{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 624,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004807692307692308,
      "grad_norm": 6.990724563598633,
      "learning_rate": 1.5873015873015874e-07,
      "loss": 1.0623,
      "step": 1
    },
    {
      "epoch": 0.009615384615384616,
      "grad_norm": 6.919309139251709,
      "learning_rate": 3.174603174603175e-07,
      "loss": 1.0437,
      "step": 2
    },
    {
      "epoch": 0.014423076923076924,
      "grad_norm": 6.894330978393555,
      "learning_rate": 4.7619047619047623e-07,
      "loss": 1.047,
      "step": 3
    },
    {
      "epoch": 0.019230769230769232,
      "grad_norm": 6.92576265335083,
      "learning_rate": 6.34920634920635e-07,
      "loss": 1.042,
      "step": 4
    },
    {
      "epoch": 0.02403846153846154,
      "grad_norm": 6.855586051940918,
      "learning_rate": 7.936507936507937e-07,
      "loss": 1.0384,
      "step": 5
    },
    {
      "epoch": 0.028846153846153848,
      "grad_norm": 6.824655532836914,
      "learning_rate": 9.523809523809525e-07,
      "loss": 1.0474,
      "step": 6
    },
    {
      "epoch": 0.03365384615384615,
      "grad_norm": 6.360400199890137,
      "learning_rate": 1.111111111111111e-06,
      "loss": 1.0354,
      "step": 7
    },
    {
      "epoch": 0.038461538461538464,
      "grad_norm": 6.237353324890137,
      "learning_rate": 1.26984126984127e-06,
      "loss": 1.035,
      "step": 8
    },
    {
      "epoch": 0.04326923076923077,
      "grad_norm": 4.962989330291748,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 0.986,
      "step": 9
    },
    {
      "epoch": 0.04807692307692308,
      "grad_norm": 4.90160608291626,
      "learning_rate": 1.5873015873015873e-06,
      "loss": 0.9925,
      "step": 10
    },
    {
      "epoch": 0.052884615384615384,
      "grad_norm": 4.697347640991211,
      "learning_rate": 1.746031746031746e-06,
      "loss": 0.9938,
      "step": 11
    },
    {
      "epoch": 0.057692307692307696,
      "grad_norm": 2.8192811012268066,
      "learning_rate": 1.904761904761905e-06,
      "loss": 0.9378,
      "step": 12
    },
    {
      "epoch": 0.0625,
      "grad_norm": 2.878005027770996,
      "learning_rate": 2.0634920634920634e-06,
      "loss": 0.9478,
      "step": 13
    },
    {
      "epoch": 0.0673076923076923,
      "grad_norm": 2.691462993621826,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.9575,
      "step": 14
    },
    {
      "epoch": 0.07211538461538461,
      "grad_norm": 2.599451780319214,
      "learning_rate": 2.380952380952381e-06,
      "loss": 0.9319,
      "step": 15
    },
    {
      "epoch": 0.07692307692307693,
      "grad_norm": 3.288738489151001,
      "learning_rate": 2.53968253968254e-06,
      "loss": 0.9198,
      "step": 16
    },
    {
      "epoch": 0.08173076923076923,
      "grad_norm": 3.5585649013519287,
      "learning_rate": 2.6984126984126986e-06,
      "loss": 0.8854,
      "step": 17
    },
    {
      "epoch": 0.08653846153846154,
      "grad_norm": 3.675098419189453,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 0.9028,
      "step": 18
    },
    {
      "epoch": 0.09134615384615384,
      "grad_norm": 3.505143642425537,
      "learning_rate": 3.015873015873016e-06,
      "loss": 0.9001,
      "step": 19
    },
    {
      "epoch": 0.09615384615384616,
      "grad_norm": 3.0340216159820557,
      "learning_rate": 3.1746031746031746e-06,
      "loss": 0.8766,
      "step": 20
    },
    {
      "epoch": 0.10096153846153846,
      "grad_norm": 2.239123821258545,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.8773,
      "step": 21
    },
    {
      "epoch": 0.10576923076923077,
      "grad_norm": 1.5766472816467285,
      "learning_rate": 3.492063492063492e-06,
      "loss": 0.8311,
      "step": 22
    },
    {
      "epoch": 0.11057692307692307,
      "grad_norm": 1.567060112953186,
      "learning_rate": 3.6507936507936507e-06,
      "loss": 0.8303,
      "step": 23
    },
    {
      "epoch": 0.11538461538461539,
      "grad_norm": 1.373935580253601,
      "learning_rate": 3.80952380952381e-06,
      "loss": 0.8011,
      "step": 24
    },
    {
      "epoch": 0.1201923076923077,
      "grad_norm": 1.1732724905014038,
      "learning_rate": 3.968253968253968e-06,
      "loss": 0.7878,
      "step": 25
    },
    {
      "epoch": 0.125,
      "grad_norm": 1.0626819133758545,
      "learning_rate": 4.126984126984127e-06,
      "loss": 0.78,
      "step": 26
    },
    {
      "epoch": 0.12980769230769232,
      "grad_norm": 1.0781354904174805,
      "learning_rate": 4.2857142857142855e-06,
      "loss": 0.7852,
      "step": 27
    },
    {
      "epoch": 0.1346153846153846,
      "grad_norm": 1.006543517112732,
      "learning_rate": 4.444444444444444e-06,
      "loss": 0.7954,
      "step": 28
    },
    {
      "epoch": 0.13942307692307693,
      "grad_norm": 0.9152565598487854,
      "learning_rate": 4.603174603174604e-06,
      "loss": 0.7573,
      "step": 29
    },
    {
      "epoch": 0.14423076923076922,
      "grad_norm": 0.8509926199913025,
      "learning_rate": 4.761904761904762e-06,
      "loss": 0.7572,
      "step": 30
    },
    {
      "epoch": 0.14903846153846154,
      "grad_norm": 0.8057727217674255,
      "learning_rate": 4.920634920634921e-06,
      "loss": 0.7536,
      "step": 31
    },
    {
      "epoch": 0.15384615384615385,
      "grad_norm": 0.8267956376075745,
      "learning_rate": 5.07936507936508e-06,
      "loss": 0.7441,
      "step": 32
    },
    {
      "epoch": 0.15865384615384615,
      "grad_norm": 0.8982202410697937,
      "learning_rate": 5.2380952380952384e-06,
      "loss": 0.7398,
      "step": 33
    },
    {
      "epoch": 0.16346153846153846,
      "grad_norm": 0.7668877840042114,
      "learning_rate": 5.396825396825397e-06,
      "loss": 0.7356,
      "step": 34
    },
    {
      "epoch": 0.16826923076923078,
      "grad_norm": 0.775622546672821,
      "learning_rate": 5.555555555555557e-06,
      "loss": 0.722,
      "step": 35
    },
    {
      "epoch": 0.17307692307692307,
      "grad_norm": 0.7095115780830383,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 0.7061,
      "step": 36
    },
    {
      "epoch": 0.1778846153846154,
      "grad_norm": 0.7553266882896423,
      "learning_rate": 5.873015873015874e-06,
      "loss": 0.7194,
      "step": 37
    },
    {
      "epoch": 0.18269230769230768,
      "grad_norm": 0.7487539052963257,
      "learning_rate": 6.031746031746032e-06,
      "loss": 0.7101,
      "step": 38
    },
    {
      "epoch": 0.1875,
      "grad_norm": 0.7720729112625122,
      "learning_rate": 6.1904761904761914e-06,
      "loss": 0.7186,
      "step": 39
    },
    {
      "epoch": 0.19230769230769232,
      "grad_norm": 0.7665765881538391,
      "learning_rate": 6.349206349206349e-06,
      "loss": 0.7262,
      "step": 40
    },
    {
      "epoch": 0.1971153846153846,
      "grad_norm": 0.8309563398361206,
      "learning_rate": 6.507936507936509e-06,
      "loss": 0.7232,
      "step": 41
    },
    {
      "epoch": 0.20192307692307693,
      "grad_norm": 0.8082911372184753,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.7006,
      "step": 42
    },
    {
      "epoch": 0.20673076923076922,
      "grad_norm": 0.6936203837394714,
      "learning_rate": 6.825396825396826e-06,
      "loss": 0.7036,
      "step": 43
    },
    {
      "epoch": 0.21153846153846154,
      "grad_norm": 0.8248268365859985,
      "learning_rate": 6.984126984126984e-06,
      "loss": 0.7022,
      "step": 44
    },
    {
      "epoch": 0.21634615384615385,
      "grad_norm": 0.7197801470756531,
      "learning_rate": 7.1428571428571436e-06,
      "loss": 0.6632,
      "step": 45
    },
    {
      "epoch": 0.22115384615384615,
      "grad_norm": 0.5920101404190063,
      "learning_rate": 7.301587301587301e-06,
      "loss": 0.682,
      "step": 46
    },
    {
      "epoch": 0.22596153846153846,
      "grad_norm": 0.6113859415054321,
      "learning_rate": 7.460317460317461e-06,
      "loss": 0.6929,
      "step": 47
    },
    {
      "epoch": 0.23076923076923078,
      "grad_norm": 0.6709004640579224,
      "learning_rate": 7.61904761904762e-06,
      "loss": 0.6728,
      "step": 48
    },
    {
      "epoch": 0.23557692307692307,
      "grad_norm": 0.6471027731895447,
      "learning_rate": 7.77777777777778e-06,
      "loss": 0.6722,
      "step": 49
    },
    {
      "epoch": 0.2403846153846154,
      "grad_norm": 0.5812222361564636,
      "learning_rate": 7.936507936507936e-06,
      "loss": 0.6787,
      "step": 50
    },
    {
      "epoch": 0.24519230769230768,
      "grad_norm": 0.7254865765571594,
      "learning_rate": 8.095238095238097e-06,
      "loss": 0.6809,
      "step": 51
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5452060103416443,
      "learning_rate": 8.253968253968254e-06,
      "loss": 0.6658,
      "step": 52
    },
    {
      "epoch": 0.2548076923076923,
      "grad_norm": 0.6729876399040222,
      "learning_rate": 8.412698412698414e-06,
      "loss": 0.6777,
      "step": 53
    },
    {
      "epoch": 0.25961538461538464,
      "grad_norm": 0.5397999286651611,
      "learning_rate": 8.571428571428571e-06,
      "loss": 0.6713,
      "step": 54
    },
    {
      "epoch": 0.2644230769230769,
      "grad_norm": 0.6134487390518188,
      "learning_rate": 8.730158730158731e-06,
      "loss": 0.6668,
      "step": 55
    },
    {
      "epoch": 0.2692307692307692,
      "grad_norm": 0.5382984280586243,
      "learning_rate": 8.888888888888888e-06,
      "loss": 0.6697,
      "step": 56
    },
    {
      "epoch": 0.27403846153846156,
      "grad_norm": 0.6409310102462769,
      "learning_rate": 9.047619047619049e-06,
      "loss": 0.6725,
      "step": 57
    },
    {
      "epoch": 0.27884615384615385,
      "grad_norm": 0.4940313398838043,
      "learning_rate": 9.206349206349207e-06,
      "loss": 0.6738,
      "step": 58
    },
    {
      "epoch": 0.28365384615384615,
      "grad_norm": 0.5982186198234558,
      "learning_rate": 9.365079365079366e-06,
      "loss": 0.6625,
      "step": 59
    },
    {
      "epoch": 0.28846153846153844,
      "grad_norm": 0.6377073526382446,
      "learning_rate": 9.523809523809525e-06,
      "loss": 0.6605,
      "step": 60
    },
    {
      "epoch": 0.2932692307692308,
      "grad_norm": 0.6730812191963196,
      "learning_rate": 9.682539682539683e-06,
      "loss": 0.6589,
      "step": 61
    },
    {
      "epoch": 0.2980769230769231,
      "grad_norm": 0.7052279710769653,
      "learning_rate": 9.841269841269842e-06,
      "loss": 0.6504,
      "step": 62
    },
    {
      "epoch": 0.30288461538461536,
      "grad_norm": 0.5276140570640564,
      "learning_rate": 1e-05,
      "loss": 0.6623,
      "step": 63
    },
    {
      "epoch": 0.3076923076923077,
      "grad_norm": 0.7044535279273987,
      "learning_rate": 9.99992160057155e-06,
      "loss": 0.6383,
      "step": 64
    },
    {
      "epoch": 0.3125,
      "grad_norm": 0.7368455529212952,
      "learning_rate": 9.999686404744782e-06,
      "loss": 0.6447,
      "step": 65
    },
    {
      "epoch": 0.3173076923076923,
      "grad_norm": 0.7093029618263245,
      "learning_rate": 9.999294419895389e-06,
      "loss": 0.6673,
      "step": 66
    },
    {
      "epoch": 0.32211538461538464,
      "grad_norm": 0.7653289437294006,
      "learning_rate": 9.998745658315924e-06,
      "loss": 0.6437,
      "step": 67
    },
    {
      "epoch": 0.3269230769230769,
      "grad_norm": 0.6811711192131042,
      "learning_rate": 9.998040137215423e-06,
      "loss": 0.656,
      "step": 68
    },
    {
      "epoch": 0.3317307692307692,
      "grad_norm": 0.6195499897003174,
      "learning_rate": 9.99717787871887e-06,
      "loss": 0.6473,
      "step": 69
    },
    {
      "epoch": 0.33653846153846156,
      "grad_norm": 0.7382338047027588,
      "learning_rate": 9.99615890986649e-06,
      "loss": 0.6484,
      "step": 70
    },
    {
      "epoch": 0.34134615384615385,
      "grad_norm": 0.7722101807594299,
      "learning_rate": 9.994983262612916e-06,
      "loss": 0.6508,
      "step": 71
    },
    {
      "epoch": 0.34615384615384615,
      "grad_norm": 0.7530032396316528,
      "learning_rate": 9.993650973826177e-06,
      "loss": 0.6346,
      "step": 72
    },
    {
      "epoch": 0.35096153846153844,
      "grad_norm": 0.7273198962211609,
      "learning_rate": 9.992162085286543e-06,
      "loss": 0.6456,
      "step": 73
    },
    {
      "epoch": 0.3557692307692308,
      "grad_norm": 0.7155181765556335,
      "learning_rate": 9.990516643685222e-06,
      "loss": 0.6553,
      "step": 74
    },
    {
      "epoch": 0.3605769230769231,
      "grad_norm": 0.7894483804702759,
      "learning_rate": 9.988714700622882e-06,
      "loss": 0.648,
      "step": 75
    },
    {
      "epoch": 0.36538461538461536,
      "grad_norm": 0.6426685452461243,
      "learning_rate": 9.986756312608048e-06,
      "loss": 0.6498,
      "step": 76
    },
    {
      "epoch": 0.3701923076923077,
      "grad_norm": 0.6765972375869751,
      "learning_rate": 9.98464154105532e-06,
      "loss": 0.6478,
      "step": 77
    },
    {
      "epoch": 0.375,
      "grad_norm": 0.6547264456748962,
      "learning_rate": 9.982370452283451e-06,
      "loss": 0.628,
      "step": 78
    },
    {
      "epoch": 0.3798076923076923,
      "grad_norm": 0.6896611452102661,
      "learning_rate": 9.979943117513265e-06,
      "loss": 0.6204,
      "step": 79
    },
    {
      "epoch": 0.38461538461538464,
      "grad_norm": 0.7491808533668518,
      "learning_rate": 9.977359612865424e-06,
      "loss": 0.6376,
      "step": 80
    },
    {
      "epoch": 0.3894230769230769,
      "grad_norm": 0.6002772450447083,
      "learning_rate": 9.974620019358046e-06,
      "loss": 0.6378,
      "step": 81
    },
    {
      "epoch": 0.3942307692307692,
      "grad_norm": 0.7602054476737976,
      "learning_rate": 9.971724422904154e-06,
      "loss": 0.643,
      "step": 82
    },
    {
      "epoch": 0.39903846153846156,
      "grad_norm": 0.5507690906524658,
      "learning_rate": 9.968672914308995e-06,
      "loss": 0.6332,
      "step": 83
    },
    {
      "epoch": 0.40384615384615385,
      "grad_norm": 0.6860342025756836,
      "learning_rate": 9.965465589267176e-06,
      "loss": 0.6489,
      "step": 84
    },
    {
      "epoch": 0.40865384615384615,
      "grad_norm": 0.5931215286254883,
      "learning_rate": 9.96210254835968e-06,
      "loss": 0.6425,
      "step": 85
    },
    {
      "epoch": 0.41346153846153844,
      "grad_norm": 0.5528920292854309,
      "learning_rate": 9.9585838970507e-06,
      "loss": 0.6361,
      "step": 86
    },
    {
      "epoch": 0.4182692307692308,
      "grad_norm": 0.6603521704673767,
      "learning_rate": 9.954909745684339e-06,
      "loss": 0.63,
      "step": 87
    },
    {
      "epoch": 0.4230769230769231,
      "grad_norm": 0.6493014693260193,
      "learning_rate": 9.951080209481138e-06,
      "loss": 0.6187,
      "step": 88
    },
    {
      "epoch": 0.42788461538461536,
      "grad_norm": 0.5607807040214539,
      "learning_rate": 9.947095408534483e-06,
      "loss": 0.6459,
      "step": 89
    },
    {
      "epoch": 0.4326923076923077,
      "grad_norm": 0.6842950582504272,
      "learning_rate": 9.94295546780682e-06,
      "loss": 0.6315,
      "step": 90
    },
    {
      "epoch": 0.4375,
      "grad_norm": 0.7010789513587952,
      "learning_rate": 9.93866051712574e-06,
      "loss": 0.6353,
      "step": 91
    },
    {
      "epoch": 0.4423076923076923,
      "grad_norm": 0.573545515537262,
      "learning_rate": 9.934210691179918e-06,
      "loss": 0.6149,
      "step": 92
    },
    {
      "epoch": 0.44711538461538464,
      "grad_norm": 0.5375520586967468,
      "learning_rate": 9.929606129514875e-06,
      "loss": 0.6227,
      "step": 93
    },
    {
      "epoch": 0.4519230769230769,
      "grad_norm": 0.6587491035461426,
      "learning_rate": 9.924846976528618e-06,
      "loss": 0.6297,
      "step": 94
    },
    {
      "epoch": 0.4567307692307692,
      "grad_norm": 0.6161583662033081,
      "learning_rate": 9.919933381467088e-06,
      "loss": 0.6421,
      "step": 95
    },
    {
      "epoch": 0.46153846153846156,
      "grad_norm": 0.5526056885719299,
      "learning_rate": 9.91486549841951e-06,
      "loss": 0.6506,
      "step": 96
    },
    {
      "epoch": 0.46634615384615385,
      "grad_norm": 0.6603134870529175,
      "learning_rate": 9.909643486313533e-06,
      "loss": 0.639,
      "step": 97
    },
    {
      "epoch": 0.47115384615384615,
      "grad_norm": 0.5784550309181213,
      "learning_rate": 9.904267508910269e-06,
      "loss": 0.6307,
      "step": 98
    },
    {
      "epoch": 0.47596153846153844,
      "grad_norm": 0.5557315945625305,
      "learning_rate": 9.898737734799134e-06,
      "loss": 0.6163,
      "step": 99
    },
    {
      "epoch": 0.4807692307692308,
      "grad_norm": 0.6521722078323364,
      "learning_rate": 9.89305433739258e-06,
      "loss": 0.6352,
      "step": 100
    },
    {
      "epoch": 0.4855769230769231,
      "grad_norm": 0.6033216118812561,
      "learning_rate": 9.887217494920655e-06,
      "loss": 0.6253,
      "step": 101
    },
    {
      "epoch": 0.49038461538461536,
      "grad_norm": 0.5049378871917725,
      "learning_rate": 9.881227390425404e-06,
      "loss": 0.6315,
      "step": 102
    },
    {
      "epoch": 0.4951923076923077,
      "grad_norm": 0.5810129046440125,
      "learning_rate": 9.875084211755127e-06,
      "loss": 0.6303,
      "step": 103
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5270701050758362,
      "learning_rate": 9.868788151558513e-06,
      "loss": 0.6275,
      "step": 104
    },
    {
      "epoch": 0.5048076923076923,
      "grad_norm": 0.5791574120521545,
      "learning_rate": 9.862339407278564e-06,
      "loss": 0.6364,
      "step": 105
    },
    {
      "epoch": 0.5096153846153846,
      "grad_norm": 0.5724871158599854,
      "learning_rate": 9.855738181146427e-06,
      "loss": 0.6222,
      "step": 106
    },
    {
      "epoch": 0.5144230769230769,
      "grad_norm": 0.5011044144630432,
      "learning_rate": 9.848984680175049e-06,
      "loss": 0.6216,
      "step": 107
    },
    {
      "epoch": 0.5192307692307693,
      "grad_norm": 0.5154832601547241,
      "learning_rate": 9.84207911615267e-06,
      "loss": 0.6224,
      "step": 108
    },
    {
      "epoch": 0.5240384615384616,
      "grad_norm": 0.5105087757110596,
      "learning_rate": 9.835021705636201e-06,
      "loss": 0.6259,
      "step": 109
    },
    {
      "epoch": 0.5288461538461539,
      "grad_norm": 0.66715407371521,
      "learning_rate": 9.827812669944423e-06,
      "loss": 0.6245,
      "step": 110
    },
    {
      "epoch": 0.5336538461538461,
      "grad_norm": 0.5680528879165649,
      "learning_rate": 9.82045223515105e-06,
      "loss": 0.6264,
      "step": 111
    },
    {
      "epoch": 0.5384615384615384,
      "grad_norm": 0.6406980156898499,
      "learning_rate": 9.812940632077629e-06,
      "loss": 0.6216,
      "step": 112
    },
    {
      "epoch": 0.5432692307692307,
      "grad_norm": 0.7648453116416931,
      "learning_rate": 9.805278096286318e-06,
      "loss": 0.6225,
      "step": 113
    },
    {
      "epoch": 0.5480769230769231,
      "grad_norm": 0.6609866619110107,
      "learning_rate": 9.797464868072489e-06,
      "loss": 0.6414,
      "step": 114
    },
    {
      "epoch": 0.5528846153846154,
      "grad_norm": 0.6097372174263,
      "learning_rate": 9.789501192457188e-06,
      "loss": 0.6301,
      "step": 115
    },
    {
      "epoch": 0.5576923076923077,
      "grad_norm": 0.6364119648933411,
      "learning_rate": 9.781387319179465e-06,
      "loss": 0.6302,
      "step": 116
    },
    {
      "epoch": 0.5625,
      "grad_norm": 0.5458793640136719,
      "learning_rate": 9.773123502688532e-06,
      "loss": 0.6119,
      "step": 117
    },
    {
      "epoch": 0.5673076923076923,
      "grad_norm": 0.7278120517730713,
      "learning_rate": 9.764710002135784e-06,
      "loss": 0.6298,
      "step": 118
    },
    {
      "epoch": 0.5721153846153846,
      "grad_norm": 0.5531219244003296,
      "learning_rate": 9.756147081366673e-06,
      "loss": 0.6174,
      "step": 119
    },
    {
      "epoch": 0.5769230769230769,
      "grad_norm": 0.6038543581962585,
      "learning_rate": 9.747435008912438e-06,
      "loss": 0.6192,
      "step": 120
    },
    {
      "epoch": 0.5817307692307693,
      "grad_norm": 0.8131069540977478,
      "learning_rate": 9.73857405798168e-06,
      "loss": 0.6198,
      "step": 121
    },
    {
      "epoch": 0.5865384615384616,
      "grad_norm": 0.5778982043266296,
      "learning_rate": 9.729564506451791e-06,
      "loss": 0.6205,
      "step": 122
    },
    {
      "epoch": 0.5913461538461539,
      "grad_norm": 0.842856764793396,
      "learning_rate": 9.720406636860252e-06,
      "loss": 0.6289,
      "step": 123
    },
    {
      "epoch": 0.5961538461538461,
      "grad_norm": 0.6863536238670349,
      "learning_rate": 9.711100736395758e-06,
      "loss": 0.6168,
      "step": 124
    },
    {
      "epoch": 0.6009615384615384,
      "grad_norm": 0.6230315566062927,
      "learning_rate": 9.70164709688922e-06,
      "loss": 0.6125,
      "step": 125
    },
    {
      "epoch": 0.6057692307692307,
      "grad_norm": 0.788438618183136,
      "learning_rate": 9.69204601480461e-06,
      "loss": 0.6296,
      "step": 126
    },
    {
      "epoch": 0.6105769230769231,
      "grad_norm": 0.6000380516052246,
      "learning_rate": 9.682297791229668e-06,
      "loss": 0.6296,
      "step": 127
    },
    {
      "epoch": 0.6153846153846154,
      "grad_norm": 0.73893141746521,
      "learning_rate": 9.67240273186646e-06,
      "loss": 0.6083,
      "step": 128
    },
    {
      "epoch": 0.6201923076923077,
      "grad_norm": 0.6741968393325806,
      "learning_rate": 9.66236114702178e-06,
      "loss": 0.6141,
      "step": 129
    },
    {
      "epoch": 0.625,
      "grad_norm": 0.7334549427032471,
      "learning_rate": 9.652173351597435e-06,
      "loss": 0.6052,
      "step": 130
    },
    {
      "epoch": 0.6298076923076923,
      "grad_norm": 0.6906297206878662,
      "learning_rate": 9.641839665080363e-06,
      "loss": 0.6196,
      "step": 131
    },
    {
      "epoch": 0.6346153846153846,
      "grad_norm": 0.7519117593765259,
      "learning_rate": 9.631360411532609e-06,
      "loss": 0.6209,
      "step": 132
    },
    {
      "epoch": 0.6394230769230769,
      "grad_norm": 0.6346924304962158,
      "learning_rate": 9.620735919581168e-06,
      "loss": 0.624,
      "step": 133
    },
    {
      "epoch": 0.6442307692307693,
      "grad_norm": 0.6513266563415527,
      "learning_rate": 9.609966522407678e-06,
      "loss": 0.6294,
      "step": 134
    },
    {
      "epoch": 0.6490384615384616,
      "grad_norm": 0.719011664390564,
      "learning_rate": 9.599052557737973e-06,
      "loss": 0.6323,
      "step": 135
    },
    {
      "epoch": 0.6538461538461539,
      "grad_norm": 0.5536561012268066,
      "learning_rate": 9.58799436783149e-06,
      "loss": 0.6081,
      "step": 136
    },
    {
      "epoch": 0.6586538461538461,
      "grad_norm": 0.6664816737174988,
      "learning_rate": 9.576792299470537e-06,
      "loss": 0.6234,
      "step": 137
    },
    {
      "epoch": 0.6634615384615384,
      "grad_norm": 0.6060488820075989,
      "learning_rate": 9.565446703949417e-06,
      "loss": 0.609,
      "step": 138
    },
    {
      "epoch": 0.6682692307692307,
      "grad_norm": 0.7735944390296936,
      "learning_rate": 9.55395793706341e-06,
      "loss": 0.609,
      "step": 139
    },
    {
      "epoch": 0.6730769230769231,
      "grad_norm": 0.5819251537322998,
      "learning_rate": 9.542326359097619e-06,
      "loss": 0.6116,
      "step": 140
    },
    {
      "epoch": 0.6778846153846154,
      "grad_norm": 0.8322558403015137,
      "learning_rate": 9.530552334815672e-06,
      "loss": 0.6071,
      "step": 141
    },
    {
      "epoch": 0.6826923076923077,
      "grad_norm": 0.744692862033844,
      "learning_rate": 9.518636233448276e-06,
      "loss": 0.5932,
      "step": 142
    },
    {
      "epoch": 0.6875,
      "grad_norm": 0.9699648022651672,
      "learning_rate": 9.506578428681648e-06,
      "loss": 0.6119,
      "step": 143
    },
    {
      "epoch": 0.6923076923076923,
      "grad_norm": 0.621521532535553,
      "learning_rate": 9.494379298645788e-06,
      "loss": 0.6121,
      "step": 144
    },
    {
      "epoch": 0.6971153846153846,
      "grad_norm": 1.0709632635116577,
      "learning_rate": 9.482039225902623e-06,
      "loss": 0.6101,
      "step": 145
    },
    {
      "epoch": 0.7019230769230769,
      "grad_norm": 0.5138935446739197,
      "learning_rate": 9.469558597434018e-06,
      "loss": 0.6283,
      "step": 146
    },
    {
      "epoch": 0.7067307692307693,
      "grad_norm": 0.8398244976997375,
      "learning_rate": 9.456937804629623e-06,
      "loss": 0.6,
      "step": 147
    },
    {
      "epoch": 0.7115384615384616,
      "grad_norm": 0.7468331456184387,
      "learning_rate": 9.444177243274619e-06,
      "loss": 0.6124,
      "step": 148
    },
    {
      "epoch": 0.7163461538461539,
      "grad_norm": 0.7130693793296814,
      "learning_rate": 9.43127731353729e-06,
      "loss": 0.6195,
      "step": 149
    },
    {
      "epoch": 0.7211538461538461,
      "grad_norm": 0.766042172908783,
      "learning_rate": 9.418238419956484e-06,
      "loss": 0.6052,
      "step": 150
    },
    {
      "epoch": 0.7259615384615384,
      "grad_norm": 0.8716906309127808,
      "learning_rate": 9.405060971428924e-06,
      "loss": 0.6209,
      "step": 151
    },
    {
      "epoch": 0.7307692307692307,
      "grad_norm": 0.6835882663726807,
      "learning_rate": 9.391745381196382e-06,
      "loss": 0.6174,
      "step": 152
    },
    {
      "epoch": 0.7355769230769231,
      "grad_norm": 0.7342336177825928,
      "learning_rate": 9.378292066832723e-06,
      "loss": 0.6243,
      "step": 153
    },
    {
      "epoch": 0.7403846153846154,
      "grad_norm": 0.6497460007667542,
      "learning_rate": 9.364701450230813e-06,
      "loss": 0.6313,
      "step": 154
    },
    {
      "epoch": 0.7451923076923077,
      "grad_norm": 0.6471742391586304,
      "learning_rate": 9.350973957589278e-06,
      "loss": 0.6124,
      "step": 155
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6364895105361938,
      "learning_rate": 9.33711001939915e-06,
      "loss": 0.614,
      "step": 156
    },
    {
      "epoch": 0.7548076923076923,
      "grad_norm": 0.6573180556297302,
      "learning_rate": 9.32311007043036e-06,
      "loss": 0.6309,
      "step": 157
    },
    {
      "epoch": 0.7596153846153846,
      "grad_norm": 0.6328399777412415,
      "learning_rate": 9.30897454971811e-06,
      "loss": 0.6101,
      "step": 158
    },
    {
      "epoch": 0.7644230769230769,
      "grad_norm": 0.6596732139587402,
      "learning_rate": 9.294703900549096e-06,
      "loss": 0.6244,
      "step": 159
    },
    {
      "epoch": 0.7692307692307693,
      "grad_norm": 0.6779056787490845,
      "learning_rate": 9.280298570447612e-06,
      "loss": 0.6095,
      "step": 160
    },
    {
      "epoch": 0.7740384615384616,
      "grad_norm": 0.6027641892433167,
      "learning_rate": 9.265759011161519e-06,
      "loss": 0.5958,
      "step": 161
    },
    {
      "epoch": 0.7788461538461539,
      "grad_norm": 0.7004779577255249,
      "learning_rate": 9.251085678648072e-06,
      "loss": 0.6221,
      "step": 162
    },
    {
      "epoch": 0.7836538461538461,
      "grad_norm": 0.7348073124885559,
      "learning_rate": 9.236279033059622e-06,
      "loss": 0.6226,
      "step": 163
    },
    {
      "epoch": 0.7884615384615384,
      "grad_norm": 0.5637420415878296,
      "learning_rate": 9.221339538729191e-06,
      "loss": 0.601,
      "step": 164
    },
    {
      "epoch": 0.7932692307692307,
      "grad_norm": 0.8568174839019775,
      "learning_rate": 9.206267664155906e-06,
      "loss": 0.6081,
      "step": 165
    },
    {
      "epoch": 0.7980769230769231,
      "grad_norm": 0.7954565286636353,
      "learning_rate": 9.191063881990308e-06,
      "loss": 0.6032,
      "step": 166
    },
    {
      "epoch": 0.8028846153846154,
      "grad_norm": 0.6611636877059937,
      "learning_rate": 9.17572866901953e-06,
      "loss": 0.6101,
      "step": 167
    },
    {
      "epoch": 0.8076923076923077,
      "grad_norm": 0.9244260787963867,
      "learning_rate": 9.160262506152343e-06,
      "loss": 0.6015,
      "step": 168
    },
    {
      "epoch": 0.8125,
      "grad_norm": 0.6074010133743286,
      "learning_rate": 9.14466587840408e-06,
      "loss": 0.6106,
      "step": 169
    },
    {
      "epoch": 0.8173076923076923,
      "grad_norm": 0.6845681071281433,
      "learning_rate": 9.12893927488142e-06,
      "loss": 0.6012,
      "step": 170
    },
    {
      "epoch": 0.8221153846153846,
      "grad_norm": 0.5826325416564941,
      "learning_rate": 9.113083188767057e-06,
      "loss": 0.6137,
      "step": 171
    },
    {
      "epoch": 0.8269230769230769,
      "grad_norm": 0.5240793228149414,
      "learning_rate": 9.097098117304223e-06,
      "loss": 0.6131,
      "step": 172
    },
    {
      "epoch": 0.8317307692307693,
      "grad_norm": 0.5843315124511719,
      "learning_rate": 9.08098456178111e-06,
      "loss": 0.6072,
      "step": 173
    },
    {
      "epoch": 0.8365384615384616,
      "grad_norm": 0.5391135215759277,
      "learning_rate": 9.064743027515127e-06,
      "loss": 0.6068,
      "step": 174
    },
    {
      "epoch": 0.8413461538461539,
      "grad_norm": 0.6380318403244019,
      "learning_rate": 9.048374023837086e-06,
      "loss": 0.6035,
      "step": 175
    },
    {
      "epoch": 0.8461538461538461,
      "grad_norm": 0.5520167350769043,
      "learning_rate": 9.03187806407519e-06,
      "loss": 0.6131,
      "step": 176
    },
    {
      "epoch": 0.8509615384615384,
      "grad_norm": 0.5716068744659424,
      "learning_rate": 9.015255665538972e-06,
      "loss": 0.6138,
      "step": 177
    },
    {
      "epoch": 0.8557692307692307,
      "grad_norm": 0.598839282989502,
      "learning_rate": 8.998507349503048e-06,
      "loss": 0.5959,
      "step": 178
    },
    {
      "epoch": 0.8605769230769231,
      "grad_norm": 0.6445143818855286,
      "learning_rate": 8.981633641190779e-06,
      "loss": 0.6076,
      "step": 179
    },
    {
      "epoch": 0.8653846153846154,
      "grad_norm": 0.6508511304855347,
      "learning_rate": 8.964635069757803e-06,
      "loss": 0.6021,
      "step": 180
    },
    {
      "epoch": 0.8701923076923077,
      "grad_norm": 0.6267576813697815,
      "learning_rate": 8.94751216827543e-06,
      "loss": 0.5993,
      "step": 181
    },
    {
      "epoch": 0.875,
      "grad_norm": 0.6657792925834656,
      "learning_rate": 8.930265473713939e-06,
      "loss": 0.601,
      "step": 182
    },
    {
      "epoch": 0.8798076923076923,
      "grad_norm": 0.6445341110229492,
      "learning_rate": 8.912895526925726e-06,
      "loss": 0.6018,
      "step": 183
    },
    {
      "epoch": 0.8846153846153846,
      "grad_norm": 0.506033718585968,
      "learning_rate": 8.895402872628352e-06,
      "loss": 0.6029,
      "step": 184
    },
    {
      "epoch": 0.8894230769230769,
      "grad_norm": 0.6721712350845337,
      "learning_rate": 8.87778805938746e-06,
      "loss": 0.5957,
      "step": 185
    },
    {
      "epoch": 0.8942307692307693,
      "grad_norm": 0.5404098033905029,
      "learning_rate": 8.86005163959956e-06,
      "loss": 0.601,
      "step": 186
    },
    {
      "epoch": 0.8990384615384616,
      "grad_norm": 0.6862823367118835,
      "learning_rate": 8.842194169474727e-06,
      "loss": 0.5961,
      "step": 187
    },
    {
      "epoch": 0.9038461538461539,
      "grad_norm": 0.6001689434051514,
      "learning_rate": 8.824216209019139e-06,
      "loss": 0.6023,
      "step": 188
    },
    {
      "epoch": 0.9086538461538461,
      "grad_norm": 0.5968487858772278,
      "learning_rate": 8.806118322017525e-06,
      "loss": 0.6087,
      "step": 189
    },
    {
      "epoch": 0.9134615384615384,
      "grad_norm": 0.5125237703323364,
      "learning_rate": 8.787901076015487e-06,
      "loss": 0.5929,
      "step": 190
    },
    {
      "epoch": 0.9182692307692307,
      "grad_norm": 0.7270171046257019,
      "learning_rate": 8.769565042301692e-06,
      "loss": 0.6209,
      "step": 191
    },
    {
      "epoch": 0.9230769230769231,
      "grad_norm": 0.5782671570777893,
      "learning_rate": 8.751110795889966e-06,
      "loss": 0.5954,
      "step": 192
    },
    {
      "epoch": 0.9278846153846154,
      "grad_norm": 0.5898118615150452,
      "learning_rate": 8.732538915501257e-06,
      "loss": 0.5993,
      "step": 193
    },
    {
      "epoch": 0.9326923076923077,
      "grad_norm": 0.5937283635139465,
      "learning_rate": 8.71384998354549e-06,
      "loss": 0.614,
      "step": 194
    },
    {
      "epoch": 0.9375,
      "grad_norm": 0.6111398935317993,
      "learning_rate": 8.695044586103297e-06,
      "loss": 0.605,
      "step": 195
    },
    {
      "epoch": 0.9423076923076923,
      "grad_norm": 0.592433750629425,
      "learning_rate": 8.676123312907641e-06,
      "loss": 0.6137,
      "step": 196
    },
    {
      "epoch": 0.9471153846153846,
      "grad_norm": 0.579230546951294,
      "learning_rate": 8.657086757325328e-06,
      "loss": 0.5918,
      "step": 197
    },
    {
      "epoch": 0.9519230769230769,
      "grad_norm": 0.7127505540847778,
      "learning_rate": 8.637935516338384e-06,
      "loss": 0.6034,
      "step": 198
    },
    {
      "epoch": 0.9567307692307693,
      "grad_norm": 0.6220248341560364,
      "learning_rate": 8.61867019052535e-06,
      "loss": 0.6043,
      "step": 199
    },
    {
      "epoch": 0.9615384615384616,
      "grad_norm": 0.523986279964447,
      "learning_rate": 8.599291384042442e-06,
      "loss": 0.5896,
      "step": 200
    },
    {
      "epoch": 0.9663461538461539,
      "grad_norm": 0.7416500449180603,
      "learning_rate": 8.579799704604597e-06,
      "loss": 0.5966,
      "step": 201
    },
    {
      "epoch": 0.9711538461538461,
      "grad_norm": 0.6234787106513977,
      "learning_rate": 8.560195763466428e-06,
      "loss": 0.6022,
      "step": 202
    },
    {
      "epoch": 0.9759615384615384,
      "grad_norm": 0.608817458152771,
      "learning_rate": 8.540480175403045e-06,
      "loss": 0.6001,
      "step": 203
    },
    {
      "epoch": 0.9807692307692307,
      "grad_norm": 0.5969523191452026,
      "learning_rate": 8.520653558690785e-06,
      "loss": 0.6143,
      "step": 204
    },
    {
      "epoch": 0.9855769230769231,
      "grad_norm": 0.6564033031463623,
      "learning_rate": 8.500716535087815e-06,
      "loss": 0.5862,
      "step": 205
    },
    {
      "epoch": 0.9903846153846154,
      "grad_norm": 0.5331311821937561,
      "learning_rate": 8.480669729814635e-06,
      "loss": 0.6073,
      "step": 206
    },
    {
      "epoch": 0.9951923076923077,
      "grad_norm": 0.5960302948951721,
      "learning_rate": 8.460513771534475e-06,
      "loss": 0.5938,
      "step": 207
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5379106402397156,
      "learning_rate": 8.440249292333583e-06,
      "loss": 0.6125,
      "step": 208
    },
    {
      "epoch": 1.0048076923076923,
      "grad_norm": 0.6571820378303528,
      "learning_rate": 8.41987692770139e-06,
      "loss": 0.5833,
      "step": 209
    },
    {
      "epoch": 1.0096153846153846,
      "grad_norm": 0.5532863736152649,
      "learning_rate": 8.399397316510596e-06,
      "loss": 0.5801,
      "step": 210
    },
    {
      "epoch": 1.0144230769230769,
      "grad_norm": 0.5147875547409058,
      "learning_rate": 8.378811100997122e-06,
      "loss": 0.5696,
      "step": 211
    },
    {
      "epoch": 1.0192307692307692,
      "grad_norm": 0.5540103316307068,
      "learning_rate": 8.358118926739984e-06,
      "loss": 0.578,
      "step": 212
    },
    {
      "epoch": 1.0240384615384615,
      "grad_norm": 0.5395882725715637,
      "learning_rate": 8.337321442641036e-06,
      "loss": 0.5836,
      "step": 213
    },
    {
      "epoch": 1.0288461538461537,
      "grad_norm": 0.5106703639030457,
      "learning_rate": 8.316419300904622e-06,
      "loss": 0.5531,
      "step": 214
    },
    {
      "epoch": 1.0336538461538463,
      "grad_norm": 0.570037841796875,
      "learning_rate": 8.295413157017127e-06,
      "loss": 0.5619,
      "step": 215
    },
    {
      "epoch": 1.0384615384615385,
      "grad_norm": 0.5976242423057556,
      "learning_rate": 8.274303669726427e-06,
      "loss": 0.559,
      "step": 216
    },
    {
      "epoch": 1.0432692307692308,
      "grad_norm": 0.5582316517829895,
      "learning_rate": 8.25309150102121e-06,
      "loss": 0.5656,
      "step": 217
    },
    {
      "epoch": 1.0480769230769231,
      "grad_norm": 0.535399854183197,
      "learning_rate": 8.231777316110245e-06,
      "loss": 0.5519,
      "step": 218
    },
    {
      "epoch": 1.0528846153846154,
      "grad_norm": 0.6076732277870178,
      "learning_rate": 8.210361783401491e-06,
      "loss": 0.5601,
      "step": 219
    },
    {
      "epoch": 1.0576923076923077,
      "grad_norm": 0.5462629199028015,
      "learning_rate": 8.188845574481162e-06,
      "loss": 0.5777,
      "step": 220
    },
    {
      "epoch": 1.0625,
      "grad_norm": 0.6692743301391602,
      "learning_rate": 8.167229364092648e-06,
      "loss": 0.5874,
      "step": 221
    },
    {
      "epoch": 1.0673076923076923,
      "grad_norm": 0.5311057567596436,
      "learning_rate": 8.145513830115367e-06,
      "loss": 0.5846,
      "step": 222
    },
    {
      "epoch": 1.0721153846153846,
      "grad_norm": 0.5106709599494934,
      "learning_rate": 8.1236996535435e-06,
      "loss": 0.5643,
      "step": 223
    },
    {
      "epoch": 1.0769230769230769,
      "grad_norm": 0.5355270504951477,
      "learning_rate": 8.101787518464634e-06,
      "loss": 0.5621,
      "step": 224
    },
    {
      "epoch": 1.0817307692307692,
      "grad_norm": 0.5128294825553894,
      "learning_rate": 8.079778112038318e-06,
      "loss": 0.5695,
      "step": 225
    },
    {
      "epoch": 1.0865384615384615,
      "grad_norm": 0.5155346989631653,
      "learning_rate": 8.057672124474508e-06,
      "loss": 0.5589,
      "step": 226
    },
    {
      "epoch": 1.0913461538461537,
      "grad_norm": 0.6576012372970581,
      "learning_rate": 8.035470249011916e-06,
      "loss": 0.5612,
      "step": 227
    },
    {
      "epoch": 1.0961538461538463,
      "grad_norm": 0.5001193284988403,
      "learning_rate": 8.013173181896283e-06,
      "loss": 0.5701,
      "step": 228
    },
    {
      "epoch": 1.1009615384615385,
      "grad_norm": 0.6337225437164307,
      "learning_rate": 7.990781622358535e-06,
      "loss": 0.558,
      "step": 229
    },
    {
      "epoch": 1.1057692307692308,
      "grad_norm": 0.622628390789032,
      "learning_rate": 7.968296272592862e-06,
      "loss": 0.5489,
      "step": 230
    },
    {
      "epoch": 1.1105769230769231,
      "grad_norm": 0.5139206051826477,
      "learning_rate": 7.945717837734688e-06,
      "loss": 0.544,
      "step": 231
    },
    {
      "epoch": 1.1153846153846154,
      "grad_norm": 0.4990007281303406,
      "learning_rate": 7.923047025838573e-06,
      "loss": 0.5771,
      "step": 232
    },
    {
      "epoch": 1.1201923076923077,
      "grad_norm": 0.43576645851135254,
      "learning_rate": 7.900284547855992e-06,
      "loss": 0.5668,
      "step": 233
    },
    {
      "epoch": 1.125,
      "grad_norm": 0.4898163378238678,
      "learning_rate": 7.87743111761305e-06,
      "loss": 0.5659,
      "step": 234
    },
    {
      "epoch": 1.1298076923076923,
      "grad_norm": 0.44757646322250366,
      "learning_rate": 7.8544874517881e-06,
      "loss": 0.5384,
      "step": 235
    },
    {
      "epoch": 1.1346153846153846,
      "grad_norm": 0.5017294883728027,
      "learning_rate": 7.831454269889251e-06,
      "loss": 0.5636,
      "step": 236
    },
    {
      "epoch": 1.1394230769230769,
      "grad_norm": 0.5470489263534546,
      "learning_rate": 7.808332294231824e-06,
      "loss": 0.5653,
      "step": 237
    },
    {
      "epoch": 1.1442307692307692,
      "grad_norm": 0.5034846067428589,
      "learning_rate": 7.785122249915688e-06,
      "loss": 0.5591,
      "step": 238
    },
    {
      "epoch": 1.1490384615384615,
      "grad_norm": 0.5333671569824219,
      "learning_rate": 7.76182486480253e-06,
      "loss": 0.5473,
      "step": 239
    },
    {
      "epoch": 1.1538461538461537,
      "grad_norm": 0.5511570572853088,
      "learning_rate": 7.738440869493018e-06,
      "loss": 0.5551,
      "step": 240
    },
    {
      "epoch": 1.1586538461538463,
      "grad_norm": 0.4971109628677368,
      "learning_rate": 7.714970997303898e-06,
      "loss": 0.5444,
      "step": 241
    },
    {
      "epoch": 1.1634615384615385,
      "grad_norm": 0.5407963991165161,
      "learning_rate": 7.691415984244998e-06,
      "loss": 0.5681,
      "step": 242
    },
    {
      "epoch": 1.1682692307692308,
      "grad_norm": 0.4940282106399536,
      "learning_rate": 7.667776568996143e-06,
      "loss": 0.5734,
      "step": 243
    },
    {
      "epoch": 1.1730769230769231,
      "grad_norm": 0.46702679991722107,
      "learning_rate": 7.64405349288399e-06,
      "loss": 0.5527,
      "step": 244
    },
    {
      "epoch": 1.1778846153846154,
      "grad_norm": 0.49713486433029175,
      "learning_rate": 7.62024749985878e-06,
      "loss": 0.5618,
      "step": 245
    },
    {
      "epoch": 1.1826923076923077,
      "grad_norm": 0.5031074285507202,
      "learning_rate": 7.596359336471015e-06,
      "loss": 0.546,
      "step": 246
    },
    {
      "epoch": 1.1875,
      "grad_norm": 0.5209451913833618,
      "learning_rate": 7.572389751848037e-06,
      "loss": 0.5599,
      "step": 247
    },
    {
      "epoch": 1.1923076923076923,
      "grad_norm": 0.5612902045249939,
      "learning_rate": 7.548339497670538e-06,
      "loss": 0.5537,
      "step": 248
    },
    {
      "epoch": 1.1971153846153846,
      "grad_norm": 0.595409631729126,
      "learning_rate": 7.524209328148995e-06,
      "loss": 0.5488,
      "step": 249
    },
    {
      "epoch": 1.2019230769230769,
      "grad_norm": 0.5032594203948975,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.5835,
      "step": 250
    },
    {
      "epoch": 1.2067307692307692,
      "grad_norm": 0.5198405385017395,
      "learning_rate": 7.4757122724225575e-06,
      "loss": 0.5783,
      "step": 251
    },
    {
      "epoch": 1.2115384615384615,
      "grad_norm": 0.552277147769928,
      "learning_rate": 7.451346907074245e-06,
      "loss": 0.5609,
      "step": 252
    },
    {
      "epoch": 1.2163461538461537,
      "grad_norm": 0.554760217666626,
      "learning_rate": 7.426904668047352e-06,
      "loss": 0.5678,
      "step": 253
    },
    {
      "epoch": 1.2211538461538463,
      "grad_norm": 0.49330809712409973,
      "learning_rate": 7.40238632184491e-06,
      "loss": 0.5779,
      "step": 254
    },
    {
      "epoch": 1.2259615384615385,
      "grad_norm": 0.6228109002113342,
      "learning_rate": 7.377792637356644e-06,
      "loss": 0.5496,
      "step": 255
    },
    {
      "epoch": 1.2307692307692308,
      "grad_norm": 0.4928394854068756,
      "learning_rate": 7.35312438583488e-06,
      "loss": 0.5928,
      "step": 256
    },
    {
      "epoch": 1.2355769230769231,
      "grad_norm": 0.5220669507980347,
      "learning_rate": 7.3283823408703466e-06,
      "loss": 0.5657,
      "step": 257
    },
    {
      "epoch": 1.2403846153846154,
      "grad_norm": 0.5418768525123596,
      "learning_rate": 7.303567278367918e-06,
      "loss": 0.5493,
      "step": 258
    },
    {
      "epoch": 1.2451923076923077,
      "grad_norm": 0.6059012413024902,
      "learning_rate": 7.278679976522279e-06,
      "loss": 0.5781,
      "step": 259
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.5703076720237732,
      "learning_rate": 7.253721215793528e-06,
      "loss": 0.5766,
      "step": 260
    },
    {
      "epoch": 1.2548076923076923,
      "grad_norm": 0.7210759520530701,
      "learning_rate": 7.2286917788826926e-06,
      "loss": 0.5951,
      "step": 261
    },
    {
      "epoch": 1.2596153846153846,
      "grad_norm": 0.49044269323349,
      "learning_rate": 7.203592450707193e-06,
      "loss": 0.5625,
      "step": 262
    },
    {
      "epoch": 1.2644230769230769,
      "grad_norm": 0.5832895636558533,
      "learning_rate": 7.178424018376224e-06,
      "loss": 0.5594,
      "step": 263
    },
    {
      "epoch": 1.2692307692307692,
      "grad_norm": 0.5973503589630127,
      "learning_rate": 7.153187271166071e-06,
      "loss": 0.5629,
      "step": 264
    },
    {
      "epoch": 1.2740384615384617,
      "grad_norm": 0.5439810752868652,
      "learning_rate": 7.127883000495353e-06,
      "loss": 0.5548,
      "step": 265
    },
    {
      "epoch": 1.2788461538461537,
      "grad_norm": 0.5828291177749634,
      "learning_rate": 7.102511999900213e-06,
      "loss": 0.5533,
      "step": 266
    },
    {
      "epoch": 1.2836538461538463,
      "grad_norm": 0.601917028427124,
      "learning_rate": 7.0770750650094335e-06,
      "loss": 0.5673,
      "step": 267
    },
    {
      "epoch": 1.2884615384615383,
      "grad_norm": 0.6008758544921875,
      "learning_rate": 7.051572993519474e-06,
      "loss": 0.557,
      "step": 268
    },
    {
      "epoch": 1.2932692307692308,
      "grad_norm": 0.49223729968070984,
      "learning_rate": 7.026006585169467e-06,
      "loss": 0.5623,
      "step": 269
    },
    {
      "epoch": 1.2980769230769231,
      "grad_norm": 0.6068645119667053,
      "learning_rate": 7.0003766417161335e-06,
      "loss": 0.5706,
      "step": 270
    },
    {
      "epoch": 1.3028846153846154,
      "grad_norm": 0.5903732776641846,
      "learning_rate": 6.974683966908642e-06,
      "loss": 0.5625,
      "step": 271
    },
    {
      "epoch": 1.3076923076923077,
      "grad_norm": 0.550208568572998,
      "learning_rate": 6.948929366463397e-06,
      "loss": 0.5873,
      "step": 272
    },
    {
      "epoch": 1.3125,
      "grad_norm": 0.4981112480163574,
      "learning_rate": 6.923113648038784e-06,
      "loss": 0.5586,
      "step": 273
    },
    {
      "epoch": 1.3173076923076923,
      "grad_norm": 0.48777687549591064,
      "learning_rate": 6.897237621209831e-06,
      "loss": 0.5635,
      "step": 274
    },
    {
      "epoch": 1.3221153846153846,
      "grad_norm": 0.540177047252655,
      "learning_rate": 6.87130209744282e-06,
      "loss": 0.5619,
      "step": 275
    },
    {
      "epoch": 1.3269230769230769,
      "grad_norm": 0.49552926421165466,
      "learning_rate": 6.845307890069851e-06,
      "loss": 0.5506,
      "step": 276
    },
    {
      "epoch": 1.3317307692307692,
      "grad_norm": 0.6340379118919373,
      "learning_rate": 6.8192558142633215e-06,
      "loss": 0.5794,
      "step": 277
    },
    {
      "epoch": 1.3365384615384617,
      "grad_norm": 0.4628427028656006,
      "learning_rate": 6.7931466870103735e-06,
      "loss": 0.5692,
      "step": 278
    },
    {
      "epoch": 1.3413461538461537,
      "grad_norm": 0.5693328380584717,
      "learning_rate": 6.766981327087271e-06,
      "loss": 0.5617,
      "step": 279
    },
    {
      "epoch": 1.3461538461538463,
      "grad_norm": 0.564184308052063,
      "learning_rate": 6.740760555033715e-06,
      "loss": 0.5501,
      "step": 280
    },
    {
      "epoch": 1.3509615384615383,
      "grad_norm": 0.4685192406177521,
      "learning_rate": 6.714485193127126e-06,
      "loss": 0.5469,
      "step": 281
    },
    {
      "epoch": 1.3557692307692308,
      "grad_norm": 0.5547739863395691,
      "learning_rate": 6.688156065356845e-06,
      "loss": 0.5787,
      "step": 282
    },
    {
      "epoch": 1.3605769230769231,
      "grad_norm": 0.5170398354530334,
      "learning_rate": 6.6617739973982985e-06,
      "loss": 0.5526,
      "step": 283
    },
    {
      "epoch": 1.3653846153846154,
      "grad_norm": 0.5516389608383179,
      "learning_rate": 6.635339816587109e-06,
      "loss": 0.5769,
      "step": 284
    },
    {
      "epoch": 1.3701923076923077,
      "grad_norm": 0.6596104502677917,
      "learning_rate": 6.60885435189314e-06,
      "loss": 0.5687,
      "step": 285
    },
    {
      "epoch": 1.375,
      "grad_norm": 0.5153890252113342,
      "learning_rate": 6.582318433894513e-06,
      "loss": 0.5623,
      "step": 286
    },
    {
      "epoch": 1.3798076923076923,
      "grad_norm": 0.7115325927734375,
      "learning_rate": 6.555732894751548e-06,
      "loss": 0.5588,
      "step": 287
    },
    {
      "epoch": 1.3846153846153846,
      "grad_norm": 0.5545114278793335,
      "learning_rate": 6.529098568180672e-06,
      "loss": 0.5605,
      "step": 288
    },
    {
      "epoch": 1.3894230769230769,
      "grad_norm": 0.617314338684082,
      "learning_rate": 6.502416289428282e-06,
      "loss": 0.5768,
      "step": 289
    },
    {
      "epoch": 1.3942307692307692,
      "grad_norm": 0.6576439142227173,
      "learning_rate": 6.475686895244534e-06,
      "loss": 0.5544,
      "step": 290
    },
    {
      "epoch": 1.3990384615384617,
      "grad_norm": 0.5532417893409729,
      "learning_rate": 6.448911223857124e-06,
      "loss": 0.5457,
      "step": 291
    },
    {
      "epoch": 1.4038461538461537,
      "grad_norm": 0.7050262689590454,
      "learning_rate": 6.422090114944982e-06,
      "loss": 0.5703,
      "step": 292
    },
    {
      "epoch": 1.4086538461538463,
      "grad_norm": 0.5105507969856262,
      "learning_rate": 6.3952244096119535e-06,
      "loss": 0.5502,
      "step": 293
    },
    {
      "epoch": 1.4134615384615383,
      "grad_norm": 0.6280075907707214,
      "learning_rate": 6.368314950360416e-06,
      "loss": 0.572,
      "step": 294
    },
    {
      "epoch": 1.4182692307692308,
      "grad_norm": 0.5413298606872559,
      "learning_rate": 6.341362581064856e-06,
      "loss": 0.5599,
      "step": 295
    },
    {
      "epoch": 1.4230769230769231,
      "grad_norm": 0.5309041738510132,
      "learning_rate": 6.314368146945418e-06,
      "loss": 0.5496,
      "step": 296
    },
    {
      "epoch": 1.4278846153846154,
      "grad_norm": 0.585471510887146,
      "learning_rate": 6.28733249454138e-06,
      "loss": 0.5519,
      "step": 297
    },
    {
      "epoch": 1.4326923076923077,
      "grad_norm": 0.4619200825691223,
      "learning_rate": 6.260256471684622e-06,
      "loss": 0.562,
      "step": 298
    },
    {
      "epoch": 1.4375,
      "grad_norm": 0.568500816822052,
      "learning_rate": 6.233140927473033e-06,
      "loss": 0.5404,
      "step": 299
    },
    {
      "epoch": 1.4423076923076923,
      "grad_norm": 0.5464410781860352,
      "learning_rate": 6.205986712243876e-06,
      "loss": 0.5724,
      "step": 300
    },
    {
      "epoch": 1.4471153846153846,
      "grad_norm": 0.5567378401756287,
      "learning_rate": 6.178794677547138e-06,
      "loss": 0.5841,
      "step": 301
    },
    {
      "epoch": 1.4519230769230769,
      "grad_norm": 0.5256141424179077,
      "learning_rate": 6.151565676118805e-06,
      "loss": 0.556,
      "step": 302
    },
    {
      "epoch": 1.4567307692307692,
      "grad_norm": 0.5142210721969604,
      "learning_rate": 6.124300561854139e-06,
      "loss": 0.5667,
      "step": 303
    },
    {
      "epoch": 1.4615384615384617,
      "grad_norm": 0.5487565398216248,
      "learning_rate": 6.097000189780893e-06,
      "loss": 0.5557,
      "step": 304
    },
    {
      "epoch": 1.4663461538461537,
      "grad_norm": 0.5075297951698303,
      "learning_rate": 6.0696654160324875e-06,
      "loss": 0.5737,
      "step": 305
    },
    {
      "epoch": 1.4711538461538463,
      "grad_norm": 0.5169580578804016,
      "learning_rate": 6.042297097821184e-06,
      "loss": 0.5487,
      "step": 306
    },
    {
      "epoch": 1.4759615384615383,
      "grad_norm": 0.5064544677734375,
      "learning_rate": 6.014896093411181e-06,
      "loss": 0.5471,
      "step": 307
    },
    {
      "epoch": 1.4807692307692308,
      "grad_norm": 0.4726231098175049,
      "learning_rate": 5.987463262091715e-06,
      "loss": 0.544,
      "step": 308
    },
    {
      "epoch": 1.4855769230769231,
      "grad_norm": 0.5868953466415405,
      "learning_rate": 5.959999464150101e-06,
      "loss": 0.5579,
      "step": 309
    },
    {
      "epoch": 1.4903846153846154,
      "grad_norm": 0.5180251598358154,
      "learning_rate": 5.932505560844766e-06,
      "loss": 0.5547,
      "step": 310
    },
    {
      "epoch": 1.4951923076923077,
      "grad_norm": 0.5364190936088562,
      "learning_rate": 5.904982414378233e-06,
      "loss": 0.5838,
      "step": 311
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.5316298007965088,
      "learning_rate": 5.877430887870081e-06,
      "loss": 0.5821,
      "step": 312
    },
    {
      "epoch": 1.5048076923076923,
      "grad_norm": 0.4852922558784485,
      "learning_rate": 5.849851845329884e-06,
      "loss": 0.5718,
      "step": 313
    },
    {
      "epoch": 1.5096153846153846,
      "grad_norm": 0.4950406551361084,
      "learning_rate": 5.822246151630109e-06,
      "loss": 0.5475,
      "step": 314
    },
    {
      "epoch": 1.5144230769230769,
      "grad_norm": 0.5721108317375183,
      "learning_rate": 5.794614672479e-06,
      "loss": 0.5682,
      "step": 315
    },
    {
      "epoch": 1.5192307692307692,
      "grad_norm": 0.43430256843566895,
      "learning_rate": 5.766958274393428e-06,
      "loss": 0.5723,
      "step": 316
    },
    {
      "epoch": 1.5240384615384617,
      "grad_norm": 0.5114768147468567,
      "learning_rate": 5.739277824671711e-06,
      "loss": 0.5724,
      "step": 317
    },
    {
      "epoch": 1.5288461538461537,
      "grad_norm": 0.5058174133300781,
      "learning_rate": 5.711574191366427e-06,
      "loss": 0.5636,
      "step": 318
    },
    {
      "epoch": 1.5336538461538463,
      "grad_norm": 0.4865401089191437,
      "learning_rate": 5.683848243257181e-06,
      "loss": 0.5444,
      "step": 319
    },
    {
      "epoch": 1.5384615384615383,
      "grad_norm": 0.5734021067619324,
      "learning_rate": 5.656100849823366e-06,
      "loss": 0.5703,
      "step": 320
    },
    {
      "epoch": 1.5432692307692308,
      "grad_norm": 0.4867899715900421,
      "learning_rate": 5.628332881216899e-06,
      "loss": 0.5728,
      "step": 321
    },
    {
      "epoch": 1.5480769230769231,
      "grad_norm": 0.46812713146209717,
      "learning_rate": 5.600545208234927e-06,
      "loss": 0.5624,
      "step": 322
    },
    {
      "epoch": 1.5528846153846154,
      "grad_norm": 0.5345550775527954,
      "learning_rate": 5.57273870229252e-06,
      "loss": 0.5487,
      "step": 323
    },
    {
      "epoch": 1.5576923076923077,
      "grad_norm": 0.45916569232940674,
      "learning_rate": 5.544914235395347e-06,
      "loss": 0.551,
      "step": 324
    },
    {
      "epoch": 1.5625,
      "grad_norm": 0.4594832956790924,
      "learning_rate": 5.517072680112332e-06,
      "loss": 0.5619,
      "step": 325
    },
    {
      "epoch": 1.5673076923076923,
      "grad_norm": 0.5421361923217773,
      "learning_rate": 5.4892149095482815e-06,
      "loss": 0.5509,
      "step": 326
    },
    {
      "epoch": 1.5721153846153846,
      "grad_norm": 0.4083757996559143,
      "learning_rate": 5.46134179731651e-06,
      "loss": 0.567,
      "step": 327
    },
    {
      "epoch": 1.5769230769230769,
      "grad_norm": 0.4842531085014343,
      "learning_rate": 5.4334542175114495e-06,
      "loss": 0.5572,
      "step": 328
    },
    {
      "epoch": 1.5817307692307692,
      "grad_norm": 0.5053375363349915,
      "learning_rate": 5.40555304468122e-06,
      "loss": 0.5653,
      "step": 329
    },
    {
      "epoch": 1.5865384615384617,
      "grad_norm": 0.5320326685905457,
      "learning_rate": 5.377639153800229e-06,
      "loss": 0.5595,
      "step": 330
    },
    {
      "epoch": 1.5913461538461537,
      "grad_norm": 0.502543568611145,
      "learning_rate": 5.34971342024171e-06,
      "loss": 0.572,
      "step": 331
    },
    {
      "epoch": 1.5961538461538463,
      "grad_norm": 0.47758862376213074,
      "learning_rate": 5.321776719750283e-06,
      "loss": 0.5742,
      "step": 332
    },
    {
      "epoch": 1.6009615384615383,
      "grad_norm": 0.49216747283935547,
      "learning_rate": 5.29382992841449e-06,
      "loss": 0.5618,
      "step": 333
    },
    {
      "epoch": 1.6057692307692308,
      "grad_norm": 0.5157171487808228,
      "learning_rate": 5.265873922639315e-06,
      "loss": 0.5513,
      "step": 334
    },
    {
      "epoch": 1.6105769230769231,
      "grad_norm": 0.48715752363204956,
      "learning_rate": 5.237909579118713e-06,
      "loss": 0.5626,
      "step": 335
    },
    {
      "epoch": 1.6153846153846154,
      "grad_norm": 0.48291143774986267,
      "learning_rate": 5.209937774808098e-06,
      "loss": 0.559,
      "step": 336
    },
    {
      "epoch": 1.6201923076923077,
      "grad_norm": 0.5301973819732666,
      "learning_rate": 5.181959386896862e-06,
      "loss": 0.5612,
      "step": 337
    },
    {
      "epoch": 1.625,
      "grad_norm": 0.4594566226005554,
      "learning_rate": 5.153975292780852e-06,
      "loss": 0.5637,
      "step": 338
    },
    {
      "epoch": 1.6298076923076923,
      "grad_norm": 0.4743099510669708,
      "learning_rate": 5.125986370034862e-06,
      "loss": 0.568,
      "step": 339
    },
    {
      "epoch": 1.6346153846153846,
      "grad_norm": 0.46242398023605347,
      "learning_rate": 5.097993496385112e-06,
      "loss": 0.5612,
      "step": 340
    },
    {
      "epoch": 1.6394230769230769,
      "grad_norm": 0.45778384804725647,
      "learning_rate": 5.069997549681718e-06,
      "loss": 0.5608,
      "step": 341
    },
    {
      "epoch": 1.6442307692307692,
      "grad_norm": 0.47530800104141235,
      "learning_rate": 5.041999407871168e-06,
      "loss": 0.5687,
      "step": 342
    },
    {
      "epoch": 1.6490384615384617,
      "grad_norm": 0.5340310335159302,
      "learning_rate": 5.01399994896879e-06,
      "loss": 0.5452,
      "step": 343
    },
    {
      "epoch": 1.6538461538461537,
      "grad_norm": 0.43904221057891846,
      "learning_rate": 4.986000051031212e-06,
      "loss": 0.5535,
      "step": 344
    },
    {
      "epoch": 1.6586538461538463,
      "grad_norm": 0.49031099677085876,
      "learning_rate": 4.958000592128834e-06,
      "loss": 0.534,
      "step": 345
    },
    {
      "epoch": 1.6634615384615383,
      "grad_norm": 0.4739511013031006,
      "learning_rate": 4.930002450318282e-06,
      "loss": 0.5489,
      "step": 346
    },
    {
      "epoch": 1.6682692307692308,
      "grad_norm": 0.4124240577220917,
      "learning_rate": 4.9020065036148885e-06,
      "loss": 0.5575,
      "step": 347
    },
    {
      "epoch": 1.6730769230769231,
      "grad_norm": 0.4636637270450592,
      "learning_rate": 4.874013629965138e-06,
      "loss": 0.5614,
      "step": 348
    },
    {
      "epoch": 1.6778846153846154,
      "grad_norm": 0.4664590060710907,
      "learning_rate": 4.846024707219149e-06,
      "loss": 0.5743,
      "step": 349
    },
    {
      "epoch": 1.6826923076923077,
      "grad_norm": 0.41771262884140015,
      "learning_rate": 4.818040613103139e-06,
      "loss": 0.5555,
      "step": 350
    },
    {
      "epoch": 1.6875,
      "grad_norm": 0.46983399987220764,
      "learning_rate": 4.790062225191902e-06,
      "loss": 0.5587,
      "step": 351
    },
    {
      "epoch": 1.6923076923076923,
      "grad_norm": 0.4576813578605652,
      "learning_rate": 4.762090420881289e-06,
      "loss": 0.5577,
      "step": 352
    },
    {
      "epoch": 1.6971153846153846,
      "grad_norm": 0.4811893403530121,
      "learning_rate": 4.734126077360685e-06,
      "loss": 0.5537,
      "step": 353
    },
    {
      "epoch": 1.7019230769230769,
      "grad_norm": 0.4063519239425659,
      "learning_rate": 4.706170071585513e-06,
      "loss": 0.5662,
      "step": 354
    },
    {
      "epoch": 1.7067307692307692,
      "grad_norm": 0.4614051580429077,
      "learning_rate": 4.678223280249718e-06,
      "loss": 0.5503,
      "step": 355
    },
    {
      "epoch": 1.7115384615384617,
      "grad_norm": 0.43803513050079346,
      "learning_rate": 4.650286579758291e-06,
      "loss": 0.5652,
      "step": 356
    },
    {
      "epoch": 1.7163461538461537,
      "grad_norm": 0.42544105648994446,
      "learning_rate": 4.622360846199772e-06,
      "loss": 0.5431,
      "step": 357
    },
    {
      "epoch": 1.7211538461538463,
      "grad_norm": 0.39839082956314087,
      "learning_rate": 4.594446955318781e-06,
      "loss": 0.555,
      "step": 358
    },
    {
      "epoch": 1.7259615384615383,
      "grad_norm": 0.38914695382118225,
      "learning_rate": 4.566545782488554e-06,
      "loss": 0.5623,
      "step": 359
    },
    {
      "epoch": 1.7307692307692308,
      "grad_norm": 0.410859078168869,
      "learning_rate": 4.53865820268349e-06,
      "loss": 0.5626,
      "step": 360
    },
    {
      "epoch": 1.7355769230769231,
      "grad_norm": 0.4289568364620209,
      "learning_rate": 4.510785090451719e-06,
      "loss": 0.5659,
      "step": 361
    },
    {
      "epoch": 1.7403846153846154,
      "grad_norm": 0.4343273639678955,
      "learning_rate": 4.482927319887669e-06,
      "loss": 0.5543,
      "step": 362
    },
    {
      "epoch": 1.7451923076923077,
      "grad_norm": 0.41860419511795044,
      "learning_rate": 4.455085764604653e-06,
      "loss": 0.5501,
      "step": 363
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.4254235625267029,
      "learning_rate": 4.427261297707482e-06,
      "loss": 0.5478,
      "step": 364
    },
    {
      "epoch": 1.7548076923076923,
      "grad_norm": 0.420207142829895,
      "learning_rate": 4.399454791765076e-06,
      "loss": 0.5683,
      "step": 365
    },
    {
      "epoch": 1.7596153846153846,
      "grad_norm": 0.4237710237503052,
      "learning_rate": 4.371667118783101e-06,
      "loss": 0.5551,
      "step": 366
    },
    {
      "epoch": 1.7644230769230769,
      "grad_norm": 0.42474254965782166,
      "learning_rate": 4.343899150176635e-06,
      "loss": 0.5508,
      "step": 367
    },
    {
      "epoch": 1.7692307692307692,
      "grad_norm": 0.4262840449810028,
      "learning_rate": 4.316151756742821e-06,
      "loss": 0.5545,
      "step": 368
    },
    {
      "epoch": 1.7740384615384617,
      "grad_norm": 0.4713822603225708,
      "learning_rate": 4.2884258086335755e-06,
      "loss": 0.5632,
      "step": 369
    },
    {
      "epoch": 1.7788461538461537,
      "grad_norm": 0.465378999710083,
      "learning_rate": 4.26072217532829e-06,
      "loss": 0.5706,
      "step": 370
    },
    {
      "epoch": 1.7836538461538463,
      "grad_norm": 0.4259505271911621,
      "learning_rate": 4.233041725606573e-06,
      "loss": 0.5653,
      "step": 371
    },
    {
      "epoch": 1.7884615384615383,
      "grad_norm": 0.4873887300491333,
      "learning_rate": 4.205385327521002e-06,
      "loss": 0.5547,
      "step": 372
    },
    {
      "epoch": 1.7932692307692308,
      "grad_norm": 0.5444599390029907,
      "learning_rate": 4.177753848369892e-06,
      "loss": 0.5635,
      "step": 373
    },
    {
      "epoch": 1.7980769230769231,
      "grad_norm": 0.4417326748371124,
      "learning_rate": 4.1501481546701185e-06,
      "loss": 0.5542,
      "step": 374
    },
    {
      "epoch": 1.8028846153846154,
      "grad_norm": 0.42065057158470154,
      "learning_rate": 4.12256911212992e-06,
      "loss": 0.5712,
      "step": 375
    },
    {
      "epoch": 1.8076923076923077,
      "grad_norm": 0.4802209436893463,
      "learning_rate": 4.095017585621767e-06,
      "loss": 0.5762,
      "step": 376
    },
    {
      "epoch": 1.8125,
      "grad_norm": 0.47133877873420715,
      "learning_rate": 4.067494439155236e-06,
      "loss": 0.5706,
      "step": 377
    },
    {
      "epoch": 1.8173076923076923,
      "grad_norm": 0.42150482535362244,
      "learning_rate": 4.0400005358499e-06,
      "loss": 0.5549,
      "step": 378
    },
    {
      "epoch": 1.8221153846153846,
      "grad_norm": 0.48820599913597107,
      "learning_rate": 4.012536737908288e-06,
      "loss": 0.5646,
      "step": 379
    },
    {
      "epoch": 1.8269230769230769,
      "grad_norm": 0.46589240431785583,
      "learning_rate": 3.985103906588821e-06,
      "loss": 0.5521,
      "step": 380
    },
    {
      "epoch": 1.8317307692307692,
      "grad_norm": 0.43386465311050415,
      "learning_rate": 3.957702902178816e-06,
      "loss": 0.5603,
      "step": 381
    },
    {
      "epoch": 1.8365384615384617,
      "grad_norm": 0.438164085149765,
      "learning_rate": 3.930334583967514e-06,
      "loss": 0.5547,
      "step": 382
    },
    {
      "epoch": 1.8413461538461537,
      "grad_norm": 0.43723925948143005,
      "learning_rate": 3.902999810219109e-06,
      "loss": 0.5277,
      "step": 383
    },
    {
      "epoch": 1.8461538461538463,
      "grad_norm": 0.4327050745487213,
      "learning_rate": 3.875699438145862e-06,
      "loss": 0.5426,
      "step": 384
    },
    {
      "epoch": 1.8509615384615383,
      "grad_norm": 0.4460081160068512,
      "learning_rate": 3.8484343238811976e-06,
      "loss": 0.5478,
      "step": 385
    },
    {
      "epoch": 1.8557692307692308,
      "grad_norm": 0.40871721506118774,
      "learning_rate": 3.821205322452863e-06,
      "loss": 0.5674,
      "step": 386
    },
    {
      "epoch": 1.8605769230769231,
      "grad_norm": 0.39245539903640747,
      "learning_rate": 3.794013287756125e-06,
      "loss": 0.5587,
      "step": 387
    },
    {
      "epoch": 1.8653846153846154,
      "grad_norm": 0.4072204828262329,
      "learning_rate": 3.766859072526969e-06,
      "loss": 0.5503,
      "step": 388
    },
    {
      "epoch": 1.8701923076923077,
      "grad_norm": 0.4407781958580017,
      "learning_rate": 3.7397435283153795e-06,
      "loss": 0.5612,
      "step": 389
    },
    {
      "epoch": 1.875,
      "grad_norm": 0.4392929673194885,
      "learning_rate": 3.712667505458622e-06,
      "loss": 0.5543,
      "step": 390
    },
    {
      "epoch": 1.8798076923076923,
      "grad_norm": 0.4082246720790863,
      "learning_rate": 3.685631853054583e-06,
      "loss": 0.5384,
      "step": 391
    },
    {
      "epoch": 1.8846153846153846,
      "grad_norm": 0.40940192341804504,
      "learning_rate": 3.658637418935146e-06,
      "loss": 0.5586,
      "step": 392
    },
    {
      "epoch": 1.8894230769230769,
      "grad_norm": 0.4302951395511627,
      "learning_rate": 3.6316850496395863e-06,
      "loss": 0.5541,
      "step": 393
    },
    {
      "epoch": 1.8942307692307692,
      "grad_norm": 0.472651869058609,
      "learning_rate": 3.6047755903880478e-06,
      "loss": 0.5523,
      "step": 394
    },
    {
      "epoch": 1.8990384615384617,
      "grad_norm": 0.4774600863456726,
      "learning_rate": 3.577909885055019e-06,
      "loss": 0.5551,
      "step": 395
    },
    {
      "epoch": 1.9038461538461537,
      "grad_norm": 0.40797290205955505,
      "learning_rate": 3.5510887761428764e-06,
      "loss": 0.5675,
      "step": 396
    },
    {
      "epoch": 1.9086538461538463,
      "grad_norm": 0.4044354557991028,
      "learning_rate": 3.524313104755468e-06,
      "loss": 0.5804,
      "step": 397
    },
    {
      "epoch": 1.9134615384615383,
      "grad_norm": 0.48422014713287354,
      "learning_rate": 3.4975837105717203e-06,
      "loss": 0.5551,
      "step": 398
    },
    {
      "epoch": 1.9182692307692308,
      "grad_norm": 0.4386497735977173,
      "learning_rate": 3.4709014318193298e-06,
      "loss": 0.5637,
      "step": 399
    },
    {
      "epoch": 1.9230769230769231,
      "grad_norm": 0.41975468397140503,
      "learning_rate": 3.4442671052484545e-06,
      "loss": 0.5714,
      "step": 400
    },
    {
      "epoch": 1.9278846153846154,
      "grad_norm": 0.37439560890197754,
      "learning_rate": 3.4176815661054884e-06,
      "loss": 0.5422,
      "step": 401
    },
    {
      "epoch": 1.9326923076923077,
      "grad_norm": 0.40816617012023926,
      "learning_rate": 3.3911456481068613e-06,
      "loss": 0.5438,
      "step": 402
    },
    {
      "epoch": 1.9375,
      "grad_norm": 0.47742047905921936,
      "learning_rate": 3.3646601834128924e-06,
      "loss": 0.5584,
      "step": 403
    },
    {
      "epoch": 1.9423076923076923,
      "grad_norm": 0.44102102518081665,
      "learning_rate": 3.3382260026017027e-06,
      "loss": 0.5481,
      "step": 404
    },
    {
      "epoch": 1.9471153846153846,
      "grad_norm": 0.39150795340538025,
      "learning_rate": 3.311843934643157e-06,
      "loss": 0.5545,
      "step": 405
    },
    {
      "epoch": 1.9519230769230769,
      "grad_norm": 0.4709937572479248,
      "learning_rate": 3.2855148068728753e-06,
      "loss": 0.5654,
      "step": 406
    },
    {
      "epoch": 1.9567307692307692,
      "grad_norm": 0.4761818051338196,
      "learning_rate": 3.2592394449662867e-06,
      "loss": 0.5343,
      "step": 407
    },
    {
      "epoch": 1.9615384615384617,
      "grad_norm": 0.44496825337409973,
      "learning_rate": 3.233018672912731e-06,
      "loss": 0.5527,
      "step": 408
    },
    {
      "epoch": 1.9663461538461537,
      "grad_norm": 0.5426850318908691,
      "learning_rate": 3.2068533129896273e-06,
      "loss": 0.5473,
      "step": 409
    },
    {
      "epoch": 1.9711538461538463,
      "grad_norm": 0.499978244304657,
      "learning_rate": 3.1807441857366798e-06,
      "loss": 0.5656,
      "step": 410
    },
    {
      "epoch": 1.9759615384615383,
      "grad_norm": 0.41915661096572876,
      "learning_rate": 3.1546921099301507e-06,
      "loss": 0.5797,
      "step": 411
    },
    {
      "epoch": 1.9807692307692308,
      "grad_norm": 0.4491116404533386,
      "learning_rate": 3.1286979025571817e-06,
      "loss": 0.5662,
      "step": 412
    },
    {
      "epoch": 1.9855769230769231,
      "grad_norm": 0.42887210845947266,
      "learning_rate": 3.1027623787901706e-06,
      "loss": 0.5593,
      "step": 413
    },
    {
      "epoch": 1.9903846153846154,
      "grad_norm": 0.445030152797699,
      "learning_rate": 3.076886351961217e-06,
      "loss": 0.547,
      "step": 414
    },
    {
      "epoch": 1.9951923076923077,
      "grad_norm": 0.47931942343711853,
      "learning_rate": 3.0510706335366034e-06,
      "loss": 0.5684,
      "step": 415
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.4522424638271332,
      "learning_rate": 3.02531603309136e-06,
      "loss": 0.5519,
      "step": 416
    },
    {
      "epoch": 2.0048076923076925,
      "grad_norm": 0.42797061800956726,
      "learning_rate": 2.9996233582838686e-06,
      "loss": 0.5399,
      "step": 417
    },
    {
      "epoch": 2.0096153846153846,
      "grad_norm": 0.46265673637390137,
      "learning_rate": 2.973993414830534e-06,
      "loss": 0.5101,
      "step": 418
    },
    {
      "epoch": 2.014423076923077,
      "grad_norm": 0.46875470876693726,
      "learning_rate": 2.948427006480528e-06,
      "loss": 0.5288,
      "step": 419
    },
    {
      "epoch": 2.019230769230769,
      "grad_norm": 0.3783641755580902,
      "learning_rate": 2.9229249349905686e-06,
      "loss": 0.5254,
      "step": 420
    },
    {
      "epoch": 2.0240384615384617,
      "grad_norm": 0.4915563464164734,
      "learning_rate": 2.897488000099788e-06,
      "loss": 0.5348,
      "step": 421
    },
    {
      "epoch": 2.0288461538461537,
      "grad_norm": 0.46728116273880005,
      "learning_rate": 2.8721169995046503e-06,
      "loss": 0.5388,
      "step": 422
    },
    {
      "epoch": 2.0336538461538463,
      "grad_norm": 0.4580628275871277,
      "learning_rate": 2.846812728833931e-06,
      "loss": 0.5255,
      "step": 423
    },
    {
      "epoch": 2.0384615384615383,
      "grad_norm": 0.42225995659828186,
      "learning_rate": 2.8215759816237748e-06,
      "loss": 0.5206,
      "step": 424
    },
    {
      "epoch": 2.043269230769231,
      "grad_norm": 0.4477675259113312,
      "learning_rate": 2.796407549292809e-06,
      "loss": 0.503,
      "step": 425
    },
    {
      "epoch": 2.048076923076923,
      "grad_norm": 0.4319400489330292,
      "learning_rate": 2.771308221117309e-06,
      "loss": 0.5187,
      "step": 426
    },
    {
      "epoch": 2.0528846153846154,
      "grad_norm": 0.4639637768268585,
      "learning_rate": 2.7462787842064753e-06,
      "loss": 0.5327,
      "step": 427
    },
    {
      "epoch": 2.0576923076923075,
      "grad_norm": 0.3952256739139557,
      "learning_rate": 2.7213200234777215e-06,
      "loss": 0.5132,
      "step": 428
    },
    {
      "epoch": 2.0625,
      "grad_norm": 0.38108691573143005,
      "learning_rate": 2.696432721632082e-06,
      "loss": 0.5536,
      "step": 429
    },
    {
      "epoch": 2.0673076923076925,
      "grad_norm": 0.3831665515899658,
      "learning_rate": 2.671617659129655e-06,
      "loss": 0.5069,
      "step": 430
    },
    {
      "epoch": 2.0721153846153846,
      "grad_norm": 0.38643160462379456,
      "learning_rate": 2.646875614165121e-06,
      "loss": 0.5222,
      "step": 431
    },
    {
      "epoch": 2.076923076923077,
      "grad_norm": 0.40774914622306824,
      "learning_rate": 2.6222073626433587e-06,
      "loss": 0.5208,
      "step": 432
    },
    {
      "epoch": 2.081730769230769,
      "grad_norm": 0.40251675248146057,
      "learning_rate": 2.597613678155092e-06,
      "loss": 0.513,
      "step": 433
    },
    {
      "epoch": 2.0865384615384617,
      "grad_norm": 0.38830575346946716,
      "learning_rate": 2.573095331952646e-06,
      "loss": 0.5445,
      "step": 434
    },
    {
      "epoch": 2.0913461538461537,
      "grad_norm": 0.4013608694076538,
      "learning_rate": 2.5486530929257574e-06,
      "loss": 0.5285,
      "step": 435
    },
    {
      "epoch": 2.0961538461538463,
      "grad_norm": 0.3629082441329956,
      "learning_rate": 2.5242877275774446e-06,
      "loss": 0.532,
      "step": 436
    },
    {
      "epoch": 2.1009615384615383,
      "grad_norm": 0.40173375606536865,
      "learning_rate": 2.5000000000000015e-06,
      "loss": 0.529,
      "step": 437
    },
    {
      "epoch": 2.105769230769231,
      "grad_norm": 0.3868687152862549,
      "learning_rate": 2.475790671851007e-06,
      "loss": 0.5278,
      "step": 438
    },
    {
      "epoch": 2.110576923076923,
      "grad_norm": 0.41604334115982056,
      "learning_rate": 2.4516605023294626e-06,
      "loss": 0.5284,
      "step": 439
    },
    {
      "epoch": 2.1153846153846154,
      "grad_norm": 0.43302401900291443,
      "learning_rate": 2.4276102481519655e-06,
      "loss": 0.5286,
      "step": 440
    },
    {
      "epoch": 2.1201923076923075,
      "grad_norm": 0.36202147603034973,
      "learning_rate": 2.403640663528986e-06,
      "loss": 0.5415,
      "step": 441
    },
    {
      "epoch": 2.125,
      "grad_norm": 0.36743971705436707,
      "learning_rate": 2.379752500141222e-06,
      "loss": 0.5429,
      "step": 442
    },
    {
      "epoch": 2.1298076923076925,
      "grad_norm": 0.4273446202278137,
      "learning_rate": 2.355946507116012e-06,
      "loss": 0.5329,
      "step": 443
    },
    {
      "epoch": 2.1346153846153846,
      "grad_norm": 0.4194796681404114,
      "learning_rate": 2.332223431003859e-06,
      "loss": 0.547,
      "step": 444
    },
    {
      "epoch": 2.139423076923077,
      "grad_norm": 0.3628264367580414,
      "learning_rate": 2.3085840157550036e-06,
      "loss": 0.5254,
      "step": 445
    },
    {
      "epoch": 2.144230769230769,
      "grad_norm": 0.3705046772956848,
      "learning_rate": 2.2850290026961032e-06,
      "loss": 0.5359,
      "step": 446
    },
    {
      "epoch": 2.1490384615384617,
      "grad_norm": 0.39662665128707886,
      "learning_rate": 2.2615591305069846e-06,
      "loss": 0.5121,
      "step": 447
    },
    {
      "epoch": 2.1538461538461537,
      "grad_norm": 0.3908136785030365,
      "learning_rate": 2.238175135197471e-06,
      "loss": 0.5438,
      "step": 448
    },
    {
      "epoch": 2.1586538461538463,
      "grad_norm": 0.36550912261009216,
      "learning_rate": 2.2148777500843125e-06,
      "loss": 0.5169,
      "step": 449
    },
    {
      "epoch": 2.1634615384615383,
      "grad_norm": 0.39665716886520386,
      "learning_rate": 2.1916677057681786e-06,
      "loss": 0.4957,
      "step": 450
    },
    {
      "epoch": 2.168269230769231,
      "grad_norm": 0.3526061475276947,
      "learning_rate": 2.1685457301107506e-06,
      "loss": 0.5355,
      "step": 451
    },
    {
      "epoch": 2.173076923076923,
      "grad_norm": 0.3616856336593628,
      "learning_rate": 2.145512548211902e-06,
      "loss": 0.5136,
      "step": 452
    },
    {
      "epoch": 2.1778846153846154,
      "grad_norm": 0.36202114820480347,
      "learning_rate": 2.1225688823869494e-06,
      "loss": 0.5107,
      "step": 453
    },
    {
      "epoch": 2.1826923076923075,
      "grad_norm": 0.38488027453422546,
      "learning_rate": 2.09971545214401e-06,
      "loss": 0.5279,
      "step": 454
    },
    {
      "epoch": 2.1875,
      "grad_norm": 0.38552579283714294,
      "learning_rate": 2.0769529741614297e-06,
      "loss": 0.5216,
      "step": 455
    },
    {
      "epoch": 2.1923076923076925,
      "grad_norm": 0.40065112709999084,
      "learning_rate": 2.054282162265313e-06,
      "loss": 0.5092,
      "step": 456
    },
    {
      "epoch": 2.1971153846153846,
      "grad_norm": 0.3517535924911499,
      "learning_rate": 2.0317037274071412e-06,
      "loss": 0.5175,
      "step": 457
    },
    {
      "epoch": 2.201923076923077,
      "grad_norm": 0.3569267988204956,
      "learning_rate": 2.009218377641466e-06,
      "loss": 0.5361,
      "step": 458
    },
    {
      "epoch": 2.206730769230769,
      "grad_norm": 0.3930048942565918,
      "learning_rate": 1.9868268181037186e-06,
      "loss": 0.5252,
      "step": 459
    },
    {
      "epoch": 2.2115384615384617,
      "grad_norm": 0.37880074977874756,
      "learning_rate": 1.964529750988086e-06,
      "loss": 0.5251,
      "step": 460
    },
    {
      "epoch": 2.2163461538461537,
      "grad_norm": 0.38495028018951416,
      "learning_rate": 1.9423278755254933e-06,
      "loss": 0.532,
      "step": 461
    },
    {
      "epoch": 2.2211538461538463,
      "grad_norm": 0.3525002598762512,
      "learning_rate": 1.9202218879616824e-06,
      "loss": 0.5215,
      "step": 462
    },
    {
      "epoch": 2.2259615384615383,
      "grad_norm": 0.3603316843509674,
      "learning_rate": 1.8982124815353665e-06,
      "loss": 0.5221,
      "step": 463
    },
    {
      "epoch": 2.230769230769231,
      "grad_norm": 0.35701784491539,
      "learning_rate": 1.8763003464565022e-06,
      "loss": 0.5117,
      "step": 464
    },
    {
      "epoch": 2.235576923076923,
      "grad_norm": 0.3503670394420624,
      "learning_rate": 1.854486169884635e-06,
      "loss": 0.5457,
      "step": 465
    },
    {
      "epoch": 2.2403846153846154,
      "grad_norm": 0.3608977198600769,
      "learning_rate": 1.8327706359073526e-06,
      "loss": 0.5061,
      "step": 466
    },
    {
      "epoch": 2.2451923076923075,
      "grad_norm": 0.3715517222881317,
      "learning_rate": 1.8111544255188402e-06,
      "loss": 0.5429,
      "step": 467
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.393916517496109,
      "learning_rate": 1.7896382165985094e-06,
      "loss": 0.5272,
      "step": 468
    },
    {
      "epoch": 2.2548076923076925,
      "grad_norm": 0.36094385385513306,
      "learning_rate": 1.768222683889757e-06,
      "loss": 0.5299,
      "step": 469
    },
    {
      "epoch": 2.2596153846153846,
      "grad_norm": 0.3601950407028198,
      "learning_rate": 1.746908498978791e-06,
      "loss": 0.5251,
      "step": 470
    },
    {
      "epoch": 2.264423076923077,
      "grad_norm": 0.3479694426059723,
      "learning_rate": 1.7256963302735752e-06,
      "loss": 0.5398,
      "step": 471
    },
    {
      "epoch": 2.269230769230769,
      "grad_norm": 0.3505624532699585,
      "learning_rate": 1.7045868429828745e-06,
      "loss": 0.4987,
      "step": 472
    },
    {
      "epoch": 2.2740384615384617,
      "grad_norm": 0.3886091411113739,
      "learning_rate": 1.6835806990953802e-06,
      "loss": 0.5267,
      "step": 473
    },
    {
      "epoch": 2.2788461538461537,
      "grad_norm": 0.37276989221572876,
      "learning_rate": 1.6626785573589667e-06,
      "loss": 0.5153,
      "step": 474
    },
    {
      "epoch": 2.2836538461538463,
      "grad_norm": 0.3589704930782318,
      "learning_rate": 1.6418810732600177e-06,
      "loss": 0.513,
      "step": 475
    },
    {
      "epoch": 2.2884615384615383,
      "grad_norm": 0.3904787302017212,
      "learning_rate": 1.6211888990028785e-06,
      "loss": 0.5202,
      "step": 476
    },
    {
      "epoch": 2.293269230769231,
      "grad_norm": 0.3644522726535797,
      "learning_rate": 1.6006026834894068e-06,
      "loss": 0.521,
      "step": 477
    },
    {
      "epoch": 2.298076923076923,
      "grad_norm": 0.3736514151096344,
      "learning_rate": 1.5801230722986104e-06,
      "loss": 0.5056,
      "step": 478
    },
    {
      "epoch": 2.3028846153846154,
      "grad_norm": 0.35614803433418274,
      "learning_rate": 1.5597507076664187e-06,
      "loss": 0.5307,
      "step": 479
    },
    {
      "epoch": 2.3076923076923075,
      "grad_norm": 0.35307708382606506,
      "learning_rate": 1.5394862284655266e-06,
      "loss": 0.5279,
      "step": 480
    },
    {
      "epoch": 2.3125,
      "grad_norm": 0.37877628207206726,
      "learning_rate": 1.5193302701853674e-06,
      "loss": 0.5286,
      "step": 481
    },
    {
      "epoch": 2.3173076923076925,
      "grad_norm": 0.35032036900520325,
      "learning_rate": 1.499283464912188e-06,
      "loss": 0.5401,
      "step": 482
    },
    {
      "epoch": 2.3221153846153846,
      "grad_norm": 0.3594701886177063,
      "learning_rate": 1.4793464413092161e-06,
      "loss": 0.5239,
      "step": 483
    },
    {
      "epoch": 2.326923076923077,
      "grad_norm": 0.35576295852661133,
      "learning_rate": 1.459519824596956e-06,
      "loss": 0.5325,
      "step": 484
    },
    {
      "epoch": 2.331730769230769,
      "grad_norm": 0.38037362694740295,
      "learning_rate": 1.4398042365335745e-06,
      "loss": 0.5181,
      "step": 485
    },
    {
      "epoch": 2.3365384615384617,
      "grad_norm": 0.3730925917625427,
      "learning_rate": 1.4202002953954042e-06,
      "loss": 0.5153,
      "step": 486
    },
    {
      "epoch": 2.3413461538461537,
      "grad_norm": 0.3507210612297058,
      "learning_rate": 1.4007086159575595e-06,
      "loss": 0.5119,
      "step": 487
    },
    {
      "epoch": 2.3461538461538463,
      "grad_norm": 0.3383399248123169,
      "learning_rate": 1.3813298094746491e-06,
      "loss": 0.5458,
      "step": 488
    },
    {
      "epoch": 2.3509615384615383,
      "grad_norm": 0.34395211935043335,
      "learning_rate": 1.362064483661617e-06,
      "loss": 0.5318,
      "step": 489
    },
    {
      "epoch": 2.355769230769231,
      "grad_norm": 0.3459123373031616,
      "learning_rate": 1.3429132426746743e-06,
      "loss": 0.5194,
      "step": 490
    },
    {
      "epoch": 2.360576923076923,
      "grad_norm": 0.38007426261901855,
      "learning_rate": 1.3238766870923592e-06,
      "loss": 0.5311,
      "step": 491
    },
    {
      "epoch": 2.3653846153846154,
      "grad_norm": 0.3795618414878845,
      "learning_rate": 1.3049554138967052e-06,
      "loss": 0.5362,
      "step": 492
    },
    {
      "epoch": 2.3701923076923075,
      "grad_norm": 0.3444887101650238,
      "learning_rate": 1.286150016454511e-06,
      "loss": 0.5263,
      "step": 493
    },
    {
      "epoch": 2.375,
      "grad_norm": 0.39823612570762634,
      "learning_rate": 1.267461084498744e-06,
      "loss": 0.5171,
      "step": 494
    },
    {
      "epoch": 2.3798076923076925,
      "grad_norm": 0.3500751852989197,
      "learning_rate": 1.2488892041100364e-06,
      "loss": 0.5099,
      "step": 495
    },
    {
      "epoch": 2.3846153846153846,
      "grad_norm": 0.34816107153892517,
      "learning_rate": 1.2304349576983094e-06,
      "loss": 0.5343,
      "step": 496
    },
    {
      "epoch": 2.389423076923077,
      "grad_norm": 0.3754243552684784,
      "learning_rate": 1.2120989239845149e-06,
      "loss": 0.5265,
      "step": 497
    },
    {
      "epoch": 2.394230769230769,
      "grad_norm": 0.32973939180374146,
      "learning_rate": 1.1938816779824753e-06,
      "loss": 0.5437,
      "step": 498
    },
    {
      "epoch": 2.3990384615384617,
      "grad_norm": 0.37367406487464905,
      "learning_rate": 1.1757837909808628e-06,
      "loss": 0.5212,
      "step": 499
    },
    {
      "epoch": 2.4038461538461537,
      "grad_norm": 0.34787219762802124,
      "learning_rate": 1.157805830525275e-06,
      "loss": 0.5225,
      "step": 500
    },
    {
      "epoch": 2.4086538461538463,
      "grad_norm": 0.3542671799659729,
      "learning_rate": 1.1399483604004403e-06,
      "loss": 0.5301,
      "step": 501
    },
    {
      "epoch": 2.4134615384615383,
      "grad_norm": 0.36626604199409485,
      "learning_rate": 1.1222119406125426e-06,
      "loss": 0.5234,
      "step": 502
    },
    {
      "epoch": 2.418269230769231,
      "grad_norm": 0.3557561933994293,
      "learning_rate": 1.1045971273716476e-06,
      "loss": 0.5316,
      "step": 503
    },
    {
      "epoch": 2.423076923076923,
      "grad_norm": 0.3680206835269928,
      "learning_rate": 1.0871044730742752e-06,
      "loss": 0.5261,
      "step": 504
    },
    {
      "epoch": 2.4278846153846154,
      "grad_norm": 0.39090457558631897,
      "learning_rate": 1.0697345262860638e-06,
      "loss": 0.5279,
      "step": 505
    },
    {
      "epoch": 2.4326923076923075,
      "grad_norm": 0.33717867732048035,
      "learning_rate": 1.0524878317245713e-06,
      "loss": 0.5218,
      "step": 506
    },
    {
      "epoch": 2.4375,
      "grad_norm": 0.357800155878067,
      "learning_rate": 1.0353649302421982e-06,
      "loss": 0.5227,
      "step": 507
    },
    {
      "epoch": 2.4423076923076925,
      "grad_norm": 0.34802302718162537,
      "learning_rate": 1.0183663588092214e-06,
      "loss": 0.5566,
      "step": 508
    },
    {
      "epoch": 2.4471153846153846,
      "grad_norm": 0.3453739583492279,
      "learning_rate": 1.0014926504969535e-06,
      "loss": 0.5259,
      "step": 509
    },
    {
      "epoch": 2.451923076923077,
      "grad_norm": 0.3763096034526825,
      "learning_rate": 9.847443344610296e-07,
      "loss": 0.5201,
      "step": 510
    },
    {
      "epoch": 2.456730769230769,
      "grad_norm": 0.3687525689601898,
      "learning_rate": 9.681219359248106e-07,
      "loss": 0.5224,
      "step": 511
    },
    {
      "epoch": 2.4615384615384617,
      "grad_norm": 0.3717356026172638,
      "learning_rate": 9.516259761629148e-07,
      "loss": 0.5145,
      "step": 512
    },
    {
      "epoch": 2.4663461538461537,
      "grad_norm": 0.36599400639533997,
      "learning_rate": 9.352569724848715e-07,
      "loss": 0.5086,
      "step": 513
    },
    {
      "epoch": 2.4711538461538463,
      "grad_norm": 0.33455225825309753,
      "learning_rate": 9.190154382188921e-07,
      "loss": 0.5148,
      "step": 514
    },
    {
      "epoch": 2.4759615384615383,
      "grad_norm": 0.3153638243675232,
      "learning_rate": 9.029018826957775e-07,
      "loss": 0.5286,
      "step": 515
    },
    {
      "epoch": 2.480769230769231,
      "grad_norm": 0.3458288609981537,
      "learning_rate": 8.86916811232944e-07,
      "loss": 0.5337,
      "step": 516
    },
    {
      "epoch": 2.485576923076923,
      "grad_norm": 0.33229777216911316,
      "learning_rate": 8.710607251185799e-07,
      "loss": 0.5443,
      "step": 517
    },
    {
      "epoch": 2.4903846153846154,
      "grad_norm": 0.3534226417541504,
      "learning_rate": 8.553341215959215e-07,
      "loss": 0.5391,
      "step": 518
    },
    {
      "epoch": 2.4951923076923075,
      "grad_norm": 0.3250564634799957,
      "learning_rate": 8.397374938476594e-07,
      "loss": 0.5482,
      "step": 519
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.3450680375099182,
      "learning_rate": 8.242713309804729e-07,
      "loss": 0.5215,
      "step": 520
    },
    {
      "epoch": 2.5048076923076925,
      "grad_norm": 0.3476369380950928,
      "learning_rate": 8.089361180096927e-07,
      "loss": 0.5308,
      "step": 521
    },
    {
      "epoch": 2.5096153846153846,
      "grad_norm": 0.35948845744132996,
      "learning_rate": 7.937323358440935e-07,
      "loss": 0.5019,
      "step": 522
    },
    {
      "epoch": 2.5144230769230766,
      "grad_norm": 0.3242650628089905,
      "learning_rate": 7.786604612708093e-07,
      "loss": 0.5317,
      "step": 523
    },
    {
      "epoch": 2.519230769230769,
      "grad_norm": 0.3232422173023224,
      "learning_rate": 7.637209669403789e-07,
      "loss": 0.5165,
      "step": 524
    },
    {
      "epoch": 2.5240384615384617,
      "grad_norm": 0.32376691699028015,
      "learning_rate": 7.489143213519301e-07,
      "loss": 0.5199,
      "step": 525
    },
    {
      "epoch": 2.5288461538461537,
      "grad_norm": 0.3350430130958557,
      "learning_rate": 7.342409888384816e-07,
      "loss": 0.5087,
      "step": 526
    },
    {
      "epoch": 2.5336538461538463,
      "grad_norm": 0.3153380751609802,
      "learning_rate": 7.197014295523879e-07,
      "loss": 0.5377,
      "step": 527
    },
    {
      "epoch": 2.5384615384615383,
      "grad_norm": 0.36297881603240967,
      "learning_rate": 7.052960994509056e-07,
      "loss": 0.5238,
      "step": 528
    },
    {
      "epoch": 2.543269230769231,
      "grad_norm": 0.34906795620918274,
      "learning_rate": 6.910254502818914e-07,
      "loss": 0.5119,
      "step": 529
    },
    {
      "epoch": 2.5480769230769234,
      "grad_norm": 0.3447514474391937,
      "learning_rate": 6.768899295696413e-07,
      "loss": 0.5261,
      "step": 530
    },
    {
      "epoch": 2.5528846153846154,
      "grad_norm": 0.3420708477497101,
      "learning_rate": 6.628899806008515e-07,
      "loss": 0.5201,
      "step": 531
    },
    {
      "epoch": 2.5576923076923075,
      "grad_norm": 0.35410743951797485,
      "learning_rate": 6.490260424107231e-07,
      "loss": 0.5136,
      "step": 532
    },
    {
      "epoch": 2.5625,
      "grad_norm": 0.3542673885822296,
      "learning_rate": 6.352985497691883e-07,
      "loss": 0.5212,
      "step": 533
    },
    {
      "epoch": 2.5673076923076925,
      "grad_norm": 0.35093072056770325,
      "learning_rate": 6.217079331672777e-07,
      "loss": 0.5121,
      "step": 534
    },
    {
      "epoch": 2.5721153846153846,
      "grad_norm": 0.33335307240486145,
      "learning_rate": 6.082546188036204e-07,
      "loss": 0.5007,
      "step": 535
    },
    {
      "epoch": 2.5769230769230766,
      "grad_norm": 0.3272242546081543,
      "learning_rate": 5.949390285710777e-07,
      "loss": 0.5443,
      "step": 536
    },
    {
      "epoch": 2.581730769230769,
      "grad_norm": 0.34555184841156006,
      "learning_rate": 5.817615800435167e-07,
      "loss": 0.5255,
      "step": 537
    },
    {
      "epoch": 2.5865384615384617,
      "grad_norm": 0.3510143458843231,
      "learning_rate": 5.687226864627115e-07,
      "loss": 0.5238,
      "step": 538
    },
    {
      "epoch": 2.5913461538461537,
      "grad_norm": 0.35968631505966187,
      "learning_rate": 5.558227567253832e-07,
      "loss": 0.5186,
      "step": 539
    },
    {
      "epoch": 2.5961538461538463,
      "grad_norm": 0.35358065366744995,
      "learning_rate": 5.430621953703785e-07,
      "loss": 0.5132,
      "step": 540
    },
    {
      "epoch": 2.6009615384615383,
      "grad_norm": 0.34705501794815063,
      "learning_rate": 5.304414025659832e-07,
      "loss": 0.5418,
      "step": 541
    },
    {
      "epoch": 2.605769230769231,
      "grad_norm": 0.34814560413360596,
      "learning_rate": 5.179607740973764e-07,
      "loss": 0.5217,
      "step": 542
    },
    {
      "epoch": 2.6105769230769234,
      "grad_norm": 0.32552292943000793,
      "learning_rate": 5.056207013542131e-07,
      "loss": 0.5224,
      "step": 543
    },
    {
      "epoch": 2.6153846153846154,
      "grad_norm": 0.3452615737915039,
      "learning_rate": 4.934215713183527e-07,
      "loss": 0.5226,
      "step": 544
    },
    {
      "epoch": 2.6201923076923075,
      "grad_norm": 0.3437098264694214,
      "learning_rate": 4.813637665517251e-07,
      "loss": 0.5258,
      "step": 545
    },
    {
      "epoch": 2.625,
      "grad_norm": 0.36558979749679565,
      "learning_rate": 4.6944766518432936e-07,
      "loss": 0.5137,
      "step": 546
    },
    {
      "epoch": 2.6298076923076925,
      "grad_norm": 0.33913153409957886,
      "learning_rate": 4.576736409023813e-07,
      "loss": 0.508,
      "step": 547
    },
    {
      "epoch": 2.6346153846153846,
      "grad_norm": 0.35266298055648804,
      "learning_rate": 4.460420629365919e-07,
      "loss": 0.5167,
      "step": 548
    },
    {
      "epoch": 2.6394230769230766,
      "grad_norm": 0.3532564640045166,
      "learning_rate": 4.3455329605058436e-07,
      "loss": 0.5145,
      "step": 549
    },
    {
      "epoch": 2.644230769230769,
      "grad_norm": 0.3402523994445801,
      "learning_rate": 4.232077005294638e-07,
      "loss": 0.5237,
      "step": 550
    },
    {
      "epoch": 2.6490384615384617,
      "grad_norm": 0.3120058476924896,
      "learning_rate": 4.120056321685101e-07,
      "loss": 0.562,
      "step": 551
    },
    {
      "epoch": 2.6538461538461537,
      "grad_norm": 0.33833712339401245,
      "learning_rate": 4.009474422620269e-07,
      "loss": 0.5257,
      "step": 552
    },
    {
      "epoch": 2.6586538461538463,
      "grad_norm": 0.333404004573822,
      "learning_rate": 3.900334775923237e-07,
      "loss": 0.5115,
      "step": 553
    },
    {
      "epoch": 2.6634615384615383,
      "grad_norm": 0.32522115111351013,
      "learning_rate": 3.7926408041883355e-07,
      "loss": 0.5286,
      "step": 554
    },
    {
      "epoch": 2.668269230769231,
      "grad_norm": 0.3394847810268402,
      "learning_rate": 3.6863958846739213e-07,
      "loss": 0.5195,
      "step": 555
    },
    {
      "epoch": 2.6730769230769234,
      "grad_norm": 0.3137637674808502,
      "learning_rate": 3.581603349196372e-07,
      "loss": 0.5339,
      "step": 556
    },
    {
      "epoch": 2.6778846153846154,
      "grad_norm": 0.35012125968933105,
      "learning_rate": 3.4782664840256387e-07,
      "loss": 0.5054,
      "step": 557
    },
    {
      "epoch": 2.6826923076923075,
      "grad_norm": 0.3344334065914154,
      "learning_rate": 3.3763885297822153e-07,
      "loss": 0.5303,
      "step": 558
    },
    {
      "epoch": 2.6875,
      "grad_norm": 0.3342319428920746,
      "learning_rate": 3.275972681335421e-07,
      "loss": 0.512,
      "step": 559
    },
    {
      "epoch": 2.6923076923076925,
      "grad_norm": 0.32932013273239136,
      "learning_rate": 3.1770220877033243e-07,
      "loss": 0.5354,
      "step": 560
    },
    {
      "epoch": 2.6971153846153846,
      "grad_norm": 0.34048202633857727,
      "learning_rate": 3.0795398519539113e-07,
      "loss": 0.5216,
      "step": 561
    },
    {
      "epoch": 2.7019230769230766,
      "grad_norm": 0.3443983495235443,
      "learning_rate": 2.9835290311078123e-07,
      "loss": 0.5131,
      "step": 562
    },
    {
      "epoch": 2.706730769230769,
      "grad_norm": 0.34546706080436707,
      "learning_rate": 2.888992636042437e-07,
      "loss": 0.4995,
      "step": 563
    },
    {
      "epoch": 2.7115384615384617,
      "grad_norm": 0.32637590169906616,
      "learning_rate": 2.7959336313974847e-07,
      "loss": 0.524,
      "step": 564
    },
    {
      "epoch": 2.7163461538461537,
      "grad_norm": 0.32179778814315796,
      "learning_rate": 2.704354935482095e-07,
      "loss": 0.5369,
      "step": 565
    },
    {
      "epoch": 2.7211538461538463,
      "grad_norm": 0.3366025388240814,
      "learning_rate": 2.6142594201832183e-07,
      "loss": 0.5021,
      "step": 566
    },
    {
      "epoch": 2.7259615384615383,
      "grad_norm": 0.3441389501094818,
      "learning_rate": 2.525649910875627e-07,
      "loss": 0.522,
      "step": 567
    },
    {
      "epoch": 2.730769230769231,
      "grad_norm": 0.3519365191459656,
      "learning_rate": 2.438529186333288e-07,
      "loss": 0.5234,
      "step": 568
    },
    {
      "epoch": 2.7355769230769234,
      "grad_norm": 0.330476313829422,
      "learning_rate": 2.3528999786421758e-07,
      "loss": 0.5309,
      "step": 569
    },
    {
      "epoch": 2.7403846153846154,
      "grad_norm": 0.3406146466732025,
      "learning_rate": 2.2687649731146844e-07,
      "loss": 0.5279,
      "step": 570
    },
    {
      "epoch": 2.7451923076923075,
      "grad_norm": 0.3351764678955078,
      "learning_rate": 2.1861268082053466e-07,
      "loss": 0.5275,
      "step": 571
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.37317317724227905,
      "learning_rate": 2.104988075428127e-07,
      "loss": 0.5285,
      "step": 572
    },
    {
      "epoch": 2.7548076923076925,
      "grad_norm": 0.3595113158226013,
      "learning_rate": 2.0253513192751374e-07,
      "loss": 0.5287,
      "step": 573
    },
    {
      "epoch": 2.7596153846153846,
      "grad_norm": 0.33269202709198,
      "learning_rate": 1.947219037136827e-07,
      "loss": 0.5151,
      "step": 574
    },
    {
      "epoch": 2.7644230769230766,
      "grad_norm": 0.341153085231781,
      "learning_rate": 1.8705936792237255e-07,
      "loss": 0.5243,
      "step": 575
    },
    {
      "epoch": 2.769230769230769,
      "grad_norm": 0.3603745102882385,
      "learning_rate": 1.7954776484895188e-07,
      "loss": 0.5228,
      "step": 576
    },
    {
      "epoch": 2.7740384615384617,
      "grad_norm": 0.3396835923194885,
      "learning_rate": 1.7218733005557707e-07,
      "loss": 0.5073,
      "step": 577
    },
    {
      "epoch": 2.7788461538461537,
      "grad_norm": 0.3207492530345917,
      "learning_rate": 1.6497829436380009e-07,
      "loss": 0.5266,
      "step": 578
    },
    {
      "epoch": 2.7836538461538463,
      "grad_norm": 0.3595437705516815,
      "learning_rate": 1.5792088384733174e-07,
      "loss": 0.5045,
      "step": 579
    },
    {
      "epoch": 2.7884615384615383,
      "grad_norm": 0.3366645574569702,
      "learning_rate": 1.510153198249531e-07,
      "loss": 0.5434,
      "step": 580
    },
    {
      "epoch": 2.793269230769231,
      "grad_norm": 0.3330222964286804,
      "learning_rate": 1.4426181885357215e-07,
      "loss": 0.5246,
      "step": 581
    },
    {
      "epoch": 2.7980769230769234,
      "grad_norm": 0.3313714861869812,
      "learning_rate": 1.376605927214364e-07,
      "loss": 0.5196,
      "step": 582
    },
    {
      "epoch": 2.8028846153846154,
      "grad_norm": 0.3222265839576721,
      "learning_rate": 1.312118484414876e-07,
      "loss": 0.5153,
      "step": 583
    },
    {
      "epoch": 2.8076923076923075,
      "grad_norm": 0.3020241856575012,
      "learning_rate": 1.2491578824487204e-07,
      "loss": 0.5489,
      "step": 584
    },
    {
      "epoch": 2.8125,
      "grad_norm": 0.3178803026676178,
      "learning_rate": 1.1877260957459835e-07,
      "loss": 0.5352,
      "step": 585
    },
    {
      "epoch": 2.8173076923076925,
      "grad_norm": 0.3120872676372528,
      "learning_rate": 1.1278250507934518e-07,
      "loss": 0.5294,
      "step": 586
    },
    {
      "epoch": 2.8221153846153846,
      "grad_norm": 0.3329028785228729,
      "learning_rate": 1.0694566260742001e-07,
      "loss": 0.5307,
      "step": 587
    },
    {
      "epoch": 2.8269230769230766,
      "grad_norm": 0.3385623097419739,
      "learning_rate": 1.0126226520086823e-07,
      "loss": 0.524,
      "step": 588
    },
    {
      "epoch": 2.831730769230769,
      "grad_norm": 0.3274289071559906,
      "learning_rate": 9.573249108973281e-08,
      "loss": 0.5196,
      "step": 589
    },
    {
      "epoch": 2.8365384615384617,
      "grad_norm": 0.31717026233673096,
      "learning_rate": 9.035651368646647e-08,
      "loss": 0.5163,
      "step": 590
    },
    {
      "epoch": 2.8413461538461537,
      "grad_norm": 0.3221368193626404,
      "learning_rate": 8.513450158049109e-08,
      "loss": 0.5256,
      "step": 591
    },
    {
      "epoch": 2.8461538461538463,
      "grad_norm": 0.3442428708076477,
      "learning_rate": 8.006661853291298e-08,
      "loss": 0.5197,
      "step": 592
    },
    {
      "epoch": 2.8509615384615383,
      "grad_norm": 0.33662882447242737,
      "learning_rate": 7.515302347138486e-08,
      "loss": 0.5188,
      "step": 593
    },
    {
      "epoch": 2.855769230769231,
      "grad_norm": 0.33708903193473816,
      "learning_rate": 7.03938704851248e-08,
      "loss": 0.5268,
      "step": 594
    },
    {
      "epoch": 2.8605769230769234,
      "grad_norm": 0.3152363896369934,
      "learning_rate": 6.578930882008283e-08,
      "loss": 0.5295,
      "step": 595
    },
    {
      "epoch": 2.8653846153846154,
      "grad_norm": 0.32401999831199646,
      "learning_rate": 6.133948287426028e-08,
      "loss": 0.511,
      "step": 596
    },
    {
      "epoch": 2.8701923076923075,
      "grad_norm": 0.32977667450904846,
      "learning_rate": 5.704453219318118e-08,
      "loss": 0.514,
      "step": 597
    },
    {
      "epoch": 2.875,
      "grad_norm": 0.32670727372169495,
      "learning_rate": 5.2904591465516855e-08,
      "loss": 0.514,
      "step": 598
    },
    {
      "epoch": 2.8798076923076925,
      "grad_norm": 0.32648882269859314,
      "learning_rate": 4.891979051886153e-08,
      "loss": 0.5181,
      "step": 599
    },
    {
      "epoch": 2.8846153846153846,
      "grad_norm": 0.324532687664032,
      "learning_rate": 4.509025431566283e-08,
      "loss": 0.5221,
      "step": 600
    },
    {
      "epoch": 2.8894230769230766,
      "grad_norm": 0.31154200434684753,
      "learning_rate": 4.141610294930043e-08,
      "loss": 0.5212,
      "step": 601
    },
    {
      "epoch": 2.894230769230769,
      "grad_norm": 0.3022911250591278,
      "learning_rate": 3.7897451640321326e-08,
      "loss": 0.5227,
      "step": 602
    },
    {
      "epoch": 2.8990384615384617,
      "grad_norm": 0.3582252860069275,
      "learning_rate": 3.4534410732825485e-08,
      "loss": 0.5237,
      "step": 603
    },
    {
      "epoch": 2.9038461538461537,
      "grad_norm": 0.3342374265193939,
      "learning_rate": 3.1327085691006954e-08,
      "loss": 0.5086,
      "step": 604
    },
    {
      "epoch": 2.9086538461538463,
      "grad_norm": 0.35545840859413147,
      "learning_rate": 2.8275577095846495e-08,
      "loss": 0.5204,
      "step": 605
    },
    {
      "epoch": 2.9134615384615383,
      "grad_norm": 0.37066027522087097,
      "learning_rate": 2.5379980641955792e-08,
      "loss": 0.519,
      "step": 606
    },
    {
      "epoch": 2.918269230769231,
      "grad_norm": 0.3329092860221863,
      "learning_rate": 2.264038713457706e-08,
      "loss": 0.5424,
      "step": 607
    },
    {
      "epoch": 2.9230769230769234,
      "grad_norm": 0.34132009744644165,
      "learning_rate": 2.0056882486736982e-08,
      "loss": 0.5209,
      "step": 608
    },
    {
      "epoch": 2.9278846153846154,
      "grad_norm": 0.32534360885620117,
      "learning_rate": 1.762954771655001e-08,
      "loss": 0.5298,
      "step": 609
    },
    {
      "epoch": 2.9326923076923075,
      "grad_norm": 0.3154315948486328,
      "learning_rate": 1.5358458944680356e-08,
      "loss": 0.5075,
      "step": 610
    },
    {
      "epoch": 2.9375,
      "grad_norm": 0.3233172595500946,
      "learning_rate": 1.3243687391952809e-08,
      "loss": 0.5264,
      "step": 611
    },
    {
      "epoch": 2.9423076923076925,
      "grad_norm": 0.3091405928134918,
      "learning_rate": 1.1285299377118974e-08,
      "loss": 0.5431,
      "step": 612
    },
    {
      "epoch": 2.9471153846153846,
      "grad_norm": 0.32396966218948364,
      "learning_rate": 9.48335631477948e-09,
      "loss": 0.5235,
      "step": 613
    },
    {
      "epoch": 2.9519230769230766,
      "grad_norm": 0.3247375786304474,
      "learning_rate": 7.837914713457184e-09,
      "loss": 0.53,
      "step": 614
    },
    {
      "epoch": 2.956730769230769,
      "grad_norm": 0.31406140327453613,
      "learning_rate": 6.349026173824713e-09,
      "loss": 0.517,
      "step": 615
    },
    {
      "epoch": 2.9615384615384617,
      "grad_norm": 0.3556351661682129,
      "learning_rate": 5.016737387085191e-09,
      "loss": 0.5218,
      "step": 616
    },
    {
      "epoch": 2.9663461538461537,
      "grad_norm": 0.34563007950782776,
      "learning_rate": 3.841090133511749e-09,
      "loss": 0.5199,
      "step": 617
    },
    {
      "epoch": 2.9711538461538463,
      "grad_norm": 0.33256563544273376,
      "learning_rate": 2.8221212811324616e-09,
      "loss": 0.511,
      "step": 618
    },
    {
      "epoch": 2.9759615384615383,
      "grad_norm": 0.32485440373420715,
      "learning_rate": 1.959862784577937e-09,
      "loss": 0.5198,
      "step": 619
    },
    {
      "epoch": 2.980769230769231,
      "grad_norm": 0.30697503685951233,
      "learning_rate": 1.2543416840771206e-09,
      "loss": 0.5484,
      "step": 620
    },
    {
      "epoch": 2.9855769230769234,
      "grad_norm": 0.3317164182662964,
      "learning_rate": 7.055801046113031e-10,
      "loss": 0.5138,
      "step": 621
    },
    {
      "epoch": 2.9903846153846154,
      "grad_norm": 0.3181401789188385,
      "learning_rate": 3.1359525521801326e-10,
      "loss": 0.5243,
      "step": 622
    },
    {
      "epoch": 2.9951923076923075,
      "grad_norm": 0.32515648007392883,
      "learning_rate": 7.839942845144777e-11,
      "loss": 0.5262,
      "step": 623
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.3372482359409332,
      "learning_rate": 0.0,
      "loss": 0.5101,
      "step": 624
    },
    {
      "epoch": 3.0,
      "step": 624,
      "total_flos": 1149901297156096.0,
      "train_loss": 0.5865381705837373,
      "train_runtime": 8382.2551,
      "train_samples_per_second": 7.124,
      "train_steps_per_second": 0.074
    }
  ],
  "logging_steps": 1,
  "max_steps": 624,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1149901297156096.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}