{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 2077,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00048155059290916753,
      "grad_norm": 1.8261996507644653,
      "learning_rate": 0.0,
      "loss": 2.6808,
      "step": 1
    },
    {
      "epoch": 0.0009631011858183351,
      "grad_norm": 3.358039617538452,
      "learning_rate": 3.1746031746031746e-06,
      "loss": 3.2673,
      "step": 2
    },
    {
      "epoch": 0.0014446517787275025,
      "grad_norm": 1.8629053831100464,
      "learning_rate": 6.349206349206349e-06,
      "loss": 1.549,
      "step": 3
    },
    {
      "epoch": 0.0019262023716366701,
      "grad_norm": 2.996866226196289,
      "learning_rate": 9.523809523809523e-06,
      "loss": 2.3225,
      "step": 4
    },
    {
      "epoch": 0.0024077529645458377,
      "grad_norm": 2.880138397216797,
      "learning_rate": 1.2698412698412699e-05,
      "loss": 2.2565,
      "step": 5
    },
    {
      "epoch": 0.002889303557455005,
      "grad_norm": 2.7909672260284424,
      "learning_rate": 1.5873015873015872e-05,
      "loss": 2.7257,
      "step": 6
    },
    {
      "epoch": 0.0033708541503641725,
      "grad_norm": 2.0042378902435303,
      "learning_rate": 1.9047619047619046e-05,
      "loss": 2.6491,
      "step": 7
    },
    {
      "epoch": 0.0038524047432733403,
      "grad_norm": 3.720799446105957,
      "learning_rate": 2.2222222222222223e-05,
      "loss": 2.9006,
      "step": 8
    },
    {
      "epoch": 0.004333955336182508,
      "grad_norm": 2.375181198120117,
      "learning_rate": 2.5396825396825397e-05,
      "loss": 2.3923,
      "step": 9
    },
    {
      "epoch": 0.004815505929091675,
      "grad_norm": 1.4350911378860474,
      "learning_rate": 2.857142857142857e-05,
      "loss": 2.3525,
      "step": 10
    },
    {
      "epoch": 0.005297056522000842,
      "grad_norm": 2.192746877670288,
      "learning_rate": 3.1746031746031745e-05,
      "loss": 2.6314,
      "step": 11
    },
    {
      "epoch": 0.00577860711491001,
      "grad_norm": 2.007908344268799,
      "learning_rate": 3.492063492063492e-05,
      "loss": 1.9395,
      "step": 12
    },
    {
      "epoch": 0.006260157707819178,
      "grad_norm": 3.7786455154418945,
      "learning_rate": 3.809523809523809e-05,
      "loss": 2.2888,
      "step": 13
    },
    {
      "epoch": 0.006741708300728345,
      "grad_norm": 3.524888515472412,
      "learning_rate": 4.126984126984127e-05,
      "loss": 1.9405,
      "step": 14
    },
    {
      "epoch": 0.007223258893637513,
      "grad_norm": 1.4730066061019897,
      "learning_rate": 4.4444444444444447e-05,
      "loss": 2.0121,
      "step": 15
    },
    {
      "epoch": 0.0077048094865466805,
      "grad_norm": 3.7371699810028076,
      "learning_rate": 4.761904761904762e-05,
      "loss": 2.3352,
      "step": 16
    },
    {
      "epoch": 0.008186360079455848,
      "grad_norm": 2.437157154083252,
      "learning_rate": 5.0793650793650794e-05,
      "loss": 1.3663,
      "step": 17
    },
    {
      "epoch": 0.008667910672365015,
      "grad_norm": 2.5214345455169678,
      "learning_rate": 5.396825396825397e-05,
      "loss": 1.757,
      "step": 18
    },
    {
      "epoch": 0.009149461265274182,
      "grad_norm": 1.2780380249023438,
      "learning_rate": 5.714285714285714e-05,
      "loss": 1.6609,
      "step": 19
    },
    {
      "epoch": 0.00963101185818335,
      "grad_norm": 2.314685821533203,
      "learning_rate": 6.0317460317460316e-05,
      "loss": 1.9831,
      "step": 20
    },
    {
      "epoch": 0.010112562451092518,
      "grad_norm": 1.9958107471466064,
      "learning_rate": 6.349206349206349e-05,
      "loss": 1.6875,
      "step": 21
    },
    {
      "epoch": 0.010594113044001685,
      "grad_norm": 2.490067958831787,
      "learning_rate": 6.666666666666667e-05,
      "loss": 2.1276,
      "step": 22
    },
    {
      "epoch": 0.011075663636910853,
      "grad_norm": 2.0263242721557617,
      "learning_rate": 6.984126984126984e-05,
      "loss": 1.5715,
      "step": 23
    },
    {
      "epoch": 0.01155721422982002,
      "grad_norm": 2.300640106201172,
      "learning_rate": 7.301587301587302e-05,
      "loss": 1.2921,
      "step": 24
    },
    {
      "epoch": 0.012038764822729187,
      "grad_norm": 2.50140380859375,
      "learning_rate": 7.619047619047618e-05,
      "loss": 1.5564,
      "step": 25
    },
    {
      "epoch": 0.012520315415638356,
      "grad_norm": 4.3480377197265625,
      "learning_rate": 7.936507936507937e-05,
      "loss": 1.5753,
      "step": 26
    },
    {
      "epoch": 0.013001866008547523,
      "grad_norm": 2.933090925216675,
      "learning_rate": 8.253968253968255e-05,
      "loss": 1.29,
      "step": 27
    },
    {
      "epoch": 0.01348341660145669,
      "grad_norm": 3.105729103088379,
      "learning_rate": 8.571428571428571e-05,
      "loss": 1.3442,
      "step": 28
    },
    {
      "epoch": 0.013964967194365858,
      "grad_norm": 1.3110624551773071,
      "learning_rate": 8.888888888888889e-05,
      "loss": 2.2436,
      "step": 29
    },
    {
      "epoch": 0.014446517787275025,
      "grad_norm": 1.9715465307235718,
      "learning_rate": 9.206349206349206e-05,
      "loss": 1.7574,
      "step": 30
    },
    {
      "epoch": 0.014928068380184192,
      "grad_norm": 1.7527216672897339,
      "learning_rate": 9.523809523809524e-05,
      "loss": 1.2682,
      "step": 31
    },
    {
      "epoch": 0.015409618973093361,
      "grad_norm": 0.927869439125061,
      "learning_rate": 9.841269841269841e-05,
      "loss": 1.7568,
      "step": 32
    },
    {
      "epoch": 0.01589116956600253,
      "grad_norm": 1.616967797279358,
      "learning_rate": 0.00010158730158730159,
      "loss": 1.7164,
      "step": 33
    },
    {
      "epoch": 0.016372720158911697,
      "grad_norm": 1.393038034439087,
      "learning_rate": 0.00010476190476190477,
      "loss": 1.4894,
      "step": 34
    },
    {
      "epoch": 0.016854270751820864,
      "grad_norm": 2.07956600189209,
      "learning_rate": 0.00010793650793650794,
      "loss": 1.621,
      "step": 35
    },
    {
      "epoch": 0.01733582134473003,
      "grad_norm": 0.7870104908943176,
      "learning_rate": 0.00011111111111111112,
      "loss": 1.7103,
      "step": 36
    },
    {
      "epoch": 0.017817371937639197,
      "grad_norm": 1.6282917261123657,
      "learning_rate": 0.00011428571428571428,
      "loss": 2.0154,
      "step": 37
    },
    {
      "epoch": 0.018298922530548364,
      "grad_norm": 2.2591805458068848,
      "learning_rate": 0.00011746031746031746,
      "loss": 2.1584,
      "step": 38
    },
    {
      "epoch": 0.018780473123457535,
      "grad_norm": 1.8572083711624146,
      "learning_rate": 0.00012063492063492063,
      "loss": 1.4158,
      "step": 39
    },
    {
      "epoch": 0.0192620237163667,
      "grad_norm": 1.4572298526763916,
      "learning_rate": 0.0001238095238095238,
      "loss": 1.9786,
      "step": 40
    },
    {
      "epoch": 0.01974357430927587,
      "grad_norm": 1.1680176258087158,
      "learning_rate": 0.00012698412698412698,
      "loss": 1.2879,
      "step": 41
    },
    {
      "epoch": 0.020225124902185036,
      "grad_norm": 1.9423965215682983,
      "learning_rate": 0.00013015873015873017,
      "loss": 1.4253,
      "step": 42
    },
    {
      "epoch": 0.020706675495094203,
      "grad_norm": 2.308589220046997,
      "learning_rate": 0.00013333333333333334,
      "loss": 1.6032,
      "step": 43
    },
    {
      "epoch": 0.02118822608800337,
      "grad_norm": 3.3484935760498047,
      "learning_rate": 0.0001365079365079365,
      "loss": 1.4502,
      "step": 44
    },
    {
      "epoch": 0.02166977668091254,
      "grad_norm": 3.0905227661132812,
      "learning_rate": 0.00013968253968253967,
      "loss": 1.6035,
      "step": 45
    },
    {
      "epoch": 0.022151327273821707,
      "grad_norm": 3.6227426528930664,
      "learning_rate": 0.00014285714285714287,
      "loss": 1.4399,
      "step": 46
    },
    {
      "epoch": 0.022632877866730874,
      "grad_norm": 3.1127336025238037,
      "learning_rate": 0.00014603174603174603,
      "loss": 0.9448,
      "step": 47
    },
    {
      "epoch": 0.02311442845964004,
      "grad_norm": 2.8288960456848145,
      "learning_rate": 0.00014920634920634923,
      "loss": 1.3749,
      "step": 48
    },
    {
      "epoch": 0.023595979052549208,
      "grad_norm": 3.2530457973480225,
      "learning_rate": 0.00015238095238095237,
      "loss": 1.9887,
      "step": 49
    },
    {
      "epoch": 0.024077529645458375,
      "grad_norm": 1.2616008520126343,
      "learning_rate": 0.00015555555555555556,
      "loss": 1.7782,
      "step": 50
    },
    {
      "epoch": 0.024559080238367545,
      "grad_norm": 1.262278437614441,
      "learning_rate": 0.00015873015873015873,
      "loss": 1.7173,
      "step": 51
    },
    {
      "epoch": 0.025040630831276712,
      "grad_norm": 1.2049165964126587,
      "learning_rate": 0.00016190476190476192,
      "loss": 1.5916,
      "step": 52
    },
    {
      "epoch": 0.02552218142418588,
      "grad_norm": 1.6197431087493896,
      "learning_rate": 0.0001650793650793651,
      "loss": 1.3619,
      "step": 53
    },
    {
      "epoch": 0.026003732017095046,
      "grad_norm": 2.1022942066192627,
      "learning_rate": 0.00016825396825396826,
      "loss": 1.0357,
      "step": 54
    },
    {
      "epoch": 0.026485282610004213,
      "grad_norm": 1.5700383186340332,
      "learning_rate": 0.00017142857142857143,
      "loss": 1.7315,
      "step": 55
    },
    {
      "epoch": 0.02696683320291338,
      "grad_norm": 1.8039864301681519,
      "learning_rate": 0.00017460317460317462,
      "loss": 1.5191,
      "step": 56
    },
    {
      "epoch": 0.02744838379582255,
      "grad_norm": 1.9089723825454712,
      "learning_rate": 0.00017777777777777779,
      "loss": 1.6027,
      "step": 57
    },
    {
      "epoch": 0.027929934388731717,
      "grad_norm": 2.2609500885009766,
      "learning_rate": 0.00018095238095238095,
      "loss": 1.839,
      "step": 58
    },
    {
      "epoch": 0.028411484981640884,
      "grad_norm": 1.3988934755325317,
      "learning_rate": 0.00018412698412698412,
      "loss": 1.4314,
      "step": 59
    },
    {
      "epoch": 0.02889303557455005,
      "grad_norm": 1.7800880670547485,
      "learning_rate": 0.00018730158730158731,
      "loss": 1.5032,
      "step": 60
    },
    {
      "epoch": 0.029374586167459218,
      "grad_norm": 1.9672104120254517,
      "learning_rate": 0.00019047619047619048,
      "loss": 1.5082,
      "step": 61
    },
    {
      "epoch": 0.029856136760368385,
      "grad_norm": 1.4464588165283203,
      "learning_rate": 0.00019365079365079365,
      "loss": 1.6496,
      "step": 62
    },
    {
      "epoch": 0.030337687353277555,
      "grad_norm": 1.5385384559631348,
      "learning_rate": 0.00019682539682539682,
      "loss": 1.648,
      "step": 63
    },
    {
      "epoch": 0.030819237946186722,
      "grad_norm": 1.5152275562286377,
      "learning_rate": 0.0002,
      "loss": 1.8346,
      "step": 64
    },
    {
      "epoch": 0.03130078853909589,
      "grad_norm": 2.0478925704956055,
      "learning_rate": 0.00019999987833918285,
      "loss": 1.2484,
      "step": 65
    },
    {
      "epoch": 0.03178233913200506,
      "grad_norm": 2.359440565109253,
      "learning_rate": 0.00019999951335702735,
      "loss": 1.5186,
      "step": 66
    },
    {
      "epoch": 0.03226388972491422,
      "grad_norm": 1.8413147926330566,
      "learning_rate": 0.00019999890505442158,
      "loss": 1.3288,
      "step": 67
    },
    {
      "epoch": 0.03274544031782339,
      "grad_norm": 1.396903395652771,
      "learning_rate": 0.0001999980534328457,
      "loss": 1.543,
      "step": 68
    },
    {
      "epoch": 0.03322699091073256,
      "grad_norm": 2.111729145050049,
      "learning_rate": 0.0001999969584943719,
      "loss": 2.0209,
      "step": 69
    },
    {
      "epoch": 0.03370854150364173,
      "grad_norm": 3.1551921367645264,
      "learning_rate": 0.00019999562024166438,
      "loss": 1.6679,
      "step": 70
    },
    {
      "epoch": 0.03419009209655089,
      "grad_norm": 1.3164373636245728,
      "learning_rate": 0.0001999940386779794,
      "loss": 1.2949,
      "step": 71
    },
    {
      "epoch": 0.03467164268946006,
      "grad_norm": 1.5184919834136963,
      "learning_rate": 0.00019999221380716527,
      "loss": 1.286,
      "step": 72
    },
    {
      "epoch": 0.03515319328236923,
      "grad_norm": 1.5143961906433105,
      "learning_rate": 0.00019999014563366226,
      "loss": 1.5927,
      "step": 73
    },
    {
      "epoch": 0.035634743875278395,
      "grad_norm": 0.9917196035385132,
      "learning_rate": 0.00019998783416250268,
      "loss": 1.2419,
      "step": 74
    },
    {
      "epoch": 0.036116294468187565,
      "grad_norm": 1.1440573930740356,
      "learning_rate": 0.0001999852793993109,
      "loss": 1.632,
      "step": 75
    },
    {
      "epoch": 0.03659784506109673,
      "grad_norm": 1.8918631076812744,
      "learning_rate": 0.00019998248135030315,
      "loss": 1.3407,
      "step": 76
    },
    {
      "epoch": 0.0370793956540059,
      "grad_norm": 1.1749062538146973,
      "learning_rate": 0.00019997944002228774,
      "loss": 1.4413,
      "step": 77
    },
    {
      "epoch": 0.03756094624691507,
      "grad_norm": 1.6219439506530762,
      "learning_rate": 0.00019997615542266482,
      "loss": 1.4515,
      "step": 78
    },
    {
      "epoch": 0.03804249683982423,
      "grad_norm": 1.4354901313781738,
      "learning_rate": 0.00019997262755942655,
      "loss": 1.1741,
      "step": 79
    },
    {
      "epoch": 0.0385240474327334,
      "grad_norm": 1.9227983951568604,
      "learning_rate": 0.000199968856441157,
      "loss": 1.3935,
      "step": 80
    },
    {
      "epoch": 0.03900559802564257,
      "grad_norm": 2.165827751159668,
      "learning_rate": 0.0001999648420770321,
      "loss": 1.7958,
      "step": 81
    },
    {
      "epoch": 0.03948714861855174,
      "grad_norm": 1.2162997722625732,
      "learning_rate": 0.0001999605844768197,
      "loss": 1.6292,
      "step": 82
    },
    {
      "epoch": 0.0399686992114609,
      "grad_norm": 0.9344050884246826,
      "learning_rate": 0.00019995608365087946,
      "loss": 1.3032,
      "step": 83
    },
    {
      "epoch": 0.04045024980437007,
      "grad_norm": 1.0530850887298584,
      "learning_rate": 0.0001999513396101628,
      "loss": 1.2898,
      "step": 84
    },
    {
      "epoch": 0.04093180039727924,
      "grad_norm": 1.737715482711792,
      "learning_rate": 0.00019994635236621306,
      "loss": 1.1906,
      "step": 85
    },
    {
      "epoch": 0.041413350990188405,
      "grad_norm": 1.2918546199798584,
      "learning_rate": 0.00019994112193116528,
      "loss": 1.3475,
      "step": 86
    },
    {
      "epoch": 0.041894901583097575,
      "grad_norm": 1.413856863975525,
      "learning_rate": 0.00019993564831774618,
      "loss": 1.2764,
      "step": 87
    },
    {
      "epoch": 0.04237645217600674,
      "grad_norm": 1.4456284046173096,
      "learning_rate": 0.00019992993153927432,
      "loss": 1.5086,
      "step": 88
    },
    {
      "epoch": 0.04285800276891591,
      "grad_norm": 2.3464932441711426,
      "learning_rate": 0.00019992397160965982,
      "loss": 1.7321,
      "step": 89
    },
    {
      "epoch": 0.04333955336182508,
      "grad_norm": 1.3598451614379883,
      "learning_rate": 0.0001999177685434045,
      "loss": 1.7024,
      "step": 90
    },
    {
      "epoch": 0.04382110395473424,
      "grad_norm": 1.1417745351791382,
      "learning_rate": 0.00019991132235560176,
      "loss": 1.5829,
      "step": 91
    },
    {
      "epoch": 0.044302654547643414,
      "grad_norm": 1.409085988998413,
      "learning_rate": 0.00019990463306193652,
      "loss": 1.5313,
      "step": 92
    },
    {
      "epoch": 0.04478420514055258,
      "grad_norm": 1.3054836988449097,
      "learning_rate": 0.00019989770067868533,
      "loss": 1.2681,
      "step": 93
    },
    {
      "epoch": 0.04526575573346175,
      "grad_norm": 1.6800811290740967,
      "learning_rate": 0.00019989052522271622,
      "loss": 1.3444,
      "step": 94
    },
    {
      "epoch": 0.04574730632637091,
      "grad_norm": 1.502502202987671,
      "learning_rate": 0.00019988310671148848,
      "loss": 1.144,
      "step": 95
    },
    {
      "epoch": 0.04622885691928008,
      "grad_norm": 2.256404399871826,
      "learning_rate": 0.00019987544516305311,
      "loss": 1.6504,
      "step": 96
    },
    {
      "epoch": 0.04671040751218925,
      "grad_norm": 1.3222041130065918,
      "learning_rate": 0.00019986754059605222,
      "loss": 1.4119,
      "step": 97
    },
    {
      "epoch": 0.047191958105098415,
      "grad_norm": 1.1852208375930786,
      "learning_rate": 0.00019985939302971938,
      "loss": 1.2823,
      "step": 98
    },
    {
      "epoch": 0.047673508698007586,
      "grad_norm": 1.2655526399612427,
      "learning_rate": 0.00019985100248387933,
      "loss": 1.4103,
      "step": 99
    },
    {
      "epoch": 0.04815505929091675,
      "grad_norm": 1.9978147745132446,
      "learning_rate": 0.00019984236897894816,
      "loss": 1.2369,
      "step": 100
    },
    {
      "epoch": 0.04863660988382592,
      "grad_norm": 2.176079034805298,
      "learning_rate": 0.000199833492535933,
      "loss": 1.5571,
      "step": 101
    },
    {
      "epoch": 0.04911816047673509,
      "grad_norm": 1.9925180673599243,
      "learning_rate": 0.00019982437317643217,
      "loss": 1.5547,
      "step": 102
    },
    {
      "epoch": 0.04959971106964425,
      "grad_norm": 2.701740264892578,
      "learning_rate": 0.00019981501092263503,
      "loss": 2.0181,
      "step": 103
    },
    {
      "epoch": 0.050081261662553424,
      "grad_norm": 2.0044503211975098,
      "learning_rate": 0.00019980540579732196,
      "loss": 1.5476,
      "step": 104
    },
    {
      "epoch": 0.05056281225546259,
      "grad_norm": 1.22849702835083,
      "learning_rate": 0.00019979555782386434,
      "loss": 1.3006,
      "step": 105
    },
    {
      "epoch": 0.05104436284837176,
      "grad_norm": 1.1254589557647705,
      "learning_rate": 0.00019978546702622443,
      "loss": 1.3877,
      "step": 106
    },
    {
      "epoch": 0.05152591344128093,
      "grad_norm": 1.9789526462554932,
      "learning_rate": 0.00019977513342895532,
      "loss": 1.4244,
      "step": 107
    },
    {
      "epoch": 0.05200746403419009,
      "grad_norm": 2.0888466835021973,
      "learning_rate": 0.00019976455705720083,
      "loss": 1.4544,
      "step": 108
    },
    {
      "epoch": 0.05248901462709926,
      "grad_norm": 1.677490234375,
      "learning_rate": 0.0001997537379366956,
      "loss": 1.7511,
      "step": 109
    },
    {
      "epoch": 0.052970565220008425,
      "grad_norm": 1.4213123321533203,
      "learning_rate": 0.00019974267609376494,
      "loss": 1.5702,
      "step": 110
    },
    {
      "epoch": 0.053452115812917596,
      "grad_norm": 1.4285441637039185,
      "learning_rate": 0.00019973137155532462,
      "loss": 1.4252,
      "step": 111
    },
    {
      "epoch": 0.05393366640582676,
      "grad_norm": 3.3786818981170654,
      "learning_rate": 0.00019971982434888107,
      "loss": 1.5166,
      "step": 112
    },
    {
      "epoch": 0.05441521699873593,
      "grad_norm": 1.7309807538986206,
      "learning_rate": 0.00019970803450253114,
      "loss": 1.3693,
      "step": 113
    },
    {
      "epoch": 0.0548967675916451,
      "grad_norm": 2.7726359367370605,
      "learning_rate": 0.0001996960020449621,
      "loss": 1.3855,
      "step": 114
    },
    {
      "epoch": 0.055378318184554264,
      "grad_norm": 0.9826974868774414,
      "learning_rate": 0.00019968372700545145,
      "loss": 1.656,
      "step": 115
    },
    {
      "epoch": 0.055859868777463434,
      "grad_norm": 0.8170602917671204,
      "learning_rate": 0.00019967120941386709,
      "loss": 1.5735,
      "step": 116
    },
    {
      "epoch": 0.0563414193703726,
      "grad_norm": 1.3170169591903687,
      "learning_rate": 0.000199658449300667,
      "loss": 1.3864,
      "step": 117
    },
    {
      "epoch": 0.05682296996328177,
      "grad_norm": 3.2949843406677246,
      "learning_rate": 0.00019964544669689928,
      "loss": 2.1717,
      "step": 118
    },
    {
      "epoch": 0.05730452055619094,
      "grad_norm": 1.2253683805465698,
      "learning_rate": 0.00019963220163420214,
      "loss": 1.8908,
      "step": 119
    },
    {
      "epoch": 0.0577860711491001,
      "grad_norm": 2.148245334625244,
      "learning_rate": 0.0001996187141448036,
      "loss": 1.1364,
      "step": 120
    },
    {
      "epoch": 0.05826762174200927,
      "grad_norm": 2.253753662109375,
      "learning_rate": 0.0001996049842615217,
      "loss": 1.2176,
      "step": 121
    },
    {
      "epoch": 0.058749172334918436,
      "grad_norm": 1.47792649269104,
      "learning_rate": 0.0001995910120177642,
      "loss": 1.2374,
      "step": 122
    },
    {
      "epoch": 0.059230722927827606,
      "grad_norm": 1.4016532897949219,
      "learning_rate": 0.00019957679744752859,
      "loss": 1.5259,
      "step": 123
    },
    {
      "epoch": 0.05971227352073677,
      "grad_norm": 1.594745397567749,
      "learning_rate": 0.00019956234058540195,
      "loss": 1.389,
      "step": 124
    },
    {
      "epoch": 0.06019382411364594,
      "grad_norm": 1.389815330505371,
      "learning_rate": 0.00019954764146656105,
      "loss": 0.941,
      "step": 125
    },
    {
      "epoch": 0.06067537470655511,
      "grad_norm": 1.7466856241226196,
      "learning_rate": 0.00019953270012677195,
      "loss": 1.6117,
      "step": 126
    },
    {
      "epoch": 0.061156925299464274,
      "grad_norm": 1.6798468828201294,
      "learning_rate": 0.00019951751660239015,
      "loss": 1.5634,
      "step": 127
    },
    {
      "epoch": 0.061638475892373444,
      "grad_norm": 1.611095905303955,
      "learning_rate": 0.00019950209093036052,
      "loss": 1.2549,
      "step": 128
    },
    {
      "epoch": 0.06212002648528261,
      "grad_norm": 1.0821895599365234,
      "learning_rate": 0.000199486423148217,
      "loss": 1.2712,
      "step": 129
    },
    {
      "epoch": 0.06260157707819178,
      "grad_norm": 1.008300542831421,
      "learning_rate": 0.00019947051329408276,
      "loss": 1.5113,
      "step": 130
    },
    {
      "epoch": 0.06308312767110094,
      "grad_norm": 0.7770178318023682,
      "learning_rate": 0.00019945436140666981,
      "loss": 1.4952,
      "step": 131
    },
    {
      "epoch": 0.06356467826401012,
      "grad_norm": 1.9329686164855957,
      "learning_rate": 0.0001994379675252793,
      "loss": 1.6034,
      "step": 132
    },
    {
      "epoch": 0.06404622885691928,
      "grad_norm": 1.346755862236023,
      "learning_rate": 0.00019942133168980103,
      "loss": 1.6609,
      "step": 133
    },
    {
      "epoch": 0.06452777944982845,
      "grad_norm": 1.096861481666565,
      "learning_rate": 0.00019940445394071355,
      "loss": 1.275,
      "step": 134
    },
    {
      "epoch": 0.06500933004273761,
      "grad_norm": 1.6034793853759766,
      "learning_rate": 0.0001993873343190842,
      "loss": 1.3262,
      "step": 135
    },
    {
      "epoch": 0.06549088063564679,
      "grad_norm": 1.4972381591796875,
      "learning_rate": 0.00019936997286656855,
      "loss": 1.6647,
      "step": 136
    },
    {
      "epoch": 0.06597243122855595,
      "grad_norm": 2.2823967933654785,
      "learning_rate": 0.00019935236962541092,
      "loss": 1.3443,
      "step": 137
    },
    {
      "epoch": 0.06645398182146511,
      "grad_norm": 1.4359623193740845,
      "learning_rate": 0.00019933452463844376,
      "loss": 1.4611,
      "step": 138
    },
    {
      "epoch": 0.06693553241437429,
      "grad_norm": 2.834228277206421,
      "learning_rate": 0.00019931643794908772,
      "loss": 1.0862,
      "step": 139
    },
    {
      "epoch": 0.06741708300728345,
      "grad_norm": 2.1683802604675293,
      "learning_rate": 0.00019929810960135172,
      "loss": 1.6587,
      "step": 140
    },
    {
      "epoch": 0.06789863360019262,
      "grad_norm": 1.7411953210830688,
      "learning_rate": 0.00019927953963983254,
      "loss": 1.4921,
      "step": 141
    },
    {
      "epoch": 0.06838018419310178,
      "grad_norm": 2.3963048458099365,
      "learning_rate": 0.00019926072810971492,
      "loss": 1.5549,
      "step": 142
    },
    {
      "epoch": 0.06886173478601096,
      "grad_norm": 1.7836371660232544,
      "learning_rate": 0.00019924167505677137,
      "loss": 1.6211,
      "step": 143
    },
    {
      "epoch": 0.06934328537892012,
      "grad_norm": 2.0127811431884766,
      "learning_rate": 0.00019922238052736215,
      "loss": 1.4484,
      "step": 144
    },
    {
      "epoch": 0.06982483597182929,
      "grad_norm": 2.0997157096862793,
      "learning_rate": 0.00019920284456843498,
      "loss": 1.7334,
      "step": 145
    },
    {
      "epoch": 0.07030638656473846,
      "grad_norm": 2.0593738555908203,
      "learning_rate": 0.00019918306722752505,
      "loss": 1.6926,
      "step": 146
    },
    {
      "epoch": 0.07078793715764763,
      "grad_norm": 2.1995582580566406,
      "learning_rate": 0.00019916304855275497,
      "loss": 1.8752,
      "step": 147
    },
    {
      "epoch": 0.07126948775055679,
      "grad_norm": 2.3022937774658203,
      "learning_rate": 0.00019914278859283445,
      "loss": 1.8061,
      "step": 148
    },
    {
      "epoch": 0.07175103834346597,
      "grad_norm": 1.742922306060791,
      "learning_rate": 0.0001991222873970604,
      "loss": 1.2207,
      "step": 149
    },
    {
      "epoch": 0.07223258893637513,
      "grad_norm": 2.0846827030181885,
      "learning_rate": 0.00019910154501531663,
      "loss": 1.1688,
      "step": 150
    },
    {
      "epoch": 0.0727141395292843,
      "grad_norm": 2.504732131958008,
      "learning_rate": 0.0001990805614980739,
      "loss": 1.7447,
      "step": 151
    },
    {
      "epoch": 0.07319569012219346,
      "grad_norm": 1.5753812789916992,
      "learning_rate": 0.00019905933689638955,
      "loss": 1.4032,
      "step": 152
    },
    {
      "epoch": 0.07367724071510263,
      "grad_norm": 0.8398808836936951,
      "learning_rate": 0.00019903787126190772,
      "loss": 1.1857,
      "step": 153
    },
    {
      "epoch": 0.0741587913080118,
      "grad_norm": 2.746708869934082,
      "learning_rate": 0.00019901616464685888,
      "loss": 1.6677,
      "step": 154
    },
    {
      "epoch": 0.07464034190092096,
      "grad_norm": 1.5562852621078491,
      "learning_rate": 0.00019899421710405996,
      "loss": 1.4773,
      "step": 155
    },
    {
      "epoch": 0.07512189249383014,
      "grad_norm": 1.3233758211135864,
      "learning_rate": 0.00019897202868691407,
      "loss": 1.4254,
      "step": 156
    },
    {
      "epoch": 0.0756034430867393,
      "grad_norm": 1.8775464296340942,
      "learning_rate": 0.00019894959944941038,
      "loss": 0.8469,
      "step": 157
    },
    {
      "epoch": 0.07608499367964847,
      "grad_norm": 1.7663127183914185,
      "learning_rate": 0.0001989269294461242,
      "loss": 1.0307,
      "step": 158
    },
    {
      "epoch": 0.07656654427255763,
      "grad_norm": 1.3111730813980103,
      "learning_rate": 0.0001989040187322164,
      "loss": 1.2632,
      "step": 159
    },
    {
      "epoch": 0.0770480948654668,
      "grad_norm": 1.4297072887420654,
      "learning_rate": 0.00019888086736343384,
      "loss": 1.8317,
      "step": 160
    },
    {
      "epoch": 0.07752964545837597,
      "grad_norm": 1.8723260164260864,
      "learning_rate": 0.0001988574753961087,
      "loss": 1.3548,
      "step": 161
    },
    {
      "epoch": 0.07801119605128513,
      "grad_norm": 1.8751100301742554,
      "learning_rate": 0.00019883384288715874,
      "loss": 1.4901,
      "step": 162
    },
    {
      "epoch": 0.07849274664419431,
      "grad_norm": 2.228203535079956,
      "learning_rate": 0.000198809969894087,
      "loss": 1.5213,
      "step": 163
    },
    {
      "epoch": 0.07897429723710347,
      "grad_norm": 2.620147228240967,
      "learning_rate": 0.0001987858564749816,
      "loss": 1.6994,
      "step": 164
    },
    {
      "epoch": 0.07945584783001264,
      "grad_norm": 0.8661990761756897,
      "learning_rate": 0.00019876150268851572,
      "loss": 1.2071,
      "step": 165
    },
    {
      "epoch": 0.0799373984229218,
      "grad_norm": 1.0385640859603882,
      "learning_rate": 0.00019873690859394737,
      "loss": 1.1957,
      "step": 166
    },
    {
      "epoch": 0.08041894901583098,
      "grad_norm": 2.100379228591919,
      "learning_rate": 0.0001987120742511193,
      "loss": 1.4744,
      "step": 167
    },
    {
      "epoch": 0.08090049960874014,
      "grad_norm": 2.0659995079040527,
      "learning_rate": 0.0001986869997204589,
      "loss": 0.9134,
      "step": 168
    },
    {
      "epoch": 0.0813820502016493,
      "grad_norm": 0.6641594767570496,
      "learning_rate": 0.00019866168506297788,
      "loss": 1.2211,
      "step": 169
    },
    {
      "epoch": 0.08186360079455848,
      "grad_norm": 2.314115285873413,
      "learning_rate": 0.00019863613034027224,
      "loss": 1.4083,
      "step": 170
    },
    {
      "epoch": 0.08234515138746765,
      "grad_norm": 1.9841386079788208,
      "learning_rate": 0.00019861033561452223,
      "loss": 1.5408,
      "step": 171
    },
    {
      "epoch": 0.08282670198037681,
      "grad_norm": 1.283460021018982,
      "learning_rate": 0.00019858430094849195,
      "loss": 1.2584,
      "step": 172
    },
    {
      "epoch": 0.08330825257328599,
      "grad_norm": 1.5605363845825195,
      "learning_rate": 0.0001985580264055294,
      "loss": 1.118,
      "step": 173
    },
    {
      "epoch": 0.08378980316619515,
      "grad_norm": 1.6310181617736816,
      "learning_rate": 0.00019853151204956616,
      "loss": 1.2551,
      "step": 174
    },
    {
      "epoch": 0.08427135375910431,
      "grad_norm": 2.816194772720337,
      "learning_rate": 0.00019850475794511749,
      "loss": 1.0223,
      "step": 175
    },
    {
      "epoch": 0.08475290435201348,
      "grad_norm": 1.1364073753356934,
      "learning_rate": 0.00019847776415728185,
      "loss": 1.0948,
      "step": 176
    },
    {
      "epoch": 0.08523445494492266,
      "grad_norm": 1.2464489936828613,
      "learning_rate": 0.000198450530751741,
      "loss": 1.4185,
      "step": 177
    },
    {
      "epoch": 0.08571600553783182,
      "grad_norm": 2.303211212158203,
      "learning_rate": 0.00019842305779475968,
      "loss": 1.7605,
      "step": 178
    },
    {
      "epoch": 0.08619755613074098,
      "grad_norm": 1.4460458755493164,
      "learning_rate": 0.00019839534535318558,
      "loss": 1.6777,
      "step": 179
    },
    {
      "epoch": 0.08667910672365016,
      "grad_norm": 1.251957893371582,
      "learning_rate": 0.00019836739349444899,
      "loss": 1.3269,
      "step": 180
    },
    {
      "epoch": 0.08716065731655932,
      "grad_norm": 1.4379816055297852,
      "learning_rate": 0.00019833920228656292,
      "loss": 1.1803,
      "step": 181
    },
    {
      "epoch": 0.08764220790946849,
      "grad_norm": 2.7593319416046143,
      "learning_rate": 0.0001983107717981226,
      "loss": 1.6815,
      "step": 182
    },
    {
      "epoch": 0.08812375850237765,
      "grad_norm": 1.2778346538543701,
      "learning_rate": 0.00019828210209830562,
      "loss": 1.1673,
      "step": 183
    },
    {
      "epoch": 0.08860530909528683,
      "grad_norm": 2.7648282051086426,
      "learning_rate": 0.00019825319325687154,
      "loss": 1.3672,
      "step": 184
    },
    {
      "epoch": 0.08908685968819599,
      "grad_norm": 1.6513874530792236,
      "learning_rate": 0.00019822404534416182,
      "loss": 1.7943,
      "step": 185
    },
    {
      "epoch": 0.08956841028110515,
      "grad_norm": 1.6982673406600952,
      "learning_rate": 0.00019819465843109963,
      "loss": 1.4816,
      "step": 186
    },
    {
      "epoch": 0.09004996087401433,
      "grad_norm": 2.23966908454895,
      "learning_rate": 0.00019816503258918969,
      "loss": 1.0362,
      "step": 187
    },
    {
      "epoch": 0.0905315114669235,
      "grad_norm": 1.636163353919983,
      "learning_rate": 0.00019813516789051808,
      "loss": 1.4224,
      "step": 188
    },
    {
      "epoch": 0.09101306205983266,
      "grad_norm": 1.8070262670516968,
      "learning_rate": 0.0001981050644077521,
      "loss": 1.5916,
      "step": 189
    },
    {
      "epoch": 0.09149461265274182,
      "grad_norm": 1.440511703491211,
      "learning_rate": 0.00019807472221414002,
      "loss": 1.4477,
      "step": 190
    },
    {
      "epoch": 0.091976163245651,
      "grad_norm": 1.704687237739563,
      "learning_rate": 0.00019804414138351094,
      "loss": 1.3935,
      "step": 191
    },
    {
      "epoch": 0.09245771383856016,
      "grad_norm": 1.2213839292526245,
      "learning_rate": 0.00019801332199027467,
      "loss": 1.4296,
      "step": 192
    },
    {
      "epoch": 0.09293926443146933,
      "grad_norm": 1.3330293893814087,
      "learning_rate": 0.00019798226410942146,
      "loss": 1.3181,
      "step": 193
    },
    {
      "epoch": 0.0934208150243785,
      "grad_norm": 1.2607433795928955,
      "learning_rate": 0.00019795096781652182,
      "loss": 1.7328,
      "step": 194
    },
    {
      "epoch": 0.09390236561728767,
      "grad_norm": 0.6895573139190674,
      "learning_rate": 0.00019791943318772643,
      "loss": 1.3479,
      "step": 195
    },
    {
      "epoch": 0.09438391621019683,
      "grad_norm": 2.196470022201538,
      "learning_rate": 0.00019788766029976587,
      "loss": 1.5081,
      "step": 196
    },
    {
      "epoch": 0.09486546680310601,
      "grad_norm": 1.9372203350067139,
      "learning_rate": 0.0001978556492299504,
      "loss": 1.1146,
      "step": 197
    },
    {
      "epoch": 0.09534701739601517,
      "grad_norm": 1.4261103868484497,
      "learning_rate": 0.00019782340005616996,
      "loss": 1.3746,
      "step": 198
    },
    {
      "epoch": 0.09582856798892433,
      "grad_norm": 1.8656367063522339,
      "learning_rate": 0.0001977909128568937,
      "loss": 1.2829,
      "step": 199
    },
    {
      "epoch": 0.0963101185818335,
      "grad_norm": 3.095597505569458,
      "learning_rate": 0.00019775818771117,
      "loss": 1.2225,
      "step": 200
    },
    {
      "epoch": 0.09679166917474268,
      "grad_norm": 1.4401271343231201,
      "learning_rate": 0.00019772522469862626,
      "loss": 1.1153,
      "step": 201
    },
    {
      "epoch": 0.09727321976765184,
      "grad_norm": 2.299915313720703,
      "learning_rate": 0.00019769202389946863,
      "loss": 1.6403,
      "step": 202
    },
    {
      "epoch": 0.097754770360561,
      "grad_norm": 1.844218134880066,
      "learning_rate": 0.0001976585853944818,
      "loss": 1.3915,
      "step": 203
    },
    {
      "epoch": 0.09823632095347018,
      "grad_norm": 1.3112562894821167,
      "learning_rate": 0.0001976249092650289,
      "loss": 1.3927,
      "step": 204
    },
    {
      "epoch": 0.09871787154637934,
      "grad_norm": 0.8005386590957642,
      "learning_rate": 0.00019759099559305124,
      "loss": 0.8893,
      "step": 205
    },
    {
      "epoch": 0.0991994221392885,
      "grad_norm": 0.8863512873649597,
      "learning_rate": 0.00019755684446106812,
      "loss": 1.1857,
      "step": 206
    },
    {
      "epoch": 0.09968097273219767,
      "grad_norm": 1.8438727855682373,
      "learning_rate": 0.00019752245595217662,
      "loss": 1.8601,
      "step": 207
    },
    {
      "epoch": 0.10016252332510685,
      "grad_norm": 1.6223257780075073,
      "learning_rate": 0.00019748783015005144,
      "loss": 1.3328,
      "step": 208
    },
    {
      "epoch": 0.10064407391801601,
      "grad_norm": 2.6211280822753906,
      "learning_rate": 0.00019745296713894465,
      "loss": 1.5416,
      "step": 209
    },
    {
      "epoch": 0.10112562451092517,
      "grad_norm": 1.1900383234024048,
      "learning_rate": 0.00019741786700368548,
      "loss": 0.977,
      "step": 210
    },
    {
      "epoch": 0.10160717510383435,
      "grad_norm": 2.4423089027404785,
      "learning_rate": 0.00019738252982968017,
      "loss": 1.3045,
      "step": 211
    },
    {
      "epoch": 0.10208872569674352,
      "grad_norm": 0.6833189129829407,
      "learning_rate": 0.00019734695570291168,
      "loss": 1.0356,
      "step": 212
    },
    {
      "epoch": 0.10257027628965268,
      "grad_norm": 2.2875657081604004,
      "learning_rate": 0.00019731114470993962,
      "loss": 1.8398,
      "step": 213
    },
    {
      "epoch": 0.10305182688256186,
      "grad_norm": 2.4047629833221436,
      "learning_rate": 0.0001972750969378998,
      "loss": 1.5066,
      "step": 214
    },
    {
      "epoch": 0.10353337747547102,
      "grad_norm": 1.9457683563232422,
      "learning_rate": 0.00019723881247450434,
      "loss": 1.0699,
      "step": 215
    },
    {
      "epoch": 0.10401492806838018,
      "grad_norm": 2.3161990642547607,
      "learning_rate": 0.0001972022914080411,
      "loss": 1.7914,
      "step": 216
    },
    {
      "epoch": 0.10449647866128935,
      "grad_norm": 2.2028820514678955,
      "learning_rate": 0.00019716553382737379,
      "loss": 1.4118,
      "step": 217
    },
    {
      "epoch": 0.10497802925419852,
      "grad_norm": 1.3063920736312866,
      "learning_rate": 0.00019712853982194152,
      "loss": 1.1901,
      "step": 218
    },
    {
      "epoch": 0.10545957984710769,
      "grad_norm": 2.71840238571167,
      "learning_rate": 0.00019709130948175876,
      "loss": 1.7944,
      "step": 219
    },
    {
      "epoch": 0.10594113044001685,
      "grad_norm": 0.7813544273376465,
      "learning_rate": 0.0001970538428974149,
      "loss": 1.1626,
      "step": 220
    },
    {
      "epoch": 0.10642268103292603,
      "grad_norm": 1.2536900043487549,
      "learning_rate": 0.00019701614016007436,
      "loss": 1.2761,
      "step": 221
    },
    {
      "epoch": 0.10690423162583519,
      "grad_norm": 2.9715793132781982,
      "learning_rate": 0.00019697820136147597,
      "loss": 1.3795,
      "step": 222
    },
    {
      "epoch": 0.10738578221874436,
      "grad_norm": 1.4083433151245117,
      "learning_rate": 0.00019694002659393305,
      "loss": 1.4559,
      "step": 223
    },
    {
      "epoch": 0.10786733281165352,
      "grad_norm": 2.055025100708008,
      "learning_rate": 0.0001969016159503331,
      "loss": 1.6453,
      "step": 224
    },
    {
      "epoch": 0.1083488834045627,
      "grad_norm": 3.2367589473724365,
      "learning_rate": 0.00019686296952413747,
      "loss": 1.11,
      "step": 225
    },
    {
      "epoch": 0.10883043399747186,
      "grad_norm": 1.7441691160202026,
      "learning_rate": 0.0001968240874093813,
      "loss": 1.15,
      "step": 226
    },
    {
      "epoch": 0.10931198459038102,
      "grad_norm": 1.1862682104110718,
      "learning_rate": 0.00019678496970067325,
      "loss": 1.2117,
      "step": 227
    },
    {
      "epoch": 0.1097935351832902,
      "grad_norm": 1.411576509475708,
      "learning_rate": 0.0001967456164931951,
      "loss": 1.1615,
      "step": 228
    },
    {
      "epoch": 0.11027508577619936,
      "grad_norm": 2.325787305831909,
      "learning_rate": 0.0001967060278827017,
      "loss": 1.7814,
      "step": 229
    },
    {
      "epoch": 0.11075663636910853,
      "grad_norm": 2.1329219341278076,
      "learning_rate": 0.00019666620396552076,
      "loss": 1.3673,
      "step": 230
    },
    {
      "epoch": 0.11123818696201769,
      "grad_norm": 1.6730871200561523,
      "learning_rate": 0.00019662614483855246,
      "loss": 1.6315,
      "step": 231
    },
    {
      "epoch": 0.11171973755492687,
      "grad_norm": 1.6149928569793701,
      "learning_rate": 0.00019658585059926934,
      "loss": 1.0816,
      "step": 232
    },
    {
      "epoch": 0.11220128814783603,
      "grad_norm": 1.2461655139923096,
      "learning_rate": 0.00019654532134571594,
      "loss": 1.2521,
      "step": 233
    },
    {
      "epoch": 0.1126828387407452,
      "grad_norm": 2.457401990890503,
      "learning_rate": 0.00019650455717650878,
      "loss": 1.1355,
      "step": 234
    },
    {
      "epoch": 0.11316438933365437,
      "grad_norm": 0.984399139881134,
      "learning_rate": 0.00019646355819083589,
      "loss": 1.0924,
      "step": 235
    },
    {
      "epoch": 0.11364593992656354,
      "grad_norm": 2.223020553588867,
      "learning_rate": 0.0001964223244884566,
      "loss": 1.179,
      "step": 236
    },
    {
      "epoch": 0.1141274905194727,
      "grad_norm": 2.634448766708374,
      "learning_rate": 0.00019638085616970153,
      "loss": 1.2024,
      "step": 237
    },
    {
      "epoch": 0.11460904111238188,
      "grad_norm": 1.9457464218139648,
      "learning_rate": 0.00019633915333547202,
      "loss": 1.2586,
      "step": 238
    },
    {
      "epoch": 0.11509059170529104,
      "grad_norm": 2.107306480407715,
      "learning_rate": 0.00019629721608724004,
      "loss": 1.2853,
      "step": 239
    },
    {
      "epoch": 0.1155721422982002,
      "grad_norm": 1.8233542442321777,
      "learning_rate": 0.0001962550445270481,
      "loss": 0.9115,
      "step": 240
    },
    {
      "epoch": 0.11605369289110937,
      "grad_norm": 1.7748448848724365,
      "learning_rate": 0.00019621263875750864,
      "loss": 1.3391,
      "step": 241
    },
    {
      "epoch": 0.11653524348401854,
      "grad_norm": 2.8353323936462402,
      "learning_rate": 0.00019616999888180406,
      "loss": 1.2732,
      "step": 242
    },
    {
      "epoch": 0.11701679407692771,
      "grad_norm": 1.657248854637146,
      "learning_rate": 0.0001961271250036865,
      "loss": 1.6226,
      "step": 243
    },
    {
      "epoch": 0.11749834466983687,
      "grad_norm": 1.8947606086730957,
      "learning_rate": 0.0001960840172274773,
      "loss": 1.3641,
      "step": 244
    },
    {
      "epoch": 0.11797989526274605,
      "grad_norm": 1.4128596782684326,
      "learning_rate": 0.00019604067565806704,
      "loss": 1.4558,
      "step": 245
    },
    {
      "epoch": 0.11846144585565521,
      "grad_norm": 1.196951985359192,
      "learning_rate": 0.00019599710040091512,
      "loss": 1.0899,
      "step": 246
    },
    {
      "epoch": 0.11894299644856438,
      "grad_norm": 2.581648349761963,
      "learning_rate": 0.00019595329156204955,
      "loss": 1.4638,
      "step": 247
    },
    {
      "epoch": 0.11942454704147354,
      "grad_norm": 3.649137258529663,
      "learning_rate": 0.00019590924924806676,
      "loss": 0.8052,
      "step": 248
    },
    {
      "epoch": 0.11990609763438272,
      "grad_norm": 1.5362485647201538,
      "learning_rate": 0.0001958649735661312,
      "loss": 1.3701,
      "step": 249
    },
    {
      "epoch": 0.12038764822729188,
      "grad_norm": 1.5539209842681885,
      "learning_rate": 0.00019582046462397515,
      "loss": 1.3271,
      "step": 250
    },
    {
      "epoch": 0.12086919882020104,
      "grad_norm": 1.6628351211547852,
      "learning_rate": 0.00019577572252989854,
      "loss": 1.2949,
      "step": 251
    },
    {
      "epoch": 0.12135074941311022,
      "grad_norm": 0.8579310774803162,
      "learning_rate": 0.00019573074739276858,
      "loss": 1.4562,
      "step": 252
    },
    {
      "epoch": 0.12183230000601938,
      "grad_norm": 1.9081571102142334,
      "learning_rate": 0.00019568553932201947,
      "loss": 1.0064,
      "step": 253
    },
    {
      "epoch": 0.12231385059892855,
      "grad_norm": 2.423394203186035,
      "learning_rate": 0.00019564009842765225,
      "loss": 1.5552,
      "step": 254
    },
    {
      "epoch": 0.12279540119183771,
      "grad_norm": 2.139652729034424,
      "learning_rate": 0.00019559442482023444,
      "loss": 1.222,
      "step": 255
    },
    {
      "epoch": 0.12327695178474689,
      "grad_norm": 1.2584847211837769,
      "learning_rate": 0.0001955485186108998,
      "loss": 1.0226,
      "step": 256
    },
    {
      "epoch": 0.12375850237765605,
      "grad_norm": 2.8226137161254883,
      "learning_rate": 0.00019550237991134805,
      "loss": 1.2213,
      "step": 257
    },
    {
      "epoch": 0.12424005297056522,
      "grad_norm": 1.5066231489181519,
      "learning_rate": 0.00019545600883384467,
      "loss": 1.3896,
      "step": 258
    },
    {
      "epoch": 0.12472160356347439,
      "grad_norm": 0.5828495621681213,
      "learning_rate": 0.0001954094054912205,
      "loss": 1.1298,
      "step": 259
    },
    {
      "epoch": 0.12520315415638356,
      "grad_norm": 0.716420590877533,
      "learning_rate": 0.00019536256999687157,
      "loss": 0.8736,
      "step": 260
    },
    {
      "epoch": 0.12568470474929272,
      "grad_norm": 1.4762283563613892,
      "learning_rate": 0.00019531550246475876,
      "loss": 1.0477,
      "step": 261
    },
    {
      "epoch": 0.12616625534220188,
      "grad_norm": 1.123988151550293,
      "learning_rate": 0.00019526820300940756,
      "loss": 1.6605,
      "step": 262
    },
    {
      "epoch": 0.12664780593511105,
      "grad_norm": 1.5415560007095337,
      "learning_rate": 0.00019522067174590778,
      "loss": 1.2029,
      "step": 263
    },
    {
      "epoch": 0.12712935652802024,
      "grad_norm": 1.7547069787979126,
      "learning_rate": 0.00019517290878991324,
      "loss": 1.1864,
      "step": 264
    },
    {
      "epoch": 0.1276109071209294,
      "grad_norm": 1.0187448263168335,
      "learning_rate": 0.0001951249142576416,
      "loss": 1.4312,
      "step": 265
    },
    {
      "epoch": 0.12809245771383856,
      "grad_norm": 1.4038090705871582,
      "learning_rate": 0.00019507668826587387,
      "loss": 1.5945,
      "step": 266
    },
    {
      "epoch": 0.12857400830674773,
      "grad_norm": 1.6867750883102417,
      "learning_rate": 0.0001950282309319544,
      "loss": 1.9116,
      "step": 267
    },
    {
      "epoch": 0.1290555588996569,
      "grad_norm": 0.8677802681922913,
      "learning_rate": 0.0001949795423737903,
      "loss": 1.4652,
      "step": 268
    },
    {
      "epoch": 0.12953710949256605,
      "grad_norm": 2.929178237915039,
      "learning_rate": 0.00019493062270985144,
      "loss": 0.9914,
      "step": 269
    },
    {
      "epoch": 0.13001866008547522,
      "grad_norm": 1.3788762092590332,
      "learning_rate": 0.00019488147205916985,
      "loss": 1.1616,
      "step": 270
    },
    {
      "epoch": 0.1305002106783844,
      "grad_norm": 1.5651005506515503,
      "learning_rate": 0.00019483209054133976,
      "loss": 1.5377,
      "step": 271
    },
    {
      "epoch": 0.13098176127129357,
      "grad_norm": 1.145813226699829,
      "learning_rate": 0.00019478247827651708,
      "loss": 1.4072,
      "step": 272
    },
    {
      "epoch": 0.13146331186420274,
      "grad_norm": 1.4603357315063477,
      "learning_rate": 0.00019473263538541914,
      "loss": 1.2003,
      "step": 273
    },
    {
      "epoch": 0.1319448624571119,
      "grad_norm": 1.5310707092285156,
      "learning_rate": 0.00019468256198932455,
      "loss": 1.168,
      "step": 274
    },
    {
      "epoch": 0.13242641305002106,
      "grad_norm": 0.6564764976501465,
      "learning_rate": 0.00019463225821007268,
      "loss": 1.4478,
      "step": 275
    },
    {
      "epoch": 0.13290796364293023,
      "grad_norm": 0.8874586820602417,
      "learning_rate": 0.00019458172417006347,
      "loss": 1.0692,
      "step": 276
    },
    {
      "epoch": 0.1333895142358394,
      "grad_norm": 1.5714865922927856,
      "learning_rate": 0.00019453095999225726,
      "loss": 1.429,
      "step": 277
    },
    {
      "epoch": 0.13387106482874858,
      "grad_norm": 1.3166011571884155,
      "learning_rate": 0.0001944799658001742,
      "loss": 1.1667,
      "step": 278
    },
    {
      "epoch": 0.13435261542165775,
      "grad_norm": 2.955052375793457,
      "learning_rate": 0.00019442874171789418,
      "loss": 1.0606,
      "step": 279
    },
    {
      "epoch": 0.1348341660145669,
      "grad_norm": 1.62515389919281,
      "learning_rate": 0.00019437728787005657,
      "loss": 1.4451,
      "step": 280
    },
    {
      "epoch": 0.13531571660747607,
      "grad_norm": 2.6105737686157227,
      "learning_rate": 0.00019432560438185963,
      "loss": 1.2848,
      "step": 281
    },
    {
      "epoch": 0.13579726720038524,
      "grad_norm": 2.1958696842193604,
      "learning_rate": 0.00019427369137906046,
      "loss": 1.0708,
      "step": 282
    },
    {
      "epoch": 0.1362788177932944,
      "grad_norm": 1.3239554166793823,
      "learning_rate": 0.00019422154898797472,
      "loss": 1.2465,
      "step": 283
    },
    {
      "epoch": 0.13676036838620356,
      "grad_norm": 1.645102620124817,
      "learning_rate": 0.00019416917733547603,
      "loss": 1.5751,
      "step": 284
    },
    {
      "epoch": 0.13724191897911275,
      "grad_norm": 1.119922161102295,
      "learning_rate": 0.00019411657654899597,
      "loss": 1.0104,
      "step": 285
    },
    {
      "epoch": 0.13772346957202192,
      "grad_norm": 1.4109903573989868,
      "learning_rate": 0.0001940637467565237,
      "loss": 1.4173,
      "step": 286
    },
    {
      "epoch": 0.13820502016493108,
      "grad_norm": 2.536267042160034,
      "learning_rate": 0.00019401068808660546,
      "loss": 1.5289,
      "step": 287
    },
    {
      "epoch": 0.13868657075784024,
      "grad_norm": 0.8169746398925781,
      "learning_rate": 0.0001939574006683445,
      "loss": 1.1146,
      "step": 288
    },
    {
      "epoch": 0.1391681213507494,
      "grad_norm": 0.9298089742660522,
      "learning_rate": 0.00019390388463140065,
      "loss": 0.7842,
      "step": 289
    },
    {
      "epoch": 0.13964967194365857,
      "grad_norm": 1.8893189430236816,
      "learning_rate": 0.00019385014010598998,
      "loss": 1.3613,
      "step": 290
    },
    {
      "epoch": 0.14013122253656773,
      "grad_norm": 1.2792352437973022,
      "learning_rate": 0.00019379616722288456,
      "loss": 0.9388,
      "step": 291
    },
    {
      "epoch": 0.14061277312947693,
      "grad_norm": 1.207482933998108,
      "learning_rate": 0.0001937419661134121,
      "loss": 1.2978,
      "step": 292
    },
    {
      "epoch": 0.1410943237223861,
      "grad_norm": 2.150141477584839,
      "learning_rate": 0.0001936875369094556,
      "loss": 1.5217,
      "step": 293
    },
    {
      "epoch": 0.14157587431529525,
      "grad_norm": 2.091521978378296,
      "learning_rate": 0.0001936328797434531,
      "loss": 1.1804,
      "step": 294
    },
    {
      "epoch": 0.14205742490820442,
      "grad_norm": 1.6191956996917725,
      "learning_rate": 0.00019357799474839735,
      "loss": 1.3867,
      "step": 295
    },
    {
      "epoch": 0.14253897550111358,
      "grad_norm": 1.475780725479126,
      "learning_rate": 0.00019352288205783536,
      "loss": 1.3459,
      "step": 296
    },
    {
      "epoch": 0.14302052609402274,
      "grad_norm": 1.8596059083938599,
      "learning_rate": 0.00019346754180586825,
      "loss": 0.9942,
      "step": 297
    },
    {
      "epoch": 0.14350207668693193,
      "grad_norm": 0.9591652154922485,
      "learning_rate": 0.00019341197412715082,
      "loss": 1.1648,
      "step": 298
    },
    {
      "epoch": 0.1439836272798411,
      "grad_norm": 1.2492796182632446,
      "learning_rate": 0.00019335617915689128,
      "loss": 1.3298,
      "step": 299
    },
    {
      "epoch": 0.14446517787275026,
      "grad_norm": 1.6927003860473633,
      "learning_rate": 0.00019330015703085082,
      "loss": 1.0156,
      "step": 300
    },
    {
      "epoch": 0.14494672846565942,
      "grad_norm": 1.809677243232727,
      "learning_rate": 0.00019324390788534343,
      "loss": 1.1279,
      "step": 301
    },
    {
      "epoch": 0.1454282790585686,
      "grad_norm": 0.9998087286949158,
      "learning_rate": 0.00019318743185723546,
      "loss": 1.2789,
      "step": 302
    },
    {
      "epoch": 0.14590982965147775,
      "grad_norm": 1.3441334962844849,
      "learning_rate": 0.00019313072908394525,
      "loss": 1.1476,
      "step": 303
    },
    {
      "epoch": 0.14639138024438691,
      "grad_norm": 1.176350474357605,
      "learning_rate": 0.00019307379970344294,
      "loss": 1.2676,
      "step": 304
    },
    {
      "epoch": 0.1468729308372961,
      "grad_norm": 1.308995008468628,
      "learning_rate": 0.00019301664385425004,
      "loss": 1.3287,
      "step": 305
    },
    {
      "epoch": 0.14735448143020527,
      "grad_norm": 1.0787750482559204,
      "learning_rate": 0.0001929592616754391,
      "loss": 1.1157,
      "step": 306
    },
    {
      "epoch": 0.14783603202311443,
      "grad_norm": 0.5602009296417236,
      "learning_rate": 0.00019290165330663336,
      "loss": 1.1641,
      "step": 307
    },
    {
      "epoch": 0.1483175826160236,
      "grad_norm": 1.4494296312332153,
      "learning_rate": 0.00019284381888800647,
      "loss": 1.3251,
      "step": 308
    },
    {
      "epoch": 0.14879913320893276,
      "grad_norm": 0.8033807277679443,
      "learning_rate": 0.00019278575856028206,
      "loss": 1.3246,
      "step": 309
    },
    {
      "epoch": 0.14928068380184192,
      "grad_norm": 0.9002365469932556,
      "learning_rate": 0.00019272747246473345,
      "loss": 0.7962,
      "step": 310
    },
    {
      "epoch": 0.1497622343947511,
      "grad_norm": 1.3145807981491089,
      "learning_rate": 0.00019266896074318334,
      "loss": 1.3565,
      "step": 311
    },
    {
      "epoch": 0.15024378498766028,
      "grad_norm": 1.9951491355895996,
      "learning_rate": 0.00019261022353800344,
      "loss": 1.4079,
      "step": 312
    },
    {
      "epoch": 0.15072533558056944,
      "grad_norm": 1.8006443977355957,
      "learning_rate": 0.00019255126099211402,
      "loss": 1.3602,
      "step": 313
    },
    {
      "epoch": 0.1512068861734786,
      "grad_norm": 1.0033605098724365,
      "learning_rate": 0.00019249207324898376,
      "loss": 1.0878,
      "step": 314
    },
    {
      "epoch": 0.15168843676638777,
      "grad_norm": 1.7852751016616821,
      "learning_rate": 0.0001924326604526292,
      "loss": 1.4196,
      "step": 315
    },
    {
      "epoch": 0.15216998735929693,
      "grad_norm": 1.5227525234222412,
      "learning_rate": 0.00019237302274761458,
      "loss": 1.1626,
      "step": 316
    },
    {
      "epoch": 0.1526515379522061,
      "grad_norm": 2.461813449859619,
      "learning_rate": 0.0001923131602790513,
      "loss": 1.4762,
      "step": 317
    },
    {
      "epoch": 0.15313308854511526,
      "grad_norm": 1.1452112197875977,
      "learning_rate": 0.00019225307319259768,
      "loss": 1.3646,
      "step": 318
    },
    {
      "epoch": 0.15361463913802445,
      "grad_norm": 2.5520575046539307,
      "learning_rate": 0.00019219276163445862,
      "loss": 1.4539,
      "step": 319
    },
    {
      "epoch": 0.1540961897309336,
      "grad_norm": 2.118633270263672,
      "learning_rate": 0.00019213222575138522,
      "loss": 1.56,
      "step": 320
    },
    {
      "epoch": 0.15457774032384278,
      "grad_norm": 1.2929282188415527,
      "learning_rate": 0.00019207146569067435,
      "loss": 0.8781,
      "step": 321
    },
    {
      "epoch": 0.15505929091675194,
      "grad_norm": 1.9068535566329956,
      "learning_rate": 0.00019201048160016838,
      "loss": 1.4251,
      "step": 322
    },
    {
      "epoch": 0.1555408415096611,
      "grad_norm": 0.9579245448112488,
      "learning_rate": 0.00019194927362825478,
      "loss": 1.2247,
      "step": 323
    },
    {
      "epoch": 0.15602239210257027,
      "grad_norm": 1.3820081949234009,
      "learning_rate": 0.00019188784192386587,
      "loss": 1.4917,
      "step": 324
    },
    {
      "epoch": 0.15650394269547943,
      "grad_norm": 2.171107769012451,
      "learning_rate": 0.00019182618663647817,
      "loss": 1.2849,
      "step": 325
    },
    {
      "epoch": 0.15698549328838862,
      "grad_norm": 2.2618470191955566,
      "learning_rate": 0.0001917643079161124,
      "loss": 1.3257,
      "step": 326
    },
    {
      "epoch": 0.15746704388129779,
      "grad_norm": 1.7617461681365967,
      "learning_rate": 0.00019170220591333283,
      "loss": 1.1713,
      "step": 327
    },
    {
      "epoch": 0.15794859447420695,
      "grad_norm": 1.572052001953125,
      "learning_rate": 0.00019163988077924713,
      "loss": 1.4365,
      "step": 328
    },
    {
      "epoch": 0.1584301450671161,
      "grad_norm": 1.9649395942687988,
      "learning_rate": 0.00019157733266550575,
      "loss": 1.0763,
      "step": 329
    },
    {
      "epoch": 0.15891169566002528,
      "grad_norm": 3.548682451248169,
      "learning_rate": 0.00019151456172430183,
      "loss": 1.5421,
      "step": 330
    },
    {
      "epoch": 0.15939324625293444,
      "grad_norm": 2.105952501296997,
      "learning_rate": 0.0001914515681083707,
      "loss": 1.4401,
      "step": 331
    },
    {
      "epoch": 0.1598747968458436,
      "grad_norm": 1.830772042274475,
      "learning_rate": 0.00019138835197098937,
      "loss": 1.2361,
      "step": 332
    },
    {
      "epoch": 0.1603563474387528,
      "grad_norm": 1.0721756219863892,
      "learning_rate": 0.00019132491346597643,
      "loss": 1.4875,
      "step": 333
    },
    {
      "epoch": 0.16083789803166196,
      "grad_norm": 2.1168439388275146,
      "learning_rate": 0.00019126125274769145,
      "loss": 1.1663,
      "step": 334
    },
    {
      "epoch": 0.16131944862457112,
      "grad_norm": 1.5975714921951294,
      "learning_rate": 0.00019119736997103476,
      "loss": 1.2608,
      "step": 335
    },
    {
      "epoch": 0.16180099921748028,
      "grad_norm": 0.5098963379859924,
      "learning_rate": 0.000191133265291447,
      "loss": 0.9824,
      "step": 336
    },
    {
      "epoch": 0.16228254981038945,
      "grad_norm": 1.3563380241394043,
      "learning_rate": 0.00019106893886490864,
      "loss": 1.4697,
      "step": 337
    },
    {
      "epoch": 0.1627641004032986,
      "grad_norm": 0.9947191476821899,
      "learning_rate": 0.00019100439084793989,
      "loss": 0.9167,
      "step": 338
    },
    {
      "epoch": 0.1632456509962078,
      "grad_norm": 2.5751492977142334,
      "learning_rate": 0.00019093962139759998,
      "loss": 1.6361,
      "step": 339
    },
    {
      "epoch": 0.16372720158911697,
      "grad_norm": 3.3359997272491455,
      "learning_rate": 0.000190874630671487,
      "loss": 1.503,
      "step": 340
    },
    {
      "epoch": 0.16420875218202613,
      "grad_norm": 0.9778684973716736,
      "learning_rate": 0.00019080941882773745,
      "loss": 1.3715,
      "step": 341
    },
    {
      "epoch": 0.1646903027749353,
      "grad_norm": 1.7242743968963623,
      "learning_rate": 0.00019074398602502584,
      "loss": 1.2782,
      "step": 342
    },
    {
      "epoch": 0.16517185336784446,
      "grad_norm": 1.7806469202041626,
      "learning_rate": 0.00019067833242256442,
      "loss": 1.5256,
      "step": 343
    },
    {
      "epoch": 0.16565340396075362,
      "grad_norm": 1.5550919771194458,
      "learning_rate": 0.0001906124581801025,
      "loss": 1.135,
      "step": 344
    },
    {
      "epoch": 0.16613495455366278,
      "grad_norm": 0.9480149745941162,
      "learning_rate": 0.0001905463634579264,
      "loss": 1.0763,
      "step": 345
    },
    {
      "epoch": 0.16661650514657197,
      "grad_norm": 2.467151641845703,
      "learning_rate": 0.00019048004841685888,
      "loss": 1.2849,
      "step": 346
    },
    {
      "epoch": 0.16709805573948114,
      "grad_norm": 8.16342544555664,
      "learning_rate": 0.00019041351321825883,
      "loss": 1.4824,
      "step": 347
    },
    {
      "epoch": 0.1675796063323903,
      "grad_norm": 2.2539398670196533,
      "learning_rate": 0.00019034675802402068,
      "loss": 1.5309,
      "step": 348
    },
    {
      "epoch": 0.16806115692529947,
      "grad_norm": 1.8528321981430054,
      "learning_rate": 0.00019027978299657436,
      "loss": 1.1862,
      "step": 349
    },
    {
      "epoch": 0.16854270751820863,
      "grad_norm": 1.8632662296295166,
      "learning_rate": 0.00019021258829888456,
      "loss": 1.7341,
      "step": 350
    },
    {
      "epoch": 0.1690242581111178,
      "grad_norm": 1.3469961881637573,
      "learning_rate": 0.00019014517409445052,
      "loss": 0.974,
      "step": 351
    },
    {
      "epoch": 0.16950580870402696,
      "grad_norm": 1.7672408819198608,
      "learning_rate": 0.00019007754054730554,
      "loss": 1.0796,
      "step": 352
    },
    {
      "epoch": 0.16998735929693615,
      "grad_norm": 1.5639548301696777,
      "learning_rate": 0.00019000968782201675,
      "loss": 1.1014,
      "step": 353
    },
    {
      "epoch": 0.1704689098898453,
      "grad_norm": 1.76442289352417,
      "learning_rate": 0.00018994161608368448,
      "loss": 1.6556,
      "step": 354
    },
    {
      "epoch": 0.17095046048275447,
      "grad_norm": 0.6001653671264648,
      "learning_rate": 0.00018987332549794196,
      "loss": 1.2077,
      "step": 355
    },
    {
      "epoch": 0.17143201107566364,
      "grad_norm": 1.3223201036453247,
      "learning_rate": 0.00018980481623095502,
      "loss": 1.1557,
      "step": 356
    },
    {
      "epoch": 0.1719135616685728,
      "grad_norm": 1.4411089420318604,
      "learning_rate": 0.00018973608844942148,
      "loss": 1.3644,
      "step": 357
    },
    {
      "epoch": 0.17239511226148196,
      "grad_norm": 1.0517265796661377,
      "learning_rate": 0.00018966714232057094,
      "loss": 1.2351,
      "step": 358
    },
    {
      "epoch": 0.17287666285439113,
      "grad_norm": 1.7375134229660034,
      "learning_rate": 0.00018959797801216418,
      "loss": 0.9886,
      "step": 359
    },
    {
      "epoch": 0.17335821344730032,
      "grad_norm": 1.6675549745559692,
      "learning_rate": 0.000189528595692493,
      "loss": 1.339,
      "step": 360
    },
    {
      "epoch": 0.17383976404020948,
      "grad_norm": 1.1284401416778564,
      "learning_rate": 0.00018945899553037956,
      "loss": 1.285,
      "step": 361
    },
    {
      "epoch": 0.17432131463311865,
      "grad_norm": 1.2890712022781372,
      "learning_rate": 0.00018938917769517613,
      "loss": 1.3916,
      "step": 362
    },
    {
      "epoch": 0.1748028652260278,
      "grad_norm": 1.4297925233840942,
      "learning_rate": 0.00018931914235676458,
      "loss": 1.3602,
      "step": 363
    },
    {
      "epoch": 0.17528441581893697,
      "grad_norm": 1.136730432510376,
      "learning_rate": 0.00018924888968555606,
      "loss": 0.8125,
      "step": 364
    },
    {
      "epoch": 0.17576596641184614,
      "grad_norm": 1.2735693454742432,
      "learning_rate": 0.00018917841985249055,
      "loss": 0.958,
      "step": 365
    },
    {
      "epoch": 0.1762475170047553,
      "grad_norm": 1.1079814434051514,
      "learning_rate": 0.0001891077330290363,
      "loss": 1.4098,
      "step": 366
    },
    {
      "epoch": 0.1767290675976645,
      "grad_norm": 1.557845950126648,
      "learning_rate": 0.00018903682938718977,
      "loss": 1.2902,
      "step": 367
    },
    {
      "epoch": 0.17721061819057365,
      "grad_norm": 2.2619524002075195,
      "learning_rate": 0.00018896570909947475,
      "loss": 1.2851,
      "step": 368
    },
    {
      "epoch": 0.17769216878348282,
      "grad_norm": 1.4451303482055664,
      "learning_rate": 0.00018889437233894234,
      "loss": 1.3328,
      "step": 369
    },
    {
      "epoch": 0.17817371937639198,
      "grad_norm": 1.1534463167190552,
      "learning_rate": 0.0001888228192791703,
      "loss": 1.3053,
      "step": 370
    },
    {
      "epoch": 0.17865526996930114,
      "grad_norm": 1.153280258178711,
      "learning_rate": 0.00018875105009426272,
      "loss": 1.3284,
      "step": 371
    },
    {
      "epoch": 0.1791368205622103,
      "grad_norm": 0.9033501148223877,
      "learning_rate": 0.00018867906495884955,
      "loss": 1.4315,
      "step": 372
    },
    {
      "epoch": 0.17961837115511947,
      "grad_norm": 2.6750900745391846,
      "learning_rate": 0.0001886068640480862,
      "loss": 1.1591,
      "step": 373
    },
    {
      "epoch": 0.18009992174802866,
      "grad_norm": 1.9415152072906494,
      "learning_rate": 0.00018853444753765306,
      "loss": 1.439,
      "step": 374
    },
    {
      "epoch": 0.18058147234093783,
      "grad_norm": 0.8949999213218689,
      "learning_rate": 0.00018846181560375525,
      "loss": 0.8605,
      "step": 375
    },
    {
      "epoch": 0.181063022933847,
      "grad_norm": 0.8251290321350098,
      "learning_rate": 0.0001883889684231219,
      "loss": 1.0652,
      "step": 376
    },
    {
      "epoch": 0.18154457352675615,
      "grad_norm": 1.2823021411895752,
      "learning_rate": 0.000188315906173006,
      "loss": 1.8293,
      "step": 377
    },
    {
      "epoch": 0.18202612411966532,
      "grad_norm": 1.1994600296020508,
      "learning_rate": 0.0001882426290311838,
      "loss": 1.1472,
      "step": 378
    },
    {
      "epoch": 0.18250767471257448,
      "grad_norm": 0.9393148422241211,
      "learning_rate": 0.00018816913717595445,
      "loss": 1.3257,
      "step": 379
    },
    {
      "epoch": 0.18298922530548364,
      "grad_norm": 1.737230658531189,
      "learning_rate": 0.00018809543078613953,
      "loss": 1.2843,
      "step": 380
    },
    {
      "epoch": 0.18347077589839283,
      "grad_norm": 0.8644424676895142,
      "learning_rate": 0.00018802151004108263,
      "loss": 0.9129,
      "step": 381
    },
    {
      "epoch": 0.183952326491302,
      "grad_norm": 1.8019263744354248,
      "learning_rate": 0.0001879473751206489,
      "loss": 1.1896,
      "step": 382
    },
    {
      "epoch": 0.18443387708421116,
      "grad_norm": 1.6992765665054321,
      "learning_rate": 0.00018787302620522467,
      "loss": 1.0729,
      "step": 383
    },
    {
      "epoch": 0.18491542767712033,
      "grad_norm": 1.2879563570022583,
      "learning_rate": 0.00018779846347571693,
      "loss": 1.1775,
      "step": 384
    },
    {
      "epoch": 0.1853969782700295,
      "grad_norm": 1.333025574684143,
      "learning_rate": 0.0001877236871135529,
      "loss": 0.9081,
      "step": 385
    },
    {
      "epoch": 0.18587852886293865,
      "grad_norm": 0.9439510107040405,
      "learning_rate": 0.00018764869730067968,
      "loss": 1.4562,
      "step": 386
    },
    {
      "epoch": 0.18636007945584784,
      "grad_norm": 1.3587722778320312,
      "learning_rate": 0.0001875734942195637,
      "loss": 0.8845,
      "step": 387
    },
    {
      "epoch": 0.186841630048757,
      "grad_norm": 1.4092299938201904,
      "learning_rate": 0.0001874980780531903,
      "loss": 1.8205,
      "step": 388
    },
    {
      "epoch": 0.18732318064166617,
      "grad_norm": 1.2788656949996948,
      "learning_rate": 0.00018742244898506337,
      "loss": 1.2593,
      "step": 389
    },
    {
      "epoch": 0.18780473123457533,
      "grad_norm": 1.3854421377182007,
      "learning_rate": 0.00018734660719920475,
      "loss": 1.2295,
      "step": 390
    },
    {
      "epoch": 0.1882862818274845,
      "grad_norm": 1.4532678127288818,
      "learning_rate": 0.00018727055288015397,
      "loss": 1.3778,
      "step": 391
    },
    {
      "epoch": 0.18876783242039366,
      "grad_norm": 0.7126154899597168,
      "learning_rate": 0.00018719428621296764,
      "loss": 1.3168,
      "step": 392
    },
    {
      "epoch": 0.18924938301330282,
      "grad_norm": 1.6068065166473389,
      "learning_rate": 0.00018711780738321897,
      "loss": 1.061,
      "step": 393
    },
    {
      "epoch": 0.18973093360621202,
      "grad_norm": 1.7989546060562134,
      "learning_rate": 0.00018704111657699758,
      "loss": 1.221,
      "step": 394
    },
    {
      "epoch": 0.19021248419912118,
      "grad_norm": 1.7175345420837402,
      "learning_rate": 0.0001869642139809088,
      "loss": 1.3736,
      "step": 395
    },
    {
      "epoch": 0.19069403479203034,
      "grad_norm": 1.4424655437469482,
      "learning_rate": 0.00018688709978207323,
      "loss": 1.5438,
      "step": 396
    },
    {
      "epoch": 0.1911755853849395,
      "grad_norm": 0.7891624569892883,
      "learning_rate": 0.00018680977416812644,
      "loss": 1.0074,
      "step": 397
    },
    {
      "epoch": 0.19165713597784867,
      "grad_norm": 1.9073477983474731,
      "learning_rate": 0.00018673223732721837,
      "loss": 1.1765,
      "step": 398
    },
    {
      "epoch": 0.19213868657075783,
      "grad_norm": 1.959021806716919,
      "learning_rate": 0.0001866544894480129,
      "loss": 1.3599,
      "step": 399
    },
    {
      "epoch": 0.192620237163667,
      "grad_norm": 1.6006906032562256,
      "learning_rate": 0.00018657653071968747,
      "loss": 1.4927,
      "step": 400
    },
    {
      "epoch": 0.1931017877565762,
      "grad_norm": 1.4201229810714722,
      "learning_rate": 0.00018649836133193253,
      "loss": 1.2108,
      "step": 401
    },
    {
      "epoch": 0.19358333834948535,
      "grad_norm": 2.104936361312866,
      "learning_rate": 0.00018641998147495112,
      "loss": 1.0064,
      "step": 402
    },
    {
      "epoch": 0.19406488894239451,
      "grad_norm": 1.7348823547363281,
      "learning_rate": 0.00018634139133945837,
      "loss": 1.0396,
      "step": 403
    },
    {
      "epoch": 0.19454643953530368,
      "grad_norm": 1.6066287755966187,
      "learning_rate": 0.00018626259111668105,
      "loss": 1.6476,
      "step": 404
    },
    {
      "epoch": 0.19502799012821284,
      "grad_norm": 1.272810697555542,
      "learning_rate": 0.00018618358099835723,
      "loss": 1.4367,
      "step": 405
    },
    {
      "epoch": 0.195509540721122,
      "grad_norm": 1.6426247358322144,
      "learning_rate": 0.00018610436117673555,
      "loss": 0.7425,
      "step": 406
    },
    {
      "epoch": 0.19599109131403117,
      "grad_norm": 1.360694169998169,
      "learning_rate": 0.00018602493184457505,
      "loss": 1.0851,
      "step": 407
    },
    {
      "epoch": 0.19647264190694036,
      "grad_norm": 1.737608790397644,
      "learning_rate": 0.00018594529319514437,
      "loss": 1.4584,
      "step": 408
    },
    {
      "epoch": 0.19695419249984952,
      "grad_norm": 1.073427438735962,
      "learning_rate": 0.00018586544542222169,
      "loss": 1.1295,
      "step": 409
    },
    {
      "epoch": 0.1974357430927587,
      "grad_norm": 1.130391240119934,
      "learning_rate": 0.00018578538872009384,
      "loss": 0.8658,
      "step": 410
    },
    {
      "epoch": 0.19791729368566785,
      "grad_norm": 0.6659079790115356,
      "learning_rate": 0.00018570512328355612,
      "loss": 0.9663,
      "step": 411
    },
    {
      "epoch": 0.198398844278577,
      "grad_norm": 1.8340193033218384,
      "learning_rate": 0.00018562464930791167,
      "loss": 1.038,
      "step": 412
    },
    {
      "epoch": 0.19888039487148618,
      "grad_norm": 2.2872612476348877,
      "learning_rate": 0.00018554396698897116,
      "loss": 0.999,
      "step": 413
    },
    {
      "epoch": 0.19936194546439534,
      "grad_norm": 0.9699216485023499,
      "learning_rate": 0.00018546307652305205,
      "loss": 1.254,
      "step": 414
    },
    {
      "epoch": 0.19984349605730453,
      "grad_norm": 1.193132996559143,
      "learning_rate": 0.00018538197810697842,
      "loss": 1.2155,
      "step": 415
    },
    {
      "epoch": 0.2003250466502137,
      "grad_norm": 1.7791582345962524,
      "learning_rate": 0.0001853006719380802,
      "loss": 1.1523,
      "step": 416
    },
    {
      "epoch": 0.20080659724312286,
      "grad_norm": 1.6898586750030518,
      "learning_rate": 0.00018521915821419284,
      "loss": 1.1484,
      "step": 417
    },
    {
      "epoch": 0.20128814783603202,
      "grad_norm": 0.758802056312561,
      "learning_rate": 0.00018513743713365698,
      "loss": 1.3679,
      "step": 418
    },
    {
      "epoch": 0.20176969842894119,
      "grad_norm": 2.3143417835235596,
      "learning_rate": 0.00018505550889531765,
      "loss": 1.4876,
      "step": 419
    },
    {
      "epoch": 0.20225124902185035,
      "grad_norm": 1.3419922590255737,
      "learning_rate": 0.00018497337369852395,
      "loss": 0.8663,
      "step": 420
    },
    {
      "epoch": 0.2027327996147595,
      "grad_norm": 1.7213921546936035,
      "learning_rate": 0.0001848910317431286,
      "loss": 1.6447,
      "step": 421
    },
    {
      "epoch": 0.2032143502076687,
      "grad_norm": 2.2190473079681396,
      "learning_rate": 0.00018480848322948739,
      "loss": 1.168,
      "step": 422
    },
    {
      "epoch": 0.20369590080057787,
      "grad_norm": 1.6743611097335815,
      "learning_rate": 0.00018472572835845873,
      "loss": 1.1955,
      "step": 423
    },
    {
      "epoch": 0.20417745139348703,
      "grad_norm": 3.2296359539031982,
      "learning_rate": 0.00018464276733140306,
      "loss": 1.1597,
      "step": 424
    },
    {
      "epoch": 0.2046590019863962,
      "grad_norm": 0.8592111468315125,
      "learning_rate": 0.0001845596003501826,
      "loss": 0.9538,
      "step": 425
    },
    {
      "epoch": 0.20514055257930536,
      "grad_norm": 0.7619894742965698,
      "learning_rate": 0.00018447622761716057,
      "loss": 1.2269,
      "step": 426
    },
    {
      "epoch": 0.20562210317221452,
      "grad_norm": 0.813829243183136,
      "learning_rate": 0.00018439264933520084,
      "loss": 1.2499,
      "step": 427
    },
    {
      "epoch": 0.2061036537651237,
      "grad_norm": 0.8501059412956238,
      "learning_rate": 0.00018430886570766747,
      "loss": 1.0844,
      "step": 428
    },
    {
      "epoch": 0.20658520435803288,
      "grad_norm": 1.3430989980697632,
      "learning_rate": 0.0001842248769384242,
      "loss": 1.8183,
      "step": 429
    },
    {
      "epoch": 0.20706675495094204,
      "grad_norm": 1.250504493713379,
      "learning_rate": 0.00018414068323183375,
      "loss": 0.9612,
      "step": 430
    },
    {
      "epoch": 0.2075483055438512,
      "grad_norm": 1.2324836254119873,
      "learning_rate": 0.00018405628479275775,
      "loss": 1.2303,
      "step": 431
    },
    {
      "epoch": 0.20802985613676037,
      "grad_norm": 1.29920494556427,
      "learning_rate": 0.00018397168182655583,
      "loss": 0.9853,
      "step": 432
    },
    {
      "epoch": 0.20851140672966953,
      "grad_norm": 1.2412772178649902,
      "learning_rate": 0.00018388687453908527,
      "loss": 1.0461,
      "step": 433
    },
    {
      "epoch": 0.2089929573225787,
      "grad_norm": 2.0380032062530518,
      "learning_rate": 0.00018380186313670058,
      "loss": 1.1744,
      "step": 434
    },
    {
      "epoch": 0.20947450791548788,
      "grad_norm": 1.1109199523925781,
      "learning_rate": 0.00018371664782625287,
      "loss": 0.9245,
      "step": 435
    },
    {
      "epoch": 0.20995605850839705,
      "grad_norm": 1.783227562904358,
      "learning_rate": 0.00018363122881508945,
      "loss": 1.4156,
      "step": 436
    },
    {
      "epoch": 0.2104376091013062,
      "grad_norm": 0.9671676754951477,
      "learning_rate": 0.00018354560631105328,
      "loss": 0.8876,
      "step": 437
    },
    {
      "epoch": 0.21091915969421537,
      "grad_norm": 1.526063323020935,
      "learning_rate": 0.00018345978052248233,
      "loss": 1.6168,
      "step": 438
    },
    {
      "epoch": 0.21140071028712454,
      "grad_norm": 0.7514840364456177,
      "learning_rate": 0.00018337375165820944,
      "loss": 0.9397,
      "step": 439
    },
    {
      "epoch": 0.2118822608800337,
      "grad_norm": 1.8357489109039307,
      "learning_rate": 0.00018328751992756137,
      "loss": 1.4963,
      "step": 440
    },
    {
      "epoch": 0.21236381147294286,
      "grad_norm": 2.761262893676758,
      "learning_rate": 0.0001832010855403586,
      "loss": 2.489,
      "step": 441
    },
    {
      "epoch": 0.21284536206585206,
      "grad_norm": 2.585672616958618,
      "learning_rate": 0.0001831144487069147,
      "loss": 0.998,
      "step": 442
    },
    {
      "epoch": 0.21332691265876122,
      "grad_norm": 1.094600796699524,
      "learning_rate": 0.0001830276096380358,
      "loss": 0.8667,
      "step": 443
    },
    {
      "epoch": 0.21380846325167038,
      "grad_norm": 1.8089797496795654,
      "learning_rate": 0.0001829405685450202,
      "loss": 1.6646,
      "step": 444
    },
    {
      "epoch": 0.21429001384457955,
      "grad_norm": 1.6599122285842896,
      "learning_rate": 0.00018285332563965765,
      "loss": 1.6531,
      "step": 445
    },
    {
      "epoch": 0.2147715644374887,
      "grad_norm": 2.1350276470184326,
      "learning_rate": 0.00018276588113422905,
      "loss": 1.219,
      "step": 446
    },
    {
      "epoch": 0.21525311503039787,
      "grad_norm": 1.2398370504379272,
      "learning_rate": 0.00018267823524150575,
      "loss": 1.2472,
      "step": 447
    },
    {
      "epoch": 0.21573466562330704,
      "grad_norm": 0.89753657579422,
      "learning_rate": 0.00018259038817474923,
      "loss": 1.4149,
      "step": 448
    },
    {
      "epoch": 0.21621621621621623,
      "grad_norm": 1.2794777154922485,
      "learning_rate": 0.0001825023401477104,
      "loss": 1.1855,
      "step": 449
    },
    {
      "epoch": 0.2166977668091254,
      "grad_norm": 1.944532871246338,
      "learning_rate": 0.0001824140913746291,
      "loss": 1.2993,
      "step": 450
    },
    {
      "epoch": 0.21717931740203456,
      "grad_norm": 1.2633885145187378,
      "learning_rate": 0.00018232564207023376,
      "loss": 0.9817,
      "step": 451
    },
    {
      "epoch": 0.21766086799494372,
      "grad_norm": 2.389735460281372,
      "learning_rate": 0.00018223699244974064,
      "loss": 1.3314,
      "step": 452
    },
    {
      "epoch": 0.21814241858785288,
      "grad_norm": 1.193429946899414,
      "learning_rate": 0.00018214814272885343,
      "loss": 1.3308,
      "step": 453
    },
    {
      "epoch": 0.21862396918076205,
      "grad_norm": 1.5840109586715698,
      "learning_rate": 0.00018205909312376276,
      "loss": 1.1668,
      "step": 454
    },
    {
      "epoch": 0.2191055197736712,
      "grad_norm": 1.4147722721099854,
      "learning_rate": 0.00018196984385114554,
      "loss": 1.3492,
      "step": 455
    },
    {
      "epoch": 0.2195870703665804,
      "grad_norm": 1.5289056301116943,
      "learning_rate": 0.0001818803951281646,
      "loss": 1.3076,
      "step": 456
    },
    {
      "epoch": 0.22006862095948956,
      "grad_norm": 1.3569334745407104,
      "learning_rate": 0.000181790747172468,
      "loss": 1.5055,
      "step": 457
    },
    {
      "epoch": 0.22055017155239873,
      "grad_norm": 0.5415627360343933,
      "learning_rate": 0.00018170090020218864,
      "loss": 1.1199,
      "step": 458
    },
    {
      "epoch": 0.2210317221453079,
      "grad_norm": 1.5818837881088257,
      "learning_rate": 0.00018161085443594365,
      "loss": 1.2897,
      "step": 459
    },
    {
      "epoch": 0.22151327273821705,
      "grad_norm": 1.1016331911087036,
      "learning_rate": 0.00018152061009283382,
      "loss": 0.9238,
      "step": 460
    },
    {
      "epoch": 0.22199482333112622,
      "grad_norm": 1.6640774011611938,
      "learning_rate": 0.00018143016739244314,
      "loss": 1.3789,
      "step": 461
    },
    {
      "epoch": 0.22247637392403538,
      "grad_norm": 1.9907211065292358,
      "learning_rate": 0.0001813395265548383,
      "loss": 1.1554,
      "step": 462
    },
    {
      "epoch": 0.22295792451694457,
      "grad_norm": 1.6290398836135864,
      "learning_rate": 0.00018124868780056814,
      "loss": 1.2628,
      "step": 463
    },
    {
      "epoch": 0.22343947510985374,
      "grad_norm": 1.5546035766601562,
      "learning_rate": 0.0001811576513506629,
      "loss": 0.8918,
      "step": 464
    },
    {
      "epoch": 0.2239210257027629,
      "grad_norm": 1.7053402662277222,
      "learning_rate": 0.00018106641742663397,
      "loss": 1.2701,
      "step": 465
    },
    {
      "epoch": 0.22440257629567206,
      "grad_norm": 2.6603705883026123,
      "learning_rate": 0.00018097498625047328,
      "loss": 1.129,
      "step": 466
    },
    {
      "epoch": 0.22488412688858123,
      "grad_norm": 1.4323021173477173,
      "learning_rate": 0.00018088335804465258,
      "loss": 1.0348,
      "step": 467
    },
    {
      "epoch": 0.2253656774814904,
      "grad_norm": 1.66669762134552,
      "learning_rate": 0.00018079153303212318,
      "loss": 1.3259,
      "step": 468
    },
    {
      "epoch": 0.22584722807439955,
      "grad_norm": 1.3757411241531372,
      "learning_rate": 0.0001806995114363152,
      "loss": 1.7636,
      "step": 469
    },
    {
      "epoch": 0.22632877866730874,
      "grad_norm": 1.1295281648635864,
      "learning_rate": 0.00018060729348113707,
      "loss": 0.8691,
      "step": 470
    },
    {
      "epoch": 0.2268103292602179,
      "grad_norm": 1.3690543174743652,
      "learning_rate": 0.00018051487939097505,
      "loss": 1.2209,
      "step": 471
    },
    {
      "epoch": 0.22729187985312707,
      "grad_norm": 1.4880307912826538,
      "learning_rate": 0.00018042226939069255,
      "loss": 1.6853,
      "step": 472
    },
    {
      "epoch": 0.22777343044603623,
      "grad_norm": 1.2337520122528076,
      "learning_rate": 0.00018032946370562982,
      "loss": 1.2583,
      "step": 473
    },
    {
      "epoch": 0.2282549810389454,
      "grad_norm": 1.9700994491577148,
      "learning_rate": 0.00018023646256160313,
      "loss": 1.0173,
      "step": 474
    },
    {
      "epoch": 0.22873653163185456,
      "grad_norm": 2.6149399280548096,
      "learning_rate": 0.00018014326618490437,
      "loss": 0.9468,
      "step": 475
    },
    {
      "epoch": 0.22921808222476375,
      "grad_norm": 1.9999854564666748,
      "learning_rate": 0.0001800498748023005,
      "loss": 1.7995,
      "step": 476
    },
    {
      "epoch": 0.22969963281767292,
      "grad_norm": 1.2951595783233643,
      "learning_rate": 0.000179956288641033,
      "loss": 1.0281,
      "step": 477
    },
    {
      "epoch": 0.23018118341058208,
      "grad_norm": 1.4535390138626099,
      "learning_rate": 0.00017986250792881718,
      "loss": 0.9053,
      "step": 478
    },
    {
      "epoch": 0.23066273400349124,
      "grad_norm": 0.9770629405975342,
      "learning_rate": 0.00017976853289384184,
      "loss": 1.2191,
      "step": 479
    },
    {
      "epoch": 0.2311442845964004,
      "grad_norm": 0.9181470274925232,
      "learning_rate": 0.00017967436376476855,
      "loss": 1.3481,
      "step": 480
    },
    {
      "epoch": 0.23162583518930957,
      "grad_norm": 1.375268578529358,
      "learning_rate": 0.0001795800007707312,
      "loss": 1.0954,
      "step": 481
    },
    {
      "epoch": 0.23210738578221873,
      "grad_norm": 1.6353603601455688,
      "learning_rate": 0.00017948544414133534,
      "loss": 1.0274,
      "step": 482
    },
    {
      "epoch": 0.23258893637512792,
      "grad_norm": 2.0417611598968506,
      "learning_rate": 0.00017939069410665773,
      "loss": 1.2636,
      "step": 483
    },
    {
      "epoch": 0.2330704869680371,
      "grad_norm": 1.7985786199569702,
      "learning_rate": 0.0001792957508972457,
      "loss": 1.1207,
      "step": 484
    },
    {
      "epoch": 0.23355203756094625,
      "grad_norm": 1.1584678888320923,
      "learning_rate": 0.00017920061474411658,
      "loss": 1.0429,
      "step": 485
    },
    {
      "epoch": 0.23403358815385542,
      "grad_norm": 1.4619596004486084,
      "learning_rate": 0.00017910528587875729,
      "loss": 1.1441,
      "step": 486
    },
    {
      "epoch": 0.23451513874676458,
      "grad_norm": 0.7729864120483398,
      "learning_rate": 0.00017900976453312352,
      "loss": 1.1577,
      "step": 487
    },
    {
      "epoch": 0.23499668933967374,
      "grad_norm": 1.8740636110305786,
      "learning_rate": 0.00017891405093963938,
      "loss": 1.2617,
      "step": 488
    },
    {
      "epoch": 0.2354782399325829,
      "grad_norm": 2.0744645595550537,
      "learning_rate": 0.00017881814533119675,
      "loss": 1.8693,
      "step": 489
    },
    {
      "epoch": 0.2359597905254921,
      "grad_norm": 1.6728328466415405,
      "learning_rate": 0.00017872204794115474,
      "loss": 0.9674,
      "step": 490
    },
    {
      "epoch": 0.23644134111840126,
      "grad_norm": 1.30010187625885,
      "learning_rate": 0.0001786257590033391,
      "loss": 1.0704,
      "step": 491
    },
    {
      "epoch": 0.23692289171131042,
      "grad_norm": 1.41806960105896,
      "learning_rate": 0.00017852927875204163,
      "loss": 1.4419,
      "step": 492
    },
    {
      "epoch": 0.2374044423042196,
      "grad_norm": 1.3471970558166504,
      "learning_rate": 0.00017843260742201963,
      "loss": 0.9775,
      "step": 493
    },
    {
      "epoch": 0.23788599289712875,
      "grad_norm": 0.782619833946228,
      "learning_rate": 0.00017833574524849535,
      "loss": 1.055,
      "step": 494
    },
    {
      "epoch": 0.23836754349003791,
      "grad_norm": 2.5765480995178223,
      "learning_rate": 0.00017823869246715553,
      "loss": 1.5947,
      "step": 495
    },
    {
      "epoch": 0.23884909408294708,
      "grad_norm": 1.9718865156173706,
      "learning_rate": 0.00017814144931415043,
      "loss": 1.9988,
      "step": 496
    },
    {
      "epoch": 0.23933064467585627,
      "grad_norm": 1.8504900932312012,
      "learning_rate": 0.0001780440160260938,
      "loss": 1.3754,
      "step": 497
    },
    {
      "epoch": 0.23981219526876543,
      "grad_norm": 1.888457179069519,
      "learning_rate": 0.00017794639284006184,
      "loss": 1.1669,
      "step": 498
    },
    {
      "epoch": 0.2402937458616746,
      "grad_norm": 1.13361394405365,
      "learning_rate": 0.0001778485799935929,
      "loss": 1.2563,
      "step": 499
    },
    {
      "epoch": 0.24077529645458376,
      "grad_norm": 2.8890769481658936,
      "learning_rate": 0.00017775057772468679,
      "loss": 1.9182,
      "step": 500
    },
    {
      "epoch": 0.24125684704749292,
      "grad_norm": 2.9799153804779053,
      "learning_rate": 0.00017765238627180424,
      "loss": 1.2403,
      "step": 501
    },
    {
      "epoch": 0.2417383976404021,
      "grad_norm": 1.1056352853775024,
      "learning_rate": 0.00017755400587386632,
      "loss": 0.9558,
      "step": 502
    },
    {
      "epoch": 0.24221994823331125,
      "grad_norm": 2.2133090496063232,
      "learning_rate": 0.00017745543677025378,
      "loss": 1.3524,
      "step": 503
    },
    {
      "epoch": 0.24270149882622044,
      "grad_norm": 1.418330430984497,
      "learning_rate": 0.00017735667920080661,
      "loss": 1.0706,
      "step": 504
    },
    {
      "epoch": 0.2431830494191296,
      "grad_norm": 1.187134861946106,
      "learning_rate": 0.0001772577334058233,
      "loss": 1.4233,
      "step": 505
    },
    {
      "epoch": 0.24366460001203877,
      "grad_norm": 1.218045949935913,
      "learning_rate": 0.00017715859962606043,
      "loss": 1.0973,
      "step": 506
    },
    {
      "epoch": 0.24414615060494793,
      "grad_norm": 2.3087682723999023,
      "learning_rate": 0.00017705927810273187,
      "loss": 1.1454,
      "step": 507
    },
    {
      "epoch": 0.2446277011978571,
      "grad_norm": 1.3424021005630493,
      "learning_rate": 0.00017695976907750844,
      "loss": 1.1544,
      "step": 508
    },
    {
      "epoch": 0.24510925179076626,
      "grad_norm": 1.9669620990753174,
      "learning_rate": 0.00017686007279251706,
      "loss": 0.9886,
      "step": 509
    },
    {
      "epoch": 0.24559080238367542,
      "grad_norm": 1.6271488666534424,
      "learning_rate": 0.00017676018949034045,
      "loss": 1.2154,
      "step": 510
    },
    {
      "epoch": 0.2460723529765846,
      "grad_norm": 1.0644521713256836,
      "learning_rate": 0.0001766601194140162,
      "loss": 1.2434,
      "step": 511
    },
    {
      "epoch": 0.24655390356949378,
      "grad_norm": 0.8739067316055298,
      "learning_rate": 0.0001765598628070365,
      "loss": 1.225,
      "step": 512
    },
    {
      "epoch": 0.24703545416240294,
      "grad_norm": 2.7235941886901855,
      "learning_rate": 0.00017645941991334732,
      "loss": 1.6851,
      "step": 513
    },
    {
      "epoch": 0.2475170047553121,
      "grad_norm": 1.497186303138733,
      "learning_rate": 0.00017635879097734804,
      "loss": 1.1316,
      "step": 514
    },
    {
      "epoch": 0.24799855534822127,
      "grad_norm": 1.9500449895858765,
      "learning_rate": 0.00017625797624389055,
      "loss": 1.2565,
      "step": 515
    },
    {
      "epoch": 0.24848010594113043,
      "grad_norm": 1.5029175281524658,
      "learning_rate": 0.00017615697595827897,
      "loss": 1.1193,
      "step": 516
    },
    {
      "epoch": 0.2489616565340396,
      "grad_norm": 1.4641175270080566,
      "learning_rate": 0.0001760557903662688,
      "loss": 1.3079,
      "step": 517
    },
    {
      "epoch": 0.24944320712694878,
      "grad_norm": 1.9112852811813354,
      "learning_rate": 0.00017595441971406648,
      "loss": 1.532,
      "step": 518
    },
    {
      "epoch": 0.24992475771985795,
      "grad_norm": 1.5127805471420288,
      "learning_rate": 0.00017585286424832874,
      "loss": 1.4572,
      "step": 519
    },
    {
      "epoch": 0.2504063083127671,
      "grad_norm": 1.1018222570419312,
      "learning_rate": 0.00017575112421616202,
      "loss": 1.101,
      "step": 520
    },
    {
      "epoch": 0.2508878589056763,
      "grad_norm": 2.4903337955474854,
      "learning_rate": 0.0001756491998651218,
      "loss": 1.3763,
      "step": 521
    },
    {
      "epoch": 0.25136940949858544,
      "grad_norm": 1.0974708795547485,
      "learning_rate": 0.0001755470914432121,
      "loss": 1.0181,
      "step": 522
    },
    {
      "epoch": 0.2518509600914946,
      "grad_norm": 0.9347370266914368,
      "learning_rate": 0.0001754447991988848,
      "loss": 1.4021,
      "step": 523
    },
    {
      "epoch": 0.25233251068440377,
      "grad_norm": 1.5271857976913452,
      "learning_rate": 0.00017534232338103903,
      "loss": 1.4552,
      "step": 524
    },
    {
      "epoch": 0.25281406127731293,
      "grad_norm": 1.0110981464385986,
      "learning_rate": 0.0001752396642390207,
      "loss": 1.5573,
      "step": 525
    },
    {
      "epoch": 0.2532956118702221,
      "grad_norm": 1.277449131011963,
      "learning_rate": 0.00017513682202262163,
      "loss": 1.2348,
      "step": 526
    },
    {
      "epoch": 0.25377716246313126,
      "grad_norm": 1.2912412881851196,
      "learning_rate": 0.00017503379698207918,
      "loss": 0.965,
      "step": 527
    },
    {
      "epoch": 0.2542587130560405,
      "grad_norm": 0.860765278339386,
      "learning_rate": 0.00017493058936807562,
      "loss": 0.881,
      "step": 528
    },
    {
      "epoch": 0.25474026364894964,
      "grad_norm": 1.1658167839050293,
      "learning_rate": 0.00017482719943173739,
      "loss": 1.1842,
      "step": 529
    },
    {
      "epoch": 0.2552218142418588,
      "grad_norm": 1.0810424089431763,
      "learning_rate": 0.00017472362742463455,
      "loss": 1.157,
      "step": 530
    },
    {
      "epoch": 0.25570336483476797,
      "grad_norm": 1.0483838319778442,
      "learning_rate": 0.0001746198735987802,
      "loss": 1.251,
      "step": 531
    },
    {
      "epoch": 0.25618491542767713,
      "grad_norm": 1.7141307592391968,
      "learning_rate": 0.00017451593820662988,
      "loss": 1.1206,
      "step": 532
    },
    {
      "epoch": 0.2566664660205863,
      "grad_norm": 1.4472633600234985,
      "learning_rate": 0.00017441182150108086,
      "loss": 1.1509,
      "step": 533
    },
    {
      "epoch": 0.25714801661349546,
      "grad_norm": 0.7835544347763062,
      "learning_rate": 0.0001743075237354716,
      "loss": 1.1639,
      "step": 534
    },
    {
      "epoch": 0.2576295672064046,
      "grad_norm": 0.7412325143814087,
      "learning_rate": 0.00017420304516358113,
      "loss": 0.95,
      "step": 535
    },
    {
      "epoch": 0.2581111177993138,
      "grad_norm": 1.523851990699768,
      "learning_rate": 0.00017409838603962843,
      "loss": 1.3458,
      "step": 536
    },
    {
      "epoch": 0.25859266839222295,
      "grad_norm": 0.7727850079536438,
      "learning_rate": 0.00017399354661827178,
      "loss": 1.1145,
      "step": 537
    },
    {
      "epoch": 0.2590742189851321,
      "grad_norm": 1.5049177408218384,
      "learning_rate": 0.00017388852715460819,
      "loss": 1.3394,
      "step": 538
    },
    {
      "epoch": 0.2595557695780413,
      "grad_norm": 1.2362557649612427,
      "learning_rate": 0.00017378332790417273,
      "loss": 0.9235,
      "step": 539
    },
    {
      "epoch": 0.26003732017095044,
      "grad_norm": 1.342603087425232,
      "learning_rate": 0.00017367794912293794,
      "loss": 1.2247,
      "step": 540
    },
    {
      "epoch": 0.26051887076385966,
      "grad_norm": 1.437679648399353,
      "learning_rate": 0.00017357239106731317,
      "loss": 0.9715,
      "step": 541
    },
    {
      "epoch": 0.2610004213567688,
      "grad_norm": 1.9657872915267944,
      "learning_rate": 0.00017346665399414405,
      "loss": 1.5494,
      "step": 542
    },
    {
      "epoch": 0.261481971949678,
      "grad_norm": 1.5127356052398682,
      "learning_rate": 0.00017336073816071168,
      "loss": 1.0598,
      "step": 543
    },
    {
      "epoch": 0.26196352254258715,
      "grad_norm": 2.316601276397705,
      "learning_rate": 0.00017325464382473226,
      "loss": 1.5169,
      "step": 544
    },
    {
      "epoch": 0.2624450731354963,
      "grad_norm": 1.2949203252792358,
      "learning_rate": 0.00017314837124435622,
      "loss": 1.3106,
      "step": 545
    },
    {
      "epoch": 0.2629266237284055,
      "grad_norm": 1.7486093044281006,
      "learning_rate": 0.00017304192067816782,
      "loss": 1.5645,
      "step": 546
    },
    {
      "epoch": 0.26340817432131464,
      "grad_norm": 1.6293529272079468,
      "learning_rate": 0.00017293529238518422,
      "loss": 0.9073,
      "step": 547
    },
    {
      "epoch": 0.2638897249142238,
      "grad_norm": 1.1016491651535034,
      "learning_rate": 0.0001728284866248552,
      "loss": 1.0756,
      "step": 548
    },
    {
      "epoch": 0.26437127550713296,
      "grad_norm": 1.2047386169433594,
      "learning_rate": 0.00017272150365706224,
      "loss": 0.986,
      "step": 549
    },
    {
      "epoch": 0.2648528261000421,
      "grad_norm": 2.18683123588562,
      "learning_rate": 0.00017261434374211802,
      "loss": 0.9876,
      "step": 550
    },
    {
      "epoch": 0.2653343766929513,
      "grad_norm": 1.7479463815689087,
      "learning_rate": 0.00017250700714076586,
      "loss": 1.2495,
      "step": 551
    },
    {
      "epoch": 0.26581592728586045,
      "grad_norm": 0.7549085021018982,
      "learning_rate": 0.00017239949411417888,
      "loss": 1.0753,
      "step": 552
    },
    {
      "epoch": 0.2662974778787696,
      "grad_norm": 1.1940267086029053,
      "learning_rate": 0.0001722918049239596,
      "loss": 1.4526,
      "step": 553
    },
    {
      "epoch": 0.2667790284716788,
      "grad_norm": 2.333388090133667,
      "learning_rate": 0.00017218393983213902,
      "loss": 0.8821,
      "step": 554
    },
    {
      "epoch": 0.267260579064588,
      "grad_norm": 1.5917562246322632,
      "learning_rate": 0.00017207589910117634,
      "loss": 0.8414,
      "step": 555
    },
    {
      "epoch": 0.26774212965749716,
      "grad_norm": 1.5154459476470947,
      "learning_rate": 0.00017196768299395797,
      "loss": 1.5247,
      "step": 556
    },
    {
      "epoch": 0.2682236802504063,
      "grad_norm": 1.0251904726028442,
      "learning_rate": 0.00017185929177379714,
      "loss": 0.9509,
      "step": 557
    },
    {
      "epoch": 0.2687052308433155,
      "grad_norm": 1.8484050035476685,
      "learning_rate": 0.00017175072570443312,
      "loss": 1.2973,
      "step": 558
    },
    {
      "epoch": 0.26918678143622465,
      "grad_norm": 1.5409196615219116,
      "learning_rate": 0.00017164198505003066,
      "loss": 1.0347,
      "step": 559
    },
    {
      "epoch": 0.2696683320291338,
      "grad_norm": 1.2860301733016968,
      "learning_rate": 0.0001715330700751793,
      "loss": 0.8327,
      "step": 560
    },
    {
      "epoch": 0.270149882622043,
      "grad_norm": 1.5070279836654663,
      "learning_rate": 0.00017142398104489273,
      "loss": 1.2737,
      "step": 561
    },
    {
      "epoch": 0.27063143321495214,
      "grad_norm": 1.648976445198059,
      "learning_rate": 0.00017131471822460814,
      "loss": 1.4919,
      "step": 562
    },
    {
      "epoch": 0.2711129838078613,
      "grad_norm": 1.5788755416870117,
      "learning_rate": 0.00017120528188018565,
      "loss": 0.9818,
      "step": 563
    },
    {
      "epoch": 0.27159453440077047,
      "grad_norm": 1.33204984664917,
      "learning_rate": 0.00017109567227790754,
      "loss": 0.813,
      "step": 564
    },
    {
      "epoch": 0.27207608499367963,
      "grad_norm": 0.8501080870628357,
      "learning_rate": 0.00017098588968447766,
      "loss": 1.1932,
      "step": 565
    },
    {
      "epoch": 0.2725576355865888,
      "grad_norm": 1.5493998527526855,
      "learning_rate": 0.00017087593436702084,
      "loss": 1.726,
      "step": 566
    },
    {
      "epoch": 0.27303918617949796,
      "grad_norm": 1.3421874046325684,
      "learning_rate": 0.00017076580659308222,
      "loss": 1.4743,
      "step": 567
    },
    {
      "epoch": 0.2735207367724071,
      "grad_norm": 2.4747328758239746,
      "learning_rate": 0.00017065550663062634,
      "loss": 1.0651,
      "step": 568
    },
    {
      "epoch": 0.27400228736531634,
      "grad_norm": 2.327418088912964,
      "learning_rate": 0.00017054503474803702,
      "loss": 1.4754,
      "step": 569
    },
    {
      "epoch": 0.2744838379582255,
      "grad_norm": 0.9829911589622498,
      "learning_rate": 0.00017043439121411618,
      "loss": 1.2749,
      "step": 570
    },
    {
      "epoch": 0.27496538855113467,
      "grad_norm": 1.0076383352279663,
      "learning_rate": 0.0001703235762980835,
      "loss": 1.1033,
      "step": 571
    },
    {
      "epoch": 0.27544693914404383,
      "grad_norm": 1.9351223707199097,
      "learning_rate": 0.00017021259026957567,
      "loss": 1.5342,
      "step": 572
    },
    {
      "epoch": 0.275928489736953,
      "grad_norm": 2.1808929443359375,
      "learning_rate": 0.00017010143339864562,
      "loss": 1.4291,
      "step": 573
    },
    {
      "epoch": 0.27641004032986216,
      "grad_norm": 1.5498580932617188,
      "learning_rate": 0.0001699901059557621,
      "loss": 1.4273,
      "step": 574
    },
    {
      "epoch": 0.2768915909227713,
      "grad_norm": 1.378250241279602,
      "learning_rate": 0.00016987860821180895,
      "loss": 1.3268,
      "step": 575
    },
    {
      "epoch": 0.2773731415156805,
      "grad_norm": 1.5918465852737427,
      "learning_rate": 0.00016976694043808416,
      "loss": 1.281,
      "step": 576
    },
    {
      "epoch": 0.27785469210858965,
      "grad_norm": 1.4815313816070557,
      "learning_rate": 0.00016965510290629972,
      "loss": 1.1872,
      "step": 577
    },
    {
      "epoch": 0.2783362427014988,
      "grad_norm": 0.7459321618080139,
      "learning_rate": 0.00016954309588858044,
      "loss": 0.6489,
      "step": 578
    },
    {
      "epoch": 0.278817793294408,
      "grad_norm": 1.3758541345596313,
      "learning_rate": 0.00016943091965746366,
      "loss": 0.8969,
      "step": 579
    },
    {
      "epoch": 0.27929934388731714,
      "grad_norm": 1.042833685874939,
      "learning_rate": 0.00016931857448589845,
      "loss": 1.0215,
      "step": 580
    },
    {
      "epoch": 0.2797808944802263,
      "grad_norm": 0.5810043811798096,
      "learning_rate": 0.00016920606064724488,
      "loss": 0.8358,
      "step": 581
    },
    {
      "epoch": 0.28026244507313547,
      "grad_norm": 0.9163647294044495,
      "learning_rate": 0.00016909337841527344,
      "loss": 1.0016,
      "step": 582
    },
    {
      "epoch": 0.2807439956660447,
      "grad_norm": 1.021188497543335,
      "learning_rate": 0.00016898052806416444,
      "loss": 1.0582,
      "step": 583
    },
    {
      "epoch": 0.28122554625895385,
      "grad_norm": 1.9669586420059204,
      "learning_rate": 0.00016886750986850718,
      "loss": 1.5056,
      "step": 584
    },
    {
      "epoch": 0.281707096851863,
      "grad_norm": 1.704929232597351,
      "learning_rate": 0.00016875432410329934,
      "loss": 1.0663,
      "step": 585
    },
    {
      "epoch": 0.2821886474447722,
      "grad_norm": 1.1752477884292603,
      "learning_rate": 0.0001686409710439464,
      "loss": 1.2723,
      "step": 586
    },
    {
      "epoch": 0.28267019803768134,
      "grad_norm": 0.782768964767456,
      "learning_rate": 0.00016852745096626088,
      "loss": 1.4009,
      "step": 587
    },
    {
      "epoch": 0.2831517486305905,
      "grad_norm": 0.8464993238449097,
      "learning_rate": 0.0001684137641464617,
      "loss": 1.2906,
      "step": 588
    },
    {
      "epoch": 0.28363329922349967,
      "grad_norm": 1.457601547241211,
      "learning_rate": 0.0001682999108611735,
      "loss": 1.3211,
      "step": 589
    },
    {
      "epoch": 0.28411484981640883,
      "grad_norm": 0.9123442769050598,
      "learning_rate": 0.00016818589138742587,
      "loss": 0.9508,
      "step": 590
    },
    {
      "epoch": 0.284596400409318,
      "grad_norm": 1.3921875953674316,
      "learning_rate": 0.00016807170600265296,
      "loss": 1.3178,
      "step": 591
    },
    {
      "epoch": 0.28507795100222716,
      "grad_norm": 1.8289682865142822,
      "learning_rate": 0.00016795735498469246,
      "loss": 1.2111,
      "step": 592
    },
    {
      "epoch": 0.2855595015951363,
      "grad_norm": 2.12333345413208,
      "learning_rate": 0.00016784283861178513,
      "loss": 1.1973,
      "step": 593
    },
    {
      "epoch": 0.2860410521880455,
      "grad_norm": 0.6705619692802429,
      "learning_rate": 0.00016772815716257412,
      "loss": 0.9265,
      "step": 594
    },
    {
      "epoch": 0.28652260278095465,
      "grad_norm": 1.2611973285675049,
      "learning_rate": 0.00016761331091610416,
      "loss": 1.3644,
      "step": 595
    },
    {
      "epoch": 0.28700415337386387,
      "grad_norm": 1.1917195320129395,
      "learning_rate": 0.00016749830015182107,
      "loss": 1.1029,
      "step": 596
    },
    {
      "epoch": 0.28748570396677303,
      "grad_norm": 1.5623480081558228,
      "learning_rate": 0.00016738312514957086,
      "loss": 0.7367,
      "step": 597
    },
    {
      "epoch": 0.2879672545596822,
      "grad_norm": 1.5358045101165771,
      "learning_rate": 0.00016726778618959926,
      "loss": 1.2407,
      "step": 598
    },
    {
      "epoch": 0.28844880515259136,
      "grad_norm": 0.6570262908935547,
      "learning_rate": 0.00016715228355255093,
      "loss": 0.6977,
      "step": 599
    },
    {
      "epoch": 0.2889303557455005,
      "grad_norm": 0.9557023644447327,
      "learning_rate": 0.00016703661751946874,
      "loss": 1.064,
      "step": 600
    },
    {
      "epoch": 0.2894119063384097,
      "grad_norm": 1.6340230703353882,
      "learning_rate": 0.00016692078837179318,
      "loss": 1.1006,
      "step": 601
    },
    {
      "epoch": 0.28989345693131885,
      "grad_norm": 1.4815481901168823,
      "learning_rate": 0.00016680479639136163,
      "loss": 1.1983,
      "step": 602
    },
    {
      "epoch": 0.290375007524228,
      "grad_norm": 1.0919681787490845,
      "learning_rate": 0.0001666886418604077,
      "loss": 0.9592,
      "step": 603
    },
    {
      "epoch": 0.2908565581171372,
      "grad_norm": 0.848789393901825,
      "learning_rate": 0.0001665723250615604,
      "loss": 0.7357,
      "step": 604
    },
    {
      "epoch": 0.29133810871004634,
      "grad_norm": 1.5520508289337158,
      "learning_rate": 0.00016645584627784381,
      "loss": 1.1477,
      "step": 605
    },
    {
      "epoch": 0.2918196593029555,
      "grad_norm": 1.3088622093200684,
      "learning_rate": 0.0001663392057926759,
      "loss": 1.1208,
      "step": 606
    },
    {
      "epoch": 0.29230120989586467,
      "grad_norm": 2.4230332374572754,
      "learning_rate": 0.00016622240388986824,
      "loss": 0.8402,
      "step": 607
    },
    {
      "epoch": 0.29278276048877383,
      "grad_norm": 1.4834061861038208,
      "learning_rate": 0.0001661054408536251,
      "loss": 1.5696,
      "step": 608
    },
    {
      "epoch": 0.293264311081683,
      "grad_norm": 1.9420775175094604,
      "learning_rate": 0.00016598831696854288,
      "loss": 1.4077,
      "step": 609
    },
    {
      "epoch": 0.2937458616745922,
      "grad_norm": 1.0769697427749634,
      "learning_rate": 0.00016587103251960937,
      "loss": 1.1339,
      "step": 610
    },
    {
      "epoch": 0.2942274122675014,
      "grad_norm": 1.8560692071914673,
      "learning_rate": 0.00016575358779220294,
      "loss": 1.357,
      "step": 611
    },
    {
      "epoch": 0.29470896286041054,
      "grad_norm": 1.758589267730713,
      "learning_rate": 0.00016563598307209204,
      "loss": 1.2776,
      "step": 612
    },
    {
      "epoch": 0.2951905134533197,
      "grad_norm": 1.1346322298049927,
      "learning_rate": 0.0001655182186454344,
      "loss": 0.8349,
      "step": 613
    },
    {
      "epoch": 0.29567206404622887,
      "grad_norm": 1.1793245077133179,
      "learning_rate": 0.00016540029479877638,
      "loss": 1.134,
      "step": 614
    },
    {
      "epoch": 0.29615361463913803,
      "grad_norm": 1.421286702156067,
      "learning_rate": 0.00016528221181905217,
      "loss": 1.292,
      "step": 615
    },
    {
      "epoch": 0.2966351652320472,
      "grad_norm": 1.9331536293029785,
      "learning_rate": 0.00016516396999358322,
      "loss": 1.3258,
      "step": 616
    },
    {
      "epoch": 0.29711671582495636,
      "grad_norm": 0.9758068919181824,
      "learning_rate": 0.00016504556961007748,
      "loss": 0.8893,
      "step": 617
    },
    {
      "epoch": 0.2975982664178655,
      "grad_norm": 1.1783885955810547,
      "learning_rate": 0.00016492701095662866,
      "loss": 1.081,
      "step": 618
    },
    {
      "epoch": 0.2980798170107747,
      "grad_norm": 1.6655336618423462,
      "learning_rate": 0.00016480829432171564,
      "loss": 1.328,
      "step": 619
    },
    {
      "epoch": 0.29856136760368385,
      "grad_norm": 0.95665442943573,
      "learning_rate": 0.0001646894199942017,
      "loss": 0.952,
      "step": 620
    },
    {
      "epoch": 0.299042918196593,
      "grad_norm": 1.0933932065963745,
      "learning_rate": 0.0001645703882633338,
      "loss": 1.2773,
      "step": 621
    },
    {
      "epoch": 0.2995244687895022,
      "grad_norm": 1.4704654216766357,
      "learning_rate": 0.00016445119941874183,
      "loss": 0.7958,
      "step": 622
    },
    {
      "epoch": 0.30000601938241134,
      "grad_norm": 1.8412909507751465,
      "learning_rate": 0.00016433185375043809,
      "loss": 1.3947,
      "step": 623
    },
    {
      "epoch": 0.30048756997532056,
      "grad_norm": 1.901084303855896,
      "learning_rate": 0.00016421235154881638,
      "loss": 1.6194,
      "step": 624
    },
    {
      "epoch": 0.3009691205682297,
      "grad_norm": 1.0536638498306274,
      "learning_rate": 0.00016409269310465146,
      "loss": 0.8081,
      "step": 625
    },
    {
      "epoch": 0.3014506711611389,
      "grad_norm": 2.2057197093963623,
      "learning_rate": 0.00016397287870909813,
      "loss": 1.7031,
      "step": 626
    },
    {
      "epoch": 0.30193222175404805,
      "grad_norm": 1.598200798034668,
      "learning_rate": 0.00016385290865369079,
      "loss": 1.7048,
      "step": 627
    },
    {
      "epoch": 0.3024137723469572,
      "grad_norm": 1.4274382591247559,
      "learning_rate": 0.00016373278323034255,
      "loss": 1.4358,
      "step": 628
    },
    {
      "epoch": 0.3028953229398664,
      "grad_norm": 0.8829864859580994,
      "learning_rate": 0.0001636125027313445,
      "loss": 1.0346,
      "step": 629
    },
    {
      "epoch": 0.30337687353277554,
      "grad_norm": 2.155449151992798,
      "learning_rate": 0.00016349206744936518,
      "loss": 1.826,
      "step": 630
    },
    {
      "epoch": 0.3038584241256847,
      "grad_norm": 1.4288023710250854,
      "learning_rate": 0.00016337147767744967,
      "loss": 1.2452,
      "step": 631
    },
    {
      "epoch": 0.30433997471859386,
      "grad_norm": 1.6957374811172485,
      "learning_rate": 0.0001632507337090189,
      "loss": 1.2266,
      "step": 632
    },
    {
      "epoch": 0.304821525311503,
      "grad_norm": 1.418856143951416,
      "learning_rate": 0.0001631298358378692,
      "loss": 1.4912,
      "step": 633
    },
    {
      "epoch": 0.3053030759044122,
      "grad_norm": 1.6799858808517456,
      "learning_rate": 0.00016300878435817113,
      "loss": 1.4365,
      "step": 634
    },
    {
      "epoch": 0.30578462649732135,
      "grad_norm": 1.2947931289672852,
      "learning_rate": 0.00016288757956446918,
      "loss": 1.3026,
      "step": 635
    },
    {
      "epoch": 0.3062661770902305,
      "grad_norm": 1.8288661241531372,
      "learning_rate": 0.00016276622175168083,
      "loss": 1.0523,
      "step": 636
    },
    {
      "epoch": 0.30674772768313974,
      "grad_norm": 1.6945993900299072,
      "learning_rate": 0.0001626447112150959,
      "loss": 1.7998,
      "step": 637
    },
    {
      "epoch": 0.3072292782760489,
      "grad_norm": 1.3144981861114502,
      "learning_rate": 0.00016252304825037576,
      "loss": 1.2975,
      "step": 638
    },
    {
      "epoch": 0.30771082886895806,
      "grad_norm": 1.617040991783142,
      "learning_rate": 0.0001624012331535528,
      "loss": 1.7468,
      "step": 639
    },
    {
      "epoch": 0.3081923794618672,
      "grad_norm": 2.113701343536377,
      "learning_rate": 0.00016227926622102947,
      "loss": 0.8698,
      "step": 640
    },
    {
      "epoch": 0.3086739300547764,
      "grad_norm": 2.9178152084350586,
      "learning_rate": 0.00016215714774957772,
      "loss": 1.7186,
      "step": 641
    },
    {
      "epoch": 0.30915548064768555,
      "grad_norm": 1.9879353046417236,
      "learning_rate": 0.00016203487803633822,
      "loss": 1.7444,
      "step": 642
    },
    {
      "epoch": 0.3096370312405947,
      "grad_norm": 1.2270036935806274,
      "learning_rate": 0.00016191245737881956,
      "loss": 1.0314,
      "step": 643
    },
    {
      "epoch": 0.3101185818335039,
      "grad_norm": 1.0305957794189453,
      "learning_rate": 0.00016178988607489777,
      "loss": 0.9286,
      "step": 644
    },
    {
      "epoch": 0.31060013242641304,
      "grad_norm": 1.9618016481399536,
      "learning_rate": 0.00016166716442281528,
      "loss": 1.552,
      "step": 645
    },
    {
      "epoch": 0.3110816830193222,
      "grad_norm": 2.0464658737182617,
      "learning_rate": 0.0001615442927211805,
      "loss": 1.2101,
      "step": 646
    },
    {
      "epoch": 0.31156323361223137,
      "grad_norm": 0.8513254523277283,
      "learning_rate": 0.0001614212712689668,
      "loss": 1.0417,
      "step": 647
    },
    {
      "epoch": 0.31204478420514054,
      "grad_norm": 1.1119502782821655,
      "learning_rate": 0.00016129810036551198,
      "loss": 1.2709,
      "step": 648
    },
    {
      "epoch": 0.3125263347980497,
      "grad_norm": 1.5346370935440063,
      "learning_rate": 0.00016117478031051755,
      "loss": 1.1562,
      "step": 649
    },
    {
      "epoch": 0.31300788539095886,
      "grad_norm": 1.3416459560394287,
      "learning_rate": 0.00016105131140404787,
      "loss": 1.2396,
      "step": 650
    },
    {
      "epoch": 0.3134894359838681,
      "grad_norm": 1.6061738729476929,
      "learning_rate": 0.00016092769394652947,
      "loss": 1.2197,
      "step": 651
    },
    {
      "epoch": 0.31397098657677724,
      "grad_norm": 1.1048046350479126,
      "learning_rate": 0.0001608039282387504,
      "loss": 0.9935,
      "step": 652
    },
    {
      "epoch": 0.3144525371696864,
      "grad_norm": 1.1886389255523682,
      "learning_rate": 0.00016068001458185936,
      "loss": 0.9865,
      "step": 653
    },
    {
      "epoch": 0.31493408776259557,
      "grad_norm": 1.1007704734802246,
      "learning_rate": 0.0001605559532773651,
      "loss": 0.8621,
      "step": 654
    },
    {
      "epoch": 0.31541563835550473,
      "grad_norm": 0.9339162707328796,
      "learning_rate": 0.00016043174462713566,
      "loss": 0.7336,
      "step": 655
    },
    {
      "epoch": 0.3158971889484139,
      "grad_norm": 1.7942975759506226,
      "learning_rate": 0.00016030738893339753,
      "loss": 1.1548,
      "step": 656
    },
    {
      "epoch": 0.31637873954132306,
      "grad_norm": 1.5289782285690308,
      "learning_rate": 0.00016018288649873497,
      "loss": 1.2926,
      "step": 657
    },
    {
      "epoch": 0.3168602901342322,
      "grad_norm": 1.0050541162490845,
      "learning_rate": 0.0001600582376260894,
      "loss": 0.9772,
      "step": 658
    },
    {
      "epoch": 0.3173418407271414,
      "grad_norm": 2.001206874847412,
      "learning_rate": 0.00015993344261875847,
      "loss": 1.3642,
      "step": 659
    },
    {
      "epoch": 0.31782339132005055,
      "grad_norm": 0.9426195621490479,
      "learning_rate": 0.00015980850178039547,
      "loss": 0.9622,
      "step": 660
    },
    {
      "epoch": 0.3183049419129597,
      "grad_norm": 0.7804467678070068,
      "learning_rate": 0.00015968341541500842,
      "loss": 1.1255,
      "step": 661
    },
    {
      "epoch": 0.3187864925058689,
      "grad_norm": 1.1761704683303833,
      "learning_rate": 0.00015955818382695953,
      "loss": 0.8444,
      "step": 662
    },
    {
      "epoch": 0.31926804309877804,
      "grad_norm": 0.9520278573036194,
      "learning_rate": 0.00015943280732096438,
      "loss": 1.1458,
      "step": 663
    },
    {
      "epoch": 0.3197495936916872,
      "grad_norm": 1.676244854927063,
      "learning_rate": 0.00015930728620209113,
      "loss": 1.532,
      "step": 664
    },
    {
      "epoch": 0.3202311442845964,
      "grad_norm": 1.5964101552963257,
      "learning_rate": 0.00015918162077575976,
      "loss": 1.2607,
      "step": 665
    },
    {
      "epoch": 0.3207126948775056,
      "grad_norm": 0.8836573958396912,
      "learning_rate": 0.00015905581134774153,
      "loss": 1.0729,
      "step": 666
    },
    {
      "epoch": 0.32119424547041475,
      "grad_norm": 1.1107227802276611,
      "learning_rate": 0.0001589298582241579,
      "loss": 1.037,
      "step": 667
    },
    {
      "epoch": 0.3216757960633239,
      "grad_norm": 1.9917302131652832,
      "learning_rate": 0.00015880376171148014,
      "loss": 1.2102,
      "step": 668
    },
    {
      "epoch": 0.3221573466562331,
      "grad_norm": 2.0715506076812744,
      "learning_rate": 0.00015867752211652831,
      "loss": 1.2414,
      "step": 669
    },
    {
      "epoch": 0.32263889724914224,
      "grad_norm": 1.0055168867111206,
      "learning_rate": 0.00015855113974647068,
      "loss": 1.3899,
      "step": 670
    },
    {
      "epoch": 0.3231204478420514,
      "grad_norm": 1.2430107593536377,
      "learning_rate": 0.0001584246149088229,
      "loss": 0.838,
      "step": 671
    },
    {
      "epoch": 0.32360199843496057,
      "grad_norm": 1.218998908996582,
      "learning_rate": 0.0001582979479114472,
      "loss": 1.0783,
      "step": 672
    },
    {
      "epoch": 0.32408354902786973,
      "grad_norm": 1.642632246017456,
      "learning_rate": 0.0001581711390625519,
      "loss": 1.4607,
      "step": 673
    },
    {
      "epoch": 0.3245650996207789,
      "grad_norm": 1.363073706626892,
      "learning_rate": 0.0001580441886706903,
      "loss": 0.9729,
      "step": 674
    },
    {
      "epoch": 0.32504665021368806,
      "grad_norm": 2.2371654510498047,
      "learning_rate": 0.00015791709704476015,
      "loss": 1.2835,
      "step": 675
    },
    {
      "epoch": 0.3255282008065972,
      "grad_norm": 1.9441112279891968,
      "learning_rate": 0.00015778986449400292,
      "loss": 1.2487,
      "step": 676
    },
    {
      "epoch": 0.3260097513995064,
      "grad_norm": 1.4167160987854004,
      "learning_rate": 0.00015766249132800292,
      "loss": 0.9911,
      "step": 677
    },
    {
      "epoch": 0.3264913019924156,
      "grad_norm": 1.316462516784668,
      "learning_rate": 0.00015753497785668663,
      "loss": 0.8893,
      "step": 678
    },
    {
      "epoch": 0.32697285258532477,
      "grad_norm": 0.9346000552177429,
      "learning_rate": 0.00015740732439032187,
      "loss": 0.9079,
      "step": 679
    },
    {
      "epoch": 0.32745440317823393,
      "grad_norm": 0.8193636536598206,
      "learning_rate": 0.00015727953123951716,
      "loss": 1.1161,
      "step": 680
    },
    {
      "epoch": 0.3279359537711431,
      "grad_norm": 1.1005698442459106,
      "learning_rate": 0.00015715159871522086,
      "loss": 1.1385,
      "step": 681
    },
    {
      "epoch": 0.32841750436405226,
      "grad_norm": 1.4008612632751465,
      "learning_rate": 0.00015702352712872056,
      "loss": 1.4036,
      "step": 682
    },
    {
      "epoch": 0.3288990549569614,
      "grad_norm": 0.950045108795166,
      "learning_rate": 0.00015689531679164204,
      "loss": 1.13,
      "step": 683
    },
    {
      "epoch": 0.3293806055498706,
      "grad_norm": 1.9419782161712646,
      "learning_rate": 0.00015676696801594886,
      "loss": 1.8128,
      "step": 684
    },
    {
      "epoch": 0.32986215614277975,
      "grad_norm": 1.1451201438903809,
      "learning_rate": 0.00015663848111394132,
      "loss": 1.2477,
      "step": 685
    },
    {
      "epoch": 0.3303437067356889,
      "grad_norm": 1.3408732414245605,
      "learning_rate": 0.00015650985639825585,
      "loss": 1.3234,
      "step": 686
    },
    {
      "epoch": 0.3308252573285981,
      "grad_norm": 1.4776135683059692,
      "learning_rate": 0.00015638109418186424,
      "loss": 0.9564,
      "step": 687
    },
    {
      "epoch": 0.33130680792150724,
      "grad_norm": 1.3781464099884033,
      "learning_rate": 0.00015625219477807277,
      "loss": 1.5948,
      "step": 688
    },
    {
      "epoch": 0.3317883585144164,
      "grad_norm": 1.5889679193496704,
      "learning_rate": 0.00015612315850052166,
      "loss": 0.9726,
      "step": 689
    },
    {
      "epoch": 0.33226990910732557,
      "grad_norm": 2.7366557121276855,
      "learning_rate": 0.00015599398566318396,
      "loss": 1.4106,
      "step": 690
    },
    {
      "epoch": 0.33275145970023473,
      "grad_norm": 1.2634618282318115,
      "learning_rate": 0.00015586467658036524,
      "loss": 0.8895,
      "step": 691
    },
    {
      "epoch": 0.33323301029314395,
      "grad_norm": 1.7835899591445923,
      "learning_rate": 0.00015573523156670244,
      "loss": 1.1469,
      "step": 692
    },
    {
      "epoch": 0.3337145608860531,
      "grad_norm": 2.0157241821289062,
      "learning_rate": 0.0001556056509371633,
      "loss": 1.6676,
      "step": 693
    },
    {
      "epoch": 0.3341961114789623,
      "grad_norm": 1.036002516746521,
      "learning_rate": 0.00015547593500704547,
      "loss": 1.1597,
      "step": 694
    },
    {
      "epoch": 0.33467766207187144,
      "grad_norm": 1.3478120565414429,
      "learning_rate": 0.00015534608409197592,
      "loss": 0.6631,
      "step": 695
    },
    {
      "epoch": 0.3351592126647806,
      "grad_norm": 1.650553584098816,
      "learning_rate": 0.00015521609850791004,
      "loss": 0.9198,
      "step": 696
    },
    {
      "epoch": 0.33564076325768977,
      "grad_norm": 1.0411776304244995,
      "learning_rate": 0.0001550859785711308,
      "loss": 1.3435,
      "step": 697
    },
    {
      "epoch": 0.33612231385059893,
      "grad_norm": 1.168008804321289,
      "learning_rate": 0.0001549557245982482,
      "loss": 1.2745,
      "step": 698
    },
    {
      "epoch": 0.3366038644435081,
      "grad_norm": 1.3533867597579956,
      "learning_rate": 0.00015482533690619837,
      "loss": 0.999,
      "step": 699
    },
    {
      "epoch": 0.33708541503641726,
      "grad_norm": 2.047961473464966,
      "learning_rate": 0.00015469481581224272,
      "loss": 0.9148,
      "step": 700
    },
    {
      "epoch": 0.3375669656293264,
      "grad_norm": 1.2058629989624023,
      "learning_rate": 0.0001545641616339673,
      "loss": 1.0357,
      "step": 701
    },
    {
      "epoch": 0.3380485162222356,
      "grad_norm": 1.341555118560791,
      "learning_rate": 0.00015443337468928206,
      "loss": 1.0662,
      "step": 702
    },
    {
      "epoch": 0.33853006681514475,
      "grad_norm": 1.428769826889038,
      "learning_rate": 0.00015430245529641986,
      "loss": 0.9347,
      "step": 703
    },
    {
      "epoch": 0.3390116174080539,
      "grad_norm": 0.882029116153717,
      "learning_rate": 0.00015417140377393596,
      "loss": 1.1294,
      "step": 704
    },
    {
      "epoch": 0.3394931680009631,
      "grad_norm": 1.0680418014526367,
      "learning_rate": 0.00015404022044070704,
      "loss": 0.7804,
      "step": 705
    },
    {
      "epoch": 0.3399747185938723,
      "grad_norm": 1.5089625120162964,
      "learning_rate": 0.00015390890561593052,
      "loss": 0.9825,
      "step": 706
    },
    {
      "epoch": 0.34045626918678146,
      "grad_norm": 1.4794280529022217,
      "learning_rate": 0.0001537774596191238,
      "loss": 1.208,
      "step": 707
    },
    {
      "epoch": 0.3409378197796906,
      "grad_norm": 1.2734895944595337,
      "learning_rate": 0.00015364588277012344,
      "loss": 1.1696,
      "step": 708
    },
    {
      "epoch": 0.3414193703725998,
      "grad_norm": 1.0913923978805542,
      "learning_rate": 0.00015351417538908435,
      "loss": 1.3214,
      "step": 709
    },
    {
      "epoch": 0.34190092096550895,
      "grad_norm": 2.045374631881714,
      "learning_rate": 0.0001533823377964791,
      "loss": 1.2074,
      "step": 710
    },
    {
      "epoch": 0.3423824715584181,
      "grad_norm": 1.3556513786315918,
      "learning_rate": 0.00015325037031309704,
      "loss": 1.6772,
      "step": 711
    },
    {
      "epoch": 0.3428640221513273,
      "grad_norm": 1.3498286008834839,
      "learning_rate": 0.00015311827326004363,
      "loss": 1.2355,
      "step": 712
    },
    {
      "epoch": 0.34334557274423644,
      "grad_norm": 1.201331615447998,
      "learning_rate": 0.0001529860469587396,
      "loss": 0.9046,
      "step": 713
    },
    {
      "epoch": 0.3438271233371456,
      "grad_norm": 2.004056930541992,
      "learning_rate": 0.00015285369173092015,
      "loss": 1.2336,
      "step": 714
    },
    {
      "epoch": 0.34430867393005476,
      "grad_norm": 0.8504548072814941,
      "learning_rate": 0.00015272120789863413,
      "loss": 0.7396,
      "step": 715
    },
    {
      "epoch": 0.34479022452296393,
      "grad_norm": 1.9286209344863892,
      "learning_rate": 0.00015258859578424342,
      "loss": 1.2599,
      "step": 716
    },
    {
      "epoch": 0.3452717751158731,
      "grad_norm": 1.2156747579574585,
      "learning_rate": 0.00015245585571042194,
      "loss": 1.0014,
      "step": 717
    },
    {
      "epoch": 0.34575332570878226,
      "grad_norm": 1.3780698776245117,
      "learning_rate": 0.00015232298800015506,
      "loss": 0.8825,
      "step": 718
    },
    {
      "epoch": 0.3462348763016914,
      "grad_norm": 1.363624930381775,
      "learning_rate": 0.00015218999297673862,
      "loss": 0.8521,
      "step": 719
    },
    {
      "epoch": 0.34671642689460064,
      "grad_norm": 1.856489896774292,
      "learning_rate": 0.0001520568709637783,
      "loss": 1.3032,
      "step": 720
    },
    {
      "epoch": 0.3471979774875098,
      "grad_norm": 2.319241523742676,
      "learning_rate": 0.00015192362228518875,
      "loss": 0.685,
      "step": 721
    },
    {
      "epoch": 0.34767952808041896,
      "grad_norm": 1.1534900665283203,
      "learning_rate": 0.00015179024726519284,
      "loss": 1.1928,
      "step": 722
    },
    {
      "epoch": 0.34816107867332813,
      "grad_norm": 2.582562208175659,
      "learning_rate": 0.00015165674622832085,
      "loss": 1.1445,
      "step": 723
    },
    {
      "epoch": 0.3486426292662373,
      "grad_norm": 1.2963651418685913,
      "learning_rate": 0.0001515231194994097,
      "loss": 1.6773,
      "step": 724
    },
    {
      "epoch": 0.34912417985914646,
      "grad_norm": 1.9767524003982544,
      "learning_rate": 0.00015138936740360207,
      "loss": 1.255,
      "step": 725
    },
    {
      "epoch": 0.3496057304520556,
      "grad_norm": 1.5948400497436523,
      "learning_rate": 0.00015125549026634585,
      "loss": 0.6921,
      "step": 726
    },
    {
      "epoch": 0.3500872810449648,
      "grad_norm": 1.4339981079101562,
      "learning_rate": 0.00015112148841339295,
      "loss": 1.2651,
      "step": 727
    },
    {
      "epoch": 0.35056883163787395,
      "grad_norm": 0.9148167967796326,
      "learning_rate": 0.000150987362170799,
      "loss": 0.9406,
      "step": 728
    },
    {
      "epoch": 0.3510503822307831,
      "grad_norm": 1.7655442953109741,
      "learning_rate": 0.00015085311186492206,
      "loss": 1.9525,
      "step": 729
    },
    {
      "epoch": 0.3515319328236923,
      "grad_norm": 1.262243390083313,
      "learning_rate": 0.00015071873782242223,
      "loss": 0.7161,
      "step": 730
    },
    {
      "epoch": 0.35201348341660144,
      "grad_norm": 1.4148969650268555,
      "learning_rate": 0.0001505842403702606,
      "loss": 1.6641,
      "step": 731
    },
    {
      "epoch": 0.3524950340095106,
      "grad_norm": 1.245644211769104,
      "learning_rate": 0.00015044961983569856,
      "loss": 0.964,
      "step": 732
    },
    {
      "epoch": 0.3529765846024198,
      "grad_norm": 1.8993353843688965,
      "learning_rate": 0.00015031487654629702,
      "loss": 1.1868,
      "step": 733
    },
    {
      "epoch": 0.353458135195329,
      "grad_norm": 3.0837442874908447,
      "learning_rate": 0.00015018001082991553,
      "loss": 1.0727,
      "step": 734
    },
    {
      "epoch": 0.35393968578823815,
      "grad_norm": 1.5420244932174683,
      "learning_rate": 0.0001500450230147116,
      "loss": 0.9753,
      "step": 735
    },
    {
      "epoch": 0.3544212363811473,
      "grad_norm": 1.2493274211883545,
      "learning_rate": 0.00014990991342913974,
      "loss": 1.4426,
      "step": 736
    },
    {
      "epoch": 0.35490278697405647,
      "grad_norm": 1.284993052482605,
      "learning_rate": 0.00014977468240195084,
      "loss": 1.1933,
      "step": 737
    },
    {
      "epoch": 0.35538433756696564,
      "grad_norm": 1.6030765771865845,
      "learning_rate": 0.0001496393302621912,
      "loss": 1.4973,
      "step": 738
    },
    {
      "epoch": 0.3558658881598748,
      "grad_norm": 1.149884581565857,
      "learning_rate": 0.00014950385733920188,
      "loss": 0.9662,
      "step": 739
    },
    {
      "epoch": 0.35634743875278396,
      "grad_norm": 1.1744946241378784,
      "learning_rate": 0.00014936826396261783,
      "loss": 1.1248,
      "step": 740
    },
    {
      "epoch": 0.3568289893456931,
      "grad_norm": 1.6480766534805298,
      "learning_rate": 0.00014923255046236705,
      "loss": 1.3487,
      "step": 741
    },
    {
      "epoch": 0.3573105399386023,
      "grad_norm": 1.3761893510818481,
      "learning_rate": 0.00014909671716866984,
      "loss": 1.068,
      "step": 742
    },
    {
      "epoch": 0.35779209053151145,
      "grad_norm": 1.6178627014160156,
      "learning_rate": 0.00014896076441203802,
      "loss": 1.1852,
      "step": 743
    },
    {
      "epoch": 0.3582736411244206,
      "grad_norm": 1.0326192378997803,
      "learning_rate": 0.000148824692523274,
      "loss": 1.3524,
      "step": 744
    },
    {
      "epoch": 0.3587551917173298,
      "grad_norm": 2.3446290493011475,
      "learning_rate": 0.0001486885018334702,
      "loss": 1.1842,
      "step": 745
    },
    {
      "epoch": 0.35923674231023894,
      "grad_norm": 0.9821613430976868,
      "learning_rate": 0.00014855219267400797,
      "loss": 1.3538,
      "step": 746
    },
    {
      "epoch": 0.35971829290314816,
      "grad_norm": 2.048778533935547,
      "learning_rate": 0.00014841576537655705,
      "loss": 1.4605,
      "step": 747
    },
    {
      "epoch": 0.3601998434960573,
      "grad_norm": 1.3085370063781738,
      "learning_rate": 0.00014827922027307451,
      "loss": 1.3022,
      "step": 748
    },
    {
      "epoch": 0.3606813940889665,
      "grad_norm": 1.3689935207366943,
      "learning_rate": 0.00014814255769580415,
      "loss": 0.956,
      "step": 749
    },
    {
      "epoch": 0.36116294468187565,
      "grad_norm": 0.7956721782684326,
      "learning_rate": 0.00014800577797727558,
      "loss": 1.2298,
      "step": 750
    },
    {
      "epoch": 0.3616444952747848,
      "grad_norm": 1.4484717845916748,
      "learning_rate": 0.00014786888145030343,
      "loss": 1.084,
      "step": 751
    },
    {
      "epoch": 0.362126045867694,
      "grad_norm": 1.233156442642212,
      "learning_rate": 0.0001477318684479866,
      "loss": 0.5948,
      "step": 752
    },
    {
      "epoch": 0.36260759646060314,
      "grad_norm": 1.5255489349365234,
      "learning_rate": 0.00014759473930370736,
      "loss": 1.1348,
      "step": 753
    },
    {
      "epoch": 0.3630891470535123,
      "grad_norm": 0.967078685760498,
      "learning_rate": 0.0001474574943511306,
      "loss": 1.0705,
      "step": 754
    },
    {
      "epoch": 0.36357069764642147,
      "grad_norm": 1.5431402921676636,
      "learning_rate": 0.0001473201339242029,
      "loss": 1.6777,
      "step": 755
    },
    {
      "epoch": 0.36405224823933063,
      "grad_norm": 1.9125128984451294,
      "learning_rate": 0.000147182658357152,
      "loss": 0.917,
      "step": 756
    },
    {
      "epoch": 0.3645337988322398,
      "grad_norm": 1.0822197198867798,
      "learning_rate": 0.00014704506798448566,
      "loss": 1.1856,
      "step": 757
    },
    {
      "epoch": 0.36501534942514896,
      "grad_norm": 2.3671305179595947,
      "learning_rate": 0.00014690736314099101,
      "loss": 1.435,
      "step": 758
    },
    {
      "epoch": 0.3654969000180581,
      "grad_norm": 1.9273123741149902,
      "learning_rate": 0.00014676954416173373,
      "loss": 1.0769,
      "step": 759
    },
    {
      "epoch": 0.3659784506109673,
      "grad_norm": 1.9701248407363892,
      "learning_rate": 0.00014663161138205724,
      "loss": 1.7443,
      "step": 760
    },
    {
      "epoch": 0.3664600012038765,
      "grad_norm": 1.7919925451278687,
      "learning_rate": 0.00014649356513758176,
      "loss": 1.1036,
      "step": 761
    },
    {
      "epoch": 0.36694155179678567,
      "grad_norm": 1.6684398651123047,
      "learning_rate": 0.00014635540576420374,
      "loss": 1.0613,
      "step": 762
    },
    {
      "epoch": 0.36742310238969483,
      "grad_norm": 1.0149741172790527,
      "learning_rate": 0.0001462171335980948,
      "loss": 0.6544,
      "step": 763
    },
    {
      "epoch": 0.367904652982604,
      "grad_norm": 0.8502764105796814,
      "learning_rate": 0.00014607874897570105,
      "loss": 0.6647,
      "step": 764
    },
    {
      "epoch": 0.36838620357551316,
      "grad_norm": 1.2991472482681274,
      "learning_rate": 0.0001459402522337422,
      "loss": 1.041,
      "step": 765
    },
    {
      "epoch": 0.3688677541684223,
      "grad_norm": 1.3300578594207764,
      "learning_rate": 0.00014580164370921078,
      "loss": 1.2481,
      "step": 766
    },
    {
      "epoch": 0.3693493047613315,
      "grad_norm": 1.838719367980957,
      "learning_rate": 0.0001456629237393713,
      "loss": 1.5574,
      "step": 767
    },
    {
      "epoch": 0.36983085535424065,
      "grad_norm": 1.98796808719635,
      "learning_rate": 0.00014552409266175952,
      "loss": 1.1958,
      "step": 768
    },
    {
      "epoch": 0.3703124059471498,
      "grad_norm": 0.9198821783065796,
      "learning_rate": 0.00014538515081418142,
      "loss": 0.9834,
      "step": 769
    },
    {
      "epoch": 0.370793956540059,
      "grad_norm": 2.295051336288452,
      "learning_rate": 0.00014524609853471264,
      "loss": 1.5568,
      "step": 770
    },
    {
      "epoch": 0.37127550713296814,
      "grad_norm": 2.1981189250946045,
      "learning_rate": 0.00014510693616169741,
      "loss": 1.5369,
      "step": 771
    },
    {
      "epoch": 0.3717570577258773,
      "grad_norm": 0.7688530683517456,
      "learning_rate": 0.0001449676640337479,
      "loss": 0.9933,
      "step": 772
    },
    {
      "epoch": 0.37223860831878647,
      "grad_norm": 1.560829997062683,
      "learning_rate": 0.00014482828248974335,
      "loss": 1.2475,
      "step": 773
    },
    {
      "epoch": 0.3727201589116957,
      "grad_norm": 1.6355130672454834,
      "learning_rate": 0.00014468879186882916,
      "loss": 1.3347,
      "step": 774
    },
    {
      "epoch": 0.37320170950460485,
      "grad_norm": 1.2548389434814453,
      "learning_rate": 0.00014454919251041622,
      "loss": 0.8129,
      "step": 775
    },
    {
      "epoch": 0.373683260097514,
      "grad_norm": 1.1506437063217163,
      "learning_rate": 0.00014440948475418,
      "loss": 1.4842,
      "step": 776
    },
    {
      "epoch": 0.3741648106904232,
      "grad_norm": 1.567155361175537,
      "learning_rate": 0.00014426966894005966,
      "loss": 1.1584,
      "step": 777
    },
    {
      "epoch": 0.37464636128333234,
      "grad_norm": 1.4268014430999756,
      "learning_rate": 0.0001441297454082573,
      "loss": 1.3995,
      "step": 778
    },
    {
      "epoch": 0.3751279118762415,
      "grad_norm": 0.6414303779602051,
      "learning_rate": 0.00014398971449923722,
      "loss": 0.6128,
      "step": 779
    },
    {
      "epoch": 0.37560946246915067,
      "grad_norm": 1.0709565877914429,
      "learning_rate": 0.00014384957655372483,
      "loss": 1.0457,
      "step": 780
    },
    {
      "epoch": 0.37609101306205983,
      "grad_norm": 1.1160578727722168,
      "learning_rate": 0.00014370933191270617,
      "loss": 0.7673,
      "step": 781
    },
    {
      "epoch": 0.376572563654969,
      "grad_norm": 0.9713869690895081,
      "learning_rate": 0.0001435689809174267,
      "loss": 1.1465,
      "step": 782
    },
    {
      "epoch": 0.37705411424787816,
      "grad_norm": 1.1051602363586426,
      "learning_rate": 0.0001434285239093908,
      "loss": 1.579,
      "step": 783
    },
    {
      "epoch": 0.3775356648407873,
      "grad_norm": 1.084945559501648,
      "learning_rate": 0.00014328796123036071,
      "loss": 0.6927,
      "step": 784
    },
    {
      "epoch": 0.3780172154336965,
      "grad_norm": 1.097021222114563,
      "learning_rate": 0.0001431472932223559,
      "loss": 1.2428,
      "step": 785
    },
    {
      "epoch": 0.37849876602660565,
      "grad_norm": 1.0359227657318115,
      "learning_rate": 0.00014300652022765207,
      "loss": 0.9259,
      "step": 786
    },
    {
      "epoch": 0.3789803166195148,
      "grad_norm": 2.208172559738159,
      "learning_rate": 0.00014286564258878033,
      "loss": 1.4583,
      "step": 787
    },
    {
      "epoch": 0.37946186721242403,
      "grad_norm": 1.7827069759368896,
      "learning_rate": 0.00014272466064852644,
      "loss": 1.2386,
      "step": 788
    },
    {
      "epoch": 0.3799434178053332,
      "grad_norm": 1.489619255065918,
      "learning_rate": 0.00014258357474993,
      "loss": 1.2554,
      "step": 789
    },
    {
      "epoch": 0.38042496839824236,
      "grad_norm": 1.9039955139160156,
      "learning_rate": 0.0001424423852362835,
      "loss": 1.076,
      "step": 790
    },
    {
      "epoch": 0.3809065189911515,
      "grad_norm": 1.2301115989685059,
      "learning_rate": 0.00014230109245113158,
      "loss": 1.1746,
      "step": 791
    },
    {
      "epoch": 0.3813880695840607,
      "grad_norm": 1.3033658266067505,
      "learning_rate": 0.00014215969673827018,
      "loss": 1.2226,
      "step": 792
    },
    {
      "epoch": 0.38186962017696985,
      "grad_norm": 1.7254654169082642,
      "learning_rate": 0.00014201819844174564,
      "loss": 1.489,
      "step": 793
    },
    {
      "epoch": 0.382351170769879,
      "grad_norm": 1.7269105911254883,
      "learning_rate": 0.0001418765979058539,
      "loss": 1.5071,
      "step": 794
    },
    {
      "epoch": 0.3828327213627882,
      "grad_norm": 1.0820810794830322,
      "learning_rate": 0.00014173489547513973,
      "loss": 1.3856,
      "step": 795
    },
    {
      "epoch": 0.38331427195569734,
      "grad_norm": 1.9783928394317627,
      "learning_rate": 0.00014159309149439582,
      "loss": 1.3175,
      "step": 796
    },
    {
      "epoch": 0.3837958225486065,
      "grad_norm": 1.1702148914337158,
      "learning_rate": 0.00014145118630866187,
      "loss": 0.9354,
      "step": 797
    },
    {
      "epoch": 0.38427737314151567,
      "grad_norm": 2.061983585357666,
      "learning_rate": 0.000141309180263224,
      "loss": 0.9897,
      "step": 798
    },
    {
      "epoch": 0.38475892373442483,
      "grad_norm": 1.6521720886230469,
      "learning_rate": 0.0001411670737036135,
      "loss": 1.0659,
      "step": 799
    },
    {
      "epoch": 0.385240474327334,
      "grad_norm": 1.5973643064498901,
      "learning_rate": 0.0001410248669756065,
      "loss": 1.3637,
      "step": 800
    },
    {
      "epoch": 0.38572202492024316,
      "grad_norm": 1.70493483543396,
      "learning_rate": 0.00014088256042522264,
      "loss": 0.7804,
      "step": 801
    },
    {
      "epoch": 0.3862035755131524,
      "grad_norm": 1.39597749710083,
      "learning_rate": 0.00014074015439872458,
      "loss": 1.2774,
      "step": 802
    },
    {
      "epoch": 0.38668512610606154,
      "grad_norm": 1.109622836112976,
      "learning_rate": 0.00014059764924261703,
      "loss": 0.6784,
      "step": 803
    },
    {
      "epoch": 0.3871666766989707,
      "grad_norm": 1.1049227714538574,
      "learning_rate": 0.00014045504530364584,
      "loss": 0.9093,
      "step": 804
    },
    {
      "epoch": 0.38764822729187987,
      "grad_norm": 1.4939508438110352,
      "learning_rate": 0.00014031234292879725,
      "loss": 0.9247,
      "step": 805
    },
    {
      "epoch": 0.38812977788478903,
      "grad_norm": 0.720761775970459,
      "learning_rate": 0.00014016954246529696,
      "loss": 0.9411,
      "step": 806
    },
    {
      "epoch": 0.3886113284776982,
      "grad_norm": 1.3061058521270752,
      "learning_rate": 0.00014002664426060942,
      "loss": 1.354,
      "step": 807
    },
    {
      "epoch": 0.38909287907060736,
      "grad_norm": 1.391789197921753,
      "learning_rate": 0.00013988364866243693,
      "loss": 1.3514,
      "step": 808
    },
    {
      "epoch": 0.3895744296635165,
      "grad_norm": 1.488972544670105,
      "learning_rate": 0.00013974055601871868,
      "loss": 0.7408,
      "step": 809
    },
    {
      "epoch": 0.3900559802564257,
      "grad_norm": 1.326461911201477,
      "learning_rate": 0.00013959736667762998,
      "loss": 1.2627,
      "step": 810
    },
    {
      "epoch": 0.39053753084933485,
      "grad_norm": 0.9552280902862549,
      "learning_rate": 0.00013945408098758156,
      "loss": 1.1061,
      "step": 811
    },
    {
      "epoch": 0.391019081442244,
      "grad_norm": 0.850977897644043,
      "learning_rate": 0.0001393106992972184,
      "loss": 0.6452,
      "step": 812
    },
    {
      "epoch": 0.3915006320351532,
      "grad_norm": 1.2269916534423828,
      "learning_rate": 0.00013916722195541926,
      "loss": 0.8735,
      "step": 813
    },
    {
      "epoch": 0.39198218262806234,
      "grad_norm": 2.1774916648864746,
      "learning_rate": 0.00013902364931129557,
      "loss": 1.0694,
      "step": 814
    },
    {
      "epoch": 0.39246373322097156,
      "grad_norm": 1.2577506303787231,
      "learning_rate": 0.00013887998171419058,
      "loss": 0.9728,
      "step": 815
    },
    {
      "epoch": 0.3929452838138807,
      "grad_norm": 1.369600772857666,
      "learning_rate": 0.00013873621951367862,
      "loss": 1.3337,
      "step": 816
    },
    {
      "epoch": 0.3934268344067899,
      "grad_norm": 1.62059485912323,
      "learning_rate": 0.00013859236305956425,
      "loss": 1.1286,
      "step": 817
    },
    {
      "epoch": 0.39390838499969905,
      "grad_norm": 1.6724095344543457,
      "learning_rate": 0.00013844841270188132,
      "loss": 1.2261,
      "step": 818
    },
    {
      "epoch": 0.3943899355926082,
      "grad_norm": 1.1767303943634033,
      "learning_rate": 0.00013830436879089228,
      "loss": 1.2912,
      "step": 819
    },
    {
      "epoch": 0.3948714861855174,
      "grad_norm": 1.427933692932129,
      "learning_rate": 0.00013816023167708704,
      "loss": 1.122,
      "step": 820
    },
    {
      "epoch": 0.39535303677842654,
      "grad_norm": 1.7607033252716064,
      "learning_rate": 0.00013801600171118244,
      "loss": 0.6785,
      "step": 821
    },
    {
      "epoch": 0.3958345873713357,
      "grad_norm": 1.9177988767623901,
      "learning_rate": 0.00013787167924412112,
      "loss": 1.3967,
      "step": 822
    },
    {
      "epoch": 0.39631613796424486,
      "grad_norm": 1.5252182483673096,
      "learning_rate": 0.0001377272646270709,
      "loss": 0.7309,
      "step": 823
    },
    {
      "epoch": 0.396797688557154,
      "grad_norm": 1.2310963869094849,
      "learning_rate": 0.00013758275821142382,
      "loss": 1.0149,
      "step": 824
    },
    {
      "epoch": 0.3972792391500632,
      "grad_norm": 1.2247278690338135,
      "learning_rate": 0.00013743816034879523,
      "loss": 0.9136,
      "step": 825
    },
    {
      "epoch": 0.39776078974297235,
      "grad_norm": 1.3076244592666626,
      "learning_rate": 0.000137293471391023,
      "loss": 1.4165,
      "step": 826
    },
    {
      "epoch": 0.3982423403358815,
      "grad_norm": 1.349042296409607,
      "learning_rate": 0.00013714869169016667,
      "loss": 0.7338,
      "step": 827
    },
    {
      "epoch": 0.3987238909287907,
      "grad_norm": 1.0609012842178345,
      "learning_rate": 0.00013700382159850656,
      "loss": 0.9101,
      "step": 828
    },
    {
      "epoch": 0.3992054415216999,
      "grad_norm": 1.4679254293441772,
      "learning_rate": 0.00013685886146854297,
      "loss": 1.0233,
      "step": 829
    },
    {
      "epoch": 0.39968699211460906,
      "grad_norm": 1.478952407836914,
      "learning_rate": 0.00013671381165299525,
      "loss": 1.2965,
      "step": 830
    },
    {
      "epoch": 0.4001685427075182,
      "grad_norm": 1.795816421508789,
      "learning_rate": 0.00013656867250480098,
      "loss": 1.015,
      "step": 831
    },
    {
      "epoch": 0.4006500933004274,
      "grad_norm": 1.4675918817520142,
      "learning_rate": 0.00013642344437711512,
      "loss": 0.5833,
      "step": 832
    },
    {
      "epoch": 0.40113164389333655,
      "grad_norm": 1.9374818801879883,
      "learning_rate": 0.00013627812762330912,
      "loss": 1.1974,
      "step": 833
    },
    {
      "epoch": 0.4016131944862457,
      "grad_norm": 1.272425651550293,
      "learning_rate": 0.00013613272259697007,
      "loss": 1.1425,
      "step": 834
    },
    {
      "epoch": 0.4020947450791549,
      "grad_norm": 1.067225694656372,
      "learning_rate": 0.00013598722965189986,
      "loss": 1.5249,
      "step": 835
    },
    {
      "epoch": 0.40257629567206404,
      "grad_norm": 0.9392385482788086,
      "learning_rate": 0.0001358416491421143,
      "loss": 1.1314,
      "step": 836
    },
    {
      "epoch": 0.4030578462649732,
      "grad_norm": 0.973700761795044,
      "learning_rate": 0.00013569598142184225,
      "loss": 1.2546,
      "step": 837
    },
    {
      "epoch": 0.40353939685788237,
      "grad_norm": 2.8119561672210693,
      "learning_rate": 0.00013555022684552483,
      "loss": 1.6709,
      "step": 838
    },
    {
      "epoch": 0.40402094745079153,
      "grad_norm": 2.1031854152679443,
      "learning_rate": 0.00013540438576781441,
      "loss": 1.0876,
      "step": 839
    },
    {
      "epoch": 0.4045024980437007,
      "grad_norm": 1.7705788612365723,
      "learning_rate": 0.0001352584585435739,
      "loss": 1.5983,
      "step": 840
    },
    {
      "epoch": 0.40498404863660986,
      "grad_norm": 2.5488033294677734,
      "learning_rate": 0.00013511244552787583,
      "loss": 1.1372,
      "step": 841
    },
    {
      "epoch": 0.405465599229519,
      "grad_norm": 1.5380879640579224,
      "learning_rate": 0.00013496634707600147,
      "loss": 1.5669,
      "step": 842
    },
    {
      "epoch": 0.40594714982242824,
      "grad_norm": 1.6294825077056885,
      "learning_rate": 0.0001348201635434399,
      "loss": 1.3751,
      "step": 843
    },
    {
      "epoch": 0.4064287004153374,
      "grad_norm": 1.5779178142547607,
      "learning_rate": 0.0001346738952858873,
      "loss": 1.3678,
      "step": 844
    },
    {
      "epoch": 0.40691025100824657,
      "grad_norm": 0.7559780478477478,
      "learning_rate": 0.000134527542659246,
      "loss": 1.7134,
      "step": 845
    },
    {
      "epoch": 0.40739180160115573,
      "grad_norm": 1.4292011260986328,
      "learning_rate": 0.00013438110601962362,
      "loss": 1.1493,
      "step": 846
    },
    {
      "epoch": 0.4078733521940649,
      "grad_norm": 0.8557339310646057,
      "learning_rate": 0.00013423458572333214,
      "loss": 0.9848,
      "step": 847
    },
    {
      "epoch": 0.40835490278697406,
      "grad_norm": 0.9579225778579712,
      "learning_rate": 0.0001340879821268872,
      "loss": 0.5911,
      "step": 848
    },
    {
      "epoch": 0.4088364533798832,
      "grad_norm": 2.2378923892974854,
      "learning_rate": 0.000133941295587007,
      "loss": 0.9402,
      "step": 849
    },
    {
      "epoch": 0.4093180039727924,
      "grad_norm": 0.9859408140182495,
      "learning_rate": 0.00013379452646061164,
      "loss": 0.7713,
      "step": 850
    },
    {
      "epoch": 0.40979955456570155,
      "grad_norm": 0.634448230266571,
      "learning_rate": 0.0001336476751048222,
      "loss": 1.2232,
      "step": 851
    },
    {
      "epoch": 0.4102811051586107,
      "grad_norm": 1.3613566160202026,
      "learning_rate": 0.00013350074187695979,
      "loss": 1.3016,
      "step": 852
    },
    {
      "epoch": 0.4107626557515199,
      "grad_norm": 1.2628653049468994,
      "learning_rate": 0.00013335372713454467,
      "loss": 0.9398,
      "step": 853
    },
    {
      "epoch": 0.41124420634442904,
      "grad_norm": 1.1055777072906494,
      "learning_rate": 0.0001332066312352956,
      "loss": 1.0807,
      "step": 854
    },
    {
      "epoch": 0.4117257569373382,
      "grad_norm": 1.241823434829712,
      "learning_rate": 0.00013305945453712868,
      "loss": 1.1685,
      "step": 855
    },
    {
      "epoch": 0.4122073075302474,
      "grad_norm": 1.7039461135864258,
      "learning_rate": 0.0001329121973981567,
      "loss": 1.1817,
      "step": 856
    },
    {
      "epoch": 0.4126888581231566,
      "grad_norm": 1.3929228782653809,
      "learning_rate": 0.00013276486017668807,
      "loss": 1.1567,
      "step": 857
    },
    {
      "epoch": 0.41317040871606575,
      "grad_norm": 1.2159061431884766,
      "learning_rate": 0.0001326174432312262,
      "loss": 1.0879,
      "step": 858
    },
    {
      "epoch": 0.4136519593089749,
      "grad_norm": 1.228167176246643,
      "learning_rate": 0.00013246994692046836,
      "loss": 0.973,
      "step": 859
    },
    {
      "epoch": 0.4141335099018841,
      "grad_norm": 1.8474361896514893,
      "learning_rate": 0.000132322371603305,
      "loss": 1.0141,
      "step": 860
    },
    {
      "epoch": 0.41461506049479324,
      "grad_norm": 0.8831610679626465,
      "learning_rate": 0.0001321747176388188,
      "loss": 0.6377,
      "step": 861
    },
    {
      "epoch": 0.4150966110877024,
      "grad_norm": 0.8661091923713684,
      "learning_rate": 0.00013202698538628376,
      "loss": 1.4466,
      "step": 862
    },
    {
      "epoch": 0.41557816168061157,
      "grad_norm": 2.1040754318237305,
      "learning_rate": 0.00013187917520516448,
      "loss": 0.7491,
      "step": 863
    },
    {
      "epoch": 0.41605971227352073,
      "grad_norm": 1.60671865940094,
      "learning_rate": 0.00013173128745511508,
      "loss": 1.2282,
      "step": 864
    },
    {
      "epoch": 0.4165412628664299,
      "grad_norm": 1.4245531558990479,
      "learning_rate": 0.0001315833224959784,
      "loss": 1.4584,
      "step": 865
    },
    {
      "epoch": 0.41702281345933906,
      "grad_norm": 0.9865677356719971,
      "learning_rate": 0.00013143528068778525,
      "loss": 1.4004,
      "step": 866
    },
    {
      "epoch": 0.4175043640522482,
      "grad_norm": 1.0726274251937866,
      "learning_rate": 0.00013128716239075338,
      "loss": 0.8711,
      "step": 867
    },
    {
      "epoch": 0.4179859146451574,
      "grad_norm": 1.2404314279556274,
      "learning_rate": 0.00013113896796528664,
      "loss": 1.0115,
      "step": 868
    },
    {
      "epoch": 0.41846746523806655,
      "grad_norm": 1.355831265449524,
      "learning_rate": 0.00013099069777197412,
      "loss": 1.0097,
      "step": 869
    },
    {
      "epoch": 0.41894901583097577,
      "grad_norm": 1.2276833057403564,
      "learning_rate": 0.0001308423521715893,
      "loss": 1.0982,
      "step": 870
    },
    {
      "epoch": 0.41943056642388493,
      "grad_norm": 0.7814748287200928,
      "learning_rate": 0.00013069393152508906,
      "loss": 0.8958,
      "step": 871
    },
    {
      "epoch": 0.4199121170167941,
      "grad_norm": 1.1194249391555786,
      "learning_rate": 0.00013054543619361303,
      "loss": 0.7148,
      "step": 872
    },
    {
      "epoch": 0.42039366760970326,
      "grad_norm": 1.2822390794754028,
      "learning_rate": 0.0001303968665384824,
      "loss": 1.0909,
      "step": 873
    },
    {
      "epoch": 0.4208752182026124,
      "grad_norm": 0.9457350969314575,
      "learning_rate": 0.00013024822292119934,
      "loss": 0.9633,
      "step": 874
    },
    {
      "epoch": 0.4213567687955216,
      "grad_norm": 1.0125480890274048,
      "learning_rate": 0.0001300995057034459,
      "loss": 1.1743,
      "step": 875
    },
    {
      "epoch": 0.42183831938843075,
      "grad_norm": 1.0287433862686157,
      "learning_rate": 0.00012995071524708325,
      "loss": 0.8275,
      "step": 876
    },
    {
      "epoch": 0.4223198699813399,
      "grad_norm": 2.6919994354248047,
      "learning_rate": 0.00012980185191415074,
      "loss": 0.9543,
      "step": 877
    },
    {
      "epoch": 0.4228014205742491,
      "grad_norm": 1.2487611770629883,
      "learning_rate": 0.0001296529160668651,
      "loss": 0.8677,
      "step": 878
    },
    {
      "epoch": 0.42328297116715824,
      "grad_norm": 0.7254701852798462,
      "learning_rate": 0.00012950390806761944,
      "loss": 0.9789,
      "step": 879
    },
    {
      "epoch": 0.4237645217600674,
      "grad_norm": 1.1634492874145508,
      "learning_rate": 0.0001293548282789825,
      "loss": 0.8729,
      "step": 880
    },
    {
      "epoch": 0.42424607235297657,
      "grad_norm": 1.3979501724243164,
      "learning_rate": 0.00012920567706369758,
      "loss": 1.9014,
      "step": 881
    },
    {
      "epoch": 0.42472762294588573,
      "grad_norm": 0.8077030777931213,
      "learning_rate": 0.00012905645478468192,
      "loss": 0.5351,
      "step": 882
    },
    {
      "epoch": 0.4252091735387949,
      "grad_norm": 1.8665834665298462,
      "learning_rate": 0.00012890716180502564,
      "loss": 0.8789,
      "step": 883
    },
    {
      "epoch": 0.4256907241317041,
      "grad_norm": 1.9203345775604248,
      "learning_rate": 0.00012875779848799078,
      "loss": 1.2982,
      "step": 884
    },
    {
      "epoch": 0.4261722747246133,
      "grad_norm": 1.531518816947937,
      "learning_rate": 0.00012860836519701063,
      "loss": 1.3991,
      "step": 885
    },
    {
      "epoch": 0.42665382531752244,
      "grad_norm": 1.7564634084701538,
      "learning_rate": 0.00012845886229568873,
      "loss": 1.0466,
      "step": 886
    },
    {
      "epoch": 0.4271353759104316,
      "grad_norm": 1.7126537561416626,
      "learning_rate": 0.00012830929014779797,
      "loss": 1.2967,
      "step": 887
    },
    {
      "epoch": 0.42761692650334077,
      "grad_norm": 1.6345220804214478,
      "learning_rate": 0.0001281596491172797,
      "loss": 0.7816,
      "step": 888
    },
    {
      "epoch": 0.42809847709624993,
      "grad_norm": 1.9738909006118774,
      "learning_rate": 0.00012800993956824303,
      "loss": 1.8702,
      "step": 889
    },
    {
      "epoch": 0.4285800276891591,
      "grad_norm": 1.4219521284103394,
      "learning_rate": 0.00012786016186496358,
      "loss": 1.0966,
      "step": 890
    },
    {
      "epoch": 0.42906157828206826,
      "grad_norm": 1.433343768119812,
      "learning_rate": 0.000127710316371883,
      "loss": 0.7067,
      "step": 891
    },
    {
      "epoch": 0.4295431288749774,
      "grad_norm": 1.803466796875,
      "learning_rate": 0.0001275604034536077,
      "loss": 0.6218,
      "step": 892
    },
    {
      "epoch": 0.4300246794678866,
      "grad_norm": 1.43092679977417,
      "learning_rate": 0.0001274104234749083,
      "loss": 1.0274,
      "step": 893
    },
    {
      "epoch": 0.43050623006079575,
      "grad_norm": 0.7985690236091614,
      "learning_rate": 0.00012726037680071853,
      "loss": 1.0084,
      "step": 894
    },
    {
      "epoch": 0.4309877806537049,
      "grad_norm": 2.045116424560547,
      "learning_rate": 0.00012711026379613434,
      "loss": 1.0859,
      "step": 895
    },
    {
      "epoch": 0.4314693312466141,
      "grad_norm": 0.8886361122131348,
      "learning_rate": 0.00012696008482641325,
      "loss": 1.0119,
      "step": 896
    },
    {
      "epoch": 0.43195088183952324,
      "grad_norm": 1.2930041551589966,
      "learning_rate": 0.00012680984025697313,
      "loss": 0.7904,
      "step": 897
    },
    {
      "epoch": 0.43243243243243246,
      "grad_norm": 2.3625149726867676,
      "learning_rate": 0.00012665953045339152,
      "loss": 1.2138,
      "step": 898
    },
    {
      "epoch": 0.4329139830253416,
      "grad_norm": 1.3330636024475098,
      "learning_rate": 0.0001265091557814047,
      "loss": 1.0498,
      "step": 899
    },
    {
      "epoch": 0.4333955336182508,
      "grad_norm": 2.4439661502838135,
      "learning_rate": 0.00012635871660690676,
      "loss": 0.6654,
      "step": 900
    },
    {
      "epoch": 0.43387708421115995,
      "grad_norm": 1.3480130434036255,
      "learning_rate": 0.0001262082132959488,
      "loss": 1.1451,
      "step": 901
    },
    {
      "epoch": 0.4343586348040691,
      "grad_norm": 2.0252513885498047,
      "learning_rate": 0.00012605764621473792,
      "loss": 1.1849,
      "step": 902
    },
    {
      "epoch": 0.4348401853969783,
      "grad_norm": 1.2678189277648926,
      "learning_rate": 0.00012590701572963642,
      "loss": 1.2898,
      "step": 903
    },
    {
      "epoch": 0.43532173598988744,
      "grad_norm": 1.384534239768982,
      "learning_rate": 0.00012575632220716078,
      "loss": 1.1128,
      "step": 904
    },
    {
      "epoch": 0.4358032865827966,
      "grad_norm": 1.5451288223266602,
      "learning_rate": 0.000125605566013981,
      "loss": 1.2512,
      "step": 905
    },
    {
      "epoch": 0.43628483717570576,
      "grad_norm": 1.8414585590362549,
      "learning_rate": 0.00012545474751691953,
      "loss": 1.3756,
      "step": 906
    },
    {
      "epoch": 0.4367663877686149,
      "grad_norm": 1.4265975952148438,
      "learning_rate": 0.00012530386708295036,
      "loss": 0.9855,
      "step": 907
    },
    {
      "epoch": 0.4372479383615241,
      "grad_norm": 2.019200086593628,
      "learning_rate": 0.00012515292507919829,
      "loss": 1.2213,
      "step": 908
    },
    {
      "epoch": 0.43772948895443325,
      "grad_norm": 1.0109248161315918,
      "learning_rate": 0.0001250019218729378,
      "loss": 0.7004,
      "step": 909
    },
    {
      "epoch": 0.4382110395473424,
      "grad_norm": 1.3910928964614868,
      "learning_rate": 0.00012485085783159238,
      "loss": 0.6213,
      "step": 910
    },
    {
      "epoch": 0.43869259014025164,
      "grad_norm": 2.540363311767578,
      "learning_rate": 0.00012469973332273354,
      "loss": 1.1664,
      "step": 911
    },
    {
      "epoch": 0.4391741407331608,
      "grad_norm": 0.8777474164962769,
      "learning_rate": 0.00012454854871407994,
      "loss": 0.7135,
      "step": 912
    },
    {
      "epoch": 0.43965569132606996,
      "grad_norm": 1.3600534200668335,
      "learning_rate": 0.00012439730437349635,
      "loss": 0.878,
      "step": 913
    },
    {
      "epoch": 0.4401372419189791,
      "grad_norm": 1.3686996698379517,
      "learning_rate": 0.00012424600066899302,
      "loss": 1.0881,
      "step": 914
    },
    {
      "epoch": 0.4406187925118883,
      "grad_norm": 1.5576976537704468,
      "learning_rate": 0.00012409463796872464,
      "loss": 1.0043,
      "step": 915
    },
    {
      "epoch": 0.44110034310479745,
      "grad_norm": 1.4311184883117676,
      "learning_rate": 0.0001239432166409893,
      "loss": 1.1763,
      "step": 916
    },
    {
      "epoch": 0.4415818936977066,
      "grad_norm": 1.2471023797988892,
      "learning_rate": 0.00012379173705422795,
      "loss": 1.2249,
      "step": 917
    },
    {
      "epoch": 0.4420634442906158,
      "grad_norm": 1.7952516078948975,
      "learning_rate": 0.00012364019957702315,
      "loss": 0.9334,
      "step": 918
    },
    {
      "epoch": 0.44254499488352494,
      "grad_norm": 1.515869379043579,
      "learning_rate": 0.00012348860457809838,
      "loss": 1.5403,
      "step": 919
    },
    {
      "epoch": 0.4430265454764341,
      "grad_norm": 1.0837361812591553,
      "learning_rate": 0.00012333695242631705,
      "loss": 1.2492,
      "step": 920
    },
    {
      "epoch": 0.44350809606934327,
      "grad_norm": 2.0795459747314453,
      "learning_rate": 0.0001231852434906817,
      "loss": 1.802,
      "step": 921
    },
    {
      "epoch": 0.44398964666225244,
      "grad_norm": 1.2322829961776733,
      "learning_rate": 0.00012303347814033292,
      "loss": 1.0496,
      "step": 922
    },
    {
      "epoch": 0.4444711972551616,
      "grad_norm": 1.3740915060043335,
      "learning_rate": 0.0001228816567445487,
      "loss": 1.2166,
      "step": 923
    },
    {
      "epoch": 0.44495274784807076,
      "grad_norm": 1.1164416074752808,
      "learning_rate": 0.0001227297796727433,
      "loss": 0.9671,
      "step": 924
    },
    {
      "epoch": 0.44543429844098,
      "grad_norm": 1.1179203987121582,
      "learning_rate": 0.00012257784729446656,
      "loss": 1.0126,
      "step": 925
    },
    {
      "epoch": 0.44591584903388914,
      "grad_norm": 1.0342260599136353,
      "learning_rate": 0.00012242585997940275,
      "loss": 1.2151,
      "step": 926
    },
    {
      "epoch": 0.4463973996267983,
      "grad_norm": 1.726683497428894,
      "learning_rate": 0.0001222738180973699,
      "loss": 0.8098,
      "step": 927
    },
    {
      "epoch": 0.44687895021970747,
      "grad_norm": 1.563063144683838,
      "learning_rate": 0.00012212172201831885,
      "loss": 0.8111,
      "step": 928
    },
    {
      "epoch": 0.44736050081261663,
      "grad_norm": 2.0912868976593018,
      "learning_rate": 0.00012196957211233222,
      "loss": 1.5385,
      "step": 929
    },
    {
      "epoch": 0.4478420514055258,
      "grad_norm": 1.1043822765350342,
      "learning_rate": 0.00012181736874962371,
      "loss": 1.3513,
      "step": 930
    },
    {
      "epoch": 0.44832360199843496,
      "grad_norm": 1.2951503992080688,
      "learning_rate": 0.00012166511230053696,
      "loss": 1.4313,
      "step": 931
    },
    {
      "epoch": 0.4488051525913441,
      "grad_norm": 0.8511658310890198,
      "learning_rate": 0.00012151280313554486,
      "loss": 1.1797,
      "step": 932
    },
    {
      "epoch": 0.4492867031842533,
      "grad_norm": 1.144500970840454,
      "learning_rate": 0.00012136044162524858,
      "loss": 1.4189,
      "step": 933
    },
    {
      "epoch": 0.44976825377716245,
      "grad_norm": 1.5903406143188477,
      "learning_rate": 0.00012120802814037663,
      "loss": 1.0242,
      "step": 934
    },
    {
      "epoch": 0.4502498043700716,
      "grad_norm": 1.4592664241790771,
      "learning_rate": 0.00012105556305178399,
      "loss": 1.3617,
      "step": 935
    },
    {
      "epoch": 0.4507313549629808,
      "grad_norm": 1.2225227355957031,
      "learning_rate": 0.00012090304673045123,
      "loss": 0.9173,
      "step": 936
    },
    {
      "epoch": 0.45121290555588994,
      "grad_norm": 1.441419243812561,
      "learning_rate": 0.00012075047954748353,
      "loss": 1.3376,
      "step": 937
    },
    {
      "epoch": 0.4516944561487991,
      "grad_norm": 1.6980377435684204,
      "learning_rate": 0.00012059786187410984,
      "loss": 1.1553,
      "step": 938
    },
    {
      "epoch": 0.4521760067417083,
      "grad_norm": 1.7701677083969116,
      "learning_rate": 0.000120445194081682,
      "loss": 1.1015,
      "step": 939
    },
    {
      "epoch": 0.4526575573346175,
      "grad_norm": 1.1329143047332764,
      "learning_rate": 0.00012029247654167379,
      "loss": 0.6795,
      "step": 940
    },
    {
      "epoch": 0.45313910792752665,
      "grad_norm": 2.149193525314331,
      "learning_rate": 0.00012013970962568002,
      "loss": 0.909,
      "step": 941
    },
    {
      "epoch": 0.4536206585204358,
      "grad_norm": 1.3953691720962524,
      "learning_rate": 0.00011998689370541562,
      "loss": 1.4807,
      "step": 942
    },
    {
      "epoch": 0.454102209113345,
      "grad_norm": 0.9279965758323669,
      "learning_rate": 0.00011983402915271478,
      "loss": 0.7435,
      "step": 943
    },
    {
      "epoch": 0.45458375970625414,
      "grad_norm": 2.231978178024292,
      "learning_rate": 0.00011968111633953007,
      "loss": 1.2841,
      "step": 944
    },
    {
      "epoch": 0.4550653102991633,
      "grad_norm": 1.15799880027771,
      "learning_rate": 0.0001195281556379314,
      "loss": 1.2933,
      "step": 945
    },
    {
      "epoch": 0.45554686089207247,
      "grad_norm": 0.8089935183525085,
      "learning_rate": 0.0001193751474201053,
      "loss": 0.7796,
      "step": 946
    },
    {
      "epoch": 0.45602841148498163,
      "grad_norm": 1.2958236932754517,
      "learning_rate": 0.00011922209205835382,
      "loss": 0.9262,
      "step": 947
    },
    {
      "epoch": 0.4565099620778908,
      "grad_norm": 1.0531879663467407,
      "learning_rate": 0.0001190689899250938,
      "loss": 1.3311,
      "step": 948
    },
    {
      "epoch": 0.45699151267079996,
      "grad_norm": 1.1241052150726318,
      "learning_rate": 0.00011891584139285582,
      "loss": 1.335,
      "step": 949
    },
    {
      "epoch": 0.4574730632637091,
      "grad_norm": 0.8808832168579102,
      "learning_rate": 0.00011876264683428344,
      "loss": 0.7111,
      "step": 950
    },
    {
      "epoch": 0.4579546138566183,
      "grad_norm": 1.5973955392837524,
      "learning_rate": 0.00011860940662213211,
      "loss": 0.5336,
      "step": 951
    },
    {
      "epoch": 0.4584361644495275,
      "grad_norm": 1.3471709489822388,
      "learning_rate": 0.00011845612112926843,
      "loss": 0.8771,
      "step": 952
    },
    {
      "epoch": 0.45891771504243667,
      "grad_norm": 1.8912404775619507,
      "learning_rate": 0.00011830279072866921,
      "loss": 1.1764,
      "step": 953
    },
    {
      "epoch": 0.45939926563534583,
      "grad_norm": 0.7309211492538452,
      "learning_rate": 0.00011814941579342044,
      "loss": 0.8934,
      "step": 954
    },
    {
      "epoch": 0.459880816228255,
      "grad_norm": 1.3789571523666382,
      "learning_rate": 0.00011799599669671654,
      "loss": 0.552,
      "step": 955
    },
    {
      "epoch": 0.46036236682116416,
      "grad_norm": 1.4538426399230957,
      "learning_rate": 0.00011784253381185937,
      "loss": 1.468,
      "step": 956
    },
    {
      "epoch": 0.4608439174140733,
      "grad_norm": 1.1952844858169556,
      "learning_rate": 0.0001176890275122573,
      "loss": 0.8286,
      "step": 957
    },
    {
      "epoch": 0.4613254680069825,
      "grad_norm": 0.9748961329460144,
      "learning_rate": 0.0001175354781714244,
      "loss": 0.9234,
      "step": 958
    },
    {
      "epoch": 0.46180701859989165,
      "grad_norm": 1.5134625434875488,
      "learning_rate": 0.0001173818861629794,
      "loss": 1.2918,
      "step": 959
    },
    {
      "epoch": 0.4622885691928008,
      "grad_norm": 1.4657390117645264,
      "learning_rate": 0.00011722825186064494,
      "loss": 0.6411,
      "step": 960
    },
    {
      "epoch": 0.46277011978571,
      "grad_norm": 0.8036825060844421,
      "learning_rate": 0.00011707457563824646,
      "loss": 1.0292,
      "step": 961
    },
    {
      "epoch": 0.46325167037861914,
      "grad_norm": 1.5182275772094727,
      "learning_rate": 0.00011692085786971149,
      "loss": 0.9529,
      "step": 962
    },
    {
      "epoch": 0.4637332209715283,
      "grad_norm": 2.0873656272888184,
      "learning_rate": 0.00011676709892906858,
      "loss": 0.9657,
      "step": 963
    },
    {
      "epoch": 0.46421477156443747,
      "grad_norm": 1.3092433214187622,
      "learning_rate": 0.00011661329919044656,
      "loss": 1.224,
      "step": 964
    },
    {
      "epoch": 0.46469632215734663,
      "grad_norm": 1.299930453300476,
      "learning_rate": 0.00011645945902807341,
      "loss": 0.6343,
      "step": 965
    },
    {
      "epoch": 0.46517787275025585,
      "grad_norm": 1.2251957654953003,
      "learning_rate": 0.00011630557881627553,
      "loss": 1.3129,
      "step": 966
    },
    {
      "epoch": 0.465659423343165,
      "grad_norm": 1.0462567806243896,
      "learning_rate": 0.0001161516589294768,
      "loss": 1.2948,
      "step": 967
    },
    {
      "epoch": 0.4661409739360742,
      "grad_norm": 1.2515294551849365,
      "learning_rate": 0.00011599769974219757,
      "loss": 1.1162,
      "step": 968
    },
    {
      "epoch": 0.46662252452898334,
      "grad_norm": 1.5011955499649048,
      "learning_rate": 0.0001158437016290539,
      "loss": 1.3869,
      "step": 969
    },
    {
      "epoch": 0.4671040751218925,
      "grad_norm": 2.204432487487793,
      "learning_rate": 0.00011568966496475649,
      "loss": 0.9236,
      "step": 970
    },
    {
      "epoch": 0.46758562571480167,
      "grad_norm": 1.0778597593307495,
      "learning_rate": 0.00011553559012410984,
      "loss": 1.3967,
      "step": 971
    },
    {
      "epoch": 0.46806717630771083,
      "grad_norm": 2.130934238433838,
      "learning_rate": 0.00011538147748201138,
      "loss": 1.5885,
      "step": 972
    },
    {
      "epoch": 0.46854872690062,
      "grad_norm": 0.8364808559417725,
      "learning_rate": 0.00011522732741345053,
      "loss": 1.039,
      "step": 973
    },
    {
      "epoch": 0.46903027749352916,
      "grad_norm": 1.7570946216583252,
      "learning_rate": 0.00011507314029350776,
      "loss": 1.2047,
      "step": 974
    },
    {
      "epoch": 0.4695118280864383,
      "grad_norm": 0.9198819398880005,
      "learning_rate": 0.00011491891649735366,
      "loss": 1.003,
      "step": 975
    },
    {
      "epoch": 0.4699933786793475,
      "grad_norm": 1.9497737884521484,
      "learning_rate": 0.00011476465640024814,
      "loss": 1.8117,
      "step": 976
    },
    {
      "epoch": 0.47047492927225665,
      "grad_norm": 0.9239754676818848,
      "learning_rate": 0.00011461036037753934,
      "loss": 0.464,
      "step": 977
    },
    {
      "epoch": 0.4709564798651658,
      "grad_norm": 0.8020768761634827,
      "learning_rate": 0.00011445602880466288,
      "loss": 0.5012,
      "step": 978
    },
    {
      "epoch": 0.471438030458075,
      "grad_norm": 1.4077554941177368,
      "learning_rate": 0.00011430166205714088,
      "loss": 1.256,
      "step": 979
    },
    {
      "epoch": 0.4719195810509842,
      "grad_norm": 0.9226470589637756,
      "learning_rate": 0.00011414726051058102,
      "loss": 0.8578,
      "step": 980
    },
    {
      "epoch": 0.47240113164389336,
      "grad_norm": 1.6840845346450806,
      "learning_rate": 0.0001139928245406757,
      "loss": 1.1237,
      "step": 981
    },
    {
      "epoch": 0.4728826822368025,
      "grad_norm": 0.9278067946434021,
      "learning_rate": 0.00011383835452320097,
      "loss": 1.0799,
      "step": 982
    },
    {
      "epoch": 0.4733642328297117,
      "grad_norm": 1.3527857065200806,
      "learning_rate": 0.00011368385083401585,
      "loss": 1.6294,
      "step": 983
    },
    {
      "epoch": 0.47384578342262085,
      "grad_norm": 1.7791368961334229,
      "learning_rate": 0.00011352931384906125,
      "loss": 0.8841,
      "step": 984
    },
    {
      "epoch": 0.47432733401553,
      "grad_norm": 1.3228471279144287,
      "learning_rate": 0.00011337474394435908,
      "loss": 0.9675,
      "step": 985
    },
    {
      "epoch": 0.4748088846084392,
      "grad_norm": 0.7633442282676697,
      "learning_rate": 0.00011322014149601136,
      "loss": 1.043,
      "step": 986
    },
    {
      "epoch": 0.47529043520134834,
      "grad_norm": 2.0492193698883057,
      "learning_rate": 0.00011306550688019926,
      "loss": 1.5654,
      "step": 987
    },
    {
      "epoch": 0.4757719857942575,
      "grad_norm": 1.4635462760925293,
      "learning_rate": 0.0001129108404731823,
      "loss": 1.14,
      "step": 988
    },
    {
      "epoch": 0.47625353638716666,
      "grad_norm": 0.9629092812538147,
      "learning_rate": 0.0001127561426512973,
      "loss": 0.9561,
      "step": 989
    },
    {
      "epoch": 0.47673508698007583,
      "grad_norm": 1.4097026586532593,
      "learning_rate": 0.0001126014137909575,
      "loss": 1.1852,
      "step": 990
    },
    {
      "epoch": 0.477216637572985,
      "grad_norm": 1.385420799255371,
      "learning_rate": 0.00011244665426865174,
      "loss": 1.0442,
      "step": 991
    },
    {
      "epoch": 0.47769818816589416,
      "grad_norm": 1.2012741565704346,
      "learning_rate": 0.00011229186446094338,
      "loss": 0.8843,
      "step": 992
    },
    {
      "epoch": 0.4781797387588034,
      "grad_norm": 1.152305245399475,
      "learning_rate": 0.00011213704474446951,
      "loss": 0.9383,
      "step": 993
    },
    {
      "epoch": 0.47866128935171254,
      "grad_norm": 1.8375195264816284,
      "learning_rate": 0.00011198219549594,
      "loss": 1.1666,
      "step": 994
    },
    {
      "epoch": 0.4791428399446217,
      "grad_norm": 1.6210880279541016,
      "learning_rate": 0.00011182731709213659,
      "loss": 0.9919,
      "step": 995
    },
    {
      "epoch": 0.47962439053753086,
      "grad_norm": 1.3899977207183838,
      "learning_rate": 0.00011167240990991192,
      "loss": 0.9018,
      "step": 996
    },
    {
      "epoch": 0.48010594113044003,
      "grad_norm": 0.9364240169525146,
      "learning_rate": 0.00011151747432618871,
      "loss": 0.8456,
      "step": 997
    },
    {
      "epoch": 0.4805874917233492,
      "grad_norm": 1.5732064247131348,
      "learning_rate": 0.00011136251071795871,
      "loss": 0.986,
      "step": 998
    },
    {
      "epoch": 0.48106904231625836,
      "grad_norm": 1.1362043619155884,
      "learning_rate": 0.00011120751946228197,
      "loss": 1.403,
      "step": 999
    },
    {
      "epoch": 0.4815505929091675,
      "grad_norm": 1.0300402641296387,
      "learning_rate": 0.00011105250093628565,
      "loss": 1.0805,
      "step": 1000
    },
    {
      "epoch": 0.4820321435020767,
      "grad_norm": 0.5833719372749329,
      "learning_rate": 0.00011089745551716344,
      "loss": 0.9848,
      "step": 1001
    },
    {
      "epoch": 0.48251369409498585,
      "grad_norm": 1.285022258758545,
      "learning_rate": 0.00011074238358217437,
      "loss": 1.4237,
      "step": 1002
    },
    {
      "epoch": 0.482995244687895,
      "grad_norm": 1.4532897472381592,
      "learning_rate": 0.00011058728550864197,
      "loss": 0.7092,
      "step": 1003
    },
    {
      "epoch": 0.4834767952808042,
      "grad_norm": 1.2283052206039429,
      "learning_rate": 0.00011043216167395344,
      "loss": 1.4935,
      "step": 1004
    },
    {
      "epoch": 0.48395834587371334,
      "grad_norm": 1.4048994779586792,
      "learning_rate": 0.00011027701245555865,
      "loss": 1.2185,
      "step": 1005
    },
    {
      "epoch": 0.4844398964666225,
      "grad_norm": 2.113971710205078,
      "learning_rate": 0.00011012183823096917,
      "loss": 1.0714,
      "step": 1006
    },
    {
      "epoch": 0.4849214470595317,
      "grad_norm": 2.0666840076446533,
      "learning_rate": 0.00010996663937775751,
      "loss": 0.9659,
      "step": 1007
    },
    {
      "epoch": 0.4854029976524409,
      "grad_norm": 1.1015936136245728,
      "learning_rate": 0.000109811416273556,
      "loss": 1.4468,
      "step": 1008
    },
    {
      "epoch": 0.48588454824535005,
      "grad_norm": 1.0336759090423584,
      "learning_rate": 0.00010965616929605609,
      "loss": 1.0347,
      "step": 1009
    },
    {
      "epoch": 0.4863660988382592,
      "grad_norm": 1.7983458042144775,
      "learning_rate": 0.0001095008988230072,
      "loss": 1.485,
      "step": 1010
    },
    {
      "epoch": 0.48684764943116837,
      "grad_norm": 1.6773706674575806,
      "learning_rate": 0.00010934560523221602,
      "loss": 1.1997,
      "step": 1011
    },
    {
      "epoch": 0.48732920002407754,
      "grad_norm": 1.3664817810058594,
      "learning_rate": 0.00010919028890154543,
      "loss": 0.9223,
      "step": 1012
    },
    {
      "epoch": 0.4878107506169867,
      "grad_norm": 1.6062132120132446,
      "learning_rate": 0.00010903495020891375,
      "loss": 1.3774,
      "step": 1013
    },
    {
      "epoch": 0.48829230120989586,
      "grad_norm": 1.1740654706954956,
      "learning_rate": 0.00010887958953229349,
      "loss": 1.3075,
      "step": 1014
    },
    {
      "epoch": 0.488773851802805,
      "grad_norm": 1.327466607093811,
      "learning_rate": 0.00010872420724971088,
      "loss": 1.3444,
      "step": 1015
    },
    {
      "epoch": 0.4892554023957142,
      "grad_norm": 1.3187872171401978,
      "learning_rate": 0.0001085688037392446,
      "loss": 0.9842,
      "step": 1016
    },
    {
      "epoch": 0.48973695298862335,
      "grad_norm": 0.9917089939117432,
      "learning_rate": 0.000108413379379025,
      "loss": 0.942,
      "step": 1017
    },
    {
      "epoch": 0.4902185035815325,
      "grad_norm": 0.9732362031936646,
      "learning_rate": 0.00010825793454723325,
      "loss": 0.9424,
      "step": 1018
    },
    {
      "epoch": 0.4907000541744417,
      "grad_norm": 0.9644367098808289,
      "learning_rate": 0.00010810246962210018,
      "loss": 1.2421,
      "step": 1019
    },
    {
      "epoch": 0.49118160476735084,
      "grad_norm": 0.9130355715751648,
      "learning_rate": 0.00010794698498190557,
      "loss": 1.2625,
      "step": 1020
    },
    {
      "epoch": 0.49166315536026006,
      "grad_norm": 1.699233889579773,
      "learning_rate": 0.00010779148100497722,
      "loss": 0.9944,
      "step": 1021
    },
    {
      "epoch": 0.4921447059531692,
      "grad_norm": 1.1781898736953735,
      "learning_rate": 0.00010763595806968996,
      "loss": 1.4611,
      "step": 1022
    },
    {
      "epoch": 0.4926262565460784,
      "grad_norm": 1.6606625318527222,
      "learning_rate": 0.00010748041655446473,
      "loss": 1.5016,
      "step": 1023
    },
    {
      "epoch": 0.49310780713898755,
      "grad_norm": 1.660539984703064,
      "learning_rate": 0.00010732485683776768,
      "loss": 1.3722,
      "step": 1024
    },
    {
      "epoch": 0.4935893577318967,
      "grad_norm": 0.8509052991867065,
      "learning_rate": 0.00010716927929810925,
      "loss": 1.1455,
      "step": 1025
    },
    {
      "epoch": 0.4940709083248059,
      "grad_norm": 6.157974720001221,
      "learning_rate": 0.00010701368431404326,
      "loss": 1.1335,
      "step": 1026
    },
    {
      "epoch": 0.49455245891771504,
      "grad_norm": 2.522059679031372,
      "learning_rate": 0.00010685807226416598,
      "loss": 0.9916,
      "step": 1027
    },
    {
      "epoch": 0.4950340095106242,
      "grad_norm": 2.0645813941955566,
      "learning_rate": 0.00010670244352711518,
      "loss": 1.1143,
      "step": 1028
    },
    {
      "epoch": 0.49551556010353337,
      "grad_norm": 1.5876270532608032,
      "learning_rate": 0.00010654679848156925,
      "loss": 0.9889,
      "step": 1029
    },
    {
      "epoch": 0.49599711069644253,
      "grad_norm": 1.8583619594573975,
      "learning_rate": 0.00010639113750624625,
      "loss": 1.2385,
      "step": 1030
    },
    {
      "epoch": 0.4964786612893517,
      "grad_norm": 1.1722699403762817,
      "learning_rate": 0.00010623546097990303,
      "loss": 1.5121,
      "step": 1031
    },
    {
      "epoch": 0.49696021188226086,
      "grad_norm": 1.5148611068725586,
      "learning_rate": 0.00010607976928133423,
      "loss": 0.7874,
      "step": 1032
    },
    {
      "epoch": 0.49744176247517,
      "grad_norm": 1.096032977104187,
      "learning_rate": 0.00010592406278937144,
      "loss": 0.8614,
      "step": 1033
    },
    {
      "epoch": 0.4979233130680792,
      "grad_norm": 1.2312672138214111,
      "learning_rate": 0.00010576834188288226,
      "loss": 1.4232,
      "step": 1034
    },
    {
      "epoch": 0.4984048636609884,
      "grad_norm": 1.202398657798767,
      "learning_rate": 0.00010561260694076935,
      "loss": 1.1642,
      "step": 1035
    },
    {
      "epoch": 0.49888641425389757,
      "grad_norm": 1.2833552360534668,
      "learning_rate": 0.00010545685834196948,
      "loss": 1.2148,
      "step": 1036
    },
    {
      "epoch": 0.49936796484680673,
      "grad_norm": 1.5922613143920898,
      "learning_rate": 0.00010530109646545272,
      "loss": 1.2816,
      "step": 1037
    },
    {
      "epoch": 0.4998495154397159,
      "grad_norm": 1.836709976196289,
      "learning_rate": 0.0001051453216902214,
      "loss": 1.376,
      "step": 1038
    },
    {
      "epoch": 0.500331066032625,
      "grad_norm": 1.5578300952911377,
      "learning_rate": 0.00010498953439530925,
      "loss": 1.2194,
      "step": 1039
    },
    {
      "epoch": 0.5008126166255342,
      "grad_norm": 0.5676748156547546,
      "learning_rate": 0.00010483373495978046,
      "loss": 0.9552,
      "step": 1040
    },
    {
      "epoch": 0.5012941672184434,
      "grad_norm": 1.2331892251968384,
      "learning_rate": 0.00010467792376272877,
      "loss": 1.1766,
      "step": 1041
    },
    {
      "epoch": 0.5017757178113526,
      "grad_norm": 1.350245714187622,
      "learning_rate": 0.00010452210118327652,
      "loss": 0.7175,
      "step": 1042
    },
    {
      "epoch": 0.5022572684042618,
      "grad_norm": 1.5291863679885864,
      "learning_rate": 0.00010436626760057378,
      "loss": 1.4097,
      "step": 1043
    },
    {
      "epoch": 0.5027388189971709,
      "grad_norm": 0.8576416373252869,
      "learning_rate": 0.00010421042339379732,
      "loss": 0.7222,
      "step": 1044
    },
    {
      "epoch": 0.5032203695900801,
      "grad_norm": 1.3233412504196167,
      "learning_rate": 0.00010405456894214987,
      "loss": 1.0909,
      "step": 1045
    },
    {
      "epoch": 0.5037019201829892,
      "grad_norm": 0.9909217953681946,
      "learning_rate": 0.00010389870462485902,
      "loss": 1.6271,
      "step": 1046
    },
    {
      "epoch": 0.5041834707758984,
      "grad_norm": 1.3215575218200684,
      "learning_rate": 0.00010374283082117635,
      "loss": 0.8316,
      "step": 1047
    },
    {
      "epoch": 0.5046650213688075,
      "grad_norm": 1.4400215148925781,
      "learning_rate": 0.00010358694791037653,
      "loss": 1.3994,
      "step": 1048
    },
    {
      "epoch": 0.5051465719617168,
      "grad_norm": 1.7992037534713745,
      "learning_rate": 0.00010343105627175644,
      "loss": 1.3219,
      "step": 1049
    },
    {
      "epoch": 0.5056281225546259,
      "grad_norm": 1.9158211946487427,
      "learning_rate": 0.00010327515628463415,
      "loss": 0.9047,
      "step": 1050
    },
    {
      "epoch": 0.5061096731475351,
      "grad_norm": 0.8823553323745728,
      "learning_rate": 0.00010311924832834808,
      "loss": 1.431,
      "step": 1051
    },
    {
      "epoch": 0.5065912237404442,
      "grad_norm": 0.9347987174987793,
      "learning_rate": 0.00010296333278225599,
      "loss": 0.7961,
      "step": 1052
    },
    {
      "epoch": 0.5070727743333534,
      "grad_norm": 2.882472038269043,
      "learning_rate": 0.00010280741002573413,
      "loss": 0.8619,
      "step": 1053
    },
    {
      "epoch": 0.5075543249262625,
      "grad_norm": 0.8229369521141052,
      "learning_rate": 0.00010265148043817632,
      "loss": 1.1733,
      "step": 1054
    },
    {
      "epoch": 0.5080358755191717,
      "grad_norm": 0.9884060025215149,
      "learning_rate": 0.00010249554439899298,
      "loss": 0.9507,
      "step": 1055
    },
    {
      "epoch": 0.508517426112081,
      "grad_norm": 0.918043851852417,
      "learning_rate": 0.00010233960228761022,
      "loss": 0.9308,
      "step": 1056
    },
    {
      "epoch": 0.5089989767049901,
      "grad_norm": 1.5720266103744507,
      "learning_rate": 0.00010218365448346893,
      "loss": 1.2772,
      "step": 1057
    },
    {
      "epoch": 0.5094805272978993,
      "grad_norm": 2.289980411529541,
      "learning_rate": 0.00010202770136602388,
      "loss": 1.4883,
      "step": 1058
    },
    {
      "epoch": 0.5099620778908084,
      "grad_norm": 0.9244562983512878,
      "learning_rate": 0.00010187174331474271,
      "loss": 0.8974,
      "step": 1059
    },
    {
      "epoch": 0.5104436284837176,
      "grad_norm": 1.3836133480072021,
      "learning_rate": 0.00010171578070910512,
      "loss": 0.7695,
      "step": 1060
    },
    {
      "epoch": 0.5109251790766267,
      "grad_norm": 0.9162578582763672,
      "learning_rate": 0.00010155981392860185,
      "loss": 1.2861,
      "step": 1061
    },
    {
      "epoch": 0.5114067296695359,
      "grad_norm": 1.125784158706665,
      "learning_rate": 0.00010140384335273386,
      "loss": 0.9343,
      "step": 1062
    },
    {
      "epoch": 0.511888280262445,
      "grad_norm": 1.069495677947998,
      "learning_rate": 0.00010124786936101127,
      "loss": 0.9909,
      "step": 1063
    },
    {
      "epoch": 0.5123698308553543,
      "grad_norm": 1.0141994953155518,
      "learning_rate": 0.00010109189233295255,
      "loss": 0.8909,
      "step": 1064
    },
    {
      "epoch": 0.5128513814482634,
      "grad_norm": 0.9358717203140259,
      "learning_rate": 0.00010093591264808358,
      "loss": 0.9834,
      "step": 1065
    },
    {
      "epoch": 0.5133329320411726,
      "grad_norm": 1.5457875728607178,
      "learning_rate": 0.00010077993068593663,
      "loss": 1.261,
      "step": 1066
    },
    {
      "epoch": 0.5138144826340817,
      "grad_norm": 1.4648905992507935,
      "learning_rate": 0.00010062394682604963,
      "loss": 1.3533,
      "step": 1067
    },
    {
      "epoch": 0.5142960332269909,
      "grad_norm": 1.1549569368362427,
      "learning_rate": 0.00010046796144796497,
      "loss": 0.8969,
      "step": 1068
    },
    {
      "epoch": 0.5147775838199001,
      "grad_norm": 1.5365687608718872,
      "learning_rate": 0.0001003119749312289,
      "loss": 0.9088,
      "step": 1069
    },
    {
      "epoch": 0.5152591344128092,
      "grad_norm": 1.3011575937271118,
      "learning_rate": 0.00010015598765539031,
      "loss": 1.6479,
      "step": 1070
    },
    {
      "epoch": 0.5157406850057185,
      "grad_norm": 1.0299460887908936,
      "learning_rate": 0.0001,
      "loss": 0.9833,
      "step": 1071
    },
    {
      "epoch": 0.5162222355986276,
      "grad_norm": 1.6121004819869995,
      "learning_rate": 9.984401234460971e-05,
      "loss": 1.0884,
      "step": 1072
    },
    {
      "epoch": 0.5167037861915368,
      "grad_norm": 1.3076612949371338,
      "learning_rate": 9.968802506877111e-05,
      "loss": 1.0745,
      "step": 1073
    },
    {
      "epoch": 0.5171853367844459,
      "grad_norm": 1.2089321613311768,
      "learning_rate": 9.953203855203504e-05,
      "loss": 0.7979,
      "step": 1074
    },
    {
      "epoch": 0.5176668873773551,
      "grad_norm": 1.5304687023162842,
      "learning_rate": 9.93760531739504e-05,
      "loss": 1.1467,
      "step": 1075
    },
    {
      "epoch": 0.5181484379702642,
      "grad_norm": 1.59665846824646,
      "learning_rate": 9.922006931406338e-05,
      "loss": 1.4846,
      "step": 1076
    },
    {
      "epoch": 0.5186299885631734,
      "grad_norm": 1.0475536584854126,
      "learning_rate": 9.906408735191643e-05,
      "loss": 0.7154,
      "step": 1077
    },
    {
      "epoch": 0.5191115391560825,
      "grad_norm": 1.07981538772583,
      "learning_rate": 9.890810766704745e-05,
      "loss": 0.8914,
      "step": 1078
    },
    {
      "epoch": 0.5195930897489918,
      "grad_norm": 3.572413682937622,
      "learning_rate": 9.875213063898875e-05,
      "loss": 1.0619,
      "step": 1079
    },
    {
      "epoch": 0.5200746403419009,
      "grad_norm": 1.268196702003479,
      "learning_rate": 9.859615664726615e-05,
      "loss": 1.1203,
      "step": 1080
    },
    {
      "epoch": 0.5205561909348101,
      "grad_norm": 1.727527379989624,
      "learning_rate": 9.844018607139818e-05,
      "loss": 1.3839,
      "step": 1081
    },
    {
      "epoch": 0.5210377415277193,
      "grad_norm": 1.8239974975585938,
      "learning_rate": 9.828421929089493e-05,
      "loss": 0.9565,
      "step": 1082
    },
    {
      "epoch": 0.5215192921206284,
      "grad_norm": 2.8285043239593506,
      "learning_rate": 9.812825668525733e-05,
      "loss": 1.4409,
      "step": 1083
    },
    {
      "epoch": 0.5220008427135376,
      "grad_norm": 1.9248615503311157,
      "learning_rate": 9.797229863397615e-05,
      "loss": 1.0256,
      "step": 1084
    },
    {
      "epoch": 0.5224823933064467,
      "grad_norm": 1.044853687286377,
      "learning_rate": 9.781634551653108e-05,
      "loss": 0.9295,
      "step": 1085
    },
    {
      "epoch": 0.522963943899356,
      "grad_norm": 1.4697555303573608,
      "learning_rate": 9.766039771238982e-05,
      "loss": 1.1206,
      "step": 1086
    },
    {
      "epoch": 0.5234454944922651,
      "grad_norm": 1.726250410079956,
      "learning_rate": 9.750445560100706e-05,
      "loss": 1.2159,
      "step": 1087
    },
    {
      "epoch": 0.5239270450851743,
      "grad_norm": 1.0951380729675293,
      "learning_rate": 9.73485195618237e-05,
      "loss": 1.3589,
      "step": 1088
    },
    {
      "epoch": 0.5244085956780834,
      "grad_norm": 0.8561838269233704,
      "learning_rate": 9.719258997426588e-05,
      "loss": 1.0009,
      "step": 1089
    },
    {
      "epoch": 0.5248901462709926,
      "grad_norm": 2.874856472015381,
      "learning_rate": 9.703666721774402e-05,
      "loss": 0.9104,
      "step": 1090
    },
    {
      "epoch": 0.5253716968639017,
      "grad_norm": 1.099172830581665,
      "learning_rate": 9.688075167165194e-05,
      "loss": 0.9832,
      "step": 1091
    },
    {
      "epoch": 0.525853247456811,
      "grad_norm": 1.1792478561401367,
      "learning_rate": 9.672484371536586e-05,
      "loss": 1.0788,
      "step": 1092
    },
    {
      "epoch": 0.52633479804972,
      "grad_norm": 1.5772693157196045,
      "learning_rate": 9.656894372824358e-05,
      "loss": 1.4684,
      "step": 1093
    },
    {
      "epoch": 0.5268163486426293,
      "grad_norm": 1.3587925434112549,
      "learning_rate": 9.64130520896235e-05,
      "loss": 1.211,
      "step": 1094
    },
    {
      "epoch": 0.5272978992355384,
      "grad_norm": 1.0316194295883179,
      "learning_rate": 9.625716917882367e-05,
      "loss": 0.7338,
      "step": 1095
    },
    {
      "epoch": 0.5277794498284476,
      "grad_norm": 1.256105899810791,
      "learning_rate": 9.6101295375141e-05,
      "loss": 1.2281,
      "step": 1096
    },
    {
      "epoch": 0.5282610004213568,
      "grad_norm": 1.0951718091964722,
      "learning_rate": 9.594543105785013e-05,
      "loss": 1.0972,
      "step": 1097
    },
    {
      "epoch": 0.5287425510142659,
      "grad_norm": 1.3974488973617554,
      "learning_rate": 9.578957660620267e-05,
      "loss": 1.0854,
      "step": 1098
    },
    {
      "epoch": 0.5292241016071751,
      "grad_norm": 2.374807596206665,
      "learning_rate": 9.563373239942623e-05,
      "loss": 1.0925,
      "step": 1099
    },
    {
      "epoch": 0.5297056522000843,
      "grad_norm": 1.185179352760315,
      "learning_rate": 9.547789881672348e-05,
      "loss": 0.8359,
      "step": 1100
    },
    {
      "epoch": 0.5301872027929935,
      "grad_norm": 0.9710121154785156,
      "learning_rate": 9.532207623727126e-05,
      "loss": 1.2331,
      "step": 1101
    },
    {
      "epoch": 0.5306687533859026,
      "grad_norm": 1.3888357877731323,
      "learning_rate": 9.516626504021957e-05,
      "loss": 1.1026,
      "step": 1102
    },
    {
      "epoch": 0.5311503039788118,
      "grad_norm": 1.2671335935592651,
      "learning_rate": 9.501046560469079e-05,
      "loss": 1.2884,
      "step": 1103
    },
    {
      "epoch": 0.5316318545717209,
      "grad_norm": 2.3087527751922607,
      "learning_rate": 9.485467830977864e-05,
      "loss": 1.4861,
      "step": 1104
    },
    {
      "epoch": 0.5321134051646301,
      "grad_norm": 1.1949607133865356,
      "learning_rate": 9.469890353454732e-05,
      "loss": 1.0482,
      "step": 1105
    },
    {
      "epoch": 0.5325949557575392,
      "grad_norm": 2.516268253326416,
      "learning_rate": 9.454314165803054e-05,
      "loss": 1.1362,
      "step": 1106
    },
    {
      "epoch": 0.5330765063504485,
      "grad_norm": 1.080284595489502,
      "learning_rate": 9.438739305923067e-05,
      "loss": 0.8045,
      "step": 1107
    },
    {
      "epoch": 0.5335580569433576,
      "grad_norm": 1.2091320753097534,
      "learning_rate": 9.423165811711777e-05,
      "loss": 1.3451,
      "step": 1108
    },
    {
      "epoch": 0.5340396075362668,
      "grad_norm": 1.2272273302078247,
      "learning_rate": 9.407593721062859e-05,
      "loss": 1.0463,
      "step": 1109
    },
    {
      "epoch": 0.534521158129176,
      "grad_norm": 1.4450758695602417,
      "learning_rate": 9.39202307186658e-05,
      "loss": 0.8372,
      "step": 1110
    },
    {
      "epoch": 0.5350027087220851,
      "grad_norm": 0.5302134156227112,
      "learning_rate": 9.3764539020097e-05,
      "loss": 1.0282,
      "step": 1111
    },
    {
      "epoch": 0.5354842593149943,
      "grad_norm": 1.6732890605926514,
      "learning_rate": 9.360886249375376e-05,
      "loss": 1.4429,
      "step": 1112
    },
    {
      "epoch": 0.5359658099079034,
      "grad_norm": 0.5877620577812195,
      "learning_rate": 9.345320151843078e-05,
      "loss": 0.5685,
      "step": 1113
    },
    {
      "epoch": 0.5364473605008127,
      "grad_norm": 2.4183499813079834,
      "learning_rate": 9.329755647288485e-05,
      "loss": 1.4885,
      "step": 1114
    },
    {
      "epoch": 0.5369289110937218,
      "grad_norm": 2.160714626312256,
      "learning_rate": 9.314192773583403e-05,
      "loss": 1.3593,
      "step": 1115
    },
    {
      "epoch": 0.537410461686631,
      "grad_norm": 1.506280779838562,
      "learning_rate": 9.298631568595674e-05,
      "loss": 1.1713,
      "step": 1116
    },
    {
      "epoch": 0.5378920122795401,
      "grad_norm": 1.4294887781143188,
      "learning_rate": 9.283072070189075e-05,
      "loss": 1.2407,
      "step": 1117
    },
    {
      "epoch": 0.5383735628724493,
      "grad_norm": 1.1967908143997192,
      "learning_rate": 9.267514316223234e-05,
      "loss": 0.7592,
      "step": 1118
    },
    {
      "epoch": 0.5388551134653584,
      "grad_norm": 1.0222610235214233,
      "learning_rate": 9.251958344553528e-05,
      "loss": 1.1028,
      "step": 1119
    },
    {
      "epoch": 0.5393366640582676,
      "grad_norm": 1.0599137544631958,
      "learning_rate": 9.23640419303101e-05,
      "loss": 0.9036,
      "step": 1120
    },
    {
      "epoch": 0.5398182146511767,
      "grad_norm": 0.8100444674491882,
      "learning_rate": 9.220851899502283e-05,
      "loss": 1.0332,
      "step": 1121
    },
    {
      "epoch": 0.540299765244086,
      "grad_norm": 1.539902925491333,
      "learning_rate": 9.205301501809448e-05,
      "loss": 1.409,
      "step": 1122
    },
    {
      "epoch": 0.5407813158369951,
      "grad_norm": 0.8249147534370422,
      "learning_rate": 9.189753037789987e-05,
      "loss": 1.1157,
      "step": 1123
    },
    {
      "epoch": 0.5412628664299043,
      "grad_norm": 1.451399326324463,
      "learning_rate": 9.174206545276677e-05,
      "loss": 1.1991,
      "step": 1124
    },
    {
      "epoch": 0.5417444170228135,
      "grad_norm": 0.6443821787834167,
      "learning_rate": 9.158662062097501e-05,
      "loss": 1.1216,
      "step": 1125
    },
    {
      "epoch": 0.5422259676157226,
      "grad_norm": 0.6511193513870239,
      "learning_rate": 9.143119626075542e-05,
      "loss": 0.6412,
      "step": 1126
    },
    {
      "epoch": 0.5427075182086318,
      "grad_norm": 0.8463265895843506,
      "learning_rate": 9.127579275028914e-05,
      "loss": 1.4851,
      "step": 1127
    },
    {
      "epoch": 0.5431890688015409,
      "grad_norm": 1.6545872688293457,
      "learning_rate": 9.112041046770653e-05,
      "loss": 1.2703,
      "step": 1128
    },
    {
      "epoch": 0.5436706193944502,
      "grad_norm": 2.8932747840881348,
      "learning_rate": 9.096504979108629e-05,
      "loss": 1.3268,
      "step": 1129
    },
    {
      "epoch": 0.5441521699873593,
      "grad_norm": 1.284162998199463,
      "learning_rate": 9.080971109845458e-05,
      "loss": 1.2082,
      "step": 1130
    },
    {
      "epoch": 0.5446337205802685,
      "grad_norm": 2.37834095954895,
      "learning_rate": 9.0654394767784e-05,
      "loss": 0.6822,
      "step": 1131
    },
    {
      "epoch": 0.5451152711731776,
      "grad_norm": 1.9115866422653198,
      "learning_rate": 9.049910117699281e-05,
      "loss": 0.6613,
      "step": 1132
    },
    {
      "epoch": 0.5455968217660868,
      "grad_norm": 1.1233134269714355,
      "learning_rate": 9.034383070394393e-05,
      "loss": 1.1494,
      "step": 1133
    },
    {
      "epoch": 0.5460783723589959,
      "grad_norm": 1.2977545261383057,
      "learning_rate": 9.0188583726444e-05,
      "loss": 1.4927,
      "step": 1134
    },
    {
      "epoch": 0.5465599229519051,
      "grad_norm": 0.5862376689910889,
      "learning_rate": 9.00333606222425e-05,
      "loss": 0.7873,
      "step": 1135
    },
    {
      "epoch": 0.5470414735448142,
      "grad_norm": 1.458271861076355,
      "learning_rate": 8.987816176903082e-05,
      "loss": 0.9886,
      "step": 1136
    },
    {
      "epoch": 0.5475230241377235,
      "grad_norm": 1.4016250371932983,
      "learning_rate": 8.972298754444136e-05,
      "loss": 1.5082,
      "step": 1137
    },
    {
      "epoch": 0.5480045747306327,
      "grad_norm": 1.2283416986465454,
      "learning_rate": 8.956783832604654e-05,
      "loss": 0.7988,
      "step": 1138
    },
    {
      "epoch": 0.5484861253235418,
      "grad_norm": 1.0414175987243652,
      "learning_rate": 8.941271449135806e-05,
      "loss": 1.1961,
      "step": 1139
    },
    {
      "epoch": 0.548967675916451,
      "grad_norm": 1.3187246322631836,
      "learning_rate": 8.925761641782567e-05,
      "loss": 0.9317,
      "step": 1140
    },
    {
      "epoch": 0.5494492265093601,
      "grad_norm": 1.9464495182037354,
      "learning_rate": 8.910254448283659e-05,
      "loss": 0.7117,
      "step": 1141
    },
    {
      "epoch": 0.5499307771022693,
      "grad_norm": 1.4207645654678345,
      "learning_rate": 8.894749906371439e-05,
      "loss": 1.0866,
      "step": 1142
    },
    {
      "epoch": 0.5504123276951784,
      "grad_norm": 1.0139679908752441,
      "learning_rate": 8.87924805377181e-05,
      "loss": 0.9417,
      "step": 1143
    },
    {
      "epoch": 0.5508938782880877,
      "grad_norm": 1.6405439376831055,
      "learning_rate": 8.863748928204131e-05,
      "loss": 0.6549,
      "step": 1144
    },
    {
      "epoch": 0.5513754288809968,
      "grad_norm": 1.4043493270874023,
      "learning_rate": 8.848252567381131e-05,
      "loss": 0.7796,
      "step": 1145
    },
    {
      "epoch": 0.551856979473906,
      "grad_norm": 1.0969845056533813,
      "learning_rate": 8.83275900900881e-05,
      "loss": 1.0326,
      "step": 1146
    },
    {
      "epoch": 0.5523385300668151,
      "grad_norm": 1.8198158740997314,
      "learning_rate": 8.817268290786343e-05,
      "loss": 1.3857,
      "step": 1147
    },
    {
      "epoch": 0.5528200806597243,
      "grad_norm": 1.9464865922927856,
      "learning_rate": 8.801780450406002e-05,
      "loss": 0.8922,
      "step": 1148
    },
    {
      "epoch": 0.5533016312526334,
      "grad_norm": 1.639780044555664,
      "learning_rate": 8.786295525553053e-05,
      "loss": 0.8284,
      "step": 1149
    },
    {
      "epoch": 0.5537831818455426,
      "grad_norm": 2.057802438735962,
      "learning_rate": 8.770813553905664e-05,
      "loss": 0.9536,
      "step": 1150
    },
    {
      "epoch": 0.5542647324384519,
      "grad_norm": 2.0702240467071533,
      "learning_rate": 8.755334573134829e-05,
      "loss": 1.0337,
      "step": 1151
    },
    {
      "epoch": 0.554746283031361,
      "grad_norm": 1.368720531463623,
      "learning_rate": 8.739858620904251e-05,
      "loss": 0.9557,
      "step": 1152
    },
    {
      "epoch": 0.5552278336242702,
      "grad_norm": 1.1172047853469849,
      "learning_rate": 8.724385734870271e-05,
      "loss": 0.9052,
      "step": 1153
    },
    {
      "epoch": 0.5557093842171793,
      "grad_norm": 1.5191566944122314,
      "learning_rate": 8.708915952681769e-05,
      "loss": 0.7963,
      "step": 1154
    },
    {
      "epoch": 0.5561909348100885,
      "grad_norm": 1.0273900032043457,
      "learning_rate": 8.693449311980074e-05,
      "loss": 1.3081,
      "step": 1155
    },
    {
      "epoch": 0.5566724854029976,
      "grad_norm": 2.036406993865967,
      "learning_rate": 8.677985850398866e-05,
      "loss": 1.301,
      "step": 1156
    },
    {
      "epoch": 0.5571540359959068,
      "grad_norm": 1.2051849365234375,
      "learning_rate": 8.662525605564093e-05,
      "loss": 1.3274,
      "step": 1157
    },
    {
      "epoch": 0.557635586588816,
      "grad_norm": 1.007328987121582,
      "learning_rate": 8.647068615093875e-05,
      "loss": 1.6707,
      "step": 1158
    },
    {
      "epoch": 0.5581171371817252,
      "grad_norm": 1.2855488061904907,
      "learning_rate": 8.631614916598419e-05,
      "loss": 0.7543,
      "step": 1159
    },
    {
      "epoch": 0.5585986877746343,
      "grad_norm": 2.2208139896392822,
      "learning_rate": 8.616164547679906e-05,
      "loss": 1.1498,
      "step": 1160
    },
    {
      "epoch": 0.5590802383675435,
      "grad_norm": 1.1285059452056885,
      "learning_rate": 8.600717545932435e-05,
      "loss": 1.2769,
      "step": 1161
    },
    {
      "epoch": 0.5595617889604526,
      "grad_norm": 0.9752216935157776,
      "learning_rate": 8.5852739489419e-05,
      "loss": 0.4862,
      "step": 1162
    },
    {
      "epoch": 0.5600433395533618,
      "grad_norm": 2.033729076385498,
      "learning_rate": 8.569833794285915e-05,
      "loss": 1.5715,
      "step": 1163
    },
    {
      "epoch": 0.5605248901462709,
      "grad_norm": 0.9259082078933716,
      "learning_rate": 8.554397119533714e-05,
      "loss": 1.0464,
      "step": 1164
    },
    {
      "epoch": 0.5610064407391802,
      "grad_norm": 1.1815052032470703,
      "learning_rate": 8.538963962246069e-05,
      "loss": 0.8728,
      "step": 1165
    },
    {
      "epoch": 0.5614879913320894,
      "grad_norm": 2.466887950897217,
      "learning_rate": 8.523534359975189e-05,
      "loss": 1.3434,
      "step": 1166
    },
    {
      "epoch": 0.5619695419249985,
      "grad_norm": 1.1548599004745483,
      "learning_rate": 8.508108350264635e-05,
      "loss": 0.9088,
      "step": 1167
    },
    {
      "epoch": 0.5624510925179077,
      "grad_norm": 1.8399606943130493,
      "learning_rate": 8.492685970649228e-05,
      "loss": 1.002,
      "step": 1168
    },
    {
      "epoch": 0.5629326431108168,
      "grad_norm": 1.794788122177124,
      "learning_rate": 8.477267258654949e-05,
      "loss": 1.4548,
      "step": 1169
    },
    {
      "epoch": 0.563414193703726,
      "grad_norm": 1.4234602451324463,
      "learning_rate": 8.461852251798866e-05,
      "loss": 1.2673,
      "step": 1170
    },
    {
      "epoch": 0.5638957442966351,
      "grad_norm": 1.721806287765503,
      "learning_rate": 8.44644098758902e-05,
      "loss": 1.1446,
      "step": 1171
    },
    {
      "epoch": 0.5643772948895444,
      "grad_norm": 1.2552214860916138,
      "learning_rate": 8.431033503524354e-05,
      "loss": 0.7614,
      "step": 1172
    },
    {
      "epoch": 0.5648588454824535,
      "grad_norm": 2.4091334342956543,
      "learning_rate": 8.415629837094611e-05,
      "loss": 1.2439,
      "step": 1173
    },
    {
      "epoch": 0.5653403960753627,
      "grad_norm": 2.0601258277893066,
      "learning_rate": 8.400230025780243e-05,
      "loss": 1.0639,
      "step": 1174
    },
    {
      "epoch": 0.5658219466682718,
      "grad_norm": 1.2945075035095215,
      "learning_rate": 8.384834107052321e-05,
      "loss": 0.8757,
      "step": 1175
    },
    {
      "epoch": 0.566303497261181,
      "grad_norm": 1.0456854104995728,
      "learning_rate": 8.369442118372447e-05,
      "loss": 0.8276,
      "step": 1176
    },
    {
      "epoch": 0.5667850478540901,
      "grad_norm": 1.1163201332092285,
      "learning_rate": 8.35405409719266e-05,
      "loss": 0.8483,
      "step": 1177
    },
    {
      "epoch": 0.5672665984469993,
      "grad_norm": 2.1336424350738525,
      "learning_rate": 8.338670080955349e-05,
      "loss": 1.4836,
      "step": 1178
    },
    {
      "epoch": 0.5677481490399086,
      "grad_norm": 1.6260133981704712,
      "learning_rate": 8.323290107093143e-05,
      "loss": 0.8971,
      "step": 1179
    },
    {
      "epoch": 0.5682296996328177,
      "grad_norm": 1.3384321928024292,
      "learning_rate": 8.307914213028856e-05,
      "loss": 0.8744,
      "step": 1180
    },
    {
      "epoch": 0.5687112502257269,
      "grad_norm": 1.5798710584640503,
      "learning_rate": 8.292542436175356e-05,
      "loss": 1.0181,
      "step": 1181
    },
    {
      "epoch": 0.569192800818636,
      "grad_norm": 1.8277827501296997,
      "learning_rate": 8.277174813935508e-05,
      "loss": 1.3165,
      "step": 1182
    },
    {
      "epoch": 0.5696743514115452,
      "grad_norm": 1.2431572675704956,
      "learning_rate": 8.261811383702061e-05,
      "loss": 1.1623,
      "step": 1183
    },
    {
      "epoch": 0.5701559020044543,
      "grad_norm": 1.9663875102996826,
      "learning_rate": 8.246452182857562e-05,
      "loss": 1.1122,
      "step": 1184
    },
    {
      "epoch": 0.5706374525973635,
      "grad_norm": 1.517680048942566,
      "learning_rate": 8.231097248774274e-05,
      "loss": 1.4465,
      "step": 1185
    },
    {
      "epoch": 0.5711190031902726,
      "grad_norm": 1.8820754289627075,
      "learning_rate": 8.215746618814067e-05,
      "loss": 1.0942,
      "step": 1186
    },
    {
      "epoch": 0.5716005537831819,
      "grad_norm": 4.469507217407227,
      "learning_rate": 8.200400330328348e-05,
      "loss": 1.9599,
      "step": 1187
    },
    {
      "epoch": 0.572082104376091,
      "grad_norm": 1.9942208528518677,
      "learning_rate": 8.185058420657957e-05,
      "loss": 1.6599,
      "step": 1188
    },
    {
      "epoch": 0.5725636549690002,
      "grad_norm": 1.0755258798599243,
      "learning_rate": 8.16972092713308e-05,
      "loss": 1.0274,
      "step": 1189
    },
    {
      "epoch": 0.5730452055619093,
      "grad_norm": 0.9476225972175598,
      "learning_rate": 8.154387887073158e-05,
      "loss": 0.7867,
      "step": 1190
    },
    {
      "epoch": 0.5735267561548185,
      "grad_norm": 1.2267851829528809,
      "learning_rate": 8.139059337786792e-05,
      "loss": 1.0899,
      "step": 1191
    },
    {
      "epoch": 0.5740083067477277,
      "grad_norm": 1.181541919708252,
      "learning_rate": 8.12373531657166e-05,
      "loss": 1.3179,
      "step": 1192
    },
    {
      "epoch": 0.5744898573406368,
      "grad_norm": 1.0868141651153564,
      "learning_rate": 8.108415860714418e-05,
      "loss": 0.7083,
      "step": 1193
    },
    {
      "epoch": 0.5749714079335461,
      "grad_norm": 1.2914657592773438,
      "learning_rate": 8.093101007490622e-05,
      "loss": 1.1711,
      "step": 1194
    },
    {
      "epoch": 0.5754529585264552,
      "grad_norm": 1.6261804103851318,
      "learning_rate": 8.077790794164619e-05,
      "loss": 0.884,
      "step": 1195
    },
    {
      "epoch": 0.5759345091193644,
      "grad_norm": 0.7875903248786926,
      "learning_rate": 8.062485257989471e-05,
      "loss": 1.2517,
      "step": 1196
    },
    {
      "epoch": 0.5764160597122735,
      "grad_norm": 1.2122163772583008,
      "learning_rate": 8.047184436206864e-05,
      "loss": 0.9933,
      "step": 1197
    },
    {
      "epoch": 0.5768976103051827,
      "grad_norm": 1.1271840333938599,
      "learning_rate": 8.031888366046998e-05,
      "loss": 0.9851,
      "step": 1198
    },
    {
      "epoch": 0.5773791608980918,
      "grad_norm": 1.2482508420944214,
      "learning_rate": 8.016597084728526e-05,
      "loss": 1.1067,
      "step": 1199
    },
    {
      "epoch": 0.577860711491001,
      "grad_norm": 1.0998296737670898,
      "learning_rate": 8.001310629458443e-05,
      "loss": 1.3181,
      "step": 1200
    },
    {
      "epoch": 0.5783422620839102,
      "grad_norm": 1.3045167922973633,
      "learning_rate": 7.986029037432002e-05,
      "loss": 1.1667,
      "step": 1201
    },
    {
      "epoch": 0.5788238126768194,
      "grad_norm": 1.3124192953109741,
      "learning_rate": 7.970752345832623e-05,
      "loss": 0.9053,
      "step": 1202
    },
    {
      "epoch": 0.5793053632697285,
      "grad_norm": 1.2553496360778809,
      "learning_rate": 7.9554805918318e-05,
      "loss": 0.7183,
      "step": 1203
    },
    {
      "epoch": 0.5797869138626377,
      "grad_norm": 1.404257893562317,
      "learning_rate": 7.940213812589018e-05,
      "loss": 1.0413,
      "step": 1204
    },
    {
      "epoch": 0.5802684644555468,
      "grad_norm": 0.8187459111213684,
      "learning_rate": 7.92495204525165e-05,
      "loss": 0.8142,
      "step": 1205
    },
    {
      "epoch": 0.580750015048456,
      "grad_norm": 1.5299630165100098,
      "learning_rate": 7.909695326954878e-05,
      "loss": 1.5361,
      "step": 1206
    },
    {
      "epoch": 0.5812315656413652,
      "grad_norm": 1.1357324123382568,
      "learning_rate": 7.894443694821602e-05,
      "loss": 0.6301,
      "step": 1207
    },
    {
      "epoch": 0.5817131162342744,
      "grad_norm": 1.4534069299697876,
      "learning_rate": 7.879197185962339e-05,
      "loss": 1.4234,
      "step": 1208
    },
    {
      "epoch": 0.5821946668271836,
      "grad_norm": 2.266435146331787,
      "learning_rate": 7.863955837475144e-05,
      "loss": 1.5866,
      "step": 1209
    },
    {
      "epoch": 0.5826762174200927,
      "grad_norm": 1.658547282218933,
      "learning_rate": 7.848719686445515e-05,
      "loss": 0.6464,
      "step": 1210
    },
    {
      "epoch": 0.5831577680130019,
      "grad_norm": 0.8215835690498352,
      "learning_rate": 7.833488769946306e-05,
      "loss": 1.2307,
      "step": 1211
    },
    {
      "epoch": 0.583639318605911,
      "grad_norm": 1.4285764694213867,
      "learning_rate": 7.818263125037633e-05,
      "loss": 1.4231,
      "step": 1212
    },
    {
      "epoch": 0.5841208691988202,
      "grad_norm": 1.2381724119186401,
      "learning_rate": 7.803042788766777e-05,
      "loss": 1.4176,
      "step": 1213
    },
    {
      "epoch": 0.5846024197917293,
      "grad_norm": 1.1669479608535767,
      "learning_rate": 7.787827798168115e-05,
      "loss": 0.677,
      "step": 1214
    },
    {
      "epoch": 0.5850839703846386,
      "grad_norm": 1.8514113426208496,
      "learning_rate": 7.772618190263009e-05,
      "loss": 0.9309,
      "step": 1215
    },
    {
      "epoch": 0.5855655209775477,
      "grad_norm": 1.007489800453186,
      "learning_rate": 7.757414002059726e-05,
      "loss": 0.7443,
      "step": 1216
    },
    {
      "epoch": 0.5860470715704569,
      "grad_norm": 1.6184797286987305,
      "learning_rate": 7.742215270553349e-05,
      "loss": 1.0925,
      "step": 1217
    },
    {
      "epoch": 0.586528622163366,
      "grad_norm": 0.9397738575935364,
      "learning_rate": 7.727022032725672e-05,
      "loss": 0.9277,
      "step": 1218
    },
    {
      "epoch": 0.5870101727562752,
      "grad_norm": 0.7832995057106018,
      "learning_rate": 7.711834325545135e-05,
      "loss": 0.6973,
      "step": 1219
    },
    {
      "epoch": 0.5874917233491844,
      "grad_norm": 1.1605414152145386,
      "learning_rate": 7.696652185966711e-05,
      "loss": 0.8558,
      "step": 1220
    },
    {
      "epoch": 0.5879732739420935,
      "grad_norm": 2.1062138080596924,
      "learning_rate": 7.681475650931834e-05,
      "loss": 0.7773,
      "step": 1221
    },
    {
      "epoch": 0.5884548245350028,
      "grad_norm": 1.954803228378296,
      "learning_rate": 7.666304757368297e-05,
      "loss": 1.3817,
      "step": 1222
    },
    {
      "epoch": 0.5889363751279119,
      "grad_norm": 1.1391233205795288,
      "learning_rate": 7.651139542190164e-05,
      "loss": 1.0308,
      "step": 1223
    },
    {
      "epoch": 0.5894179257208211,
      "grad_norm": 1.5288896560668945,
      "learning_rate": 7.635980042297687e-05,
      "loss": 1.1794,
      "step": 1224
    },
    {
      "epoch": 0.5898994763137302,
      "grad_norm": 1.1054024696350098,
      "learning_rate": 7.620826294577208e-05,
      "loss": 0.9524,
      "step": 1225
    },
    {
      "epoch": 0.5903810269066394,
      "grad_norm": 1.7065656185150146,
      "learning_rate": 7.605678335901071e-05,
      "loss": 0.9285,
      "step": 1226
    },
    {
      "epoch": 0.5908625774995485,
      "grad_norm": 1.3592902421951294,
      "learning_rate": 7.59053620312754e-05,
      "loss": 1.4872,
      "step": 1227
    },
    {
      "epoch": 0.5913441280924577,
      "grad_norm": 0.9916879534721375,
      "learning_rate": 7.575399933100697e-05,
      "loss": 0.9319,
      "step": 1228
    },
    {
      "epoch": 0.5918256786853668,
      "grad_norm": 2.821290969848633,
      "learning_rate": 7.560269562650368e-05,
      "loss": 0.8145,
      "step": 1229
    },
    {
      "epoch": 0.5923072292782761,
      "grad_norm": 1.5236966609954834,
      "learning_rate": 7.54514512859201e-05,
      "loss": 1.0816,
      "step": 1230
    },
    {
      "epoch": 0.5927887798711852,
      "grad_norm": 1.2185916900634766,
      "learning_rate": 7.530026667726645e-05,
      "loss": 1.2082,
      "step": 1231
    },
    {
      "epoch": 0.5932703304640944,
      "grad_norm": 1.198987364768982,
      "learning_rate": 7.51491421684076e-05,
      "loss": 1.2195,
      "step": 1232
    },
    {
      "epoch": 0.5937518810570036,
      "grad_norm": 2.704132556915283,
      "learning_rate": 7.49980781270622e-05,
      "loss": 1.6415,
      "step": 1233
    },
    {
      "epoch": 0.5942334316499127,
      "grad_norm": 1.1184351444244385,
      "learning_rate": 7.484707492080172e-05,
      "loss": 1.2502,
      "step": 1234
    },
    {
      "epoch": 0.5947149822428219,
      "grad_norm": 1.0581955909729004,
      "learning_rate": 7.469613291704962e-05,
      "loss": 1.3217,
      "step": 1235
    },
    {
      "epoch": 0.595196532835731,
      "grad_norm": 2.034511089324951,
      "learning_rate": 7.45452524830805e-05,
      "loss": 1.1031,
      "step": 1236
    },
    {
      "epoch": 0.5956780834286403,
      "grad_norm": 0.6704680919647217,
      "learning_rate": 7.439443398601903e-05,
      "loss": 0.5278,
      "step": 1237
    },
    {
      "epoch": 0.5961596340215494,
      "grad_norm": 1.1241718530654907,
      "learning_rate": 7.424367779283926e-05,
      "loss": 1.4311,
      "step": 1238
    },
    {
      "epoch": 0.5966411846144586,
      "grad_norm": 0.6718910336494446,
      "learning_rate": 7.409298427036364e-05,
      "loss": 0.79,
      "step": 1239
    },
    {
      "epoch": 0.5971227352073677,
      "grad_norm": 1.4623241424560547,
      "learning_rate": 7.39423537852621e-05,
      "loss": 1.2784,
      "step": 1240
    },
    {
      "epoch": 0.5976042858002769,
      "grad_norm": 0.8255816698074341,
      "learning_rate": 7.379178670405123e-05,
      "loss": 1.1137,
      "step": 1241
    },
    {
      "epoch": 0.598085836393186,
      "grad_norm": 1.7775801420211792,
      "learning_rate": 7.364128339309326e-05,
      "loss": 1.1779,
      "step": 1242
    },
    {
      "epoch": 0.5985673869860952,
      "grad_norm": 1.4368200302124023,
      "learning_rate": 7.349084421859533e-05,
      "loss": 1.1234,
      "step": 1243
    },
    {
      "epoch": 0.5990489375790043,
      "grad_norm": 1.2881683111190796,
      "learning_rate": 7.334046954660852e-05,
      "loss": 1.0434,
      "step": 1244
    },
    {
      "epoch": 0.5995304881719136,
      "grad_norm": 1.3741695880889893,
      "learning_rate": 7.31901597430269e-05,
      "loss": 0.9918,
      "step": 1245
    },
    {
      "epoch": 0.6000120387648227,
      "grad_norm": 2.0060150623321533,
      "learning_rate": 7.303991517358678e-05,
      "loss": 1.1128,
      "step": 1246
    },
    {
      "epoch": 0.6004935893577319,
      "grad_norm": 1.7438344955444336,
      "learning_rate": 7.288973620386568e-05,
      "loss": 1.3951,
      "step": 1247
    },
    {
      "epoch": 0.6009751399506411,
      "grad_norm": 1.3696821928024292,
      "learning_rate": 7.273962319928151e-05,
      "loss": 1.1776,
      "step": 1248
    },
    {
      "epoch": 0.6014566905435502,
      "grad_norm": 0.9706969857215881,
      "learning_rate": 7.258957652509171e-05,
      "loss": 0.8859,
      "step": 1249
    },
    {
      "epoch": 0.6019382411364594,
      "grad_norm": 1.1778486967086792,
      "learning_rate": 7.24395965463923e-05,
      "loss": 0.9378,
      "step": 1250
    },
    {
      "epoch": 0.6024197917293685,
      "grad_norm": 1.9411897659301758,
      "learning_rate": 7.228968362811702e-05,
      "loss": 0.8424,
      "step": 1251
    },
    {
      "epoch": 0.6029013423222778,
      "grad_norm": 1.1882762908935547,
      "learning_rate": 7.21398381350364e-05,
      "loss": 1.235,
      "step": 1252
    },
    {
      "epoch": 0.6033828929151869,
      "grad_norm": 1.270648717880249,
      "learning_rate": 7.199006043175698e-05,
      "loss": 0.889,
      "step": 1253
    },
    {
      "epoch": 0.6038644435080961,
      "grad_norm": 1.0557968616485596,
      "learning_rate": 7.184035088272028e-05,
      "loss": 0.7097,
      "step": 1254
    },
    {
      "epoch": 0.6043459941010052,
      "grad_norm": 1.042986273765564,
      "learning_rate": 7.169070985220208e-05,
      "loss": 1.0926,
      "step": 1255
    },
    {
      "epoch": 0.6048275446939144,
      "grad_norm": 1.5754332542419434,
      "learning_rate": 7.154113770431132e-05,
      "loss": 1.2821,
      "step": 1256
    },
    {
      "epoch": 0.6053090952868235,
      "grad_norm": 1.8790380954742432,
      "learning_rate": 7.13916348029894e-05,
      "loss": 0.9518,
      "step": 1257
    },
    {
      "epoch": 0.6057906458797327,
      "grad_norm": 1.3659653663635254,
      "learning_rate": 7.124220151200926e-05,
      "loss": 1.1807,
      "step": 1258
    },
    {
      "epoch": 0.6062721964726419,
      "grad_norm": 1.2705018520355225,
      "learning_rate": 7.10928381949744e-05,
      "loss": 0.788,
      "step": 1259
    },
    {
      "epoch": 0.6067537470655511,
      "grad_norm": 1.4139997959136963,
      "learning_rate": 7.094354521531807e-05,
      "loss": 1.3152,
      "step": 1260
    },
    {
      "epoch": 0.6072352976584603,
      "grad_norm": 0.8229020237922668,
      "learning_rate": 7.079432293630244e-05,
      "loss": 1.013,
      "step": 1261
    },
    {
      "epoch": 0.6077168482513694,
      "grad_norm": 1.4998596906661987,
      "learning_rate": 7.064517172101753e-05,
      "loss": 1.1517,
      "step": 1262
    },
    {
      "epoch": 0.6081983988442786,
      "grad_norm": 1.490497350692749,
      "learning_rate": 7.04960919323806e-05,
      "loss": 1.2411,
      "step": 1263
    },
    {
      "epoch": 0.6086799494371877,
      "grad_norm": 2.3827402591705322,
      "learning_rate": 7.034708393313493e-05,
      "loss": 1.4421,
      "step": 1264
    },
    {
      "epoch": 0.609161500030097,
      "grad_norm": 1.6210700273513794,
      "learning_rate": 7.019814808584928e-05,
      "loss": 1.409,
      "step": 1265
    },
    {
      "epoch": 0.609643050623006,
      "grad_norm": 2.0713021755218506,
      "learning_rate": 7.004928475291678e-05,
      "loss": 0.9599,
      "step": 1266
    },
    {
      "epoch": 0.6101246012159153,
      "grad_norm": 1.434581995010376,
      "learning_rate": 6.990049429655412e-05,
      "loss": 1.0161,
      "step": 1267
    },
    {
      "epoch": 0.6106061518088244,
      "grad_norm": 1.2966500520706177,
      "learning_rate": 6.97517770788007e-05,
      "loss": 1.0563,
      "step": 1268
    },
    {
      "epoch": 0.6110877024017336,
      "grad_norm": 1.2908307313919067,
      "learning_rate": 6.960313346151761e-05,
      "loss": 1.3258,
      "step": 1269
    },
    {
      "epoch": 0.6115692529946427,
      "grad_norm": 1.4354865550994873,
      "learning_rate": 6.9454563806387e-05,
      "loss": 1.361,
      "step": 1270
    },
    {
      "epoch": 0.6120508035875519,
      "grad_norm": 0.7785577178001404,
      "learning_rate": 6.930606847491094e-05,
      "loss": 0.8053,
      "step": 1271
    },
    {
      "epoch": 0.612532354180461,
      "grad_norm": 1.784480094909668,
      "learning_rate": 6.915764782841072e-05,
      "loss": 1.3545,
      "step": 1272
    },
    {
      "epoch": 0.6130139047733703,
      "grad_norm": 2.5070528984069824,
      "learning_rate": 6.900930222802588e-05,
      "loss": 1.2498,
      "step": 1273
    },
    {
      "epoch": 0.6134954553662795,
      "grad_norm": 1.3918461799621582,
      "learning_rate": 6.886103203471337e-05,
      "loss": 0.9216,
      "step": 1274
    },
    {
      "epoch": 0.6139770059591886,
      "grad_norm": 1.2397583723068237,
      "learning_rate": 6.871283760924665e-05,
      "loss": 1.0083,
      "step": 1275
    },
    {
      "epoch": 0.6144585565520978,
      "grad_norm": 1.1042224168777466,
      "learning_rate": 6.856471931221478e-05,
      "loss": 1.1162,
      "step": 1276
    },
    {
      "epoch": 0.6149401071450069,
      "grad_norm": 0.9897892475128174,
      "learning_rate": 6.841667750402162e-05,
      "loss": 0.7177,
      "step": 1277
    },
    {
      "epoch": 0.6154216577379161,
      "grad_norm": 2.1587398052215576,
      "learning_rate": 6.826871254488496e-05,
      "loss": 1.0383,
      "step": 1278
    },
    {
      "epoch": 0.6159032083308252,
      "grad_norm": 0.7059440612792969,
      "learning_rate": 6.812082479483553e-05,
      "loss": 0.611,
      "step": 1279
    },
    {
      "epoch": 0.6163847589237345,
      "grad_norm": 1.4390486478805542,
      "learning_rate": 6.797301461371625e-05,
      "loss": 0.9625,
      "step": 1280
    },
    {
      "epoch": 0.6168663095166436,
      "grad_norm": 1.2871108055114746,
      "learning_rate": 6.782528236118124e-05,
      "loss": 1.1854,
      "step": 1281
    },
    {
      "epoch": 0.6173478601095528,
      "grad_norm": 0.9264230132102966,
      "learning_rate": 6.767762839669503e-05,
      "loss": 1.3269,
      "step": 1282
    },
    {
      "epoch": 0.6178294107024619,
      "grad_norm": 0.9619544148445129,
      "learning_rate": 6.753005307953167e-05,
      "loss": 0.9452,
      "step": 1283
    },
    {
      "epoch": 0.6183109612953711,
      "grad_norm": 1.4726909399032593,
      "learning_rate": 6.738255676877381e-05,
      "loss": 1.1915,
      "step": 1284
    },
    {
      "epoch": 0.6187925118882802,
      "grad_norm": 1.1944189071655273,
      "learning_rate": 6.723513982331195e-05,
      "loss": 1.1482,
      "step": 1285
    },
    {
      "epoch": 0.6192740624811894,
      "grad_norm": 1.9658671617507935,
      "learning_rate": 6.708780260184333e-05,
      "loss": 1.1999,
      "step": 1286
    },
    {
      "epoch": 0.6197556130740985,
      "grad_norm": 2.1050336360931396,
      "learning_rate": 6.694054546287132e-05,
      "loss": 0.96,
      "step": 1287
    },
    {
      "epoch": 0.6202371636670078,
      "grad_norm": 1.7945077419281006,
      "learning_rate": 6.679336876470441e-05,
      "loss": 1.3478,
      "step": 1288
    },
    {
      "epoch": 0.620718714259917,
      "grad_norm": 1.6055036783218384,
      "learning_rate": 6.664627286545535e-05,
      "loss": 0.9602,
      "step": 1289
    },
    {
      "epoch": 0.6212002648528261,
      "grad_norm": 1.3143390417099,
      "learning_rate": 6.649925812304025e-05,
      "loss": 0.9817,
      "step": 1290
    },
    {
      "epoch": 0.6216818154457353,
      "grad_norm": 1.0135152339935303,
      "learning_rate": 6.635232489517782e-05,
      "loss": 0.8729,
      "step": 1291
    },
    {
      "epoch": 0.6221633660386444,
      "grad_norm": 0.9975107908248901,
      "learning_rate": 6.620547353938836e-05,
      "loss": 1.0122,
      "step": 1292
    },
    {
      "epoch": 0.6226449166315536,
      "grad_norm": 1.2772380113601685,
      "learning_rate": 6.605870441299302e-05,
      "loss": 0.9714,
      "step": 1293
    },
    {
      "epoch": 0.6231264672244627,
      "grad_norm": 1.0343154668807983,
      "learning_rate": 6.591201787311285e-05,
      "loss": 1.5245,
      "step": 1294
    },
    {
      "epoch": 0.623608017817372,
      "grad_norm": 1.774516224861145,
      "learning_rate": 6.57654142766679e-05,
      "loss": 0.9565,
      "step": 1295
    },
    {
      "epoch": 0.6240895684102811,
      "grad_norm": 2.1767048835754395,
      "learning_rate": 6.561889398037643e-05,
      "loss": 0.9869,
      "step": 1296
    },
    {
      "epoch": 0.6245711190031903,
      "grad_norm": 1.8510329723358154,
      "learning_rate": 6.547245734075403e-05,
      "loss": 1.03,
      "step": 1297
    },
    {
      "epoch": 0.6250526695960994,
      "grad_norm": 1.354679822921753,
      "learning_rate": 6.532610471411274e-05,
      "loss": 0.9354,
      "step": 1298
    },
    {
      "epoch": 0.6255342201890086,
      "grad_norm": 1.2885664701461792,
      "learning_rate": 6.517983645656014e-05,
      "loss": 0.8063,
      "step": 1299
    },
    {
      "epoch": 0.6260157707819177,
      "grad_norm": 1.3011651039123535,
      "learning_rate": 6.503365292399857e-05,
      "loss": 1.2994,
      "step": 1300
    },
    {
      "epoch": 0.6264973213748269,
      "grad_norm": 1.2887555360794067,
      "learning_rate": 6.488755447212418e-05,
      "loss": 1.1833,
      "step": 1301
    },
    {
      "epoch": 0.6269788719677362,
      "grad_norm": 0.688773512840271,
      "learning_rate": 6.474154145642612e-05,
      "loss": 1.2065,
      "step": 1302
    },
    {
      "epoch": 0.6274604225606453,
      "grad_norm": 1.158751130104065,
      "learning_rate": 6.459561423218561e-05,
      "loss": 1.4046,
      "step": 1303
    },
    {
      "epoch": 0.6279419731535545,
      "grad_norm": 0.8562422394752502,
      "learning_rate": 6.444977315447521e-05,
      "loss": 0.7242,
      "step": 1304
    },
    {
      "epoch": 0.6284235237464636,
      "grad_norm": 1.4259194135665894,
      "learning_rate": 6.430401857815776e-05,
      "loss": 1.0269,
      "step": 1305
    },
    {
      "epoch": 0.6289050743393728,
      "grad_norm": 1.824709415435791,
      "learning_rate": 6.415835085788575e-05,
      "loss": 1.46,
      "step": 1306
    },
    {
      "epoch": 0.6293866249322819,
      "grad_norm": 1.494925618171692,
      "learning_rate": 6.401277034810017e-05,
      "loss": 0.9106,
      "step": 1307
    },
    {
      "epoch": 0.6298681755251911,
      "grad_norm": 1.1361433267593384,
      "learning_rate": 6.386727740302994e-05,
      "loss": 1.0614,
      "step": 1308
    },
    {
      "epoch": 0.6303497261181003,
      "grad_norm": 0.8548444509506226,
      "learning_rate": 6.37218723766909e-05,
      "loss": 0.9466,
      "step": 1309
    },
    {
      "epoch": 0.6308312767110095,
      "grad_norm": 1.5237458944320679,
      "learning_rate": 6.357655562288488e-05,
      "loss": 1.3003,
      "step": 1310
    },
    {
      "epoch": 0.6313128273039186,
      "grad_norm": 2.1414363384246826,
      "learning_rate": 6.343132749519902e-05,
      "loss": 1.3837,
      "step": 1311
    },
    {
      "epoch": 0.6317943778968278,
      "grad_norm": 1.4853914976119995,
      "learning_rate": 6.328618834700474e-05,
      "loss": 0.9136,
      "step": 1312
    },
    {
      "epoch": 0.6322759284897369,
      "grad_norm": 3.514492988586426,
      "learning_rate": 6.314113853145703e-05,
      "loss": 1.321,
      "step": 1313
    },
    {
      "epoch": 0.6327574790826461,
      "grad_norm": 1.0241303443908691,
      "learning_rate": 6.299617840149349e-05,
      "loss": 0.9012,
      "step": 1314
    },
    {
      "epoch": 0.6332390296755553,
      "grad_norm": 1.2108702659606934,
      "learning_rate": 6.285130830983339e-05,
      "loss": 1.0515,
      "step": 1315
    },
    {
      "epoch": 0.6337205802684645,
      "grad_norm": 1.4265339374542236,
      "learning_rate": 6.270652860897704e-05,
      "loss": 1.1594,
      "step": 1316
    },
    {
      "epoch": 0.6342021308613737,
      "grad_norm": 1.2736425399780273,
      "learning_rate": 6.25618396512048e-05,
      "loss": 1.2042,
      "step": 1317
    },
    {
      "epoch": 0.6346836814542828,
      "grad_norm": 1.6517406702041626,
      "learning_rate": 6.24172417885762e-05,
      "loss": 1.2523,
      "step": 1318
    },
    {
      "epoch": 0.635165232047192,
      "grad_norm": 0.9992088079452515,
      "learning_rate": 6.227273537292911e-05,
      "loss": 1.1092,
      "step": 1319
    },
    {
      "epoch": 0.6356467826401011,
      "grad_norm": 0.9736806154251099,
      "learning_rate": 6.212832075587891e-05,
      "loss": 1.0101,
      "step": 1320
    },
    {
      "epoch": 0.6361283332330103,
      "grad_norm": 1.1378737688064575,
      "learning_rate": 6.19839982888176e-05,
      "loss": 1.2925,
      "step": 1321
    },
    {
      "epoch": 0.6366098838259194,
      "grad_norm": 1.354857325553894,
      "learning_rate": 6.183976832291296e-05,
      "loss": 0.94,
      "step": 1322
    },
    {
      "epoch": 0.6370914344188287,
      "grad_norm": 0.7981762290000916,
      "learning_rate": 6.169563120910775e-05,
      "loss": 1.0068,
      "step": 1323
    },
    {
      "epoch": 0.6375729850117378,
      "grad_norm": 1.3615357875823975,
      "learning_rate": 6.155158729811867e-05,
      "loss": 1.4241,
      "step": 1324
    },
    {
      "epoch": 0.638054535604647,
      "grad_norm": 0.9818617701530457,
      "learning_rate": 6.140763694043578e-05,
      "loss": 0.7129,
      "step": 1325
    },
    {
      "epoch": 0.6385360861975561,
      "grad_norm": 0.8565928339958191,
      "learning_rate": 6.126378048632139e-05,
      "loss": 0.982,
      "step": 1326
    },
    {
      "epoch": 0.6390176367904653,
      "grad_norm": 0.8885542154312134,
      "learning_rate": 6.112001828580944e-05,
      "loss": 1.4397,
      "step": 1327
    },
    {
      "epoch": 0.6394991873833744,
      "grad_norm": 1.2945500612258911,
      "learning_rate": 6.0976350688704455e-05,
      "loss": 0.7162,
      "step": 1328
    },
    {
      "epoch": 0.6399807379762836,
      "grad_norm": 1.0205796957015991,
      "learning_rate": 6.083277804458072e-05,
      "loss": 1.2608,
      "step": 1329
    },
    {
      "epoch": 0.6404622885691929,
      "grad_norm": 1.148720622062683,
      "learning_rate": 6.068930070278159e-05,
      "loss": 1.1016,
      "step": 1330
    },
    {
      "epoch": 0.640943839162102,
      "grad_norm": 0.8900640606880188,
      "learning_rate": 6.054591901241846e-05,
      "loss": 0.8941,
      "step": 1331
    },
    {
      "epoch": 0.6414253897550112,
      "grad_norm": 1.1613692045211792,
      "learning_rate": 6.040263332237002e-05,
      "loss": 1.0666,
      "step": 1332
    },
    {
      "epoch": 0.6419069403479203,
      "grad_norm": 1.0480207204818726,
      "learning_rate": 6.025944398128137e-05,
      "loss": 0.8137,
      "step": 1333
    },
    {
      "epoch": 0.6423884909408295,
      "grad_norm": 1.5120476484298706,
      "learning_rate": 6.011635133756309e-05,
      "loss": 1.3695,
      "step": 1334
    },
    {
      "epoch": 0.6428700415337386,
      "grad_norm": 1.2082058191299438,
      "learning_rate": 5.99733557393906e-05,
      "loss": 0.8123,
      "step": 1335
    },
    {
      "epoch": 0.6433515921266478,
      "grad_norm": 1.4196912050247192,
      "learning_rate": 5.983045753470308e-05,
      "loss": 1.4308,
      "step": 1336
    },
    {
      "epoch": 0.6438331427195569,
      "grad_norm": 0.7111194133758545,
      "learning_rate": 5.96876570712028e-05,
      "loss": 1.1979,
      "step": 1337
    },
    {
      "epoch": 0.6443146933124662,
      "grad_norm": 1.9173833131790161,
      "learning_rate": 5.954495469635417e-05,
      "loss": 0.9656,
      "step": 1338
    },
    {
      "epoch": 0.6447962439053753,
      "grad_norm": 1.5747874975204468,
      "learning_rate": 5.940235075738296e-05,
      "loss": 1.4372,
      "step": 1339
    },
    {
      "epoch": 0.6452777944982845,
      "grad_norm": 1.7322578430175781,
      "learning_rate": 5.925984560127542e-05,
      "loss": 1.5148,
      "step": 1340
    },
    {
      "epoch": 0.6457593450911936,
      "grad_norm": 1.4524576663970947,
      "learning_rate": 5.911743957477739e-05,
      "loss": 1.8077,
      "step": 1341
    },
    {
      "epoch": 0.6462408956841028,
      "grad_norm": 1.4451179504394531,
      "learning_rate": 5.897513302439355e-05,
      "loss": 1.0097,
      "step": 1342
    },
    {
      "epoch": 0.646722446277012,
      "grad_norm": 1.0157692432403564,
      "learning_rate": 5.883292629638651e-05,
      "loss": 1.7309,
      "step": 1343
    },
    {
      "epoch": 0.6472039968699211,
      "grad_norm": 1.1843279600143433,
      "learning_rate": 5.869081973677604e-05,
      "loss": 0.935,
      "step": 1344
    },
    {
      "epoch": 0.6476855474628304,
      "grad_norm": 1.1999677419662476,
      "learning_rate": 5.8548813691338134e-05,
      "loss": 1.224,
      "step": 1345
    },
    {
      "epoch": 0.6481670980557395,
      "grad_norm": 0.7884172797203064,
      "learning_rate": 5.84069085056042e-05,
      "loss": 1.374,
      "step": 1346
    },
    {
      "epoch": 0.6486486486486487,
      "grad_norm": 2.4568867683410645,
      "learning_rate": 5.826510452486027e-05,
      "loss": 1.2019,
      "step": 1347
    },
    {
      "epoch": 0.6491301992415578,
      "grad_norm": 2.2645375728607178,
      "learning_rate": 5.81234020941461e-05,
      "loss": 1.0378,
      "step": 1348
    },
    {
      "epoch": 0.649611749834467,
      "grad_norm": 1.5647883415222168,
      "learning_rate": 5.798180155825437e-05,
      "loss": 1.2287,
      "step": 1349
    },
    {
      "epoch": 0.6500933004273761,
      "grad_norm": 1.1952080726623535,
      "learning_rate": 5.784030326172981e-05,
      "loss": 1.2422,
      "step": 1350
    },
    {
      "epoch": 0.6505748510202853,
      "grad_norm": 1.26864755153656,
      "learning_rate": 5.7698907548868395e-05,
      "loss": 1.2312,
      "step": 1351
    },
    {
      "epoch": 0.6510564016131944,
      "grad_norm": 0.8909727334976196,
      "learning_rate": 5.755761476371653e-05,
      "loss": 0.8675,
      "step": 1352
    },
    {
      "epoch": 0.6515379522061037,
      "grad_norm": 1.3474384546279907,
      "learning_rate": 5.741642525007003e-05,
      "loss": 1.2167,
      "step": 1353
    },
    {
      "epoch": 0.6520195027990128,
      "grad_norm": 1.6467247009277344,
      "learning_rate": 5.727533935147359e-05,
      "loss": 0.9913,
      "step": 1354
    },
    {
      "epoch": 0.652501053391922,
      "grad_norm": 0.7765967845916748,
      "learning_rate": 5.713435741121975e-05,
      "loss": 0.9967,
      "step": 1355
    },
    {
      "epoch": 0.6529826039848312,
      "grad_norm": 1.4382472038269043,
      "learning_rate": 5.699347977234799e-05,
      "loss": 1.4647,
      "step": 1356
    },
    {
      "epoch": 0.6534641545777403,
      "grad_norm": 1.5743249654769897,
      "learning_rate": 5.685270677764412e-05,
      "loss": 0.8362,
      "step": 1357
    },
    {
      "epoch": 0.6539457051706495,
      "grad_norm": 0.8253456354141235,
      "learning_rate": 5.671203876963931e-05,
      "loss": 0.9583,
      "step": 1358
    },
    {
      "epoch": 0.6544272557635586,
      "grad_norm": 0.8837342858314514,
      "learning_rate": 5.657147609060924e-05,
      "loss": 1.1434,
      "step": 1359
    },
    {
      "epoch": 0.6549088063564679,
      "grad_norm": 1.224574327468872,
      "learning_rate": 5.643101908257333e-05,
      "loss": 1.013,
      "step": 1360
    },
    {
      "epoch": 0.655390356949377,
      "grad_norm": 1.3580174446105957,
      "learning_rate": 5.629066808729385e-05,
      "loss": 0.7689,
      "step": 1361
    },
    {
      "epoch": 0.6558719075422862,
      "grad_norm": 1.3258894681930542,
      "learning_rate": 5.6150423446275144e-05,
      "loss": 1.1871,
      "step": 1362
    },
    {
      "epoch": 0.6563534581351953,
      "grad_norm": 1.1320303678512573,
      "learning_rate": 5.601028550076277e-05,
      "loss": 1.0056,
      "step": 1363
    },
    {
      "epoch": 0.6568350087281045,
      "grad_norm": 2.4723637104034424,
      "learning_rate": 5.587025459174271e-05,
      "loss": 1.0261,
      "step": 1364
    },
    {
      "epoch": 0.6573165593210136,
      "grad_norm": 0.9704416990280151,
      "learning_rate": 5.573033105994038e-05,
      "loss": 0.9322,
      "step": 1365
    },
    {
      "epoch": 0.6577981099139228,
      "grad_norm": 0.9944827556610107,
      "learning_rate": 5.559051524582002e-05,
      "loss": 1.0991,
      "step": 1366
    },
    {
      "epoch": 0.658279660506832,
      "grad_norm": 1.1883389949798584,
      "learning_rate": 5.5450807489583777e-05,
      "loss": 1.2106,
      "step": 1367
    },
    {
      "epoch": 0.6587612110997412,
      "grad_norm": 0.9173049330711365,
      "learning_rate": 5.531120813117085e-05,
      "loss": 0.8985,
      "step": 1368
    },
    {
      "epoch": 0.6592427616926503,
      "grad_norm": 1.328553557395935,
      "learning_rate": 5.517171751025667e-05,
      "loss": 1.3791,
      "step": 1369
    },
    {
      "epoch": 0.6597243122855595,
      "grad_norm": 0.934523344039917,
      "learning_rate": 5.5032335966252103e-05,
      "loss": 0.793,
      "step": 1370
    },
    {
      "epoch": 0.6602058628784687,
      "grad_norm": 1.46604323387146,
      "learning_rate": 5.489306383830258e-05,
      "loss": 1.0358,
      "step": 1371
    },
    {
      "epoch": 0.6606874134713778,
      "grad_norm": 0.9401931762695312,
      "learning_rate": 5.475390146528738e-05,
      "loss": 0.7538,
      "step": 1372
    },
    {
      "epoch": 0.661168964064287,
      "grad_norm": 1.0969661474227905,
      "learning_rate": 5.461484918581858e-05,
      "loss": 0.5788,
      "step": 1373
    },
    {
      "epoch": 0.6616505146571962,
      "grad_norm": 2.0446221828460693,
      "learning_rate": 5.4475907338240494e-05,
      "loss": 0.6462,
      "step": 1374
    },
    {
      "epoch": 0.6621320652501054,
      "grad_norm": 1.3886526823043823,
      "learning_rate": 5.43370762606287e-05,
      "loss": 1.2701,
      "step": 1375
    },
    {
      "epoch": 0.6626136158430145,
      "grad_norm": 0.8271850943565369,
      "learning_rate": 5.4198356290789276e-05,
      "loss": 1.0241,
      "step": 1376
    },
    {
      "epoch": 0.6630951664359237,
      "grad_norm": 0.8809240460395813,
      "learning_rate": 5.405974776625785e-05,
      "loss": 0.7635,
      "step": 1377
    },
    {
      "epoch": 0.6635767170288328,
      "grad_norm": 1.8833659887313843,
      "learning_rate": 5.392125102429899e-05,
      "loss": 0.9512,
      "step": 1378
    },
    {
      "epoch": 0.664058267621742,
      "grad_norm": 1.6587456464767456,
      "learning_rate": 5.378286640190522e-05,
      "loss": 0.7275,
      "step": 1379
    },
    {
      "epoch": 0.6645398182146511,
      "grad_norm": 1.6763464212417603,
      "learning_rate": 5.364459423579629e-05,
      "loss": 1.2133,
      "step": 1380
    },
    {
      "epoch": 0.6650213688075604,
      "grad_norm": 1.2368160486221313,
      "learning_rate": 5.350643486241825e-05,
      "loss": 0.8025,
      "step": 1381
    },
    {
      "epoch": 0.6655029194004695,
      "grad_norm": 2.6323835849761963,
      "learning_rate": 5.33683886179428e-05,
      "loss": 1.008,
      "step": 1382
    },
    {
      "epoch": 0.6659844699933787,
      "grad_norm": 2.192049026489258,
      "learning_rate": 5.3230455838266266e-05,
      "loss": 0.8079,
      "step": 1383
    },
    {
      "epoch": 0.6664660205862879,
      "grad_norm": 1.4030390977859497,
      "learning_rate": 5.309263685900898e-05,
      "loss": 1.1317,
      "step": 1384
    },
    {
      "epoch": 0.666947571179197,
      "grad_norm": 0.9323775172233582,
      "learning_rate": 5.295493201551433e-05,
      "loss": 0.6487,
      "step": 1385
    },
    {
      "epoch": 0.6674291217721062,
      "grad_norm": 1.4248300790786743,
      "learning_rate": 5.281734164284802e-05,
      "loss": 1.1483,
      "step": 1386
    },
    {
      "epoch": 0.6679106723650153,
      "grad_norm": 1.3177506923675537,
      "learning_rate": 5.26798660757971e-05,
      "loss": 1.1627,
      "step": 1387
    },
    {
      "epoch": 0.6683922229579246,
      "grad_norm": 1.6739046573638916,
      "learning_rate": 5.2542505648869434e-05,
      "loss": 1.2749,
      "step": 1388
    },
    {
      "epoch": 0.6688737735508337,
      "grad_norm": 0.7236259579658508,
      "learning_rate": 5.240526069629265e-05,
      "loss": 0.8827,
      "step": 1389
    },
    {
      "epoch": 0.6693553241437429,
      "grad_norm": 1.3506990671157837,
      "learning_rate": 5.22681315520134e-05,
      "loss": 1.288,
      "step": 1390
    },
    {
      "epoch": 0.669836874736652,
      "grad_norm": 0.9272012114524841,
      "learning_rate": 5.213111854969661e-05,
      "loss": 0.8747,
      "step": 1391
    },
    {
      "epoch": 0.6703184253295612,
      "grad_norm": 1.5468556880950928,
      "learning_rate": 5.199422202272448e-05,
      "loss": 0.8931,
      "step": 1392
    },
    {
      "epoch": 0.6707999759224703,
      "grad_norm": 1.164279818534851,
      "learning_rate": 5.185744230419589e-05,
      "loss": 0.8449,
      "step": 1393
    },
    {
      "epoch": 0.6712815265153795,
      "grad_norm": 0.6217030882835388,
      "learning_rate": 5.172077972692553e-05,
      "loss": 0.881,
      "step": 1394
    },
    {
      "epoch": 0.6717630771082886,
      "grad_norm": 1.1114040613174438,
      "learning_rate": 5.1584234623442974e-05,
      "loss": 1.2327,
      "step": 1395
    },
    {
      "epoch": 0.6722446277011979,
      "grad_norm": 0.9713524580001831,
      "learning_rate": 5.1447807325992025e-05,
      "loss": 0.549,
      "step": 1396
    },
    {
      "epoch": 0.6727261782941071,
      "grad_norm": 1.384885311126709,
      "learning_rate": 5.13114981665298e-05,
      "loss": 1.46,
      "step": 1397
    },
    {
      "epoch": 0.6732077288870162,
      "grad_norm": 0.7873196601867676,
      "learning_rate": 5.117530747672603e-05,
      "loss": 0.7555,
      "step": 1398
    },
    {
      "epoch": 0.6736892794799254,
      "grad_norm": 1.2935165166854858,
      "learning_rate": 5.103923558796203e-05,
      "loss": 1.1429,
      "step": 1399
    },
    {
      "epoch": 0.6741708300728345,
      "grad_norm": 1.563491940498352,
      "learning_rate": 5.090328283133019e-05,
      "loss": 1.1768,
      "step": 1400
    },
    {
      "epoch": 0.6746523806657437,
      "grad_norm": 2.1122798919677734,
      "learning_rate": 5.0767449537632986e-05,
      "loss": 1.324,
      "step": 1401
    },
    {
      "epoch": 0.6751339312586528,
      "grad_norm": 4.458497524261475,
      "learning_rate": 5.06317360373822e-05,
      "loss": 1.1747,
      "step": 1402
    },
    {
      "epoch": 0.6756154818515621,
      "grad_norm": 1.8291646242141724,
      "learning_rate": 5.049614266079813e-05,
      "loss": 0.917,
      "step": 1403
    },
    {
      "epoch": 0.6760970324444712,
      "grad_norm": 0.9280420541763306,
      "learning_rate": 5.036066973780882e-05,
      "loss": 0.7539,
      "step": 1404
    },
    {
      "epoch": 0.6765785830373804,
      "grad_norm": 1.6787300109863281,
      "learning_rate": 5.022531759804918e-05,
      "loss": 0.8946,
      "step": 1405
    },
    {
      "epoch": 0.6770601336302895,
      "grad_norm": 1.7975965738296509,
      "learning_rate": 5.009008657086025e-05,
      "loss": 1.7972,
      "step": 1406
    },
    {
      "epoch": 0.6775416842231987,
      "grad_norm": 1.3834428787231445,
      "learning_rate": 4.9954976985288395e-05,
      "loss": 1.1798,
      "step": 1407
    },
    {
      "epoch": 0.6780232348161078,
      "grad_norm": 1.5930489301681519,
      "learning_rate": 4.981998917008448e-05,
      "loss": 1.1226,
      "step": 1408
    },
    {
      "epoch": 0.678504785409017,
      "grad_norm": 1.5451221466064453,
      "learning_rate": 4.9685123453703e-05,
      "loss": 1.0632,
      "step": 1409
    },
    {
      "epoch": 0.6789863360019261,
      "grad_norm": 1.1124958992004395,
      "learning_rate": 4.955038016430149e-05,
      "loss": 0.99,
      "step": 1410
    },
    {
      "epoch": 0.6794678865948354,
      "grad_norm": 1.382188081741333,
      "learning_rate": 4.9415759629739455e-05,
      "loss": 0.7417,
      "step": 1411
    },
    {
      "epoch": 0.6799494371877446,
      "grad_norm": 1.0870238542556763,
      "learning_rate": 4.928126217757782e-05,
      "loss": 1.4366,
      "step": 1412
    },
    {
      "epoch": 0.6804309877806537,
      "grad_norm": 1.2263619899749756,
      "learning_rate": 4.914688813507797e-05,
      "loss": 1.3964,
      "step": 1413
    },
    {
      "epoch": 0.6809125383735629,
      "grad_norm": 1.1502625942230225,
      "learning_rate": 4.901263782920105e-05,
      "loss": 1.2702,
      "step": 1414
    },
    {
      "epoch": 0.681394088966472,
      "grad_norm": 2.7644057273864746,
      "learning_rate": 4.887851158660706e-05,
      "loss": 0.9569,
      "step": 1415
    },
    {
      "epoch": 0.6818756395593812,
      "grad_norm": 0.8721311092376709,
      "learning_rate": 4.8744509733654184e-05,
      "loss": 1.1576,
      "step": 1416
    },
    {
      "epoch": 0.6823571901522903,
      "grad_norm": 1.1992177963256836,
      "learning_rate": 4.861063259639793e-05,
      "loss": 1.0653,
      "step": 1417
    },
    {
      "epoch": 0.6828387407451996,
      "grad_norm": 1.2239335775375366,
      "learning_rate": 4.847688050059033e-05,
      "loss": 1.3825,
      "step": 1418
    },
    {
      "epoch": 0.6833202913381087,
      "grad_norm": 1.9246747493743896,
      "learning_rate": 4.8343253771679155e-05,
      "loss": 1.6294,
      "step": 1419
    },
    {
      "epoch": 0.6838018419310179,
      "grad_norm": 1.2689882516860962,
      "learning_rate": 4.82097527348072e-05,
      "loss": 1.3249,
      "step": 1420
    },
    {
      "epoch": 0.684283392523927,
      "grad_norm": 0.7828402519226074,
      "learning_rate": 4.8076377714811284e-05,
      "loss": 0.8218,
      "step": 1421
    },
    {
      "epoch": 0.6847649431168362,
      "grad_norm": 0.8386598229408264,
      "learning_rate": 4.7943129036221735e-05,
      "loss": 0.9151,
      "step": 1422
    },
    {
      "epoch": 0.6852464937097453,
      "grad_norm": 1.5944420099258423,
      "learning_rate": 4.781000702326142e-05,
      "loss": 0.995,
      "step": 1423
    },
    {
      "epoch": 0.6857280443026545,
      "grad_norm": 1.5625433921813965,
      "learning_rate": 4.767701199984497e-05,
      "loss": 0.9407,
      "step": 1424
    },
    {
      "epoch": 0.6862095948955638,
      "grad_norm": 1.4244143962860107,
      "learning_rate": 4.7544144289578066e-05,
      "loss": 0.8419,
      "step": 1425
    },
    {
      "epoch": 0.6866911454884729,
      "grad_norm": 0.9626308679580688,
      "learning_rate": 4.7411404215756594e-05,
      "loss": 0.7041,
      "step": 1426
    },
    {
      "epoch": 0.6871726960813821,
      "grad_norm": 1.711273193359375,
      "learning_rate": 4.7278792101365866e-05,
      "loss": 1.0512,
      "step": 1427
    },
    {
      "epoch": 0.6876542466742912,
      "grad_norm": 1.0641824007034302,
      "learning_rate": 4.714630826907985e-05,
      "loss": 0.9934,
      "step": 1428
    },
    {
      "epoch": 0.6881357972672004,
      "grad_norm": 1.2644219398498535,
      "learning_rate": 4.701395304126038e-05,
      "loss": 1.5435,
      "step": 1429
    },
    {
      "epoch": 0.6886173478601095,
      "grad_norm": 1.0635844469070435,
      "learning_rate": 4.6881726739956375e-05,
      "loss": 0.8749,
      "step": 1430
    },
    {
      "epoch": 0.6890988984530187,
      "grad_norm": 1.022322416305542,
      "learning_rate": 4.6749629686902984e-05,
      "loss": 1.2948,
      "step": 1431
    },
    {
      "epoch": 0.6895804490459279,
      "grad_norm": 0.8008168935775757,
      "learning_rate": 4.661766220352097e-05,
      "loss": 0.5503,
      "step": 1432
    },
    {
      "epoch": 0.6900619996388371,
      "grad_norm": 1.877286434173584,
      "learning_rate": 4.64858246109157e-05,
      "loss": 1.5707,
      "step": 1433
    },
    {
      "epoch": 0.6905435502317462,
      "grad_norm": 1.440529465675354,
      "learning_rate": 4.63541172298766e-05,
      "loss": 1.1388,
      "step": 1434
    },
    {
      "epoch": 0.6910251008246554,
      "grad_norm": 0.7773860692977905,
      "learning_rate": 4.622254038087622e-05,
      "loss": 1.2407,
      "step": 1435
    },
    {
      "epoch": 0.6915066514175645,
      "grad_norm": 1.2136985063552856,
      "learning_rate": 4.60910943840695e-05,
      "loss": 0.9126,
      "step": 1436
    },
    {
      "epoch": 0.6919882020104737,
      "grad_norm": 1.2893611192703247,
      "learning_rate": 4.5959779559292985e-05,
      "loss": 0.5336,
      "step": 1437
    },
    {
      "epoch": 0.6924697526033828,
      "grad_norm": 1.3301244974136353,
      "learning_rate": 4.582859622606406e-05,
      "loss": 0.9188,
      "step": 1438
    },
    {
      "epoch": 0.6929513031962921,
      "grad_norm": 1.0734645128250122,
      "learning_rate": 4.569754470358014e-05,
      "loss": 1.083,
      "step": 1439
    },
    {
      "epoch": 0.6934328537892013,
      "grad_norm": 1.5678324699401855,
      "learning_rate": 4.556662531071796e-05,
      "loss": 1.1346,
      "step": 1440
    },
    {
      "epoch": 0.6939144043821104,
      "grad_norm": 0.7291122078895569,
      "learning_rate": 4.54358383660327e-05,
      "loss": 0.7212,
      "step": 1441
    },
    {
      "epoch": 0.6943959549750196,
      "grad_norm": 1.8412057161331177,
      "learning_rate": 4.530518418775733e-05,
      "loss": 1.1831,
      "step": 1442
    },
    {
      "epoch": 0.6948775055679287,
      "grad_norm": 1.5250680446624756,
      "learning_rate": 4.5174663093801674e-05,
      "loss": 1.5554,
      "step": 1443
    },
    {
      "epoch": 0.6953590561608379,
      "grad_norm": 1.55964994430542,
      "learning_rate": 4.504427540175181e-05,
      "loss": 0.8259,
      "step": 1444
    },
    {
      "epoch": 0.695840606753747,
      "grad_norm": 0.8687563538551331,
      "learning_rate": 4.491402142886922e-05,
      "loss": 0.8217,
      "step": 1445
    },
    {
      "epoch": 0.6963221573466563,
      "grad_norm": 1.1135036945343018,
      "learning_rate": 4.4783901492089984e-05,
      "loss": 1.021,
      "step": 1446
    },
    {
      "epoch": 0.6968037079395654,
      "grad_norm": 0.6413815021514893,
      "learning_rate": 4.465391590802407e-05,
      "loss": 0.859,
      "step": 1447
    },
    {
      "epoch": 0.6972852585324746,
      "grad_norm": 2.0961172580718994,
      "learning_rate": 4.4524064992954516e-05,
      "loss": 1.164,
      "step": 1448
    },
    {
      "epoch": 0.6977668091253837,
      "grad_norm": 1.389328956604004,
      "learning_rate": 4.4394349062836736e-05,
      "loss": 1.1326,
      "step": 1449
    },
    {
      "epoch": 0.6982483597182929,
      "grad_norm": 2.074847936630249,
      "learning_rate": 4.4264768433297565e-05,
      "loss": 0.9993,
      "step": 1450
    },
    {
      "epoch": 0.698729910311202,
      "grad_norm": 0.9919458627700806,
      "learning_rate": 4.4135323419634766e-05,
      "loss": 1.0598,
      "step": 1451
    },
    {
      "epoch": 0.6992114609041112,
      "grad_norm": 1.7709884643554688,
      "learning_rate": 4.4006014336816035e-05,
      "loss": 1.1706,
      "step": 1452
    },
    {
      "epoch": 0.6996930114970205,
      "grad_norm": 1.410871148109436,
      "learning_rate": 4.387684149947837e-05,
      "loss": 0.6805,
      "step": 1453
    },
    {
      "epoch": 0.7001745620899296,
      "grad_norm": 1.3541136980056763,
      "learning_rate": 4.374780522192726e-05,
      "loss": 1.3764,
      "step": 1454
    },
    {
      "epoch": 0.7006561126828388,
      "grad_norm": 1.1134296655654907,
      "learning_rate": 4.3618905818135805e-05,
      "loss": 1.0796,
      "step": 1455
    },
    {
      "epoch": 0.7011376632757479,
      "grad_norm": 1.741981863975525,
      "learning_rate": 4.349014360174417e-05,
      "loss": 1.0365,
      "step": 1456
    },
    {
      "epoch": 0.7016192138686571,
      "grad_norm": 1.332885503768921,
      "learning_rate": 4.336151888605871e-05,
      "loss": 1.0993,
      "step": 1457
    },
    {
      "epoch": 0.7021007644615662,
      "grad_norm": 0.8882187008857727,
      "learning_rate": 4.323303198405117e-05,
      "loss": 1.3629,
      "step": 1458
    },
    {
      "epoch": 0.7025823150544754,
      "grad_norm": 1.9233535528182983,
      "learning_rate": 4.310468320835796e-05,
      "loss": 1.3752,
      "step": 1459
    },
    {
      "epoch": 0.7030638656473845,
      "grad_norm": 1.049921989440918,
      "learning_rate": 4.297647287127946e-05,
      "loss": 0.7382,
      "step": 1460
    },
    {
      "epoch": 0.7035454162402938,
      "grad_norm": 1.0032683610916138,
      "learning_rate": 4.284840128477913e-05,
      "loss": 1.3867,
      "step": 1461
    },
    {
      "epoch": 0.7040269668332029,
      "grad_norm": 0.6361508369445801,
      "learning_rate": 4.2720468760482854e-05,
      "loss": 0.6359,
      "step": 1462
    },
    {
      "epoch": 0.7045085174261121,
      "grad_norm": 1.1217848062515259,
      "learning_rate": 4.2592675609678135e-05,
      "loss": 0.8278,
      "step": 1463
    },
    {
      "epoch": 0.7049900680190212,
      "grad_norm": 0.9972174167633057,
      "learning_rate": 4.24650221433134e-05,
      "loss": 0.991,
      "step": 1464
    },
    {
      "epoch": 0.7054716186119304,
      "grad_norm": 0.7988883852958679,
      "learning_rate": 4.2337508671997086e-05,
      "loss": 0.748,
      "step": 1465
    },
    {
      "epoch": 0.7059531692048396,
      "grad_norm": 1.302498459815979,
      "learning_rate": 4.221013550599707e-05,
      "loss": 0.8978,
      "step": 1466
    },
    {
      "epoch": 0.7064347197977487,
      "grad_norm": 2.2597219944000244,
      "learning_rate": 4.208290295523984e-05,
      "loss": 0.9361,
      "step": 1467
    },
    {
      "epoch": 0.706916270390658,
      "grad_norm": 1.706140160560608,
      "learning_rate": 4.1955811329309746e-05,
      "loss": 1.3327,
      "step": 1468
    },
    {
      "epoch": 0.7073978209835671,
      "grad_norm": 1.2415703535079956,
      "learning_rate": 4.182886093744813e-05,
      "loss": 1.3903,
      "step": 1469
    },
    {
      "epoch": 0.7078793715764763,
      "grad_norm": 1.3145755529403687,
      "learning_rate": 4.170205208855281e-05,
      "loss": 1.123,
      "step": 1470
    },
    {
      "epoch": 0.7083609221693854,
      "grad_norm": 0.6678088307380676,
      "learning_rate": 4.157538509117714e-05,
      "loss": 0.8622,
      "step": 1471
    },
    {
      "epoch": 0.7088424727622946,
      "grad_norm": 1.3157100677490234,
      "learning_rate": 4.144886025352934e-05,
      "loss": 1.3002,
      "step": 1472
    },
    {
      "epoch": 0.7093240233552037,
      "grad_norm": 1.1868373155593872,
      "learning_rate": 4.13224778834717e-05,
      "loss": 0.8192,
      "step": 1473
    },
    {
      "epoch": 0.7098055739481129,
      "grad_norm": 1.7007673978805542,
      "learning_rate": 4.1196238288519874e-05,
      "loss": 1.0183,
      "step": 1474
    },
    {
      "epoch": 0.710287124541022,
      "grad_norm": 1.5121554136276245,
      "learning_rate": 4.107014177584211e-05,
      "loss": 0.7933,
      "step": 1475
    },
    {
      "epoch": 0.7107686751339313,
      "grad_norm": 1.0907368659973145,
      "learning_rate": 4.094418865225853e-05,
      "loss": 1.0658,
      "step": 1476
    },
    {
      "epoch": 0.7112502257268404,
      "grad_norm": 1.2506096363067627,
      "learning_rate": 4.081837922424027e-05,
      "loss": 0.7582,
      "step": 1477
    },
    {
      "epoch": 0.7117317763197496,
      "grad_norm": 0.9508859515190125,
      "learning_rate": 4.069271379790891e-05,
      "loss": 0.6525,
      "step": 1478
    },
    {
      "epoch": 0.7122133269126587,
      "grad_norm": 1.5834177732467651,
      "learning_rate": 4.0567192679035636e-05,
      "loss": 1.3429,
      "step": 1479
    },
    {
      "epoch": 0.7126948775055679,
      "grad_norm": 1.0601916313171387,
      "learning_rate": 4.044181617304048e-05,
      "loss": 0.6514,
      "step": 1480
    },
    {
      "epoch": 0.7131764280984771,
      "grad_norm": 0.7968013882637024,
      "learning_rate": 4.03165845849916e-05,
      "loss": 0.6831,
      "step": 1481
    },
    {
      "epoch": 0.7136579786913863,
      "grad_norm": 1.6858556270599365,
      "learning_rate": 4.019149821960455e-05,
      "loss": 1.1264,
      "step": 1482
    },
    {
      "epoch": 0.7141395292842955,
      "grad_norm": 1.9675416946411133,
      "learning_rate": 4.006655738124152e-05,
      "loss": 1.4078,
      "step": 1483
    },
    {
      "epoch": 0.7146210798772046,
      "grad_norm": 1.0574262142181396,
      "learning_rate": 3.9941762373910586e-05,
      "loss": 0.8147,
      "step": 1484
    },
    {
      "epoch": 0.7151026304701138,
      "grad_norm": 1.302774429321289,
      "learning_rate": 3.9817113501265016e-05,
      "loss": 0.9749,
      "step": 1485
    },
    {
      "epoch": 0.7155841810630229,
      "grad_norm": 0.947575032711029,
      "learning_rate": 3.9692611066602516e-05,
      "loss": 1.1432,
      "step": 1486
    },
    {
      "epoch": 0.7160657316559321,
      "grad_norm": 1.0743399858474731,
      "learning_rate": 3.956825537286436e-05,
      "loss": 0.5454,
      "step": 1487
    },
    {
      "epoch": 0.7165472822488412,
      "grad_norm": 0.817898154258728,
      "learning_rate": 3.944404672263494e-05,
      "loss": 1.0663,
      "step": 1488
    },
    {
      "epoch": 0.7170288328417505,
      "grad_norm": 1.6338201761245728,
      "learning_rate": 3.931998541814069e-05,
      "loss": 0.9855,
      "step": 1489
    },
    {
      "epoch": 0.7175103834346596,
      "grad_norm": 1.1315727233886719,
      "learning_rate": 3.919607176124966e-05,
      "loss": 1.1625,
      "step": 1490
    },
    {
      "epoch": 0.7179919340275688,
      "grad_norm": 2.0492475032806396,
      "learning_rate": 3.9072306053470566e-05,
      "loss": 0.9772,
      "step": 1491
    },
    {
      "epoch": 0.7184734846204779,
      "grad_norm": 1.2393330335617065,
      "learning_rate": 3.8948688595952164e-05,
      "loss": 0.5491,
      "step": 1492
    },
    {
      "epoch": 0.7189550352133871,
      "grad_norm": 1.2757458686828613,
      "learning_rate": 3.882521968948246e-05,
      "loss": 1.1177,
      "step": 1493
    },
    {
      "epoch": 0.7194365858062963,
      "grad_norm": 0.9154059290885925,
      "learning_rate": 3.8701899634488014e-05,
      "loss": 1.1136,
      "step": 1494
    },
    {
      "epoch": 0.7199181363992054,
      "grad_norm": 0.9303089380264282,
      "learning_rate": 3.857872873103322e-05,
      "loss": 0.625,
      "step": 1495
    },
    {
      "epoch": 0.7203996869921147,
      "grad_norm": 1.1693419218063354,
      "learning_rate": 3.8455707278819507e-05,
      "loss": 0.8755,
      "step": 1496
    },
    {
      "epoch": 0.7208812375850238,
      "grad_norm": 1.4991776943206787,
      "learning_rate": 3.833283557718471e-05,
      "loss": 0.8142,
      "step": 1497
    },
    {
      "epoch": 0.721362788177933,
      "grad_norm": 0.860416054725647,
      "learning_rate": 3.821011392510228e-05,
      "loss": 0.8,
      "step": 1498
    },
    {
      "epoch": 0.7218443387708421,
      "grad_norm": 1.4386273622512817,
      "learning_rate": 3.808754262118046e-05,
      "loss": 0.7314,
      "step": 1499
    },
    {
      "epoch": 0.7223258893637513,
      "grad_norm": 0.8807114362716675,
      "learning_rate": 3.796512196366182e-05,
      "loss": 0.9046,
      "step": 1500
    },
    {
      "epoch": 0.7228074399566604,
      "grad_norm": 1.303338646888733,
      "learning_rate": 3.784285225042229e-05,
      "loss": 1.2848,
      "step": 1501
    },
    {
      "epoch": 0.7232889905495696,
      "grad_norm": 1.5512970685958862,
      "learning_rate": 3.772073377897052e-05,
      "loss": 1.1972,
      "step": 1502
    },
    {
      "epoch": 0.7237705411424787,
      "grad_norm": 1.2404366731643677,
      "learning_rate": 3.7598766846447184e-05,
      "loss": 1.0546,
      "step": 1503
    },
    {
      "epoch": 0.724252091735388,
      "grad_norm": 0.9745569229125977,
      "learning_rate": 3.747695174962423e-05,
      "loss": 1.2306,
      "step": 1504
    },
    {
      "epoch": 0.7247336423282971,
      "grad_norm": 1.5072669982910156,
      "learning_rate": 3.7355288784904116e-05,
      "loss": 1.1781,
      "step": 1505
    },
    {
      "epoch": 0.7252151929212063,
      "grad_norm": 1.6194456815719604,
      "learning_rate": 3.7233778248319176e-05,
      "loss": 1.0238,
      "step": 1506
    },
    {
      "epoch": 0.7256967435141155,
      "grad_norm": 0.8337032198905945,
      "learning_rate": 3.7112420435530845e-05,
      "loss": 0.8484,
      "step": 1507
    },
    {
      "epoch": 0.7261782941070246,
      "grad_norm": 2.076263666152954,
      "learning_rate": 3.69912156418289e-05,
      "loss": 0.9451,
      "step": 1508
    },
    {
      "epoch": 0.7266598446999338,
      "grad_norm": 0.8466574549674988,
      "learning_rate": 3.687016416213084e-05,
      "loss": 0.9896,
      "step": 1509
    },
    {
      "epoch": 0.7271413952928429,
      "grad_norm": 1.45832359790802,
      "learning_rate": 3.674926629098113e-05,
      "loss": 0.6598,
      "step": 1510
    },
    {
      "epoch": 0.7276229458857522,
      "grad_norm": 1.6322743892669678,
      "learning_rate": 3.6628522322550394e-05,
      "loss": 0.9753,
      "step": 1511
    },
    {
      "epoch": 0.7281044964786613,
      "grad_norm": 1.3642991781234741,
      "learning_rate": 3.6507932550634846e-05,
      "loss": 0.8041,
      "step": 1512
    },
    {
      "epoch": 0.7285860470715705,
      "grad_norm": 2.366534948348999,
      "learning_rate": 3.638749726865552e-05,
      "loss": 1.0187,
      "step": 1513
    },
    {
      "epoch": 0.7290675976644796,
      "grad_norm": 0.8270635604858398,
      "learning_rate": 3.6267216769657485e-05,
      "loss": 1.2884,
      "step": 1514
    },
    {
      "epoch": 0.7295491482573888,
      "grad_norm": 1.1024192571640015,
      "learning_rate": 3.6147091346309224e-05,
      "loss": 1.1454,
      "step": 1515
    },
    {
      "epoch": 0.7300306988502979,
      "grad_norm": 1.2304437160491943,
      "learning_rate": 3.602712129090189e-05,
      "loss": 0.808,
      "step": 1516
    },
    {
      "epoch": 0.7305122494432071,
      "grad_norm": 1.2767695188522339,
      "learning_rate": 3.590730689534857e-05,
      "loss": 1.0378,
      "step": 1517
    },
    {
      "epoch": 0.7309938000361162,
      "grad_norm": 2.23051381111145,
      "learning_rate": 3.578764845118362e-05,
      "loss": 1.0889,
      "step": 1518
    },
    {
      "epoch": 0.7314753506290255,
      "grad_norm": 3.2993433475494385,
      "learning_rate": 3.566814624956194e-05,
      "loss": 1.2595,
      "step": 1519
    },
    {
      "epoch": 0.7319569012219346,
      "grad_norm": 1.3207457065582275,
      "learning_rate": 3.554880058125819e-05,
      "loss": 0.6184,
      "step": 1520
    },
    {
      "epoch": 0.7324384518148438,
      "grad_norm": 1.3402554988861084,
      "learning_rate": 3.5429611736666235e-05,
      "loss": 0.8969,
      "step": 1521
    },
    {
      "epoch": 0.732920002407753,
      "grad_norm": 0.8190476894378662,
      "learning_rate": 3.53105800057983e-05,
      "loss": 0.7872,
      "step": 1522
    },
    {
      "epoch": 0.7334015530006621,
      "grad_norm": 2.015207290649414,
      "learning_rate": 3.519170567828435e-05,
      "loss": 0.9681,
      "step": 1523
    },
    {
      "epoch": 0.7338831035935713,
      "grad_norm": 2.461707353591919,
      "learning_rate": 3.507298904337134e-05,
      "loss": 1.5624,
      "step": 1524
    },
    {
      "epoch": 0.7343646541864804,
      "grad_norm": 1.14073646068573,
      "learning_rate": 3.495443038992253e-05,
      "loss": 0.7439,
      "step": 1525
    },
    {
      "epoch": 0.7348462047793897,
      "grad_norm": 1.6873282194137573,
      "learning_rate": 3.4836030006416775e-05,
      "loss": 0.9676,
      "step": 1526
    },
    {
      "epoch": 0.7353277553722988,
      "grad_norm": 1.208768367767334,
      "learning_rate": 3.471778818094785e-05,
      "loss": 1.3361,
      "step": 1527
    },
    {
      "epoch": 0.735809305965208,
      "grad_norm": 0.9515869617462158,
      "learning_rate": 3.459970520122364e-05,
      "loss": 0.8256,
      "step": 1528
    },
    {
      "epoch": 0.7362908565581171,
      "grad_norm": 1.1123392581939697,
      "learning_rate": 3.44817813545656e-05,
      "loss": 0.7005,
      "step": 1529
    },
    {
      "epoch": 0.7367724071510263,
      "grad_norm": 2.2039904594421387,
      "learning_rate": 3.4364016927907974e-05,
      "loss": 1.1032,
      "step": 1530
    },
    {
      "epoch": 0.7372539577439354,
      "grad_norm": 1.3004488945007324,
      "learning_rate": 3.424641220779711e-05,
      "loss": 1.5031,
      "step": 1531
    },
    {
      "epoch": 0.7377355083368446,
      "grad_norm": 1.4476922750473022,
      "learning_rate": 3.412896748039067e-05,
      "loss": 1.0745,
      "step": 1532
    },
    {
      "epoch": 0.7382170589297538,
      "grad_norm": 0.7874704003334045,
      "learning_rate": 3.401168303145713e-05,
      "loss": 0.8033,
      "step": 1533
    },
    {
      "epoch": 0.738698609522663,
      "grad_norm": 0.9005951285362244,
      "learning_rate": 3.3894559146374924e-05,
      "loss": 0.6565,
      "step": 1534
    },
    {
      "epoch": 0.7391801601155722,
      "grad_norm": 1.3831499814987183,
      "learning_rate": 3.37775961101318e-05,
      "loss": 0.8558,
      "step": 1535
    },
    {
      "epoch": 0.7396617107084813,
      "grad_norm": 1.3143084049224854,
      "learning_rate": 3.366079420732413e-05,
      "loss": 0.9753,
      "step": 1536
    },
    {
      "epoch": 0.7401432613013905,
      "grad_norm": 1.4303947687149048,
      "learning_rate": 3.3544153722156216e-05,
      "loss": 0.9523,
      "step": 1537
    },
    {
      "epoch": 0.7406248118942996,
      "grad_norm": 2.141397476196289,
      "learning_rate": 3.3427674938439594e-05,
      "loss": 0.9265,
      "step": 1538
    },
    {
      "epoch": 0.7411063624872088,
      "grad_norm": 0.8365777730941772,
      "learning_rate": 3.3311358139592317e-05,
      "loss": 0.6724,
      "step": 1539
    },
    {
      "epoch": 0.741587913080118,
      "grad_norm": 1.1216137409210205,
      "learning_rate": 3.319520360863837e-05,
      "loss": 0.8722,
      "step": 1540
    },
    {
      "epoch": 0.7420694636730272,
      "grad_norm": 1.93210768699646,
      "learning_rate": 3.3079211628206854e-05,
      "loss": 0.6907,
      "step": 1541
    },
    {
      "epoch": 0.7425510142659363,
      "grad_norm": 1.2532786130905151,
      "learning_rate": 3.296338248053129e-05,
      "loss": 0.7464,
      "step": 1542
    },
    {
      "epoch": 0.7430325648588455,
      "grad_norm": 1.1532790660858154,
      "learning_rate": 3.2847716447449096e-05,
      "loss": 1.1418,
      "step": 1543
    },
    {
      "epoch": 0.7435141154517546,
      "grad_norm": 1.905920386314392,
      "learning_rate": 3.2732213810400745e-05,
      "loss": 0.9258,
      "step": 1544
    },
    {
      "epoch": 0.7439956660446638,
      "grad_norm": 0.7798826098442078,
      "learning_rate": 3.261687485042915e-05,
      "loss": 0.6651,
      "step": 1545
    },
    {
      "epoch": 0.7444772166375729,
      "grad_norm": 1.4036656618118286,
      "learning_rate": 3.250169984817897e-05,
      "loss": 1.5675,
      "step": 1546
    },
    {
      "epoch": 0.7449587672304822,
      "grad_norm": 1.1912508010864258,
      "learning_rate": 3.238668908389586e-05,
      "loss": 1.0666,
      "step": 1547
    },
    {
      "epoch": 0.7454403178233914,
      "grad_norm": 1.462277889251709,
      "learning_rate": 3.227184283742591e-05,
      "loss": 1.0765,
      "step": 1548
    },
    {
      "epoch": 0.7459218684163005,
      "grad_norm": 1.4211452007293701,
      "learning_rate": 3.215716138821488e-05,
      "loss": 0.9951,
      "step": 1549
    },
    {
      "epoch": 0.7464034190092097,
      "grad_norm": 2.275908946990967,
      "learning_rate": 3.204264501530756e-05,
      "loss": 0.9807,
      "step": 1550
    },
    {
      "epoch": 0.7468849696021188,
      "grad_norm": 1.334725022315979,
      "learning_rate": 3.192829399734706e-05,
      "loss": 1.2879,
      "step": 1551
    },
    {
      "epoch": 0.747366520195028,
      "grad_norm": 1.395117163658142,
      "learning_rate": 3.181410861257413e-05,
      "loss": 1.0134,
      "step": 1552
    },
    {
      "epoch": 0.7478480707879371,
      "grad_norm": 1.5626182556152344,
      "learning_rate": 3.170008913882656e-05,
      "loss": 1.012,
      "step": 1553
    },
    {
      "epoch": 0.7483296213808464,
      "grad_norm": 0.6520123481750488,
      "learning_rate": 3.1586235853538325e-05,
      "loss": 0.7782,
      "step": 1554
    },
    {
      "epoch": 0.7488111719737555,
      "grad_norm": 2.306577682495117,
      "learning_rate": 3.1472549033739126e-05,
      "loss": 0.6796,
      "step": 1555
    },
    {
      "epoch": 0.7492927225666647,
      "grad_norm": 1.3525209426879883,
      "learning_rate": 3.1359028956053615e-05,
      "loss": 1.0408,
      "step": 1556
    },
    {
      "epoch": 0.7497742731595738,
      "grad_norm": 1.0601783990859985,
      "learning_rate": 3.1245675896700685e-05,
      "loss": 1.6405,
      "step": 1557
    },
    {
      "epoch": 0.750255823752483,
      "grad_norm": 1.5352672338485718,
      "learning_rate": 3.113249013149284e-05,
      "loss": 0.8528,
      "step": 1558
    },
    {
      "epoch": 0.7507373743453921,
      "grad_norm": 1.416002869606018,
      "learning_rate": 3.101947193583557e-05,
      "loss": 1.1975,
      "step": 1559
    },
    {
      "epoch": 0.7512189249383013,
      "grad_norm": 1.0686691999435425,
      "learning_rate": 3.0906621584726546e-05,
      "loss": 1.0953,
      "step": 1560
    },
    {
      "epoch": 0.7517004755312104,
      "grad_norm": 1.7240899801254272,
      "learning_rate": 3.079393935275513e-05,
      "loss": 1.2278,
      "step": 1561
    },
    {
      "epoch": 0.7521820261241197,
      "grad_norm": 0.9483497142791748,
      "learning_rate": 3.068142551410155e-05,
      "loss": 0.5209,
      "step": 1562
    },
    {
      "epoch": 0.7526635767170289,
      "grad_norm": 1.7931760549545288,
      "learning_rate": 3.0569080342536347e-05,
      "loss": 1.0836,
      "step": 1563
    },
    {
      "epoch": 0.753145127309938,
      "grad_norm": 1.7884643077850342,
      "learning_rate": 3.0456904111419572e-05,
      "loss": 1.4918,
      "step": 1564
    },
    {
      "epoch": 0.7536266779028472,
      "grad_norm": 1.1318650245666504,
      "learning_rate": 3.034489709370033e-05,
      "loss": 0.7584,
      "step": 1565
    },
    {
      "epoch": 0.7541082284957563,
      "grad_norm": 1.5290600061416626,
      "learning_rate": 3.0233059561915855e-05,
      "loss": 0.8625,
      "step": 1566
    },
    {
      "epoch": 0.7545897790886655,
      "grad_norm": 1.882394790649414,
      "learning_rate": 3.01213917881911e-05,
      "loss": 1.1046,
      "step": 1567
    },
    {
      "epoch": 0.7550713296815746,
      "grad_norm": 2.8446288108825684,
      "learning_rate": 3.0009894044237907e-05,
      "loss": 1.2174,
      "step": 1568
    },
    {
      "epoch": 0.7555528802744839,
      "grad_norm": 0.8607590794563293,
      "learning_rate": 2.9898566601354418e-05,
      "loss": 0.8202,
      "step": 1569
    },
    {
      "epoch": 0.756034430867393,
      "grad_norm": 1.5302932262420654,
      "learning_rate": 2.9787409730424374e-05,
      "loss": 1.1249,
      "step": 1570
    },
    {
      "epoch": 0.7565159814603022,
      "grad_norm": 1.0047945976257324,
      "learning_rate": 2.96764237019165e-05,
      "loss": 1.0182,
      "step": 1571
    },
    {
      "epoch": 0.7569975320532113,
      "grad_norm": 1.4508897066116333,
      "learning_rate": 2.9565608785883815e-05,
      "loss": 1.1479,
      "step": 1572
    },
    {
      "epoch": 0.7574790826461205,
      "grad_norm": 1.592687964439392,
      "learning_rate": 2.9454965251962973e-05,
      "loss": 1.0707,
      "step": 1573
    },
    {
      "epoch": 0.7579606332390296,
      "grad_norm": 1.9797749519348145,
      "learning_rate": 2.9344493369373637e-05,
      "loss": 1.0988,
      "step": 1574
    },
    {
      "epoch": 0.7584421838319388,
      "grad_norm": 1.1662919521331787,
      "learning_rate": 2.9234193406917833e-05,
      "loss": 1.6237,
      "step": 1575
    },
    {
      "epoch": 0.7589237344248481,
      "grad_norm": 2.1466081142425537,
      "learning_rate": 2.912406563297916e-05,
      "loss": 1.8034,
      "step": 1576
    },
    {
      "epoch": 0.7594052850177572,
      "grad_norm": 0.9439541101455688,
      "learning_rate": 2.901411031552236e-05,
      "loss": 0.7539,
      "step": 1577
    },
    {
      "epoch": 0.7598868356106664,
      "grad_norm": 0.795150101184845,
      "learning_rate": 2.8904327722092495e-05,
      "loss": 1.0441,
      "step": 1578
    },
    {
      "epoch": 0.7603683862035755,
      "grad_norm": 1.7991502285003662,
      "learning_rate": 2.879471811981437e-05,
      "loss": 1.3322,
      "step": 1579
    },
    {
      "epoch": 0.7608499367964847,
      "grad_norm": 1.3999747037887573,
      "learning_rate": 2.868528177539187e-05,
      "loss": 1.3627,
      "step": 1580
    },
    {
      "epoch": 0.7613314873893938,
      "grad_norm": 1.203881859779358,
      "learning_rate": 2.8576018955107285e-05,
      "loss": 1.1353,
      "step": 1581
    },
    {
      "epoch": 0.761813037982303,
      "grad_norm": 1.1931161880493164,
      "learning_rate": 2.8466929924820705e-05,
      "loss": 1.301,
      "step": 1582
    },
    {
      "epoch": 0.7622945885752122,
      "grad_norm": 1.9760181903839111,
      "learning_rate": 2.8358014949969334e-05,
      "loss": 1.282,
      "step": 1583
    },
    {
      "epoch": 0.7627761391681214,
      "grad_norm": 1.9036422967910767,
      "learning_rate": 2.8249274295566864e-05,
      "loss": 0.9995,
      "step": 1584
    },
    {
      "epoch": 0.7632576897610305,
      "grad_norm": 1.434086799621582,
      "learning_rate": 2.8140708226202884e-05,
      "loss": 1.0437,
      "step": 1585
    },
    {
      "epoch": 0.7637392403539397,
      "grad_norm": 2.7155611515045166,
      "learning_rate": 2.803231700604204e-05,
      "loss": 0.9243,
      "step": 1586
    },
    {
      "epoch": 0.7642207909468488,
      "grad_norm": 1.4511245489120483,
      "learning_rate": 2.7924100898823702e-05,
      "loss": 1.0818,
      "step": 1587
    },
    {
      "epoch": 0.764702341539758,
      "grad_norm": 0.6047967672348022,
      "learning_rate": 2.7816060167861002e-05,
      "loss": 0.8779,
      "step": 1588
    },
    {
      "epoch": 0.7651838921326672,
      "grad_norm": 1.2775282859802246,
      "learning_rate": 2.7708195076040445e-05,
      "loss": 0.8294,
      "step": 1589
    },
    {
      "epoch": 0.7656654427255764,
      "grad_norm": 0.8086519837379456,
      "learning_rate": 2.760050588582114e-05,
      "loss": 1.2524,
      "step": 1590
    },
    {
      "epoch": 0.7661469933184856,
      "grad_norm": 1.572804570198059,
      "learning_rate": 2.749299285923417e-05,
      "loss": 1.1163,
      "step": 1591
    },
    {
      "epoch": 0.7666285439113947,
      "grad_norm": 1.5714657306671143,
      "learning_rate": 2.7385656257881997e-05,
      "loss": 0.7362,
      "step": 1592
    },
    {
      "epoch": 0.7671100945043039,
      "grad_norm": 7.148636341094971,
      "learning_rate": 2.7278496342937788e-05,
      "loss": 0.4732,
      "step": 1593
    },
    {
      "epoch": 0.767591645097213,
      "grad_norm": 1.4571722745895386,
      "learning_rate": 2.717151337514482e-05,
      "loss": 0.8289,
      "step": 1594
    },
    {
      "epoch": 0.7680731956901222,
      "grad_norm": 1.1533465385437012,
      "learning_rate": 2.7064707614815776e-05,
      "loss": 0.9716,
      "step": 1595
    },
    {
      "epoch": 0.7685547462830313,
      "grad_norm": 2.4930906295776367,
      "learning_rate": 2.6958079321832185e-05,
      "loss": 0.9099,
      "step": 1596
    },
    {
      "epoch": 0.7690362968759406,
      "grad_norm": 1.533561110496521,
      "learning_rate": 2.6851628755643776e-05,
      "loss": 1.0338,
      "step": 1597
    },
    {
      "epoch": 0.7695178474688497,
      "grad_norm": 1.4437587261199951,
      "learning_rate": 2.6745356175267765e-05,
      "loss": 1.0458,
      "step": 1598
    },
    {
      "epoch": 0.7699993980617589,
      "grad_norm": 2.031998872756958,
      "learning_rate": 2.6639261839288343e-05,
      "loss": 0.9946,
      "step": 1599
    },
    {
      "epoch": 0.770480948654668,
      "grad_norm": 1.4426003694534302,
      "learning_rate": 2.6533346005855987e-05,
      "loss": 1.2781,
      "step": 1600
    },
    {
      "epoch": 0.7709624992475772,
      "grad_norm": 0.8161165714263916,
      "learning_rate": 2.6427608932686843e-05,
      "loss": 0.5746,
      "step": 1601
    },
    {
      "epoch": 0.7714440498404863,
      "grad_norm": 2.3404717445373535,
      "learning_rate": 2.6322050877062064e-05,
      "loss": 1.0919,
      "step": 1602
    },
    {
      "epoch": 0.7719256004333955,
      "grad_norm": 1.3652180433273315,
      "learning_rate": 2.6216672095827266e-05,
      "loss": 1.1376,
      "step": 1603
    },
    {
      "epoch": 0.7724071510263048,
      "grad_norm": 1.6225255727767944,
      "learning_rate": 2.6111472845391827e-05,
      "loss": 0.9624,
      "step": 1604
    },
    {
      "epoch": 0.7728887016192139,
      "grad_norm": 0.6491601467132568,
      "learning_rate": 2.6006453381728236e-05,
      "loss": 0.5823,
      "step": 1605
    },
    {
      "epoch": 0.7733702522121231,
      "grad_norm": 1.5913759469985962,
      "learning_rate": 2.5901613960371585e-05,
      "loss": 0.7756,
      "step": 1606
    },
    {
      "epoch": 0.7738518028050322,
      "grad_norm": 1.037990689277649,
      "learning_rate": 2.5796954836418884e-05,
      "loss": 0.9103,
      "step": 1607
    },
    {
      "epoch": 0.7743333533979414,
      "grad_norm": 1.1184632778167725,
      "learning_rate": 2.569247626452842e-05,
      "loss": 1.1583,
      "step": 1608
    },
    {
      "epoch": 0.7748149039908505,
      "grad_norm": 1.5549561977386475,
      "learning_rate": 2.558817849891918e-05,
      "loss": 0.886,
      "step": 1609
    },
    {
      "epoch": 0.7752964545837597,
      "grad_norm": 1.2459309101104736,
      "learning_rate": 2.548406179337015e-05,
      "loss": 1.2806,
      "step": 1610
    },
    {
      "epoch": 0.7757780051766688,
      "grad_norm": 1.240311622619629,
      "learning_rate": 2.5380126401219807e-05,
      "loss": 0.9664,
      "step": 1611
    },
    {
      "epoch": 0.7762595557695781,
      "grad_norm": 1.4923044443130493,
      "learning_rate": 2.527637257536547e-05,
      "loss": 1.0005,
      "step": 1612
    },
    {
      "epoch": 0.7767411063624872,
      "grad_norm": 1.6498850584030151,
      "learning_rate": 2.517280056826262e-05,
      "loss": 0.9726,
      "step": 1613
    },
    {
      "epoch": 0.7772226569553964,
      "grad_norm": 0.9657262563705444,
      "learning_rate": 2.5069410631924385e-05,
      "loss": 0.9405,
      "step": 1614
    },
    {
      "epoch": 0.7777042075483055,
      "grad_norm": 1.3658956289291382,
      "learning_rate": 2.4966203017920818e-05,
      "loss": 0.8192,
      "step": 1615
    },
    {
      "epoch": 0.7781857581412147,
      "grad_norm": 0.9119521379470825,
      "learning_rate": 2.4863177977378392e-05,
      "loss": 0.9192,
      "step": 1616
    },
    {
      "epoch": 0.7786673087341239,
      "grad_norm": 0.41423511505126953,
      "learning_rate": 2.4760335760979312e-05,
      "loss": 0.7734,
      "step": 1617
    },
    {
      "epoch": 0.779148859327033,
      "grad_norm": 1.0119727849960327,
      "learning_rate": 2.4657676618960944e-05,
      "loss": 0.9848,
      "step": 1618
    },
    {
      "epoch": 0.7796304099199423,
      "grad_norm": 1.1602240800857544,
      "learning_rate": 2.455520080111522e-05,
      "loss": 1.1952,
      "step": 1619
    },
    {
      "epoch": 0.7801119605128514,
      "grad_norm": 1.4785152673721313,
      "learning_rate": 2.4452908556787912e-05,
      "loss": 1.3403,
      "step": 1620
    },
    {
      "epoch": 0.7805935111057606,
      "grad_norm": 1.1291165351867676,
      "learning_rate": 2.4350800134878203e-05,
      "loss": 1.0129,
      "step": 1621
    },
    {
      "epoch": 0.7810750616986697,
      "grad_norm": 1.3477588891983032,
      "learning_rate": 2.4248875783837987e-05,
      "loss": 0.6666,
      "step": 1622
    },
    {
      "epoch": 0.7815566122915789,
      "grad_norm": 1.6150904893875122,
      "learning_rate": 2.414713575167129e-05,
      "loss": 0.904,
      "step": 1623
    },
    {
      "epoch": 0.782038162884488,
      "grad_norm": 1.4399563074111938,
      "learning_rate": 2.4045580285933557e-05,
      "loss": 0.891,
      "step": 1624
    },
    {
      "epoch": 0.7825197134773972,
      "grad_norm": 2.0917887687683105,
      "learning_rate": 2.3944209633731242e-05,
      "loss": 1.1222,
      "step": 1625
    },
    {
      "epoch": 0.7830012640703063,
      "grad_norm": 1.661434531211853,
      "learning_rate": 2.3843024041721053e-05,
      "loss": 1.0306,
      "step": 1626
    },
    {
      "epoch": 0.7834828146632156,
      "grad_norm": 1.2105722427368164,
      "learning_rate": 2.3742023756109456e-05,
      "loss": 1.2007,
      "step": 1627
    },
    {
      "epoch": 0.7839643652561247,
      "grad_norm": 2.4509129524230957,
      "learning_rate": 2.3641209022651976e-05,
      "loss": 1.5157,
      "step": 1628
    },
    {
      "epoch": 0.7844459158490339,
      "grad_norm": 1.8058174848556519,
      "learning_rate": 2.3540580086652675e-05,
      "loss": 1.0862,
      "step": 1629
    },
    {
      "epoch": 0.7849274664419431,
      "grad_norm": 0.9262884855270386,
      "learning_rate": 2.344013719296353e-05,
      "loss": 0.8812,
      "step": 1630
    },
    {
      "epoch": 0.7854090170348522,
      "grad_norm": 0.670758843421936,
      "learning_rate": 2.3339880585983842e-05,
      "loss": 1.2164,
      "step": 1631
    },
    {
      "epoch": 0.7858905676277614,
      "grad_norm": 0.5160911679267883,
      "learning_rate": 2.3239810509659597e-05,
      "loss": 0.6216,
      "step": 1632
    },
    {
      "epoch": 0.7863721182206705,
      "grad_norm": 1.8558579683303833,
      "learning_rate": 2.313992720748295e-05,
      "loss": 1.3,
      "step": 1633
    },
    {
      "epoch": 0.7868536688135798,
      "grad_norm": 1.7251777648925781,
      "learning_rate": 2.304023092249159e-05,
      "loss": 1.025,
      "step": 1634
    },
    {
      "epoch": 0.7873352194064889,
      "grad_norm": 1.749752163887024,
      "learning_rate": 2.2940721897268136e-05,
      "loss": 1.6923,
      "step": 1635
    },
    {
      "epoch": 0.7878167699993981,
      "grad_norm": 1.5679839849472046,
      "learning_rate": 2.2841400373939592e-05,
      "loss": 1.1516,
      "step": 1636
    },
    {
      "epoch": 0.7882983205923072,
      "grad_norm": 1.0447665452957153,
      "learning_rate": 2.274226659417671e-05,
      "loss": 1.0232,
      "step": 1637
    },
    {
      "epoch": 0.7887798711852164,
      "grad_norm": 1.1124646663665771,
      "learning_rate": 2.2643320799193402e-05,
      "loss": 0.5733,
      "step": 1638
    },
    {
      "epoch": 0.7892614217781255,
      "grad_norm": 1.0584079027175903,
      "learning_rate": 2.2544563229746218e-05,
      "loss": 1.0848,
      "step": 1639
    },
    {
      "epoch": 0.7897429723710347,
      "grad_norm": 1.412977933883667,
      "learning_rate": 2.2445994126133708e-05,
      "loss": 1.1197,
      "step": 1640
    },
    {
      "epoch": 0.7902245229639439,
      "grad_norm": 1.5864454507827759,
      "learning_rate": 2.234761372819577e-05,
      "loss": 0.7381,
      "step": 1641
    },
    {
      "epoch": 0.7907060735568531,
      "grad_norm": 1.4704523086547852,
      "learning_rate": 2.2249422275313214e-05,
      "loss": 0.902,
      "step": 1642
    },
    {
      "epoch": 0.7911876241497622,
      "grad_norm": 1.0914117097854614,
      "learning_rate": 2.215142000640714e-05,
      "loss": 0.7538,
      "step": 1643
    },
    {
      "epoch": 0.7916691747426714,
      "grad_norm": 1.7431955337524414,
      "learning_rate": 2.2053607159938195e-05,
      "loss": 1.1072,
      "step": 1644
    },
    {
      "epoch": 0.7921507253355806,
      "grad_norm": 1.1136835813522339,
      "learning_rate": 2.1955983973906236e-05,
      "loss": 1.16,
      "step": 1645
    },
    {
      "epoch": 0.7926322759284897,
      "grad_norm": 1.987646460533142,
      "learning_rate": 2.1858550685849578e-05,
      "loss": 1.5347,
      "step": 1646
    },
    {
      "epoch": 0.793113826521399,
      "grad_norm": 2.8198578357696533,
      "learning_rate": 2.17613075328445e-05,
      "loss": 1.049,
      "step": 1647
    },
    {
      "epoch": 0.793595377114308,
      "grad_norm": 1.6491267681121826,
      "learning_rate": 2.1664254751504642e-05,
      "loss": 1.2853,
      "step": 1648
    },
    {
      "epoch": 0.7940769277072173,
      "grad_norm": 0.9724451899528503,
      "learning_rate": 2.1567392577980393e-05,
      "loss": 0.7166,
      "step": 1649
    },
    {
      "epoch": 0.7945584783001264,
      "grad_norm": 1.3596974611282349,
      "learning_rate": 2.1470721247958404e-05,
      "loss": 1.4258,
      "step": 1650
    },
    {
      "epoch": 0.7950400288930356,
      "grad_norm": 0.8474813103675842,
      "learning_rate": 2.137424099666091e-05,
      "loss": 1.1261,
      "step": 1651
    },
    {
      "epoch": 0.7955215794859447,
      "grad_norm": 1.0039057731628418,
      "learning_rate": 2.1277952058845284e-05,
      "loss": 0.7619,
      "step": 1652
    },
    {
      "epoch": 0.7960031300788539,
      "grad_norm": 1.2193447351455688,
      "learning_rate": 2.118185466880327e-05,
      "loss": 1.1294,
      "step": 1653
    },
    {
      "epoch": 0.796484680671763,
      "grad_norm": 0.8053333163261414,
      "learning_rate": 2.1085949060360654e-05,
      "loss": 1.0368,
      "step": 1654
    },
    {
      "epoch": 0.7969662312646723,
      "grad_norm": 1.967658519744873,
      "learning_rate": 2.0990235466876517e-05,
      "loss": 0.8924,
      "step": 1655
    },
    {
      "epoch": 0.7974477818575814,
      "grad_norm": 1.2955012321472168,
      "learning_rate": 2.089471412124274e-05,
      "loss": 0.8684,
      "step": 1656
    },
    {
      "epoch": 0.7979293324504906,
      "grad_norm": 0.8939071297645569,
      "learning_rate": 2.079938525588342e-05,
      "loss": 0.6564,
      "step": 1657
    },
    {
      "epoch": 0.7984108830433998,
      "grad_norm": 0.7493122220039368,
      "learning_rate": 2.0704249102754324e-05,
      "loss": 1.2537,
      "step": 1658
    },
    {
      "epoch": 0.7988924336363089,
      "grad_norm": 1.0269842147827148,
      "learning_rate": 2.0609305893342278e-05,
      "loss": 0.9487,
      "step": 1659
    },
    {
      "epoch": 0.7993739842292181,
      "grad_norm": 1.1473485231399536,
      "learning_rate": 2.0514555858664663e-05,
      "loss": 1.0869,
      "step": 1660
    },
    {
      "epoch": 0.7998555348221272,
      "grad_norm": 1.1629431247711182,
      "learning_rate": 2.0419999229268805e-05,
      "loss": 0.7488,
      "step": 1661
    },
    {
      "epoch": 0.8003370854150365,
      "grad_norm": 0.9898182153701782,
      "learning_rate": 2.032563623523147e-05,
      "loss": 1.4029,
      "step": 1662
    },
    {
      "epoch": 0.8008186360079456,
      "grad_norm": 1.0312323570251465,
      "learning_rate": 2.0231467106158186e-05,
      "loss": 0.6381,
      "step": 1663
    },
    {
      "epoch": 0.8013001866008548,
      "grad_norm": 2.3493716716766357,
      "learning_rate": 2.0137492071182863e-05,
      "loss": 1.3007,
      "step": 1664
    },
    {
      "epoch": 0.8017817371937639,
      "grad_norm": 1.5460721254348755,
      "learning_rate": 2.0043711358967043e-05,
      "loss": 1.7703,
      "step": 1665
    },
    {
      "epoch": 0.8022632877866731,
      "grad_norm": 1.2260479927062988,
      "learning_rate": 1.9950125197699508e-05,
      "loss": 0.9535,
      "step": 1666
    },
    {
      "epoch": 0.8027448383795822,
      "grad_norm": 1.2574350833892822,
      "learning_rate": 1.985673381509565e-05,
      "loss": 1.3317,
      "step": 1667
    },
    {
      "epoch": 0.8032263889724914,
      "grad_norm": 1.0495798587799072,
      "learning_rate": 1.9763537438396894e-05,
      "loss": 1.1513,
      "step": 1668
    },
    {
      "epoch": 0.8037079395654005,
      "grad_norm": 1.1890629529953003,
      "learning_rate": 1.96705362943702e-05,
      "loss": 0.9384,
      "step": 1669
    },
    {
      "epoch": 0.8041894901583098,
      "grad_norm": 1.176146388053894,
      "learning_rate": 1.9577730609307454e-05,
      "loss": 1.0353,
      "step": 1670
    },
    {
      "epoch": 0.804671040751219,
      "grad_norm": 1.3619112968444824,
      "learning_rate": 1.9485120609024975e-05,
      "loss": 0.7944,
      "step": 1671
    },
    {
      "epoch": 0.8051525913441281,
      "grad_norm": 1.557839274406433,
      "learning_rate": 1.9392706518862935e-05,
      "loss": 1.0984,
      "step": 1672
    },
    {
      "epoch": 0.8056341419370373,
      "grad_norm": 1.076859951019287,
      "learning_rate": 1.9300488563684804e-05,
      "loss": 0.8068,
      "step": 1673
    },
    {
      "epoch": 0.8061156925299464,
      "grad_norm": 0.7001736760139465,
      "learning_rate": 1.920846696787684e-05,
      "loss": 1.0244,
      "step": 1674
    },
    {
      "epoch": 0.8065972431228556,
      "grad_norm": 0.9017906785011292,
      "learning_rate": 1.9116641955347446e-05,
      "loss": 0.7633,
      "step": 1675
    },
    {
      "epoch": 0.8070787937157647,
      "grad_norm": 1.4248368740081787,
      "learning_rate": 1.9025013749526767e-05,
      "loss": 1.3303,
      "step": 1676
    },
    {
      "epoch": 0.807560344308674,
      "grad_norm": 1.5006868839263916,
      "learning_rate": 1.8933582573366036e-05,
      "loss": 1.3253,
      "step": 1677
    },
    {
      "epoch": 0.8080418949015831,
      "grad_norm": 0.9528875946998596,
      "learning_rate": 1.8842348649337116e-05,
      "loss": 0.9253,
      "step": 1678
    },
    {
      "epoch": 0.8085234454944923,
      "grad_norm": 0.9706194996833801,
      "learning_rate": 1.875131219943187e-05,
      "loss": 1.4926,
      "step": 1679
    },
    {
      "epoch": 0.8090049960874014,
      "grad_norm": 1.6707158088684082,
      "learning_rate": 1.8660473445161663e-05,
      "loss": 0.9237,
      "step": 1680
    },
    {
      "epoch": 0.8094865466803106,
      "grad_norm": 0.8742366433143616,
      "learning_rate": 1.856983260755686e-05,
      "loss": 1.0458,
      "step": 1681
    },
    {
      "epoch": 0.8099680972732197,
      "grad_norm": 1.1289740800857544,
      "learning_rate": 1.8479389907166223e-05,
      "loss": 0.7,
      "step": 1682
    },
    {
      "epoch": 0.8104496478661289,
      "grad_norm": 1.4022997617721558,
      "learning_rate": 1.8389145564056387e-05,
      "loss": 0.7136,
      "step": 1683
    },
    {
      "epoch": 0.810931198459038,
      "grad_norm": 1.2287734746932983,
      "learning_rate": 1.829909979781137e-05,
      "loss": 0.678,
      "step": 1684
    },
    {
      "epoch": 0.8114127490519473,
      "grad_norm": 1.589320421218872,
      "learning_rate": 1.820925282753201e-05,
      "loss": 1.0536,
      "step": 1685
    },
    {
      "epoch": 0.8118942996448565,
      "grad_norm": 1.084364891052246,
      "learning_rate": 1.8119604871835437e-05,
      "loss": 1.0798,
      "step": 1686
    },
    {
      "epoch": 0.8123758502377656,
      "grad_norm": 1.2457737922668457,
      "learning_rate": 1.8030156148854492e-05,
      "loss": 1.0926,
      "step": 1687
    },
    {
      "epoch": 0.8128574008306748,
      "grad_norm": 1.6498080492019653,
      "learning_rate": 1.7940906876237284e-05,
      "loss": 1.1191,
      "step": 1688
    },
    {
      "epoch": 0.8133389514235839,
      "grad_norm": 1.0763680934906006,
      "learning_rate": 1.78518572711466e-05,
      "loss": 0.9941,
      "step": 1689
    },
    {
      "epoch": 0.8138205020164931,
      "grad_norm": 1.5605993270874023,
      "learning_rate": 1.776300755025939e-05,
      "loss": 1.3372,
      "step": 1690
    },
    {
      "epoch": 0.8143020526094022,
      "grad_norm": 1.491334319114685,
      "learning_rate": 1.767435792976626e-05,
      "loss": 0.9861,
      "step": 1691
    },
    {
      "epoch": 0.8147836032023115,
      "grad_norm": 0.9926377534866333,
      "learning_rate": 1.7585908625370905e-05,
      "loss": 1.1993,
      "step": 1692
    },
    {
      "epoch": 0.8152651537952206,
      "grad_norm": 1.2687879800796509,
      "learning_rate": 1.749765985228963e-05,
      "loss": 1.2423,
      "step": 1693
    },
    {
      "epoch": 0.8157467043881298,
      "grad_norm": 1.3995667695999146,
      "learning_rate": 1.740961182525077e-05,
      "loss": 0.666,
      "step": 1694
    },
    {
      "epoch": 0.8162282549810389,
      "grad_norm": 0.7029280662536621,
      "learning_rate": 1.7321764758494252e-05,
      "loss": 1.0854,
      "step": 1695
    },
    {
      "epoch": 0.8167098055739481,
      "grad_norm": 1.4649654626846313,
      "learning_rate": 1.7234118865770987e-05,
      "loss": 1.2835,
      "step": 1696
    },
    {
      "epoch": 0.8171913561668572,
      "grad_norm": 1.2847552299499512,
      "learning_rate": 1.7146674360342373e-05,
      "loss": 1.1177,
      "step": 1697
    },
    {
      "epoch": 0.8176729067597664,
      "grad_norm": 1.0304591655731201,
      "learning_rate": 1.7059431454979824e-05,
      "loss": 1.2883,
      "step": 1698
    },
    {
      "epoch": 0.8181544573526757,
      "grad_norm": 4.674599647521973,
      "learning_rate": 1.6972390361964195e-05,
      "loss": 1.5203,
      "step": 1699
    },
    {
      "epoch": 0.8186360079455848,
      "grad_norm": 1.4923611879348755,
      "learning_rate": 1.688555129308531e-05,
      "loss": 1.0396,
      "step": 1700
    },
    {
      "epoch": 0.819117558538494,
      "grad_norm": 1.5413893461227417,
      "learning_rate": 1.6798914459641434e-05,
      "loss": 1.1047,
      "step": 1701
    },
    {
      "epoch": 0.8195991091314031,
      "grad_norm": 1.0391546487808228,
      "learning_rate": 1.6712480072438662e-05,
      "loss": 1.0181,
      "step": 1702
    },
    {
      "epoch": 0.8200806597243123,
      "grad_norm": 1.6693581342697144,
      "learning_rate": 1.6626248341790596e-05,
      "loss": 1.2507,
      "step": 1703
    },
    {
      "epoch": 0.8205622103172214,
      "grad_norm": 1.4764776229858398,
      "learning_rate": 1.6540219477517684e-05,
      "loss": 1.1429,
      "step": 1704
    },
    {
      "epoch": 0.8210437609101306,
      "grad_norm": 1.2193595170974731,
      "learning_rate": 1.6454393688946767e-05,
      "loss": 0.7094,
      "step": 1705
    },
    {
      "epoch": 0.8215253115030398,
      "grad_norm": 1.7810328006744385,
      "learning_rate": 1.6368771184910557e-05,
      "loss": 1.0903,
      "step": 1706
    },
    {
      "epoch": 0.822006862095949,
      "grad_norm": 1.2276769876480103,
      "learning_rate": 1.6283352173747145e-05,
      "loss": 1.0527,
      "step": 1707
    },
    {
      "epoch": 0.8224884126888581,
      "grad_norm": 0.9521345496177673,
      "learning_rate": 1.619813686329946e-05,
      "loss": 1.063,
      "step": 1708
    },
    {
      "epoch": 0.8229699632817673,
      "grad_norm": 0.9406653046607971,
      "learning_rate": 1.611312546091476e-05,
      "loss": 1.1315,
      "step": 1709
    },
    {
      "epoch": 0.8234515138746764,
      "grad_norm": 2.2442519664764404,
      "learning_rate": 1.6028318173444202e-05,
      "loss": 1.3555,
      "step": 1710
    },
    {
      "epoch": 0.8239330644675856,
      "grad_norm": 1.6496635675430298,
      "learning_rate": 1.594371520724226e-05,
      "loss": 0.8781,
      "step": 1711
    },
    {
      "epoch": 0.8244146150604948,
      "grad_norm": 0.7756935358047485,
      "learning_rate": 1.5859316768166244e-05,
      "loss": 0.8861,
      "step": 1712
    },
    {
      "epoch": 0.824896165653404,
      "grad_norm": 1.3517109155654907,
      "learning_rate": 1.5775123061575836e-05,
      "loss": 1.4605,
      "step": 1713
    },
    {
      "epoch": 0.8253777162463132,
      "grad_norm": 1.1789662837982178,
      "learning_rate": 1.569113429233252e-05,
      "loss": 0.8333,
      "step": 1714
    },
    {
      "epoch": 0.8258592668392223,
      "grad_norm": 1.2283135652542114,
      "learning_rate": 1.5607350664799157e-05,
      "loss": 1.3822,
      "step": 1715
    },
    {
      "epoch": 0.8263408174321315,
      "grad_norm": 2.2275712490081787,
      "learning_rate": 1.552377238283943e-05,
      "loss": 1.0787,
      "step": 1716
    },
    {
      "epoch": 0.8268223680250406,
      "grad_norm": 1.14742112159729,
      "learning_rate": 1.5440399649817385e-05,
      "loss": 0.7207,
      "step": 1717
    },
    {
      "epoch": 0.8273039186179498,
      "grad_norm": 1.1468846797943115,
      "learning_rate": 1.5357232668596933e-05,
      "loss": 1.607,
      "step": 1718
    },
    {
      "epoch": 0.8277854692108589,
      "grad_norm": 1.2647056579589844,
      "learning_rate": 1.5274271641541295e-05,
      "loss": 0.6784,
      "step": 1719
    },
    {
      "epoch": 0.8282670198037682,
      "grad_norm": 1.2414592504501343,
      "learning_rate": 1.5191516770512649e-05,
      "loss": 1.1059,
      "step": 1720
    },
    {
      "epoch": 0.8287485703966773,
      "grad_norm": 1.4528673887252808,
      "learning_rate": 1.5108968256871437e-05,
      "loss": 0.959,
      "step": 1721
    },
    {
      "epoch": 0.8292301209895865,
      "grad_norm": 0.9891937971115112,
      "learning_rate": 1.5026626301476087e-05,
      "loss": 0.9553,
      "step": 1722
    },
    {
      "epoch": 0.8297116715824956,
      "grad_norm": 1.4874505996704102,
      "learning_rate": 1.4944491104682379e-05,
      "loss": 0.9951,
      "step": 1723
    },
    {
      "epoch": 0.8301932221754048,
      "grad_norm": 0.9920333027839661,
      "learning_rate": 1.4862562866343034e-05,
      "loss": 1.0162,
      "step": 1724
    },
    {
      "epoch": 0.8306747727683139,
      "grad_norm": 1.9017934799194336,
      "learning_rate": 1.4780841785807164e-05,
      "loss": 1.5139,
      "step": 1725
    },
    {
      "epoch": 0.8311563233612231,
      "grad_norm": 1.7579786777496338,
      "learning_rate": 1.4699328061919848e-05,
      "loss": 0.7744,
      "step": 1726
    },
    {
      "epoch": 0.8316378739541324,
      "grad_norm": 1.472794532775879,
      "learning_rate": 1.4618021893021605e-05,
      "loss": 0.789,
      "step": 1727
    },
    {
      "epoch": 0.8321194245470415,
      "grad_norm": 0.7566042542457581,
      "learning_rate": 1.453692347694794e-05,
      "loss": 0.7419,
      "step": 1728
    },
    {
      "epoch": 0.8326009751399507,
      "grad_norm": 1.2207714319229126,
      "learning_rate": 1.4456033011028835e-05,
      "loss": 0.756,
      "step": 1729
    },
    {
      "epoch": 0.8330825257328598,
      "grad_norm": 3.2781076431274414,
      "learning_rate": 1.437535069208833e-05,
      "loss": 1.0199,
      "step": 1730
    },
    {
      "epoch": 0.833564076325769,
      "grad_norm": 1.6142027378082275,
      "learning_rate": 1.4294876716443906e-05,
      "loss": 1.0405,
      "step": 1731
    },
    {
      "epoch": 0.8340456269186781,
      "grad_norm": 1.2991055250167847,
      "learning_rate": 1.4214611279906187e-05,
      "loss": 0.7913,
      "step": 1732
    },
    {
      "epoch": 0.8345271775115873,
      "grad_norm": 0.7640783190727234,
      "learning_rate": 1.4134554577778337e-05,
      "loss": 0.9908,
      "step": 1733
    },
    {
      "epoch": 0.8350087281044964,
      "grad_norm": 0.8637135624885559,
      "learning_rate": 1.4054706804855634e-05,
      "loss": 1.0656,
      "step": 1734
    },
    {
      "epoch": 0.8354902786974057,
      "grad_norm": 0.9299548268318176,
      "learning_rate": 1.3975068155424976e-05,
      "loss": 1.5108,
      "step": 1735
    },
    {
      "epoch": 0.8359718292903148,
      "grad_norm": 1.3927792310714722,
      "learning_rate": 1.3895638823264446e-05,
      "loss": 1.2068,
      "step": 1736
    },
    {
      "epoch": 0.836453379883224,
      "grad_norm": 1.495082139968872,
      "learning_rate": 1.3816419001642777e-05,
      "loss": 1.2147,
      "step": 1737
    },
    {
      "epoch": 0.8369349304761331,
      "grad_norm": 2.2586164474487305,
      "learning_rate": 1.3737408883318948e-05,
      "loss": 1.3118,
      "step": 1738
    },
    {
      "epoch": 0.8374164810690423,
      "grad_norm": 1.5495485067367554,
      "learning_rate": 1.365860866054165e-05,
      "loss": 1.238,
      "step": 1739
    },
    {
      "epoch": 0.8378980316619515,
      "grad_norm": 1.0529330968856812,
      "learning_rate": 1.358001852504891e-05,
      "loss": 0.7745,
      "step": 1740
    },
    {
      "epoch": 0.8383795822548606,
      "grad_norm": 1.3768119812011719,
      "learning_rate": 1.3501638668067485e-05,
      "loss": 1.1778,
      "step": 1741
    },
    {
      "epoch": 0.8388611328477699,
      "grad_norm": 1.8559973239898682,
      "learning_rate": 1.3423469280312562e-05,
      "loss": 0.9764,
      "step": 1742
    },
    {
      "epoch": 0.839342683440679,
      "grad_norm": 1.4691377878189087,
      "learning_rate": 1.3345510551987128e-05,
      "loss": 0.9006,
      "step": 1743
    },
    {
      "epoch": 0.8398242340335882,
      "grad_norm": 1.3930045366287231,
      "learning_rate": 1.326776267278167e-05,
      "loss": 1.0521,
      "step": 1744
    },
    {
      "epoch": 0.8403057846264973,
      "grad_norm": 2.14896559715271,
      "learning_rate": 1.3190225831873581e-05,
      "loss": 1.2787,
      "step": 1745
    },
    {
      "epoch": 0.8407873352194065,
      "grad_norm": 1.1368664503097534,
      "learning_rate": 1.3112900217926782e-05,
      "loss": 1.2348,
      "step": 1746
    },
    {
      "epoch": 0.8412688858123156,
      "grad_norm": 1.338537573814392,
      "learning_rate": 1.3035786019091223e-05,
      "loss": 1.2228,
      "step": 1747
    },
    {
      "epoch": 0.8417504364052248,
      "grad_norm": 1.509090542793274,
      "learning_rate": 1.2958883423002422e-05,
      "loss": 1.0257,
      "step": 1748
    },
    {
      "epoch": 0.842231986998134,
      "grad_norm": 1.3313130140304565,
      "learning_rate": 1.288219261678103e-05,
      "loss": 0.9713,
      "step": 1749
    },
    {
      "epoch": 0.8427135375910432,
      "grad_norm": 1.3211489915847778,
      "learning_rate": 1.2805713787032381e-05,
      "loss": 0.6921,
      "step": 1750
    },
    {
      "epoch": 0.8431950881839523,
      "grad_norm": 0.9114959836006165,
      "learning_rate": 1.2729447119846016e-05,
      "loss": 0.7964,
      "step": 1751
    },
    {
      "epoch": 0.8436766387768615,
      "grad_norm": 1.2766203880310059,
      "learning_rate": 1.265339280079525e-05,
      "loss": 1.1503,
      "step": 1752
    },
    {
      "epoch": 0.8441581893697706,
      "grad_norm": 1.9224025011062622,
      "learning_rate": 1.257755101493665e-05,
      "loss": 1.0172,
      "step": 1753
    },
    {
      "epoch": 0.8446397399626798,
      "grad_norm": 1.0871329307556152,
      "learning_rate": 1.2501921946809714e-05,
      "loss": 0.9062,
      "step": 1754
    },
    {
      "epoch": 0.845121290555589,
      "grad_norm": 1.3384684324264526,
      "learning_rate": 1.2426505780436326e-05,
      "loss": 1.2903,
      "step": 1755
    },
    {
      "epoch": 0.8456028411484982,
      "grad_norm": 1.1397919654846191,
      "learning_rate": 1.2351302699320332e-05,
      "loss": 1.3649,
      "step": 1756
    },
    {
      "epoch": 0.8460843917414074,
      "grad_norm": 1.9695689678192139,
      "learning_rate": 1.2276312886447106e-05,
      "loss": 1.1589,
      "step": 1757
    },
    {
      "epoch": 0.8465659423343165,
      "grad_norm": 1.1453593969345093,
      "learning_rate": 1.2201536524283074e-05,
      "loss": 1.464,
      "step": 1758
    },
    {
      "epoch": 0.8470474929272257,
      "grad_norm": 0.7946385145187378,
      "learning_rate": 1.2126973794775343e-05,
      "loss": 0.9698,
      "step": 1759
    },
    {
      "epoch": 0.8475290435201348,
      "grad_norm": 2.089094638824463,
      "learning_rate": 1.2052624879351104e-05,
      "loss": 1.4394,
      "step": 1760
    },
    {
      "epoch": 0.848010594113044,
      "grad_norm": 1.9007961750030518,
      "learning_rate": 1.1978489958917382e-05,
      "loss": 1.2042,
      "step": 1761
    },
    {
      "epoch": 0.8484921447059531,
      "grad_norm": 0.9683759808540344,
      "learning_rate": 1.1904569213860472e-05,
      "loss": 0.871,
      "step": 1762
    },
    {
      "epoch": 0.8489736952988624,
      "grad_norm": 0.6783609986305237,
      "learning_rate": 1.1830862824045552e-05,
      "loss": 0.9506,
      "step": 1763
    },
    {
      "epoch": 0.8494552458917715,
      "grad_norm": 0.7889495491981506,
      "learning_rate": 1.1757370968816217e-05,
      "loss": 0.8332,
      "step": 1764
    },
    {
      "epoch": 0.8499367964846807,
      "grad_norm": 1.396812915802002,
      "learning_rate": 1.1684093826994024e-05,
      "loss": 0.6454,
      "step": 1765
    },
    {
      "epoch": 0.8504183470775898,
      "grad_norm": 1.170555591583252,
      "learning_rate": 1.1611031576878117e-05,
      "loss": 1.1174,
      "step": 1766
    },
    {
      "epoch": 0.850899897670499,
      "grad_norm": 1.023296594619751,
      "learning_rate": 1.1538184396244778e-05,
      "loss": 0.7332,
      "step": 1767
    },
    {
      "epoch": 0.8513814482634082,
      "grad_norm": 1.1360104084014893,
      "learning_rate": 1.146555246234694e-05,
      "loss": 1.1022,
      "step": 1768
    },
    {
      "epoch": 0.8518629988563173,
      "grad_norm": 1.6912881135940552,
      "learning_rate": 1.1393135951913824e-05,
      "loss": 1.163,
      "step": 1769
    },
    {
      "epoch": 0.8523445494492266,
      "grad_norm": 1.2755537033081055,
      "learning_rate": 1.132093504115046e-05,
      "loss": 0.9156,
      "step": 1770
    },
    {
      "epoch": 0.8528261000421357,
      "grad_norm": 1.4152694940567017,
      "learning_rate": 1.1248949905737283e-05,
      "loss": 0.8227,
      "step": 1771
    },
    {
      "epoch": 0.8533076506350449,
      "grad_norm": 0.7402531504631042,
      "learning_rate": 1.1177180720829694e-05,
      "loss": 0.4274,
      "step": 1772
    },
    {
      "epoch": 0.853789201227954,
      "grad_norm": 1.4452061653137207,
      "learning_rate": 1.1105627661057671e-05,
      "loss": 0.6975,
      "step": 1773
    },
    {
      "epoch": 0.8542707518208632,
      "grad_norm": 1.3618583679199219,
      "learning_rate": 1.103429090052528e-05,
      "loss": 1.3002,
      "step": 1774
    },
    {
      "epoch": 0.8547523024137723,
      "grad_norm": 0.8775191307067871,
      "learning_rate": 1.096317061281027e-05,
      "loss": 0.9172,
      "step": 1775
    },
    {
      "epoch": 0.8552338530066815,
      "grad_norm": 1.9647924900054932,
      "learning_rate": 1.0892266970963704e-05,
      "loss": 0.8293,
      "step": 1776
    },
    {
      "epoch": 0.8557154035995906,
      "grad_norm": 1.4634672403335571,
      "learning_rate": 1.082158014750948e-05,
      "loss": 1.704,
      "step": 1777
    },
    {
      "epoch": 0.8561969541924999,
      "grad_norm": 1.2343543767929077,
      "learning_rate": 1.0751110314443958e-05,
      "loss": 1.1633,
      "step": 1778
    },
    {
      "epoch": 0.856678504785409,
      "grad_norm": 1.3354381322860718,
      "learning_rate": 1.0680857643235431e-05,
      "loss": 1.1231,
      "step": 1779
    },
    {
      "epoch": 0.8571600553783182,
      "grad_norm": 1.1379916667938232,
      "learning_rate": 1.0610822304823887e-05,
      "loss": 0.7879,
      "step": 1780
    },
    {
      "epoch": 0.8576416059712274,
      "grad_norm": 2.157014846801758,
      "learning_rate": 1.0541004469620452e-05,
      "loss": 1.3558,
      "step": 1781
    },
    {
      "epoch": 0.8581231565641365,
      "grad_norm": 1.2952896356582642,
      "learning_rate": 1.0471404307507016e-05,
      "loss": 1.2211,
      "step": 1782
    },
    {
      "epoch": 0.8586047071570457,
      "grad_norm": 1.8142328262329102,
      "learning_rate": 1.0402021987835831e-05,
      "loss": 1.2093,
      "step": 1783
    },
    {
      "epoch": 0.8590862577499548,
      "grad_norm": 1.540513038635254,
      "learning_rate": 1.0332857679429098e-05,
      "loss": 0.7386,
      "step": 1784
    },
    {
      "epoch": 0.8595678083428641,
      "grad_norm": 1.5768598318099976,
      "learning_rate": 1.0263911550578531e-05,
      "loss": 1.0644,
      "step": 1785
    },
    {
      "epoch": 0.8600493589357732,
      "grad_norm": 1.8080589771270752,
      "learning_rate": 1.0195183769045013e-05,
      "loss": 0.5846,
      "step": 1786
    },
    {
      "epoch": 0.8605309095286824,
      "grad_norm": 1.0664002895355225,
      "learning_rate": 1.0126674502058054e-05,
      "loss": 1.0136,
      "step": 1787
    },
    {
      "epoch": 0.8610124601215915,
      "grad_norm": 1.2095695734024048,
      "learning_rate": 1.005838391631555e-05,
      "loss": 1.1975,
      "step": 1788
    },
    {
      "epoch": 0.8614940107145007,
      "grad_norm": 1.0594955682754517,
      "learning_rate": 9.990312177983263e-06,
      "loss": 1.0239,
      "step": 1789
    },
    {
      "epoch": 0.8619755613074098,
      "grad_norm": 1.3126804828643799,
      "learning_rate": 9.922459452694466e-06,
      "loss": 1.3505,
      "step": 1790
    },
    {
      "epoch": 0.862457111900319,
      "grad_norm": 0.9183663129806519,
      "learning_rate": 9.854825905549503e-06,
      "loss": 0.8823,
      "step": 1791
    },
    {
      "epoch": 0.8629386624932281,
      "grad_norm": 1.757525086402893,
      "learning_rate": 9.787411701115456e-06,
      "loss": 1.0742,
      "step": 1792
    },
    {
      "epoch": 0.8634202130861374,
      "grad_norm": 1.3535352945327759,
      "learning_rate": 9.720217003425647e-06,
      "loss": 0.94,
      "step": 1793
    },
    {
      "epoch": 0.8639017636790465,
      "grad_norm": 1.4137847423553467,
      "learning_rate": 9.65324197597931e-06,
      "loss": 1.0441,
      "step": 1794
    },
    {
      "epoch": 0.8643833142719557,
      "grad_norm": 0.855448842048645,
      "learning_rate": 9.58648678174121e-06,
      "loss": 0.8851,
      "step": 1795
    },
    {
      "epoch": 0.8648648648648649,
      "grad_norm": 1.6125357151031494,
      "learning_rate": 9.51995158314113e-06,
      "loss": 1.0864,
      "step": 1796
    },
    {
      "epoch": 0.865346415457774,
      "grad_norm": 1.308194637298584,
      "learning_rate": 9.45363654207363e-06,
      "loss": 1.1207,
      "step": 1797
    },
    {
      "epoch": 0.8658279660506832,
      "grad_norm": 1.1923720836639404,
      "learning_rate": 9.387541819897549e-06,
      "loss": 0.6107,
      "step": 1798
    },
    {
      "epoch": 0.8663095166435923,
      "grad_norm": 2.00758957862854,
      "learning_rate": 9.321667577435634e-06,
      "loss": 1.066,
      "step": 1799
    },
    {
      "epoch": 0.8667910672365016,
      "grad_norm": 1.4394862651824951,
      "learning_rate": 9.256013974974175e-06,
      "loss": 0.7757,
      "step": 1800
    },
    {
      "epoch": 0.8672726178294107,
      "grad_norm": 1.922194480895996,
      "learning_rate": 9.19058117226258e-06,
      "loss": 0.8847,
      "step": 1801
    },
    {
      "epoch": 0.8677541684223199,
      "grad_norm": 1.5127134323120117,
      "learning_rate": 9.125369328513034e-06,
      "loss": 1.1871,
      "step": 1802
    },
    {
      "epoch": 0.868235719015229,
      "grad_norm": 2.3378794193267822,
      "learning_rate": 9.060378602400054e-06,
      "loss": 1.3067,
      "step": 1803
    },
    {
      "epoch": 0.8687172696081382,
      "grad_norm": 0.9934513568878174,
      "learning_rate": 8.995609152060136e-06,
      "loss": 0.873,
      "step": 1804
    },
    {
      "epoch": 0.8691988202010473,
      "grad_norm": 1.0122309923171997,
      "learning_rate": 8.931061135091357e-06,
      "loss": 1.0634,
      "step": 1805
    },
    {
      "epoch": 0.8696803707939565,
      "grad_norm": 0.6063061952590942,
      "learning_rate": 8.866734708553015e-06,
      "loss": 1.2849,
      "step": 1806
    },
    {
      "epoch": 0.8701619213868657,
      "grad_norm": 1.3410141468048096,
      "learning_rate": 8.802630028965242e-06,
      "loss": 0.8513,
      "step": 1807
    },
    {
      "epoch": 0.8706434719797749,
      "grad_norm": 1.8235042095184326,
      "learning_rate": 8.738747252308555e-06,
      "loss": 0.8182,
      "step": 1808
    },
    {
      "epoch": 0.8711250225726841,
      "grad_norm": 1.1212221384048462,
      "learning_rate": 8.675086534023591e-06,
      "loss": 0.584,
      "step": 1809
    },
    {
      "epoch": 0.8716065731655932,
      "grad_norm": 1.4920377731323242,
      "learning_rate": 8.611648029010643e-06,
      "loss": 0.4761,
      "step": 1810
    },
    {
      "epoch": 0.8720881237585024,
      "grad_norm": 1.820839285850525,
      "learning_rate": 8.548431891629316e-06,
      "loss": 1.0748,
      "step": 1811
    },
    {
      "epoch": 0.8725696743514115,
      "grad_norm": 2.4675819873809814,
      "learning_rate": 8.485438275698154e-06,
      "loss": 0.9195,
      "step": 1812
    },
    {
      "epoch": 0.8730512249443207,
      "grad_norm": 2.1817128658294678,
      "learning_rate": 8.422667334494249e-06,
      "loss": 1.5723,
      "step": 1813
    },
    {
      "epoch": 0.8735327755372299,
      "grad_norm": 2.472827196121216,
      "learning_rate": 8.360119220752893e-06,
      "loss": 1.1663,
      "step": 1814
    },
    {
      "epoch": 0.8740143261301391,
      "grad_norm": 0.7892443537712097,
      "learning_rate": 8.297794086667165e-06,
      "loss": 1.2433,
      "step": 1815
    },
    {
      "epoch": 0.8744958767230482,
      "grad_norm": 1.1648625135421753,
      "learning_rate": 8.235692083887613e-06,
      "loss": 1.0757,
      "step": 1816
    },
    {
      "epoch": 0.8749774273159574,
      "grad_norm": 0.8333413004875183,
      "learning_rate": 8.173813363521843e-06,
      "loss": 1.3362,
      "step": 1817
    },
    {
      "epoch": 0.8754589779088665,
      "grad_norm": 1.970495581626892,
      "learning_rate": 8.112158076134157e-06,
      "loss": 1.2067,
      "step": 1818
    },
    {
      "epoch": 0.8759405285017757,
      "grad_norm": 0.8945471048355103,
      "learning_rate": 8.05072637174522e-06,
      "loss": 0.9888,
      "step": 1819
    },
    {
      "epoch": 0.8764220790946848,
      "grad_norm": 1.283353567123413,
      "learning_rate": 7.989518399831641e-06,
      "loss": 1.145,
      "step": 1820
    },
    {
      "epoch": 0.876903629687594,
      "grad_norm": 0.7924458980560303,
      "learning_rate": 7.928534309325675e-06,
      "loss": 0.806,
      "step": 1821
    },
    {
      "epoch": 0.8773851802805033,
      "grad_norm": 0.9375920295715332,
      "learning_rate": 7.8677742486148e-06,
      "loss": 0.6394,
      "step": 1822
    },
    {
      "epoch": 0.8778667308734124,
      "grad_norm": 1.1882402896881104,
      "learning_rate": 7.807238365541391e-06,
      "loss": 0.6891,
      "step": 1823
    },
    {
      "epoch": 0.8783482814663216,
      "grad_norm": 1.498372197151184,
      "learning_rate": 7.746926807402344e-06,
      "loss": 1.0293,
      "step": 1824
    },
    {
      "epoch": 0.8788298320592307,
      "grad_norm": 1.8423895835876465,
      "learning_rate": 7.686839720948736e-06,
      "loss": 1.9301,
      "step": 1825
    },
    {
      "epoch": 0.8793113826521399,
      "grad_norm": 1.4550963640213013,
      "learning_rate": 7.6269772523854365e-06,
      "loss": 0.8888,
      "step": 1826
    },
    {
      "epoch": 0.879792933245049,
      "grad_norm": 1.3613805770874023,
      "learning_rate": 7.567339547370789e-06,
      "loss": 0.8487,
      "step": 1827
    },
    {
      "epoch": 0.8802744838379583,
      "grad_norm": 0.8463455438613892,
      "learning_rate": 7.507926751016248e-06,
      "loss": 1.0509,
      "step": 1828
    },
    {
      "epoch": 0.8807560344308674,
      "grad_norm": 1.0641120672225952,
      "learning_rate": 7.4487390078859855e-06,
      "loss": 1.0464,
      "step": 1829
    },
    {
      "epoch": 0.8812375850237766,
      "grad_norm": 1.2556158304214478,
      "learning_rate": 7.389776461996578e-06,
      "loss": 0.9199,
      "step": 1830
    },
    {
      "epoch": 0.8817191356166857,
      "grad_norm": 1.6580806970596313,
      "learning_rate": 7.331039256816663e-06,
      "loss": 1.4582,
      "step": 1831
    },
    {
      "epoch": 0.8822006862095949,
      "grad_norm": 1.3408979177474976,
      "learning_rate": 7.27252753526656e-06,
      "loss": 1.3094,
      "step": 1832
    },
    {
      "epoch": 0.882682236802504,
      "grad_norm": 1.5318692922592163,
      "learning_rate": 7.214241439717962e-06,
      "loss": 1.4137,
      "step": 1833
    },
    {
      "epoch": 0.8831637873954132,
      "grad_norm": 1.2791460752487183,
      "learning_rate": 7.1561811119935425e-06,
      "loss": 0.6995,
      "step": 1834
    },
    {
      "epoch": 0.8836453379883223,
      "grad_norm": 1.1193867921829224,
      "learning_rate": 7.098346693366642e-06,
      "loss": 0.9256,
      "step": 1835
    },
    {
      "epoch": 0.8841268885812316,
      "grad_norm": 1.5691089630126953,
      "learning_rate": 7.0407383245609136e-06,
      "loss": 0.612,
      "step": 1836
    },
    {
      "epoch": 0.8846084391741408,
      "grad_norm": 0.8311463594436646,
      "learning_rate": 6.983356145749975e-06,
      "loss": 0.7028,
      "step": 1837
    },
    {
      "epoch": 0.8850899897670499,
      "grad_norm": 1.351941466331482,
      "learning_rate": 6.9262002965570835e-06,
      "loss": 0.9814,
      "step": 1838
    },
    {
      "epoch": 0.8855715403599591,
      "grad_norm": 1.1044578552246094,
      "learning_rate": 6.869270916054782e-06,
      "loss": 0.9075,
      "step": 1839
    },
    {
      "epoch": 0.8860530909528682,
      "grad_norm": 1.295556664466858,
      "learning_rate": 6.812568142764575e-06,
      "loss": 0.6876,
      "step": 1840
    },
    {
      "epoch": 0.8865346415457774,
      "grad_norm": 0.9764739274978638,
      "learning_rate": 6.756092114656587e-06,
      "loss": 1.1875,
      "step": 1841
    },
    {
      "epoch": 0.8870161921386865,
      "grad_norm": 1.551121711730957,
      "learning_rate": 6.699842969149195e-06,
      "loss": 1.0332,
      "step": 1842
    },
    {
      "epoch": 0.8874977427315958,
      "grad_norm": 1.2547286748886108,
      "learning_rate": 6.64382084310875e-06,
      "loss": 0.9953,
      "step": 1843
    },
    {
      "epoch": 0.8879792933245049,
      "grad_norm": 1.2803791761398315,
      "learning_rate": 6.5880258728491905e-06,
      "loss": 0.6631,
      "step": 1844
    },
    {
      "epoch": 0.8884608439174141,
      "grad_norm": 1.3346611261367798,
      "learning_rate": 6.532458194131763e-06,
      "loss": 1.2052,
      "step": 1845
    },
    {
      "epoch": 0.8889423945103232,
      "grad_norm": 2.678377866744995,
      "learning_rate": 6.477117942164657e-06,
      "loss": 1.261,
      "step": 1846
    },
    {
      "epoch": 0.8894239451032324,
      "grad_norm": 0.9634042382240295,
      "learning_rate": 6.422005251602658e-06,
      "loss": 0.6186,
      "step": 1847
    },
    {
      "epoch": 0.8899054956961415,
      "grad_norm": 0.8209473490715027,
      "learning_rate": 6.367120256546888e-06,
      "loss": 0.661,
      "step": 1848
    },
    {
      "epoch": 0.8903870462890507,
      "grad_norm": 1.1738144159317017,
      "learning_rate": 6.312463090544396e-06,
      "loss": 0.678,
      "step": 1849
    },
    {
      "epoch": 0.89086859688196,
      "grad_norm": 1.1035970449447632,
      "learning_rate": 6.258033886587911e-06,
      "loss": 1.3459,
      "step": 1850
    },
    {
      "epoch": 0.8913501474748691,
      "grad_norm": 1.742279052734375,
      "learning_rate": 6.2038327771154485e-06,
      "loss": 0.9814,
      "step": 1851
    },
    {
      "epoch": 0.8918316980677783,
      "grad_norm": 1.6004955768585205,
      "learning_rate": 6.1498598940100346e-06,
      "loss": 0.9128,
      "step": 1852
    },
    {
      "epoch": 0.8923132486606874,
      "grad_norm": 0.7587423324584961,
      "learning_rate": 6.0961153685993646e-06,
      "loss": 0.671,
      "step": 1853
    },
    {
      "epoch": 0.8927947992535966,
      "grad_norm": 2.1707489490509033,
      "learning_rate": 6.0425993316554965e-06,
      "loss": 0.921,
      "step": 1854
    },
    {
      "epoch": 0.8932763498465057,
      "grad_norm": 1.363389015197754,
      "learning_rate": 5.989311913394546e-06,
      "loss": 0.734,
      "step": 1855
    },
    {
      "epoch": 0.8937579004394149,
      "grad_norm": 1.0534840822219849,
      "learning_rate": 5.93625324347632e-06,
      "loss": 0.9686,
      "step": 1856
    },
    {
      "epoch": 0.894239451032324,
      "grad_norm": 1.3182815313339233,
      "learning_rate": 5.8834234510040335e-06,
      "loss": 1.3721,
      "step": 1857
    },
    {
      "epoch": 0.8947210016252333,
      "grad_norm": 1.9481348991394043,
      "learning_rate": 5.830822664523994e-06,
      "loss": 0.4303,
      "step": 1858
    },
    {
      "epoch": 0.8952025522181424,
      "grad_norm": 1.6406868696212769,
      "learning_rate": 5.77845101202531e-06,
      "loss": 1.2238,
      "step": 1859
    },
    {
      "epoch": 0.8956841028110516,
      "grad_norm": 1.5610930919647217,
      "learning_rate": 5.726308620939536e-06,
      "loss": 0.9428,
      "step": 1860
    },
    {
      "epoch": 0.8961656534039607,
      "grad_norm": 1.9844375848770142,
      "learning_rate": 5.674395618140393e-06,
      "loss": 0.7494,
      "step": 1861
    },
    {
      "epoch": 0.8966472039968699,
      "grad_norm": 1.8582005500793457,
      "learning_rate": 5.622712129943453e-06,
      "loss": 1.1963,
      "step": 1862
    },
    {
      "epoch": 0.8971287545897791,
      "grad_norm": 1.3388550281524658,
      "learning_rate": 5.571258282105829e-06,
      "loss": 1.041,
      "step": 1863
    },
    {
      "epoch": 0.8976103051826883,
      "grad_norm": 1.5679712295532227,
      "learning_rate": 5.520034199825841e-06,
      "loss": 1.2233,
      "step": 1864
    },
    {
      "epoch": 0.8980918557755975,
      "grad_norm": 1.4661005735397339,
      "learning_rate": 5.469040007742776e-06,
      "loss": 1.0228,
      "step": 1865
    },
    {
      "epoch": 0.8985734063685066,
      "grad_norm": 1.356406569480896,
      "learning_rate": 5.418275829936537e-06,
      "loss": 1.7113,
      "step": 1866
    },
    {
      "epoch": 0.8990549569614158,
      "grad_norm": 1.2747793197631836,
      "learning_rate": 5.36774178992735e-06,
      "loss": 0.9003,
      "step": 1867
    },
    {
      "epoch": 0.8995365075543249,
      "grad_norm": 1.65596342086792,
      "learning_rate": 5.317438010675469e-06,
      "loss": 1.1362,
      "step": 1868
    },
    {
      "epoch": 0.9000180581472341,
      "grad_norm": 1.531292200088501,
      "learning_rate": 5.267364614580861e-06,
      "loss": 0.9057,
      "step": 1869
    },
    {
      "epoch": 0.9004996087401432,
      "grad_norm": 1.6751679182052612,
      "learning_rate": 5.217521723482943e-06,
      "loss": 0.9452,
      "step": 1870
    },
    {
      "epoch": 0.9009811593330525,
      "grad_norm": 1.1299400329589844,
      "learning_rate": 5.167909458660258e-06,
      "loss": 1.4416,
      "step": 1871
    },
    {
      "epoch": 0.9014627099259616,
      "grad_norm": 1.8792394399642944,
      "learning_rate": 5.118527940830165e-06,
      "loss": 1.2021,
      "step": 1872
    },
    {
      "epoch": 0.9019442605188708,
      "grad_norm": 1.3564519882202148,
      "learning_rate": 5.069377290148602e-06,
      "loss": 0.7179,
      "step": 1873
    },
    {
      "epoch": 0.9024258111117799,
      "grad_norm": 1.6830651760101318,
      "learning_rate": 5.020457626209707e-06,
      "loss": 0.9145,
      "step": 1874
    },
    {
      "epoch": 0.9029073617046891,
      "grad_norm": 0.93392014503479,
      "learning_rate": 4.971769068045628e-06,
      "loss": 1.0233,
      "step": 1875
    },
    {
      "epoch": 0.9033889122975982,
      "grad_norm": 1.0822404623031616,
      "learning_rate": 4.923311734126135e-06,
      "loss": 0.8249,
      "step": 1876
    },
    {
      "epoch": 0.9038704628905074,
      "grad_norm": 1.0630532503128052,
      "learning_rate": 4.875085742358432e-06,
      "loss": 0.8261,
      "step": 1877
    },
    {
      "epoch": 0.9043520134834167,
      "grad_norm": 1.0887762308120728,
      "learning_rate": 4.827091210086776e-06,
      "loss": 0.9689,
      "step": 1878
    },
    {
      "epoch": 0.9048335640763258,
      "grad_norm": 1.1887446641921997,
      "learning_rate": 4.779328254092252e-06,
      "loss": 1.285,
      "step": 1879
    },
    {
      "epoch": 0.905315114669235,
      "grad_norm": 1.5426466464996338,
      "learning_rate": 4.731796990592452e-06,
      "loss": 0.8858,
      "step": 1880
    },
    {
      "epoch": 0.9057966652621441,
      "grad_norm": 1.3883047103881836,
      "learning_rate": 4.68449753524125e-06,
      "loss": 1.3978,
      "step": 1881
    },
    {
      "epoch": 0.9062782158550533,
      "grad_norm": 1.1940714120864868,
      "learning_rate": 4.637430003128429e-06,
      "loss": 1.0143,
      "step": 1882
    },
    {
      "epoch": 0.9067597664479624,
      "grad_norm": 2.22959041595459,
      "learning_rate": 4.5905945087794996e-06,
      "loss": 1.3727,
      "step": 1883
    },
    {
      "epoch": 0.9072413170408716,
      "grad_norm": 0.8964865803718567,
      "learning_rate": 4.543991166155337e-06,
      "loss": 1.1727,
      "step": 1884
    },
    {
      "epoch": 0.9077228676337807,
      "grad_norm": 1.5005121231079102,
      "learning_rate": 4.497620088651966e-06,
      "loss": 1.0019,
      "step": 1885
    },
    {
      "epoch": 0.90820441822669,
      "grad_norm": 2.345700740814209,
      "learning_rate": 4.451481389100232e-06,
      "loss": 1.2451,
      "step": 1886
    },
    {
      "epoch": 0.9086859688195991,
      "grad_norm": 1.666134238243103,
      "learning_rate": 4.405575179765586e-06,
      "loss": 1.1797,
      "step": 1887
    },
    {
      "epoch": 0.9091675194125083,
      "grad_norm": 0.8888693451881409,
      "learning_rate": 4.359901572347758e-06,
      "loss": 1.3268,
      "step": 1888
    },
    {
      "epoch": 0.9096490700054174,
      "grad_norm": 1.7369005680084229,
      "learning_rate": 4.314460677980537e-06,
      "loss": 0.8564,
      "step": 1889
    },
    {
      "epoch": 0.9101306205983266,
      "grad_norm": 1.2043582201004028,
      "learning_rate": 4.269252607231422e-06,
      "loss": 0.7559,
      "step": 1890
    },
    {
      "epoch": 0.9106121711912358,
      "grad_norm": 1.3113512992858887,
      "learning_rate": 4.224277470101445e-06,
      "loss": 0.9213,
      "step": 1891
    },
    {
      "epoch": 0.9110937217841449,
      "grad_norm": 1.3803439140319824,
      "learning_rate": 4.179535376024857e-06,
      "loss": 1.2073,
      "step": 1892
    },
    {
      "epoch": 0.9115752723770542,
      "grad_norm": 1.2347534894943237,
      "learning_rate": 4.135026433868827e-06,
      "loss": 1.0552,
      "step": 1893
    },
    {
      "epoch": 0.9120568229699633,
      "grad_norm": 0.8483551144599915,
      "learning_rate": 4.090750751933248e-06,
      "loss": 1.2542,
      "step": 1894
    },
    {
      "epoch": 0.9125383735628725,
      "grad_norm": 1.440475344657898,
      "learning_rate": 4.046708437950464e-06,
      "loss": 1.0222,
      "step": 1895
    },
    {
      "epoch": 0.9130199241557816,
      "grad_norm": 1.2423763275146484,
      "learning_rate": 4.0028995990849084e-06,
      "loss": 0.761,
      "step": 1896
    },
    {
      "epoch": 0.9135014747486908,
      "grad_norm": 1.6167302131652832,
      "learning_rate": 3.95932434193299e-06,
      "loss": 0.8379,
      "step": 1897
    },
    {
      "epoch": 0.9139830253415999,
      "grad_norm": 1.9288986921310425,
      "learning_rate": 3.915982772522719e-06,
      "loss": 1.2538,
      "step": 1898
    },
    {
      "epoch": 0.9144645759345091,
      "grad_norm": 1.0971380472183228,
      "learning_rate": 3.872874996313513e-06,
      "loss": 1.0831,
      "step": 1899
    },
    {
      "epoch": 0.9149461265274182,
      "grad_norm": 0.8678745031356812,
      "learning_rate": 3.830001118195936e-06,
      "loss": 0.9509,
      "step": 1900
    },
    {
      "epoch": 0.9154276771203275,
      "grad_norm": 1.3216204643249512,
      "learning_rate": 3.787361242491394e-06,
      "loss": 0.6416,
      "step": 1901
    },
    {
      "epoch": 0.9159092277132366,
      "grad_norm": 1.0907399654388428,
      "learning_rate": 3.744955472951928e-06,
      "loss": 0.7098,
      "step": 1902
    },
    {
      "epoch": 0.9163907783061458,
      "grad_norm": 1.445749044418335,
      "learning_rate": 3.702783912759955e-06,
      "loss": 0.7986,
      "step": 1903
    },
    {
      "epoch": 0.916872328899055,
      "grad_norm": 1.0240713357925415,
      "learning_rate": 3.660846664528006e-06,
      "loss": 0.856,
      "step": 1904
    },
    {
      "epoch": 0.9173538794919641,
      "grad_norm": 1.3976671695709229,
      "learning_rate": 3.6191438302984772e-06,
      "loss": 1.1663,
      "step": 1905
    },
    {
      "epoch": 0.9178354300848733,
      "grad_norm": 1.3567636013031006,
      "learning_rate": 3.577675511543388e-06,
      "loss": 1.1491,
      "step": 1906
    },
    {
      "epoch": 0.9183169806777824,
      "grad_norm": 0.8432836532592773,
      "learning_rate": 3.5364418091641373e-06,
      "loss": 1.1473,
      "step": 1907
    },
    {
      "epoch": 0.9187985312706917,
      "grad_norm": 1.0946440696716309,
      "learning_rate": 3.495442823491224e-06,
      "loss": 1.0727,
      "step": 1908
    },
    {
      "epoch": 0.9192800818636008,
      "grad_norm": 2.576817512512207,
      "learning_rate": 3.4546786542840605e-06,
      "loss": 1.2277,
      "step": 1909
    },
    {
      "epoch": 0.91976163245651,
      "grad_norm": 1.1032428741455078,
      "learning_rate": 3.4141494007306816e-06,
      "loss": 0.9587,
      "step": 1910
    },
    {
      "epoch": 0.9202431830494191,
      "grad_norm": 1.4798821210861206,
      "learning_rate": 3.373855161447548e-06,
      "loss": 1.1395,
      "step": 1911
    },
    {
      "epoch": 0.9207247336423283,
      "grad_norm": 1.403666615486145,
      "learning_rate": 3.333796034479242e-06,
      "loss": 0.8923,
      "step": 1912
    },
    {
      "epoch": 0.9212062842352374,
      "grad_norm": 1.405367374420166,
      "learning_rate": 3.293972117298294e-06,
      "loss": 0.744,
      "step": 1913
    },
    {
      "epoch": 0.9216878348281466,
      "grad_norm": 1.716288447380066,
      "learning_rate": 3.2543835068049255e-06,
      "loss": 0.8059,
      "step": 1914
    },
    {
      "epoch": 0.9221693854210558,
      "grad_norm": 1.2227338552474976,
      "learning_rate": 3.21503029932676e-06,
      "loss": 0.8197,
      "step": 1915
    },
    {
      "epoch": 0.922650936013965,
      "grad_norm": 1.0366851091384888,
      "learning_rate": 3.1759125906186793e-06,
      "loss": 0.8565,
      "step": 1916
    },
    {
      "epoch": 0.9231324866068741,
      "grad_norm": 1.973418951034546,
      "learning_rate": 3.137030475862535e-06,
      "loss": 1.6489,
      "step": 1917
    },
    {
      "epoch": 0.9236140371997833,
      "grad_norm": 1.4878705739974976,
      "learning_rate": 3.098384049666925e-06,
      "loss": 0.907,
      "step": 1918
    },
    {
      "epoch": 0.9240955877926925,
      "grad_norm": 2.2031521797180176,
      "learning_rate": 3.059973406066963e-06,
      "loss": 1.1393,
      "step": 1919
    },
    {
      "epoch": 0.9245771383856016,
      "grad_norm": 1.5338754653930664,
      "learning_rate": 3.0217986385240537e-06,
      "loss": 0.7839,
      "step": 1920
    },
    {
      "epoch": 0.9250586889785108,
      "grad_norm": 1.1197309494018555,
      "learning_rate": 2.983859839925662e-06,
      "loss": 0.9048,
      "step": 1921
    },
    {
      "epoch": 0.92554023957142,
      "grad_norm": 1.0619890689849854,
      "learning_rate": 2.94615710258509e-06,
      "loss": 1.2019,
      "step": 1922
    },
    {
      "epoch": 0.9260217901643292,
      "grad_norm": 0.9183767437934875,
      "learning_rate": 2.908690518241275e-06,
      "loss": 1.0725,
      "step": 1923
    },
    {
      "epoch": 0.9265033407572383,
      "grad_norm": 1.091325283050537,
      "learning_rate": 2.8714601780584937e-06,
      "loss": 0.7111,
      "step": 1924
    },
    {
      "epoch": 0.9269848913501475,
      "grad_norm": 0.835948646068573,
      "learning_rate": 2.834466172626238e-06,
      "loss": 0.5663,
      "step": 1925
    },
    {
      "epoch": 0.9274664419430566,
      "grad_norm": 1.3675706386566162,
      "learning_rate": 2.7977085919589254e-06,
      "loss": 0.96,
      "step": 1926
    },
    {
      "epoch": 0.9279479925359658,
      "grad_norm": 3.7371230125427246,
      "learning_rate": 2.76118752549569e-06,
      "loss": 0.8921,
      "step": 1927
    },
    {
      "epoch": 0.9284295431288749,
      "grad_norm": 0.9313801527023315,
      "learning_rate": 2.7249030621001924e-06,
      "loss": 1.183,
      "step": 1928
    },
    {
      "epoch": 0.9289110937217842,
      "grad_norm": 1.3623645305633545,
      "learning_rate": 2.688855290060399e-06,
      "loss": 0.9608,
      "step": 1929
    },
    {
      "epoch": 0.9293926443146933,
      "grad_norm": 1.3960970640182495,
      "learning_rate": 2.653044297088314e-06,
      "loss": 1.4327,
      "step": 1930
    },
    {
      "epoch": 0.9298741949076025,
      "grad_norm": 1.4595905542373657,
      "learning_rate": 2.6174701703198468e-06,
      "loss": 1.0167,
      "step": 1931
    },
    {
      "epoch": 0.9303557455005117,
      "grad_norm": 1.7392947673797607,
      "learning_rate": 2.5821329963145347e-06,
      "loss": 0.9313,
      "step": 1932
    },
    {
      "epoch": 0.9308372960934208,
      "grad_norm": 1.1131982803344727,
      "learning_rate": 2.547032861055376e-06,
      "loss": 1.202,
      "step": 1933
    },
    {
      "epoch": 0.93131884668633,
      "grad_norm": 0.8748666048049927,
      "learning_rate": 2.5121698499485757e-06,
      "loss": 1.0469,
      "step": 1934
    },
    {
      "epoch": 0.9318003972792391,
      "grad_norm": 1.2871897220611572,
      "learning_rate": 2.4775440478233993e-06,
      "loss": 1.0139,
      "step": 1935
    },
    {
      "epoch": 0.9322819478721484,
      "grad_norm": 1.4765926599502563,
      "learning_rate": 2.4431555389319074e-06,
      "loss": 1.3705,
      "step": 1936
    },
    {
      "epoch": 0.9327634984650575,
      "grad_norm": 1.10368812084198,
      "learning_rate": 2.4090044069487784e-06,
      "loss": 1.3003,
      "step": 1937
    },
    {
      "epoch": 0.9332450490579667,
      "grad_norm": 2.3013839721679688,
      "learning_rate": 2.3750907349711084e-06,
      "loss": 1.1265,
      "step": 1938
    },
    {
      "epoch": 0.9337265996508758,
      "grad_norm": 1.317530870437622,
      "learning_rate": 2.3414146055182106e-06,
      "loss": 1.4314,
      "step": 1939
    },
    {
      "epoch": 0.934208150243785,
      "grad_norm": 1.1160175800323486,
      "learning_rate": 2.307976100531384e-06,
      "loss": 0.9804,
      "step": 1940
    },
    {
      "epoch": 0.9346897008366941,
      "grad_norm": 1.495829701423645,
      "learning_rate": 2.274775301373744e-06,
      "loss": 1.0281,
      "step": 1941
    },
    {
      "epoch": 0.9351712514296033,
      "grad_norm": 1.4368069171905518,
      "learning_rate": 2.241812288830003e-06,
      "loss": 0.8521,
      "step": 1942
    },
    {
      "epoch": 0.9356528020225124,
      "grad_norm": 1.4072201251983643,
      "learning_rate": 2.2090871431063253e-06,
      "loss": 0.6815,
      "step": 1943
    },
    {
      "epoch": 0.9361343526154217,
      "grad_norm": 1.6371530294418335,
      "learning_rate": 2.176599943830071e-06,
      "loss": 1.0713,
      "step": 1944
    },
    {
      "epoch": 0.9366159032083309,
      "grad_norm": 1.2672380208969116,
      "learning_rate": 2.144350770049597e-06,
      "loss": 0.7537,
      "step": 1945
    },
    {
      "epoch": 0.93709745380124,
      "grad_norm": 1.6637040376663208,
      "learning_rate": 2.112339700234156e-06,
      "loss": 1.1629,
      "step": 1946
    },
    {
      "epoch": 0.9375790043941492,
      "grad_norm": 1.7890605926513672,
      "learning_rate": 2.0805668122735767e-06,
      "loss": 1.1404,
      "step": 1947
    },
    {
      "epoch": 0.9380605549870583,
      "grad_norm": 1.7288395166397095,
      "learning_rate": 2.0490321834781833e-06,
      "loss": 0.8669,
      "step": 1948
    },
    {
      "epoch": 0.9385421055799675,
      "grad_norm": 1.0026174783706665,
      "learning_rate": 2.0177358905785537e-06,
      "loss": 0.8872,
      "step": 1949
    },
    {
      "epoch": 0.9390236561728766,
      "grad_norm": 0.8083702325820923,
      "learning_rate": 1.986678009725329e-06,
      "loss": 0.7799,
      "step": 1950
    },
    {
      "epoch": 0.9395052067657859,
      "grad_norm": 1.2848485708236694,
      "learning_rate": 1.955858616489059e-06,
      "loss": 1.1604,
      "step": 1951
    },
    {
      "epoch": 0.939986757358695,
      "grad_norm": 1.6027108430862427,
      "learning_rate": 1.9252777858599915e-06,
      "loss": 0.9935,
      "step": 1952
    },
    {
      "epoch": 0.9404683079516042,
      "grad_norm": 1.0570029020309448,
      "learning_rate": 1.8949355922479151e-06,
      "loss": 0.7828,
      "step": 1953
    },
    {
      "epoch": 0.9409498585445133,
      "grad_norm": 1.8423429727554321,
      "learning_rate": 1.8648321094819287e-06,
      "loss": 1.0436,
      "step": 1954
    },
    {
      "epoch": 0.9414314091374225,
      "grad_norm": 0.6815853714942932,
      "learning_rate": 1.8349674108103288e-06,
      "loss": 0.8578,
      "step": 1955
    },
    {
      "epoch": 0.9419129597303316,
      "grad_norm": 1.155103087425232,
      "learning_rate": 1.8053415689003872e-06,
      "loss": 1.032,
      "step": 1956
    },
    {
      "epoch": 0.9423945103232408,
      "grad_norm": 1.0122603178024292,
      "learning_rate": 1.7759546558381967e-06,
      "loss": 0.9273,
      "step": 1957
    },
    {
      "epoch": 0.94287606091615,
      "grad_norm": 1.395577073097229,
      "learning_rate": 1.7468067431284707e-06,
      "loss": 0.7053,
      "step": 1958
    },
    {
      "epoch": 0.9433576115090592,
      "grad_norm": 1.2888978719711304,
      "learning_rate": 1.7178979016943764e-06,
      "loss": 1.0323,
      "step": 1959
    },
    {
      "epoch": 0.9438391621019684,
      "grad_norm": 1.5507824420928955,
      "learning_rate": 1.6892282018773908e-06,
      "loss": 1.1558,
      "step": 1960
    },
    {
      "epoch": 0.9443207126948775,
      "grad_norm": 1.320107102394104,
      "learning_rate": 1.6607977134370789e-06,
      "loss": 1.1615,
      "step": 1961
    },
    {
      "epoch": 0.9448022632877867,
      "grad_norm": 2.0098369121551514,
      "learning_rate": 1.6326065055510043e-06,
      "loss": 1.8009,
      "step": 1962
    },
    {
      "epoch": 0.9452838138806958,
      "grad_norm": 0.6828423142433167,
      "learning_rate": 1.6046546468144407e-06,
      "loss": 0.8715,
      "step": 1963
    },
    {
      "epoch": 0.945765364473605,
      "grad_norm": 1.1351264715194702,
      "learning_rate": 1.576942205240317e-06,
      "loss": 0.7263,
      "step": 1964
    },
    {
      "epoch": 0.9462469150665141,
      "grad_norm": 1.4850733280181885,
      "learning_rate": 1.5494692482590057e-06,
      "loss": 0.8289,
      "step": 1965
    },
    {
      "epoch": 0.9467284656594234,
      "grad_norm": 1.0143333673477173,
      "learning_rate": 1.522235842718156e-06,
      "loss": 0.935,
      "step": 1966
    },
    {
      "epoch": 0.9472100162523325,
      "grad_norm": 1.2888576984405518,
      "learning_rate": 1.4952420548825285e-06,
      "loss": 1.1602,
      "step": 1967
    },
    {
      "epoch": 0.9476915668452417,
      "grad_norm": 1.4866303205490112,
      "learning_rate": 1.468487950433839e-06,
      "loss": 1.2576,
      "step": 1968
    },
    {
      "epoch": 0.9481731174381508,
      "grad_norm": 1.2636531591415405,
      "learning_rate": 1.441973594470636e-06,
      "loss": 1.0837,
      "step": 1969
    },
    {
      "epoch": 0.94865466803106,
      "grad_norm": 1.834274411201477,
      "learning_rate": 1.415699051508068e-06,
      "loss": 1.4064,
      "step": 1970
    },
    {
      "epoch": 0.9491362186239691,
      "grad_norm": 1.6916282176971436,
      "learning_rate": 1.3896643854777847e-06,
      "loss": 1.0554,
      "step": 1971
    },
    {
      "epoch": 0.9496177692168783,
      "grad_norm": 1.3553987741470337,
      "learning_rate": 1.3638696597277679e-06,
      "loss": 0.5161,
      "step": 1972
    },
    {
      "epoch": 0.9500993198097876,
      "grad_norm": 1.032878041267395,
      "learning_rate": 1.3383149370221449e-06,
      "loss": 0.7661,
      "step": 1973
    },
    {
      "epoch": 0.9505808704026967,
      "grad_norm": 1.258754014968872,
      "learning_rate": 1.313000279541121e-06,
      "loss": 1.088,
      "step": 1974
    },
    {
      "epoch": 0.9510624209956059,
      "grad_norm": 0.7557881474494934,
      "learning_rate": 1.287925748880703e-06,
      "loss": 0.7384,
      "step": 1975
    },
    {
      "epoch": 0.951543971588515,
      "grad_norm": 1.5555517673492432,
      "learning_rate": 1.2630914060526522e-06,
      "loss": 1.3905,
      "step": 1976
    },
    {
      "epoch": 0.9520255221814242,
      "grad_norm": 1.456100344657898,
      "learning_rate": 1.2384973114843101e-06,
      "loss": 0.963,
      "step": 1977
    },
    {
      "epoch": 0.9525070727743333,
      "grad_norm": 0.8599039912223816,
      "learning_rate": 1.2141435250184185e-06,
      "loss": 0.6164,
      "step": 1978
    },
    {
      "epoch": 0.9529886233672425,
      "grad_norm": 1.2738579511642456,
      "learning_rate": 1.1900301059130093e-06,
      "loss": 1.0677,
      "step": 1979
    },
    {
      "epoch": 0.9534701739601517,
      "grad_norm": 0.7908176183700562,
      "learning_rate": 1.1661571128412596e-06,
      "loss": 0.5351,
      "step": 1980
    },
    {
      "epoch": 0.9539517245530609,
      "grad_norm": 0.9536134004592896,
      "learning_rate": 1.142524603891315e-06,
      "loss": 0.9714,
      "step": 1981
    },
    {
      "epoch": 0.95443327514597,
      "grad_norm": 0.9034488201141357,
      "learning_rate": 1.1191326365661892e-06,
      "loss": 0.8813,
      "step": 1982
    },
    {
      "epoch": 0.9549148257388792,
      "grad_norm": 1.997523546218872,
      "learning_rate": 1.0959812677835968e-06,
      "loss": 0.8413,
      "step": 1983
    },
    {
      "epoch": 0.9553963763317883,
      "grad_norm": 2.741316795349121,
      "learning_rate": 1.0730705538758322e-06,
      "loss": 1.2431,
      "step": 1984
    },
    {
      "epoch": 0.9558779269246975,
      "grad_norm": 1.9072446823120117,
      "learning_rate": 1.0504005505896141e-06,
      "loss": 0.8411,
      "step": 1985
    },
    {
      "epoch": 0.9563594775176067,
      "grad_norm": 1.5593496561050415,
      "learning_rate": 1.0279713130859514e-06,
      "loss": 1.2058,
      "step": 1986
    },
    {
      "epoch": 0.9568410281105159,
      "grad_norm": 1.0455855131149292,
      "learning_rate": 1.005782895940055e-06,
      "loss": 1.4487,
      "step": 1987
    },
    {
      "epoch": 0.9573225787034251,
      "grad_norm": 1.1664166450500488,
      "learning_rate": 9.838353531411272e-07,
      "loss": 1.4458,
      "step": 1988
    },
    {
      "epoch": 0.9578041292963342,
      "grad_norm": 1.0400137901306152,
      "learning_rate": 9.62128738092294e-07,
      "loss": 1.2702,
      "step": 1989
    },
    {
      "epoch": 0.9582856798892434,
      "grad_norm": 1.628124713897705,
      "learning_rate": 9.406631036104508e-07,
      "loss": 1.0891,
      "step": 1990
    },
    {
      "epoch": 0.9587672304821525,
      "grad_norm": 1.1455107927322388,
      "learning_rate": 9.194385019261287e-07,
      "loss": 1.2041,
      "step": 1991
    },
    {
      "epoch": 0.9592487810750617,
      "grad_norm": 1.1641745567321777,
      "learning_rate": 8.984549846833612e-07,
      "loss": 1.0147,
      "step": 1992
    },
    {
      "epoch": 0.9597303316679708,
      "grad_norm": 0.9494290947914124,
      "learning_rate": 8.777126029396065e-07,
      "loss": 0.9006,
      "step": 1993
    },
    {
      "epoch": 0.9602118822608801,
      "grad_norm": 1.9469881057739258,
      "learning_rate": 8.572114071655479e-07,
      "loss": 0.809,
      "step": 1994
    },
    {
      "epoch": 0.9606934328537892,
      "grad_norm": 1.4969104528427124,
      "learning_rate": 8.369514472450379e-07,
      "loss": 0.7611,
      "step": 1995
    },
    {
      "epoch": 0.9611749834466984,
      "grad_norm": 0.9892942905426025,
      "learning_rate": 8.169327724749543e-07,
      "loss": 0.8763,
      "step": 1996
    },
    {
      "epoch": 0.9616565340396075,
      "grad_norm": 1.335258960723877,
      "learning_rate": 7.971554315650442e-07,
      "loss": 0.9522,
      "step": 1997
    },
    {
      "epoch": 0.9621380846325167,
      "grad_norm": 0.8572422862052917,
      "learning_rate": 7.776194726378583e-07,
      "loss": 1.078,
      "step": 1998
    },
    {
      "epoch": 0.9626196352254258,
      "grad_norm": 1.271378755569458,
      "learning_rate": 7.583249432286277e-07,
      "loss": 1.1247,
      "step": 1999
    },
    {
      "epoch": 0.963101185818335,
      "grad_norm": 1.356995701789856,
      "learning_rate": 7.392718902850981e-07,
      "loss": 1.0718,
      "step": 2000
    },
    {
      "epoch": 0.9635827364112443,
      "grad_norm": 1.4200941324234009,
      "learning_rate": 7.204603601674853e-07,
      "loss": 0.9008,
      "step": 2001
    },
    {
      "epoch": 0.9640642870041534,
      "grad_norm": 1.2518941164016724,
      "learning_rate": 7.018903986483083e-07,
      "loss": 1.111,
      "step": 2002
    },
    {
      "epoch": 0.9645458375970626,
      "grad_norm": 2.1957178115844727,
      "learning_rate": 6.835620509122897e-07,
      "loss": 0.9958,
      "step": 2003
    },
    {
      "epoch": 0.9650273881899717,
      "grad_norm": 1.606023907661438,
      "learning_rate": 6.65475361556267e-07,
      "loss": 0.8813,
      "step": 2004
    },
    {
      "epoch": 0.9655089387828809,
      "grad_norm": 1.77558171749115,
      "learning_rate": 6.47630374589081e-07,
      "loss": 1.2309,
      "step": 2005
    },
    {
      "epoch": 0.96599048937579,
      "grad_norm": 1.3803273439407349,
      "learning_rate": 6.300271334314434e-07,
      "loss": 0.8396,
      "step": 2006
    },
    {
      "epoch": 0.9664720399686992,
      "grad_norm": 1.3587279319763184,
      "learning_rate": 6.126656809158359e-07,
      "loss": 0.9843,
      "step": 2007
    },
    {
      "epoch": 0.9669535905616083,
      "grad_norm": 1.0519191026687622,
      "learning_rate": 5.955460592864337e-07,
      "loss": 0.9414,
      "step": 2008
    },
    {
      "epoch": 0.9674351411545176,
      "grad_norm": 1.1373958587646484,
      "learning_rate": 5.78668310198982e-07,
      "loss": 1.1012,
      "step": 2009
    },
    {
      "epoch": 0.9679166917474267,
      "grad_norm": 1.0758074522018433,
      "learning_rate": 5.620324747207084e-07,
      "loss": 0.6684,
      "step": 2010
    },
    {
      "epoch": 0.9683982423403359,
      "grad_norm": 1.418303370475769,
      "learning_rate": 5.456385933301777e-07,
      "loss": 0.7151,
      "step": 2011
    },
    {
      "epoch": 0.968879792933245,
      "grad_norm": 1.3951810598373413,
      "learning_rate": 5.294867059172592e-07,
      "loss": 0.9564,
      "step": 2012
    },
    {
      "epoch": 0.9693613435261542,
      "grad_norm": 0.8763538002967834,
      "learning_rate": 5.135768517829819e-07,
      "loss": 1.0727,
      "step": 2013
    },
    {
      "epoch": 0.9698428941190634,
      "grad_norm": 1.143296718597412,
      "learning_rate": 4.979090696394795e-07,
      "loss": 0.717,
      "step": 2014
    },
    {
      "epoch": 0.9703244447119725,
      "grad_norm": 1.6854242086410522,
      "learning_rate": 4.824833976098453e-07,
      "loss": 1.0674,
      "step": 2015
    },
    {
      "epoch": 0.9708059953048818,
      "grad_norm": 1.3107540607452393,
      "learning_rate": 4.6729987322807757e-07,
      "loss": 1.0084,
      "step": 2016
    },
    {
      "epoch": 0.9712875458977909,
      "grad_norm": 1.0913753509521484,
      "learning_rate": 4.523585334389679e-07,
      "loss": 0.9276,
      "step": 2017
    },
    {
      "epoch": 0.9717690964907001,
      "grad_norm": 1.4453269243240356,
      "learning_rate": 4.3765941459804614e-07,
      "loss": 1.1271,
      "step": 2018
    },
    {
      "epoch": 0.9722506470836092,
      "grad_norm": 1.5734392404556274,
      "learning_rate": 4.232025524714356e-07,
      "loss": 1.1291,
      "step": 2019
    },
    {
      "epoch": 0.9727321976765184,
      "grad_norm": 0.8632877469062805,
      "learning_rate": 4.0898798223582e-07,
      "loss": 1.0434,
      "step": 2020
    },
    {
      "epoch": 0.9732137482694275,
      "grad_norm": 0.6184337735176086,
      "learning_rate": 3.950157384783104e-07,
      "loss": 1.0932,
      "step": 2021
    },
    {
      "epoch": 0.9736952988623367,
      "grad_norm": 1.2346669435501099,
      "learning_rate": 3.8128585519640046e-07,
      "loss": 0.8532,
      "step": 2022
    },
    {
      "epoch": 0.9741768494552459,
      "grad_norm": 1.6233429908752441,
      "learning_rate": 3.677983657978779e-07,
      "loss": 1.2925,
      "step": 2023
    },
    {
      "epoch": 0.9746584000481551,
      "grad_norm": 1.7683333158493042,
      "learning_rate": 3.545533031007131e-07,
      "loss": 1.3652,
      "step": 2024
    },
    {
      "epoch": 0.9751399506410642,
      "grad_norm": 3.0898826122283936,
      "learning_rate": 3.415506993330153e-07,
      "loss": 0.6764,
      "step": 2025
    },
    {
      "epoch": 0.9756215012339734,
      "grad_norm": 1.5899797677993774,
      "learning_rate": 3.2879058613292105e-07,
      "loss": 0.9369,
      "step": 2026
    },
    {
      "epoch": 0.9761030518268826,
      "grad_norm": 1.158605933189392,
      "learning_rate": 3.1627299454856095e-07,
      "loss": 0.8562,
      "step": 2027
    },
    {
      "epoch": 0.9765846024197917,
      "grad_norm": 0.9602823853492737,
      "learning_rate": 3.0399795503793793e-07,
      "loss": 1.5028,
      "step": 2028
    },
    {
      "epoch": 0.9770661530127009,
      "grad_norm": 1.6838467121124268,
      "learning_rate": 2.9196549746888235e-07,
      "loss": 1.048,
      "step": 2029
    },
    {
      "epoch": 0.97754770360561,
      "grad_norm": 1.2902735471725464,
      "learning_rate": 2.801756511189524e-07,
      "loss": 0.512,
      "step": 2030
    },
    {
      "epoch": 0.9780292541985193,
      "grad_norm": 1.4468449354171753,
      "learning_rate": 2.686284446754006e-07,
      "loss": 1.1971,
      "step": 2031
    },
    {
      "epoch": 0.9785108047914284,
      "grad_norm": 1.0377534627914429,
      "learning_rate": 2.573239062350963e-07,
      "loss": 1.2347,
      "step": 2032
    },
    {
      "epoch": 0.9789923553843376,
      "grad_norm": 3.3899059295654297,
      "learning_rate": 2.4626206330440326e-07,
      "loss": 1.2533,
      "step": 2033
    },
    {
      "epoch": 0.9794739059772467,
      "grad_norm": 1.4247949123382568,
      "learning_rate": 2.3544294279918e-07,
      "loss": 1.323,
      "step": 2034
    },
    {
      "epoch": 0.9799554565701559,
      "grad_norm": 0.8033959865570068,
      "learning_rate": 2.2486657104471286e-07,
      "loss": 1.1199,
      "step": 2035
    },
    {
      "epoch": 0.980437007163065,
      "grad_norm": 2.6981232166290283,
      "learning_rate": 2.1453297377557191e-07,
      "loss": 0.9129,
      "step": 2036
    },
    {
      "epoch": 0.9809185577559743,
      "grad_norm": 1.971558690071106,
      "learning_rate": 2.044421761356552e-07,
      "loss": 1.4128,
      "step": 2037
    },
    {
      "epoch": 0.9814001083488834,
      "grad_norm": 1.693182110786438,
      "learning_rate": 1.9459420267804452e-07,
      "loss": 0.9637,
      "step": 2038
    },
    {
      "epoch": 0.9818816589417926,
      "grad_norm": 1.7164835929870605,
      "learning_rate": 1.8498907736499426e-07,
      "loss": 1.2333,
      "step": 2039
    },
    {
      "epoch": 0.9823632095347017,
      "grad_norm": 1.0046031475067139,
      "learning_rate": 1.7562682356786487e-07,
      "loss": 0.8905,
      "step": 2040
    },
    {
      "epoch": 0.9828447601276109,
      "grad_norm": 1.4502230882644653,
      "learning_rate": 1.665074640670228e-07,
      "loss": 1.3474,
      "step": 2041
    },
    {
      "epoch": 0.9833263107205201,
      "grad_norm": 1.2290542125701904,
      "learning_rate": 1.576310210518517e-07,
      "loss": 0.7918,
      "step": 2042
    },
    {
      "epoch": 0.9838078613134292,
      "grad_norm": 1.0192421674728394,
      "learning_rate": 1.489975161206636e-07,
      "loss": 0.5974,
      "step": 2043
    },
    {
      "epoch": 0.9842894119063385,
      "grad_norm": 1.1567174196243286,
      "learning_rate": 1.406069702806323e-07,
      "loss": 1.1081,
      "step": 2044
    },
    {
      "epoch": 0.9847709624992476,
      "grad_norm": 1.6893396377563477,
      "learning_rate": 1.324594039477822e-07,
      "loss": 1.4763,
      "step": 2045
    },
    {
      "epoch": 0.9852525130921568,
      "grad_norm": 1.0625635385513306,
      "learning_rate": 1.2455483694689962e-07,
      "loss": 1.2492,
      "step": 2046
    },
    {
      "epoch": 0.9857340636850659,
      "grad_norm": 1.486932396888733,
      "learning_rate": 1.1689328851151038e-07,
      "loss": 1.0017,
      "step": 2047
    },
    {
      "epoch": 0.9862156142779751,
      "grad_norm": 1.95296311378479,
      "learning_rate": 1.0947477728381339e-07,
      "loss": 1.1354,
      "step": 2048
    },
    {
      "epoch": 0.9866971648708842,
      "grad_norm": 1.9899911880493164,
      "learning_rate": 1.0229932131465836e-07,
      "loss": 1.2576,
      "step": 2049
    },
    {
      "epoch": 0.9871787154637934,
      "grad_norm": 1.1334552764892578,
      "learning_rate": 9.536693806347919e-08,
      "loss": 1.5894,
      "step": 2050
    },
    {
      "epoch": 0.9876602660567025,
      "grad_norm": 1.404974102973938,
      "learning_rate": 8.867764439826065e-08,
      "loss": 1.0897,
      "step": 2051
    },
    {
      "epoch": 0.9881418166496118,
      "grad_norm": 0.6934735178947449,
      "learning_rate": 8.223145659550513e-08,
      "loss": 0.9795,
      "step": 2052
    },
    {
      "epoch": 0.9886233672425209,
      "grad_norm": 1.419352412223816,
      "learning_rate": 7.602839034017706e-08,
      "loss": 1.1985,
      "step": 2053
    },
    {
      "epoch": 0.9891049178354301,
      "grad_norm": 1.403533935546875,
      "learning_rate": 7.006846072568074e-08,
      "loss": 0.8797,
      "step": 2054
    },
    {
      "epoch": 0.9895864684283393,
      "grad_norm": 1.2746680974960327,
      "learning_rate": 6.435168225381594e-08,
      "loss": 0.9737,
      "step": 2055
    },
    {
      "epoch": 0.9900680190212484,
      "grad_norm": 0.8976141810417175,
      "learning_rate": 5.887806883474456e-08,
      "loss": 0.5275,
      "step": 2056
    },
    {
      "epoch": 0.9905495696141576,
      "grad_norm": 1.4184216260910034,
      "learning_rate": 5.364763378694626e-08,
      "loss": 1.1672,
      "step": 2057
    },
    {
      "epoch": 0.9910311202070667,
      "grad_norm": 0.5540860891342163,
      "learning_rate": 4.8660389837207334e-08,
      "loss": 1.0871,
      "step": 2058
    },
    {
      "epoch": 0.991512670799976,
      "grad_norm": 1.2567036151885986,
      "learning_rate": 4.391634912056519e-08,
      "loss": 1.1482,
      "step": 2059
    },
    {
      "epoch": 0.9919942213928851,
      "grad_norm": 1.0820990800857544,
      "learning_rate": 3.9415523180297286e-08,
      "loss": 0.9257,
      "step": 2060
    },
    {
      "epoch": 0.9924757719857943,
      "grad_norm": 1.7405600547790527,
      "learning_rate": 3.515792296789888e-08,
      "loss": 0.9414,
      "step": 2061
    },
    {
      "epoch": 0.9929573225787034,
      "grad_norm": 1.0613324642181396,
      "learning_rate": 3.114355884301645e-08,
      "loss": 1.0658,
      "step": 2062
    },
    {
      "epoch": 0.9934388731716126,
      "grad_norm": 1.1937965154647827,
      "learning_rate": 2.7372440573469883e-08,
      "loss": 1.1805,
      "step": 2063
    },
    {
      "epoch": 0.9939204237645217,
      "grad_norm": 1.3320246934890747,
      "learning_rate": 2.384457733520806e-08,
      "loss": 0.7096,
      "step": 2064
    },
    {
      "epoch": 0.9944019743574309,
      "grad_norm": 1.6771990060806274,
      "learning_rate": 2.0559977712297785e-08,
      "loss": 0.7598,
      "step": 2065
    },
    {
      "epoch": 0.99488352495034,
      "grad_norm": 1.689074993133545,
      "learning_rate": 1.7518649696857126e-08,
      "loss": 1.1655,
      "step": 2066
    },
    {
      "epoch": 0.9953650755432493,
      "grad_norm": 1.3726707696914673,
      "learning_rate": 1.4720600689110963e-08,
      "loss": 1.1435,
      "step": 2067
    },
    {
      "epoch": 0.9958466261361584,
      "grad_norm": 1.1037760972976685,
      "learning_rate": 1.216583749731326e-08,
      "loss": 0.6674,
      "step": 2068
    },
    {
      "epoch": 0.9963281767290676,
      "grad_norm": 1.4881978034973145,
      "learning_rate": 9.854366337758159e-09,
      "loss": 1.2383,
      "step": 2069
    },
    {
      "epoch": 0.9968097273219768,
      "grad_norm": 1.2955620288848877,
      "learning_rate": 7.786192834746686e-09,
      "loss": 1.315,
      "step": 2070
    },
    {
      "epoch": 0.9972912779148859,
      "grad_norm": 1.222991704940796,
      "learning_rate": 5.961322020608951e-09,
      "loss": 0.8927,
      "step": 2071
    },
    {
      "epoch": 0.9977728285077951,
      "grad_norm": 0.9904059171676636,
      "learning_rate": 4.3797583356264275e-09,
      "loss": 0.646,
      "step": 2072
    },
    {
      "epoch": 0.9982543791007042,
      "grad_norm": 1.0938152074813843,
      "learning_rate": 3.0415056281096755e-09,
      "loss": 1.0317,
      "step": 2073
    },
    {
      "epoch": 0.9987359296936135,
      "grad_norm": 0.9345012903213501,
      "learning_rate": 1.9465671543095197e-09,
      "loss": 1.2173,
      "step": 2074
    },
    {
      "epoch": 0.9992174802865226,
      "grad_norm": 1.38668954372406,
      "learning_rate": 1.094945578439255e-09,
      "loss": 1.2566,
      "step": 2075
    },
    {
      "epoch": 0.9996990308794318,
      "grad_norm": 1.8665199279785156,
      "learning_rate": 4.866429726857469e-10,
      "loss": 1.2174,
      "step": 2076
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2143700122833252,
      "learning_rate": 1.2166081717612797e-10,
      "loss": 1.0254,
      "step": 2077
    },
    {
      "epoch": 1.0,
      "step": 2077,
      "total_flos": 1.0779033600324403e+18,
      "train_loss": 1.1537158609756668,
      "train_runtime": 6614.9689,
      "train_samples_per_second": 5.023,
      "train_steps_per_second": 0.314
    }
  ],
  "logging_steps": 1,
  "max_steps": 2077,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 2400000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.0779033600324403e+18,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}