.969160926737633,
      "grad_norm": 4.901893615722656,
      "learning_rate": 5.2297165200391e-06,
      "loss": 1.2816,
      "step": 12579
    },
    {
      "epoch": 1.9693174702567313,
      "grad_norm": 3.7496819496154785,
      "learning_rate": 5.205278592375365e-06,
      "loss": 0.8846,
      "step": 12580
    },
    {
      "epoch": 1.9694740137758298,
      "grad_norm": 6.612450122833252,
      "learning_rate": 5.180840664711632e-06,
      "loss": 1.17,
      "step": 12581
    },
    {
      "epoch": 1.969630557294928,
      "grad_norm": 5.878314018249512,
      "learning_rate": 5.1564027370478976e-06,
      "loss": 1.9362,
      "step": 12582
    },
    {
      "epoch": 1.9697871008140262,
      "grad_norm": 4.377343654632568,
      "learning_rate": 5.131964809384163e-06,
      "loss": 1.3451,
      "step": 12583
    },
    {
      "epoch": 1.9699436443331246,
      "grad_norm": 2.8641626834869385,
      "learning_rate": 5.10752688172043e-06,
      "loss": 0.6299,
      "step": 12584
    },
    {
      "epoch": 1.970100187852223,
      "grad_norm": 2.5689475536346436,
      "learning_rate": 5.083088954056695e-06,
      "loss": 0.5064,
      "step": 12585
    },
    {
      "epoch": 1.9702567313713213,
      "grad_norm": 2.1100947856903076,
      "learning_rate": 5.058651026392961e-06,
      "loss": 0.6611,
      "step": 12586
    },
    {
      "epoch": 1.9704132748904195,
      "grad_norm": 2.1359899044036865,
      "learning_rate": 5.0342130987292275e-06,
      "loss": 0.8358,
      "step": 12587
    },
    {
      "epoch": 1.9705698184095177,
      "grad_norm": 2.32698392868042,
      "learning_rate": 5.009775171065493e-06,
      "loss": 0.864,
      "step": 12588
    },
    {
      "epoch": 1.9707263619286162,
      "grad_norm": 0.9872207641601562,
      "learning_rate": 4.985337243401759e-06,
      "loss": 1.5319,
      "step": 12589
    },
    {
      "epoch": 1.9708829054477146,
      "grad_norm": 1.0228644609451294,
      "learning_rate": 4.960899315738025e-06,
      "loss": 1.4884,
      "step": 12590
    },
    {
      "epoch": 1.9710394489668128,
      "grad_norm": 0.822214663028717,
      "learning_rate": 4.936461388074291e-06,
      "loss": 1.5075,
      "step": 12591
    },
    {
      "epoch": 1.971195992485911,
      "grad_norm": 1.3645509481430054,
      "learning_rate": 4.9120234604105566e-06,
      "loss": 1.6097,
      "step": 12592
    },
    {
      "epoch": 1.9713525360050093,
      "grad_norm": 1.2646301984786987,
      "learning_rate": 4.887585532746823e-06,
      "loss": 1.5494,
      "step": 12593
    },
    {
      "epoch": 1.9715090795241077,
      "grad_norm": 1.6212059259414673,
      "learning_rate": 4.863147605083089e-06,
      "loss": 1.4176,
      "step": 12594
    },
    {
      "epoch": 1.9716656230432061,
      "grad_norm": 2.566788673400879,
      "learning_rate": 4.838709677419354e-06,
      "loss": 1.444,
      "step": 12595
    },
    {
      "epoch": 1.9718221665623044,
      "grad_norm": 1.1608645915985107,
      "learning_rate": 4.81427174975562e-06,
      "loss": 1.3891,
      "step": 12596
    },
    {
      "epoch": 1.9719787100814026,
      "grad_norm": 2.177610397338867,
      "learning_rate": 4.7898338220918865e-06,
      "loss": 1.3828,
      "step": 12597
    },
    {
      "epoch": 1.9721352536005008,
      "grad_norm": 5.3571977615356445,
      "learning_rate": 4.765395894428152e-06,
      "loss": 1.3869,
      "step": 12598
    },
    {
      "epoch": 1.9722917971195992,
      "grad_norm": 1.231022834777832,
      "learning_rate": 4.740957966764418e-06,
      "loss": 1.5305,
      "step": 12599
    },
    {
      "epoch": 1.9724483406386977,
      "grad_norm": 1.084704875946045,
      "learning_rate": 4.716520039100684e-06,
      "loss": 1.4779,
      "step": 12600
    },
    {
      "epoch": 1.972604884157796,
      "grad_norm": 2.484323024749756,
      "learning_rate": 4.69208211143695e-06,
      "loss": 1.4689,
      "step": 12601
    },
    {
      "epoch": 1.972761427676894,
      "grad_norm": 6.887755870819092,
      "learning_rate": 4.667644183773216e-06,
      "loss": 1.3976,
      "step": 12602
    },
    {
      "epoch": 1.9729179711959923,
      "grad_norm": 2.1070051193237305,
      "learning_rate": 4.643206256109481e-06,
      "loss": 1.4639,
      "step": 12603
    },
    {
      "epoch": 1.9730745147150908,
      "grad_norm": 8.759196281433105,
      "learning_rate": 4.618768328445748e-06,
      "loss": 1.6144,
      "step": 12604
    },
    {
      "epoch": 1.9732310582341892,
      "grad_norm": 2.5562074184417725,
      "learning_rate": 4.594330400782013e-06,
      "loss": 1.4443,
      "step": 12605
    },
    {
      "epoch": 1.9733876017532874,
      "grad_norm": 1.4550449848175049,
      "learning_rate": 4.569892473118279e-06,
      "loss": 1.4747,
      "step": 12606
    },
    {
      "epoch": 1.9735441452723856,
      "grad_norm": 3.1739792823791504,
      "learning_rate": 4.5454545454545455e-06,
      "loss": 1.3716,
      "step": 12607
    },
    {
      "epoch": 1.973700688791484,
      "grad_norm": 2.384730577468872,
      "learning_rate": 4.521016617790811e-06,
      "loss": 1.4844,
      "step": 12608
    },
    {
      "epoch": 1.9738572323105823,
      "grad_norm": 1.8984757661819458,
      "learning_rate": 4.496578690127077e-06,
      "loss": 1.5907,
      "step": 12609
    },
    {
      "epoch": 1.9740137758296807,
      "grad_norm": 2.8245761394500732,
      "learning_rate": 4.472140762463343e-06,
      "loss": 1.5429,
      "step": 12610
    },
    {
      "epoch": 1.974170319348779,
      "grad_norm": 3.4093105792999268,
      "learning_rate": 4.447702834799608e-06,
      "loss": 1.408,
      "step": 12611
    },
    {
      "epoch": 1.9743268628678772,
      "grad_norm": 2.0443670749664307,
      "learning_rate": 4.423264907135875e-06,
      "loss": 1.2528,
      "step": 12612
    },
    {
      "epoch": 1.9744834063869756,
      "grad_norm": 2.573211193084717,
      "learning_rate": 4.39882697947214e-06,
      "loss": 1.1372,
      "step": 12613
    },
    {
      "epoch": 1.974639949906074,
      "grad_norm": 3.4609265327453613,
      "learning_rate": 4.374389051808406e-06,
      "loss": 1.1394,
      "step": 12614
    },
    {
      "epoch": 1.9747964934251723,
      "grad_norm": 5.100785255432129,
      "learning_rate": 4.349951124144672e-06,
      "loss": 1.2423,
      "step": 12615
    },
    {
      "epoch": 1.9749530369442705,
      "grad_norm": 2.3350820541381836,
      "learning_rate": 4.325513196480938e-06,
      "loss": 1.1019,
      "step": 12616
    },
    {
      "epoch": 1.9751095804633687,
      "grad_norm": 6.419863224029541,
      "learning_rate": 4.301075268817204e-06,
      "loss": 1.3405,
      "step": 12617
    },
    {
      "epoch": 1.9752661239824671,
      "grad_norm": 2.2042758464813232,
      "learning_rate": 4.27663734115347e-06,
      "loss": 1.5057,
      "step": 12618
    },
    {
      "epoch": 1.9754226675015656,
      "grad_norm": 4.460781574249268,
      "learning_rate": 4.252199413489736e-06,
      "loss": 1.1672,
      "step": 12619
    },
    {
      "epoch": 1.9755792110206638,
      "grad_norm": 3.557311773300171,
      "learning_rate": 4.2277614858260015e-06,
      "loss": 1.1246,
      "step": 12620
    },
    {
      "epoch": 1.975735754539762,
      "grad_norm": 1.4838236570358276,
      "learning_rate": 4.203323558162268e-06,
      "loss": 1.149,
      "step": 12621
    },
    {
      "epoch": 1.9758922980588602,
      "grad_norm": 6.371537685394287,
      "learning_rate": 4.178885630498534e-06,
      "loss": 1.1258,
      "step": 12622
    },
    {
      "epoch": 1.9760488415779587,
      "grad_norm": 4.541806221008301,
      "learning_rate": 4.154447702834799e-06,
      "loss": 1.1318,
      "step": 12623
    },
    {
      "epoch": 1.9762053850970571,
      "grad_norm": 4.556163787841797,
      "learning_rate": 4.130009775171065e-06,
      "loss": 1.4951,
      "step": 12624
    },
    {
      "epoch": 1.9763619286161553,
      "grad_norm": 6.643764495849609,
      "learning_rate": 4.105571847507331e-06,
      "loss": 1.6353,
      "step": 12625
    },
    {
      "epoch": 1.9765184721352536,
      "grad_norm": 3.9981839656829834,
      "learning_rate": 4.081133919843597e-06,
      "loss": 1.3794,
      "step": 12626
    },
    {
      "epoch": 1.9766750156543518,
      "grad_norm": 2.1788060665130615,
      "learning_rate": 4.056695992179863e-06,
      "loss": 1.2775,
      "step": 12627
    },
    {
      "epoch": 1.9768315591734502,
      "grad_norm": 3.0667903423309326,
      "learning_rate": 4.032258064516128e-06,
      "loss": 1.3801,
      "step": 12628
    },
    {
      "epoch": 1.9769881026925487,
      "grad_norm": 5.267396926879883,
      "learning_rate": 4.007820136852395e-06,
      "loss": 1.6699,
      "step": 12629
    },
    {
      "epoch": 1.9771446462116469,
      "grad_norm": 3.0507497787475586,
      "learning_rate": 3.9833822091886605e-06,
      "loss": 1.2728,
      "step": 12630
    },
    {
      "epoch": 1.977301189730745,
      "grad_norm": 2.740083694458008,
      "learning_rate": 3.958944281524926e-06,
      "loss": 0.9162,
      "step": 12631
    },
    {
      "epoch": 1.9774577332498433,
      "grad_norm": 3.808227300643921,
      "learning_rate": 3.934506353861192e-06,
      "loss": 1.2745,
      "step": 12632
    },
    {
      "epoch": 1.9776142767689417,
      "grad_norm": 2.6046314239501953,
      "learning_rate": 3.910068426197458e-06,
      "loss": 1.1256,
      "step": 12633
    },
    {
      "epoch": 1.9777708202880402,
      "grad_norm": 2.029060125350952,
      "learning_rate": 3.885630498533724e-06,
      "loss": 0.5949,
      "step": 12634
    },
    {
      "epoch": 1.9779273638071384,
      "grad_norm": 4.188864707946777,
      "learning_rate": 3.86119257086999e-06,
      "loss": 0.8376,
      "step": 12635
    },
    {
      "epoch": 1.9780839073262366,
      "grad_norm": 1.602839708328247,
      "learning_rate": 3.836754643206256e-06,
      "loss": 0.4762,
      "step": 12636
    },
    {
      "epoch": 1.9782404508453348,
      "grad_norm": 1.9798146486282349,
      "learning_rate": 3.8123167155425217e-06,
      "loss": 0.7491,
      "step": 12637
    },
    {
      "epoch": 1.9783969943644333,
      "grad_norm": 7.58799409866333,
      "learning_rate": 3.7878787878787874e-06,
      "loss": 0.9718,
      "step": 12638
    },
    {
      "epoch": 1.9785535378835317,
      "grad_norm": 0.6994940638542175,
      "learning_rate": 3.7634408602150534e-06,
      "loss": 1.5842,
      "step": 12639
    },
    {
      "epoch": 1.97871008140263,
      "grad_norm": 1.8685157299041748,
      "learning_rate": 3.739002932551319e-06,
      "loss": 1.6148,
      "step": 12640
    },
    {
      "epoch": 1.9788666249217282,
      "grad_norm": 0.7222037315368652,
      "learning_rate": 3.714565004887585e-06,
      "loss": 1.5231,
      "step": 12641
    },
    {
      "epoch": 1.9790231684408266,
      "grad_norm": 0.7957490682601929,
      "learning_rate": 3.6901270772238512e-06,
      "loss": 1.5259,
      "step": 12642
    },
    {
      "epoch": 1.9791797119599248,
      "grad_norm": 0.8446835875511169,
      "learning_rate": 3.665689149560117e-06,
      "loss": 1.5504,
      "step": 12643
    },
    {
      "epoch": 1.9793362554790233,
      "grad_norm": 1.2421315908432007,
      "learning_rate": 3.641251221896383e-06,
      "loss": 1.5892,
      "step": 12644
    },
    {
      "epoch": 1.9794927989981215,
      "grad_norm": 1.1285243034362793,
      "learning_rate": 3.616813294232649e-06,
      "loss": 1.4458,
      "step": 12645
    },
    {
      "epoch": 1.9796493425172197,
      "grad_norm": 0.9664294123649597,
      "learning_rate": 3.5923753665689147e-06,
      "loss": 1.4354,
      "step": 12646
    },
    {
      "epoch": 1.9798058860363181,
      "grad_norm": 1.5649850368499756,
      "learning_rate": 3.5679374389051807e-06,
      "loss": 1.4355,
      "step": 12647
    },
    {
      "epoch": 1.9799624295554166,
      "grad_norm": 1.3277194499969482,
      "learning_rate": 3.5434995112414464e-06,
      "loss": 1.6038,
      "step": 12648
    },
    {
      "epoch": 1.9801189730745148,
      "grad_norm": 1.0830150842666626,
      "learning_rate": 3.5190615835777125e-06,
      "loss": 1.4237,
      "step": 12649
    },
    {
      "epoch": 1.980275516593613,
      "grad_norm": 1.434964895248413,
      "learning_rate": 3.4946236559139785e-06,
      "loss": 1.4098,
      "step": 12650
    },
    {
      "epoch": 1.9804320601127112,
      "grad_norm": 1.861877202987671,
      "learning_rate": 3.4701857282502437e-06,
      "loss": 1.5271,
      "step": 12651
    },
    {
      "epoch": 1.9805886036318097,
      "grad_norm": 2.560774326324463,
      "learning_rate": 3.44574780058651e-06,
      "loss": 1.3852,
      "step": 12652
    },
    {
      "epoch": 1.980745147150908,
      "grad_norm": 1.684672236442566,
      "learning_rate": 3.4213098729227763e-06,
      "loss": 1.3889,
      "step": 12653
    },
    {
      "epoch": 1.9809016906700063,
      "grad_norm": 1.0975278615951538,
      "learning_rate": 3.3968719452590415e-06,
      "loss": 1.3657,
      "step": 12654
    },
    {
      "epoch": 1.9810582341891045,
      "grad_norm": 1.292589783668518,
      "learning_rate": 3.3724340175953076e-06,
      "loss": 1.2904,
      "step": 12655
    },
    {
      "epoch": 1.9812147777082028,
      "grad_norm": 4.135151386260986,
      "learning_rate": 3.3479960899315733e-06,
      "loss": 1.4146,
      "step": 12656
    },
    {
      "epoch": 1.9813713212273012,
      "grad_norm": 2.1695103645324707,
      "learning_rate": 3.3235581622678393e-06,
      "loss": 1.4241,
      "step": 12657
    },
    {
      "epoch": 1.9815278647463996,
      "grad_norm": 5.103790283203125,
      "learning_rate": 3.2991202346041054e-06,
      "loss": 1.5548,
      "step": 12658
    },
    {
      "epoch": 1.9816844082654979,
      "grad_norm": 3.985460042953491,
      "learning_rate": 3.274682306940371e-06,
      "loss": 1.4314,
      "step": 12659
    },
    {
      "epoch": 1.981840951784596,
      "grad_norm": 1.8362746238708496,
      "learning_rate": 3.250244379276637e-06,
      "loss": 1.3154,
      "step": 12660
    },
    {
      "epoch": 1.9819974953036943,
      "grad_norm": 5.180610179901123,
      "learning_rate": 3.225806451612903e-06,
      "loss": 1.4386,
      "step": 12661
    },
    {
      "epoch": 1.9821540388227927,
      "grad_norm": 3.1565520763397217,
      "learning_rate": 3.201368523949169e-06,
      "loss": 1.3667,
      "step": 12662
    },
    {
      "epoch": 1.9823105823418912,
      "grad_norm": 1.2697768211364746,
      "learning_rate": 3.176930596285435e-06,
      "loss": 1.1725,
      "step": 12663
    },
    {
      "epoch": 1.9824671258609894,
      "grad_norm": 12.818161964416504,
      "learning_rate": 3.1524926686217006e-06,
      "loss": 1.7743,
      "step": 12664
    },
    {
      "epoch": 1.9826236693800876,
      "grad_norm": 4.739766597747803,
      "learning_rate": 3.1280547409579666e-06,
      "loss": 1.4785,
      "step": 12665
    },
    {
      "epoch": 1.9827802128991858,
      "grad_norm": 2.6713364124298096,
      "learning_rate": 3.1036168132942327e-06,
      "loss": 1.4496,
      "step": 12666
    },
    {
      "epoch": 1.9829367564182843,
      "grad_norm": 2.46557879447937,
      "learning_rate": 3.079178885630498e-06,
      "loss": 1.255,
      "step": 12667
    },
    {
      "epoch": 1.9830932999373827,
      "grad_norm": 3.2623789310455322,
      "learning_rate": 3.0547409579667644e-06,
      "loss": 1.3316,
      "step": 12668
    },
    {
      "epoch": 1.983249843456481,
      "grad_norm": 3.391611337661743,
      "learning_rate": 3.0303030303030305e-06,
      "loss": 1.1054,
      "step": 12669
    },
    {
      "epoch": 1.9834063869755791,
      "grad_norm": 2.285482168197632,
      "learning_rate": 3.0058651026392957e-06,
      "loss": 1.035,
      "step": 12670
    },
    {
      "epoch": 1.9835629304946774,
      "grad_norm": 1.301674246788025,
      "learning_rate": 2.9814271749755618e-06,
      "loss": 0.9496,
      "step": 12671
    },
    {
      "epoch": 1.9837194740137758,
      "grad_norm": 21.29591178894043,
      "learning_rate": 2.9569892473118274e-06,
      "loss": 1.2513,
      "step": 12672
    },
    {
      "epoch": 1.9838760175328742,
      "grad_norm": 3.679635763168335,
      "learning_rate": 2.9325513196480935e-06,
      "loss": 0.8991,
      "step": 12673
    },
    {
      "epoch": 1.9840325610519725,
      "grad_norm": 3.6654906272888184,
      "learning_rate": 2.9081133919843596e-06,
      "loss": 1.2406,
      "step": 12674
    },
    {
      "epoch": 1.9841891045710707,
      "grad_norm": 3.2650372982025146,
      "learning_rate": 2.8836754643206252e-06,
      "loss": 1.1958,
      "step": 12675
    },
    {
      "epoch": 1.9843456480901691,
      "grad_norm": 4.416281223297119,
      "learning_rate": 2.8592375366568913e-06,
      "loss": 1.1306,
      "step": 12676
    },
    {
      "epoch": 1.9845021916092673,
      "grad_norm": 2.6011712551116943,
      "learning_rate": 2.8347996089931574e-06,
      "loss": 1.1018,
      "step": 12677
    },
    {
      "epoch": 1.9846587351283658,
      "grad_norm": 3.921834945678711,
      "learning_rate": 2.810361681329423e-06,
      "loss": 0.9713,
      "step": 12678
    },
    {
      "epoch": 1.984815278647464,
      "grad_norm": 5.239248275756836,
      "learning_rate": 2.785923753665689e-06,
      "loss": 1.1969,
      "step": 12679
    },
    {
      "epoch": 1.9849718221665622,
      "grad_norm": 7.275485038757324,
      "learning_rate": 2.7614858260019547e-06,
      "loss": 0.8116,
      "step": 12680
    },
    {
      "epoch": 1.9851283656856606,
      "grad_norm": 1.9333630800247192,
      "learning_rate": 2.737047898338221e-06,
      "loss": 1.3322,
      "step": 12681
    },
    {
      "epoch": 1.985284909204759,
      "grad_norm": 3.5819573402404785,
      "learning_rate": 2.712609970674487e-06,
      "loss": 1.0275,
      "step": 12682
    },
    {
      "epoch": 1.9854414527238573,
      "grad_norm": 3.910200834274292,
      "learning_rate": 2.6881720430107525e-06,
      "loss": 0.8067,
      "step": 12683
    },
    {
      "epoch": 1.9855979962429555,
      "grad_norm": 3.8609650135040283,
      "learning_rate": 2.6637341153470186e-06,
      "loss": 0.7034,
      "step": 12684
    },
    {
      "epoch": 1.9857545397620537,
      "grad_norm": 2.1950485706329346,
      "learning_rate": 2.6392961876832847e-06,
      "loss": 0.8698,
      "step": 12685
    },
    {
      "epoch": 1.9859110832811522,
      "grad_norm": 3.267005205154419,
      "learning_rate": 2.61485826001955e-06,
      "loss": 0.9502,
      "step": 12686
    },
    {
      "epoch": 1.9860676268002506,
      "grad_norm": 1.1924513578414917,
      "learning_rate": 2.590420332355816e-06,
      "loss": 0.6124,
      "step": 12687
    },
    {
      "epoch": 1.9862241703193488,
      "grad_norm": 1.8844116926193237,
      "learning_rate": 2.5659824046920816e-06,
      "loss": 0.7941,
      "step": 12688
    },
    {
      "epoch": 1.986380713838447,
      "grad_norm": 0.9816786050796509,
      "learning_rate": 2.5415444770283477e-06,
      "loss": 1.4848,
      "step": 12689
    },
    {
      "epoch": 1.9865372573575453,
      "grad_norm": 0.7692421078681946,
      "learning_rate": 2.5171065493646137e-06,
      "loss": 1.4116,
      "step": 12690
    },
    {
      "epoch": 1.9866938008766437,
      "grad_norm": 0.8593037724494934,
      "learning_rate": 2.4926686217008794e-06,
      "loss": 1.4803,
      "step": 12691
    },
    {
      "epoch": 1.9868503443957422,
      "grad_norm": 1.4960730075836182,
      "learning_rate": 2.4682306940371455e-06,
      "loss": 1.4621,
      "step": 12692
    },
    {
      "epoch": 1.9870068879148404,
      "grad_norm": 1.2484159469604492,
      "learning_rate": 2.4437927663734115e-06,
      "loss": 1.3931,
      "step": 12693
    },
    {
      "epoch": 1.9871634314339386,
      "grad_norm": 2.241774320602417,
      "learning_rate": 2.419354838709677e-06,
      "loss": 1.4726,
      "step": 12694
    },
    {
      "epoch": 1.9873199749530368,
      "grad_norm": 1.0046658515930176,
      "learning_rate": 2.3949169110459433e-06,
      "loss": 1.3704,
      "step": 12695
    },
    {
      "epoch": 1.9874765184721352,
      "grad_norm": 1.600348949432373,
      "learning_rate": 2.370478983382209e-06,
      "loss": 1.4287,
      "step": 12696
    },
    {
      "epoch": 1.9876330619912337,
      "grad_norm": 0.8872906565666199,
      "learning_rate": 2.346041055718475e-06,
      "loss": 1.3028,
      "step": 12697
    },
    {
      "epoch": 1.987789605510332,
      "grad_norm": 1.608568787574768,
      "learning_rate": 2.3216031280547406e-06,
      "loss": 1.4942,
      "step": 12698
    },
    {
      "epoch": 1.9879461490294301,
      "grad_norm": 2.1755223274230957,
      "learning_rate": 2.2971652003910067e-06,
      "loss": 1.3595,
      "step": 12699
    },
    {
      "epoch": 1.9881026925485283,
      "grad_norm": 1.844046711921692,
      "learning_rate": 2.2727272727272728e-06,
      "loss": 1.396,
      "step": 12700
    },
    {
      "epoch": 1.9882592360676268,
      "grad_norm": 1.6896350383758545,
      "learning_rate": 2.2482893450635384e-06,
      "loss": 1.4409,
      "step": 12701
    },
    {
      "epoch": 1.9884157795867252,
      "grad_norm": 2.31750226020813,
      "learning_rate": 2.223851417399804e-06,
      "loss": 1.3609,
      "step": 12702
    },
    {
      "epoch": 1.9885723231058234,
      "grad_norm": 2.511352062225342,
      "learning_rate": 2.19941348973607e-06,
      "loss": 1.3918,
      "step": 12703
    },
    {
      "epoch": 1.9887288666249217,
      "grad_norm": 1.6797364950180054,
      "learning_rate": 2.174975562072336e-06,
      "loss": 1.3564,
      "step": 12704
    },
    {
      "epoch": 1.9888854101440199,
      "grad_norm": 3.3981409072875977,
      "learning_rate": 2.150537634408602e-06,
      "loss": 1.4078,
      "step": 12705
    },
    {
      "epoch": 1.9890419536631183,
      "grad_norm": 1.422497272491455,
      "learning_rate": 2.126099706744868e-06,
      "loss": 1.2969,
      "step": 12706
    },
    {
      "epoch": 1.9891984971822168,
      "grad_norm": 6.27196741104126,
      "learning_rate": 2.101661779081134e-06,
      "loss": 1.3743,
      "step": 12707
    },
    {
      "epoch": 1.989355040701315,
      "grad_norm": 2.9742422103881836,
      "learning_rate": 2.0772238514173996e-06,
      "loss": 1.1895,
      "step": 12708
    },
    {
      "epoch": 1.9895115842204132,
      "grad_norm": 2.017789363861084,
      "learning_rate": 2.0527859237536657e-06,
      "loss": 1.3513,
      "step": 12709
    },
    {
      "epoch": 1.9896681277395116,
      "grad_norm": 4.888723850250244,
      "learning_rate": 2.0283479960899314e-06,
      "loss": 1.5357,
      "step": 12710
    },
    {
      "epoch": 1.9898246712586098,
      "grad_norm": 2.907987594604492,
      "learning_rate": 2.0039100684261974e-06,
      "loss": 1.4441,
      "step": 12711
    },
    {
      "epoch": 1.9899812147777083,
      "grad_norm": 4.175799369812012,
      "learning_rate": 1.979472140762463e-06,
      "loss": 1.3473,
      "step": 12712
    },
    {
      "epoch": 1.9901377582968065,
      "grad_norm": 1.9999713897705078,
      "learning_rate": 1.955034213098729e-06,
      "loss": 1.4132,
      "step": 12713
    },
    {
      "epoch": 1.9902943018159047,
      "grad_norm": 2.5126242637634277,
      "learning_rate": 1.930596285434995e-06,
      "loss": 1.2115,
      "step": 12714
    },
    {
      "epoch": 1.9904508453350032,
      "grad_norm": 1.8449431657791138,
      "learning_rate": 1.9061583577712609e-06,
      "loss": 1.1961,
      "step": 12715
    },
    {
      "epoch": 1.9906073888541016,
      "grad_norm": 8.649978637695312,
      "learning_rate": 1.8817204301075267e-06,
      "loss": 1.2299,
      "step": 12716
    },
    {
      "epoch": 1.9907639323731998,
      "grad_norm": 12.748600006103516,
      "learning_rate": 1.8572825024437926e-06,
      "loss": 1.2318,
      "step": 12717
    },
    {
      "epoch": 1.990920475892298,
      "grad_norm": 9.06901741027832,
      "learning_rate": 1.8328445747800584e-06,
      "loss": 1.4941,
      "step": 12718
    },
    {
      "epoch": 1.9910770194113963,
      "grad_norm": 6.057063102722168,
      "learning_rate": 1.8084066471163245e-06,
      "loss": 1.0969,
      "step": 12719
    },
    {
      "epoch": 1.9912335629304947,
      "grad_norm": 3.5154621601104736,
      "learning_rate": 1.7839687194525904e-06,
      "loss": 1.2691,
      "step": 12720
    },
    {
      "epoch": 1.9913901064495931,
      "grad_norm": 3.144012451171875,
      "learning_rate": 1.7595307917888562e-06,
      "loss": 1.0831,
      "step": 12721
    },
    {
      "epoch": 1.9915466499686914,
      "grad_norm": 7.510293483734131,
      "learning_rate": 1.7350928641251219e-06,
      "loss": 1.0199,
      "step": 12722
    },
    {
      "epoch": 1.9917031934877896,
      "grad_norm": 3.6680872440338135,
      "learning_rate": 1.7106549364613882e-06,
      "loss": 1.3326,
      "step": 12723
    },
    {
      "epoch": 1.9918597370068878,
      "grad_norm": 7.296219348907471,
      "learning_rate": 1.6862170087976538e-06,
      "loss": 1.1472,
      "step": 12724
    },
    {
      "epoch": 1.9920162805259862,
      "grad_norm": 3.799044132232666,
      "learning_rate": 1.6617790811339197e-06,
      "loss": 0.9459,
      "step": 12725
    },
    {
      "epoch": 1.9921728240450847,
      "grad_norm": 4.250785827636719,
      "learning_rate": 1.6373411534701855e-06,
      "loss": 1.1241,
      "step": 12726
    },
    {
      "epoch": 1.9923293675641829,
      "grad_norm": 5.237523078918457,
      "learning_rate": 1.6129032258064516e-06,
      "loss": 1.1708,
      "step": 12727
    },
    {
      "epoch": 1.992485911083281,
      "grad_norm": 2.55849552154541,
      "learning_rate": 1.5884652981427175e-06,
      "loss": 1.2067,
      "step": 12728
    },
    {
      "epoch": 1.9926424546023793,
      "grad_norm": 4.564910411834717,
      "learning_rate": 1.5640273704789833e-06,
      "loss": 1.8229,
      "step": 12729
    },
    {
      "epoch": 1.9927989981214778,
      "grad_norm": 3.0857863426208496,
      "learning_rate": 1.539589442815249e-06,
      "loss": 1.2861,
      "step": 12730
    },
    {
      "epoch": 1.9929555416405762,
      "grad_norm": 3.102818489074707,
      "learning_rate": 1.5151515151515152e-06,
      "loss": 1.1077,
      "step": 12731
    },
    {
      "epoch": 1.9931120851596744,
      "grad_norm": 1.7498348951339722,
      "learning_rate": 1.4907135874877809e-06,
      "loss": 0.6339,
      "step": 12732
    },
    {
      "epoch": 1.9932686286787726,
      "grad_norm": NaN,
      "learning_rate": 1.4907135874877809e-06,
      "loss": 0.0,
      "step": 12733
    },
    {
      "epoch": 1.9934251721978709,
      "grad_norm": 8.903223991394043,
      "learning_rate": 1.4662756598240468e-06,
      "loss": 0.9257,
      "step": 12734
    },
    {
      "epoch": 1.9935817157169693,
      "grad_norm": 9.738998413085938,
      "learning_rate": 1.4418377321603126e-06,
      "loss": 1.1337,
      "step": 12735
    },
    {
      "epoch": 1.9937382592360677,
      "grad_norm": 2.1444084644317627,
      "learning_rate": 1.4173998044965787e-06,
      "loss": 0.5903,
      "step": 12736
    },
    {
      "epoch": 1.993894802755166,
      "grad_norm": 3.0119729042053223,
      "learning_rate": 1.3929618768328445e-06,
      "loss": 1.2444,
      "step": 12737
    },
    {
      "epoch": 1.9940513462742642,
      "grad_norm": 4.827343940734863,
      "learning_rate": 1.3685239491691104e-06,
      "loss": 1.2683,
      "step": 12738
    },
    {
      "epoch": 1.9942078897933626,
      "grad_norm": 1.7269351482391357,
      "learning_rate": 1.3440860215053763e-06,
      "loss": 1.4892,
      "step": 12739
    },
    {
      "epoch": 1.9943644333124608,
      "grad_norm": 1.236871600151062,
      "learning_rate": 1.3196480938416423e-06,
      "loss": 1.4841,
      "step": 12740
    },
    {
      "epoch": 1.9945209768315593,
      "grad_norm": 1.0202114582061768,
      "learning_rate": 1.295210166177908e-06,
      "loss": 1.4703,
      "step": 12741
    },
    {
      "epoch": 1.9946775203506575,
      "grad_norm": 0.5978159308433533,
      "learning_rate": 1.2707722385141738e-06,
      "loss": 1.4379,
      "step": 12742
    },
    {
      "epoch": 1.9948340638697557,
      "grad_norm": 1.8672462701797485,
      "learning_rate": 1.2463343108504397e-06,
      "loss": 1.3344,
      "step": 12743
    },
    {
      "epoch": 1.9949906073888541,
      "grad_norm": 0.9997517466545105,
      "learning_rate": 1.2218963831867058e-06,
      "loss": 1.4433,
      "step": 12744
    },
    {
      "epoch": 1.9951471509079524,
      "grad_norm": 1.3582754135131836,
      "learning_rate": 1.1974584555229716e-06,
      "loss": 1.4282,
      "step": 12745
    },
    {
      "epoch": 1.9953036944270508,
      "grad_norm": 1.0937734842300415,
      "learning_rate": 1.1730205278592375e-06,
      "loss": 1.4486,
      "step": 12746
    },
    {
      "epoch": 1.995460237946149,
      "grad_norm": 0.7297368049621582,
      "learning_rate": 1.1485826001955033e-06,
      "loss": 1.3394,
      "step": 12747
    },
    {
      "epoch": 1.9956167814652472,
      "grad_norm": 1.304893136024475,
      "learning_rate": 1.1241446725317692e-06,
      "loss": 1.3208,
      "step": 12748
    },
    {
      "epoch": 1.9957733249843457,
      "grad_norm": 0.9732430577278137,
      "learning_rate": 1.099706744868035e-06,
      "loss": 1.3768,
      "step": 12749
    },
    {
      "epoch": 1.9959298685034441,
      "grad_norm": 2.0447745323181152,
      "learning_rate": 1.075268817204301e-06,
      "loss": 1.3748,
      "step": 12750
    },
    {
      "epoch": 1.9960864120225423,
      "grad_norm": 1.8304771184921265,
      "learning_rate": 1.050830889540567e-06,
      "loss": 1.4384,
      "step": 12751
    },
    {
      "epoch": 1.9962429555416406,
      "grad_norm": 1.1020729541778564,
      "learning_rate": 1.0263929618768329e-06,
      "loss": 1.3156,
      "step": 12752
    },
    {
      "epoch": 1.9963994990607388,
      "grad_norm": 2.1293833255767822,
      "learning_rate": 1.0019550342130987e-06,
      "loss": 1.2381,
      "step": 12753
    },
    {
      "epoch": 1.9965560425798372,
      "grad_norm": 2.4541094303131104,
      "learning_rate": 9.775171065493646e-07,
      "loss": 1.2771,
      "step": 12754
    },
    {
      "epoch": 1.9967125860989356,
      "grad_norm": 3.102612257003784,
      "learning_rate": 9.530791788856304e-07,
      "loss": 1.3604,
      "step": 12755
    },
    {
      "epoch": 1.9968691296180339,
      "grad_norm": 1.3522253036499023,
      "learning_rate": 9.286412512218963e-07,
      "loss": 1.2247,
      "step": 12756
    },
    {
      "epoch": 1.997025673137132,
      "grad_norm": 4.309826850891113,
      "learning_rate": 9.042033235581623e-07,
      "loss": 1.3129,
      "step": 12757
    },
    {
      "epoch": 1.9971822166562303,
      "grad_norm": 1.946575403213501,
      "learning_rate": 8.797653958944281e-07,
      "loss": 1.1901,
      "step": 12758
    },
    {
      "epoch": 1.9973387601753287,
      "grad_norm": 2.3310208320617676,
      "learning_rate": 8.553274682306941e-07,
      "loss": 1.0399,
      "step": 12759
    },
    {
      "epoch": 1.9974953036944272,
      "grad_norm": 1.9439860582351685,
      "learning_rate": 8.308895405669598e-07,
      "loss": 1.1825,
      "step": 12760
    },
    {
      "epoch": 1.9976518472135254,
      "grad_norm": 2.59775972366333,
      "learning_rate": 8.064516129032258e-07,
      "loss": 1.2622,
      "step": 12761
    },
    {
      "epoch": 1.9978083907326236,
      "grad_norm": 4.020702838897705,
      "learning_rate": 7.820136852394917e-07,
      "loss": 1.4661,
      "step": 12762
    },
    {
      "epoch": 1.9979649342517218,
      "grad_norm": 4.491816997528076,
      "learning_rate": 7.575757575757576e-07,
      "loss": 1.2114,
      "step": 12763
    },
    {
      "epoch": 1.9981214777708203,
      "grad_norm": 3.0681772232055664,
      "learning_rate": 7.331378299120234e-07,
      "loss": 1.169,
      "step": 12764
    },
    {
      "epoch": 1.9982780212899187,
      "grad_norm": 11.430411338806152,
      "learning_rate": 7.086999022482893e-07,
      "loss": 1.5927,
      "step": 12765
    },
    {
      "epoch": 1.998434564809017,
      "grad_norm": 8.486896514892578,
      "learning_rate": 6.842619745845552e-07,
      "loss": 1.3744,
      "step": 12766
    },
    {
      "epoch": 1.9985911083281152,
      "grad_norm": 14.378063201904297,
      "learning_rate": 6.598240469208212e-07,
      "loss": 1.1301,
      "step": 12767
    },
    {
      "epoch": 1.9987476518472134,
      "grad_norm": 15.947986602783203,
      "learning_rate": 6.353861192570869e-07,
      "loss": 1.7693,
      "step": 12768
    },
    {
      "epoch": 1.9989041953663118,
      "grad_norm": 10.535764694213867,
      "learning_rate": 6.109481915933529e-07,
      "loss": 1.2029,
      "step": 12769
    },
    {
      "epoch": 1.9990607388854102,
      "grad_norm": 3.89440655708313,
      "learning_rate": 5.865102639296187e-07,
      "loss": 1.4126,
      "step": 12770
    },
    {
      "epoch": 1.9992172824045085,
      "grad_norm": 2.050438642501831,
      "learning_rate": 5.620723362658846e-07,
      "loss": 1.1434,
      "step": 12771
    },
    {
      "epoch": 1.9993738259236067,
      "grad_norm": 1.9969635009765625,
      "learning_rate": 5.376344086021505e-07,
      "loss": 1.0368,
      "step": 12772
    },
    {
      "epoch": 1.9995303694427051,
      "grad_norm": 4.676234722137451,
      "learning_rate": 5.131964809384164e-07,
      "loss": 0.965,
      "step": 12773
    },
    {
      "epoch": 1.9996869129618033,
      "grad_norm": 1.3593485355377197,
      "learning_rate": 4.887585532746823e-07,
      "loss": 0.5821,
      "step": 12774
    },
    {
      "epoch": 1.9998434564809018,
      "grad_norm": 2.8748373985290527,
      "learning_rate": 4.6432062561094814e-07,
      "loss": 1.5815,
      "step": 12775
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.058191299438477,
      "learning_rate": 4.3988269794721406e-07,
      "loss": 1.2623,
      "step": 12776
    },
    {
      "epoch": 2.0,
      "step": 12776,
      "total_flos": 1.799905283771071e+19,
      "train_loss": 0.9086307350367672,
      "train_runtime": 8068.0189,
      "train_samples_per_second": 25.335,
      "train_steps_per_second": 1.584
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 12776,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 400,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.799905283771071e+19,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}