{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9999223300970874,
  "eval_steps": 500,
  "global_step": 6437,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0001553398058252427,
      "grad_norm": 0.0,
      "learning_rate": 5.154639175257732e-06,
      "loss": 8.1968,
      "step": 1
    },
    {
      "epoch": 0.0003106796116504854,
      "grad_norm": 0.0,
      "learning_rate": 1.0309278350515464e-05,
      "loss": 8.2563,
      "step": 2
    },
    {
      "epoch": 0.00046601941747572817,
      "grad_norm": 0.0,
      "learning_rate": 1.5463917525773197e-05,
      "loss": 8.1035,
      "step": 3
    },
    {
      "epoch": 0.0006213592233009709,
      "grad_norm": 0.0,
      "learning_rate": 2.0618556701030927e-05,
      "loss": 7.4243,
      "step": 4
    },
    {
      "epoch": 0.0007766990291262136,
      "grad_norm": 0.0,
      "learning_rate": 2.5773195876288658e-05,
      "loss": 6.606,
      "step": 5
    },
    {
      "epoch": 0.0009320388349514563,
      "grad_norm": 0.0,
      "learning_rate": 3.0927835051546395e-05,
      "loss": 6.2373,
      "step": 6
    },
    {
      "epoch": 0.001087378640776699,
      "grad_norm": 0.0,
      "learning_rate": 3.608247422680412e-05,
      "loss": 5.8862,
      "step": 7
    },
    {
      "epoch": 0.0012427184466019417,
      "grad_norm": 0.0,
      "learning_rate": 4.1237113402061855e-05,
      "loss": 5.4849,
      "step": 8
    },
    {
      "epoch": 0.0013980582524271844,
      "grad_norm": 0.0,
      "learning_rate": 4.639175257731959e-05,
      "loss": 5.064,
      "step": 9
    },
    {
      "epoch": 0.0015533980582524273,
      "grad_norm": 0.0,
      "learning_rate": 5.1546391752577315e-05,
      "loss": 4.8291,
      "step": 10
    },
    {
      "epoch": 0.00170873786407767,
      "grad_norm": 0.0,
      "learning_rate": 5.670103092783505e-05,
      "loss": 4.6116,
      "step": 11
    },
    {
      "epoch": 0.0018640776699029127,
      "grad_norm": 0.0,
      "learning_rate": 6.185567010309279e-05,
      "loss": 4.4705,
      "step": 12
    },
    {
      "epoch": 0.002019417475728155,
      "grad_norm": 0.0,
      "learning_rate": 6.701030927835051e-05,
      "loss": 4.3416,
      "step": 13
    },
    {
      "epoch": 0.002174757281553398,
      "grad_norm": 0.0,
      "learning_rate": 7.216494845360824e-05,
      "loss": 4.1985,
      "step": 14
    },
    {
      "epoch": 0.002330097087378641,
      "grad_norm": 0.0,
      "learning_rate": 7.731958762886599e-05,
      "loss": 4.1763,
      "step": 15
    },
    {
      "epoch": 0.0024854368932038834,
      "grad_norm": 0.0,
      "learning_rate": 8.247422680412371e-05,
      "loss": 4.092,
      "step": 16
    },
    {
      "epoch": 0.0026407766990291263,
      "grad_norm": 0.0,
      "learning_rate": 8.762886597938143e-05,
      "loss": 3.9919,
      "step": 17
    },
    {
      "epoch": 0.0027961165048543688,
      "grad_norm": 0.0,
      "learning_rate": 9.278350515463918e-05,
      "loss": 4.0205,
      "step": 18
    },
    {
      "epoch": 0.0029514563106796117,
      "grad_norm": 0.0,
      "learning_rate": 9.793814432989691e-05,
      "loss": 3.9963,
      "step": 19
    },
    {
      "epoch": 0.0031067961165048546,
      "grad_norm": 0.0,
      "learning_rate": 0.00010309278350515463,
      "loss": 3.9705,
      "step": 20
    },
    {
      "epoch": 0.003262135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.00010824742268041238,
      "loss": 3.9255,
      "step": 21
    },
    {
      "epoch": 0.00341747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0001134020618556701,
      "loss": 3.9651,
      "step": 22
    },
    {
      "epoch": 0.0035728155339805824,
      "grad_norm": 0.0,
      "learning_rate": 0.00011855670103092783,
      "loss": 3.8562,
      "step": 23
    },
    {
      "epoch": 0.0037281553398058253,
      "grad_norm": 0.0,
      "learning_rate": 0.00012371134020618558,
      "loss": 3.8276,
      "step": 24
    },
    {
      "epoch": 0.003883495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.00012886597938144327,
      "loss": 3.7881,
      "step": 25
    },
    {
      "epoch": 0.00403883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.00013402061855670103,
      "loss": 3.7969,
      "step": 26
    },
    {
      "epoch": 0.004194174757281554,
      "grad_norm": 0.0,
      "learning_rate": 0.00013917525773195878,
      "loss": 3.7642,
      "step": 27
    },
    {
      "epoch": 0.004349514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.00014432989690721647,
      "loss": 3.675,
      "step": 28
    },
    {
      "epoch": 0.0045048543689320385,
      "grad_norm": 0.0,
      "learning_rate": 0.00014948453608247422,
      "loss": 3.7158,
      "step": 29
    },
    {
      "epoch": 0.004660194174757282,
      "grad_norm": 0.0,
      "learning_rate": 0.00015463917525773197,
      "loss": 3.7356,
      "step": 30
    },
    {
      "epoch": 0.004815533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.00015979381443298967,
      "loss": 3.7419,
      "step": 31
    },
    {
      "epoch": 0.004970873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.00016494845360824742,
      "loss": 3.5962,
      "step": 32
    },
    {
      "epoch": 0.005126213592233009,
      "grad_norm": 0.0,
      "learning_rate": 0.00017010309278350517,
      "loss": 3.687,
      "step": 33
    },
    {
      "epoch": 0.005281553398058253,
      "grad_norm": 0.0,
      "learning_rate": 0.00017525773195876287,
      "loss": 3.6255,
      "step": 34
    },
    {
      "epoch": 0.005436893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.00018041237113402062,
      "loss": 3.6187,
      "step": 35
    },
    {
      "epoch": 0.0055922330097087376,
      "grad_norm": 0.0,
      "learning_rate": 0.00018556701030927837,
      "loss": 3.667,
      "step": 36
    },
    {
      "epoch": 0.005747572815533981,
      "grad_norm": 0.0,
      "learning_rate": 0.00019072164948453606,
      "loss": 3.5139,
      "step": 37
    },
    {
      "epoch": 0.005902912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.00019587628865979381,
      "loss": 3.4663,
      "step": 38
    },
    {
      "epoch": 0.006058252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.00020103092783505157,
      "loss": 3.6233,
      "step": 39
    },
    {
      "epoch": 0.006213592233009709,
      "grad_norm": 0.0,
      "learning_rate": 0.00020618556701030926,
      "loss": 3.4338,
      "step": 40
    },
    {
      "epoch": 0.006368932038834952,
      "grad_norm": 0.0,
      "learning_rate": 0.000211340206185567,
      "loss": 3.4768,
      "step": 41
    },
    {
      "epoch": 0.006524271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.00021649484536082476,
      "loss": 3.4473,
      "step": 42
    },
    {
      "epoch": 0.006679611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.00022164948453608246,
      "loss": 3.499,
      "step": 43
    },
    {
      "epoch": 0.00683495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0002268041237113402,
      "loss": 3.4119,
      "step": 44
    },
    {
      "epoch": 0.006990291262135922,
      "grad_norm": 0.0,
      "learning_rate": 0.00023195876288659796,
      "loss": 3.3789,
      "step": 45
    },
    {
      "epoch": 0.007145631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.00023711340206185566,
      "loss": 3.4819,
      "step": 46
    },
    {
      "epoch": 0.007300970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.0002422680412371134,
      "loss": 3.3887,
      "step": 47
    },
    {
      "epoch": 0.007456310679611651,
      "grad_norm": 0.0,
      "learning_rate": 0.00024742268041237116,
      "loss": 3.3013,
      "step": 48
    },
    {
      "epoch": 0.007611650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.00025257731958762885,
      "loss": 3.4338,
      "step": 49
    },
    {
      "epoch": 0.007766990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.00025773195876288655,
      "loss": 3.2969,
      "step": 50
    },
    {
      "epoch": 0.007922330097087378,
      "grad_norm": 0.0,
      "learning_rate": 0.00026288659793814435,
      "loss": 3.4441,
      "step": 51
    },
    {
      "epoch": 0.00807766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.00026804123711340205,
      "loss": 3.3716,
      "step": 52
    },
    {
      "epoch": 0.008233009708737865,
      "grad_norm": 0.0,
      "learning_rate": 0.00027319587628865975,
      "loss": 3.3181,
      "step": 53
    },
    {
      "epoch": 0.008388349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.00027835051546391755,
      "loss": 3.4175,
      "step": 54
    },
    {
      "epoch": 0.00854368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.00028350515463917525,
      "loss": 3.2531,
      "step": 55
    },
    {
      "epoch": 0.008699029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.00028865979381443294,
      "loss": 3.3149,
      "step": 56
    },
    {
      "epoch": 0.008854368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.00029381443298969075,
      "loss": 3.302,
      "step": 57
    },
    {
      "epoch": 0.009009708737864077,
      "grad_norm": 0.0,
      "learning_rate": 0.00029896907216494845,
      "loss": 3.3584,
      "step": 58
    },
    {
      "epoch": 0.00916504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.00030412371134020614,
      "loss": 3.4529,
      "step": 59
    },
    {
      "epoch": 0.009320388349514564,
      "grad_norm": 0.0,
      "learning_rate": 0.00030927835051546395,
      "loss": 3.239,
      "step": 60
    },
    {
      "epoch": 0.009475728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.00031443298969072164,
      "loss": 3.3191,
      "step": 61
    },
    {
      "epoch": 0.009631067961165049,
      "grad_norm": 0.0,
      "learning_rate": 0.00031958762886597934,
      "loss": 3.3235,
      "step": 62
    },
    {
      "epoch": 0.009786407766990291,
      "grad_norm": 0.0,
      "learning_rate": 0.00032474226804123714,
      "loss": 3.3047,
      "step": 63
    },
    {
      "epoch": 0.009941747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.00032989690721649484,
      "loss": 3.2771,
      "step": 64
    },
    {
      "epoch": 0.010097087378640776,
      "grad_norm": 0.0,
      "learning_rate": 0.00033505154639175254,
      "loss": 3.2759,
      "step": 65
    },
    {
      "epoch": 0.010252427184466019,
      "grad_norm": 0.0,
      "learning_rate": 0.00034020618556701034,
      "loss": 3.3188,
      "step": 66
    },
    {
      "epoch": 0.010407766990291263,
      "grad_norm": 0.0,
      "learning_rate": 0.00034536082474226804,
      "loss": 3.1436,
      "step": 67
    },
    {
      "epoch": 0.010563106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.00035051546391752573,
      "loss": 3.2307,
      "step": 68
    },
    {
      "epoch": 0.010718446601941748,
      "grad_norm": 0.0,
      "learning_rate": 0.00035567010309278354,
      "loss": 3.3345,
      "step": 69
    },
    {
      "epoch": 0.01087378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.00036082474226804123,
      "loss": 3.1509,
      "step": 70
    },
    {
      "epoch": 0.011029126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.00036597938144329893,
      "loss": 3.2898,
      "step": 71
    },
    {
      "epoch": 0.011184466019417475,
      "grad_norm": 0.0,
      "learning_rate": 0.00037113402061855674,
      "loss": 3.2815,
      "step": 72
    },
    {
      "epoch": 0.011339805825242718,
      "grad_norm": 0.0,
      "learning_rate": 0.00037628865979381443,
      "loss": 3.2271,
      "step": 73
    },
    {
      "epoch": 0.011495145631067962,
      "grad_norm": 0.0,
      "learning_rate": 0.00038144329896907213,
      "loss": 3.134,
      "step": 74
    },
    {
      "epoch": 0.011650485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.00038659793814432993,
      "loss": 3.2234,
      "step": 75
    },
    {
      "epoch": 0.011805825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.00039175257731958763,
      "loss": 3.2588,
      "step": 76
    },
    {
      "epoch": 0.01196116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0003969072164948453,
      "loss": 3.1765,
      "step": 77
    },
    {
      "epoch": 0.012116504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.00040206185567010313,
      "loss": 3.0955,
      "step": 78
    },
    {
      "epoch": 0.012271844660194174,
      "grad_norm": 0.0,
      "learning_rate": 0.0004072164948453608,
      "loss": 3.1746,
      "step": 79
    },
    {
      "epoch": 0.012427184466019418,
      "grad_norm": 0.0,
      "learning_rate": 0.0004123711340206185,
      "loss": 3.3232,
      "step": 80
    },
    {
      "epoch": 0.01258252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.00041752577319587633,
      "loss": 3.0793,
      "step": 81
    },
    {
      "epoch": 0.012737864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.000422680412371134,
      "loss": 3.2744,
      "step": 82
    },
    {
      "epoch": 0.012893203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.0004278350515463917,
      "loss": 3.2646,
      "step": 83
    },
    {
      "epoch": 0.013048543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.0004329896907216495,
      "loss": 3.2283,
      "step": 84
    },
    {
      "epoch": 0.01320388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0004381443298969072,
      "loss": 3.2275,
      "step": 85
    },
    {
      "epoch": 0.013359223300970873,
      "grad_norm": 0.0,
      "learning_rate": 0.0004432989690721649,
      "loss": 3.1519,
      "step": 86
    },
    {
      "epoch": 0.013514563106796117,
      "grad_norm": 0.0,
      "learning_rate": 0.0004484536082474227,
      "loss": 3.1775,
      "step": 87
    },
    {
      "epoch": 0.01366990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0004536082474226804,
      "loss": 3.1711,
      "step": 88
    },
    {
      "epoch": 0.013825242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0004587628865979381,
      "loss": 3.2131,
      "step": 89
    },
    {
      "epoch": 0.013980582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.0004639175257731959,
      "loss": 3.125,
      "step": 90
    },
    {
      "epoch": 0.014135922330097087,
      "grad_norm": 0.0,
      "learning_rate": 0.0004690721649484536,
      "loss": 3.1724,
      "step": 91
    },
    {
      "epoch": 0.01429126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0004742268041237113,
      "loss": 3.1704,
      "step": 92
    },
    {
      "epoch": 0.014446601941747572,
      "grad_norm": 0.0,
      "learning_rate": 0.0004793814432989691,
      "loss": 3.1956,
      "step": 93
    },
    {
      "epoch": 0.014601941747572816,
      "grad_norm": 0.0,
      "learning_rate": 0.0004845360824742268,
      "loss": 3.105,
      "step": 94
    },
    {
      "epoch": 0.014757281553398059,
      "grad_norm": 0.0,
      "learning_rate": 0.0004896907216494845,
      "loss": 3.2577,
      "step": 95
    },
    {
      "epoch": 0.014912621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0004948453608247423,
      "loss": 3.2585,
      "step": 96
    },
    {
      "epoch": 0.015067961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.0005,
      "loss": 3.2004,
      "step": 97
    },
    {
      "epoch": 0.015223300970873786,
      "grad_norm": 0.0,
      "learning_rate": 0.0005051546391752577,
      "loss": 3.0686,
      "step": 98
    },
    {
      "epoch": 0.015378640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.0005103092783505154,
      "loss": 3.1707,
      "step": 99
    },
    {
      "epoch": 0.015533980582524271,
      "grad_norm": 0.0,
      "learning_rate": 0.0005154639175257731,
      "loss": 3.1514,
      "step": 100
    },
    {
      "epoch": 0.015689320388349515,
      "grad_norm": 0.0,
      "learning_rate": 0.000520618556701031,
      "loss": 3.115,
      "step": 101
    },
    {
      "epoch": 0.015844660194174756,
      "grad_norm": 0.0,
      "learning_rate": 0.0005257731958762887,
      "loss": 3.0322,
      "step": 102
    },
    {
      "epoch": 0.016,
      "grad_norm": 0.0,
      "learning_rate": 0.0005309278350515464,
      "loss": 3.1333,
      "step": 103
    },
    {
      "epoch": 0.01615533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.0005360824742268041,
      "loss": 3.1316,
      "step": 104
    },
    {
      "epoch": 0.016310679611650485,
      "grad_norm": 0.0,
      "learning_rate": 0.0005412371134020618,
      "loss": 3.1309,
      "step": 105
    },
    {
      "epoch": 0.01646601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.0005463917525773195,
      "loss": 2.9719,
      "step": 106
    },
    {
      "epoch": 0.01662135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0005515463917525774,
      "loss": 3.019,
      "step": 107
    },
    {
      "epoch": 0.016776699029126214,
      "grad_norm": 0.0,
      "learning_rate": 0.0005567010309278351,
      "loss": 3.1243,
      "step": 108
    },
    {
      "epoch": 0.016932038834951455,
      "grad_norm": 0.0,
      "learning_rate": 0.0005618556701030928,
      "loss": 3.1033,
      "step": 109
    },
    {
      "epoch": 0.0170873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0005670103092783505,
      "loss": 3.0631,
      "step": 110
    },
    {
      "epoch": 0.01724271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.0005721649484536082,
      "loss": 3.0643,
      "step": 111
    },
    {
      "epoch": 0.017398058252427184,
      "grad_norm": 0.0,
      "learning_rate": 0.0005773195876288659,
      "loss": 3.0256,
      "step": 112
    },
    {
      "epoch": 0.01755339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0005824742268041238,
      "loss": 3.0374,
      "step": 113
    },
    {
      "epoch": 0.01770873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0005876288659793815,
      "loss": 3.0608,
      "step": 114
    },
    {
      "epoch": 0.017864077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.0005927835051546392,
      "loss": 3.05,
      "step": 115
    },
    {
      "epoch": 0.018019417475728154,
      "grad_norm": 0.0,
      "learning_rate": 0.0005979381443298969,
      "loss": 2.905,
      "step": 116
    },
    {
      "epoch": 0.0181747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0006030927835051546,
      "loss": 2.941,
      "step": 117
    },
    {
      "epoch": 0.01833009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0006082474226804123,
      "loss": 3.0063,
      "step": 118
    },
    {
      "epoch": 0.018485436893203883,
      "grad_norm": 0.0,
      "learning_rate": 0.0006134020618556702,
      "loss": 2.9348,
      "step": 119
    },
    {
      "epoch": 0.018640776699029128,
      "grad_norm": 0.0,
      "learning_rate": 0.0006185567010309279,
      "loss": 3.0037,
      "step": 120
    },
    {
      "epoch": 0.018796116504854368,
      "grad_norm": 0.0,
      "learning_rate": 0.0006237113402061856,
      "loss": 3.0195,
      "step": 121
    },
    {
      "epoch": 0.018951456310679612,
      "grad_norm": 0.0,
      "learning_rate": 0.0006288659793814433,
      "loss": 2.9628,
      "step": 122
    },
    {
      "epoch": 0.019106796116504853,
      "grad_norm": 0.0,
      "learning_rate": 0.000634020618556701,
      "loss": 3.0339,
      "step": 123
    },
    {
      "epoch": 0.019262135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0006391752577319587,
      "loss": 2.9731,
      "step": 124
    },
    {
      "epoch": 0.019417475728155338,
      "grad_norm": 0.0,
      "learning_rate": 0.0006443298969072166,
      "loss": 2.9984,
      "step": 125
    },
    {
      "epoch": 0.019572815533980582,
      "grad_norm": 0.0,
      "learning_rate": 0.0006494845360824743,
      "loss": 2.949,
      "step": 126
    },
    {
      "epoch": 0.019728155339805827,
      "grad_norm": 0.0,
      "learning_rate": 0.000654639175257732,
      "loss": 2.9854,
      "step": 127
    },
    {
      "epoch": 0.019883495145631067,
      "grad_norm": 0.0,
      "learning_rate": 0.0006597938144329897,
      "loss": 3.0273,
      "step": 128
    },
    {
      "epoch": 0.02003883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0006649484536082474,
      "loss": 3.0984,
      "step": 129
    },
    {
      "epoch": 0.020194174757281552,
      "grad_norm": 0.0,
      "learning_rate": 0.0006701030927835051,
      "loss": 2.942,
      "step": 130
    },
    {
      "epoch": 0.020349514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.000675257731958763,
      "loss": 3.0686,
      "step": 131
    },
    {
      "epoch": 0.020504854368932037,
      "grad_norm": 0.0,
      "learning_rate": 0.0006804123711340207,
      "loss": 2.9308,
      "step": 132
    },
    {
      "epoch": 0.02066019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.0006855670103092784,
      "loss": 2.9592,
      "step": 133
    },
    {
      "epoch": 0.020815533980582526,
      "grad_norm": 0.0,
      "learning_rate": 0.0006907216494845361,
      "loss": 2.9316,
      "step": 134
    },
    {
      "epoch": 0.020970873786407766,
      "grad_norm": 0.0,
      "learning_rate": 0.0006958762886597938,
      "loss": 3.0466,
      "step": 135
    },
    {
      "epoch": 0.02112621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0007010309278350515,
      "loss": 2.9431,
      "step": 136
    },
    {
      "epoch": 0.02128155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.0007061855670103094,
      "loss": 2.9712,
      "step": 137
    },
    {
      "epoch": 0.021436893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0007113402061855671,
      "loss": 2.9509,
      "step": 138
    },
    {
      "epoch": 0.021592233009708736,
      "grad_norm": 0.0,
      "learning_rate": 0.0007164948453608248,
      "loss": 2.9028,
      "step": 139
    },
    {
      "epoch": 0.02174757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0007216494845360825,
      "loss": 2.9377,
      "step": 140
    },
    {
      "epoch": 0.021902912621359225,
      "grad_norm": 0.0,
      "learning_rate": 0.0007268041237113402,
      "loss": 2.8905,
      "step": 141
    },
    {
      "epoch": 0.022058252427184465,
      "grad_norm": 0.0,
      "learning_rate": 0.0007319587628865979,
      "loss": 2.8621,
      "step": 142
    },
    {
      "epoch": 0.02221359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0007371134020618558,
      "loss": 2.8909,
      "step": 143
    },
    {
      "epoch": 0.02236893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0007422680412371135,
      "loss": 3.0339,
      "step": 144
    },
    {
      "epoch": 0.022524271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.0007474226804123712,
      "loss": 3.0204,
      "step": 145
    },
    {
      "epoch": 0.022679611650485435,
      "grad_norm": 0.0,
      "learning_rate": 0.0007525773195876289,
      "loss": 2.9089,
      "step": 146
    },
    {
      "epoch": 0.02283495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0007577319587628866,
      "loss": 2.9014,
      "step": 147
    },
    {
      "epoch": 0.022990291262135924,
      "grad_norm": 0.0,
      "learning_rate": 0.0007628865979381443,
      "loss": 3.0449,
      "step": 148
    },
    {
      "epoch": 0.023145631067961164,
      "grad_norm": 0.0,
      "learning_rate": 0.0007680412371134022,
      "loss": 2.8558,
      "step": 149
    },
    {
      "epoch": 0.02330097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.0007731958762886599,
      "loss": 2.8488,
      "step": 150
    },
    {
      "epoch": 0.02345631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0007783505154639176,
      "loss": 2.9097,
      "step": 151
    },
    {
      "epoch": 0.023611650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.0007835051546391753,
      "loss": 2.9692,
      "step": 152
    },
    {
      "epoch": 0.023766990291262134,
      "grad_norm": 0.0,
      "learning_rate": 0.000788659793814433,
      "loss": 2.8451,
      "step": 153
    },
    {
      "epoch": 0.02392233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.0007938144329896907,
      "loss": 2.907,
      "step": 154
    },
    {
      "epoch": 0.024077669902912623,
      "grad_norm": 0.0,
      "learning_rate": 0.0007989690721649486,
      "loss": 2.8838,
      "step": 155
    },
    {
      "epoch": 0.024233009708737863,
      "grad_norm": 0.0,
      "learning_rate": 0.0008041237113402063,
      "loss": 2.8906,
      "step": 156
    },
    {
      "epoch": 0.024388349514563108,
      "grad_norm": 0.0,
      "learning_rate": 0.000809278350515464,
      "loss": 2.7881,
      "step": 157
    },
    {
      "epoch": 0.02454368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0008144329896907217,
      "loss": 2.936,
      "step": 158
    },
    {
      "epoch": 0.024699029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.0008195876288659793,
      "loss": 2.9169,
      "step": 159
    },
    {
      "epoch": 0.024854368932038837,
      "grad_norm": 0.0,
      "learning_rate": 0.000824742268041237,
      "loss": 2.9565,
      "step": 160
    },
    {
      "epoch": 0.025009708737864077,
      "grad_norm": 0.0,
      "learning_rate": 0.000829896907216495,
      "loss": 2.8949,
      "step": 161
    },
    {
      "epoch": 0.02516504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0008350515463917527,
      "loss": 2.922,
      "step": 162
    },
    {
      "epoch": 0.025320388349514562,
      "grad_norm": 0.0,
      "learning_rate": 0.0008402061855670104,
      "loss": 2.8656,
      "step": 163
    },
    {
      "epoch": 0.025475728155339807,
      "grad_norm": 0.0,
      "learning_rate": 0.000845360824742268,
      "loss": 2.8855,
      "step": 164
    },
    {
      "epoch": 0.025631067961165047,
      "grad_norm": 0.0,
      "learning_rate": 0.0008505154639175257,
      "loss": 2.8689,
      "step": 165
    },
    {
      "epoch": 0.02578640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.0008556701030927834,
      "loss": 2.8794,
      "step": 166
    },
    {
      "epoch": 0.025941747572815536,
      "grad_norm": 0.0,
      "learning_rate": 0.0008608247422680414,
      "loss": 2.8406,
      "step": 167
    },
    {
      "epoch": 0.026097087378640776,
      "grad_norm": 0.0,
      "learning_rate": 0.000865979381443299,
      "loss": 2.8647,
      "step": 168
    },
    {
      "epoch": 0.02625242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0008711340206185567,
      "loss": 2.9595,
      "step": 169
    },
    {
      "epoch": 0.02640776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.0008762886597938144,
      "loss": 2.7693,
      "step": 170
    },
    {
      "epoch": 0.026563106796116506,
      "grad_norm": 0.0,
      "learning_rate": 0.0008814432989690721,
      "loss": 2.9348,
      "step": 171
    },
    {
      "epoch": 0.026718446601941746,
      "grad_norm": 0.0,
      "learning_rate": 0.0008865979381443298,
      "loss": 2.8423,
      "step": 172
    },
    {
      "epoch": 0.02687378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0008917525773195877,
      "loss": 2.8882,
      "step": 173
    },
    {
      "epoch": 0.027029126213592235,
      "grad_norm": 0.0,
      "learning_rate": 0.0008969072164948454,
      "loss": 2.8113,
      "step": 174
    },
    {
      "epoch": 0.027184466019417475,
      "grad_norm": 0.0,
      "learning_rate": 0.0009020618556701031,
      "loss": 2.8497,
      "step": 175
    },
    {
      "epoch": 0.02733980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0009072164948453608,
      "loss": 2.9445,
      "step": 176
    },
    {
      "epoch": 0.02749514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0009123711340206185,
      "loss": 2.9061,
      "step": 177
    },
    {
      "epoch": 0.027650485436893205,
      "grad_norm": 0.0,
      "learning_rate": 0.0009175257731958762,
      "loss": 2.9902,
      "step": 178
    },
    {
      "epoch": 0.027805825242718445,
      "grad_norm": 0.0,
      "learning_rate": 0.0009226804123711341,
      "loss": 2.8208,
      "step": 179
    },
    {
      "epoch": 0.02796116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0009278350515463918,
      "loss": 2.8523,
      "step": 180
    },
    {
      "epoch": 0.028116504854368934,
      "grad_norm": 0.0,
      "learning_rate": 0.0009329896907216495,
      "loss": 2.8561,
      "step": 181
    },
    {
      "epoch": 0.028271844660194174,
      "grad_norm": 0.0,
      "learning_rate": 0.0009381443298969072,
      "loss": 2.8162,
      "step": 182
    },
    {
      "epoch": 0.02842718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.0009432989690721649,
      "loss": 2.845,
      "step": 183
    },
    {
      "epoch": 0.02858252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0009484536082474226,
      "loss": 2.719,
      "step": 184
    },
    {
      "epoch": 0.028737864077669904,
      "grad_norm": 0.0,
      "learning_rate": 0.0009536082474226805,
      "loss": 2.8721,
      "step": 185
    },
    {
      "epoch": 0.028893203883495144,
      "grad_norm": 0.0,
      "learning_rate": 0.0009587628865979382,
      "loss": 2.7393,
      "step": 186
    },
    {
      "epoch": 0.02904854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.0009639175257731959,
      "loss": 2.8182,
      "step": 187
    },
    {
      "epoch": 0.029203883495145633,
      "grad_norm": 0.0,
      "learning_rate": 0.0009690721649484536,
      "loss": 2.8342,
      "step": 188
    },
    {
      "epoch": 0.029359223300970873,
      "grad_norm": 0.0,
      "learning_rate": 0.0009742268041237113,
      "loss": 2.7499,
      "step": 189
    },
    {
      "epoch": 0.029514563106796118,
      "grad_norm": 0.0,
      "learning_rate": 0.000979381443298969,
      "loss": 2.7865,
      "step": 190
    },
    {
      "epoch": 0.02966990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.000984536082474227,
      "loss": 2.845,
      "step": 191
    },
    {
      "epoch": 0.029825242718446603,
      "grad_norm": 0.0,
      "learning_rate": 0.0009896907216494846,
      "loss": 3.0044,
      "step": 192
    },
    {
      "epoch": 0.029980582524271843,
      "grad_norm": 0.0,
      "learning_rate": 0.0009948453608247423,
      "loss": 2.7278,
      "step": 193
    },
    {
      "epoch": 0.030135922330097088,
      "grad_norm": 0.0,
      "learning_rate": 0.001,
      "loss": 2.906,
      "step": 194
    },
    {
      "epoch": 0.030291262135922332,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999999366928046,
      "loss": 2.8739,
      "step": 195
    },
    {
      "epoch": 0.030446601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.000999999746771234,
      "loss": 2.7939,
      "step": 196
    },
    {
      "epoch": 0.030601941747572817,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999994302353364,
      "loss": 2.7694,
      "step": 197
    },
    {
      "epoch": 0.030757281553398057,
      "grad_norm": 0.0,
      "learning_rate": 0.000999998987085192,
      "loss": 2.8555,
      "step": 198
    },
    {
      "epoch": 0.0309126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.000999998417320913,
      "loss": 2.7201,
      "step": 199
    },
    {
      "epoch": 0.031067961165048542,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999977209426438,
      "loss": 2.8572,
      "step": 200
    },
    {
      "epoch": 0.031223300970873787,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999968979505607,
      "loss": 2.8499,
      "step": 201
    },
    {
      "epoch": 0.03137864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999959483448719,
      "loss": 2.7721,
      "step": 202
    },
    {
      "epoch": 0.031533980582524275,
      "grad_norm": 0.0,
      "learning_rate": 0.000999994872125818,
      "loss": 2.7571,
      "step": 203
    },
    {
      "epoch": 0.03168932038834951,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999936692936717,
      "loss": 2.7684,
      "step": 204
    },
    {
      "epoch": 0.031844660194174756,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999923398487376,
      "loss": 2.7914,
      "step": 205
    },
    {
      "epoch": 0.032,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999908837913518,
      "loss": 2.6978,
      "step": 206
    },
    {
      "epoch": 0.032155339805825245,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999893011218837,
      "loss": 2.7506,
      "step": 207
    },
    {
      "epoch": 0.03231067961165048,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999875918407335,
      "loss": 2.7743,
      "step": 208
    },
    {
      "epoch": 0.032466019417475726,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999857559483344,
      "loss": 2.7126,
      "step": 209
    },
    {
      "epoch": 0.03262135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999837934451515,
      "loss": 2.7681,
      "step": 210
    },
    {
      "epoch": 0.032776699029126215,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999817043316811,
      "loss": 2.7893,
      "step": 211
    },
    {
      "epoch": 0.03293203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.000999979488608453,
      "loss": 2.7778,
      "step": 212
    },
    {
      "epoch": 0.033087378640776696,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999771462760277,
      "loss": 2.8297,
      "step": 213
    },
    {
      "epoch": 0.03324271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999746773349985,
      "loss": 2.796,
      "step": 214
    },
    {
      "epoch": 0.033398058252427185,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999720817859908,
      "loss": 2.7683,
      "step": 215
    },
    {
      "epoch": 0.03355339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999693596296614,
      "loss": 2.8497,
      "step": 216
    },
    {
      "epoch": 0.03370873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999665108667002,
      "loss": 2.8156,
      "step": 217
    },
    {
      "epoch": 0.03386407766990291,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999635354978284,
      "loss": 2.6743,
      "step": 218
    },
    {
      "epoch": 0.034019417475728154,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999604335237994,
      "loss": 2.7367,
      "step": 219
    },
    {
      "epoch": 0.0341747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999572049453984,
      "loss": 2.7892,
      "step": 220
    },
    {
      "epoch": 0.03433009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999538497634435,
      "loss": 2.7192,
      "step": 221
    },
    {
      "epoch": 0.03448543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999503679787838,
      "loss": 2.8201,
      "step": 222
    },
    {
      "epoch": 0.034640776699029124,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999467595923016,
      "loss": 2.661,
      "step": 223
    },
    {
      "epoch": 0.03479611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999430246049102,
      "loss": 2.8405,
      "step": 224
    },
    {
      "epoch": 0.03495145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999391630175557,
      "loss": 2.8474,
      "step": 225
    },
    {
      "epoch": 0.03510679611650486,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999351748312155,
      "loss": 2.7268,
      "step": 226
    },
    {
      "epoch": 0.035262135922330094,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999310600469,
      "loss": 2.7985,
      "step": 227
    },
    {
      "epoch": 0.03541747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999268186656508,
      "loss": 2.7527,
      "step": 228
    },
    {
      "epoch": 0.03557281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999224506885426,
      "loss": 2.7654,
      "step": 229
    },
    {
      "epoch": 0.03572815533980583,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999179561166805,
      "loss": 2.6958,
      "step": 230
    },
    {
      "epoch": 0.03588349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999133349512036,
      "loss": 2.7573,
      "step": 231
    },
    {
      "epoch": 0.03603883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999085871932815,
      "loss": 2.7756,
      "step": 232
    },
    {
      "epoch": 0.03619417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.0009999037128441165,
      "loss": 2.6593,
      "step": 233
    },
    {
      "epoch": 0.0363495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0009998987119049433,
      "loss": 2.6689,
      "step": 234
    },
    {
      "epoch": 0.03650485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0009998935843770283,
      "loss": 2.8662,
      "step": 235
    },
    {
      "epoch": 0.03666019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.0009998883302616696,
      "loss": 2.7645,
      "step": 236
    },
    {
      "epoch": 0.03681553398058252,
      "grad_norm": 0.0,
      "learning_rate": 0.0009998829495601977,
      "loss": 2.7358,
      "step": 237
    },
    {
      "epoch": 0.03697087378640777,
      "grad_norm": 0.0,
      "learning_rate": 0.0009998774422739752,
      "loss": 2.7185,
      "step": 238
    },
    {
      "epoch": 0.03712621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0009998718084043969,
      "loss": 2.7625,
      "step": 239
    },
    {
      "epoch": 0.037281553398058255,
      "grad_norm": 0.0,
      "learning_rate": 0.0009998660479528894,
      "loss": 2.7269,
      "step": 240
    },
    {
      "epoch": 0.03743689320388349,
      "grad_norm": 0.0,
      "learning_rate": 0.0009998601609209111,
      "loss": 2.7753,
      "step": 241
    },
    {
      "epoch": 0.037592233009708736,
      "grad_norm": 0.0,
      "learning_rate": 0.0009998541473099532,
      "loss": 2.6881,
      "step": 242
    },
    {
      "epoch": 0.03774757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0009998480071215382,
      "loss": 2.7743,
      "step": 243
    },
    {
      "epoch": 0.037902912621359225,
      "grad_norm": 0.0,
      "learning_rate": 0.000999841740357221,
      "loss": 2.7588,
      "step": 244
    },
    {
      "epoch": 0.03805825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.000999835347018589,
      "loss": 2.7358,
      "step": 245
    },
    {
      "epoch": 0.038213592233009706,
      "grad_norm": 0.0,
      "learning_rate": 0.0009998288271072602,
      "loss": 2.7765,
      "step": 246
    },
    {
      "epoch": 0.03836893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0009998221806248866,
      "loss": 2.7511,
      "step": 247
    },
    {
      "epoch": 0.038524271844660195,
      "grad_norm": 0.0,
      "learning_rate": 0.0009998154075731508,
      "loss": 2.7809,
      "step": 248
    },
    {
      "epoch": 0.03867961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.0009998085079537683,
      "loss": 2.7296,
      "step": 249
    },
    {
      "epoch": 0.038834951456310676,
      "grad_norm": 0.0,
      "learning_rate": 0.0009998014817684858,
      "loss": 2.7432,
      "step": 250
    },
    {
      "epoch": 0.03899029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.0009997943290190828,
      "loss": 2.7527,
      "step": 251
    },
    {
      "epoch": 0.039145631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0009997870497073704,
      "loss": 2.7861,
      "step": 252
    },
    {
      "epoch": 0.03930097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.0009997796438351923,
      "loss": 2.7966,
      "step": 253
    },
    {
      "epoch": 0.03945631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0009997721114044236,
      "loss": 2.6838,
      "step": 254
    },
    {
      "epoch": 0.03961165048543689,
      "grad_norm": 0.0,
      "learning_rate": 0.0009997644524169715,
      "loss": 2.8391,
      "step": 255
    },
    {
      "epoch": 0.039766990291262135,
      "grad_norm": 0.0,
      "learning_rate": 0.000999756666874776,
      "loss": 2.7507,
      "step": 256
    },
    {
      "epoch": 0.03992233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.0009997487547798083,
      "loss": 2.7092,
      "step": 257
    },
    {
      "epoch": 0.04007766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.000999740716134072,
      "loss": 2.6934,
      "step": 258
    },
    {
      "epoch": 0.04023300970873787,
      "grad_norm": 0.0,
      "learning_rate": 0.0009997325509396029,
      "loss": 2.7378,
      "step": 259
    },
    {
      "epoch": 0.040388349514563104,
      "grad_norm": 0.0,
      "learning_rate": 0.0009997242591984683,
      "loss": 2.7816,
      "step": 260
    },
    {
      "epoch": 0.04054368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0009997158409127684,
      "loss": 2.7563,
      "step": 261
    },
    {
      "epoch": 0.04069902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.0009997072960846345,
      "loss": 2.7006,
      "step": 262
    },
    {
      "epoch": 0.04085436893203884,
      "grad_norm": 0.0,
      "learning_rate": 0.0009996986247162306,
      "loss": 2.7229,
      "step": 263
    },
    {
      "epoch": 0.041009708737864074,
      "grad_norm": 0.0,
      "learning_rate": 0.0009996898268097526,
      "loss": 2.7828,
      "step": 264
    },
    {
      "epoch": 0.04116504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0009996809023674281,
      "loss": 2.84,
      "step": 265
    },
    {
      "epoch": 0.04132038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.0009996718513915176,
      "loss": 2.7164,
      "step": 266
    },
    {
      "epoch": 0.04147572815533981,
      "grad_norm": 0.0,
      "learning_rate": 0.0009996626738843124,
      "loss": 2.644,
      "step": 267
    },
    {
      "epoch": 0.04163106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0009996533698481368,
      "loss": 2.8024,
      "step": 268
    },
    {
      "epoch": 0.04178640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.000999643939285347,
      "loss": 2.8522,
      "step": 269
    },
    {
      "epoch": 0.04194174757281553,
      "grad_norm": 0.0,
      "learning_rate": 0.0009996343821983306,
      "loss": 2.7351,
      "step": 270
    },
    {
      "epoch": 0.04209708737864078,
      "grad_norm": 0.0,
      "learning_rate": 0.0009996246985895085,
      "loss": 2.7886,
      "step": 271
    },
    {
      "epoch": 0.04225242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0009996148884613323,
      "loss": 2.6584,
      "step": 272
    },
    {
      "epoch": 0.042407766990291265,
      "grad_norm": 0.0,
      "learning_rate": 0.0009996049518162862,
      "loss": 2.7271,
      "step": 273
    },
    {
      "epoch": 0.0425631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0009995948886568867,
      "loss": 2.7469,
      "step": 274
    },
    {
      "epoch": 0.04271844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.000999584698985682,
      "loss": 2.7356,
      "step": 275
    },
    {
      "epoch": 0.04287378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0009995743828052524,
      "loss": 2.7869,
      "step": 276
    },
    {
      "epoch": 0.043029126213592235,
      "grad_norm": 0.0,
      "learning_rate": 0.0009995639401182102,
      "loss": 2.715,
      "step": 277
    },
    {
      "epoch": 0.04318446601941747,
      "grad_norm": 0.0,
      "learning_rate": 0.0009995533709271998,
      "loss": 2.7832,
      "step": 278
    },
    {
      "epoch": 0.043339805825242717,
      "grad_norm": 0.0,
      "learning_rate": 0.0009995426752348977,
      "loss": 2.6815,
      "step": 279
    },
    {
      "epoch": 0.04349514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0009995318530440123,
      "loss": 2.7589,
      "step": 280
    },
    {
      "epoch": 0.043650485436893205,
      "grad_norm": 0.0,
      "learning_rate": 0.0009995209043572842,
      "loss": 2.7856,
      "step": 281
    },
    {
      "epoch": 0.04380582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.0009995098291774858,
      "loss": 2.7019,
      "step": 282
    },
    {
      "epoch": 0.043961165048543686,
      "grad_norm": 0.0,
      "learning_rate": 0.0009994986275074214,
      "loss": 2.7089,
      "step": 283
    },
    {
      "epoch": 0.04411650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.0009994872993499282,
      "loss": 2.6525,
      "step": 284
    },
    {
      "epoch": 0.044271844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.0009994758447078745,
      "loss": 2.7555,
      "step": 285
    },
    {
      "epoch": 0.04442718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.0009994642635841606,
      "loss": 2.734,
      "step": 286
    },
    {
      "epoch": 0.04458252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0009994525559817197,
      "loss": 2.7883,
      "step": 287
    },
    {
      "epoch": 0.0447378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0009994407219035164,
      "loss": 2.7511,
      "step": 288
    },
    {
      "epoch": 0.044893203883495145,
      "grad_norm": 0.0,
      "learning_rate": 0.0009994287613525473,
      "loss": 2.7687,
      "step": 289
    },
    {
      "epoch": 0.04504854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.0009994166743318412,
      "loss": 2.7582,
      "step": 290
    },
    {
      "epoch": 0.04520388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0009994044608444587,
      "loss": 2.7275,
      "step": 291
    },
    {
      "epoch": 0.04535922330097087,
      "grad_norm": 0.0,
      "learning_rate": 0.000999392120893493,
      "loss": 2.6978,
      "step": 292
    },
    {
      "epoch": 0.045514563106796115,
      "grad_norm": 0.0,
      "learning_rate": 0.0009993796544820683,
      "loss": 2.7272,
      "step": 293
    },
    {
      "epoch": 0.04566990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0009993670616133422,
      "loss": 2.6941,
      "step": 294
    },
    {
      "epoch": 0.0458252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0009993543422905033,
      "loss": 2.6917,
      "step": 295
    },
    {
      "epoch": 0.04598058252427185,
      "grad_norm": 0.0,
      "learning_rate": 0.0009993414965167723,
      "loss": 2.7059,
      "step": 296
    },
    {
      "epoch": 0.046135922330097084,
      "grad_norm": 0.0,
      "learning_rate": 0.0009993285242954023,
      "loss": 2.8276,
      "step": 297
    },
    {
      "epoch": 0.04629126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0009993154256296782,
      "loss": 2.7623,
      "step": 298
    },
    {
      "epoch": 0.04644660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.000999302200522917,
      "loss": 2.7219,
      "step": 299
    },
    {
      "epoch": 0.04660194174757282,
      "grad_norm": 0.0,
      "learning_rate": 0.0009992888489784675,
      "loss": 2.7091,
      "step": 300
    },
    {
      "epoch": 0.04675728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.0009992753709997108,
      "loss": 2.681,
      "step": 301
    },
    {
      "epoch": 0.0469126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.00099926176659006,
      "loss": 2.7367,
      "step": 302
    },
    {
      "epoch": 0.04706796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.0009992480357529602,
      "loss": 2.79,
      "step": 303
    },
    {
      "epoch": 0.04722330097087379,
      "grad_norm": 0.0,
      "learning_rate": 0.000999234178491888,
      "loss": 2.6096,
      "step": 304
    },
    {
      "epoch": 0.04737864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.000999220194810353,
      "loss": 2.7275,
      "step": 305
    },
    {
      "epoch": 0.04753398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.000999206084711896,
      "loss": 2.7074,
      "step": 306
    },
    {
      "epoch": 0.04768932038834951,
      "grad_norm": 0.0,
      "learning_rate": 0.0009991918482000902,
      "loss": 2.7676,
      "step": 307
    },
    {
      "epoch": 0.04784466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.0009991774852785406,
      "loss": 2.7214,
      "step": 308
    },
    {
      "epoch": 0.048,
      "grad_norm": 0.0,
      "learning_rate": 0.0009991629959508843,
      "loss": 2.7916,
      "step": 309
    },
    {
      "epoch": 0.048155339805825245,
      "grad_norm": 0.0,
      "learning_rate": 0.0009991483802207904,
      "loss": 2.7419,
      "step": 310
    },
    {
      "epoch": 0.04831067961165048,
      "grad_norm": 0.0,
      "learning_rate": 0.00099913363809196,
      "loss": 2.7073,
      "step": 311
    },
    {
      "epoch": 0.04846601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.0009991187695681265,
      "loss": 2.6149,
      "step": 312
    },
    {
      "epoch": 0.04862135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0009991037746530545,
      "loss": 2.7109,
      "step": 313
    },
    {
      "epoch": 0.048776699029126215,
      "grad_norm": 0.0,
      "learning_rate": 0.0009990886533505418,
      "loss": 2.6937,
      "step": 314
    },
    {
      "epoch": 0.04893203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.000999073405664417,
      "loss": 2.6068,
      "step": 315
    },
    {
      "epoch": 0.0490873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0009990580315985416,
      "loss": 2.7104,
      "step": 316
    },
    {
      "epoch": 0.04924271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.0009990425311568085,
      "loss": 2.7,
      "step": 317
    },
    {
      "epoch": 0.049398058252427185,
      "grad_norm": 0.0,
      "learning_rate": 0.0009990269043431432,
      "loss": 2.7769,
      "step": 318
    },
    {
      "epoch": 0.04955339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0009990111511615025,
      "loss": 2.7008,
      "step": 319
    },
    {
      "epoch": 0.04970873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0009989952716158758,
      "loss": 2.7664,
      "step": 320
    },
    {
      "epoch": 0.04986407766990291,
      "grad_norm": 0.0,
      "learning_rate": 0.0009989792657102842,
      "loss": 2.6472,
      "step": 321
    },
    {
      "epoch": 0.050019417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.0009989631334487807,
      "loss": 2.6541,
      "step": 322
    },
    {
      "epoch": 0.0501747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0009989468748354505,
      "loss": 2.7285,
      "step": 323
    },
    {
      "epoch": 0.05033009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0009989304898744109,
      "loss": 2.6901,
      "step": 324
    },
    {
      "epoch": 0.05048543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.000998913978569811,
      "loss": 2.6611,
      "step": 325
    },
    {
      "epoch": 0.050640776699029125,
      "grad_norm": 0.0,
      "learning_rate": 0.000998897340925832,
      "loss": 2.7343,
      "step": 326
    },
    {
      "epoch": 0.05079611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0009988805769466868,
      "loss": 2.7197,
      "step": 327
    },
    {
      "epoch": 0.05095145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.0009988636866366208,
      "loss": 2.7902,
      "step": 328
    },
    {
      "epoch": 0.05110679611650486,
      "grad_norm": 0.0,
      "learning_rate": 0.0009988466699999109,
      "loss": 2.6873,
      "step": 329
    },
    {
      "epoch": 0.051262135922330095,
      "grad_norm": 0.0,
      "learning_rate": 0.0009988295270408662,
      "loss": 2.6552,
      "step": 330
    },
    {
      "epoch": 0.05141747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0009988122577638278,
      "loss": 2.7031,
      "step": 331
    },
    {
      "epoch": 0.05157281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.0009987948621731691,
      "loss": 2.7284,
      "step": 332
    },
    {
      "epoch": 0.05172815533980583,
      "grad_norm": 0.0,
      "learning_rate": 0.0009987773402732947,
      "loss": 2.6825,
      "step": 333
    },
    {
      "epoch": 0.05188349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.0009987596920686419,
      "loss": 2.8175,
      "step": 334
    },
    {
      "epoch": 0.05203883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0009987419175636796,
      "loss": 2.6216,
      "step": 335
    },
    {
      "epoch": 0.05219417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.000998724016762909,
      "loss": 2.6462,
      "step": 336
    },
    {
      "epoch": 0.0523495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.000998705989670863,
      "loss": 2.6948,
      "step": 337
    },
    {
      "epoch": 0.05250485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0009986878362921066,
      "loss": 2.6713,
      "step": 338
    },
    {
      "epoch": 0.05266019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.0009986695566312368,
      "loss": 2.6711,
      "step": 339
    },
    {
      "epoch": 0.05281553398058252,
      "grad_norm": 0.0,
      "learning_rate": 0.0009986511506928824,
      "loss": 2.6454,
      "step": 340
    },
    {
      "epoch": 0.05297087378640777,
      "grad_norm": 0.0,
      "learning_rate": 0.0009986326184817044,
      "loss": 2.7805,
      "step": 341
    },
    {
      "epoch": 0.05312621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0009986139600023955,
      "loss": 2.6992,
      "step": 342
    },
    {
      "epoch": 0.053281553398058255,
      "grad_norm": 0.0,
      "learning_rate": 0.000998595175259681,
      "loss": 2.5815,
      "step": 343
    },
    {
      "epoch": 0.05343689320388349,
      "grad_norm": 0.0,
      "learning_rate": 0.0009985762642583174,
      "loss": 2.7235,
      "step": 344
    },
    {
      "epoch": 0.05359223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0009985572270030936,
      "loss": 2.6144,
      "step": 345
    },
    {
      "epoch": 0.05374757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0009985380634988303,
      "loss": 2.6934,
      "step": 346
    },
    {
      "epoch": 0.053902912621359225,
      "grad_norm": 0.0,
      "learning_rate": 0.0009985187737503805,
      "loss": 2.6976,
      "step": 347
    },
    {
      "epoch": 0.05405825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.0009984993577626285,
      "loss": 2.6649,
      "step": 348
    },
    {
      "epoch": 0.05421359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0009984798155404913,
      "loss": 2.6295,
      "step": 349
    },
    {
      "epoch": 0.05436893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0009984601470889175,
      "loss": 2.7434,
      "step": 350
    },
    {
      "epoch": 0.054524271844660195,
      "grad_norm": 0.0,
      "learning_rate": 0.0009984403524128877,
      "loss": 2.7185,
      "step": 351
    },
    {
      "epoch": 0.05467961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.0009984204315174145,
      "loss": 2.7319,
      "step": 352
    },
    {
      "epoch": 0.05483495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0009984003844075425,
      "loss": 2.7354,
      "step": 353
    },
    {
      "epoch": 0.05499029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.000998380211088348,
      "loss": 2.686,
      "step": 354
    },
    {
      "epoch": 0.055145631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0009983599115649397,
      "loss": 2.6735,
      "step": 355
    },
    {
      "epoch": 0.05530097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.0009983394858424576,
      "loss": 2.7264,
      "step": 356
    },
    {
      "epoch": 0.05545631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.000998318933926075,
      "loss": 2.6908,
      "step": 357
    },
    {
      "epoch": 0.05561165048543689,
      "grad_norm": 0.0,
      "learning_rate": 0.0009982982558209951,
      "loss": 2.7578,
      "step": 358
    },
    {
      "epoch": 0.055766990291262135,
      "grad_norm": 0.0,
      "learning_rate": 0.000998277451532455,
      "loss": 2.6813,
      "step": 359
    },
    {
      "epoch": 0.05592233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.0009982565210657226,
      "loss": 2.7319,
      "step": 360
    },
    {
      "epoch": 0.05607766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.000998235464426098,
      "loss": 2.7983,
      "step": 361
    },
    {
      "epoch": 0.05623300970873787,
      "grad_norm": 0.0,
      "learning_rate": 0.0009982142816189136,
      "loss": 2.7012,
      "step": 362
    },
    {
      "epoch": 0.056388349514563105,
      "grad_norm": 0.0,
      "learning_rate": 0.0009981929726495336,
      "loss": 2.6869,
      "step": 363
    },
    {
      "epoch": 0.05654368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0009981715375233537,
      "loss": 2.7023,
      "step": 364
    },
    {
      "epoch": 0.05669902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.0009981499762458021,
      "loss": 2.6136,
      "step": 365
    },
    {
      "epoch": 0.05685436893203884,
      "grad_norm": 0.0,
      "learning_rate": 0.0009981282888223387,
      "loss": 2.686,
      "step": 366
    },
    {
      "epoch": 0.057009708737864075,
      "grad_norm": 0.0,
      "learning_rate": 0.0009981064752584555,
      "loss": 2.6912,
      "step": 367
    },
    {
      "epoch": 0.05716504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.000998084535559676,
      "loss": 2.6855,
      "step": 368
    },
    {
      "epoch": 0.05732038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.000998062469731556,
      "loss": 2.7356,
      "step": 369
    },
    {
      "epoch": 0.05747572815533981,
      "grad_norm": 0.0,
      "learning_rate": 0.0009980402777796835,
      "loss": 2.682,
      "step": 370
    },
    {
      "epoch": 0.05763106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0009980179597096782,
      "loss": 2.6055,
      "step": 371
    },
    {
      "epoch": 0.05778640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.0009979955155271913,
      "loss": 2.5657,
      "step": 372
    },
    {
      "epoch": 0.05794174757281553,
      "grad_norm": 0.0,
      "learning_rate": 0.0009979729452379064,
      "loss": 2.6721,
      "step": 373
    },
    {
      "epoch": 0.05809708737864078,
      "grad_norm": 0.0,
      "learning_rate": 0.000997950248847539,
      "loss": 2.6971,
      "step": 374
    },
    {
      "epoch": 0.05825242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0009979274263618367,
      "loss": 2.6118,
      "step": 375
    },
    {
      "epoch": 0.058407766990291266,
      "grad_norm": 0.0,
      "learning_rate": 0.0009979044777865785,
      "loss": 2.5935,
      "step": 376
    },
    {
      "epoch": 0.0585631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0009978814031275758,
      "loss": 2.6577,
      "step": 377
    },
    {
      "epoch": 0.05871844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.0009978582023906716,
      "loss": 2.7052,
      "step": 378
    },
    {
      "epoch": 0.05887378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0009978348755817412,
      "loss": 2.6217,
      "step": 379
    },
    {
      "epoch": 0.059029126213592235,
      "grad_norm": 0.0,
      "learning_rate": 0.0009978114227066916,
      "loss": 2.632,
      "step": 380
    },
    {
      "epoch": 0.05918446601941747,
      "grad_norm": 0.0,
      "learning_rate": 0.0009977878437714618,
      "loss": 2.6774,
      "step": 381
    },
    {
      "epoch": 0.05933980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0009977641387820223,
      "loss": 2.6981,
      "step": 382
    },
    {
      "epoch": 0.05949514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0009977403077443762,
      "loss": 2.653,
      "step": 383
    },
    {
      "epoch": 0.059650485436893205,
      "grad_norm": 0.0,
      "learning_rate": 0.0009977163506645582,
      "loss": 2.6759,
      "step": 384
    },
    {
      "epoch": 0.05980582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.0009976922675486348,
      "loss": 2.7179,
      "step": 385
    },
    {
      "epoch": 0.05996116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0009976680584027047,
      "loss": 2.686,
      "step": 386
    },
    {
      "epoch": 0.06011650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.0009976437232328981,
      "loss": 2.5736,
      "step": 387
    },
    {
      "epoch": 0.060271844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.0009976192620453777,
      "loss": 2.6487,
      "step": 388
    },
    {
      "epoch": 0.06042718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.0009975946748463375,
      "loss": 2.6353,
      "step": 389
    },
    {
      "epoch": 0.060582524271844664,
      "grad_norm": 0.0,
      "learning_rate": 0.0009975699616420037,
      "loss": 2.6334,
      "step": 390
    },
    {
      "epoch": 0.0607378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0009975451224386346,
      "loss": 2.6584,
      "step": 391
    },
    {
      "epoch": 0.060893203883495145,
      "grad_norm": 0.0,
      "learning_rate": 0.00099752015724252,
      "loss": 2.7341,
      "step": 392
    },
    {
      "epoch": 0.06104854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.000997495066059982,
      "loss": 2.625,
      "step": 393
    },
    {
      "epoch": 0.06120388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0009974698488973742,
      "loss": 2.59,
      "step": 394
    },
    {
      "epoch": 0.06135922330097087,
      "grad_norm": 0.0,
      "learning_rate": 0.0009974445057610824,
      "loss": 2.6782,
      "step": 395
    },
    {
      "epoch": 0.061514563106796115,
      "grad_norm": 0.0,
      "learning_rate": 0.0009974190366575243,
      "loss": 2.6288,
      "step": 396
    },
    {
      "epoch": 0.06166990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.000997393441593149,
      "loss": 2.6296,
      "step": 397
    },
    {
      "epoch": 0.0618252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0009973677205744387,
      "loss": 2.7249,
      "step": 398
    },
    {
      "epoch": 0.06198058252427185,
      "grad_norm": 0.0,
      "learning_rate": 0.000997341873607906,
      "loss": 2.6603,
      "step": 399
    },
    {
      "epoch": 0.062135922330097085,
      "grad_norm": 0.0,
      "learning_rate": 0.0009973159007000962,
      "loss": 2.6639,
      "step": 400
    },
    {
      "epoch": 0.06229126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0009972898018575868,
      "loss": 2.5863,
      "step": 401
    },
    {
      "epoch": 0.06244660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.0009972635770869862,
      "loss": 2.7096,
      "step": 402
    },
    {
      "epoch": 0.06260194174757282,
      "grad_norm": 0.0,
      "learning_rate": 0.0009972372263949358,
      "loss": 2.6157,
      "step": 403
    },
    {
      "epoch": 0.06275728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.000997210749788108,
      "loss": 2.5997,
      "step": 404
    },
    {
      "epoch": 0.0629126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0009971841472732077,
      "loss": 2.6548,
      "step": 405
    },
    {
      "epoch": 0.06306796116504855,
      "grad_norm": 0.0,
      "learning_rate": 0.0009971574188569713,
      "loss": 2.6979,
      "step": 406
    },
    {
      "epoch": 0.06322330097087378,
      "grad_norm": 0.0,
      "learning_rate": 0.000997130564546167,
      "loss": 2.605,
      "step": 407
    },
    {
      "epoch": 0.06337864077669902,
      "grad_norm": 0.0,
      "learning_rate": 0.0009971035843475953,
      "loss": 2.6931,
      "step": 408
    },
    {
      "epoch": 0.06353398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.0009970764782680883,
      "loss": 2.6406,
      "step": 409
    },
    {
      "epoch": 0.06368932038834951,
      "grad_norm": 0.0,
      "learning_rate": 0.00099704924631451,
      "loss": 2.6794,
      "step": 410
    },
    {
      "epoch": 0.06384466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.0009970218884937565,
      "loss": 2.6875,
      "step": 411
    },
    {
      "epoch": 0.064,
      "grad_norm": 0.0,
      "learning_rate": 0.0009969944048127555,
      "loss": 2.6613,
      "step": 412
    },
    {
      "epoch": 0.06415533980582525,
      "grad_norm": 0.0,
      "learning_rate": 0.0009969667952784667,
      "loss": 2.642,
      "step": 413
    },
    {
      "epoch": 0.06431067961165049,
      "grad_norm": 0.0,
      "learning_rate": 0.0009969390598978814,
      "loss": 2.605,
      "step": 414
    },
    {
      "epoch": 0.06446601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.0009969111986780233,
      "loss": 2.5731,
      "step": 415
    },
    {
      "epoch": 0.06462135922330096,
      "grad_norm": 0.0,
      "learning_rate": 0.0009968832116259475,
      "loss": 2.6074,
      "step": 416
    },
    {
      "epoch": 0.06477669902912621,
      "grad_norm": 0.0,
      "learning_rate": 0.000996855098748741,
      "loss": 2.6058,
      "step": 417
    },
    {
      "epoch": 0.06493203883495145,
      "grad_norm": 0.0,
      "learning_rate": 0.000996826860053523,
      "loss": 2.6909,
      "step": 418
    },
    {
      "epoch": 0.0650873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0009967984955474444,
      "loss": 2.6475,
      "step": 419
    },
    {
      "epoch": 0.06524271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.0009967700052376876,
      "loss": 2.6841,
      "step": 420
    },
    {
      "epoch": 0.06539805825242719,
      "grad_norm": 0.0,
      "learning_rate": 0.0009967413891314675,
      "loss": 2.6296,
      "step": 421
    },
    {
      "epoch": 0.06555339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0009967126472360303,
      "loss": 2.5524,
      "step": 422
    },
    {
      "epoch": 0.06570873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0009966837795586543,
      "loss": 2.5685,
      "step": 423
    },
    {
      "epoch": 0.06586407766990292,
      "grad_norm": 0.0,
      "learning_rate": 0.0009966547861066499,
      "loss": 2.7332,
      "step": 424
    },
    {
      "epoch": 0.06601941747572816,
      "grad_norm": 0.0,
      "learning_rate": 0.0009966256668873586,
      "loss": 2.686,
      "step": 425
    },
    {
      "epoch": 0.06617475728155339,
      "grad_norm": 0.0,
      "learning_rate": 0.0009965964219081547,
      "loss": 2.6547,
      "step": 426
    },
    {
      "epoch": 0.06633009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0009965670511764436,
      "loss": 2.6454,
      "step": 427
    },
    {
      "epoch": 0.06648543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.000996537554699663,
      "loss": 2.6494,
      "step": 428
    },
    {
      "epoch": 0.06664077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.000996507932485282,
      "loss": 2.6199,
      "step": 429
    },
    {
      "epoch": 0.06679611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0009964781845408018,
      "loss": 2.6228,
      "step": 430
    },
    {
      "epoch": 0.06695145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.0009964483108737556,
      "loss": 2.6112,
      "step": 431
    },
    {
      "epoch": 0.06710679611650486,
      "grad_norm": 0.0,
      "learning_rate": 0.0009964183114917083,
      "loss": 2.5746,
      "step": 432
    },
    {
      "epoch": 0.0672621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0009963881864022567,
      "loss": 2.7133,
      "step": 433
    },
    {
      "epoch": 0.06741747572815535,
      "grad_norm": 0.0,
      "learning_rate": 0.000996357935613029,
      "loss": 2.6105,
      "step": 434
    },
    {
      "epoch": 0.06757281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.0009963275591316856,
      "loss": 2.6216,
      "step": 435
    },
    {
      "epoch": 0.06772815533980582,
      "grad_norm": 0.0,
      "learning_rate": 0.0009962970569659191,
      "loss": 2.6733,
      "step": 436
    },
    {
      "epoch": 0.06788349514563106,
      "grad_norm": 0.0,
      "learning_rate": 0.0009962664291234531,
      "loss": 2.6035,
      "step": 437
    },
    {
      "epoch": 0.06803883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0009962356756120438,
      "loss": 2.5948,
      "step": 438
    },
    {
      "epoch": 0.06819417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.0009962047964394787,
      "loss": 2.5956,
      "step": 439
    },
    {
      "epoch": 0.0683495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0009961737916135771,
      "loss": 2.5957,
      "step": 440
    },
    {
      "epoch": 0.06850485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0009961426611421907,
      "loss": 2.6304,
      "step": 441
    },
    {
      "epoch": 0.06866019417475729,
      "grad_norm": 0.0,
      "learning_rate": 0.0009961114050332023,
      "loss": 2.6396,
      "step": 442
    },
    {
      "epoch": 0.06881553398058253,
      "grad_norm": 0.0,
      "learning_rate": 0.000996080023294527,
      "loss": 2.6101,
      "step": 443
    },
    {
      "epoch": 0.06897087378640776,
      "grad_norm": 0.0,
      "learning_rate": 0.0009960485159341117,
      "loss": 2.6591,
      "step": 444
    },
    {
      "epoch": 0.069126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0009960168829599347,
      "loss": 2.5691,
      "step": 445
    },
    {
      "epoch": 0.06928155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.0009959851243800067,
      "loss": 2.6478,
      "step": 446
    },
    {
      "epoch": 0.06943689320388349,
      "grad_norm": 0.0,
      "learning_rate": 0.0009959532402023694,
      "loss": 2.6257,
      "step": 447
    },
    {
      "epoch": 0.06959223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.000995921230435097,
      "loss": 2.6041,
      "step": 448
    },
    {
      "epoch": 0.06974757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0009958890950862958,
      "loss": 2.6437,
      "step": 449
    },
    {
      "epoch": 0.06990291262135923,
      "grad_norm": 0.0,
      "learning_rate": 0.0009958568341641026,
      "loss": 2.6195,
      "step": 450
    },
    {
      "epoch": 0.07005825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.0009958244476766874,
      "loss": 2.476,
      "step": 451
    },
    {
      "epoch": 0.07021359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.000995791935632251,
      "loss": 2.6508,
      "step": 452
    },
    {
      "epoch": 0.07036893203883496,
      "grad_norm": 0.0,
      "learning_rate": 0.0009957592980390268,
      "loss": 2.6493,
      "step": 453
    },
    {
      "epoch": 0.07052427184466019,
      "grad_norm": 0.0,
      "learning_rate": 0.0009957265349052789,
      "loss": 2.6702,
      "step": 454
    },
    {
      "epoch": 0.07067961165048543,
      "grad_norm": 0.0,
      "learning_rate": 0.0009956936462393044,
      "loss": 2.6295,
      "step": 455
    },
    {
      "epoch": 0.07083495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0009956606320494317,
      "loss": 2.593,
      "step": 456
    },
    {
      "epoch": 0.07099029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.0009956274923440208,
      "loss": 2.6188,
      "step": 457
    },
    {
      "epoch": 0.07114563106796117,
      "grad_norm": 0.0,
      "learning_rate": 0.0009955942271314632,
      "loss": 2.627,
      "step": 458
    },
    {
      "epoch": 0.07130097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.0009955608364201835,
      "loss": 2.626,
      "step": 459
    },
    {
      "epoch": 0.07145631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0009955273202186365,
      "loss": 2.611,
      "step": 460
    },
    {
      "epoch": 0.0716116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0009954936785353096,
      "loss": 2.6233,
      "step": 461
    },
    {
      "epoch": 0.07176699029126214,
      "grad_norm": 0.0,
      "learning_rate": 0.0009954599113787222,
      "loss": 2.6523,
      "step": 462
    },
    {
      "epoch": 0.07192233009708737,
      "grad_norm": 0.0,
      "learning_rate": 0.0009954260187574246,
      "loss": 2.6382,
      "step": 463
    },
    {
      "epoch": 0.07207766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0009953920006799998,
      "loss": 2.5994,
      "step": 464
    },
    {
      "epoch": 0.07223300970873786,
      "grad_norm": 0.0,
      "learning_rate": 0.0009953578571550617,
      "loss": 2.6184,
      "step": 465
    },
    {
      "epoch": 0.0723883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0009953235881912568,
      "loss": 2.5896,
      "step": 466
    },
    {
      "epoch": 0.07254368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.000995289193797263,
      "loss": 2.6622,
      "step": 467
    },
    {
      "epoch": 0.0726990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0009952546739817898,
      "loss": 2.593,
      "step": 468
    },
    {
      "epoch": 0.07285436893203884,
      "grad_norm": 0.0,
      "learning_rate": 0.0009952200287535787,
      "loss": 2.5494,
      "step": 469
    },
    {
      "epoch": 0.07300970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.0009951852581214027,
      "loss": 2.5175,
      "step": 470
    },
    {
      "epoch": 0.07316504854368933,
      "grad_norm": 0.0,
      "learning_rate": 0.0009951503620940667,
      "loss": 2.5856,
      "step": 471
    },
    {
      "epoch": 0.07332038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.000995115340680408,
      "loss": 2.6508,
      "step": 472
    },
    {
      "epoch": 0.0734757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0009950801938892941,
      "loss": 2.688,
      "step": 473
    },
    {
      "epoch": 0.07363106796116504,
      "grad_norm": 0.0,
      "learning_rate": 0.0009950449217296258,
      "loss": 2.5602,
      "step": 474
    },
    {
      "epoch": 0.07378640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.000995009524210335,
      "loss": 2.5645,
      "step": 475
    },
    {
      "epoch": 0.07394174757281553,
      "grad_norm": 0.0,
      "learning_rate": 0.0009949740013403849,
      "loss": 2.6409,
      "step": 476
    },
    {
      "epoch": 0.07409708737864078,
      "grad_norm": 0.0,
      "learning_rate": 0.0009949383531287713,
      "loss": 2.6758,
      "step": 477
    },
    {
      "epoch": 0.07425242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0009949025795845216,
      "loss": 2.5547,
      "step": 478
    },
    {
      "epoch": 0.07440776699029127,
      "grad_norm": 0.0,
      "learning_rate": 0.0009948666807166942,
      "loss": 2.6139,
      "step": 479
    },
    {
      "epoch": 0.07456310679611651,
      "grad_norm": 0.0,
      "learning_rate": 0.00099483065653438,
      "loss": 2.5957,
      "step": 480
    },
    {
      "epoch": 0.07471844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.000994794507046701,
      "loss": 2.5919,
      "step": 481
    },
    {
      "epoch": 0.07487378640776698,
      "grad_norm": 0.0,
      "learning_rate": 0.0009947582322628117,
      "loss": 2.6093,
      "step": 482
    },
    {
      "epoch": 0.07502912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.0009947218321918981,
      "loss": 2.6162,
      "step": 483
    },
    {
      "epoch": 0.07518446601941747,
      "grad_norm": 0.0,
      "learning_rate": 0.000994685306843177,
      "loss": 2.6348,
      "step": 484
    },
    {
      "epoch": 0.07533980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0009946486562258984,
      "loss": 2.6357,
      "step": 485
    },
    {
      "epoch": 0.07549514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0009946118803493427,
      "loss": 2.5511,
      "step": 486
    },
    {
      "epoch": 0.0756504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.000994574979222823,
      "loss": 2.5441,
      "step": 487
    },
    {
      "epoch": 0.07580582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.0009945379528556837,
      "loss": 2.6897,
      "step": 488
    },
    {
      "epoch": 0.0759611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.000994500801257301,
      "loss": 2.6184,
      "step": 489
    },
    {
      "epoch": 0.07611650485436894,
      "grad_norm": 0.0,
      "learning_rate": 0.0009944635244370824,
      "loss": 2.5498,
      "step": 490
    },
    {
      "epoch": 0.07627184466019417,
      "grad_norm": 0.0,
      "learning_rate": 0.0009944261224044677,
      "loss": 2.6978,
      "step": 491
    },
    {
      "epoch": 0.07642718446601941,
      "grad_norm": 0.0,
      "learning_rate": 0.0009943885951689282,
      "loss": 2.5974,
      "step": 492
    },
    {
      "epoch": 0.07658252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0009943509427399668,
      "loss": 2.6343,
      "step": 493
    },
    {
      "epoch": 0.0767378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0009943131651271183,
      "loss": 2.5353,
      "step": 494
    },
    {
      "epoch": 0.07689320388349515,
      "grad_norm": 0.0,
      "learning_rate": 0.0009942752623399488,
      "loss": 2.642,
      "step": 495
    },
    {
      "epoch": 0.07704854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.0009942372343880567,
      "loss": 2.509,
      "step": 496
    },
    {
      "epoch": 0.07720388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0009941990812810716,
      "loss": 2.6425,
      "step": 497
    },
    {
      "epoch": 0.07735922330097088,
      "grad_norm": 0.0,
      "learning_rate": 0.000994160803028655,
      "loss": 2.559,
      "step": 498
    },
    {
      "epoch": 0.07751456310679612,
      "grad_norm": 0.0,
      "learning_rate": 0.0009941223996405002,
      "loss": 2.6083,
      "step": 499
    },
    {
      "epoch": 0.07766990291262135,
      "grad_norm": 0.0,
      "learning_rate": 0.0009940838711263319,
      "loss": 2.4847,
      "step": 500
    },
    {
      "epoch": 0.0778252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0009940452174959063,
      "loss": 2.6021,
      "step": 501
    },
    {
      "epoch": 0.07798058252427184,
      "grad_norm": 0.0,
      "learning_rate": 0.0009940064387590122,
      "loss": 2.6819,
      "step": 502
    },
    {
      "epoch": 0.07813592233009709,
      "grad_norm": 0.0,
      "learning_rate": 0.0009939675349254693,
      "loss": 2.6096,
      "step": 503
    },
    {
      "epoch": 0.07829126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.000993928506005129,
      "loss": 2.5786,
      "step": 504
    },
    {
      "epoch": 0.07844660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.0009938893520078746,
      "loss": 2.4753,
      "step": 505
    },
    {
      "epoch": 0.07860194174757282,
      "grad_norm": 0.0,
      "learning_rate": 0.000993850072943621,
      "loss": 2.6564,
      "step": 506
    },
    {
      "epoch": 0.07875728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.0009938106688223152,
      "loss": 2.6278,
      "step": 507
    },
    {
      "epoch": 0.0789126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0009937711396539347,
      "loss": 2.5969,
      "step": 508
    },
    {
      "epoch": 0.07906796116504855,
      "grad_norm": 0.0,
      "learning_rate": 0.0009937314854484903,
      "loss": 2.6304,
      "step": 509
    },
    {
      "epoch": 0.07922330097087378,
      "grad_norm": 0.0,
      "learning_rate": 0.0009936917062160228,
      "loss": 2.5514,
      "step": 510
    },
    {
      "epoch": 0.07937864077669902,
      "grad_norm": 0.0,
      "learning_rate": 0.0009936518019666059,
      "loss": 2.5988,
      "step": 511
    },
    {
      "epoch": 0.07953398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.0009936117727103442,
      "loss": 2.5953,
      "step": 512
    },
    {
      "epoch": 0.07968932038834951,
      "grad_norm": 0.0,
      "learning_rate": 0.0009935716184573747,
      "loss": 2.5159,
      "step": 513
    },
    {
      "epoch": 0.07984466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.0009935313392178651,
      "loss": 2.6185,
      "step": 514
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.0,
      "learning_rate": 0.0009934909350020158,
      "loss": 2.6047,
      "step": 515
    },
    {
      "epoch": 0.08015533980582525,
      "grad_norm": 0.0,
      "learning_rate": 0.0009934504058200578,
      "loss": 2.7123,
      "step": 516
    },
    {
      "epoch": 0.08031067961165049,
      "grad_norm": 0.0,
      "learning_rate": 0.0009934097516822544,
      "loss": 2.556,
      "step": 517
    },
    {
      "epoch": 0.08046601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.0009933689725989006,
      "loss": 2.5664,
      "step": 518
    },
    {
      "epoch": 0.08062135922330096,
      "grad_norm": 0.0,
      "learning_rate": 0.0009933280685803228,
      "loss": 2.5715,
      "step": 519
    },
    {
      "epoch": 0.08077669902912621,
      "grad_norm": 0.0,
      "learning_rate": 0.0009932870396368789,
      "loss": 2.6458,
      "step": 520
    },
    {
      "epoch": 0.08093203883495145,
      "grad_norm": 0.0,
      "learning_rate": 0.0009932458857789589,
      "loss": 2.6451,
      "step": 521
    },
    {
      "epoch": 0.0810873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0009932046070169835,
      "loss": 2.6062,
      "step": 522
    },
    {
      "epoch": 0.08124271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.0009931632033614065,
      "loss": 2.5106,
      "step": 523
    },
    {
      "epoch": 0.08139805825242719,
      "grad_norm": 0.0,
      "learning_rate": 0.0009931216748227118,
      "loss": 2.6069,
      "step": 524
    },
    {
      "epoch": 0.08155339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0009930800214114162,
      "loss": 2.5818,
      "step": 525
    },
    {
      "epoch": 0.08170873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.000993038243138067,
      "loss": 2.63,
      "step": 526
    },
    {
      "epoch": 0.08186407766990292,
      "grad_norm": 0.0,
      "learning_rate": 0.0009929963400132442,
      "loss": 2.6733,
      "step": 527
    },
    {
      "epoch": 0.08201941747572815,
      "grad_norm": 0.0,
      "learning_rate": 0.0009929543120475587,
      "loss": 2.5845,
      "step": 528
    },
    {
      "epoch": 0.08217475728155339,
      "grad_norm": 0.0,
      "learning_rate": 0.0009929121592516528,
      "loss": 2.6118,
      "step": 529
    },
    {
      "epoch": 0.08233009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0009928698816362013,
      "loss": 2.5073,
      "step": 530
    },
    {
      "epoch": 0.08248543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.0009928274792119097,
      "loss": 2.5425,
      "step": 531
    },
    {
      "epoch": 0.08264077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.000992784951989516,
      "loss": 2.5739,
      "step": 532
    },
    {
      "epoch": 0.08279611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0009927422999797889,
      "loss": 2.6029,
      "step": 533
    },
    {
      "epoch": 0.08295145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.0009926995231935292,
      "loss": 2.5031,
      "step": 534
    },
    {
      "epoch": 0.08310679611650486,
      "grad_norm": 0.0,
      "learning_rate": 0.0009926566216415691,
      "loss": 2.498,
      "step": 535
    },
    {
      "epoch": 0.0832621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.000992613595334773,
      "loss": 2.5573,
      "step": 536
    },
    {
      "epoch": 0.08341747572815535,
      "grad_norm": 0.0,
      "learning_rate": 0.000992570444284036,
      "loss": 2.5349,
      "step": 537
    },
    {
      "epoch": 0.08357281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.0009925271685002853,
      "loss": 2.5281,
      "step": 538
    },
    {
      "epoch": 0.08372815533980582,
      "grad_norm": 0.0,
      "learning_rate": 0.0009924837679944794,
      "loss": 2.5709,
      "step": 539
    },
    {
      "epoch": 0.08388349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.0009924402427776088,
      "loss": 2.5667,
      "step": 540
    },
    {
      "epoch": 0.08403883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0009923965928606952,
      "loss": 2.5149,
      "step": 541
    },
    {
      "epoch": 0.08419417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.0009923528182547922,
      "loss": 2.6052,
      "step": 542
    },
    {
      "epoch": 0.0843495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0009923089189709845,
      "loss": 2.5885,
      "step": 543
    },
    {
      "epoch": 0.08450485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0009922648950203888,
      "loss": 2.5935,
      "step": 544
    },
    {
      "epoch": 0.08466019417475729,
      "grad_norm": 0.0,
      "learning_rate": 0.0009922207464141532,
      "loss": 2.5255,
      "step": 545
    },
    {
      "epoch": 0.08481553398058253,
      "grad_norm": 0.0,
      "learning_rate": 0.0009921764731634574,
      "loss": 2.493,
      "step": 546
    },
    {
      "epoch": 0.08497087378640776,
      "grad_norm": 0.0,
      "learning_rate": 0.0009921320752795128,
      "loss": 2.5083,
      "step": 547
    },
    {
      "epoch": 0.085126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0009920875527735622,
      "loss": 2.5836,
      "step": 548
    },
    {
      "epoch": 0.08528155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.0009920429056568798,
      "loss": 2.537,
      "step": 549
    },
    {
      "epoch": 0.0854368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0009919981339407715,
      "loss": 2.5304,
      "step": 550
    },
    {
      "epoch": 0.08559223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0009919532376365752,
      "loss": 2.5752,
      "step": 551
    },
    {
      "epoch": 0.08574757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0009919082167556595,
      "loss": 2.5845,
      "step": 552
    },
    {
      "epoch": 0.08590291262135923,
      "grad_norm": 0.0,
      "learning_rate": 0.000991863071309425,
      "loss": 2.6252,
      "step": 553
    },
    {
      "epoch": 0.08605825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.0009918178013093042,
      "loss": 2.5538,
      "step": 554
    },
    {
      "epoch": 0.08621359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0009917724067667604,
      "loss": 2.5345,
      "step": 555
    },
    {
      "epoch": 0.08636893203883494,
      "grad_norm": 0.0,
      "learning_rate": 0.000991726887693289,
      "loss": 2.5056,
      "step": 556
    },
    {
      "epoch": 0.08652427184466019,
      "grad_norm": 0.0,
      "learning_rate": 0.0009916812441004166,
      "loss": 2.5057,
      "step": 557
    },
    {
      "epoch": 0.08667961165048543,
      "grad_norm": 0.0,
      "learning_rate": 0.0009916354759997015,
      "loss": 2.5801,
      "step": 558
    },
    {
      "epoch": 0.08683495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0009915895834027339,
      "loss": 2.538,
      "step": 559
    },
    {
      "epoch": 0.08699029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.0009915435663211342,
      "loss": 2.5447,
      "step": 560
    },
    {
      "epoch": 0.08714563106796117,
      "grad_norm": 0.0,
      "learning_rate": 0.0009914974247665562,
      "loss": 2.6632,
      "step": 561
    },
    {
      "epoch": 0.08730097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.0009914511587506838,
      "loss": 2.5944,
      "step": 562
    },
    {
      "epoch": 0.08745631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0009914047682852332,
      "loss": 2.5179,
      "step": 563
    },
    {
      "epoch": 0.0876116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0009913582533819514,
      "loss": 2.5751,
      "step": 564
    },
    {
      "epoch": 0.08776699029126214,
      "grad_norm": 0.0,
      "learning_rate": 0.0009913116140526173,
      "loss": 2.5765,
      "step": 565
    },
    {
      "epoch": 0.08792233009708737,
      "grad_norm": 0.0,
      "learning_rate": 0.0009912648503090419,
      "loss": 2.4523,
      "step": 566
    },
    {
      "epoch": 0.08807766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0009912179621630664,
      "loss": 2.3948,
      "step": 567
    },
    {
      "epoch": 0.08823300970873786,
      "grad_norm": 0.0,
      "learning_rate": 0.0009911709496265647,
      "loss": 2.493,
      "step": 568
    },
    {
      "epoch": 0.0883883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0009911238127114416,
      "loss": 2.4911,
      "step": 569
    },
    {
      "epoch": 0.08854368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0009910765514296333,
      "loss": 2.5713,
      "step": 570
    },
    {
      "epoch": 0.0886990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0009910291657931082,
      "loss": 2.5389,
      "step": 571
    },
    {
      "epoch": 0.08885436893203884,
      "grad_norm": 0.0,
      "learning_rate": 0.0009909816558138654,
      "loss": 2.5282,
      "step": 572
    },
    {
      "epoch": 0.08900970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.000990934021503936,
      "loss": 2.6815,
      "step": 573
    },
    {
      "epoch": 0.08916504854368933,
      "grad_norm": 0.0,
      "learning_rate": 0.000990886262875382,
      "loss": 2.5994,
      "step": 574
    },
    {
      "epoch": 0.08932038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.0009908383799402975,
      "loss": 2.6595,
      "step": 575
    },
    {
      "epoch": 0.0894757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0009907903727108078,
      "loss": 2.554,
      "step": 576
    },
    {
      "epoch": 0.08963106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0009907422411990696,
      "loss": 2.5005,
      "step": 577
    },
    {
      "epoch": 0.08978640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.0009906939854172716,
      "loss": 2.5616,
      "step": 578
    },
    {
      "epoch": 0.08994174757281553,
      "grad_norm": 0.0,
      "learning_rate": 0.000990645605377633,
      "loss": 2.5596,
      "step": 579
    },
    {
      "epoch": 0.09009708737864078,
      "grad_norm": 0.0,
      "learning_rate": 0.0009905971010924055,
      "loss": 2.5216,
      "step": 580
    },
    {
      "epoch": 0.09025242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0009905484725738715,
      "loss": 2.5349,
      "step": 581
    },
    {
      "epoch": 0.09040776699029127,
      "grad_norm": 0.0,
      "learning_rate": 0.000990499719834345,
      "loss": 2.5648,
      "step": 582
    },
    {
      "epoch": 0.09056310679611651,
      "grad_norm": 0.0,
      "learning_rate": 0.000990450842886172,
      "loss": 2.5363,
      "step": 583
    },
    {
      "epoch": 0.09071844660194174,
      "grad_norm": 0.0,
      "learning_rate": 0.0009904018417417294,
      "loss": 2.5659,
      "step": 584
    },
    {
      "epoch": 0.09087378640776698,
      "grad_norm": 0.0,
      "learning_rate": 0.0009903527164134254,
      "loss": 2.6538,
      "step": 585
    },
    {
      "epoch": 0.09102912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.0009903034669137005,
      "loss": 2.5582,
      "step": 586
    },
    {
      "epoch": 0.09118446601941747,
      "grad_norm": 0.0,
      "learning_rate": 0.0009902540932550255,
      "loss": 2.5536,
      "step": 587
    },
    {
      "epoch": 0.09133980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0009902045954499037,
      "loss": 2.4709,
      "step": 588
    },
    {
      "epoch": 0.09149514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.000990154973510869,
      "loss": 2.4946,
      "step": 589
    },
    {
      "epoch": 0.0916504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0009901052274504875,
      "loss": 2.5936,
      "step": 590
    },
    {
      "epoch": 0.09180582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.000990055357281356,
      "loss": 2.5319,
      "step": 591
    },
    {
      "epoch": 0.0919611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0009900053630161033,
      "loss": 2.5111,
      "step": 592
    },
    {
      "epoch": 0.09211650485436894,
      "grad_norm": 0.0,
      "learning_rate": 0.0009899552446673893,
      "loss": 2.5815,
      "step": 593
    },
    {
      "epoch": 0.09227184466019417,
      "grad_norm": 0.0,
      "learning_rate": 0.0009899050022479053,
      "loss": 2.5995,
      "step": 594
    },
    {
      "epoch": 0.09242718446601941,
      "grad_norm": 0.0,
      "learning_rate": 0.0009898546357703741,
      "loss": 2.5422,
      "step": 595
    },
    {
      "epoch": 0.09258252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.00098980414524755,
      "loss": 2.4867,
      "step": 596
    },
    {
      "epoch": 0.0927378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.000989753530692219,
      "loss": 2.5835,
      "step": 597
    },
    {
      "epoch": 0.09289320388349515,
      "grad_norm": 0.0,
      "learning_rate": 0.0009897027921171976,
      "loss": 2.547,
      "step": 598
    },
    {
      "epoch": 0.09304854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.0009896519295353346,
      "loss": 2.5072,
      "step": 599
    },
    {
      "epoch": 0.09320388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.00098960094295951,
      "loss": 2.4794,
      "step": 600
    },
    {
      "epoch": 0.09335922330097088,
      "grad_norm": 0.0,
      "learning_rate": 0.0009895498324026346,
      "loss": 2.5845,
      "step": 601
    },
    {
      "epoch": 0.09351456310679612,
      "grad_norm": 0.0,
      "learning_rate": 0.0009894985978776516,
      "loss": 2.6394,
      "step": 602
    },
    {
      "epoch": 0.09366990291262135,
      "grad_norm": 0.0,
      "learning_rate": 0.0009894472393975344,
      "loss": 2.5127,
      "step": 603
    },
    {
      "epoch": 0.0938252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0009893957569752892,
      "loss": 2.5072,
      "step": 604
    },
    {
      "epoch": 0.09398058252427184,
      "grad_norm": 0.0,
      "learning_rate": 0.0009893441506239524,
      "loss": 2.5249,
      "step": 605
    },
    {
      "epoch": 0.09413592233009709,
      "grad_norm": 0.0,
      "learning_rate": 0.000989292420356592,
      "loss": 2.5133,
      "step": 606
    },
    {
      "epoch": 0.09429126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0009892405661863083,
      "loss": 2.5093,
      "step": 607
    },
    {
      "epoch": 0.09444660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.0009891885881262315,
      "loss": 2.5618,
      "step": 608
    },
    {
      "epoch": 0.09460194174757282,
      "grad_norm": 0.0,
      "learning_rate": 0.0009891364861895245,
      "loss": 2.6665,
      "step": 609
    },
    {
      "epoch": 0.09475728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.0009890842603893807,
      "loss": 2.5685,
      "step": 610
    },
    {
      "epoch": 0.0949126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0009890319107390252,
      "loss": 2.5597,
      "step": 611
    },
    {
      "epoch": 0.09506796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.0009889794372517147,
      "loss": 2.5907,
      "step": 612
    },
    {
      "epoch": 0.09522330097087378,
      "grad_norm": 0.0,
      "learning_rate": 0.0009889268399407368,
      "loss": 2.5049,
      "step": 613
    },
    {
      "epoch": 0.09537864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0009888741188194105,
      "loss": 2.6002,
      "step": 614
    },
    {
      "epoch": 0.09553398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.0009888212739010865,
      "loss": 2.583,
      "step": 615
    },
    {
      "epoch": 0.09568932038834951,
      "grad_norm": 0.0,
      "learning_rate": 0.0009887683051991467,
      "loss": 2.5616,
      "step": 616
    },
    {
      "epoch": 0.09584466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.000988715212727004,
      "loss": 2.4945,
      "step": 617
    },
    {
      "epoch": 0.096,
      "grad_norm": 0.0,
      "learning_rate": 0.0009886619964981035,
      "loss": 2.5714,
      "step": 618
    },
    {
      "epoch": 0.09615533980582525,
      "grad_norm": 0.0,
      "learning_rate": 0.0009886086565259204,
      "loss": 2.4803,
      "step": 619
    },
    {
      "epoch": 0.09631067961165049,
      "grad_norm": 0.0,
      "learning_rate": 0.0009885551928239624,
      "loss": 2.5276,
      "step": 620
    },
    {
      "epoch": 0.09646601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.000988501605405768,
      "loss": 2.5284,
      "step": 621
    },
    {
      "epoch": 0.09662135922330096,
      "grad_norm": 0.0,
      "learning_rate": 0.000988447894284907,
      "loss": 2.5245,
      "step": 622
    },
    {
      "epoch": 0.09677669902912621,
      "grad_norm": 0.0,
      "learning_rate": 0.0009883940594749804,
      "loss": 2.5302,
      "step": 623
    },
    {
      "epoch": 0.09693203883495145,
      "grad_norm": 0.0,
      "learning_rate": 0.0009883401009896208,
      "loss": 2.4738,
      "step": 624
    },
    {
      "epoch": 0.0970873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0009882860188424926,
      "loss": 2.5651,
      "step": 625
    },
    {
      "epoch": 0.09724271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.00098823181304729,
      "loss": 2.5524,
      "step": 626
    },
    {
      "epoch": 0.09739805825242719,
      "grad_norm": 0.0,
      "learning_rate": 0.0009881774836177403,
      "loss": 2.5786,
      "step": 627
    },
    {
      "epoch": 0.09755339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0009881230305676008,
      "loss": 2.5917,
      "step": 628
    },
    {
      "epoch": 0.09770873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0009880684539106607,
      "loss": 2.5721,
      "step": 629
    },
    {
      "epoch": 0.09786407766990292,
      "grad_norm": 0.0,
      "learning_rate": 0.0009880137536607406,
      "loss": 2.5522,
      "step": 630
    },
    {
      "epoch": 0.09801941747572815,
      "grad_norm": 0.0,
      "learning_rate": 0.0009879589298316918,
      "loss": 2.5345,
      "step": 631
    },
    {
      "epoch": 0.0981747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0009879039824373978,
      "loss": 2.5253,
      "step": 632
    },
    {
      "epoch": 0.09833009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.000987848911491772,
      "loss": 2.5304,
      "step": 633
    },
    {
      "epoch": 0.09848543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.0009877937170087608,
      "loss": 2.405,
      "step": 634
    },
    {
      "epoch": 0.09864077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.0009877383990023404,
      "loss": 2.4384,
      "step": 635
    },
    {
      "epoch": 0.09879611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0009876829574865197,
      "loss": 2.4733,
      "step": 636
    },
    {
      "epoch": 0.09895145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.0009876273924753372,
      "loss": 2.4979,
      "step": 637
    },
    {
      "epoch": 0.09910679611650486,
      "grad_norm": 0.0,
      "learning_rate": 0.000987571703982864,
      "loss": 2.5332,
      "step": 638
    },
    {
      "epoch": 0.0992621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.000987515892023202,
      "loss": 2.5278,
      "step": 639
    },
    {
      "epoch": 0.09941747572815535,
      "grad_norm": 0.0,
      "learning_rate": 0.0009874599566104844,
      "loss": 2.5446,
      "step": 640
    },
    {
      "epoch": 0.09957281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.0009874038977588757,
      "loss": 2.5165,
      "step": 641
    },
    {
      "epoch": 0.09972815533980582,
      "grad_norm": 0.0,
      "learning_rate": 0.0009873477154825716,
      "loss": 2.541,
      "step": 642
    },
    {
      "epoch": 0.09988349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.000987291409795799,
      "loss": 2.5614,
      "step": 643
    },
    {
      "epoch": 0.10003883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.000987234980712816,
      "loss": 2.5596,
      "step": 644
    },
    {
      "epoch": 0.10019417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.0009871784282479124,
      "loss": 2.5439,
      "step": 645
    },
    {
      "epoch": 0.1003495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0009871217524154085,
      "loss": 2.4904,
      "step": 646
    },
    {
      "epoch": 0.10050485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0009870649532296566,
      "loss": 2.5765,
      "step": 647
    },
    {
      "epoch": 0.10066019417475729,
      "grad_norm": 0.0,
      "learning_rate": 0.0009870080307050398,
      "loss": 2.5929,
      "step": 648
    },
    {
      "epoch": 0.10081553398058253,
      "grad_norm": 0.0,
      "learning_rate": 0.0009869509848559723,
      "loss": 2.4614,
      "step": 649
    },
    {
      "epoch": 0.10097087378640776,
      "grad_norm": 0.0,
      "learning_rate": 0.0009868938156969,
      "loss": 2.5574,
      "step": 650
    },
    {
      "epoch": 0.101126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0009868365232423,
      "loss": 2.4595,
      "step": 651
    },
    {
      "epoch": 0.10128155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.00098677910750668,
      "loss": 2.5233,
      "step": 652
    },
    {
      "epoch": 0.1014368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0009867215685045792,
      "loss": 2.5839,
      "step": 653
    },
    {
      "epoch": 0.10159223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0009866639062505685,
      "loss": 2.5342,
      "step": 654
    },
    {
      "epoch": 0.10174757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0009866061207592493,
      "loss": 2.5256,
      "step": 655
    },
    {
      "epoch": 0.10190291262135923,
      "grad_norm": 0.0,
      "learning_rate": 0.000986548212045255,
      "loss": 2.4823,
      "step": 656
    },
    {
      "epoch": 0.10205825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.0009864901801232495,
      "loss": 2.442,
      "step": 657
    },
    {
      "epoch": 0.10221359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.000986432025007928,
      "loss": 2.4845,
      "step": 658
    },
    {
      "epoch": 0.10236893203883494,
      "grad_norm": 0.0,
      "learning_rate": 0.0009863737467140172,
      "loss": 2.4812,
      "step": 659
    },
    {
      "epoch": 0.10252427184466019,
      "grad_norm": 0.0,
      "learning_rate": 0.000986315345256275,
      "loss": 2.454,
      "step": 660
    },
    {
      "epoch": 0.10267961165048543,
      "grad_norm": 0.0,
      "learning_rate": 0.00098625682064949,
      "loss": 2.5189,
      "step": 661
    },
    {
      "epoch": 0.10283495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0009861981729084826,
      "loss": 2.5284,
      "step": 662
    },
    {
      "epoch": 0.10299029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.000986139402048104,
      "loss": 2.519,
      "step": 663
    },
    {
      "epoch": 0.10314563106796117,
      "grad_norm": 0.0,
      "learning_rate": 0.0009860805080832366,
      "loss": 2.5001,
      "step": 664
    },
    {
      "epoch": 0.10330097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.0009860214910287941,
      "loss": 2.5365,
      "step": 665
    },
    {
      "epoch": 0.10345631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0009859623508997216,
      "loss": 2.4591,
      "step": 666
    },
    {
      "epoch": 0.1036116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0009859030877109945,
      "loss": 2.4695,
      "step": 667
    },
    {
      "epoch": 0.10376699029126214,
      "grad_norm": 0.0,
      "learning_rate": 0.0009858437014776203,
      "loss": 2.4977,
      "step": 668
    },
    {
      "epoch": 0.10392233009708737,
      "grad_norm": 0.0,
      "learning_rate": 0.0009857841922146374,
      "loss": 2.4373,
      "step": 669
    },
    {
      "epoch": 0.10407766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.000985724559937115,
      "loss": 2.4788,
      "step": 670
    },
    {
      "epoch": 0.10423300970873786,
      "grad_norm": 0.0,
      "learning_rate": 0.0009856648046601536,
      "loss": 2.5565,
      "step": 671
    },
    {
      "epoch": 0.1043883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0009856049263988855,
      "loss": 2.5381,
      "step": 672
    },
    {
      "epoch": 0.10454368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0009855449251684732,
      "loss": 2.4215,
      "step": 673
    },
    {
      "epoch": 0.1046990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0009854848009841109,
      "loss": 2.4786,
      "step": 674
    },
    {
      "epoch": 0.10485436893203884,
      "grad_norm": 0.0,
      "learning_rate": 0.0009854245538610236,
      "loss": 2.4653,
      "step": 675
    },
    {
      "epoch": 0.10500970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.0009853641838144676,
      "loss": 2.5792,
      "step": 676
    },
    {
      "epoch": 0.10516504854368933,
      "grad_norm": 0.0,
      "learning_rate": 0.0009853036908597306,
      "loss": 2.468,
      "step": 677
    },
    {
      "epoch": 0.10532038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.0009852430750121309,
      "loss": 2.5237,
      "step": 678
    },
    {
      "epoch": 0.1054757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0009851823362870182,
      "loss": 2.5714,
      "step": 679
    },
    {
      "epoch": 0.10563106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0009851214746997735,
      "loss": 2.4639,
      "step": 680
    },
    {
      "epoch": 0.10578640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.0009850604902658085,
      "loss": 2.453,
      "step": 681
    },
    {
      "epoch": 0.10594174757281553,
      "grad_norm": 0.0,
      "learning_rate": 0.0009849993830005662,
      "loss": 2.5071,
      "step": 682
    },
    {
      "epoch": 0.10609708737864078,
      "grad_norm": 0.0,
      "learning_rate": 0.0009849381529195209,
      "loss": 2.5068,
      "step": 683
    },
    {
      "epoch": 0.10625242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0009848768000381776,
      "loss": 2.5597,
      "step": 684
    },
    {
      "epoch": 0.10640776699029127,
      "grad_norm": 0.0,
      "learning_rate": 0.0009848153243720728,
      "loss": 2.4718,
      "step": 685
    },
    {
      "epoch": 0.10656310679611651,
      "grad_norm": 0.0,
      "learning_rate": 0.000984753725936774,
      "loss": 2.4044,
      "step": 686
    },
    {
      "epoch": 0.10671844660194174,
      "grad_norm": 0.0,
      "learning_rate": 0.0009846920047478793,
      "loss": 2.469,
      "step": 687
    },
    {
      "epoch": 0.10687378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0009846301608210186,
      "loss": 2.4189,
      "step": 688
    },
    {
      "epoch": 0.10702912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.0009845681941718523,
      "loss": 2.4244,
      "step": 689
    },
    {
      "epoch": 0.10718446601941747,
      "grad_norm": 0.0,
      "learning_rate": 0.0009845061048160727,
      "loss": 2.4767,
      "step": 690
    },
    {
      "epoch": 0.10733980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.000984443892769402,
      "loss": 2.5309,
      "step": 691
    },
    {
      "epoch": 0.10749514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0009843815580475941,
      "loss": 2.4916,
      "step": 692
    },
    {
      "epoch": 0.1076504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0009843191006664344,
      "loss": 2.572,
      "step": 693
    },
    {
      "epoch": 0.10780582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.0009842565206417387,
      "loss": 2.5242,
      "step": 694
    },
    {
      "epoch": 0.1079611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.000984193817989354,
      "loss": 2.4244,
      "step": 695
    },
    {
      "epoch": 0.10811650485436894,
      "grad_norm": 0.0,
      "learning_rate": 0.0009841309927251585,
      "loss": 2.4489,
      "step": 696
    },
    {
      "epoch": 0.10827184466019417,
      "grad_norm": 0.0,
      "learning_rate": 0.0009840680448650612,
      "loss": 2.4858,
      "step": 697
    },
    {
      "epoch": 0.10842718446601941,
      "grad_norm": 0.0,
      "learning_rate": 0.0009840049744250024,
      "loss": 2.48,
      "step": 698
    },
    {
      "epoch": 0.10858252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0009839417814209534,
      "loss": 2.4994,
      "step": 699
    },
    {
      "epoch": 0.1087378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0009838784658689165,
      "loss": 2.5049,
      "step": 700
    },
    {
      "epoch": 0.10889320388349515,
      "grad_norm": 0.0,
      "learning_rate": 0.0009838150277849248,
      "loss": 2.4519,
      "step": 701
    },
    {
      "epoch": 0.10904854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.000983751467185043,
      "loss": 2.5403,
      "step": 702
    },
    {
      "epoch": 0.10920388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0009836877840853663,
      "loss": 2.5244,
      "step": 703
    },
    {
      "epoch": 0.10935922330097088,
      "grad_norm": 0.0,
      "learning_rate": 0.000983623978502021,
      "loss": 2.5239,
      "step": 704
    },
    {
      "epoch": 0.10951456310679612,
      "grad_norm": 0.0,
      "learning_rate": 0.0009835600504511648,
      "loss": 2.5247,
      "step": 705
    },
    {
      "epoch": 0.10966990291262135,
      "grad_norm": 0.0,
      "learning_rate": 0.0009834959999489856,
      "loss": 2.4373,
      "step": 706
    },
    {
      "epoch": 0.1098252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0009834318270117034,
      "loss": 2.5649,
      "step": 707
    },
    {
      "epoch": 0.10998058252427184,
      "grad_norm": 0.0,
      "learning_rate": 0.0009833675316555683,
      "loss": 2.552,
      "step": 708
    },
    {
      "epoch": 0.11013592233009709,
      "grad_norm": 0.0,
      "learning_rate": 0.0009833031138968618,
      "loss": 2.4327,
      "step": 709
    },
    {
      "epoch": 0.11029126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0009832385737518965,
      "loss": 2.4065,
      "step": 710
    },
    {
      "epoch": 0.11044660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.0009831739112370155,
      "loss": 2.5623,
      "step": 711
    },
    {
      "epoch": 0.11060194174757282,
      "grad_norm": 0.0,
      "learning_rate": 0.0009831091263685936,
      "loss": 2.4717,
      "step": 712
    },
    {
      "epoch": 0.11075728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.0009830442191630358,
      "loss": 2.4071,
      "step": 713
    },
    {
      "epoch": 0.1109126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0009829791896367786,
      "loss": 2.5648,
      "step": 714
    },
    {
      "epoch": 0.11106796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.0009829140378062894,
      "loss": 2.5234,
      "step": 715
    },
    {
      "epoch": 0.11122330097087378,
      "grad_norm": 0.0,
      "learning_rate": 0.0009828487636880667,
      "loss": 2.3832,
      "step": 716
    },
    {
      "epoch": 0.11137864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0009827833672986395,
      "loss": 2.4642,
      "step": 717
    },
    {
      "epoch": 0.11153398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.0009827178486545682,
      "loss": 2.576,
      "step": 718
    },
    {
      "epoch": 0.11168932038834951,
      "grad_norm": 0.0,
      "learning_rate": 0.000982652207772444,
      "loss": 2.5389,
      "step": 719
    },
    {
      "epoch": 0.11184466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.0009825864446688889,
      "loss": 2.5548,
      "step": 720
    },
    {
      "epoch": 0.112,
      "grad_norm": 0.0,
      "learning_rate": 0.000982520559360556,
      "loss": 2.5347,
      "step": 721
    },
    {
      "epoch": 0.11215533980582525,
      "grad_norm": 0.0,
      "learning_rate": 0.00098245455186413,
      "loss": 2.5317,
      "step": 722
    },
    {
      "epoch": 0.11231067961165049,
      "grad_norm": 0.0,
      "learning_rate": 0.0009823884221963248,
      "loss": 2.4324,
      "step": 723
    },
    {
      "epoch": 0.11246601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.0009823221703738874,
      "loss": 2.4847,
      "step": 724
    },
    {
      "epoch": 0.11262135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.000982255796413594,
      "loss": 2.4778,
      "step": 725
    },
    {
      "epoch": 0.11277669902912621,
      "grad_norm": 0.0,
      "learning_rate": 0.0009821893003322524,
      "loss": 2.5511,
      "step": 726
    },
    {
      "epoch": 0.11293203883495145,
      "grad_norm": 0.0,
      "learning_rate": 0.0009821226821467017,
      "loss": 2.4977,
      "step": 727
    },
    {
      "epoch": 0.1130873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0009820559418738113,
      "loss": 2.5264,
      "step": 728
    },
    {
      "epoch": 0.11324271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.0009819890795304818,
      "loss": 2.4785,
      "step": 729
    },
    {
      "epoch": 0.11339805825242719,
      "grad_norm": 0.0,
      "learning_rate": 0.0009819220951336448,
      "loss": 2.4691,
      "step": 730
    },
    {
      "epoch": 0.11355339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0009818549887002624,
      "loss": 2.4819,
      "step": 731
    },
    {
      "epoch": 0.11370873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.000981787760247328,
      "loss": 2.5507,
      "step": 732
    },
    {
      "epoch": 0.11386407766990292,
      "grad_norm": 0.0,
      "learning_rate": 0.000981720409791866,
      "loss": 2.4652,
      "step": 733
    },
    {
      "epoch": 0.11401941747572815,
      "grad_norm": 0.0,
      "learning_rate": 0.000981652937350931,
      "loss": 2.5275,
      "step": 734
    },
    {
      "epoch": 0.1141747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0009815853429416095,
      "loss": 2.626,
      "step": 735
    },
    {
      "epoch": 0.11433009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0009815176265810179,
      "loss": 2.5304,
      "step": 736
    },
    {
      "epoch": 0.11448543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.0009814497882863044,
      "loss": 2.4946,
      "step": 737
    },
    {
      "epoch": 0.11464077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.0009813818280746472,
      "loss": 2.4764,
      "step": 738
    },
    {
      "epoch": 0.11479611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.000981313745963256,
      "loss": 2.5302,
      "step": 739
    },
    {
      "epoch": 0.11495145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.0009812455419693708,
      "loss": 2.6061,
      "step": 740
    },
    {
      "epoch": 0.11510679611650486,
      "grad_norm": 0.0,
      "learning_rate": 0.0009811772161102635,
      "loss": 2.5315,
      "step": 741
    },
    {
      "epoch": 0.1152621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0009811087684032354,
      "loss": 2.5604,
      "step": 742
    },
    {
      "epoch": 0.11541747572815533,
      "grad_norm": 0.0,
      "learning_rate": 0.00098104019886562,
      "loss": 2.4286,
      "step": 743
    },
    {
      "epoch": 0.11557281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.0009809715075147808,
      "loss": 2.5183,
      "step": 744
    },
    {
      "epoch": 0.11572815533980582,
      "grad_norm": 0.0,
      "learning_rate": 0.0009809026943681126,
      "loss": 2.4872,
      "step": 745
    },
    {
      "epoch": 0.11588349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.0009808337594430408,
      "loss": 2.4441,
      "step": 746
    },
    {
      "epoch": 0.11603883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0009807647027570215,
      "loss": 2.5179,
      "step": 747
    },
    {
      "epoch": 0.11619417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.0009806955243275418,
      "loss": 2.5299,
      "step": 748
    },
    {
      "epoch": 0.1163495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0009806262241721204,
      "loss": 2.5299,
      "step": 749
    },
    {
      "epoch": 0.11650485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0009805568023083053,
      "loss": 2.4342,
      "step": 750
    },
    {
      "epoch": 0.11666019417475729,
      "grad_norm": 0.0,
      "learning_rate": 0.0009804872587536764,
      "loss": 2.4962,
      "step": 751
    },
    {
      "epoch": 0.11681553398058253,
      "grad_norm": 0.0,
      "learning_rate": 0.000980417593525844,
      "loss": 2.5161,
      "step": 752
    },
    {
      "epoch": 0.11697087378640776,
      "grad_norm": 0.0,
      "learning_rate": 0.0009803478066424498,
      "loss": 2.4403,
      "step": 753
    },
    {
      "epoch": 0.117126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0009802778981211654,
      "loss": 2.4973,
      "step": 754
    },
    {
      "epoch": 0.11728155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.0009802078679796938,
      "loss": 2.4751,
      "step": 755
    },
    {
      "epoch": 0.1174368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0009801377162357685,
      "loss": 2.5258,
      "step": 756
    },
    {
      "epoch": 0.11759223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.000980067442907154,
      "loss": 2.4702,
      "step": 757
    },
    {
      "epoch": 0.11774757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0009799970480116455,
      "loss": 2.5134,
      "step": 758
    },
    {
      "epoch": 0.11790291262135923,
      "grad_norm": 0.0,
      "learning_rate": 0.0009799265315670693,
      "loss": 2.5416,
      "step": 759
    },
    {
      "epoch": 0.11805825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.000979855893591282,
      "loss": 2.5236,
      "step": 760
    },
    {
      "epoch": 0.11821359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0009797851341021712,
      "loss": 2.4373,
      "step": 761
    },
    {
      "epoch": 0.11836893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.000979714253117655,
      "loss": 2.3916,
      "step": 762
    },
    {
      "epoch": 0.11852427184466019,
      "grad_norm": 0.0,
      "learning_rate": 0.0009796432506556827,
      "loss": 2.4083,
      "step": 763
    },
    {
      "epoch": 0.11867961165048543,
      "grad_norm": 0.0,
      "learning_rate": 0.0009795721267342342,
      "loss": 2.5663,
      "step": 764
    },
    {
      "epoch": 0.11883495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.00097950088137132,
      "loss": 2.5442,
      "step": 765
    },
    {
      "epoch": 0.11899029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.0009794295145849818,
      "loss": 2.4485,
      "step": 766
    },
    {
      "epoch": 0.11914563106796117,
      "grad_norm": 0.0,
      "learning_rate": 0.0009793580263932912,
      "loss": 2.514,
      "step": 767
    },
    {
      "epoch": 0.11930097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.0009792864168143517,
      "loss": 2.6053,
      "step": 768
    },
    {
      "epoch": 0.11945631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0009792146858662964,
      "loss": 2.4712,
      "step": 769
    },
    {
      "epoch": 0.1196116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0009791428335672897,
      "loss": 2.4961,
      "step": 770
    },
    {
      "epoch": 0.11976699029126213,
      "grad_norm": 0.0,
      "learning_rate": 0.000979070859935527,
      "loss": 2.4985,
      "step": 771
    },
    {
      "epoch": 0.11992233009708737,
      "grad_norm": 0.0,
      "learning_rate": 0.0009789987649892339,
      "loss": 2.4712,
      "step": 772
    },
    {
      "epoch": 0.12007766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0009789265487466666,
      "loss": 2.3973,
      "step": 773
    },
    {
      "epoch": 0.12023300970873786,
      "grad_norm": 0.0,
      "learning_rate": 0.0009788542112261127,
      "loss": 2.5507,
      "step": 774
    },
    {
      "epoch": 0.1203883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0009787817524458903,
      "loss": 2.5062,
      "step": 775
    },
    {
      "epoch": 0.12054368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0009787091724243475,
      "loss": 2.6224,
      "step": 776
    },
    {
      "epoch": 0.1206990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0009786364711798642,
      "loss": 2.5383,
      "step": 777
    },
    {
      "epoch": 0.12085436893203884,
      "grad_norm": 0.0,
      "learning_rate": 0.00097856364873085,
      "loss": 2.5095,
      "step": 778
    },
    {
      "epoch": 0.12100970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.000978490705095746,
      "loss": 2.4393,
      "step": 779
    },
    {
      "epoch": 0.12116504854368933,
      "grad_norm": 0.0,
      "learning_rate": 0.0009784176402930233,
      "loss": 2.52,
      "step": 780
    },
    {
      "epoch": 0.12132038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.0009783444543411844,
      "loss": 2.4662,
      "step": 781
    },
    {
      "epoch": 0.1214757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0009782711472587618,
      "loss": 2.3647,
      "step": 782
    },
    {
      "epoch": 0.12163106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.000978197719064319,
      "loss": 2.6273,
      "step": 783
    },
    {
      "epoch": 0.12178640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.0009781241697764502,
      "loss": 2.4874,
      "step": 784
    },
    {
      "epoch": 0.12194174757281553,
      "grad_norm": 0.0,
      "learning_rate": 0.00097805049941378,
      "loss": 2.3964,
      "step": 785
    },
    {
      "epoch": 0.12209708737864078,
      "grad_norm": 0.0,
      "learning_rate": 0.0009779767079949643,
      "loss": 2.4662,
      "step": 786
    },
    {
      "epoch": 0.12225242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0009779027955386889,
      "loss": 2.4742,
      "step": 787
    },
    {
      "epoch": 0.12240776699029127,
      "grad_norm": 0.0,
      "learning_rate": 0.0009778287620636702,
      "loss": 2.47,
      "step": 788
    },
    {
      "epoch": 0.12256310679611651,
      "grad_norm": 0.0,
      "learning_rate": 0.0009777546075886566,
      "loss": 2.4603,
      "step": 789
    },
    {
      "epoch": 0.12271844660194174,
      "grad_norm": 0.0,
      "learning_rate": 0.0009776803321324252,
      "loss": 2.5184,
      "step": 790
    },
    {
      "epoch": 0.12287378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.000977605935713785,
      "loss": 2.5276,
      "step": 791
    },
    {
      "epoch": 0.12302912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.0009775314183515754,
      "loss": 2.5487,
      "step": 792
    },
    {
      "epoch": 0.12318446601941747,
      "grad_norm": 0.0,
      "learning_rate": 0.0009774567800646663,
      "loss": 2.4968,
      "step": 793
    },
    {
      "epoch": 0.12333980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0009773820208719584,
      "loss": 2.4341,
      "step": 794
    },
    {
      "epoch": 0.12349514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0009773071407923825,
      "loss": 2.5131,
      "step": 795
    },
    {
      "epoch": 0.1236504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0009772321398449008,
      "loss": 2.5547,
      "step": 796
    },
    {
      "epoch": 0.12380582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.0009771570180485054,
      "loss": 2.5481,
      "step": 797
    },
    {
      "epoch": 0.1239611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0009770817754222196,
      "loss": 2.3519,
      "step": 798
    },
    {
      "epoch": 0.12411650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.0009770064119850967,
      "loss": 2.4248,
      "step": 799
    },
    {
      "epoch": 0.12427184466019417,
      "grad_norm": 0.0,
      "learning_rate": 0.000976930927756221,
      "loss": 2.495,
      "step": 800
    },
    {
      "epoch": 0.12442718446601941,
      "grad_norm": 0.0,
      "learning_rate": 0.0009768553227547073,
      "loss": 2.5706,
      "step": 801
    },
    {
      "epoch": 0.12458252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0009767795969997009,
      "loss": 2.4497,
      "step": 802
    },
    {
      "epoch": 0.1247378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0009767037505103778,
      "loss": 2.5645,
      "step": 803
    },
    {
      "epoch": 0.12489320388349515,
      "grad_norm": 0.0,
      "learning_rate": 0.0009766277833059445,
      "loss": 2.5128,
      "step": 804
    },
    {
      "epoch": 0.1250485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0009765516954056381,
      "loss": 2.4352,
      "step": 805
    },
    {
      "epoch": 0.12520388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0009764754868287264,
      "loss": 2.4368,
      "step": 806
    },
    {
      "epoch": 0.12535922330097088,
      "grad_norm": 0.0,
      "learning_rate": 0.0009763991575945073,
      "loss": 2.4333,
      "step": 807
    },
    {
      "epoch": 0.12551456310679612,
      "grad_norm": 0.0,
      "learning_rate": 0.0009763227077223097,
      "loss": 2.373,
      "step": 808
    },
    {
      "epoch": 0.12566990291262137,
      "grad_norm": 0.0,
      "learning_rate": 0.000976246137231493,
      "loss": 2.4623,
      "step": 809
    },
    {
      "epoch": 0.1258252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0009761694461414468,
      "loss": 2.5157,
      "step": 810
    },
    {
      "epoch": 0.12598058252427186,
      "grad_norm": 0.0,
      "learning_rate": 0.0009760926344715918,
      "loss": 2.5743,
      "step": 811
    },
    {
      "epoch": 0.1261359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0009760157022413787,
      "loss": 2.4454,
      "step": 812
    },
    {
      "epoch": 0.12629126213592232,
      "grad_norm": 0.0,
      "learning_rate": 0.000975938649470289,
      "loss": 2.4744,
      "step": 813
    },
    {
      "epoch": 0.12644660194174756,
      "grad_norm": 0.0,
      "learning_rate": 0.0009758614761778349,
      "loss": 2.472,
      "step": 814
    },
    {
      "epoch": 0.1266019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.0009757841823835585,
      "loss": 2.5347,
      "step": 815
    },
    {
      "epoch": 0.12675728155339805,
      "grad_norm": 0.0,
      "learning_rate": 0.0009757067681070332,
      "loss": 2.4254,
      "step": 816
    },
    {
      "epoch": 0.1269126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0009756292333678622,
      "loss": 2.5027,
      "step": 817
    },
    {
      "epoch": 0.12706796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.0009755515781856797,
      "loss": 2.514,
      "step": 818
    },
    {
      "epoch": 0.12722330097087378,
      "grad_norm": 0.0,
      "learning_rate": 0.0009754738025801503,
      "loss": 2.5067,
      "step": 819
    },
    {
      "epoch": 0.12737864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0009753959065709688,
      "loss": 2.505,
      "step": 820
    },
    {
      "epoch": 0.12753398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.000975317890177861,
      "loss": 2.4681,
      "step": 821
    },
    {
      "epoch": 0.12768932038834951,
      "grad_norm": 0.0,
      "learning_rate": 0.0009752397534205826,
      "loss": 2.5277,
      "step": 822
    },
    {
      "epoch": 0.12784466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.0009751614963189201,
      "loss": 2.5011,
      "step": 823
    },
    {
      "epoch": 0.128,
      "grad_norm": 0.0,
      "learning_rate": 0.0009750831188926908,
      "loss": 2.423,
      "step": 824
    },
    {
      "epoch": 0.12815533980582525,
      "grad_norm": 0.0,
      "learning_rate": 0.0009750046211617419,
      "loss": 2.4987,
      "step": 825
    },
    {
      "epoch": 0.1283106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.000974926003145951,
      "loss": 2.5356,
      "step": 826
    },
    {
      "epoch": 0.12846601941747574,
      "grad_norm": 0.0,
      "learning_rate": 0.0009748472648652269,
      "loss": 2.4452,
      "step": 827
    },
    {
      "epoch": 0.12862135922330098,
      "grad_norm": 0.0,
      "learning_rate": 0.0009747684063395081,
      "loss": 2.4905,
      "step": 828
    },
    {
      "epoch": 0.12877669902912622,
      "grad_norm": 0.0,
      "learning_rate": 0.000974689427588764,
      "loss": 2.5062,
      "step": 829
    },
    {
      "epoch": 0.12893203883495147,
      "grad_norm": 0.0,
      "learning_rate": 0.0009746103286329943,
      "loss": 2.4436,
      "step": 830
    },
    {
      "epoch": 0.1290873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.000974531109492229,
      "loss": 2.5278,
      "step": 831
    },
    {
      "epoch": 0.12924271844660193,
      "grad_norm": 0.0,
      "learning_rate": 0.0009744517701865288,
      "loss": 2.3538,
      "step": 832
    },
    {
      "epoch": 0.12939805825242717,
      "grad_norm": 0.0,
      "learning_rate": 0.0009743723107359845,
      "loss": 2.4264,
      "step": 833
    },
    {
      "epoch": 0.12955339805825242,
      "grad_norm": 0.0,
      "learning_rate": 0.0009742927311607179,
      "loss": 2.4655,
      "step": 834
    },
    {
      "epoch": 0.12970873786407766,
      "grad_norm": 0.0,
      "learning_rate": 0.0009742130314808804,
      "loss": 2.5419,
      "step": 835
    },
    {
      "epoch": 0.1298640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.0009741332117166546,
      "loss": 2.5599,
      "step": 836
    },
    {
      "epoch": 0.13001941747572815,
      "grad_norm": 0.0,
      "learning_rate": 0.000974053271888253,
      "loss": 2.4421,
      "step": 837
    },
    {
      "epoch": 0.1301747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0009739732120159186,
      "loss": 2.4678,
      "step": 838
    },
    {
      "epoch": 0.13033009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0009738930321199251,
      "loss": 2.5262,
      "step": 839
    },
    {
      "epoch": 0.13048543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.0009738127322205761,
      "loss": 2.4751,
      "step": 840
    },
    {
      "epoch": 0.13064077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.0009737323123382058,
      "loss": 2.4395,
      "step": 841
    },
    {
      "epoch": 0.13079611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0009736517724931793,
      "loss": 2.4845,
      "step": 842
    },
    {
      "epoch": 0.13095145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.0009735711127058911,
      "loss": 2.4999,
      "step": 843
    },
    {
      "epoch": 0.13110679611650486,
      "grad_norm": 0.0,
      "learning_rate": 0.0009734903329967666,
      "loss": 2.4143,
      "step": 844
    },
    {
      "epoch": 0.1312621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0009734094333862619,
      "loss": 2.5657,
      "step": 845
    },
    {
      "epoch": 0.13141747572815535,
      "grad_norm": 0.0,
      "learning_rate": 0.0009733284138948629,
      "loss": 2.4896,
      "step": 846
    },
    {
      "epoch": 0.1315728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.0009732472745430862,
      "loss": 2.49,
      "step": 847
    },
    {
      "epoch": 0.13172815533980584,
      "grad_norm": 0.0,
      "learning_rate": 0.0009731660153514783,
      "loss": 2.4736,
      "step": 848
    },
    {
      "epoch": 0.13188349514563108,
      "grad_norm": 0.0,
      "learning_rate": 0.0009730846363406167,
      "loss": 2.3929,
      "step": 849
    },
    {
      "epoch": 0.13203883495145632,
      "grad_norm": 0.0,
      "learning_rate": 0.0009730031375311089,
      "loss": 2.4768,
      "step": 850
    },
    {
      "epoch": 0.13219417475728154,
      "grad_norm": 0.0,
      "learning_rate": 0.0009729215189435924,
      "loss": 2.5106,
      "step": 851
    },
    {
      "epoch": 0.13234951456310678,
      "grad_norm": 0.0,
      "learning_rate": 0.0009728397805987357,
      "loss": 2.467,
      "step": 852
    },
    {
      "epoch": 0.13250485436893203,
      "grad_norm": 0.0,
      "learning_rate": 0.0009727579225172372,
      "loss": 2.4136,
      "step": 853
    },
    {
      "epoch": 0.13266019417475727,
      "grad_norm": 0.0,
      "learning_rate": 0.0009726759447198257,
      "loss": 2.4777,
      "step": 854
    },
    {
      "epoch": 0.13281553398058252,
      "grad_norm": 0.0,
      "learning_rate": 0.0009725938472272602,
      "loss": 2.4401,
      "step": 855
    },
    {
      "epoch": 0.13297087378640776,
      "grad_norm": 0.0,
      "learning_rate": 0.0009725116300603305,
      "loss": 2.4506,
      "step": 856
    },
    {
      "epoch": 0.133126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0009724292932398561,
      "loss": 2.4293,
      "step": 857
    },
    {
      "epoch": 0.13328155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.0009723468367866873,
      "loss": 2.4497,
      "step": 858
    },
    {
      "epoch": 0.1334368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0009722642607217041,
      "loss": 2.4937,
      "step": 859
    },
    {
      "epoch": 0.13359223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0009721815650658173,
      "loss": 2.4463,
      "step": 860
    },
    {
      "epoch": 0.13374757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0009720987498399679,
      "loss": 2.4717,
      "step": 861
    },
    {
      "epoch": 0.13390291262135923,
      "grad_norm": 0.0,
      "learning_rate": 0.0009720158150651271,
      "loss": 2.3561,
      "step": 862
    },
    {
      "epoch": 0.13405825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.000971932760762296,
      "loss": 2.4904,
      "step": 863
    },
    {
      "epoch": 0.13421359223300972,
      "grad_norm": 0.0,
      "learning_rate": 0.0009718495869525068,
      "loss": 2.5035,
      "step": 864
    },
    {
      "epoch": 0.13436893203883496,
      "grad_norm": 0.0,
      "learning_rate": 0.0009717662936568214,
      "loss": 2.481,
      "step": 865
    },
    {
      "epoch": 0.1345242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0009716828808963319,
      "loss": 2.4144,
      "step": 866
    },
    {
      "epoch": 0.13467961165048545,
      "grad_norm": 0.0,
      "learning_rate": 0.0009715993486921609,
      "loss": 2.4158,
      "step": 867
    },
    {
      "epoch": 0.1348349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.0009715156970654613,
      "loss": 2.3887,
      "step": 868
    },
    {
      "epoch": 0.1349902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.0009714319260374158,
      "loss": 2.4468,
      "step": 869
    },
    {
      "epoch": 0.13514563106796115,
      "grad_norm": 0.0,
      "learning_rate": 0.0009713480356292379,
      "loss": 2.5046,
      "step": 870
    },
    {
      "epoch": 0.1353009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0009712640258621709,
      "loss": 2.476,
      "step": 871
    },
    {
      "epoch": 0.13545631067961164,
      "grad_norm": 0.0,
      "learning_rate": 0.0009711798967574885,
      "loss": 2.4781,
      "step": 872
    },
    {
      "epoch": 0.13561165048543689,
      "grad_norm": 0.0,
      "learning_rate": 0.0009710956483364948,
      "loss": 2.5266,
      "step": 873
    },
    {
      "epoch": 0.13576699029126213,
      "grad_norm": 0.0,
      "learning_rate": 0.0009710112806205238,
      "loss": 2.3896,
      "step": 874
    },
    {
      "epoch": 0.13592233009708737,
      "grad_norm": 0.0,
      "learning_rate": 0.0009709267936309396,
      "loss": 2.5521,
      "step": 875
    },
    {
      "epoch": 0.13607766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0009708421873891372,
      "loss": 2.512,
      "step": 876
    },
    {
      "epoch": 0.13623300970873786,
      "grad_norm": 0.0,
      "learning_rate": 0.0009707574619165411,
      "loss": 2.4603,
      "step": 877
    },
    {
      "epoch": 0.1363883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0009706726172346061,
      "loss": 2.4783,
      "step": 878
    },
    {
      "epoch": 0.13654368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0009705876533648175,
      "loss": 2.5139,
      "step": 879
    },
    {
      "epoch": 0.1366990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0009705025703286905,
      "loss": 2.4447,
      "step": 880
    },
    {
      "epoch": 0.13685436893203884,
      "grad_norm": 0.0,
      "learning_rate": 0.0009704173681477706,
      "loss": 2.4973,
      "step": 881
    },
    {
      "epoch": 0.13700970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.0009703320468436336,
      "loss": 2.4357,
      "step": 882
    },
    {
      "epoch": 0.13716504854368933,
      "grad_norm": 0.0,
      "learning_rate": 0.000970246606437885,
      "loss": 2.4247,
      "step": 883
    },
    {
      "epoch": 0.13732038834951457,
      "grad_norm": 0.0,
      "learning_rate": 0.0009701610469521611,
      "loss": 2.4951,
      "step": 884
    },
    {
      "epoch": 0.13747572815533982,
      "grad_norm": 0.0,
      "learning_rate": 0.0009700753684081277,
      "loss": 2.4149,
      "step": 885
    },
    {
      "epoch": 0.13763106796116506,
      "grad_norm": 0.0,
      "learning_rate": 0.0009699895708274813,
      "loss": 2.5203,
      "step": 886
    },
    {
      "epoch": 0.1377864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0009699036542319484,
      "loss": 2.4485,
      "step": 887
    },
    {
      "epoch": 0.13794174757281552,
      "grad_norm": 0.0,
      "learning_rate": 0.0009698176186432852,
      "loss": 2.4703,
      "step": 888
    },
    {
      "epoch": 0.13809708737864076,
      "grad_norm": 0.0,
      "learning_rate": 0.0009697314640832786,
      "loss": 2.5183,
      "step": 889
    },
    {
      "epoch": 0.138252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0009696451905737456,
      "loss": 2.4388,
      "step": 890
    },
    {
      "epoch": 0.13840776699029125,
      "grad_norm": 0.0,
      "learning_rate": 0.0009695587981365328,
      "loss": 2.4515,
      "step": 891
    },
    {
      "epoch": 0.1385631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0009694722867935173,
      "loss": 2.4188,
      "step": 892
    },
    {
      "epoch": 0.13871844660194174,
      "grad_norm": 0.0,
      "learning_rate": 0.0009693856565666064,
      "loss": 2.5212,
      "step": 893
    },
    {
      "epoch": 0.13887378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0009692989074777375,
      "loss": 2.4557,
      "step": 894
    },
    {
      "epoch": 0.13902912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.0009692120395488777,
      "loss": 2.505,
      "step": 895
    },
    {
      "epoch": 0.13918446601941747,
      "grad_norm": 0.0,
      "learning_rate": 0.0009691250528020245,
      "loss": 2.4103,
      "step": 896
    },
    {
      "epoch": 0.13933980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0009690379472592056,
      "loss": 2.4177,
      "step": 897
    },
    {
      "epoch": 0.13949514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0009689507229424784,
      "loss": 2.5507,
      "step": 898
    },
    {
      "epoch": 0.1396504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0009688633798739309,
      "loss": 2.4691,
      "step": 899
    },
    {
      "epoch": 0.13980582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.0009687759180756805,
      "loss": 2.3694,
      "step": 900
    },
    {
      "epoch": 0.1399611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0009686883375698754,
      "loss": 2.4697,
      "step": 901
    },
    {
      "epoch": 0.14011650485436894,
      "grad_norm": 0.0,
      "learning_rate": 0.0009686006383786933,
      "loss": 2.5023,
      "step": 902
    },
    {
      "epoch": 0.14027184466019418,
      "grad_norm": 0.0,
      "learning_rate": 0.0009685128205243423,
      "loss": 2.5541,
      "step": 903
    },
    {
      "epoch": 0.14042718446601943,
      "grad_norm": 0.0,
      "learning_rate": 0.0009684248840290604,
      "loss": 2.4496,
      "step": 904
    },
    {
      "epoch": 0.14058252427184467,
      "grad_norm": 0.0,
      "learning_rate": 0.0009683368289151156,
      "loss": 2.4453,
      "step": 905
    },
    {
      "epoch": 0.14073786407766992,
      "grad_norm": 0.0,
      "learning_rate": 0.0009682486552048058,
      "loss": 2.3519,
      "step": 906
    },
    {
      "epoch": 0.14089320388349513,
      "grad_norm": 0.0,
      "learning_rate": 0.0009681603629204592,
      "loss": 2.4979,
      "step": 907
    },
    {
      "epoch": 0.14104854368932038,
      "grad_norm": 0.0,
      "learning_rate": 0.0009680719520844343,
      "loss": 2.4523,
      "step": 908
    },
    {
      "epoch": 0.14120388349514562,
      "grad_norm": 0.0,
      "learning_rate": 0.0009679834227191189,
      "loss": 2.478,
      "step": 909
    },
    {
      "epoch": 0.14135922330097087,
      "grad_norm": 0.0,
      "learning_rate": 0.0009678947748469314,
      "loss": 2.3096,
      "step": 910
    },
    {
      "epoch": 0.1415145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.0009678060084903198,
      "loss": 2.5011,
      "step": 911
    },
    {
      "epoch": 0.14166990291262135,
      "grad_norm": 0.0,
      "learning_rate": 0.0009677171236717624,
      "loss": 2.3915,
      "step": 912
    },
    {
      "epoch": 0.1418252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0009676281204137673,
      "loss": 2.4191,
      "step": 913
    },
    {
      "epoch": 0.14198058252427184,
      "grad_norm": 0.0,
      "learning_rate": 0.0009675389987388727,
      "loss": 2.469,
      "step": 914
    },
    {
      "epoch": 0.1421359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0009674497586696472,
      "loss": 2.4928,
      "step": 915
    },
    {
      "epoch": 0.14229126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0009673604002286882,
      "loss": 2.4482,
      "step": 916
    },
    {
      "epoch": 0.14244660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.0009672709234386243,
      "loss": 2.4635,
      "step": 917
    },
    {
      "epoch": 0.14260194174757282,
      "grad_norm": 0.0,
      "learning_rate": 0.0009671813283221136,
      "loss": 2.4557,
      "step": 918
    },
    {
      "epoch": 0.14275728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.0009670916149018442,
      "loss": 2.4099,
      "step": 919
    },
    {
      "epoch": 0.1429126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0009670017832005338,
      "loss": 2.4346,
      "step": 920
    },
    {
      "epoch": 0.14306796116504855,
      "grad_norm": 0.0,
      "learning_rate": 0.0009669118332409307,
      "loss": 2.3918,
      "step": 921
    },
    {
      "epoch": 0.1432233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.0009668217650458127,
      "loss": 2.4326,
      "step": 922
    },
    {
      "epoch": 0.14337864077669904,
      "grad_norm": 0.0,
      "learning_rate": 0.0009667315786379875,
      "loss": 2.4534,
      "step": 923
    },
    {
      "epoch": 0.14353398058252428,
      "grad_norm": 0.0,
      "learning_rate": 0.0009666412740402932,
      "loss": 2.4972,
      "step": 924
    },
    {
      "epoch": 0.1436893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0009665508512755973,
      "loss": 2.4875,
      "step": 925
    },
    {
      "epoch": 0.14384466019417474,
      "grad_norm": 0.0,
      "learning_rate": 0.0009664603103667975,
      "loss": 2.4827,
      "step": 926
    },
    {
      "epoch": 0.144,
      "grad_norm": 0.0,
      "learning_rate": 0.0009663696513368214,
      "loss": 2.4769,
      "step": 927
    },
    {
      "epoch": 0.14415533980582523,
      "grad_norm": 0.0,
      "learning_rate": 0.0009662788742086266,
      "loss": 2.5215,
      "step": 928
    },
    {
      "epoch": 0.14431067961165048,
      "grad_norm": 0.0,
      "learning_rate": 0.0009661879790052001,
      "loss": 2.4982,
      "step": 929
    },
    {
      "epoch": 0.14446601941747572,
      "grad_norm": 0.0,
      "learning_rate": 0.0009660969657495595,
      "loss": 2.3695,
      "step": 930
    },
    {
      "epoch": 0.14462135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.000966005834464752,
      "loss": 2.5542,
      "step": 931
    },
    {
      "epoch": 0.1447766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0009659145851738545,
      "loss": 2.4993,
      "step": 932
    },
    {
      "epoch": 0.14493203883495145,
      "grad_norm": 0.0,
      "learning_rate": 0.000965823217899974,
      "loss": 2.436,
      "step": 933
    },
    {
      "epoch": 0.1450873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0009657317326662473,
      "loss": 2.4784,
      "step": 934
    },
    {
      "epoch": 0.14524271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.0009656401294958412,
      "loss": 2.3896,
      "step": 935
    },
    {
      "epoch": 0.1453980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0009655484084119521,
      "loss": 2.4752,
      "step": 936
    },
    {
      "epoch": 0.14555339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0009654565694378064,
      "loss": 2.4283,
      "step": 937
    },
    {
      "epoch": 0.14570873786407768,
      "grad_norm": 0.0,
      "learning_rate": 0.0009653646125966606,
      "loss": 2.436,
      "step": 938
    },
    {
      "epoch": 0.14586407766990292,
      "grad_norm": 0.0,
      "learning_rate": 0.0009652725379118008,
      "loss": 2.4319,
      "step": 939
    },
    {
      "epoch": 0.14601941747572816,
      "grad_norm": 0.0,
      "learning_rate": 0.0009651803454065427,
      "loss": 2.4517,
      "step": 940
    },
    {
      "epoch": 0.1461747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0009650880351042324,
      "loss": 2.4954,
      "step": 941
    },
    {
      "epoch": 0.14633009708737865,
      "grad_norm": 0.0,
      "learning_rate": 0.0009649956070282453,
      "loss": 2.4545,
      "step": 942
    },
    {
      "epoch": 0.1464854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.0009649030612019868,
      "loss": 2.4956,
      "step": 943
    },
    {
      "epoch": 0.1466407766990291,
      "grad_norm": 0.0,
      "learning_rate": 0.0009648103976488924,
      "loss": 2.4351,
      "step": 944
    },
    {
      "epoch": 0.14679611650485436,
      "grad_norm": 0.0,
      "learning_rate": 0.000964717616392427,
      "loss": 2.481,
      "step": 945
    },
    {
      "epoch": 0.1469514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0009646247174560856,
      "loss": 2.496,
      "step": 946
    },
    {
      "epoch": 0.14710679611650485,
      "grad_norm": 0.0,
      "learning_rate": 0.000964531700863393,
      "loss": 2.3937,
      "step": 947
    },
    {
      "epoch": 0.1472621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0009644385666379033,
      "loss": 2.4243,
      "step": 948
    },
    {
      "epoch": 0.14741747572815533,
      "grad_norm": 0.0,
      "learning_rate": 0.0009643453148032011,
      "loss": 2.4844,
      "step": 949
    },
    {
      "epoch": 0.14757281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.0009642519453829002,
      "loss": 2.4844,
      "step": 950
    },
    {
      "epoch": 0.14772815533980582,
      "grad_norm": 0.0,
      "learning_rate": 0.0009641584584006446,
      "loss": 2.501,
      "step": 951
    },
    {
      "epoch": 0.14788349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.000964064853880108,
      "loss": 2.4219,
      "step": 952
    },
    {
      "epoch": 0.1480388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0009639711318449936,
      "loss": 2.4633,
      "step": 953
    },
    {
      "epoch": 0.14819417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.0009638772923190342,
      "loss": 2.4996,
      "step": 954
    },
    {
      "epoch": 0.1483495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0009637833353259933,
      "loss": 2.4159,
      "step": 955
    },
    {
      "epoch": 0.14850485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0009636892608896632,
      "loss": 2.4779,
      "step": 956
    },
    {
      "epoch": 0.1486601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.0009635950690338662,
      "loss": 2.4391,
      "step": 957
    },
    {
      "epoch": 0.14881553398058253,
      "grad_norm": 0.0,
      "learning_rate": 0.0009635007597824543,
      "loss": 2.4674,
      "step": 958
    },
    {
      "epoch": 0.14897087378640778,
      "grad_norm": 0.0,
      "learning_rate": 0.0009634063331593097,
      "loss": 2.5388,
      "step": 959
    },
    {
      "epoch": 0.14912621359223302,
      "grad_norm": 0.0,
      "learning_rate": 0.0009633117891883436,
      "loss": 2.4879,
      "step": 960
    },
    {
      "epoch": 0.14928155339805826,
      "grad_norm": 0.0,
      "learning_rate": 0.0009632171278934974,
      "loss": 2.3917,
      "step": 961
    },
    {
      "epoch": 0.1494368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.000963122349298742,
      "loss": 2.4009,
      "step": 962
    },
    {
      "epoch": 0.14959223300970872,
      "grad_norm": 0.0,
      "learning_rate": 0.000963027453428078,
      "loss": 2.4435,
      "step": 963
    },
    {
      "epoch": 0.14974757281553397,
      "grad_norm": 0.0,
      "learning_rate": 0.0009629324403055362,
      "loss": 2.4619,
      "step": 964
    },
    {
      "epoch": 0.1499029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.000962837309955176,
      "loss": 2.4657,
      "step": 965
    },
    {
      "epoch": 0.15005825242718446,
      "grad_norm": 0.0,
      "learning_rate": 0.0009627420624010876,
      "loss": 2.4797,
      "step": 966
    },
    {
      "epoch": 0.1502135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0009626466976673902,
      "loss": 2.5398,
      "step": 967
    },
    {
      "epoch": 0.15036893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.000962551215778233,
      "loss": 2.4292,
      "step": 968
    },
    {
      "epoch": 0.1505242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0009624556167577949,
      "loss": 2.4249,
      "step": 969
    },
    {
      "epoch": 0.15067961165048543,
      "grad_norm": 0.0,
      "learning_rate": 0.000962359900630284,
      "loss": 2.4507,
      "step": 970
    },
    {
      "epoch": 0.15083495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0009622640674199386,
      "loss": 2.4125,
      "step": 971
    },
    {
      "epoch": 0.15099029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.0009621681171510264,
      "loss": 2.5049,
      "step": 972
    },
    {
      "epoch": 0.15114563106796117,
      "grad_norm": 0.0,
      "learning_rate": 0.0009620720498478446,
      "loss": 2.4515,
      "step": 973
    },
    {
      "epoch": 0.1513009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0009619758655347205,
      "loss": 2.4414,
      "step": 974
    },
    {
      "epoch": 0.15145631067961166,
      "grad_norm": 0.0,
      "learning_rate": 0.0009618795642360106,
      "loss": 2.4388,
      "step": 975
    },
    {
      "epoch": 0.1516116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0009617831459761009,
      "loss": 2.4926,
      "step": 976
    },
    {
      "epoch": 0.15176699029126214,
      "grad_norm": 0.0,
      "learning_rate": 0.0009616866107794077,
      "loss": 2.4908,
      "step": 977
    },
    {
      "epoch": 0.1519223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0009615899586703764,
      "loss": 2.5341,
      "step": 978
    },
    {
      "epoch": 0.15207766990291263,
      "grad_norm": 0.0,
      "learning_rate": 0.000961493189673482,
      "loss": 2.4446,
      "step": 979
    },
    {
      "epoch": 0.15223300970873788,
      "grad_norm": 0.0,
      "learning_rate": 0.0009613963038132289,
      "loss": 2.5045,
      "step": 980
    },
    {
      "epoch": 0.15238834951456312,
      "grad_norm": 0.0,
      "learning_rate": 0.0009612993011141519,
      "loss": 2.4348,
      "step": 981
    },
    {
      "epoch": 0.15254368932038834,
      "grad_norm": 0.0,
      "learning_rate": 0.0009612021816008149,
      "loss": 2.5323,
      "step": 982
    },
    {
      "epoch": 0.15269902912621358,
      "grad_norm": 0.0,
      "learning_rate": 0.0009611049452978109,
      "loss": 2.3979,
      "step": 983
    },
    {
      "epoch": 0.15285436893203883,
      "grad_norm": 0.0,
      "learning_rate": 0.0009610075922297632,
      "loss": 2.4801,
      "step": 984
    },
    {
      "epoch": 0.15300970873786407,
      "grad_norm": 0.0,
      "learning_rate": 0.0009609101224213242,
      "loss": 2.418,
      "step": 985
    },
    {
      "epoch": 0.1531650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.0009608125358971765,
      "loss": 2.439,
      "step": 986
    },
    {
      "epoch": 0.15332038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.0009607148326820312,
      "loss": 2.5261,
      "step": 987
    },
    {
      "epoch": 0.1534757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0009606170128006301,
      "loss": 2.5215,
      "step": 988
    },
    {
      "epoch": 0.15363106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0009605190762777438,
      "loss": 2.4553,
      "step": 989
    },
    {
      "epoch": 0.1537864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0009604210231381726,
      "loss": 2.4121,
      "step": 990
    },
    {
      "epoch": 0.15394174757281553,
      "grad_norm": 0.0,
      "learning_rate": 0.0009603228534067465,
      "loss": 2.5115,
      "step": 991
    },
    {
      "epoch": 0.15409708737864078,
      "grad_norm": 0.0,
      "learning_rate": 0.0009602245671083247,
      "loss": 2.4517,
      "step": 992
    },
    {
      "epoch": 0.15425242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0009601261642677965,
      "loss": 2.5308,
      "step": 993
    },
    {
      "epoch": 0.15440776699029127,
      "grad_norm": 0.0,
      "learning_rate": 0.0009600276449100798,
      "loss": 2.4489,
      "step": 994
    },
    {
      "epoch": 0.1545631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.000959929009060123,
      "loss": 2.5215,
      "step": 995
    },
    {
      "epoch": 0.15471844660194176,
      "grad_norm": 0.0,
      "learning_rate": 0.0009598302567429033,
      "loss": 2.4883,
      "step": 996
    },
    {
      "epoch": 0.154873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0009597313879834277,
      "loss": 2.3793,
      "step": 997
    },
    {
      "epoch": 0.15502912621359224,
      "grad_norm": 0.0,
      "learning_rate": 0.0009596324028067325,
      "loss": 2.473,
      "step": 998
    },
    {
      "epoch": 0.1551844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.0009595333012378839,
      "loss": 2.4257,
      "step": 999
    },
    {
      "epoch": 0.1553398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.0009594340833019769,
      "loss": 2.4243,
      "step": 1000
    },
    {
      "epoch": 0.15549514563106795,
      "grad_norm": 0.0,
      "learning_rate": 0.0009593347490241365,
      "loss": 2.4453,
      "step": 1001
    },
    {
      "epoch": 0.1556504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0009592352984295172,
      "loss": 2.4933,
      "step": 1002
    },
    {
      "epoch": 0.15580582524271844,
      "grad_norm": 0.0,
      "learning_rate": 0.0009591357315433022,
      "loss": 2.5322,
      "step": 1003
    },
    {
      "epoch": 0.15596116504854368,
      "grad_norm": 0.0,
      "learning_rate": 0.0009590360483907054,
      "loss": 2.4305,
      "step": 1004
    },
    {
      "epoch": 0.15611650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.000958936248996969,
      "loss": 2.5562,
      "step": 1005
    },
    {
      "epoch": 0.15627184466019417,
      "grad_norm": 0.0,
      "learning_rate": 0.000958836333387365,
      "loss": 2.4481,
      "step": 1006
    },
    {
      "epoch": 0.15642718446601941,
      "grad_norm": 0.0,
      "learning_rate": 0.0009587363015871953,
      "loss": 2.5343,
      "step": 1007
    },
    {
      "epoch": 0.15658252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0009586361536217905,
      "loss": 2.4413,
      "step": 1008
    },
    {
      "epoch": 0.1567378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0009585358895165113,
      "loss": 2.5447,
      "step": 1009
    },
    {
      "epoch": 0.15689320388349515,
      "grad_norm": 0.0,
      "learning_rate": 0.0009584355092967469,
      "loss": 2.4479,
      "step": 1010
    },
    {
      "epoch": 0.1570485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0009583350129879169,
      "loss": 2.4984,
      "step": 1011
    },
    {
      "epoch": 0.15720388349514564,
      "grad_norm": 0.0,
      "learning_rate": 0.0009582344006154698,
      "loss": 2.41,
      "step": 1012
    },
    {
      "epoch": 0.15735922330097088,
      "grad_norm": 0.0,
      "learning_rate": 0.0009581336722048835,
      "loss": 2.4177,
      "step": 1013
    },
    {
      "epoch": 0.15751456310679612,
      "grad_norm": 0.0,
      "learning_rate": 0.0009580328277816652,
      "loss": 2.4623,
      "step": 1014
    },
    {
      "epoch": 0.15766990291262137,
      "grad_norm": 0.0,
      "learning_rate": 0.0009579318673713518,
      "loss": 2.4475,
      "step": 1015
    },
    {
      "epoch": 0.1578252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0009578307909995094,
      "loss": 2.4359,
      "step": 1016
    },
    {
      "epoch": 0.15798058252427186,
      "grad_norm": 0.0,
      "learning_rate": 0.0009577295986917333,
      "loss": 2.5063,
      "step": 1017
    },
    {
      "epoch": 0.1581359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0009576282904736483,
      "loss": 2.4442,
      "step": 1018
    },
    {
      "epoch": 0.15829126213592232,
      "grad_norm": 0.0,
      "learning_rate": 0.0009575268663709087,
      "loss": 2.5548,
      "step": 1019
    },
    {
      "epoch": 0.15844660194174756,
      "grad_norm": 0.0,
      "learning_rate": 0.0009574253264091979,
      "loss": 2.4042,
      "step": 1020
    },
    {
      "epoch": 0.1586019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.0009573236706142288,
      "loss": 2.4158,
      "step": 1021
    },
    {
      "epoch": 0.15875728155339805,
      "grad_norm": 0.0,
      "learning_rate": 0.0009572218990117434,
      "loss": 2.481,
      "step": 1022
    },
    {
      "epoch": 0.1589126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0009571200116275134,
      "loss": 2.446,
      "step": 1023
    },
    {
      "epoch": 0.15906796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.0009570180084873396,
      "loss": 2.4064,
      "step": 1024
    },
    {
      "epoch": 0.15922330097087378,
      "grad_norm": 0.0,
      "learning_rate": 0.0009569158896170521,
      "loss": 2.3248,
      "step": 1025
    },
    {
      "epoch": 0.15937864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0009568136550425104,
      "loss": 2.5029,
      "step": 1026
    },
    {
      "epoch": 0.15953398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.000956711304789603,
      "loss": 2.5083,
      "step": 1027
    },
    {
      "epoch": 0.15968932038834951,
      "grad_norm": 0.0,
      "learning_rate": 0.000956608838884248,
      "loss": 2.4384,
      "step": 1028
    },
    {
      "epoch": 0.15984466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.0009565062573523931,
      "loss": 2.421,
      "step": 1029
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.0,
      "learning_rate": 0.0009564035602200144,
      "loss": 2.4102,
      "step": 1030
    },
    {
      "epoch": 0.16015533980582525,
      "grad_norm": 0.0,
      "learning_rate": 0.000956300747513118,
      "loss": 2.501,
      "step": 1031
    },
    {
      "epoch": 0.1603106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0009561978192577391,
      "loss": 2.3918,
      "step": 1032
    },
    {
      "epoch": 0.16046601941747574,
      "grad_norm": 0.0,
      "learning_rate": 0.000956094775479942,
      "loss": 2.4574,
      "step": 1033
    },
    {
      "epoch": 0.16062135922330098,
      "grad_norm": 0.0,
      "learning_rate": 0.0009559916162058202,
      "loss": 2.3966,
      "step": 1034
    },
    {
      "epoch": 0.16077669902912622,
      "grad_norm": 0.0,
      "learning_rate": 0.000955888341461497,
      "loss": 2.3875,
      "step": 1035
    },
    {
      "epoch": 0.16093203883495147,
      "grad_norm": 0.0,
      "learning_rate": 0.0009557849512731241,
      "loss": 2.42,
      "step": 1036
    },
    {
      "epoch": 0.1610873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0009556814456668831,
      "loss": 2.4132,
      "step": 1037
    },
    {
      "epoch": 0.16124271844660193,
      "grad_norm": 0.0,
      "learning_rate": 0.0009555778246689846,
      "loss": 2.4281,
      "step": 1038
    },
    {
      "epoch": 0.16139805825242717,
      "grad_norm": 0.0,
      "learning_rate": 0.0009554740883056683,
      "loss": 2.4731,
      "step": 1039
    },
    {
      "epoch": 0.16155339805825242,
      "grad_norm": 0.0,
      "learning_rate": 0.0009553702366032034,
      "loss": 2.4668,
      "step": 1040
    },
    {
      "epoch": 0.16170873786407766,
      "grad_norm": 0.0,
      "learning_rate": 0.000955266269587888,
      "loss": 2.4685,
      "step": 1041
    },
    {
      "epoch": 0.1618640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.0009551621872860497,
      "loss": 2.4423,
      "step": 1042
    },
    {
      "epoch": 0.16201941747572815,
      "grad_norm": 0.0,
      "learning_rate": 0.0009550579897240447,
      "loss": 2.4265,
      "step": 1043
    },
    {
      "epoch": 0.1621747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0009549536769282597,
      "loss": 2.3021,
      "step": 1044
    },
    {
      "epoch": 0.16233009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0009548492489251087,
      "loss": 2.3684,
      "step": 1045
    },
    {
      "epoch": 0.16248543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.0009547447057410366,
      "loss": 2.4402,
      "step": 1046
    },
    {
      "epoch": 0.16264077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.0009546400474025164,
      "loss": 2.4159,
      "step": 1047
    },
    {
      "epoch": 0.16279611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0009545352739360504,
      "loss": 2.4142,
      "step": 1048
    },
    {
      "epoch": 0.16295145631067962,
      "grad_norm": 0.0,
      "learning_rate": 0.0009544303853681708,
      "loss": 2.5223,
      "step": 1049
    },
    {
      "epoch": 0.16310679611650486,
      "grad_norm": 0.0,
      "learning_rate": 0.0009543253817254382,
      "loss": 2.4553,
      "step": 1050
    },
    {
      "epoch": 0.1632621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0009542202630344423,
      "loss": 2.4797,
      "step": 1051
    },
    {
      "epoch": 0.16341747572815535,
      "grad_norm": 0.0,
      "learning_rate": 0.0009541150293218025,
      "loss": 2.4015,
      "step": 1052
    },
    {
      "epoch": 0.1635728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.0009540096806141667,
      "loss": 2.4252,
      "step": 1053
    },
    {
      "epoch": 0.16372815533980584,
      "grad_norm": 0.0,
      "learning_rate": 0.0009539042169382126,
      "loss": 2.5342,
      "step": 1054
    },
    {
      "epoch": 0.16388349514563108,
      "grad_norm": 0.0,
      "learning_rate": 0.0009537986383206463,
      "loss": 2.3237,
      "step": 1055
    },
    {
      "epoch": 0.1640388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0009536929447882034,
      "loss": 2.4946,
      "step": 1056
    },
    {
      "epoch": 0.16419417475728154,
      "grad_norm": 0.0,
      "learning_rate": 0.0009535871363676488,
      "loss": 2.4154,
      "step": 1057
    },
    {
      "epoch": 0.16434951456310679,
      "grad_norm": 0.0,
      "learning_rate": 0.000953481213085776,
      "loss": 2.3904,
      "step": 1058
    },
    {
      "epoch": 0.16450485436893203,
      "grad_norm": 0.0,
      "learning_rate": 0.0009533751749694078,
      "loss": 2.4651,
      "step": 1059
    },
    {
      "epoch": 0.16466019417475727,
      "grad_norm": 0.0,
      "learning_rate": 0.0009532690220453963,
      "loss": 2.448,
      "step": 1060
    },
    {
      "epoch": 0.16481553398058252,
      "grad_norm": 0.0,
      "learning_rate": 0.0009531627543406223,
      "loss": 2.469,
      "step": 1061
    },
    {
      "epoch": 0.16497087378640776,
      "grad_norm": 0.0,
      "learning_rate": 0.0009530563718819958,
      "loss": 2.4707,
      "step": 1062
    },
    {
      "epoch": 0.165126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0009529498746964562,
      "loss": 2.4296,
      "step": 1063
    },
    {
      "epoch": 0.16528155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.0009528432628109713,
      "loss": 2.4772,
      "step": 1064
    },
    {
      "epoch": 0.1654368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0009527365362525384,
      "loss": 2.4277,
      "step": 1065
    },
    {
      "epoch": 0.16559223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.000952629695048184,
      "loss": 2.5277,
      "step": 1066
    },
    {
      "epoch": 0.16574757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.000952522739224963,
      "loss": 2.4673,
      "step": 1067
    },
    {
      "epoch": 0.16590291262135923,
      "grad_norm": 0.0,
      "learning_rate": 0.0009524156688099599,
      "loss": 2.5303,
      "step": 1068
    },
    {
      "epoch": 0.16605825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.0009523084838302879,
      "loss": 2.4257,
      "step": 1069
    },
    {
      "epoch": 0.16621359223300972,
      "grad_norm": 0.0,
      "learning_rate": 0.0009522011843130895,
      "loss": 2.3612,
      "step": 1070
    },
    {
      "epoch": 0.16636893203883496,
      "grad_norm": 0.0,
      "learning_rate": 0.0009520937702855357,
      "loss": 2.4426,
      "step": 1071
    },
    {
      "epoch": 0.1665242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0009519862417748272,
      "loss": 2.4821,
      "step": 1072
    },
    {
      "epoch": 0.16667961165048545,
      "grad_norm": 0.0,
      "learning_rate": 0.0009518785988081931,
      "loss": 2.4666,
      "step": 1073
    },
    {
      "epoch": 0.1668349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.0009517708414128917,
      "loss": 2.4017,
      "step": 1074
    },
    {
      "epoch": 0.1669902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.0009516629696162104,
      "loss": 2.4419,
      "step": 1075
    },
    {
      "epoch": 0.16714563106796115,
      "grad_norm": 0.0,
      "learning_rate": 0.0009515549834454653,
      "loss": 2.4542,
      "step": 1076
    },
    {
      "epoch": 0.1673009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0009514468829280017,
      "loss": 2.3854,
      "step": 1077
    },
    {
      "epoch": 0.16745631067961164,
      "grad_norm": 0.0,
      "learning_rate": 0.0009513386680911937,
      "loss": 2.5748,
      "step": 1078
    },
    {
      "epoch": 0.16761165048543689,
      "grad_norm": 0.0,
      "learning_rate": 0.0009512303389624443,
      "loss": 2.4287,
      "step": 1079
    },
    {
      "epoch": 0.16776699029126213,
      "grad_norm": 0.0,
      "learning_rate": 0.000951121895569186,
      "loss": 2.483,
      "step": 1080
    },
    {
      "epoch": 0.16792233009708737,
      "grad_norm": 0.0,
      "learning_rate": 0.0009510133379388793,
      "loss": 2.3441,
      "step": 1081
    },
    {
      "epoch": 0.16807766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0009509046660990143,
      "loss": 2.3862,
      "step": 1082
    },
    {
      "epoch": 0.16823300970873786,
      "grad_norm": 0.0,
      "learning_rate": 0.0009507958800771099,
      "loss": 2.481,
      "step": 1083
    },
    {
      "epoch": 0.1683883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0009506869799007136,
      "loss": 2.3302,
      "step": 1084
    },
    {
      "epoch": 0.16854368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0009505779655974024,
      "loss": 2.4269,
      "step": 1085
    },
    {
      "epoch": 0.1686990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0009504688371947816,
      "loss": 2.4315,
      "step": 1086
    },
    {
      "epoch": 0.16885436893203884,
      "grad_norm": 0.0,
      "learning_rate": 0.000950359594720486,
      "loss": 2.4769,
      "step": 1087
    },
    {
      "epoch": 0.16900970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.0009502502382021785,
      "loss": 2.4518,
      "step": 1088
    },
    {
      "epoch": 0.16916504854368933,
      "grad_norm": 0.0,
      "learning_rate": 0.0009501407676675515,
      "loss": 2.4778,
      "step": 1089
    },
    {
      "epoch": 0.16932038834951457,
      "grad_norm": 0.0,
      "learning_rate": 0.0009500311831443261,
      "loss": 2.4542,
      "step": 1090
    },
    {
      "epoch": 0.16947572815533982,
      "grad_norm": 0.0,
      "learning_rate": 0.0009499214846602524,
      "loss": 2.4036,
      "step": 1091
    },
    {
      "epoch": 0.16963106796116506,
      "grad_norm": 0.0,
      "learning_rate": 0.0009498116722431089,
      "loss": 2.3628,
      "step": 1092
    },
    {
      "epoch": 0.1697864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0009497017459207036,
      "loss": 2.4664,
      "step": 1093
    },
    {
      "epoch": 0.16994174757281552,
      "grad_norm": 0.0,
      "learning_rate": 0.0009495917057208728,
      "loss": 2.4258,
      "step": 1094
    },
    {
      "epoch": 0.17009708737864077,
      "grad_norm": 0.0,
      "learning_rate": 0.0009494815516714818,
      "loss": 2.3719,
      "step": 1095
    },
    {
      "epoch": 0.170252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.000949371283800425,
      "loss": 2.4036,
      "step": 1096
    },
    {
      "epoch": 0.17040776699029125,
      "grad_norm": 0.0,
      "learning_rate": 0.0009492609021356252,
      "loss": 2.3992,
      "step": 1097
    },
    {
      "epoch": 0.1705631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0009491504067050342,
      "loss": 2.3283,
      "step": 1098
    },
    {
      "epoch": 0.17071844660194174,
      "grad_norm": 0.0,
      "learning_rate": 0.0009490397975366327,
      "loss": 2.4364,
      "step": 1099
    },
    {
      "epoch": 0.170873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0009489290746584303,
      "loss": 2.3978,
      "step": 1100
    },
    {
      "epoch": 0.17102912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.0009488182380984649,
      "loss": 2.4094,
      "step": 1101
    },
    {
      "epoch": 0.17118446601941747,
      "grad_norm": 0.0,
      "learning_rate": 0.0009487072878848038,
      "loss": 2.4875,
      "step": 1102
    },
    {
      "epoch": 0.17133980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0009485962240455424,
      "loss": 2.4404,
      "step": 1103
    },
    {
      "epoch": 0.17149514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0009484850466088058,
      "loss": 2.4153,
      "step": 1104
    },
    {
      "epoch": 0.1716504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0009483737556027468,
      "loss": 2.4695,
      "step": 1105
    },
    {
      "epoch": 0.17180582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.0009482623510555478,
      "loss": 2.4144,
      "step": 1106
    },
    {
      "epoch": 0.1719611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0009481508329954196,
      "loss": 2.3848,
      "step": 1107
    },
    {
      "epoch": 0.17211650485436894,
      "grad_norm": 0.0,
      "learning_rate": 0.0009480392014506017,
      "loss": 2.4335,
      "step": 1108
    },
    {
      "epoch": 0.17227184466019418,
      "grad_norm": 0.0,
      "learning_rate": 0.0009479274564493626,
      "loss": 2.3956,
      "step": 1109
    },
    {
      "epoch": 0.17242718446601943,
      "grad_norm": 0.0,
      "learning_rate": 0.0009478155980199989,
      "loss": 2.2996,
      "step": 1110
    },
    {
      "epoch": 0.17258252427184467,
      "grad_norm": 0.0,
      "learning_rate": 0.0009477036261908369,
      "loss": 2.4629,
      "step": 1111
    },
    {
      "epoch": 0.1727378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0009475915409902307,
      "loss": 2.4215,
      "step": 1112
    },
    {
      "epoch": 0.17289320388349513,
      "grad_norm": 0.0,
      "learning_rate": 0.0009474793424465638,
      "loss": 2.3638,
      "step": 1113
    },
    {
      "epoch": 0.17304854368932038,
      "grad_norm": 0.0,
      "learning_rate": 0.000947367030588248,
      "loss": 2.4468,
      "step": 1114
    },
    {
      "epoch": 0.17320388349514562,
      "grad_norm": 0.0,
      "learning_rate": 0.0009472546054437239,
      "loss": 2.5363,
      "step": 1115
    },
    {
      "epoch": 0.17335922330097087,
      "grad_norm": 0.0,
      "learning_rate": 0.0009471420670414606,
      "loss": 2.4279,
      "step": 1116
    },
    {
      "epoch": 0.1735145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.0009470294154099564,
      "loss": 2.463,
      "step": 1117
    },
    {
      "epoch": 0.17366990291262135,
      "grad_norm": 0.0,
      "learning_rate": 0.0009469166505777375,
      "loss": 2.5039,
      "step": 1118
    },
    {
      "epoch": 0.1738252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0009468037725733596,
      "loss": 2.4271,
      "step": 1119
    },
    {
      "epoch": 0.17398058252427184,
      "grad_norm": 0.0,
      "learning_rate": 0.0009466907814254064,
      "loss": 2.4011,
      "step": 1120
    },
    {
      "epoch": 0.1741359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0009465776771624908,
      "loss": 2.3929,
      "step": 1121
    },
    {
      "epoch": 0.17429126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0009464644598132538,
      "loss": 2.4646,
      "step": 1122
    },
    {
      "epoch": 0.17444660194174758,
      "grad_norm": 0.0,
      "learning_rate": 0.0009463511294063654,
      "loss": 2.3325,
      "step": 1123
    },
    {
      "epoch": 0.17460194174757282,
      "grad_norm": 0.0,
      "learning_rate": 0.0009462376859705239,
      "loss": 2.4467,
      "step": 1124
    },
    {
      "epoch": 0.17475728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.000946124129534457,
      "loss": 2.4119,
      "step": 1125
    },
    {
      "epoch": 0.1749126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0009460104601269197,
      "loss": 2.4283,
      "step": 1126
    },
    {
      "epoch": 0.17506796116504855,
      "grad_norm": 0.0,
      "learning_rate": 0.0009458966777766969,
      "loss": 2.5442,
      "step": 1127
    },
    {
      "epoch": 0.1752233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.0009457827825126014,
      "loss": 2.5237,
      "step": 1128
    },
    {
      "epoch": 0.17537864077669904,
      "grad_norm": 0.0,
      "learning_rate": 0.0009456687743634749,
      "loss": 2.505,
      "step": 1129
    },
    {
      "epoch": 0.17553398058252428,
      "grad_norm": 0.0,
      "learning_rate": 0.0009455546533581872,
      "loss": 2.4137,
      "step": 1130
    },
    {
      "epoch": 0.1756893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0009454404195256374,
      "loss": 2.3784,
      "step": 1131
    },
    {
      "epoch": 0.17584466019417475,
      "grad_norm": 0.0,
      "learning_rate": 0.0009453260728947525,
      "loss": 2.3921,
      "step": 1132
    },
    {
      "epoch": 0.176,
      "grad_norm": 0.0,
      "learning_rate": 0.0009452116134944887,
      "loss": 2.4178,
      "step": 1133
    },
    {
      "epoch": 0.17615533980582523,
      "grad_norm": 0.0,
      "learning_rate": 0.0009450970413538301,
      "loss": 2.4261,
      "step": 1134
    },
    {
      "epoch": 0.17631067961165048,
      "grad_norm": 0.0,
      "learning_rate": 0.0009449823565017899,
      "loss": 2.4474,
      "step": 1135
    },
    {
      "epoch": 0.17646601941747572,
      "grad_norm": 0.0,
      "learning_rate": 0.0009448675589674093,
      "loss": 2.4493,
      "step": 1136
    },
    {
      "epoch": 0.17662135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0009447526487797585,
      "loss": 2.4553,
      "step": 1137
    },
    {
      "epoch": 0.1767766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0009446376259679362,
      "loss": 2.483,
      "step": 1138
    },
    {
      "epoch": 0.17693203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.0009445224905610695,
      "loss": 2.4624,
      "step": 1139
    },
    {
      "epoch": 0.1770873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0009444072425883137,
      "loss": 2.3597,
      "step": 1140
    },
    {
      "epoch": 0.17724271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.0009442918820788531,
      "loss": 2.4553,
      "step": 1141
    },
    {
      "epoch": 0.1773980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0009441764090619003,
      "loss": 2.4407,
      "step": 1142
    },
    {
      "epoch": 0.17755339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0009440608235666963,
      "loss": 2.3915,
      "step": 1143
    },
    {
      "epoch": 0.17770873786407768,
      "grad_norm": 0.0,
      "learning_rate": 0.0009439451256225107,
      "loss": 2.4525,
      "step": 1144
    },
    {
      "epoch": 0.17786407766990292,
      "grad_norm": 0.0,
      "learning_rate": 0.0009438293152586417,
      "loss": 2.4596,
      "step": 1145
    },
    {
      "epoch": 0.17801941747572816,
      "grad_norm": 0.0,
      "learning_rate": 0.0009437133925044158,
      "loss": 2.4951,
      "step": 1146
    },
    {
      "epoch": 0.1781747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0009435973573891877,
      "loss": 2.5068,
      "step": 1147
    },
    {
      "epoch": 0.17833009708737865,
      "grad_norm": 0.0,
      "learning_rate": 0.0009434812099423411,
      "loss": 2.4514,
      "step": 1148
    },
    {
      "epoch": 0.1784854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.0009433649501932879,
      "loss": 2.4578,
      "step": 1149
    },
    {
      "epoch": 0.1786407766990291,
      "grad_norm": 0.0,
      "learning_rate": 0.0009432485781714681,
      "loss": 2.3431,
      "step": 1150
    },
    {
      "epoch": 0.17879611650485436,
      "grad_norm": 0.0,
      "learning_rate": 0.0009431320939063507,
      "loss": 2.4509,
      "step": 1151
    },
    {
      "epoch": 0.1789514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0009430154974274331,
      "loss": 2.4946,
      "step": 1152
    },
    {
      "epoch": 0.17910679611650485,
      "grad_norm": 0.0,
      "learning_rate": 0.0009428987887642404,
      "loss": 2.5764,
      "step": 1153
    },
    {
      "epoch": 0.1792621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0009427819679463268,
      "loss": 2.4166,
      "step": 1154
    },
    {
      "epoch": 0.17941747572815533,
      "grad_norm": 0.0,
      "learning_rate": 0.0009426650350032746,
      "loss": 2.3785,
      "step": 1155
    },
    {
      "epoch": 0.17957281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.0009425479899646949,
      "loss": 2.376,
      "step": 1156
    },
    {
      "epoch": 0.17972815533980582,
      "grad_norm": 0.0,
      "learning_rate": 0.0009424308328602265,
      "loss": 2.4871,
      "step": 1157
    },
    {
      "epoch": 0.17988349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.0009423135637195372,
      "loss": 2.3281,
      "step": 1158
    },
    {
      "epoch": 0.1800388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0009421961825723228,
      "loss": 2.4583,
      "step": 1159
    },
    {
      "epoch": 0.18019417475728156,
      "grad_norm": 0.0,
      "learning_rate": 0.0009420786894483075,
      "loss": 2.3966,
      "step": 1160
    },
    {
      "epoch": 0.1803495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0009419610843772442,
      "loss": 2.407,
      "step": 1161
    },
    {
      "epoch": 0.18050485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0009418433673889136,
      "loss": 2.4941,
      "step": 1162
    },
    {
      "epoch": 0.1806601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.0009417255385131254,
      "loss": 2.4847,
      "step": 1163
    },
    {
      "epoch": 0.18081553398058253,
      "grad_norm": 0.0,
      "learning_rate": 0.0009416075977797167,
      "loss": 2.3165,
      "step": 1164
    },
    {
      "epoch": 0.18097087378640778,
      "grad_norm": 0.0,
      "learning_rate": 0.0009414895452185541,
      "loss": 2.4457,
      "step": 1165
    },
    {
      "epoch": 0.18112621359223302,
      "grad_norm": 0.0,
      "learning_rate": 0.0009413713808595315,
      "loss": 2.4614,
      "step": 1166
    },
    {
      "epoch": 0.18128155339805826,
      "grad_norm": 0.0,
      "learning_rate": 0.0009412531047325718,
      "loss": 2.4117,
      "step": 1167
    },
    {
      "epoch": 0.18143689320388348,
      "grad_norm": 0.0,
      "learning_rate": 0.0009411347168676256,
      "loss": 2.4738,
      "step": 1168
    },
    {
      "epoch": 0.18159223300970873,
      "grad_norm": 0.0,
      "learning_rate": 0.0009410162172946723,
      "loss": 2.4869,
      "step": 1169
    },
    {
      "epoch": 0.18174757281553397,
      "grad_norm": 0.0,
      "learning_rate": 0.0009408976060437194,
      "loss": 2.4222,
      "step": 1170
    },
    {
      "epoch": 0.18190291262135921,
      "grad_norm": 0.0,
      "learning_rate": 0.0009407788831448028,
      "loss": 2.4148,
      "step": 1171
    },
    {
      "epoch": 0.18205825242718446,
      "grad_norm": 0.0,
      "learning_rate": 0.0009406600486279861,
      "loss": 2.424,
      "step": 1172
    },
    {
      "epoch": 0.1822135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0009405411025233622,
      "loss": 2.5189,
      "step": 1173
    },
    {
      "epoch": 0.18236893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0009404220448610514,
      "loss": 2.4487,
      "step": 1174
    },
    {
      "epoch": 0.1825242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0009403028756712024,
      "loss": 2.3711,
      "step": 1175
    },
    {
      "epoch": 0.18267961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.0009401835949839926,
      "loss": 2.4095,
      "step": 1176
    },
    {
      "epoch": 0.18283495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0009400642028296271,
      "loss": 2.4219,
      "step": 1177
    },
    {
      "epoch": 0.18299029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.0009399446992383394,
      "loss": 2.4308,
      "step": 1178
    },
    {
      "epoch": 0.18314563106796117,
      "grad_norm": 0.0,
      "learning_rate": 0.0009398250842403913,
      "loss": 2.4862,
      "step": 1179
    },
    {
      "epoch": 0.1833009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0009397053578660726,
      "loss": 2.3729,
      "step": 1180
    },
    {
      "epoch": 0.18345631067961166,
      "grad_norm": 0.0,
      "learning_rate": 0.0009395855201457017,
      "loss": 2.3566,
      "step": 1181
    },
    {
      "epoch": 0.1836116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.000939465571109625,
      "loss": 2.3856,
      "step": 1182
    },
    {
      "epoch": 0.18376699029126214,
      "grad_norm": 0.0,
      "learning_rate": 0.0009393455107882168,
      "loss": 2.364,
      "step": 1183
    },
    {
      "epoch": 0.1839223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0009392253392118802,
      "loss": 2.4484,
      "step": 1184
    },
    {
      "epoch": 0.18407766990291263,
      "grad_norm": 0.0,
      "learning_rate": 0.0009391050564110455,
      "loss": 2.4233,
      "step": 1185
    },
    {
      "epoch": 0.18423300970873788,
      "grad_norm": 0.0,
      "learning_rate": 0.0009389846624161723,
      "loss": 2.3477,
      "step": 1186
    },
    {
      "epoch": 0.1843883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0009388641572577477,
      "loss": 2.3453,
      "step": 1187
    },
    {
      "epoch": 0.18454368932038834,
      "grad_norm": 0.0,
      "learning_rate": 0.000938743540966287,
      "loss": 2.3864,
      "step": 1188
    },
    {
      "epoch": 0.18469902912621358,
      "grad_norm": 0.0,
      "learning_rate": 0.0009386228135723336,
      "loss": 2.3778,
      "step": 1189
    },
    {
      "epoch": 0.18485436893203883,
      "grad_norm": 0.0,
      "learning_rate": 0.0009385019751064595,
      "loss": 2.3787,
      "step": 1190
    },
    {
      "epoch": 0.18500970873786407,
      "grad_norm": 0.0,
      "learning_rate": 0.0009383810255992641,
      "loss": 2.5925,
      "step": 1191
    },
    {
      "epoch": 0.18516504854368931,
      "grad_norm": 0.0,
      "learning_rate": 0.0009382599650813756,
      "loss": 2.4756,
      "step": 1192
    },
    {
      "epoch": 0.18532038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.0009381387935834497,
      "loss": 2.446,
      "step": 1193
    },
    {
      "epoch": 0.1854757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0009380175111361708,
      "loss": 2.475,
      "step": 1194
    },
    {
      "epoch": 0.18563106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.000937896117770251,
      "loss": 2.4073,
      "step": 1195
    },
    {
      "epoch": 0.1857864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0009377746135164306,
      "loss": 2.4507,
      "step": 1196
    },
    {
      "epoch": 0.18594174757281554,
      "grad_norm": 0.0,
      "learning_rate": 0.0009376529984054781,
      "loss": 2.489,
      "step": 1197
    },
    {
      "epoch": 0.18609708737864078,
      "grad_norm": 0.0,
      "learning_rate": 0.0009375312724681895,
      "loss": 2.3102,
      "step": 1198
    },
    {
      "epoch": 0.18625242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0009374094357353898,
      "loss": 2.4515,
      "step": 1199
    },
    {
      "epoch": 0.18640776699029127,
      "grad_norm": 0.0,
      "learning_rate": 0.0009372874882379313,
      "loss": 2.3997,
      "step": 1200
    },
    {
      "epoch": 0.1865631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0009371654300066948,
      "loss": 2.4384,
      "step": 1201
    },
    {
      "epoch": 0.18671844660194176,
      "grad_norm": 0.0,
      "learning_rate": 0.0009370432610725887,
      "loss": 2.5094,
      "step": 1202
    },
    {
      "epoch": 0.186873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.00093692098146655,
      "loss": 2.4781,
      "step": 1203
    },
    {
      "epoch": 0.18702912621359224,
      "grad_norm": 0.0,
      "learning_rate": 0.0009367985912195431,
      "loss": 2.5303,
      "step": 1204
    },
    {
      "epoch": 0.1871844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.0009366760903625611,
      "loss": 2.4833,
      "step": 1205
    },
    {
      "epoch": 0.1873398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.0009365534789266245,
      "loss": 2.3969,
      "step": 1206
    },
    {
      "epoch": 0.18749514563106795,
      "grad_norm": 0.0,
      "learning_rate": 0.000936430756942782,
      "loss": 2.4766,
      "step": 1207
    },
    {
      "epoch": 0.1876504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0009363079244421105,
      "loss": 2.4755,
      "step": 1208
    },
    {
      "epoch": 0.18780582524271844,
      "grad_norm": 0.0,
      "learning_rate": 0.0009361849814557146,
      "loss": 2.4194,
      "step": 1209
    },
    {
      "epoch": 0.18796116504854368,
      "grad_norm": 0.0,
      "learning_rate": 0.000936061928014727,
      "loss": 2.4018,
      "step": 1210
    },
    {
      "epoch": 0.18811650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.0009359387641503086,
      "loss": 2.4978,
      "step": 1211
    },
    {
      "epoch": 0.18827184466019417,
      "grad_norm": 0.0,
      "learning_rate": 0.0009358154898936477,
      "loss": 2.3669,
      "step": 1212
    },
    {
      "epoch": 0.18842718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.000935692105275961,
      "loss": 2.5195,
      "step": 1213
    },
    {
      "epoch": 0.18858252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0009355686103284932,
      "loss": 2.3683,
      "step": 1214
    },
    {
      "epoch": 0.1887378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0009354450050825166,
      "loss": 2.3567,
      "step": 1215
    },
    {
      "epoch": 0.18889320388349515,
      "grad_norm": 0.0,
      "learning_rate": 0.0009353212895693318,
      "loss": 2.5249,
      "step": 1216
    },
    {
      "epoch": 0.1890485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0009351974638202668,
      "loss": 2.416,
      "step": 1217
    },
    {
      "epoch": 0.18920388349514564,
      "grad_norm": 0.0,
      "learning_rate": 0.0009350735278666782,
      "loss": 2.4109,
      "step": 1218
    },
    {
      "epoch": 0.18935922330097088,
      "grad_norm": 0.0,
      "learning_rate": 0.0009349494817399498,
      "loss": 2.3943,
      "step": 1219
    },
    {
      "epoch": 0.18951456310679612,
      "grad_norm": 0.0,
      "learning_rate": 0.000934825325471494,
      "loss": 2.3925,
      "step": 1220
    },
    {
      "epoch": 0.18966990291262137,
      "grad_norm": 0.0,
      "learning_rate": 0.0009347010590927505,
      "loss": 2.4557,
      "step": 1221
    },
    {
      "epoch": 0.1898252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0009345766826351872,
      "loss": 2.4165,
      "step": 1222
    },
    {
      "epoch": 0.18998058252427186,
      "grad_norm": 0.0,
      "learning_rate": 0.0009344521961302999,
      "loss": 2.4236,
      "step": 1223
    },
    {
      "epoch": 0.19013592233009707,
      "grad_norm": 0.0,
      "learning_rate": 0.000934327599609612,
      "loss": 2.4441,
      "step": 1224
    },
    {
      "epoch": 0.19029126213592232,
      "grad_norm": 0.0,
      "learning_rate": 0.0009342028931046751,
      "loss": 2.4091,
      "step": 1225
    },
    {
      "epoch": 0.19044660194174756,
      "grad_norm": 0.0,
      "learning_rate": 0.0009340780766470682,
      "loss": 2.4395,
      "step": 1226
    },
    {
      "epoch": 0.1906019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.0009339531502683987,
      "loss": 2.4736,
      "step": 1227
    },
    {
      "epoch": 0.19075728155339805,
      "grad_norm": 0.0,
      "learning_rate": 0.0009338281140003013,
      "loss": 2.3081,
      "step": 1228
    },
    {
      "epoch": 0.1909126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0009337029678744391,
      "loss": 2.433,
      "step": 1229
    },
    {
      "epoch": 0.19106796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.0009335777119225024,
      "loss": 2.3591,
      "step": 1230
    },
    {
      "epoch": 0.19122330097087378,
      "grad_norm": 0.0,
      "learning_rate": 0.0009334523461762099,
      "loss": 2.4552,
      "step": 1231
    },
    {
      "epoch": 0.19137864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0009333268706673075,
      "loss": 2.3867,
      "step": 1232
    },
    {
      "epoch": 0.19153398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.0009332012854275694,
      "loss": 2.4185,
      "step": 1233
    },
    {
      "epoch": 0.19168932038834952,
      "grad_norm": 0.0,
      "learning_rate": 0.0009330755904887974,
      "loss": 2.3979,
      "step": 1234
    },
    {
      "epoch": 0.19184466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.0009329497858828211,
      "loss": 2.3701,
      "step": 1235
    },
    {
      "epoch": 0.192,
      "grad_norm": 0.0,
      "learning_rate": 0.0009328238716414976,
      "loss": 2.3364,
      "step": 1236
    },
    {
      "epoch": 0.19215533980582525,
      "grad_norm": 0.0,
      "learning_rate": 0.0009326978477967123,
      "loss": 2.5205,
      "step": 1237
    },
    {
      "epoch": 0.1923106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.000932571714380378,
      "loss": 2.3909,
      "step": 1238
    },
    {
      "epoch": 0.19246601941747574,
      "grad_norm": 0.0,
      "learning_rate": 0.0009324454714244352,
      "loss": 2.4396,
      "step": 1239
    },
    {
      "epoch": 0.19262135922330098,
      "grad_norm": 0.0,
      "learning_rate": 0.0009323191189608524,
      "loss": 2.4559,
      "step": 1240
    },
    {
      "epoch": 0.19277669902912623,
      "grad_norm": 0.0,
      "learning_rate": 0.0009321926570216255,
      "loss": 2.4097,
      "step": 1241
    },
    {
      "epoch": 0.19293203883495147,
      "grad_norm": 0.0,
      "learning_rate": 0.0009320660856387785,
      "loss": 2.51,
      "step": 1242
    },
    {
      "epoch": 0.19308737864077669,
      "grad_norm": 0.0,
      "learning_rate": 0.0009319394048443628,
      "loss": 2.4023,
      "step": 1243
    },
    {
      "epoch": 0.19324271844660193,
      "grad_norm": 0.0,
      "learning_rate": 0.0009318126146704575,
      "loss": 2.4363,
      "step": 1244
    },
    {
      "epoch": 0.19339805825242717,
      "grad_norm": 0.0,
      "learning_rate": 0.0009316857151491699,
      "loss": 2.3788,
      "step": 1245
    },
    {
      "epoch": 0.19355339805825242,
      "grad_norm": 0.0,
      "learning_rate": 0.0009315587063126343,
      "loss": 2.5039,
      "step": 1246
    },
    {
      "epoch": 0.19370873786407766,
      "grad_norm": 0.0,
      "learning_rate": 0.000931431588193013,
      "loss": 2.2949,
      "step": 1247
    },
    {
      "epoch": 0.1938640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.000931304360822496,
      "loss": 2.4271,
      "step": 1248
    },
    {
      "epoch": 0.19401941747572815,
      "grad_norm": 0.0,
      "learning_rate": 0.0009311770242333011,
      "loss": 2.4512,
      "step": 1249
    },
    {
      "epoch": 0.1941747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0009310495784576733,
      "loss": 2.5204,
      "step": 1250
    },
    {
      "epoch": 0.19433009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.000930922023527886,
      "loss": 2.4436,
      "step": 1251
    },
    {
      "epoch": 0.19448543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.0009307943594762392,
      "loss": 2.4967,
      "step": 1252
    },
    {
      "epoch": 0.19464077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.0009306665863350614,
      "loss": 2.3655,
      "step": 1253
    },
    {
      "epoch": 0.19479611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0009305387041367086,
      "loss": 2.4028,
      "step": 1254
    },
    {
      "epoch": 0.19495145631067962,
      "grad_norm": 0.0,
      "learning_rate": 0.0009304107129135639,
      "loss": 2.4266,
      "step": 1255
    },
    {
      "epoch": 0.19510679611650486,
      "grad_norm": 0.0,
      "learning_rate": 0.0009302826126980386,
      "loss": 2.4329,
      "step": 1256
    },
    {
      "epoch": 0.1952621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0009301544035225715,
      "loss": 2.408,
      "step": 1257
    },
    {
      "epoch": 0.19541747572815535,
      "grad_norm": 0.0,
      "learning_rate": 0.0009300260854196283,
      "loss": 2.3992,
      "step": 1258
    },
    {
      "epoch": 0.1955728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.0009298976584217034,
      "loss": 2.3809,
      "step": 1259
    },
    {
      "epoch": 0.19572815533980584,
      "grad_norm": 0.0,
      "learning_rate": 0.0009297691225613181,
      "loss": 2.3264,
      "step": 1260
    },
    {
      "epoch": 0.19588349514563108,
      "grad_norm": 0.0,
      "learning_rate": 0.000929640477871021,
      "loss": 2.5497,
      "step": 1261
    },
    {
      "epoch": 0.1960388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0009295117243833892,
      "loss": 2.506,
      "step": 1262
    },
    {
      "epoch": 0.19619417475728154,
      "grad_norm": 0.0,
      "learning_rate": 0.0009293828621310263,
      "loss": 2.4622,
      "step": 1263
    },
    {
      "epoch": 0.1963495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0009292538911465642,
      "loss": 2.4098,
      "step": 1264
    },
    {
      "epoch": 0.19650485436893203,
      "grad_norm": 0.0,
      "learning_rate": 0.0009291248114626621,
      "loss": 2.4937,
      "step": 1265
    },
    {
      "epoch": 0.19666019417475727,
      "grad_norm": 0.0,
      "learning_rate": 0.0009289956231120066,
      "loss": 2.3962,
      "step": 1266
    },
    {
      "epoch": 0.19681553398058252,
      "grad_norm": 0.0,
      "learning_rate": 0.0009288663261273118,
      "loss": 2.4862,
      "step": 1267
    },
    {
      "epoch": 0.19697087378640776,
      "grad_norm": 0.0,
      "learning_rate": 0.0009287369205413196,
      "loss": 2.3658,
      "step": 1268
    },
    {
      "epoch": 0.197126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0009286074063867991,
      "loss": 2.386,
      "step": 1269
    },
    {
      "epoch": 0.19728155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.0009284777836965471,
      "loss": 2.4458,
      "step": 1270
    },
    {
      "epoch": 0.1974368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0009283480525033878,
      "loss": 2.377,
      "step": 1271
    },
    {
      "epoch": 0.19759223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0009282182128401728,
      "loss": 2.3101,
      "step": 1272
    },
    {
      "epoch": 0.19774757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0009280882647397812,
      "loss": 2.4419,
      "step": 1273
    },
    {
      "epoch": 0.19790291262135923,
      "grad_norm": 0.0,
      "learning_rate": 0.0009279582082351197,
      "loss": 2.347,
      "step": 1274
    },
    {
      "epoch": 0.19805825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.0009278280433591222,
      "loss": 2.4368,
      "step": 1275
    },
    {
      "epoch": 0.19821359223300972,
      "grad_norm": 0.0,
      "learning_rate": 0.0009276977701447504,
      "loss": 2.4496,
      "step": 1276
    },
    {
      "epoch": 0.19836893203883496,
      "grad_norm": 0.0,
      "learning_rate": 0.0009275673886249931,
      "loss": 2.4846,
      "step": 1277
    },
    {
      "epoch": 0.1985242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0009274368988328668,
      "loss": 2.3951,
      "step": 1278
    },
    {
      "epoch": 0.19867961165048545,
      "grad_norm": 0.0,
      "learning_rate": 0.0009273063008014151,
      "loss": 2.3293,
      "step": 1279
    },
    {
      "epoch": 0.1988349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.0009271755945637093,
      "loss": 2.3925,
      "step": 1280
    },
    {
      "epoch": 0.1989902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.0009270447801528477,
      "loss": 2.3419,
      "step": 1281
    },
    {
      "epoch": 0.19914563106796115,
      "grad_norm": 0.0,
      "learning_rate": 0.0009269138576019567,
      "loss": 2.4958,
      "step": 1282
    },
    {
      "epoch": 0.1993009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0009267828269441894,
      "loss": 2.4761,
      "step": 1283
    },
    {
      "epoch": 0.19945631067961164,
      "grad_norm": 0.0,
      "learning_rate": 0.0009266516882127266,
      "loss": 2.3966,
      "step": 1284
    },
    {
      "epoch": 0.1996116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0009265204414407764,
      "loss": 2.468,
      "step": 1285
    },
    {
      "epoch": 0.19976699029126213,
      "grad_norm": 0.0,
      "learning_rate": 0.0009263890866615742,
      "loss": 2.5388,
      "step": 1286
    },
    {
      "epoch": 0.19992233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.0009262576239083828,
      "loss": 2.3608,
      "step": 1287
    },
    {
      "epoch": 0.20007766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0009261260532144926,
      "loss": 2.4375,
      "step": 1288
    },
    {
      "epoch": 0.20023300970873786,
      "grad_norm": 0.0,
      "learning_rate": 0.0009259943746132207,
      "loss": 2.5037,
      "step": 1289
    },
    {
      "epoch": 0.2003883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0009258625881379123,
      "loss": 2.4064,
      "step": 1290
    },
    {
      "epoch": 0.20054368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0009257306938219392,
      "loss": 2.457,
      "step": 1291
    },
    {
      "epoch": 0.2006990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.000925598691698701,
      "loss": 2.4321,
      "step": 1292
    },
    {
      "epoch": 0.20085436893203884,
      "grad_norm": 0.0,
      "learning_rate": 0.0009254665818016245,
      "loss": 2.4034,
      "step": 1293
    },
    {
      "epoch": 0.20100970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.0009253343641641637,
      "loss": 2.4259,
      "step": 1294
    },
    {
      "epoch": 0.20116504854368933,
      "grad_norm": 0.0,
      "learning_rate": 0.0009252020388197994,
      "loss": 2.3857,
      "step": 1295
    },
    {
      "epoch": 0.20132038834951457,
      "grad_norm": 0.0,
      "learning_rate": 0.0009250696058020412,
      "loss": 2.4307,
      "step": 1296
    },
    {
      "epoch": 0.20147572815533982,
      "grad_norm": 0.0,
      "learning_rate": 0.0009249370651444241,
      "loss": 2.3883,
      "step": 1297
    },
    {
      "epoch": 0.20163106796116506,
      "grad_norm": 0.0,
      "learning_rate": 0.0009248044168805115,
      "loss": 2.4382,
      "step": 1298
    },
    {
      "epoch": 0.20178640776699028,
      "grad_norm": 0.0,
      "learning_rate": 0.0009246716610438938,
      "loss": 2.5403,
      "step": 1299
    },
    {
      "epoch": 0.20194174757281552,
      "grad_norm": 0.0,
      "learning_rate": 0.0009245387976681886,
      "loss": 2.36,
      "step": 1300
    },
    {
      "epoch": 0.20209708737864077,
      "grad_norm": 0.0,
      "learning_rate": 0.0009244058267870406,
      "loss": 2.326,
      "step": 1301
    },
    {
      "epoch": 0.202252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.000924272748434122,
      "loss": 2.4424,
      "step": 1302
    },
    {
      "epoch": 0.20240776699029125,
      "grad_norm": 0.0,
      "learning_rate": 0.0009241395626431321,
      "loss": 2.4205,
      "step": 1303
    },
    {
      "epoch": 0.2025631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0009240062694477972,
      "loss": 2.496,
      "step": 1304
    },
    {
      "epoch": 0.20271844660194174,
      "grad_norm": 0.0,
      "learning_rate": 0.0009238728688818709,
      "loss": 2.375,
      "step": 1305
    },
    {
      "epoch": 0.202873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0009237393609791344,
      "loss": 2.4492,
      "step": 1306
    },
    {
      "epoch": 0.20302912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.0009236057457733956,
      "loss": 2.4485,
      "step": 1307
    },
    {
      "epoch": 0.20318446601941748,
      "grad_norm": 0.0,
      "learning_rate": 0.0009234720232984897,
      "loss": 2.4008,
      "step": 1308
    },
    {
      "epoch": 0.20333980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.000923338193588279,
      "loss": 2.4257,
      "step": 1309
    },
    {
      "epoch": 0.20349514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0009232042566766531,
      "loss": 2.4191,
      "step": 1310
    },
    {
      "epoch": 0.2036504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0009230702125975287,
      "loss": 2.4691,
      "step": 1311
    },
    {
      "epoch": 0.20380582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.0009229360613848494,
      "loss": 2.4431,
      "step": 1312
    },
    {
      "epoch": 0.2039611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0009228018030725866,
      "loss": 2.4305,
      "step": 1313
    },
    {
      "epoch": 0.20411650485436894,
      "grad_norm": 0.0,
      "learning_rate": 0.000922667437694738,
      "loss": 2.4082,
      "step": 1314
    },
    {
      "epoch": 0.20427184466019419,
      "grad_norm": 0.0,
      "learning_rate": 0.0009225329652853289,
      "loss": 2.3749,
      "step": 1315
    },
    {
      "epoch": 0.20442718446601943,
      "grad_norm": 0.0,
      "learning_rate": 0.0009223983858784116,
      "loss": 2.3479,
      "step": 1316
    },
    {
      "epoch": 0.20458252427184467,
      "grad_norm": 0.0,
      "learning_rate": 0.0009222636995080655,
      "loss": 2.4015,
      "step": 1317
    },
    {
      "epoch": 0.2047378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0009221289062083971,
      "loss": 2.3611,
      "step": 1318
    },
    {
      "epoch": 0.20489320388349513,
      "grad_norm": 0.0,
      "learning_rate": 0.0009219940060135397,
      "loss": 2.3799,
      "step": 1319
    },
    {
      "epoch": 0.20504854368932038,
      "grad_norm": 0.0,
      "learning_rate": 0.000921858998957654,
      "loss": 2.4004,
      "step": 1320
    },
    {
      "epoch": 0.20520388349514562,
      "grad_norm": 0.0,
      "learning_rate": 0.0009217238850749279,
      "loss": 2.4929,
      "step": 1321
    },
    {
      "epoch": 0.20535922330097087,
      "grad_norm": 0.0,
      "learning_rate": 0.0009215886643995759,
      "loss": 2.3799,
      "step": 1322
    },
    {
      "epoch": 0.2055145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.00092145333696584,
      "loss": 2.3876,
      "step": 1323
    },
    {
      "epoch": 0.20566990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0009213179028079888,
      "loss": 2.353,
      "step": 1324
    },
    {
      "epoch": 0.2058252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.000921182361960318,
      "loss": 2.4457,
      "step": 1325
    },
    {
      "epoch": 0.20598058252427184,
      "grad_norm": 0.0,
      "learning_rate": 0.0009210467144571507,
      "loss": 2.3776,
      "step": 1326
    },
    {
      "epoch": 0.2061359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0009209109603328367,
      "loss": 2.3893,
      "step": 1327
    },
    {
      "epoch": 0.20629126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0009207750996217528,
      "loss": 2.4152,
      "step": 1328
    },
    {
      "epoch": 0.20644660194174758,
      "grad_norm": 0.0,
      "learning_rate": 0.000920639132358303,
      "loss": 2.4856,
      "step": 1329
    },
    {
      "epoch": 0.20660194174757282,
      "grad_norm": 0.0,
      "learning_rate": 0.0009205030585769178,
      "loss": 2.4498,
      "step": 1330
    },
    {
      "epoch": 0.20675728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.0009203668783120551,
      "loss": 2.3621,
      "step": 1331
    },
    {
      "epoch": 0.2069126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0009202305915982,
      "loss": 2.4692,
      "step": 1332
    },
    {
      "epoch": 0.20706796116504855,
      "grad_norm": 0.0,
      "learning_rate": 0.0009200941984698639,
      "loss": 2.3921,
      "step": 1333
    },
    {
      "epoch": 0.2072233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.0009199576989615854,
      "loss": 2.3837,
      "step": 1334
    },
    {
      "epoch": 0.20737864077669904,
      "grad_norm": 0.0,
      "learning_rate": 0.0009198210931079302,
      "loss": 2.4233,
      "step": 1335
    },
    {
      "epoch": 0.20753398058252429,
      "grad_norm": 0.0,
      "learning_rate": 0.0009196843809434911,
      "loss": 2.2957,
      "step": 1336
    },
    {
      "epoch": 0.2076893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0009195475625028872,
      "loss": 2.3435,
      "step": 1337
    },
    {
      "epoch": 0.20784466019417475,
      "grad_norm": 0.0,
      "learning_rate": 0.0009194106378207649,
      "loss": 2.3845,
      "step": 1338
    },
    {
      "epoch": 0.208,
      "grad_norm": 0.0,
      "learning_rate": 0.0009192736069317978,
      "loss": 2.446,
      "step": 1339
    },
    {
      "epoch": 0.20815533980582523,
      "grad_norm": 0.0,
      "learning_rate": 0.0009191364698706856,
      "loss": 2.3351,
      "step": 1340
    },
    {
      "epoch": 0.20831067961165048,
      "grad_norm": 0.0,
      "learning_rate": 0.0009189992266721556,
      "loss": 2.4952,
      "step": 1341
    },
    {
      "epoch": 0.20846601941747572,
      "grad_norm": 0.0,
      "learning_rate": 0.0009188618773709618,
      "loss": 2.4601,
      "step": 1342
    },
    {
      "epoch": 0.20862135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0009187244220018849,
      "loss": 2.3252,
      "step": 1343
    },
    {
      "epoch": 0.2087766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0009185868605997325,
      "loss": 2.3152,
      "step": 1344
    },
    {
      "epoch": 0.20893203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.000918449193199339,
      "loss": 2.3125,
      "step": 1345
    },
    {
      "epoch": 0.2090873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0009183114198355662,
      "loss": 2.4714,
      "step": 1346
    },
    {
      "epoch": 0.20924271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.000918173540543302,
      "loss": 2.4022,
      "step": 1347
    },
    {
      "epoch": 0.2093980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0009180355553574612,
      "loss": 2.5167,
      "step": 1348
    },
    {
      "epoch": 0.20955339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0009178974643129859,
      "loss": 2.5106,
      "step": 1349
    },
    {
      "epoch": 0.20970873786407768,
      "grad_norm": 0.0,
      "learning_rate": 0.0009177592674448446,
      "loss": 2.4042,
      "step": 1350
    },
    {
      "epoch": 0.20986407766990292,
      "grad_norm": 0.0,
      "learning_rate": 0.000917620964788033,
      "loss": 2.3251,
      "step": 1351
    },
    {
      "epoch": 0.21001941747572817,
      "grad_norm": 0.0,
      "learning_rate": 0.0009174825563775729,
      "loss": 2.3135,
      "step": 1352
    },
    {
      "epoch": 0.2101747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0009173440422485133,
      "loss": 2.443,
      "step": 1353
    },
    {
      "epoch": 0.21033009708737865,
      "grad_norm": 0.0,
      "learning_rate": 0.0009172054224359304,
      "loss": 2.3679,
      "step": 1354
    },
    {
      "epoch": 0.21048543689320387,
      "grad_norm": 0.0,
      "learning_rate": 0.0009170666969749264,
      "loss": 2.4451,
      "step": 1355
    },
    {
      "epoch": 0.21064077669902911,
      "grad_norm": 0.0,
      "learning_rate": 0.0009169278659006306,
      "loss": 2.4796,
      "step": 1356
    },
    {
      "epoch": 0.21079611650485436,
      "grad_norm": 0.0,
      "learning_rate": 0.000916788929248199,
      "loss": 2.3447,
      "step": 1357
    },
    {
      "epoch": 0.2109514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0009166498870528146,
      "loss": 2.4304,
      "step": 1358
    },
    {
      "epoch": 0.21110679611650485,
      "grad_norm": 0.0,
      "learning_rate": 0.0009165107393496866,
      "loss": 2.3846,
      "step": 1359
    },
    {
      "epoch": 0.2112621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0009163714861740513,
      "loss": 2.3497,
      "step": 1360
    },
    {
      "epoch": 0.21141747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0009162321275611718,
      "loss": 2.3856,
      "step": 1361
    },
    {
      "epoch": 0.21157281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.0009160926635463373,
      "loss": 2.425,
      "step": 1362
    },
    {
      "epoch": 0.21172815533980582,
      "grad_norm": 0.0,
      "learning_rate": 0.0009159530941648647,
      "loss": 2.4259,
      "step": 1363
    },
    {
      "epoch": 0.21188349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.0009158134194520962,
      "loss": 2.379,
      "step": 1364
    },
    {
      "epoch": 0.2120388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0009156736394434022,
      "loss": 2.3947,
      "step": 1365
    },
    {
      "epoch": 0.21219417475728156,
      "grad_norm": 0.0,
      "learning_rate": 0.0009155337541741786,
      "loss": 2.4458,
      "step": 1366
    },
    {
      "epoch": 0.2123495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0009153937636798484,
      "loss": 2.4659,
      "step": 1367
    },
    {
      "epoch": 0.21250485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0009152536679958614,
      "loss": 2.3627,
      "step": 1368
    },
    {
      "epoch": 0.2126601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.0009151134671576939,
      "loss": 2.4247,
      "step": 1369
    },
    {
      "epoch": 0.21281553398058253,
      "grad_norm": 0.0,
      "learning_rate": 0.0009149731612008485,
      "loss": 2.4386,
      "step": 1370
    },
    {
      "epoch": 0.21297087378640778,
      "grad_norm": 0.0,
      "learning_rate": 0.0009148327501608548,
      "loss": 2.3925,
      "step": 1371
    },
    {
      "epoch": 0.21312621359223302,
      "grad_norm": 0.0,
      "learning_rate": 0.0009146922340732691,
      "loss": 2.4446,
      "step": 1372
    },
    {
      "epoch": 0.21328155339805827,
      "grad_norm": 0.0,
      "learning_rate": 0.000914551612973674,
      "loss": 2.5211,
      "step": 1373
    },
    {
      "epoch": 0.21343689320388348,
      "grad_norm": 0.0,
      "learning_rate": 0.0009144108868976787,
      "loss": 2.3396,
      "step": 1374
    },
    {
      "epoch": 0.21359223300970873,
      "grad_norm": 0.0,
      "learning_rate": 0.0009142700558809193,
      "loss": 2.3784,
      "step": 1375
    },
    {
      "epoch": 0.21374757281553397,
      "grad_norm": 0.0,
      "learning_rate": 0.0009141291199590581,
      "loss": 2.4473,
      "step": 1376
    },
    {
      "epoch": 0.21390291262135921,
      "grad_norm": 0.0,
      "learning_rate": 0.0009139880791677841,
      "loss": 2.3102,
      "step": 1377
    },
    {
      "epoch": 0.21405825242718446,
      "grad_norm": 0.0,
      "learning_rate": 0.0009138469335428132,
      "loss": 2.4689,
      "step": 1378
    },
    {
      "epoch": 0.2142135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0009137056831198872,
      "loss": 2.3507,
      "step": 1379
    },
    {
      "epoch": 0.21436893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.000913564327934775,
      "loss": 2.3795,
      "step": 1380
    },
    {
      "epoch": 0.2145242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0009134228680232714,
      "loss": 2.4069,
      "step": 1381
    },
    {
      "epoch": 0.21467961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.0009132813034211987,
      "loss": 2.4666,
      "step": 1382
    },
    {
      "epoch": 0.21483495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0009131396341644047,
      "loss": 2.3583,
      "step": 1383
    },
    {
      "epoch": 0.21499029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.0009129978602887644,
      "loss": 2.4098,
      "step": 1384
    },
    {
      "epoch": 0.21514563106796117,
      "grad_norm": 0.0,
      "learning_rate": 0.000912855981830179,
      "loss": 2.3562,
      "step": 1385
    },
    {
      "epoch": 0.2153009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0009127139988245759,
      "loss": 2.5125,
      "step": 1386
    },
    {
      "epoch": 0.21545631067961166,
      "grad_norm": 0.0,
      "learning_rate": 0.0009125719113079095,
      "loss": 2.3969,
      "step": 1387
    },
    {
      "epoch": 0.2156116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0009124297193161604,
      "loss": 2.3812,
      "step": 1388
    },
    {
      "epoch": 0.21576699029126215,
      "grad_norm": 0.0,
      "learning_rate": 0.0009122874228853359,
      "loss": 2.3827,
      "step": 1389
    },
    {
      "epoch": 0.2159223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0009121450220514694,
      "loss": 2.3235,
      "step": 1390
    },
    {
      "epoch": 0.21607766990291263,
      "grad_norm": 0.0,
      "learning_rate": 0.000912002516850621,
      "loss": 2.3091,
      "step": 1391
    },
    {
      "epoch": 0.21623300970873788,
      "grad_norm": 0.0,
      "learning_rate": 0.0009118599073188768,
      "loss": 2.446,
      "step": 1392
    },
    {
      "epoch": 0.2163883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0009117171934923498,
      "loss": 2.4154,
      "step": 1393
    },
    {
      "epoch": 0.21654368932038834,
      "grad_norm": 0.0,
      "learning_rate": 0.0009115743754071795,
      "loss": 2.3806,
      "step": 1394
    },
    {
      "epoch": 0.21669902912621358,
      "grad_norm": 0.0,
      "learning_rate": 0.0009114314530995313,
      "loss": 2.4044,
      "step": 1395
    },
    {
      "epoch": 0.21685436893203883,
      "grad_norm": 0.0,
      "learning_rate": 0.0009112884266055972,
      "loss": 2.4033,
      "step": 1396
    },
    {
      "epoch": 0.21700970873786407,
      "grad_norm": 0.0,
      "learning_rate": 0.0009111452959615958,
      "loss": 2.3518,
      "step": 1397
    },
    {
      "epoch": 0.21716504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0009110020612037717,
      "loss": 2.4561,
      "step": 1398
    },
    {
      "epoch": 0.21732038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.0009108587223683964,
      "loss": 2.4576,
      "step": 1399
    },
    {
      "epoch": 0.2174757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0009107152794917671,
      "loss": 2.298,
      "step": 1400
    },
    {
      "epoch": 0.21763106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0009105717326102076,
      "loss": 2.3273,
      "step": 1401
    },
    {
      "epoch": 0.2177864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0009104280817600684,
      "loss": 2.3629,
      "step": 1402
    },
    {
      "epoch": 0.21794174757281554,
      "grad_norm": 0.0,
      "learning_rate": 0.0009102843269777258,
      "loss": 2.3624,
      "step": 1403
    },
    {
      "epoch": 0.21809708737864078,
      "grad_norm": 0.0,
      "learning_rate": 0.0009101404682995829,
      "loss": 2.4728,
      "step": 1404
    },
    {
      "epoch": 0.21825242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0009099965057620684,
      "loss": 2.3761,
      "step": 1405
    },
    {
      "epoch": 0.21840776699029127,
      "grad_norm": 0.0,
      "learning_rate": 0.0009098524394016385,
      "loss": 2.3832,
      "step": 1406
    },
    {
      "epoch": 0.2185631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0009097082692547743,
      "loss": 2.2968,
      "step": 1407
    },
    {
      "epoch": 0.21871844660194176,
      "grad_norm": 0.0,
      "learning_rate": 0.0009095639953579839,
      "loss": 2.4064,
      "step": 1408
    },
    {
      "epoch": 0.218873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0009094196177478019,
      "loss": 2.4542,
      "step": 1409
    },
    {
      "epoch": 0.21902912621359225,
      "grad_norm": 0.0,
      "learning_rate": 0.0009092751364607886,
      "loss": 2.3079,
      "step": 1410
    },
    {
      "epoch": 0.21918446601941746,
      "grad_norm": 0.0,
      "learning_rate": 0.000909130551533531,
      "loss": 2.4081,
      "step": 1411
    },
    {
      "epoch": 0.2193398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.0009089858630026422,
      "loss": 2.3684,
      "step": 1412
    },
    {
      "epoch": 0.21949514563106795,
      "grad_norm": 0.0,
      "learning_rate": 0.0009088410709047612,
      "loss": 2.3713,
      "step": 1413
    },
    {
      "epoch": 0.2196504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0009086961752765538,
      "loss": 2.4247,
      "step": 1414
    },
    {
      "epoch": 0.21980582524271844,
      "grad_norm": 0.0,
      "learning_rate": 0.0009085511761547116,
      "loss": 2.365,
      "step": 1415
    },
    {
      "epoch": 0.21996116504854368,
      "grad_norm": 0.0,
      "learning_rate": 0.0009084060735759527,
      "loss": 2.4214,
      "step": 1416
    },
    {
      "epoch": 0.22011650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.0009082608675770211,
      "loss": 2.4166,
      "step": 1417
    },
    {
      "epoch": 0.22027184466019417,
      "grad_norm": 0.0,
      "learning_rate": 0.0009081155581946871,
      "loss": 2.3883,
      "step": 1418
    },
    {
      "epoch": 0.22042718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.0009079701454657477,
      "loss": 2.3712,
      "step": 1419
    },
    {
      "epoch": 0.22058252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0009078246294270248,
      "loss": 2.4463,
      "step": 1420
    },
    {
      "epoch": 0.2207378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0009076790101153678,
      "loss": 2.463,
      "step": 1421
    },
    {
      "epoch": 0.22089320388349515,
      "grad_norm": 0.0,
      "learning_rate": 0.0009075332875676516,
      "loss": 2.4564,
      "step": 1422
    },
    {
      "epoch": 0.2210485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0009073874618207772,
      "loss": 2.434,
      "step": 1423
    },
    {
      "epoch": 0.22120388349514564,
      "grad_norm": 0.0,
      "learning_rate": 0.0009072415329116721,
      "loss": 2.4275,
      "step": 1424
    },
    {
      "epoch": 0.22135922330097088,
      "grad_norm": 0.0,
      "learning_rate": 0.0009070955008772895,
      "loss": 2.3283,
      "step": 1425
    },
    {
      "epoch": 0.22151456310679613,
      "grad_norm": 0.0,
      "learning_rate": 0.0009069493657546091,
      "loss": 2.3516,
      "step": 1426
    },
    {
      "epoch": 0.22166990291262137,
      "grad_norm": 0.0,
      "learning_rate": 0.0009068031275806363,
      "loss": 2.4467,
      "step": 1427
    },
    {
      "epoch": 0.2218252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.000906656786392403,
      "loss": 2.3541,
      "step": 1428
    },
    {
      "epoch": 0.22198058252427186,
      "grad_norm": 0.0,
      "learning_rate": 0.0009065103422269667,
      "loss": 2.3179,
      "step": 1429
    },
    {
      "epoch": 0.22213592233009707,
      "grad_norm": 0.0,
      "learning_rate": 0.0009063637951214116,
      "loss": 2.4039,
      "step": 1430
    },
    {
      "epoch": 0.22229126213592232,
      "grad_norm": 0.0,
      "learning_rate": 0.0009062171451128478,
      "loss": 2.4446,
      "step": 1431
    },
    {
      "epoch": 0.22244660194174756,
      "grad_norm": 0.0,
      "learning_rate": 0.0009060703922384108,
      "loss": 2.4381,
      "step": 1432
    },
    {
      "epoch": 0.2226019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.0009059235365352629,
      "loss": 2.3828,
      "step": 1433
    },
    {
      "epoch": 0.22275728155339805,
      "grad_norm": 0.0,
      "learning_rate": 0.0009057765780405925,
      "loss": 2.3945,
      "step": 1434
    },
    {
      "epoch": 0.2229126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0009056295167916131,
      "loss": 2.3878,
      "step": 1435
    },
    {
      "epoch": 0.22306796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.0009054823528255651,
      "loss": 2.3286,
      "step": 1436
    },
    {
      "epoch": 0.22322330097087378,
      "grad_norm": 0.0,
      "learning_rate": 0.0009053350861797149,
      "loss": 2.4059,
      "step": 1437
    },
    {
      "epoch": 0.22337864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0009051877168913544,
      "loss": 2.4368,
      "step": 1438
    },
    {
      "epoch": 0.22353398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.000905040244997802,
      "loss": 2.4567,
      "step": 1439
    },
    {
      "epoch": 0.22368932038834952,
      "grad_norm": 0.0,
      "learning_rate": 0.0009048926705364014,
      "loss": 2.486,
      "step": 1440
    },
    {
      "epoch": 0.22384466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.0009047449935445229,
      "loss": 2.3944,
      "step": 1441
    },
    {
      "epoch": 0.224,
      "grad_norm": 0.0,
      "learning_rate": 0.0009045972140595627,
      "loss": 2.4171,
      "step": 1442
    },
    {
      "epoch": 0.22415533980582525,
      "grad_norm": 0.0,
      "learning_rate": 0.0009044493321189427,
      "loss": 2.3425,
      "step": 1443
    },
    {
      "epoch": 0.2243106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0009043013477601109,
      "loss": 2.4065,
      "step": 1444
    },
    {
      "epoch": 0.22446601941747574,
      "grad_norm": 0.0,
      "learning_rate": 0.0009041532610205411,
      "loss": 2.3547,
      "step": 1445
    },
    {
      "epoch": 0.22462135922330098,
      "grad_norm": 0.0,
      "learning_rate": 0.0009040050719377333,
      "loss": 2.4135,
      "step": 1446
    },
    {
      "epoch": 0.22477669902912623,
      "grad_norm": 0.0,
      "learning_rate": 0.000903856780549213,
      "loss": 2.3248,
      "step": 1447
    },
    {
      "epoch": 0.22493203883495147,
      "grad_norm": 0.0,
      "learning_rate": 0.0009037083868925323,
      "loss": 2.4087,
      "step": 1448
    },
    {
      "epoch": 0.2250873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0009035598910052681,
      "loss": 2.2979,
      "step": 1449
    },
    {
      "epoch": 0.22524271844660193,
      "grad_norm": 0.0,
      "learning_rate": 0.0009034112929250244,
      "loss": 2.4745,
      "step": 1450
    },
    {
      "epoch": 0.22539805825242717,
      "grad_norm": 0.0,
      "learning_rate": 0.0009032625926894302,
      "loss": 2.3645,
      "step": 1451
    },
    {
      "epoch": 0.22555339805825242,
      "grad_norm": 0.0,
      "learning_rate": 0.0009031137903361408,
      "loss": 2.3701,
      "step": 1452
    },
    {
      "epoch": 0.22570873786407766,
      "grad_norm": 0.0,
      "learning_rate": 0.0009029648859028372,
      "loss": 2.371,
      "step": 1453
    },
    {
      "epoch": 0.2258640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.0009028158794272263,
      "loss": 2.4171,
      "step": 1454
    },
    {
      "epoch": 0.22601941747572815,
      "grad_norm": 0.0,
      "learning_rate": 0.0009026667709470408,
      "loss": 2.4431,
      "step": 1455
    },
    {
      "epoch": 0.2261747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0009025175605000392,
      "loss": 2.4775,
      "step": 1456
    },
    {
      "epoch": 0.22633009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0009023682481240061,
      "loss": 2.4174,
      "step": 1457
    },
    {
      "epoch": 0.22648543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.0009022188338567514,
      "loss": 2.3602,
      "step": 1458
    },
    {
      "epoch": 0.22664077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.0009020693177361112,
      "loss": 2.3616,
      "step": 1459
    },
    {
      "epoch": 0.22679611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0009019196997999475,
      "loss": 2.4531,
      "step": 1460
    },
    {
      "epoch": 0.22695145631067962,
      "grad_norm": 0.0,
      "learning_rate": 0.0009017699800861476,
      "loss": 2.3374,
      "step": 1461
    },
    {
      "epoch": 0.22710679611650486,
      "grad_norm": 0.0,
      "learning_rate": 0.000901620158632625,
      "loss": 2.3253,
      "step": 1462
    },
    {
      "epoch": 0.2272621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0009014702354773188,
      "loss": 2.3827,
      "step": 1463
    },
    {
      "epoch": 0.22741747572815535,
      "grad_norm": 0.0,
      "learning_rate": 0.0009013202106581937,
      "loss": 2.4365,
      "step": 1464
    },
    {
      "epoch": 0.2275728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.0009011700842132404,
      "loss": 2.4054,
      "step": 1465
    },
    {
      "epoch": 0.22772815533980584,
      "grad_norm": 0.0,
      "learning_rate": 0.0009010198561804752,
      "loss": 2.332,
      "step": 1466
    },
    {
      "epoch": 0.22788349514563105,
      "grad_norm": 0.0,
      "learning_rate": 0.0009008695265979403,
      "loss": 2.3962,
      "step": 1467
    },
    {
      "epoch": 0.2280388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0009007190955037033,
      "loss": 2.4124,
      "step": 1468
    },
    {
      "epoch": 0.22819417475728154,
      "grad_norm": 0.0,
      "learning_rate": 0.0009005685629358578,
      "loss": 2.4258,
      "step": 1469
    },
    {
      "epoch": 0.2283495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0009004179289325228,
      "loss": 2.3547,
      "step": 1470
    },
    {
      "epoch": 0.22850485436893203,
      "grad_norm": 0.0,
      "learning_rate": 0.0009002671935318434,
      "loss": 2.4131,
      "step": 1471
    },
    {
      "epoch": 0.22866019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.0009001163567719901,
      "loss": 2.4156,
      "step": 1472
    },
    {
      "epoch": 0.22881553398058252,
      "grad_norm": 0.0,
      "learning_rate": 0.000899965418691159,
      "loss": 2.436,
      "step": 1473
    },
    {
      "epoch": 0.22897087378640776,
      "grad_norm": 0.0,
      "learning_rate": 0.000899814379327572,
      "loss": 2.4397,
      "step": 1474
    },
    {
      "epoch": 0.229126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0008996632387194768,
      "loss": 2.4689,
      "step": 1475
    },
    {
      "epoch": 0.22928155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.0008995119969051462,
      "loss": 2.401,
      "step": 1476
    },
    {
      "epoch": 0.2294368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0008993606539228792,
      "loss": 2.4037,
      "step": 1477
    },
    {
      "epoch": 0.22959223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.000899209209811,
      "loss": 2.4783,
      "step": 1478
    },
    {
      "epoch": 0.22974757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0008990576646078591,
      "loss": 2.3154,
      "step": 1479
    },
    {
      "epoch": 0.22990291262135923,
      "grad_norm": 0.0,
      "learning_rate": 0.0008989060183518317,
      "loss": 2.3446,
      "step": 1480
    },
    {
      "epoch": 0.23005825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.000898754271081319,
      "loss": 2.3763,
      "step": 1481
    },
    {
      "epoch": 0.23021359223300972,
      "grad_norm": 0.0,
      "learning_rate": 0.0008986024228347478,
      "loss": 2.3812,
      "step": 1482
    },
    {
      "epoch": 0.23036893203883496,
      "grad_norm": 0.0,
      "learning_rate": 0.0008984504736505706,
      "loss": 2.3721,
      "step": 1483
    },
    {
      "epoch": 0.2305242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0008982984235672651,
      "loss": 2.3246,
      "step": 1484
    },
    {
      "epoch": 0.23067961165048545,
      "grad_norm": 0.0,
      "learning_rate": 0.0008981462726233349,
      "loss": 2.3386,
      "step": 1485
    },
    {
      "epoch": 0.23083495145631067,
      "grad_norm": 0.0,
      "learning_rate": 0.0008979940208573091,
      "loss": 2.3636,
      "step": 1486
    },
    {
      "epoch": 0.2309902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.000897841668307742,
      "loss": 2.4088,
      "step": 1487
    },
    {
      "epoch": 0.23114563106796115,
      "grad_norm": 0.0,
      "learning_rate": 0.0008976892150132137,
      "loss": 2.4648,
      "step": 1488
    },
    {
      "epoch": 0.2313009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0008975366610123299,
      "loss": 2.2942,
      "step": 1489
    },
    {
      "epoch": 0.23145631067961164,
      "grad_norm": 0.0,
      "learning_rate": 0.0008973840063437215,
      "loss": 2.4041,
      "step": 1490
    },
    {
      "epoch": 0.2316116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0008972312510460452,
      "loss": 2.3789,
      "step": 1491
    },
    {
      "epoch": 0.23176699029126213,
      "grad_norm": 0.0,
      "learning_rate": 0.0008970783951579829,
      "loss": 2.4131,
      "step": 1492
    },
    {
      "epoch": 0.23192233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.0008969254387182421,
      "loss": 2.3379,
      "step": 1493
    },
    {
      "epoch": 0.23207766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0008967723817655559,
      "loss": 2.4276,
      "step": 1494
    },
    {
      "epoch": 0.23223300970873786,
      "grad_norm": 0.0,
      "learning_rate": 0.0008966192243386826,
      "loss": 2.3673,
      "step": 1495
    },
    {
      "epoch": 0.2323883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0008964659664764063,
      "loss": 2.4752,
      "step": 1496
    },
    {
      "epoch": 0.23254368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0008963126082175361,
      "loss": 2.3937,
      "step": 1497
    },
    {
      "epoch": 0.2326990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0008961591496009067,
      "loss": 2.4124,
      "step": 1498
    },
    {
      "epoch": 0.23285436893203884,
      "grad_norm": 0.0,
      "learning_rate": 0.0008960055906653783,
      "loss": 2.4369,
      "step": 1499
    },
    {
      "epoch": 0.23300970873786409,
      "grad_norm": 0.0,
      "learning_rate": 0.0008958519314498366,
      "loss": 2.3551,
      "step": 1500
    },
    {
      "epoch": 0.23316504854368933,
      "grad_norm": 0.0,
      "learning_rate": 0.0008956981719931923,
      "loss": 2.5204,
      "step": 1501
    },
    {
      "epoch": 0.23332038834951457,
      "grad_norm": 0.0,
      "learning_rate": 0.0008955443123343816,
      "loss": 2.3871,
      "step": 1502
    },
    {
      "epoch": 0.23347572815533982,
      "grad_norm": 0.0,
      "learning_rate": 0.0008953903525123667,
      "loss": 2.2946,
      "step": 1503
    },
    {
      "epoch": 0.23363106796116506,
      "grad_norm": 0.0,
      "learning_rate": 0.0008952362925661342,
      "loss": 2.5227,
      "step": 1504
    },
    {
      "epoch": 0.23378640776699028,
      "grad_norm": 0.0,
      "learning_rate": 0.0008950821325346967,
      "loss": 2.4805,
      "step": 1505
    },
    {
      "epoch": 0.23394174757281552,
      "grad_norm": 0.0,
      "learning_rate": 0.0008949278724570919,
      "loss": 2.4377,
      "step": 1506
    },
    {
      "epoch": 0.23409708737864077,
      "grad_norm": 0.0,
      "learning_rate": 0.000894773512372383,
      "loss": 2.4554,
      "step": 1507
    },
    {
      "epoch": 0.234252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0008946190523196582,
      "loss": 2.3031,
      "step": 1508
    },
    {
      "epoch": 0.23440776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.0008944644923380315,
      "loss": 2.2686,
      "step": 1509
    },
    {
      "epoch": 0.2345631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0008943098324666417,
      "loss": 2.2977,
      "step": 1510
    },
    {
      "epoch": 0.23471844660194174,
      "grad_norm": 0.0,
      "learning_rate": 0.0008941550727446533,
      "loss": 2.3856,
      "step": 1511
    },
    {
      "epoch": 0.234873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0008940002132112559,
      "loss": 2.3059,
      "step": 1512
    },
    {
      "epoch": 0.23502912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.0008938452539056642,
      "loss": 2.3693,
      "step": 1513
    },
    {
      "epoch": 0.23518446601941748,
      "grad_norm": 0.0,
      "learning_rate": 0.0008936901948671186,
      "loss": 2.4016,
      "step": 1514
    },
    {
      "epoch": 0.23533980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0008935350361348842,
      "loss": 2.4254,
      "step": 1515
    },
    {
      "epoch": 0.23549514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0008933797777482522,
      "loss": 2.4293,
      "step": 1516
    },
    {
      "epoch": 0.2356504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.000893224419746538,
      "loss": 2.3358,
      "step": 1517
    },
    {
      "epoch": 0.23580582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.0008930689621690827,
      "loss": 2.3705,
      "step": 1518
    },
    {
      "epoch": 0.2359611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.000892913405055253,
      "loss": 2.3986,
      "step": 1519
    },
    {
      "epoch": 0.23611650485436894,
      "grad_norm": 0.0,
      "learning_rate": 0.0008927577484444402,
      "loss": 2.3762,
      "step": 1520
    },
    {
      "epoch": 0.23627184466019419,
      "grad_norm": 0.0,
      "learning_rate": 0.0008926019923760611,
      "loss": 2.42,
      "step": 1521
    },
    {
      "epoch": 0.23642718446601943,
      "grad_norm": 0.0,
      "learning_rate": 0.0008924461368895576,
      "loss": 2.3737,
      "step": 1522
    },
    {
      "epoch": 0.23658252427184465,
      "grad_norm": 0.0,
      "learning_rate": 0.0008922901820243968,
      "loss": 2.4219,
      "step": 1523
    },
    {
      "epoch": 0.2367378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0008921341278200709,
      "loss": 2.4019,
      "step": 1524
    },
    {
      "epoch": 0.23689320388349513,
      "grad_norm": 0.0,
      "learning_rate": 0.0008919779743160974,
      "loss": 2.3947,
      "step": 1525
    },
    {
      "epoch": 0.23704854368932038,
      "grad_norm": 0.0,
      "learning_rate": 0.0008918217215520188,
      "loss": 2.3508,
      "step": 1526
    },
    {
      "epoch": 0.23720388349514562,
      "grad_norm": 0.0,
      "learning_rate": 0.0008916653695674029,
      "loss": 2.4153,
      "step": 1527
    },
    {
      "epoch": 0.23735922330097087,
      "grad_norm": 0.0,
      "learning_rate": 0.0008915089184018424,
      "loss": 2.4108,
      "step": 1528
    },
    {
      "epoch": 0.2375145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.0008913523680949553,
      "loss": 2.457,
      "step": 1529
    },
    {
      "epoch": 0.23766990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0008911957186863846,
      "loss": 2.4163,
      "step": 1530
    },
    {
      "epoch": 0.2378252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0008910389702157985,
      "loss": 2.4008,
      "step": 1531
    },
    {
      "epoch": 0.23798058252427184,
      "grad_norm": 0.0,
      "learning_rate": 0.0008908821227228902,
      "loss": 2.3643,
      "step": 1532
    },
    {
      "epoch": 0.2381359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.000890725176247378,
      "loss": 2.2697,
      "step": 1533
    },
    {
      "epoch": 0.23829126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0008905681308290052,
      "loss": 2.3542,
      "step": 1534
    },
    {
      "epoch": 0.23844660194174758,
      "grad_norm": 0.0,
      "learning_rate": 0.0008904109865075404,
      "loss": 2.3867,
      "step": 1535
    },
    {
      "epoch": 0.23860194174757282,
      "grad_norm": 0.0,
      "learning_rate": 0.0008902537433227768,
      "loss": 2.3104,
      "step": 1536
    },
    {
      "epoch": 0.23875728155339807,
      "grad_norm": 0.0,
      "learning_rate": 0.0008900964013145332,
      "loss": 2.4883,
      "step": 1537
    },
    {
      "epoch": 0.2389126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0008899389605226529,
      "loss": 2.4342,
      "step": 1538
    },
    {
      "epoch": 0.23906796116504855,
      "grad_norm": 0.0,
      "learning_rate": 0.0008897814209870046,
      "loss": 2.2936,
      "step": 1539
    },
    {
      "epoch": 0.2392233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.0008896237827474816,
      "loss": 2.3478,
      "step": 1540
    },
    {
      "epoch": 0.23937864077669904,
      "grad_norm": 0.0,
      "learning_rate": 0.0008894660458440026,
      "loss": 2.4139,
      "step": 1541
    },
    {
      "epoch": 0.23953398058252426,
      "grad_norm": 0.0,
      "learning_rate": 0.0008893082103165111,
      "loss": 2.2944,
      "step": 1542
    },
    {
      "epoch": 0.2396893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0008891502762049757,
      "loss": 2.4021,
      "step": 1543
    },
    {
      "epoch": 0.23984466019417475,
      "grad_norm": 0.0,
      "learning_rate": 0.0008889922435493898,
      "loss": 2.3524,
      "step": 1544
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.0,
      "learning_rate": 0.0008888341123897718,
      "loss": 2.3622,
      "step": 1545
    },
    {
      "epoch": 0.24015533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.000888675882766165,
      "loss": 2.3905,
      "step": 1546
    },
    {
      "epoch": 0.24031067961165048,
      "grad_norm": 0.0,
      "learning_rate": 0.0008885175547186377,
      "loss": 2.4432,
      "step": 1547
    },
    {
      "epoch": 0.24046601941747572,
      "grad_norm": 0.0,
      "learning_rate": 0.0008883591282872834,
      "loss": 2.3485,
      "step": 1548
    },
    {
      "epoch": 0.24062135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0008882006035122199,
      "loss": 2.4438,
      "step": 1549
    },
    {
      "epoch": 0.2407766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0008880419804335902,
      "loss": 2.3466,
      "step": 1550
    },
    {
      "epoch": 0.24093203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.0008878832590915625,
      "loss": 2.3927,
      "step": 1551
    },
    {
      "epoch": 0.2410873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0008877244395263294,
      "loss": 2.3323,
      "step": 1552
    },
    {
      "epoch": 0.24124271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.0008875655217781087,
      "loss": 2.3138,
      "step": 1553
    },
    {
      "epoch": 0.2413980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.000887406505887143,
      "loss": 2.3737,
      "step": 1554
    },
    {
      "epoch": 0.24155339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0008872473918936994,
      "loss": 2.3519,
      "step": 1555
    },
    {
      "epoch": 0.24170873786407768,
      "grad_norm": 0.0,
      "learning_rate": 0.0008870881798380705,
      "loss": 2.3926,
      "step": 1556
    },
    {
      "epoch": 0.24186407766990292,
      "grad_norm": 0.0,
      "learning_rate": 0.0008869288697605733,
      "loss": 2.3396,
      "step": 1557
    },
    {
      "epoch": 0.24201941747572817,
      "grad_norm": 0.0,
      "learning_rate": 0.0008867694617015496,
      "loss": 2.4274,
      "step": 1558
    },
    {
      "epoch": 0.2421747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.000886609955701366,
      "loss": 2.3245,
      "step": 1559
    },
    {
      "epoch": 0.24233009708737865,
      "grad_norm": 0.0,
      "learning_rate": 0.0008864503518004143,
      "loss": 2.3258,
      "step": 1560
    },
    {
      "epoch": 0.24248543689320387,
      "grad_norm": 0.0,
      "learning_rate": 0.0008862906500391106,
      "loss": 2.4142,
      "step": 1561
    },
    {
      "epoch": 0.24264077669902911,
      "grad_norm": 0.0,
      "learning_rate": 0.000886130850457896,
      "loss": 2.4125,
      "step": 1562
    },
    {
      "epoch": 0.24279611650485436,
      "grad_norm": 0.0,
      "learning_rate": 0.0008859709530972364,
      "loss": 2.4287,
      "step": 1563
    },
    {
      "epoch": 0.2429514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0008858109579976223,
      "loss": 2.3628,
      "step": 1564
    },
    {
      "epoch": 0.24310679611650485,
      "grad_norm": 0.0,
      "learning_rate": 0.0008856508651995694,
      "loss": 2.3737,
      "step": 1565
    },
    {
      "epoch": 0.2432621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0008854906747436175,
      "loss": 2.3523,
      "step": 1566
    },
    {
      "epoch": 0.24341747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0008853303866703314,
      "loss": 2.2973,
      "step": 1567
    },
    {
      "epoch": 0.24357281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.0008851700010203007,
      "loss": 2.4434,
      "step": 1568
    },
    {
      "epoch": 0.24372815533980582,
      "grad_norm": 0.0,
      "learning_rate": 0.0008850095178341399,
      "loss": 2.3451,
      "step": 1569
    },
    {
      "epoch": 0.24388349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.0008848489371524877,
      "loss": 2.4347,
      "step": 1570
    },
    {
      "epoch": 0.2440388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0008846882590160078,
      "loss": 2.474,
      "step": 1571
    },
    {
      "epoch": 0.24419417475728156,
      "grad_norm": 0.0,
      "learning_rate": 0.0008845274834653886,
      "loss": 2.3317,
      "step": 1572
    },
    {
      "epoch": 0.2443495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.000884366610541343,
      "loss": 2.4866,
      "step": 1573
    },
    {
      "epoch": 0.24450485436893205,
      "grad_norm": 0.0,
      "learning_rate": 0.0008842056402846089,
      "loss": 2.3895,
      "step": 1574
    },
    {
      "epoch": 0.2446601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.000884044572735948,
      "loss": 2.4148,
      "step": 1575
    },
    {
      "epoch": 0.24481553398058253,
      "grad_norm": 0.0,
      "learning_rate": 0.000883883407936148,
      "loss": 2.3672,
      "step": 1576
    },
    {
      "epoch": 0.24497087378640778,
      "grad_norm": 0.0,
      "learning_rate": 0.00088372214592602,
      "loss": 2.3672,
      "step": 1577
    },
    {
      "epoch": 0.24512621359223302,
      "grad_norm": 0.0,
      "learning_rate": 0.0008835607867464004,
      "loss": 2.498,
      "step": 1578
    },
    {
      "epoch": 0.24528155339805827,
      "grad_norm": 0.0,
      "learning_rate": 0.0008833993304381498,
      "loss": 2.3247,
      "step": 1579
    },
    {
      "epoch": 0.24543689320388348,
      "grad_norm": 0.0,
      "learning_rate": 0.0008832377770421536,
      "loss": 2.4146,
      "step": 1580
    },
    {
      "epoch": 0.24559223300970873,
      "grad_norm": 0.0,
      "learning_rate": 0.0008830761265993221,
      "loss": 2.3145,
      "step": 1581
    },
    {
      "epoch": 0.24574757281553397,
      "grad_norm": 0.0,
      "learning_rate": 0.0008829143791505893,
      "loss": 2.3805,
      "step": 1582
    },
    {
      "epoch": 0.24590291262135922,
      "grad_norm": 0.0,
      "learning_rate": 0.0008827525347369148,
      "loss": 2.3224,
      "step": 1583
    },
    {
      "epoch": 0.24605825242718446,
      "grad_norm": 0.0,
      "learning_rate": 0.0008825905933992819,
      "loss": 2.3394,
      "step": 1584
    },
    {
      "epoch": 0.2462135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.000882428555178699,
      "loss": 2.4688,
      "step": 1585
    },
    {
      "epoch": 0.24636893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0008822664201161989,
      "loss": 2.4117,
      "step": 1586
    },
    {
      "epoch": 0.2465242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0008821041882528386,
      "loss": 2.349,
      "step": 1587
    },
    {
      "epoch": 0.24667961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.0008819418596297004,
      "loss": 2.3434,
      "step": 1588
    },
    {
      "epoch": 0.24683495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.00088177943428789,
      "loss": 2.3105,
      "step": 1589
    },
    {
      "epoch": 0.24699029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.0008816169122685384,
      "loss": 2.3878,
      "step": 1590
    },
    {
      "epoch": 0.24714563106796117,
      "grad_norm": 0.0,
      "learning_rate": 0.0008814542936128008,
      "loss": 2.4263,
      "step": 1591
    },
    {
      "epoch": 0.2473009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0008812915783618571,
      "loss": 2.3794,
      "step": 1592
    },
    {
      "epoch": 0.24745631067961166,
      "grad_norm": 0.0,
      "learning_rate": 0.0008811287665569112,
      "loss": 2.2898,
      "step": 1593
    },
    {
      "epoch": 0.2476116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0008809658582391921,
      "loss": 2.4462,
      "step": 1594
    },
    {
      "epoch": 0.24776699029126215,
      "grad_norm": 0.0,
      "learning_rate": 0.0008808028534499526,
      "loss": 2.337,
      "step": 1595
    },
    {
      "epoch": 0.2479223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0008806397522304702,
      "loss": 2.26,
      "step": 1596
    },
    {
      "epoch": 0.24807766990291263,
      "grad_norm": 0.0,
      "learning_rate": 0.000880476554622047,
      "loss": 2.4266,
      "step": 1597
    },
    {
      "epoch": 0.24823300970873785,
      "grad_norm": 0.0,
      "learning_rate": 0.0008803132606660091,
      "loss": 2.3945,
      "step": 1598
    },
    {
      "epoch": 0.2483883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0008801498704037075,
      "loss": 2.3975,
      "step": 1599
    },
    {
      "epoch": 0.24854368932038834,
      "grad_norm": 0.0,
      "learning_rate": 0.0008799863838765172,
      "loss": 2.3406,
      "step": 1600
    },
    {
      "epoch": 0.24869902912621358,
      "grad_norm": 0.0,
      "learning_rate": 0.0008798228011258376,
      "loss": 2.3997,
      "step": 1601
    },
    {
      "epoch": 0.24885436893203883,
      "grad_norm": 0.0,
      "learning_rate": 0.0008796591221930926,
      "loss": 2.432,
      "step": 1602
    },
    {
      "epoch": 0.24900970873786407,
      "grad_norm": 0.0,
      "learning_rate": 0.0008794953471197305,
      "loss": 2.4231,
      "step": 1603
    },
    {
      "epoch": 0.24916504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0008793314759472239,
      "loss": 2.3623,
      "step": 1604
    },
    {
      "epoch": 0.24932038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.0008791675087170696,
      "loss": 2.3643,
      "step": 1605
    },
    {
      "epoch": 0.2494757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0008790034454707887,
      "loss": 2.4177,
      "step": 1606
    },
    {
      "epoch": 0.24963106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0008788392862499269,
      "loss": 2.4857,
      "step": 1607
    },
    {
      "epoch": 0.2497864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0008786750310960541,
      "loss": 2.2955,
      "step": 1608
    },
    {
      "epoch": 0.24994174757281554,
      "grad_norm": 0.0,
      "learning_rate": 0.0008785106800507643,
      "loss": 2.405,
      "step": 1609
    },
    {
      "epoch": 0.2500970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.000878346233155676,
      "loss": 2.3461,
      "step": 1610
    },
    {
      "epoch": 0.250252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0008781816904524317,
      "loss": 2.4073,
      "step": 1611
    },
    {
      "epoch": 0.25040776699029127,
      "grad_norm": 0.0,
      "learning_rate": 0.0008780170519826986,
      "loss": 2.3503,
      "step": 1612
    },
    {
      "epoch": 0.2505631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0008778523177881677,
      "loss": 2.2698,
      "step": 1613
    },
    {
      "epoch": 0.25071844660194176,
      "grad_norm": 0.0,
      "learning_rate": 0.0008776874879105547,
      "loss": 2.2697,
      "step": 1614
    },
    {
      "epoch": 0.250873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0008775225623915989,
      "loss": 2.4366,
      "step": 1615
    },
    {
      "epoch": 0.25102912621359225,
      "grad_norm": 0.0,
      "learning_rate": 0.0008773575412730645,
      "loss": 2.3635,
      "step": 1616
    },
    {
      "epoch": 0.25118446601941746,
      "grad_norm": 0.0,
      "learning_rate": 0.0008771924245967395,
      "loss": 2.4552,
      "step": 1617
    },
    {
      "epoch": 0.25133980582524273,
      "grad_norm": 0.0,
      "learning_rate": 0.0008770272124044361,
      "loss": 2.4369,
      "step": 1618
    },
    {
      "epoch": 0.25149514563106795,
      "grad_norm": 0.0,
      "learning_rate": 0.0008768619047379908,
      "loss": 2.3318,
      "step": 1619
    },
    {
      "epoch": 0.2516504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0008766965016392646,
      "loss": 2.4614,
      "step": 1620
    },
    {
      "epoch": 0.25180582524271844,
      "grad_norm": 0.0,
      "learning_rate": 0.0008765310031501419,
      "loss": 2.3889,
      "step": 1621
    },
    {
      "epoch": 0.2519611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0008763654093125317,
      "loss": 2.4089,
      "step": 1622
    },
    {
      "epoch": 0.25211650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.0008761997201683675,
      "loss": 2.4066,
      "step": 1623
    },
    {
      "epoch": 0.2522718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.0008760339357596061,
      "loss": 2.3193,
      "step": 1624
    },
    {
      "epoch": 0.2524271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.0008758680561282293,
      "loss": 2.3337,
      "step": 1625
    },
    {
      "epoch": 0.25258252427184463,
      "grad_norm": 0.0,
      "learning_rate": 0.0008757020813162422,
      "loss": 2.3484,
      "step": 1626
    },
    {
      "epoch": 0.2527378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0008755360113656746,
      "loss": 2.3491,
      "step": 1627
    },
    {
      "epoch": 0.2528932038834951,
      "grad_norm": 0.0,
      "learning_rate": 0.0008753698463185804,
      "loss": 2.3928,
      "step": 1628
    },
    {
      "epoch": 0.2530485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0008752035862170369,
      "loss": 2.3813,
      "step": 1629
    },
    {
      "epoch": 0.2532038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.0008750372311031462,
      "loss": 2.4172,
      "step": 1630
    },
    {
      "epoch": 0.2533592233009709,
      "grad_norm": 0.0,
      "learning_rate": 0.0008748707810190344,
      "loss": 2.4875,
      "step": 1631
    },
    {
      "epoch": 0.2535145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.0008747042360068512,
      "loss": 2.3455,
      "step": 1632
    },
    {
      "epoch": 0.25366990291262137,
      "grad_norm": 0.0,
      "learning_rate": 0.0008745375961087707,
      "loss": 2.328,
      "step": 1633
    },
    {
      "epoch": 0.2538252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0008743708613669907,
      "loss": 2.3137,
      "step": 1634
    },
    {
      "epoch": 0.25398058252427186,
      "grad_norm": 0.0,
      "learning_rate": 0.0008742040318237334,
      "loss": 2.3612,
      "step": 1635
    },
    {
      "epoch": 0.2541359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.000874037107521245,
      "loss": 2.3228,
      "step": 1636
    },
    {
      "epoch": 0.25429126213592235,
      "grad_norm": 0.0,
      "learning_rate": 0.0008738700885017954,
      "loss": 2.2366,
      "step": 1637
    },
    {
      "epoch": 0.25444660194174756,
      "grad_norm": 0.0,
      "learning_rate": 0.0008737029748076784,
      "loss": 2.4382,
      "step": 1638
    },
    {
      "epoch": 0.25460194174757284,
      "grad_norm": 0.0,
      "learning_rate": 0.0008735357664812122,
      "loss": 2.332,
      "step": 1639
    },
    {
      "epoch": 0.25475728155339805,
      "grad_norm": 0.0,
      "learning_rate": 0.000873368463564739,
      "loss": 2.3567,
      "step": 1640
    },
    {
      "epoch": 0.2549126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0008732010661006244,
      "loss": 2.3932,
      "step": 1641
    },
    {
      "epoch": 0.25506796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.0008730335741312583,
      "loss": 2.2864,
      "step": 1642
    },
    {
      "epoch": 0.2552233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.0008728659876990545,
      "loss": 2.3319,
      "step": 1643
    },
    {
      "epoch": 0.25537864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0008726983068464507,
      "loss": 2.3989,
      "step": 1644
    },
    {
      "epoch": 0.25553398058252424,
      "grad_norm": 0.0,
      "learning_rate": 0.0008725305316159086,
      "loss": 2.4222,
      "step": 1645
    },
    {
      "epoch": 0.2556893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0008723626620499136,
      "loss": 2.3475,
      "step": 1646
    },
    {
      "epoch": 0.25584466019417473,
      "grad_norm": 0.0,
      "learning_rate": 0.0008721946981909752,
      "loss": 2.3315,
      "step": 1647
    },
    {
      "epoch": 0.256,
      "grad_norm": 0.0,
      "learning_rate": 0.0008720266400816266,
      "loss": 2.3937,
      "step": 1648
    },
    {
      "epoch": 0.2561553398058252,
      "grad_norm": 0.0,
      "learning_rate": 0.0008718584877644251,
      "loss": 2.3405,
      "step": 1649
    },
    {
      "epoch": 0.2563106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0008716902412819515,
      "loss": 2.3589,
      "step": 1650
    },
    {
      "epoch": 0.2564660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.0008715219006768109,
      "loss": 2.3767,
      "step": 1651
    },
    {
      "epoch": 0.256621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0008713534659916318,
      "loss": 2.3225,
      "step": 1652
    },
    {
      "epoch": 0.2567766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0008711849372690666,
      "loss": 2.2584,
      "step": 1653
    },
    {
      "epoch": 0.25693203883495147,
      "grad_norm": 0.0,
      "learning_rate": 0.000871016314551792,
      "loss": 2.4312,
      "step": 1654
    },
    {
      "epoch": 0.2570873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0008708475978825078,
      "loss": 2.3197,
      "step": 1655
    },
    {
      "epoch": 0.25724271844660196,
      "grad_norm": 0.0,
      "learning_rate": 0.0008706787873039379,
      "loss": 2.3521,
      "step": 1656
    },
    {
      "epoch": 0.2573980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0008705098828588303,
      "loss": 2.3228,
      "step": 1657
    },
    {
      "epoch": 0.25755339805825245,
      "grad_norm": 0.0,
      "learning_rate": 0.0008703408845899563,
      "loss": 2.3503,
      "step": 1658
    },
    {
      "epoch": 0.25770873786407766,
      "grad_norm": 0.0,
      "learning_rate": 0.0008701717925401109,
      "loss": 2.3982,
      "step": 1659
    },
    {
      "epoch": 0.25786407766990294,
      "grad_norm": 0.0,
      "learning_rate": 0.0008700026067521135,
      "loss": 2.4425,
      "step": 1660
    },
    {
      "epoch": 0.25801941747572815,
      "grad_norm": 0.0,
      "learning_rate": 0.0008698333272688065,
      "loss": 2.41,
      "step": 1661
    },
    {
      "epoch": 0.2581747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0008696639541330565,
      "loss": 2.3848,
      "step": 1662
    },
    {
      "epoch": 0.25833009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0008694944873877536,
      "loss": 2.3015,
      "step": 1663
    },
    {
      "epoch": 0.25848543689320386,
      "grad_norm": 0.0,
      "learning_rate": 0.0008693249270758116,
      "loss": 2.3687,
      "step": 1664
    },
    {
      "epoch": 0.25864077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.000869155273240168,
      "loss": 2.3876,
      "step": 1665
    },
    {
      "epoch": 0.25879611650485435,
      "grad_norm": 0.0,
      "learning_rate": 0.0008689855259237843,
      "loss": 2.386,
      "step": 1666
    },
    {
      "epoch": 0.2589514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0008688156851696452,
      "loss": 2.392,
      "step": 1667
    },
    {
      "epoch": 0.25910679611650483,
      "grad_norm": 0.0,
      "learning_rate": 0.0008686457510207593,
      "loss": 2.4176,
      "step": 1668
    },
    {
      "epoch": 0.2592621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0008684757235201588,
      "loss": 2.312,
      "step": 1669
    },
    {
      "epoch": 0.2594174757281553,
      "grad_norm": 0.0,
      "learning_rate": 0.0008683056027108995,
      "loss": 2.326,
      "step": 1670
    },
    {
      "epoch": 0.2595728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.000868135388636061,
      "loss": 2.4213,
      "step": 1671
    },
    {
      "epoch": 0.2597281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.0008679650813387463,
      "loss": 2.3896,
      "step": 1672
    },
    {
      "epoch": 0.2598834951456311,
      "grad_norm": 0.0,
      "learning_rate": 0.0008677946808620824,
      "loss": 2.3892,
      "step": 1673
    },
    {
      "epoch": 0.2600388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0008676241872492192,
      "loss": 2.3665,
      "step": 1674
    },
    {
      "epoch": 0.26019417475728157,
      "grad_norm": 0.0,
      "learning_rate": 0.0008674536005433309,
      "loss": 2.3096,
      "step": 1675
    },
    {
      "epoch": 0.2603495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0008672829207876147,
      "loss": 2.3737,
      "step": 1676
    },
    {
      "epoch": 0.26050485436893206,
      "grad_norm": 0.0,
      "learning_rate": 0.0008671121480252919,
      "loss": 2.3719,
      "step": 1677
    },
    {
      "epoch": 0.2606601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.0008669412822996067,
      "loss": 2.2968,
      "step": 1678
    },
    {
      "epoch": 0.26081553398058255,
      "grad_norm": 0.0,
      "learning_rate": 0.0008667703236538277,
      "loss": 2.2773,
      "step": 1679
    },
    {
      "epoch": 0.26097087378640776,
      "grad_norm": 0.0,
      "learning_rate": 0.0008665992721312463,
      "loss": 2.3866,
      "step": 1680
    },
    {
      "epoch": 0.26112621359223304,
      "grad_norm": 0.0,
      "learning_rate": 0.0008664281277751775,
      "loss": 2.3811,
      "step": 1681
    },
    {
      "epoch": 0.26128155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.0008662568906289603,
      "loss": 2.3019,
      "step": 1682
    },
    {
      "epoch": 0.26143689320388347,
      "grad_norm": 0.0,
      "learning_rate": 0.0008660855607359566,
      "loss": 2.3154,
      "step": 1683
    },
    {
      "epoch": 0.26159223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0008659141381395524,
      "loss": 2.3336,
      "step": 1684
    },
    {
      "epoch": 0.26174757281553396,
      "grad_norm": 0.0,
      "learning_rate": 0.0008657426228831564,
      "loss": 2.3752,
      "step": 1685
    },
    {
      "epoch": 0.26190291262135923,
      "grad_norm": 0.0,
      "learning_rate": 0.0008655710150102016,
      "loss": 2.2999,
      "step": 1686
    },
    {
      "epoch": 0.26205825242718445,
      "grad_norm": 0.0,
      "learning_rate": 0.0008653993145641438,
      "loss": 2.3636,
      "step": 1687
    },
    {
      "epoch": 0.2622135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0008652275215884626,
      "loss": 2.4161,
      "step": 1688
    },
    {
      "epoch": 0.26236893203883493,
      "grad_norm": 0.0,
      "learning_rate": 0.0008650556361266608,
      "loss": 2.3978,
      "step": 1689
    },
    {
      "epoch": 0.2625242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.000864883658222265,
      "loss": 2.3158,
      "step": 1690
    },
    {
      "epoch": 0.2626796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.0008647115879188246,
      "loss": 2.3718,
      "step": 1691
    },
    {
      "epoch": 0.2628349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.0008645394252599131,
      "loss": 2.3425,
      "step": 1692
    },
    {
      "epoch": 0.2629902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.0008643671702891269,
      "loss": 2.3406,
      "step": 1693
    },
    {
      "epoch": 0.2631456310679612,
      "grad_norm": 0.0,
      "learning_rate": 0.0008641948230500858,
      "loss": 2.4261,
      "step": 1694
    },
    {
      "epoch": 0.2633009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0008640223835864333,
      "loss": 2.4305,
      "step": 1695
    },
    {
      "epoch": 0.26345631067961167,
      "grad_norm": 0.0,
      "learning_rate": 0.0008638498519418358,
      "loss": 2.3478,
      "step": 1696
    },
    {
      "epoch": 0.2636116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0008636772281599836,
      "loss": 2.3083,
      "step": 1697
    },
    {
      "epoch": 0.26376699029126216,
      "grad_norm": 0.0,
      "learning_rate": 0.0008635045122845897,
      "loss": 2.2932,
      "step": 1698
    },
    {
      "epoch": 0.2639223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0008633317043593909,
      "loss": 2.4327,
      "step": 1699
    },
    {
      "epoch": 0.26407766990291265,
      "grad_norm": 0.0,
      "learning_rate": 0.000863158804428147,
      "loss": 2.3136,
      "step": 1700
    },
    {
      "epoch": 0.26423300970873786,
      "grad_norm": 0.0,
      "learning_rate": 0.0008629858125346413,
      "loss": 2.369,
      "step": 1701
    },
    {
      "epoch": 0.2643883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0008628127287226804,
      "loss": 2.3236,
      "step": 1702
    },
    {
      "epoch": 0.26454368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0008626395530360941,
      "loss": 2.3739,
      "step": 1703
    },
    {
      "epoch": 0.26469902912621357,
      "grad_norm": 0.0,
      "learning_rate": 0.0008624662855187354,
      "loss": 2.3365,
      "step": 1704
    },
    {
      "epoch": 0.26485436893203884,
      "grad_norm": 0.0,
      "learning_rate": 0.0008622929262144806,
      "loss": 2.36,
      "step": 1705
    },
    {
      "epoch": 0.26500970873786406,
      "grad_norm": 0.0,
      "learning_rate": 0.0008621194751672294,
      "loss": 2.4387,
      "step": 1706
    },
    {
      "epoch": 0.26516504854368933,
      "grad_norm": 0.0,
      "learning_rate": 0.0008619459324209045,
      "loss": 2.4048,
      "step": 1707
    },
    {
      "epoch": 0.26532038834951455,
      "grad_norm": 0.0,
      "learning_rate": 0.0008617722980194518,
      "loss": 2.3923,
      "step": 1708
    },
    {
      "epoch": 0.2654757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0008615985720068408,
      "loss": 2.2623,
      "step": 1709
    },
    {
      "epoch": 0.26563106796116503,
      "grad_norm": 0.0,
      "learning_rate": 0.0008614247544270637,
      "loss": 2.3602,
      "step": 1710
    },
    {
      "epoch": 0.2657864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0008612508453241361,
      "loss": 2.2825,
      "step": 1711
    },
    {
      "epoch": 0.2659417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.0008610768447420971,
      "loss": 2.3448,
      "step": 1712
    },
    {
      "epoch": 0.2660970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.0008609027527250083,
      "loss": 2.3386,
      "step": 1713
    },
    {
      "epoch": 0.266252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0008607285693169551,
      "loss": 2.4061,
      "step": 1714
    },
    {
      "epoch": 0.2664077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.0008605542945620454,
      "loss": 2.3696,
      "step": 1715
    },
    {
      "epoch": 0.2665631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0008603799285044108,
      "loss": 2.4172,
      "step": 1716
    },
    {
      "epoch": 0.26671844660194177,
      "grad_norm": 0.0,
      "learning_rate": 0.0008602054711882059,
      "loss": 2.3035,
      "step": 1717
    },
    {
      "epoch": 0.266873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0008600309226576082,
      "loss": 2.3152,
      "step": 1718
    },
    {
      "epoch": 0.26702912621359226,
      "grad_norm": 0.0,
      "learning_rate": 0.0008598562829568184,
      "loss": 2.3683,
      "step": 1719
    },
    {
      "epoch": 0.2671844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.0008596815521300603,
      "loss": 2.3502,
      "step": 1720
    },
    {
      "epoch": 0.2673398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.0008595067302215808,
      "loss": 2.3091,
      "step": 1721
    },
    {
      "epoch": 0.26749514563106797,
      "grad_norm": 0.0,
      "learning_rate": 0.0008593318172756497,
      "loss": 2.3605,
      "step": 1722
    },
    {
      "epoch": 0.2676504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0008591568133365603,
      "loss": 2.3956,
      "step": 1723
    },
    {
      "epoch": 0.26780582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.0008589817184486284,
      "loss": 2.3698,
      "step": 1724
    },
    {
      "epoch": 0.26796116504854367,
      "grad_norm": 0.0,
      "learning_rate": 0.0008588065326561932,
      "loss": 2.2596,
      "step": 1725
    },
    {
      "epoch": 0.26811650485436894,
      "grad_norm": 0.0,
      "learning_rate": 0.0008586312560036164,
      "loss": 2.3273,
      "step": 1726
    },
    {
      "epoch": 0.26827184466019416,
      "grad_norm": 0.0,
      "learning_rate": 0.0008584558885352838,
      "loss": 2.3882,
      "step": 1727
    },
    {
      "epoch": 0.26842718446601943,
      "grad_norm": 0.0,
      "learning_rate": 0.0008582804302956029,
      "loss": 2.3899,
      "step": 1728
    },
    {
      "epoch": 0.26858252427184465,
      "grad_norm": 0.0,
      "learning_rate": 0.0008581048813290048,
      "loss": 2.2489,
      "step": 1729
    },
    {
      "epoch": 0.2687378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.000857929241679944,
      "loss": 2.3337,
      "step": 1730
    },
    {
      "epoch": 0.26889320388349514,
      "grad_norm": 0.0,
      "learning_rate": 0.0008577535113928972,
      "loss": 2.3713,
      "step": 1731
    },
    {
      "epoch": 0.2690485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0008575776905123642,
      "loss": 2.3318,
      "step": 1732
    },
    {
      "epoch": 0.2692038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.0008574017790828684,
      "loss": 2.3829,
      "step": 1733
    },
    {
      "epoch": 0.2693592233009709,
      "grad_norm": 0.0,
      "learning_rate": 0.000857225777148955,
      "loss": 2.3619,
      "step": 1734
    },
    {
      "epoch": 0.2695145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.0008570496847551932,
      "loss": 2.3346,
      "step": 1735
    },
    {
      "epoch": 0.2696699029126214,
      "grad_norm": 0.0,
      "learning_rate": 0.0008568735019461746,
      "loss": 2.4203,
      "step": 1736
    },
    {
      "epoch": 0.2698252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0008566972287665138,
      "loss": 2.371,
      "step": 1737
    },
    {
      "epoch": 0.2699805825242718,
      "grad_norm": 0.0,
      "learning_rate": 0.000856520865260848,
      "loss": 2.3971,
      "step": 1738
    },
    {
      "epoch": 0.2701359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.000856344411473838,
      "loss": 2.3495,
      "step": 1739
    },
    {
      "epoch": 0.2702912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.0008561678674501663,
      "loss": 2.2972,
      "step": 1740
    },
    {
      "epoch": 0.2704466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.0008559912332345394,
      "loss": 2.3488,
      "step": 1741
    },
    {
      "epoch": 0.2706019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.0008558145088716861,
      "loss": 2.4146,
      "step": 1742
    },
    {
      "epoch": 0.27075728155339807,
      "grad_norm": 0.0,
      "learning_rate": 0.000855637694406358,
      "loss": 2.295,
      "step": 1743
    },
    {
      "epoch": 0.2709126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0008554607898833296,
      "loss": 2.3766,
      "step": 1744
    },
    {
      "epoch": 0.27106796116504855,
      "grad_norm": 0.0,
      "learning_rate": 0.0008552837953473984,
      "loss": 2.3673,
      "step": 1745
    },
    {
      "epoch": 0.27122330097087377,
      "grad_norm": 0.0,
      "learning_rate": 0.0008551067108433843,
      "loss": 2.264,
      "step": 1746
    },
    {
      "epoch": 0.27137864077669904,
      "grad_norm": 0.0,
      "learning_rate": 0.0008549295364161302,
      "loss": 2.3486,
      "step": 1747
    },
    {
      "epoch": 0.27153398058252426,
      "grad_norm": 0.0,
      "learning_rate": 0.000854752272110502,
      "loss": 2.3397,
      "step": 1748
    },
    {
      "epoch": 0.27168932038834953,
      "grad_norm": 0.0,
      "learning_rate": 0.0008545749179713879,
      "loss": 2.3687,
      "step": 1749
    },
    {
      "epoch": 0.27184466019417475,
      "grad_norm": 0.0,
      "learning_rate": 0.0008543974740436991,
      "loss": 2.3554,
      "step": 1750
    },
    {
      "epoch": 0.272,
      "grad_norm": 0.0,
      "learning_rate": 0.0008542199403723696,
      "loss": 2.3418,
      "step": 1751
    },
    {
      "epoch": 0.27215533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.000854042317002356,
      "loss": 2.3994,
      "step": 1752
    },
    {
      "epoch": 0.2723106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0008538646039786375,
      "loss": 2.3807,
      "step": 1753
    },
    {
      "epoch": 0.2724660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.0008536868013462165,
      "loss": 2.4442,
      "step": 1754
    },
    {
      "epoch": 0.272621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0008535089091501174,
      "loss": 2.3242,
      "step": 1755
    },
    {
      "epoch": 0.2727766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0008533309274353877,
      "loss": 2.2402,
      "step": 1756
    },
    {
      "epoch": 0.27293203883495143,
      "grad_norm": 0.0,
      "learning_rate": 0.0008531528562470976,
      "loss": 2.3748,
      "step": 1757
    },
    {
      "epoch": 0.2730873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.00085297469563034,
      "loss": 2.4678,
      "step": 1758
    },
    {
      "epoch": 0.2732427184466019,
      "grad_norm": 0.0,
      "learning_rate": 0.0008527964456302297,
      "loss": 2.3113,
      "step": 1759
    },
    {
      "epoch": 0.2733980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0008526181062919054,
      "loss": 2.3535,
      "step": 1760
    },
    {
      "epoch": 0.2735533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.0008524396776605274,
      "loss": 2.3067,
      "step": 1761
    },
    {
      "epoch": 0.2737087378640777,
      "grad_norm": 0.0,
      "learning_rate": 0.000852261159781279,
      "loss": 2.371,
      "step": 1762
    },
    {
      "epoch": 0.2738640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.0008520825526993662,
      "loss": 2.3995,
      "step": 1763
    },
    {
      "epoch": 0.27401941747572817,
      "grad_norm": 0.0,
      "learning_rate": 0.0008519038564600174,
      "loss": 2.3506,
      "step": 1764
    },
    {
      "epoch": 0.2741747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0008517250711084834,
      "loss": 2.4509,
      "step": 1765
    },
    {
      "epoch": 0.27433009708737865,
      "grad_norm": 0.0,
      "learning_rate": 0.0008515461966900381,
      "loss": 2.4225,
      "step": 1766
    },
    {
      "epoch": 0.27448543689320387,
      "grad_norm": 0.0,
      "learning_rate": 0.0008513672332499776,
      "loss": 2.4139,
      "step": 1767
    },
    {
      "epoch": 0.27464077669902914,
      "grad_norm": 0.0,
      "learning_rate": 0.0008511881808336203,
      "loss": 2.3772,
      "step": 1768
    },
    {
      "epoch": 0.27479611650485436,
      "grad_norm": 0.0,
      "learning_rate": 0.0008510090394863079,
      "loss": 2.4514,
      "step": 1769
    },
    {
      "epoch": 0.27495145631067963,
      "grad_norm": 0.0,
      "learning_rate": 0.0008508298092534039,
      "loss": 2.418,
      "step": 1770
    },
    {
      "epoch": 0.27510679611650485,
      "grad_norm": 0.0,
      "learning_rate": 0.0008506504901802945,
      "loss": 2.3153,
      "step": 1771
    },
    {
      "epoch": 0.2752621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0008504710823123885,
      "loss": 2.3663,
      "step": 1772
    },
    {
      "epoch": 0.27541747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0008502915856951172,
      "loss": 2.3246,
      "step": 1773
    },
    {
      "epoch": 0.2755728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.0008501120003739342,
      "loss": 2.4451,
      "step": 1774
    },
    {
      "epoch": 0.2757281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.0008499323263943156,
      "loss": 2.369,
      "step": 1775
    },
    {
      "epoch": 0.27588349514563104,
      "grad_norm": 0.0,
      "learning_rate": 0.0008497525638017602,
      "loss": 2.3851,
      "step": 1776
    },
    {
      "epoch": 0.2760388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0008495727126417892,
      "loss": 2.3868,
      "step": 1777
    },
    {
      "epoch": 0.27619417475728153,
      "grad_norm": 0.0,
      "learning_rate": 0.0008493927729599458,
      "loss": 2.3179,
      "step": 1778
    },
    {
      "epoch": 0.2763495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.000849212744801796,
      "loss": 2.4613,
      "step": 1779
    },
    {
      "epoch": 0.276504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0008490326282129282,
      "loss": 2.396,
      "step": 1780
    },
    {
      "epoch": 0.2766601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.0008488524232389529,
      "loss": 2.3768,
      "step": 1781
    },
    {
      "epoch": 0.2768155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.0008486721299255034,
      "loss": 2.3496,
      "step": 1782
    },
    {
      "epoch": 0.2769708737864078,
      "grad_norm": 0.0,
      "learning_rate": 0.0008484917483182348,
      "loss": 2.3774,
      "step": 1783
    },
    {
      "epoch": 0.277126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0008483112784628255,
      "loss": 2.3148,
      "step": 1784
    },
    {
      "epoch": 0.27728155339805827,
      "grad_norm": 0.0,
      "learning_rate": 0.0008481307204049753,
      "loss": 2.3678,
      "step": 1785
    },
    {
      "epoch": 0.2774368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0008479500741904067,
      "loss": 2.3585,
      "step": 1786
    },
    {
      "epoch": 0.27759223300970876,
      "grad_norm": 0.0,
      "learning_rate": 0.0008477693398648645,
      "loss": 2.3699,
      "step": 1787
    },
    {
      "epoch": 0.27774757281553397,
      "grad_norm": 0.0,
      "learning_rate": 0.0008475885174741159,
      "loss": 2.4009,
      "step": 1788
    },
    {
      "epoch": 0.27790291262135924,
      "grad_norm": 0.0,
      "learning_rate": 0.0008474076070639504,
      "loss": 2.3561,
      "step": 1789
    },
    {
      "epoch": 0.27805825242718446,
      "grad_norm": 0.0,
      "learning_rate": 0.0008472266086801797,
      "loss": 2.4205,
      "step": 1790
    },
    {
      "epoch": 0.27821359223300973,
      "grad_norm": 0.0,
      "learning_rate": 0.0008470455223686375,
      "loss": 2.3375,
      "step": 1791
    },
    {
      "epoch": 0.27836893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0008468643481751805,
      "loss": 2.4156,
      "step": 1792
    },
    {
      "epoch": 0.2785242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0008466830861456871,
      "loss": 2.304,
      "step": 1793
    },
    {
      "epoch": 0.27867961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.0008465017363260579,
      "loss": 2.3235,
      "step": 1794
    },
    {
      "epoch": 0.27883495145631065,
      "grad_norm": 0.0,
      "learning_rate": 0.000846320298762216,
      "loss": 2.3373,
      "step": 1795
    },
    {
      "epoch": 0.2789902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.0008461387735001067,
      "loss": 2.4359,
      "step": 1796
    },
    {
      "epoch": 0.27914563106796114,
      "grad_norm": 0.0,
      "learning_rate": 0.0008459571605856972,
      "loss": 2.3835,
      "step": 1797
    },
    {
      "epoch": 0.2793009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0008457754600649773,
      "loss": 2.396,
      "step": 1798
    },
    {
      "epoch": 0.27945631067961163,
      "grad_norm": 0.0,
      "learning_rate": 0.0008455936719839588,
      "loss": 2.411,
      "step": 1799
    },
    {
      "epoch": 0.2796116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0008454117963886755,
      "loss": 2.2755,
      "step": 1800
    },
    {
      "epoch": 0.2797669902912621,
      "grad_norm": 0.0,
      "learning_rate": 0.0008452298333251838,
      "loss": 2.3368,
      "step": 1801
    },
    {
      "epoch": 0.2799223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0008450477828395618,
      "loss": 2.2899,
      "step": 1802
    },
    {
      "epoch": 0.2800776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.0008448656449779098,
      "loss": 2.2647,
      "step": 1803
    },
    {
      "epoch": 0.2802330097087379,
      "grad_norm": 0.0,
      "learning_rate": 0.0008446834197863507,
      "loss": 2.4185,
      "step": 1804
    },
    {
      "epoch": 0.2803883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0008445011073110289,
      "loss": 2.292,
      "step": 1805
    },
    {
      "epoch": 0.28054368932038837,
      "grad_norm": 0.0,
      "learning_rate": 0.0008443187075981112,
      "loss": 2.4376,
      "step": 1806
    },
    {
      "epoch": 0.2806990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0008441362206937866,
      "loss": 2.4071,
      "step": 1807
    },
    {
      "epoch": 0.28085436893203886,
      "grad_norm": 0.0,
      "learning_rate": 0.000843953646644266,
      "loss": 2.3597,
      "step": 1808
    },
    {
      "epoch": 0.2810097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.0008437709854957823,
      "loss": 2.4036,
      "step": 1809
    },
    {
      "epoch": 0.28116504854368934,
      "grad_norm": 0.0,
      "learning_rate": 0.0008435882372945905,
      "loss": 2.3905,
      "step": 1810
    },
    {
      "epoch": 0.28132038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.0008434054020869678,
      "loss": 2.4191,
      "step": 1811
    },
    {
      "epoch": 0.28147572815533983,
      "grad_norm": 0.0,
      "learning_rate": 0.0008432224799192134,
      "loss": 2.3327,
      "step": 1812
    },
    {
      "epoch": 0.28163106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0008430394708376486,
      "loss": 2.2466,
      "step": 1813
    },
    {
      "epoch": 0.28178640776699027,
      "grad_norm": 0.0,
      "learning_rate": 0.000842856374888616,
      "loss": 2.3274,
      "step": 1814
    },
    {
      "epoch": 0.28194174757281554,
      "grad_norm": 0.0,
      "learning_rate": 0.0008426731921184813,
      "loss": 2.3521,
      "step": 1815
    },
    {
      "epoch": 0.28209708737864075,
      "grad_norm": 0.0,
      "learning_rate": 0.0008424899225736315,
      "loss": 2.4189,
      "step": 1816
    },
    {
      "epoch": 0.282252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0008423065663004756,
      "loss": 2.3369,
      "step": 1817
    },
    {
      "epoch": 0.28240776699029124,
      "grad_norm": 0.0,
      "learning_rate": 0.0008421231233454449,
      "loss": 2.5309,
      "step": 1818
    },
    {
      "epoch": 0.2825631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0008419395937549923,
      "loss": 2.4694,
      "step": 1819
    },
    {
      "epoch": 0.28271844660194173,
      "grad_norm": 0.0,
      "learning_rate": 0.0008417559775755929,
      "loss": 2.2937,
      "step": 1820
    },
    {
      "epoch": 0.282873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0008415722748537433,
      "loss": 2.3619,
      "step": 1821
    },
    {
      "epoch": 0.2830291262135922,
      "grad_norm": 0.0,
      "learning_rate": 0.0008413884856359627,
      "loss": 2.3691,
      "step": 1822
    },
    {
      "epoch": 0.2831844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.0008412046099687915,
      "loss": 2.3651,
      "step": 1823
    },
    {
      "epoch": 0.2833398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.0008410206478987926,
      "loss": 2.3197,
      "step": 1824
    },
    {
      "epoch": 0.283495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0008408365994725503,
      "loss": 2.343,
      "step": 1825
    },
    {
      "epoch": 0.2836504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0008406524647366709,
      "loss": 2.2963,
      "step": 1826
    },
    {
      "epoch": 0.28380582524271847,
      "grad_norm": 0.0,
      "learning_rate": 0.0008404682437377828,
      "loss": 2.3478,
      "step": 1827
    },
    {
      "epoch": 0.2839611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0008402839365225359,
      "loss": 2.3234,
      "step": 1828
    },
    {
      "epoch": 0.28411650485436896,
      "grad_norm": 0.0,
      "learning_rate": 0.0008400995431376024,
      "loss": 2.325,
      "step": 1829
    },
    {
      "epoch": 0.2842718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.0008399150636296755,
      "loss": 2.2362,
      "step": 1830
    },
    {
      "epoch": 0.28442718446601944,
      "grad_norm": 0.0,
      "learning_rate": 0.0008397304980454711,
      "loss": 2.3312,
      "step": 1831
    },
    {
      "epoch": 0.28458252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0008395458464317264,
      "loss": 2.2786,
      "step": 1832
    },
    {
      "epoch": 0.2847378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0008393611088352006,
      "loss": 2.3348,
      "step": 1833
    },
    {
      "epoch": 0.28489320388349515,
      "grad_norm": 0.0,
      "learning_rate": 0.0008391762853026744,
      "loss": 2.3905,
      "step": 1834
    },
    {
      "epoch": 0.28504854368932037,
      "grad_norm": 0.0,
      "learning_rate": 0.0008389913758809506,
      "loss": 2.3867,
      "step": 1835
    },
    {
      "epoch": 0.28520388349514564,
      "grad_norm": 0.0,
      "learning_rate": 0.0008388063806168535,
      "loss": 2.3969,
      "step": 1836
    },
    {
      "epoch": 0.28535922330097085,
      "grad_norm": 0.0,
      "learning_rate": 0.0008386212995572292,
      "loss": 2.4769,
      "step": 1837
    },
    {
      "epoch": 0.2855145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.0008384361327489455,
      "loss": 2.3522,
      "step": 1838
    },
    {
      "epoch": 0.28566990291262134,
      "grad_norm": 0.0,
      "learning_rate": 0.0008382508802388924,
      "loss": 2.3259,
      "step": 1839
    },
    {
      "epoch": 0.2858252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0008380655420739806,
      "loss": 2.3579,
      "step": 1840
    },
    {
      "epoch": 0.28598058252427183,
      "grad_norm": 0.0,
      "learning_rate": 0.0008378801183011433,
      "loss": 2.3353,
      "step": 1841
    },
    {
      "epoch": 0.2861359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0008376946089673352,
      "loss": 2.3296,
      "step": 1842
    },
    {
      "epoch": 0.2862912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.0008375090141195325,
      "loss": 2.3837,
      "step": 1843
    },
    {
      "epoch": 0.2864466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.0008373233338047333,
      "loss": 2.3516,
      "step": 1844
    },
    {
      "epoch": 0.2866019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.0008371375680699569,
      "loss": 2.4219,
      "step": 1845
    },
    {
      "epoch": 0.2867572815533981,
      "grad_norm": 0.0,
      "learning_rate": 0.0008369517169622447,
      "loss": 2.394,
      "step": 1846
    },
    {
      "epoch": 0.2869126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0008367657805286597,
      "loss": 2.3149,
      "step": 1847
    },
    {
      "epoch": 0.28706796116504857,
      "grad_norm": 0.0,
      "learning_rate": 0.0008365797588162862,
      "loss": 2.4161,
      "step": 1848
    },
    {
      "epoch": 0.2872233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.0008363936518722302,
      "loss": 2.3228,
      "step": 1849
    },
    {
      "epoch": 0.287378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0008362074597436194,
      "loss": 2.3397,
      "step": 1850
    },
    {
      "epoch": 0.2875339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0008360211824776032,
      "loss": 2.2142,
      "step": 1851
    },
    {
      "epoch": 0.2876893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.000835834820121352,
      "loss": 2.385,
      "step": 1852
    },
    {
      "epoch": 0.28784466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.0008356483727220584,
      "loss": 2.4431,
      "step": 1853
    },
    {
      "epoch": 0.288,
      "grad_norm": 0.0,
      "learning_rate": 0.0008354618403269361,
      "loss": 2.392,
      "step": 1854
    },
    {
      "epoch": 0.28815533980582525,
      "grad_norm": 0.0,
      "learning_rate": 0.0008352752229832204,
      "loss": 2.3422,
      "step": 1855
    },
    {
      "epoch": 0.28831067961165047,
      "grad_norm": 0.0,
      "learning_rate": 0.0008350885207381684,
      "loss": 2.2974,
      "step": 1856
    },
    {
      "epoch": 0.28846601941747574,
      "grad_norm": 0.0,
      "learning_rate": 0.0008349017336390586,
      "loss": 2.3611,
      "step": 1857
    },
    {
      "epoch": 0.28862135922330096,
      "grad_norm": 0.0,
      "learning_rate": 0.0008347148617331903,
      "loss": 2.376,
      "step": 1858
    },
    {
      "epoch": 0.2887766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0008345279050678854,
      "loss": 2.2882,
      "step": 1859
    },
    {
      "epoch": 0.28893203883495144,
      "grad_norm": 0.0,
      "learning_rate": 0.0008343408636904864,
      "loss": 2.2961,
      "step": 1860
    },
    {
      "epoch": 0.2890873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0008341537376483577,
      "loss": 2.3788,
      "step": 1861
    },
    {
      "epoch": 0.28924271844660193,
      "grad_norm": 0.0,
      "learning_rate": 0.0008339665269888848,
      "loss": 2.3566,
      "step": 1862
    },
    {
      "epoch": 0.2893980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0008337792317594752,
      "loss": 2.249,
      "step": 1863
    },
    {
      "epoch": 0.2895533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.0008335918520075572,
      "loss": 2.368,
      "step": 1864
    },
    {
      "epoch": 0.2897087378640777,
      "grad_norm": 0.0,
      "learning_rate": 0.0008334043877805808,
      "loss": 2.3885,
      "step": 1865
    },
    {
      "epoch": 0.2898640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.0008332168391260173,
      "loss": 2.3805,
      "step": 1866
    },
    {
      "epoch": 0.2900194174757282,
      "grad_norm": 0.0,
      "learning_rate": 0.0008330292060913594,
      "loss": 2.3685,
      "step": 1867
    },
    {
      "epoch": 0.2901747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0008328414887241211,
      "loss": 2.4155,
      "step": 1868
    },
    {
      "epoch": 0.2903300970873786,
      "grad_norm": 0.0,
      "learning_rate": 0.0008326536870718382,
      "loss": 2.3661,
      "step": 1869
    },
    {
      "epoch": 0.2904854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.0008324658011820672,
      "loss": 2.3656,
      "step": 1870
    },
    {
      "epoch": 0.2906407766990291,
      "grad_norm": 0.0,
      "learning_rate": 0.0008322778311023861,
      "loss": 2.3623,
      "step": 1871
    },
    {
      "epoch": 0.2907961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.0008320897768803946,
      "loss": 2.3136,
      "step": 1872
    },
    {
      "epoch": 0.2909514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0008319016385637132,
      "loss": 2.3853,
      "step": 1873
    },
    {
      "epoch": 0.29110679611650486,
      "grad_norm": 0.0,
      "learning_rate": 0.0008317134161999842,
      "loss": 2.2839,
      "step": 1874
    },
    {
      "epoch": 0.2912621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0008315251098368708,
      "loss": 2.303,
      "step": 1875
    },
    {
      "epoch": 0.29141747572815535,
      "grad_norm": 0.0,
      "learning_rate": 0.0008313367195220576,
      "loss": 2.3336,
      "step": 1876
    },
    {
      "epoch": 0.29157281553398057,
      "grad_norm": 0.0,
      "learning_rate": 0.0008311482453032503,
      "loss": 2.3746,
      "step": 1877
    },
    {
      "epoch": 0.29172815533980584,
      "grad_norm": 0.0,
      "learning_rate": 0.0008309596872281762,
      "loss": 2.3821,
      "step": 1878
    },
    {
      "epoch": 0.29188349514563106,
      "grad_norm": 0.0,
      "learning_rate": 0.0008307710453445836,
      "loss": 2.3605,
      "step": 1879
    },
    {
      "epoch": 0.2920388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.000830582319700242,
      "loss": 2.2584,
      "step": 1880
    },
    {
      "epoch": 0.29219417475728154,
      "grad_norm": 0.0,
      "learning_rate": 0.0008303935103429421,
      "loss": 2.3894,
      "step": 1881
    },
    {
      "epoch": 0.2923495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.000830204617320496,
      "loss": 2.3638,
      "step": 1882
    },
    {
      "epoch": 0.29250485436893203,
      "grad_norm": 0.0,
      "learning_rate": 0.0008300156406807369,
      "loss": 2.3812,
      "step": 1883
    },
    {
      "epoch": 0.2926601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.0008298265804715188,
      "loss": 2.4205,
      "step": 1884
    },
    {
      "epoch": 0.2928155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.0008296374367407175,
      "loss": 2.3335,
      "step": 1885
    },
    {
      "epoch": 0.2929708737864078,
      "grad_norm": 0.0,
      "learning_rate": 0.0008294482095362296,
      "loss": 2.3813,
      "step": 1886
    },
    {
      "epoch": 0.293126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0008292588989059728,
      "loss": 2.2084,
      "step": 1887
    },
    {
      "epoch": 0.2932815533980582,
      "grad_norm": 0.0,
      "learning_rate": 0.000829069504897886,
      "loss": 2.3931,
      "step": 1888
    },
    {
      "epoch": 0.2934368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0008288800275599292,
      "loss": 2.4441,
      "step": 1889
    },
    {
      "epoch": 0.2935922330097087,
      "grad_norm": 0.0,
      "learning_rate": 0.0008286904669400836,
      "loss": 2.4078,
      "step": 1890
    },
    {
      "epoch": 0.293747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0008285008230863513,
      "loss": 2.3668,
      "step": 1891
    },
    {
      "epoch": 0.2939029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.0008283110960467557,
      "loss": 2.3688,
      "step": 1892
    },
    {
      "epoch": 0.2940582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.000828121285869341,
      "loss": 2.2448,
      "step": 1893
    },
    {
      "epoch": 0.2942135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0008279313926021726,
      "loss": 2.3351,
      "step": 1894
    },
    {
      "epoch": 0.29436893203883496,
      "grad_norm": 0.0,
      "learning_rate": 0.0008277414162933373,
      "loss": 2.3193,
      "step": 1895
    },
    {
      "epoch": 0.2945242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.000827551356990942,
      "loss": 2.371,
      "step": 1896
    },
    {
      "epoch": 0.29467961165048545,
      "grad_norm": 0.0,
      "learning_rate": 0.0008273612147431155,
      "loss": 2.373,
      "step": 1897
    },
    {
      "epoch": 0.29483495145631067,
      "grad_norm": 0.0,
      "learning_rate": 0.0008271709895980075,
      "loss": 2.2877,
      "step": 1898
    },
    {
      "epoch": 0.29499029126213594,
      "grad_norm": 0.0,
      "learning_rate": 0.000826980681603788,
      "loss": 2.4323,
      "step": 1899
    },
    {
      "epoch": 0.29514563106796116,
      "grad_norm": 0.0,
      "learning_rate": 0.0008267902908086489,
      "loss": 2.3922,
      "step": 1900
    },
    {
      "epoch": 0.29530097087378643,
      "grad_norm": 0.0,
      "learning_rate": 0.0008265998172608022,
      "loss": 2.4065,
      "step": 1901
    },
    {
      "epoch": 0.29545631067961164,
      "grad_norm": 0.0,
      "learning_rate": 0.0008264092610084817,
      "loss": 2.3381,
      "step": 1902
    },
    {
      "epoch": 0.2956116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0008262186220999414,
      "loss": 2.2073,
      "step": 1903
    },
    {
      "epoch": 0.29576699029126213,
      "grad_norm": 0.0,
      "learning_rate": 0.000826027900583457,
      "loss": 2.3496,
      "step": 1904
    },
    {
      "epoch": 0.2959223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0008258370965073242,
      "loss": 2.3722,
      "step": 1905
    },
    {
      "epoch": 0.2960776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.0008256462099198603,
      "loss": 2.3142,
      "step": 1906
    },
    {
      "epoch": 0.29623300970873784,
      "grad_norm": 0.0,
      "learning_rate": 0.0008254552408694032,
      "loss": 2.2421,
      "step": 1907
    },
    {
      "epoch": 0.2963883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0008252641894043118,
      "loss": 2.4812,
      "step": 1908
    },
    {
      "epoch": 0.2965436893203883,
      "grad_norm": 0.0,
      "learning_rate": 0.000825073055572966,
      "loss": 2.3564,
      "step": 1909
    },
    {
      "epoch": 0.2966990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.000824881839423766,
      "loss": 2.3324,
      "step": 1910
    },
    {
      "epoch": 0.2968543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.0008246905410051335,
      "loss": 2.4364,
      "step": 1911
    },
    {
      "epoch": 0.2970097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.0008244991603655107,
      "loss": 2.4119,
      "step": 1912
    },
    {
      "epoch": 0.2971650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.0008243076975533609,
      "loss": 2.3204,
      "step": 1913
    },
    {
      "epoch": 0.2973203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.0008241161526171677,
      "loss": 2.3281,
      "step": 1914
    },
    {
      "epoch": 0.2974757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.000823924525605436,
      "loss": 2.2577,
      "step": 1915
    },
    {
      "epoch": 0.29763106796116506,
      "grad_norm": 0.0,
      "learning_rate": 0.000823732816566691,
      "loss": 2.254,
      "step": 1916
    },
    {
      "epoch": 0.2977864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0008235410255494791,
      "loss": 2.2799,
      "step": 1917
    },
    {
      "epoch": 0.29794174757281555,
      "grad_norm": 0.0,
      "learning_rate": 0.0008233491526023674,
      "loss": 2.3097,
      "step": 1918
    },
    {
      "epoch": 0.29809708737864077,
      "grad_norm": 0.0,
      "learning_rate": 0.0008231571977739436,
      "loss": 2.3792,
      "step": 1919
    },
    {
      "epoch": 0.29825242718446604,
      "grad_norm": 0.0,
      "learning_rate": 0.0008229651611128162,
      "loss": 2.3682,
      "step": 1920
    },
    {
      "epoch": 0.29840776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.0008227730426676142,
      "loss": 2.3356,
      "step": 1921
    },
    {
      "epoch": 0.29856310679611653,
      "grad_norm": 0.0,
      "learning_rate": 0.0008225808424869878,
      "loss": 2.3949,
      "step": 1922
    },
    {
      "epoch": 0.29871844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.0008223885606196076,
      "loss": 2.3374,
      "step": 1923
    },
    {
      "epoch": 0.298873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0008221961971141646,
      "loss": 2.3909,
      "step": 1924
    },
    {
      "epoch": 0.29902912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.000822003752019371,
      "loss": 2.3335,
      "step": 1925
    },
    {
      "epoch": 0.29918446601941745,
      "grad_norm": 0.0,
      "learning_rate": 0.0008218112253839597,
      "loss": 2.3372,
      "step": 1926
    },
    {
      "epoch": 0.2993398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.0008216186172566836,
      "loss": 2.3424,
      "step": 1927
    },
    {
      "epoch": 0.29949514563106794,
      "grad_norm": 0.0,
      "learning_rate": 0.0008214259276863165,
      "loss": 2.2461,
      "step": 1928
    },
    {
      "epoch": 0.2996504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0008212331567216534,
      "loss": 2.4315,
      "step": 1929
    },
    {
      "epoch": 0.2998058252427184,
      "grad_norm": 0.0,
      "learning_rate": 0.0008210403044115092,
      "loss": 2.3901,
      "step": 1930
    },
    {
      "epoch": 0.2999611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0008208473708047197,
      "loss": 2.3433,
      "step": 1931
    },
    {
      "epoch": 0.3001165048543689,
      "grad_norm": 0.0,
      "learning_rate": 0.0008206543559501413,
      "loss": 2.3843,
      "step": 1932
    },
    {
      "epoch": 0.3002718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.0008204612598966507,
      "loss": 2.3412,
      "step": 1933
    },
    {
      "epoch": 0.3004271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.0008202680826931456,
      "loss": 2.4421,
      "step": 1934
    },
    {
      "epoch": 0.3005825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.0008200748243885438,
      "loss": 2.344,
      "step": 1935
    },
    {
      "epoch": 0.3007378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0008198814850317843,
      "loss": 2.3402,
      "step": 1936
    },
    {
      "epoch": 0.30089320388349516,
      "grad_norm": 0.0,
      "learning_rate": 0.0008196880646718257,
      "loss": 2.3337,
      "step": 1937
    },
    {
      "epoch": 0.3010485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.000819494563357648,
      "loss": 2.4226,
      "step": 1938
    },
    {
      "epoch": 0.30120388349514565,
      "grad_norm": 0.0,
      "learning_rate": 0.0008193009811382511,
      "loss": 2.2979,
      "step": 1939
    },
    {
      "epoch": 0.30135922330097087,
      "grad_norm": 0.0,
      "learning_rate": 0.0008191073180626553,
      "loss": 2.3457,
      "step": 1940
    },
    {
      "epoch": 0.30151456310679614,
      "grad_norm": 0.0,
      "learning_rate": 0.0008189135741799022,
      "loss": 2.3203,
      "step": 1941
    },
    {
      "epoch": 0.30166990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0008187197495390531,
      "loss": 2.4091,
      "step": 1942
    },
    {
      "epoch": 0.30182524271844663,
      "grad_norm": 0.0,
      "learning_rate": 0.0008185258441891899,
      "loss": 2.3918,
      "step": 1943
    },
    {
      "epoch": 0.30198058252427185,
      "grad_norm": 0.0,
      "learning_rate": 0.0008183318581794153,
      "loss": 2.34,
      "step": 1944
    },
    {
      "epoch": 0.30213592233009706,
      "grad_norm": 0.0,
      "learning_rate": 0.0008181377915588517,
      "loss": 2.4006,
      "step": 1945
    },
    {
      "epoch": 0.30229126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0008179436443766426,
      "loss": 2.3666,
      "step": 1946
    },
    {
      "epoch": 0.30244660194174755,
      "grad_norm": 0.0,
      "learning_rate": 0.0008177494166819517,
      "loss": 2.3966,
      "step": 1947
    },
    {
      "epoch": 0.3026019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.000817555108523963,
      "loss": 2.2841,
      "step": 1948
    },
    {
      "epoch": 0.30275728155339804,
      "grad_norm": 0.0,
      "learning_rate": 0.0008173607199518809,
      "loss": 2.2972,
      "step": 1949
    },
    {
      "epoch": 0.3029126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0008171662510149301,
      "loss": 2.316,
      "step": 1950
    },
    {
      "epoch": 0.3030679611650485,
      "grad_norm": 0.0,
      "learning_rate": 0.0008169717017623559,
      "loss": 2.3126,
      "step": 1951
    },
    {
      "epoch": 0.3032233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.0008167770722434236,
      "loss": 2.2844,
      "step": 1952
    },
    {
      "epoch": 0.303378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0008165823625074191,
      "loss": 2.3457,
      "step": 1953
    },
    {
      "epoch": 0.3035339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0008163875726036486,
      "loss": 2.4861,
      "step": 1954
    },
    {
      "epoch": 0.3036893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0008161927025814384,
      "loss": 2.2897,
      "step": 1955
    },
    {
      "epoch": 0.3038446601941748,
      "grad_norm": 0.0,
      "learning_rate": 0.0008159977524901349,
      "loss": 2.4128,
      "step": 1956
    },
    {
      "epoch": 0.304,
      "grad_norm": 0.0,
      "learning_rate": 0.0008158027223791058,
      "loss": 2.3528,
      "step": 1957
    },
    {
      "epoch": 0.30415533980582526,
      "grad_norm": 0.0,
      "learning_rate": 0.0008156076122977376,
      "loss": 2.3616,
      "step": 1958
    },
    {
      "epoch": 0.3043106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0008154124222954383,
      "loss": 2.3044,
      "step": 1959
    },
    {
      "epoch": 0.30446601941747575,
      "grad_norm": 0.0,
      "learning_rate": 0.0008152171524216352,
      "loss": 2.3925,
      "step": 1960
    },
    {
      "epoch": 0.30462135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0008150218027257768,
      "loss": 2.33,
      "step": 1961
    },
    {
      "epoch": 0.30477669902912624,
      "grad_norm": 0.0,
      "learning_rate": 0.0008148263732573307,
      "loss": 2.3885,
      "step": 1962
    },
    {
      "epoch": 0.30493203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.0008146308640657856,
      "loss": 2.4209,
      "step": 1963
    },
    {
      "epoch": 0.3050873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0008144352752006499,
      "loss": 2.3491,
      "step": 1964
    },
    {
      "epoch": 0.30524271844660195,
      "grad_norm": 0.0,
      "learning_rate": 0.0008142396067114524,
      "loss": 2.2825,
      "step": 1965
    },
    {
      "epoch": 0.30539805825242716,
      "grad_norm": 0.0,
      "learning_rate": 0.0008140438586477419,
      "loss": 2.3065,
      "step": 1966
    },
    {
      "epoch": 0.30555339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0008138480310590876,
      "loss": 2.3992,
      "step": 1967
    },
    {
      "epoch": 0.30570873786407765,
      "grad_norm": 0.0,
      "learning_rate": 0.0008136521239950787,
      "loss": 2.3767,
      "step": 1968
    },
    {
      "epoch": 0.3058640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.0008134561375053245,
      "loss": 2.2942,
      "step": 1969
    },
    {
      "epoch": 0.30601941747572814,
      "grad_norm": 0.0,
      "learning_rate": 0.000813260071639454,
      "loss": 2.3792,
      "step": 1970
    },
    {
      "epoch": 0.3061747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0008130639264471173,
      "loss": 2.3105,
      "step": 1971
    },
    {
      "epoch": 0.3063300970873786,
      "grad_norm": 0.0,
      "learning_rate": 0.0008128677019779838,
      "loss": 2.3947,
      "step": 1972
    },
    {
      "epoch": 0.3064854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.000812671398281743,
      "loss": 2.307,
      "step": 1973
    },
    {
      "epoch": 0.3066407766990291,
      "grad_norm": 0.0,
      "learning_rate": 0.0008124750154081049,
      "loss": 2.3691,
      "step": 1974
    },
    {
      "epoch": 0.3067961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.0008122785534067992,
      "loss": 2.3955,
      "step": 1975
    },
    {
      "epoch": 0.3069514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0008120820123275756,
      "loss": 2.3496,
      "step": 1976
    },
    {
      "epoch": 0.3071067961165049,
      "grad_norm": 0.0,
      "learning_rate": 0.0008118853922202043,
      "loss": 2.2843,
      "step": 1977
    },
    {
      "epoch": 0.3072621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0008116886931344749,
      "loss": 2.3796,
      "step": 1978
    },
    {
      "epoch": 0.30741747572815536,
      "grad_norm": 0.0,
      "learning_rate": 0.0008114919151201972,
      "loss": 2.2571,
      "step": 1979
    },
    {
      "epoch": 0.3075728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.0008112950582272011,
      "loss": 2.2156,
      "step": 1980
    },
    {
      "epoch": 0.3077281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.0008110981225053367,
      "loss": 2.3608,
      "step": 1981
    },
    {
      "epoch": 0.30788349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.0008109011080044737,
      "loss": 2.4329,
      "step": 1982
    },
    {
      "epoch": 0.3080388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0008107040147745015,
      "loss": 2.316,
      "step": 1983
    },
    {
      "epoch": 0.30819417475728156,
      "grad_norm": 0.0,
      "learning_rate": 0.0008105068428653302,
      "loss": 2.3865,
      "step": 1984
    },
    {
      "epoch": 0.3083495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0008103095923268892,
      "loss": 2.3077,
      "step": 1985
    },
    {
      "epoch": 0.30850485436893205,
      "grad_norm": 0.0,
      "learning_rate": 0.000810112263209128,
      "loss": 2.2859,
      "step": 1986
    },
    {
      "epoch": 0.30866019417475726,
      "grad_norm": 0.0,
      "learning_rate": 0.0008099148555620161,
      "loss": 2.3359,
      "step": 1987
    },
    {
      "epoch": 0.30881553398058254,
      "grad_norm": 0.0,
      "learning_rate": 0.0008097173694355426,
      "loss": 2.3773,
      "step": 1988
    },
    {
      "epoch": 0.30897087378640775,
      "grad_norm": 0.0,
      "learning_rate": 0.000809519804879717,
      "loss": 2.3494,
      "step": 1989
    },
    {
      "epoch": 0.309126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0008093221619445681,
      "loss": 2.3751,
      "step": 1990
    },
    {
      "epoch": 0.30928155339805824,
      "grad_norm": 0.0,
      "learning_rate": 0.0008091244406801449,
      "loss": 2.4666,
      "step": 1991
    },
    {
      "epoch": 0.3094368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0008089266411365159,
      "loss": 2.2921,
      "step": 1992
    },
    {
      "epoch": 0.30959223300970873,
      "grad_norm": 0.0,
      "learning_rate": 0.0008087287633637699,
      "loss": 2.2821,
      "step": 1993
    },
    {
      "epoch": 0.309747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0008085308074120151,
      "loss": 2.298,
      "step": 1994
    },
    {
      "epoch": 0.3099029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.0008083327733313798,
      "loss": 2.2848,
      "step": 1995
    },
    {
      "epoch": 0.3100582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.0008081346611720117,
      "loss": 2.3505,
      "step": 1996
    },
    {
      "epoch": 0.3102135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0008079364709840787,
      "loss": 2.3417,
      "step": 1997
    },
    {
      "epoch": 0.310368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.000807738202817768,
      "loss": 2.2609,
      "step": 1998
    },
    {
      "epoch": 0.3105242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.000807539856723287,
      "loss": 2.4999,
      "step": 1999
    },
    {
      "epoch": 0.3106796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.0008073414327508626,
      "loss": 2.418,
      "step": 2000
    },
    {
      "epoch": 0.3108349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.0008071429309507417,
      "loss": 2.3435,
      "step": 2001
    },
    {
      "epoch": 0.3109902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.0008069443513731903,
      "loss": 2.3229,
      "step": 2002
    },
    {
      "epoch": 0.31114563106796117,
      "grad_norm": 0.0,
      "learning_rate": 0.0008067456940684946,
      "loss": 2.347,
      "step": 2003
    },
    {
      "epoch": 0.3113009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0008065469590869603,
      "loss": 2.255,
      "step": 2004
    },
    {
      "epoch": 0.31145631067961166,
      "grad_norm": 0.0,
      "learning_rate": 0.0008063481464789131,
      "loss": 2.4384,
      "step": 2005
    },
    {
      "epoch": 0.3116116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0008061492562946977,
      "loss": 2.2548,
      "step": 2006
    },
    {
      "epoch": 0.31176699029126215,
      "grad_norm": 0.0,
      "learning_rate": 0.000805950288584679,
      "loss": 2.3229,
      "step": 2007
    },
    {
      "epoch": 0.31192233009708736,
      "grad_norm": 0.0,
      "learning_rate": 0.0008057512433992414,
      "loss": 2.2567,
      "step": 2008
    },
    {
      "epoch": 0.31207766990291264,
      "grad_norm": 0.0,
      "learning_rate": 0.0008055521207887884,
      "loss": 2.4073,
      "step": 2009
    },
    {
      "epoch": 0.31223300970873785,
      "grad_norm": 0.0,
      "learning_rate": 0.0008053529208037443,
      "loss": 2.2505,
      "step": 2010
    },
    {
      "epoch": 0.3123883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0008051536434945519,
      "loss": 2.3975,
      "step": 2011
    },
    {
      "epoch": 0.31254368932038834,
      "grad_norm": 0.0,
      "learning_rate": 0.0008049542889116741,
      "loss": 2.3492,
      "step": 2012
    },
    {
      "epoch": 0.3126990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.000804754857105593,
      "loss": 2.4049,
      "step": 2013
    },
    {
      "epoch": 0.31285436893203883,
      "grad_norm": 0.0,
      "learning_rate": 0.0008045553481268105,
      "loss": 2.3024,
      "step": 2014
    },
    {
      "epoch": 0.3130097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.0008043557620258481,
      "loss": 2.2908,
      "step": 2015
    },
    {
      "epoch": 0.3131650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.0008041560988532468,
      "loss": 2.4313,
      "step": 2016
    },
    {
      "epoch": 0.3133203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.0008039563586595671,
      "loss": 2.2659,
      "step": 2017
    },
    {
      "epoch": 0.3134757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0008037565414953887,
      "loss": 2.3013,
      "step": 2018
    },
    {
      "epoch": 0.313631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0008035566474113113,
      "loss": 2.3485,
      "step": 2019
    },
    {
      "epoch": 0.3137864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0008033566764579538,
      "loss": 2.3744,
      "step": 2020
    },
    {
      "epoch": 0.3139417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.0008031566286859544,
      "loss": 2.3019,
      "step": 2021
    },
    {
      "epoch": 0.3140970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.0008029565041459712,
      "loss": 2.3585,
      "step": 2022
    },
    {
      "epoch": 0.314252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0008027563028886815,
      "loss": 2.3334,
      "step": 2023
    },
    {
      "epoch": 0.31440776699029127,
      "grad_norm": 0.0,
      "learning_rate": 0.0008025560249647818,
      "loss": 2.3439,
      "step": 2024
    },
    {
      "epoch": 0.3145631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0008023556704249885,
      "loss": 2.4052,
      "step": 2025
    },
    {
      "epoch": 0.31471844660194176,
      "grad_norm": 0.0,
      "learning_rate": 0.0008021552393200369,
      "loss": 2.3555,
      "step": 2026
    },
    {
      "epoch": 0.314873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0008019547317006822,
      "loss": 2.3536,
      "step": 2027
    },
    {
      "epoch": 0.31502912621359225,
      "grad_norm": 0.0,
      "learning_rate": 0.0008017541476176981,
      "loss": 2.3115,
      "step": 2028
    },
    {
      "epoch": 0.31518446601941746,
      "grad_norm": 0.0,
      "learning_rate": 0.0008015534871218792,
      "loss": 2.3264,
      "step": 2029
    },
    {
      "epoch": 0.31533980582524274,
      "grad_norm": 0.0,
      "learning_rate": 0.0008013527502640376,
      "loss": 2.2745,
      "step": 2030
    },
    {
      "epoch": 0.31549514563106795,
      "grad_norm": 0.0,
      "learning_rate": 0.0008011519370950062,
      "loss": 2.3622,
      "step": 2031
    },
    {
      "epoch": 0.3156504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0008009510476656366,
      "loss": 2.4148,
      "step": 2032
    },
    {
      "epoch": 0.31580582524271844,
      "grad_norm": 0.0,
      "learning_rate": 0.0008007500820267997,
      "loss": 2.3247,
      "step": 2033
    },
    {
      "epoch": 0.3159611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0008005490402293856,
      "loss": 2.4346,
      "step": 2034
    },
    {
      "epoch": 0.31611650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.0008003479223243044,
      "loss": 2.3522,
      "step": 2035
    },
    {
      "epoch": 0.3162718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.0008001467283624843,
      "loss": 2.3724,
      "step": 2036
    },
    {
      "epoch": 0.3164271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.0007999454583948737,
      "loss": 2.384,
      "step": 2037
    },
    {
      "epoch": 0.31658252427184463,
      "grad_norm": 0.0,
      "learning_rate": 0.0007997441124724401,
      "loss": 2.3169,
      "step": 2038
    },
    {
      "epoch": 0.3167378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.00079954269064617,
      "loss": 2.3839,
      "step": 2039
    },
    {
      "epoch": 0.3168932038834951,
      "grad_norm": 0.0,
      "learning_rate": 0.000799341192967069,
      "loss": 2.3795,
      "step": 2040
    },
    {
      "epoch": 0.3170485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0007991396194861624,
      "loss": 2.4091,
      "step": 2041
    },
    {
      "epoch": 0.3172038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.0007989379702544942,
      "loss": 2.2352,
      "step": 2042
    },
    {
      "epoch": 0.3173592233009709,
      "grad_norm": 0.0,
      "learning_rate": 0.0007987362453231277,
      "loss": 2.3813,
      "step": 2043
    },
    {
      "epoch": 0.3175145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.0007985344447431459,
      "loss": 2.2712,
      "step": 2044
    },
    {
      "epoch": 0.31766990291262137,
      "grad_norm": 0.0,
      "learning_rate": 0.00079833256856565,
      "loss": 2.401,
      "step": 2045
    },
    {
      "epoch": 0.3178252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0007981306168417611,
      "loss": 2.4691,
      "step": 2046
    },
    {
      "epoch": 0.31798058252427186,
      "grad_norm": 0.0,
      "learning_rate": 0.0007979285896226192,
      "loss": 2.2446,
      "step": 2047
    },
    {
      "epoch": 0.3181359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0007977264869593834,
      "loss": 2.3092,
      "step": 2048
    },
    {
      "epoch": 0.31829126213592235,
      "grad_norm": 0.0,
      "learning_rate": 0.000797524308903232,
      "loss": 2.295,
      "step": 2049
    },
    {
      "epoch": 0.31844660194174756,
      "grad_norm": 0.0,
      "learning_rate": 0.000797322055505362,
      "loss": 2.3372,
      "step": 2050
    },
    {
      "epoch": 0.31860194174757284,
      "grad_norm": 0.0,
      "learning_rate": 0.00079711972681699,
      "loss": 2.3112,
      "step": 2051
    },
    {
      "epoch": 0.31875728155339805,
      "grad_norm": 0.0,
      "learning_rate": 0.0007969173228893516,
      "loss": 2.3823,
      "step": 2052
    },
    {
      "epoch": 0.3189126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0007967148437737011,
      "loss": 2.3313,
      "step": 2053
    },
    {
      "epoch": 0.31906796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.0007965122895213119,
      "loss": 2.3718,
      "step": 2054
    },
    {
      "epoch": 0.3192233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.0007963096601834767,
      "loss": 2.3999,
      "step": 2055
    },
    {
      "epoch": 0.31937864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0007961069558115073,
      "loss": 2.3138,
      "step": 2056
    },
    {
      "epoch": 0.31953398058252425,
      "grad_norm": 0.0,
      "learning_rate": 0.0007959041764567338,
      "loss": 2.3726,
      "step": 2057
    },
    {
      "epoch": 0.3196893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0007957013221705065,
      "loss": 2.3426,
      "step": 2058
    },
    {
      "epoch": 0.31984466019417473,
      "grad_norm": 0.0,
      "learning_rate": 0.0007954983930041933,
      "loss": 2.3845,
      "step": 2059
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.0,
      "learning_rate": 0.0007952953890091819,
      "loss": 2.3065,
      "step": 2060
    },
    {
      "epoch": 0.3201553398058252,
      "grad_norm": 0.0,
      "learning_rate": 0.0007950923102368786,
      "loss": 2.3945,
      "step": 2061
    },
    {
      "epoch": 0.3203106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.000794889156738709,
      "loss": 2.2639,
      "step": 2062
    },
    {
      "epoch": 0.3204660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.0007946859285661176,
      "loss": 2.4139,
      "step": 2063
    },
    {
      "epoch": 0.320621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0007944826257705673,
      "loss": 2.3875,
      "step": 2064
    },
    {
      "epoch": 0.3207766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0007942792484035402,
      "loss": 2.2439,
      "step": 2065
    },
    {
      "epoch": 0.32093203883495147,
      "grad_norm": 0.0,
      "learning_rate": 0.0007940757965165375,
      "loss": 2.4044,
      "step": 2066
    },
    {
      "epoch": 0.3210873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.000793872270161079,
      "loss": 2.298,
      "step": 2067
    },
    {
      "epoch": 0.32124271844660196,
      "grad_norm": 0.0,
      "learning_rate": 0.0007936686693887033,
      "loss": 2.3628,
      "step": 2068
    },
    {
      "epoch": 0.3213980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0007934649942509681,
      "loss": 2.3954,
      "step": 2069
    },
    {
      "epoch": 0.32155339805825245,
      "grad_norm": 0.0,
      "learning_rate": 0.00079326124479945,
      "loss": 2.3014,
      "step": 2070
    },
    {
      "epoch": 0.32170873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0007930574210857439,
      "loss": 2.3443,
      "step": 2071
    },
    {
      "epoch": 0.32186407766990294,
      "grad_norm": 0.0,
      "learning_rate": 0.000792853523161464,
      "loss": 2.3696,
      "step": 2072
    },
    {
      "epoch": 0.32201941747572815,
      "grad_norm": 0.0,
      "learning_rate": 0.0007926495510782429,
      "loss": 2.341,
      "step": 2073
    },
    {
      "epoch": 0.3221747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0007924455048877326,
      "loss": 2.3877,
      "step": 2074
    },
    {
      "epoch": 0.32233009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0007922413846416031,
      "loss": 2.3596,
      "step": 2075
    },
    {
      "epoch": 0.32248543689320386,
      "grad_norm": 0.0,
      "learning_rate": 0.0007920371903915438,
      "loss": 2.3685,
      "step": 2076
    },
    {
      "epoch": 0.32264077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.0007918329221892624,
      "loss": 2.33,
      "step": 2077
    },
    {
      "epoch": 0.32279611650485435,
      "grad_norm": 0.0,
      "learning_rate": 0.0007916285800864853,
      "loss": 2.4199,
      "step": 2078
    },
    {
      "epoch": 0.3229514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0007914241641349582,
      "loss": 2.3766,
      "step": 2079
    },
    {
      "epoch": 0.32310679611650484,
      "grad_norm": 0.0,
      "learning_rate": 0.000791219674386445,
      "loss": 2.2598,
      "step": 2080
    },
    {
      "epoch": 0.3232621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0007910151108927283,
      "loss": 2.3514,
      "step": 2081
    },
    {
      "epoch": 0.3234174757281553,
      "grad_norm": 0.0,
      "learning_rate": 0.0007908104737056094,
      "loss": 2.3724,
      "step": 2082
    },
    {
      "epoch": 0.3235728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.0007906057628769085,
      "loss": 2.3604,
      "step": 2083
    },
    {
      "epoch": 0.3237281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.000790400978458464,
      "loss": 2.2911,
      "step": 2084
    },
    {
      "epoch": 0.3238834951456311,
      "grad_norm": 0.0,
      "learning_rate": 0.0007901961205021335,
      "loss": 2.3397,
      "step": 2085
    },
    {
      "epoch": 0.3240388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.000789991189059793,
      "loss": 2.271,
      "step": 2086
    },
    {
      "epoch": 0.3241941747572816,
      "grad_norm": 0.0,
      "learning_rate": 0.0007897861841833366,
      "loss": 2.3796,
      "step": 2087
    },
    {
      "epoch": 0.3243495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0007895811059246778,
      "loss": 2.2294,
      "step": 2088
    },
    {
      "epoch": 0.32450485436893206,
      "grad_norm": 0.0,
      "learning_rate": 0.0007893759543357482,
      "loss": 2.48,
      "step": 2089
    },
    {
      "epoch": 0.3246601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.000789170729468498,
      "loss": 2.3598,
      "step": 2090
    },
    {
      "epoch": 0.32481553398058255,
      "grad_norm": 0.0,
      "learning_rate": 0.0007889654313748962,
      "loss": 2.3394,
      "step": 2091
    },
    {
      "epoch": 0.32497087378640777,
      "grad_norm": 0.0,
      "learning_rate": 0.0007887600601069302,
      "loss": 2.1744,
      "step": 2092
    },
    {
      "epoch": 0.325126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0007885546157166056,
      "loss": 2.3524,
      "step": 2093
    },
    {
      "epoch": 0.32528155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.0007883490982559474,
      "loss": 2.3236,
      "step": 2094
    },
    {
      "epoch": 0.32543689320388347,
      "grad_norm": 0.0,
      "learning_rate": 0.000788143507776998,
      "loss": 2.4034,
      "step": 2095
    },
    {
      "epoch": 0.32559223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0007879378443318191,
      "loss": 2.2787,
      "step": 2096
    },
    {
      "epoch": 0.32574757281553396,
      "grad_norm": 0.0,
      "learning_rate": 0.0007877321079724906,
      "loss": 2.3599,
      "step": 2097
    },
    {
      "epoch": 0.32590291262135923,
      "grad_norm": 0.0,
      "learning_rate": 0.000787526298751111,
      "loss": 2.3752,
      "step": 2098
    },
    {
      "epoch": 0.32605825242718445,
      "grad_norm": 0.0,
      "learning_rate": 0.0007873204167197965,
      "loss": 2.3599,
      "step": 2099
    },
    {
      "epoch": 0.3262135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0007871144619306829,
      "loss": 2.4417,
      "step": 2100
    },
    {
      "epoch": 0.32636893203883494,
      "grad_norm": 0.0,
      "learning_rate": 0.0007869084344359239,
      "loss": 2.3804,
      "step": 2101
    },
    {
      "epoch": 0.3265242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0007867023342876913,
      "loss": 2.2981,
      "step": 2102
    },
    {
      "epoch": 0.3266796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.0007864961615381761,
      "loss": 2.405,
      "step": 2103
    },
    {
      "epoch": 0.3268349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.0007862899162395865,
      "loss": 2.3969,
      "step": 2104
    },
    {
      "epoch": 0.3269902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.0007860835984441502,
      "loss": 2.3587,
      "step": 2105
    },
    {
      "epoch": 0.3271456310679612,
      "grad_norm": 0.0,
      "learning_rate": 0.0007858772082041124,
      "loss": 2.4196,
      "step": 2106
    },
    {
      "epoch": 0.3273009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0007856707455717375,
      "loss": 2.2954,
      "step": 2107
    },
    {
      "epoch": 0.3274563106796117,
      "grad_norm": 0.0,
      "learning_rate": 0.0007854642105993077,
      "loss": 2.3335,
      "step": 2108
    },
    {
      "epoch": 0.3276116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0007852576033391235,
      "loss": 2.3263,
      "step": 2109
    },
    {
      "epoch": 0.32776699029126216,
      "grad_norm": 0.0,
      "learning_rate": 0.0007850509238435038,
      "loss": 2.2302,
      "step": 2110
    },
    {
      "epoch": 0.3279223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0007848441721647856,
      "loss": 2.3622,
      "step": 2111
    },
    {
      "epoch": 0.3280776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.0007846373483553248,
      "loss": 2.4053,
      "step": 2112
    },
    {
      "epoch": 0.32823300970873787,
      "grad_norm": 0.0,
      "learning_rate": 0.0007844304524674948,
      "loss": 2.3161,
      "step": 2113
    },
    {
      "epoch": 0.3283883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0007842234845536878,
      "loss": 2.3364,
      "step": 2114
    },
    {
      "epoch": 0.32854368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0007840164446663139,
      "loss": 2.2629,
      "step": 2115
    },
    {
      "epoch": 0.32869902912621357,
      "grad_norm": 0.0,
      "learning_rate": 0.0007838093328578015,
      "loss": 2.3296,
      "step": 2116
    },
    {
      "epoch": 0.32885436893203884,
      "grad_norm": 0.0,
      "learning_rate": 0.0007836021491805976,
      "loss": 2.3124,
      "step": 2117
    },
    {
      "epoch": 0.32900970873786406,
      "grad_norm": 0.0,
      "learning_rate": 0.0007833948936871667,
      "loss": 2.3839,
      "step": 2118
    },
    {
      "epoch": 0.32916504854368933,
      "grad_norm": 0.0,
      "learning_rate": 0.0007831875664299919,
      "loss": 2.3911,
      "step": 2119
    },
    {
      "epoch": 0.32932038834951455,
      "grad_norm": 0.0,
      "learning_rate": 0.0007829801674615746,
      "loss": 2.3809,
      "step": 2120
    },
    {
      "epoch": 0.3294757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0007827726968344343,
      "loss": 2.288,
      "step": 2121
    },
    {
      "epoch": 0.32963106796116504,
      "grad_norm": 0.0,
      "learning_rate": 0.0007825651546011081,
      "loss": 2.3257,
      "step": 2122
    },
    {
      "epoch": 0.3297864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0007823575408141522,
      "loss": 2.2935,
      "step": 2123
    },
    {
      "epoch": 0.3299417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.0007821498555261399,
      "loss": 2.3545,
      "step": 2124
    },
    {
      "epoch": 0.3300970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.0007819420987896633,
      "loss": 2.2897,
      "step": 2125
    },
    {
      "epoch": 0.330252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0007817342706573324,
      "loss": 2.3763,
      "step": 2126
    },
    {
      "epoch": 0.3304077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.0007815263711817753,
      "loss": 2.3241,
      "step": 2127
    },
    {
      "epoch": 0.3305631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0007813184004156382,
      "loss": 2.2919,
      "step": 2128
    },
    {
      "epoch": 0.3307184466019418,
      "grad_norm": 0.0,
      "learning_rate": 0.0007811103584115848,
      "loss": 2.3186,
      "step": 2129
    },
    {
      "epoch": 0.330873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0007809022452222981,
      "loss": 2.2922,
      "step": 2130
    },
    {
      "epoch": 0.3310291262135922,
      "grad_norm": 0.0,
      "learning_rate": 0.0007806940609004777,
      "loss": 2.4126,
      "step": 2131
    },
    {
      "epoch": 0.3311844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.0007804858054988421,
      "loss": 2.4155,
      "step": 2132
    },
    {
      "epoch": 0.3313398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.0007802774790701277,
      "loss": 2.2811,
      "step": 2133
    },
    {
      "epoch": 0.33149514563106797,
      "grad_norm": 0.0,
      "learning_rate": 0.0007800690816670884,
      "loss": 2.3774,
      "step": 2134
    },
    {
      "epoch": 0.3316504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0007798606133424969,
      "loss": 2.2651,
      "step": 2135
    },
    {
      "epoch": 0.33180582524271846,
      "grad_norm": 0.0,
      "learning_rate": 0.0007796520741491428,
      "loss": 2.416,
      "step": 2136
    },
    {
      "epoch": 0.33196116504854367,
      "grad_norm": 0.0,
      "learning_rate": 0.0007794434641398346,
      "loss": 2.2878,
      "step": 2137
    },
    {
      "epoch": 0.33211650485436894,
      "grad_norm": 0.0,
      "learning_rate": 0.0007792347833673985,
      "loss": 2.3542,
      "step": 2138
    },
    {
      "epoch": 0.33227184466019416,
      "grad_norm": 0.0,
      "learning_rate": 0.0007790260318846782,
      "loss": 2.3973,
      "step": 2139
    },
    {
      "epoch": 0.33242718446601943,
      "grad_norm": 0.0,
      "learning_rate": 0.0007788172097445357,
      "loss": 2.3291,
      "step": 2140
    },
    {
      "epoch": 0.33258252427184465,
      "grad_norm": 0.0,
      "learning_rate": 0.0007786083169998506,
      "loss": 2.324,
      "step": 2141
    },
    {
      "epoch": 0.3327378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0007783993537035206,
      "loss": 2.3781,
      "step": 2142
    },
    {
      "epoch": 0.33289320388349514,
      "grad_norm": 0.0,
      "learning_rate": 0.0007781903199084616,
      "loss": 2.4003,
      "step": 2143
    },
    {
      "epoch": 0.3330485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0007779812156676065,
      "loss": 2.3608,
      "step": 2144
    },
    {
      "epoch": 0.3332038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.0007777720410339067,
      "loss": 2.3428,
      "step": 2145
    },
    {
      "epoch": 0.3333592233009709,
      "grad_norm": 0.0,
      "learning_rate": 0.0007775627960603312,
      "loss": 2.3882,
      "step": 2146
    },
    {
      "epoch": 0.3335145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.0007773534807998669,
      "loss": 2.3413,
      "step": 2147
    },
    {
      "epoch": 0.3336699029126214,
      "grad_norm": 0.0,
      "learning_rate": 0.0007771440953055185,
      "loss": 2.39,
      "step": 2148
    },
    {
      "epoch": 0.3338252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0007769346396303083,
      "loss": 2.3475,
      "step": 2149
    },
    {
      "epoch": 0.3339805825242718,
      "grad_norm": 0.0,
      "learning_rate": 0.0007767251138272766,
      "loss": 2.2984,
      "step": 2150
    },
    {
      "epoch": 0.3341359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0007765155179494813,
      "loss": 2.2334,
      "step": 2151
    },
    {
      "epoch": 0.3342912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.0007763058520499982,
      "loss": 2.3577,
      "step": 2152
    },
    {
      "epoch": 0.3344466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.0007760961161819206,
      "loss": 2.4243,
      "step": 2153
    },
    {
      "epoch": 0.3346019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.0007758863103983597,
      "loss": 2.4244,
      "step": 2154
    },
    {
      "epoch": 0.33475728155339807,
      "grad_norm": 0.0,
      "learning_rate": 0.0007756764347524446,
      "loss": 2.3484,
      "step": 2155
    },
    {
      "epoch": 0.3349126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0007754664892973216,
      "loss": 2.2711,
      "step": 2156
    },
    {
      "epoch": 0.33506796116504856,
      "grad_norm": 0.0,
      "learning_rate": 0.000775256474086155,
      "loss": 2.2314,
      "step": 2157
    },
    {
      "epoch": 0.33522330097087377,
      "grad_norm": 0.0,
      "learning_rate": 0.0007750463891721264,
      "loss": 2.3292,
      "step": 2158
    },
    {
      "epoch": 0.33537864077669904,
      "grad_norm": 0.0,
      "learning_rate": 0.000774836234608436,
      "loss": 2.2834,
      "step": 2159
    },
    {
      "epoch": 0.33553398058252426,
      "grad_norm": 0.0,
      "learning_rate": 0.0007746260104483005,
      "loss": 2.3608,
      "step": 2160
    },
    {
      "epoch": 0.33568932038834953,
      "grad_norm": 0.0,
      "learning_rate": 0.0007744157167449548,
      "loss": 2.3477,
      "step": 2161
    },
    {
      "epoch": 0.33584466019417475,
      "grad_norm": 0.0,
      "learning_rate": 0.0007742053535516513,
      "loss": 2.332,
      "step": 2162
    },
    {
      "epoch": 0.336,
      "grad_norm": 0.0,
      "learning_rate": 0.0007739949209216602,
      "loss": 2.3595,
      "step": 2163
    },
    {
      "epoch": 0.33615533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.0007737844189082688,
      "loss": 2.3094,
      "step": 2164
    },
    {
      "epoch": 0.3363106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0007735738475647824,
      "loss": 2.3483,
      "step": 2165
    },
    {
      "epoch": 0.3364660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.000773363206944524,
      "loss": 2.2854,
      "step": 2166
    },
    {
      "epoch": 0.336621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0007731524971008334,
      "loss": 2.302,
      "step": 2167
    },
    {
      "epoch": 0.3367766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0007729417180870686,
      "loss": 2.3228,
      "step": 2168
    },
    {
      "epoch": 0.33693203883495143,
      "grad_norm": 0.0,
      "learning_rate": 0.0007727308699566051,
      "loss": 2.3271,
      "step": 2169
    },
    {
      "epoch": 0.3370873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0007725199527628354,
      "loss": 2.3381,
      "step": 2170
    },
    {
      "epoch": 0.3372427184466019,
      "grad_norm": 0.0,
      "learning_rate": 0.00077230896655917,
      "loss": 2.3507,
      "step": 2171
    },
    {
      "epoch": 0.3373980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0007720979113990367,
      "loss": 2.1754,
      "step": 2172
    },
    {
      "epoch": 0.3375533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.0007718867873358806,
      "loss": 2.2958,
      "step": 2173
    },
    {
      "epoch": 0.3377087378640777,
      "grad_norm": 0.0,
      "learning_rate": 0.0007716755944231643,
      "loss": 2.3429,
      "step": 2174
    },
    {
      "epoch": 0.3378640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.0007714643327143682,
      "loss": 2.4536,
      "step": 2175
    },
    {
      "epoch": 0.33801941747572817,
      "grad_norm": 0.0,
      "learning_rate": 0.0007712530022629898,
      "loss": 2.301,
      "step": 2176
    },
    {
      "epoch": 0.3381747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.000771041603122544,
      "loss": 2.2992,
      "step": 2177
    },
    {
      "epoch": 0.33833009708737866,
      "grad_norm": 0.0,
      "learning_rate": 0.0007708301353465631,
      "loss": 2.3309,
      "step": 2178
    },
    {
      "epoch": 0.3384854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.0007706185989885967,
      "loss": 2.334,
      "step": 2179
    },
    {
      "epoch": 0.33864077669902914,
      "grad_norm": 0.0,
      "learning_rate": 0.0007704069941022123,
      "loss": 2.3696,
      "step": 2180
    },
    {
      "epoch": 0.33879611650485436,
      "grad_norm": 0.0,
      "learning_rate": 0.000770195320740994,
      "loss": 2.3544,
      "step": 2181
    },
    {
      "epoch": 0.33895145631067963,
      "grad_norm": 0.0,
      "learning_rate": 0.0007699835789585436,
      "loss": 2.2196,
      "step": 2182
    },
    {
      "epoch": 0.33910679611650485,
      "grad_norm": 0.0,
      "learning_rate": 0.0007697717688084804,
      "loss": 2.3165,
      "step": 2183
    },
    {
      "epoch": 0.3392621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0007695598903444405,
      "loss": 2.3199,
      "step": 2184
    },
    {
      "epoch": 0.33941747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0007693479436200781,
      "loss": 2.363,
      "step": 2185
    },
    {
      "epoch": 0.3395728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.0007691359286890639,
      "loss": 2.3175,
      "step": 2186
    },
    {
      "epoch": 0.3397281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.0007689238456050862,
      "loss": 2.3613,
      "step": 2187
    },
    {
      "epoch": 0.33988349514563104,
      "grad_norm": 0.0,
      "learning_rate": 0.0007687116944218505,
      "loss": 2.1611,
      "step": 2188
    },
    {
      "epoch": 0.3400388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0007684994751930798,
      "loss": 2.3173,
      "step": 2189
    },
    {
      "epoch": 0.34019417475728153,
      "grad_norm": 0.0,
      "learning_rate": 0.0007682871879725141,
      "loss": 2.2706,
      "step": 2190
    },
    {
      "epoch": 0.3403495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0007680748328139105,
      "loss": 2.3567,
      "step": 2191
    },
    {
      "epoch": 0.340504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0007678624097710432,
      "loss": 2.2988,
      "step": 2192
    },
    {
      "epoch": 0.3406601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.0007676499188977042,
      "loss": 2.2256,
      "step": 2193
    },
    {
      "epoch": 0.3408155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.0007674373602477022,
      "loss": 2.3806,
      "step": 2194
    },
    {
      "epoch": 0.3409708737864078,
      "grad_norm": 0.0,
      "learning_rate": 0.0007672247338748633,
      "loss": 2.3523,
      "step": 2195
    },
    {
      "epoch": 0.341126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0007670120398330304,
      "loss": 2.354,
      "step": 2196
    },
    {
      "epoch": 0.34128155339805827,
      "grad_norm": 0.0,
      "learning_rate": 0.0007667992781760636,
      "loss": 2.3175,
      "step": 2197
    },
    {
      "epoch": 0.3414368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0007665864489578407,
      "loss": 2.4243,
      "step": 2198
    },
    {
      "epoch": 0.34159223300970876,
      "grad_norm": 0.0,
      "learning_rate": 0.0007663735522322561,
      "loss": 2.2979,
      "step": 2199
    },
    {
      "epoch": 0.341747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0007661605880532212,
      "loss": 2.3201,
      "step": 2200
    },
    {
      "epoch": 0.34190291262135925,
      "grad_norm": 0.0,
      "learning_rate": 0.0007659475564746647,
      "loss": 2.2188,
      "step": 2201
    },
    {
      "epoch": 0.34205825242718446,
      "grad_norm": 0.0,
      "learning_rate": 0.0007657344575505323,
      "loss": 2.3146,
      "step": 2202
    },
    {
      "epoch": 0.34221359223300973,
      "grad_norm": 0.0,
      "learning_rate": 0.0007655212913347868,
      "loss": 2.2815,
      "step": 2203
    },
    {
      "epoch": 0.34236893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0007653080578814082,
      "loss": 2.306,
      "step": 2204
    },
    {
      "epoch": 0.3425242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0007650947572443932,
      "loss": 2.3805,
      "step": 2205
    },
    {
      "epoch": 0.34267961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.0007648813894777555,
      "loss": 2.3516,
      "step": 2206
    },
    {
      "epoch": 0.34283495145631065,
      "grad_norm": 0.0,
      "learning_rate": 0.0007646679546355262,
      "loss": 2.2793,
      "step": 2207
    },
    {
      "epoch": 0.3429902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.0007644544527717532,
      "loss": 2.2932,
      "step": 2208
    },
    {
      "epoch": 0.34314563106796114,
      "grad_norm": 0.0,
      "learning_rate": 0.0007642408839405011,
      "loss": 2.319,
      "step": 2209
    },
    {
      "epoch": 0.3433009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0007640272481958518,
      "loss": 2.312,
      "step": 2210
    },
    {
      "epoch": 0.34345631067961163,
      "grad_norm": 0.0,
      "learning_rate": 0.0007638135455919039,
      "loss": 2.3231,
      "step": 2211
    },
    {
      "epoch": 0.3436116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0007635997761827733,
      "loss": 2.4709,
      "step": 2212
    },
    {
      "epoch": 0.3437669902912621,
      "grad_norm": 0.0,
      "learning_rate": 0.0007633859400225922,
      "loss": 2.3746,
      "step": 2213
    },
    {
      "epoch": 0.3439223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0007631720371655105,
      "loss": 2.3199,
      "step": 2214
    },
    {
      "epoch": 0.3440776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.0007629580676656941,
      "loss": 2.3427,
      "step": 2215
    },
    {
      "epoch": 0.3442330097087379,
      "grad_norm": 0.0,
      "learning_rate": 0.0007627440315773266,
      "loss": 2.2797,
      "step": 2216
    },
    {
      "epoch": 0.3443883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0007625299289546079,
      "loss": 2.2356,
      "step": 2217
    },
    {
      "epoch": 0.34454368932038837,
      "grad_norm": 0.0,
      "learning_rate": 0.0007623157598517549,
      "loss": 2.2867,
      "step": 2218
    },
    {
      "epoch": 0.3446990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0007621015243230016,
      "loss": 2.3243,
      "step": 2219
    },
    {
      "epoch": 0.34485436893203886,
      "grad_norm": 0.0,
      "learning_rate": 0.0007618872224225985,
      "loss": 2.3207,
      "step": 2220
    },
    {
      "epoch": 0.3450097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.000761672854204813,
      "loss": 2.3179,
      "step": 2221
    },
    {
      "epoch": 0.34516504854368935,
      "grad_norm": 0.0,
      "learning_rate": 0.0007614584197239293,
      "loss": 2.3843,
      "step": 2222
    },
    {
      "epoch": 0.34532038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.0007612439190342486,
      "loss": 2.2693,
      "step": 2223
    },
    {
      "epoch": 0.3454757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0007610293521900882,
      "loss": 2.3432,
      "step": 2224
    },
    {
      "epoch": 0.34563106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0007608147192457828,
      "loss": 2.2756,
      "step": 2225
    },
    {
      "epoch": 0.34578640776699027,
      "grad_norm": 0.0,
      "learning_rate": 0.0007606000202556838,
      "loss": 2.3882,
      "step": 2226
    },
    {
      "epoch": 0.34594174757281554,
      "grad_norm": 0.0,
      "learning_rate": 0.000760385255274159,
      "loss": 2.347,
      "step": 2227
    },
    {
      "epoch": 0.34609708737864076,
      "grad_norm": 0.0,
      "learning_rate": 0.0007601704243555931,
      "loss": 2.3102,
      "step": 2228
    },
    {
      "epoch": 0.346252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0007599555275543875,
      "loss": 2.3519,
      "step": 2229
    },
    {
      "epoch": 0.34640776699029124,
      "grad_norm": 0.0,
      "learning_rate": 0.0007597405649249603,
      "loss": 2.3458,
      "step": 2230
    },
    {
      "epoch": 0.3465631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.000759525536521746,
      "loss": 2.3751,
      "step": 2231
    },
    {
      "epoch": 0.34671844660194173,
      "grad_norm": 0.0,
      "learning_rate": 0.0007593104423991963,
      "loss": 2.3477,
      "step": 2232
    },
    {
      "epoch": 0.346873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0007590952826117789,
      "loss": 2.3135,
      "step": 2233
    },
    {
      "epoch": 0.3470291262135922,
      "grad_norm": 0.0,
      "learning_rate": 0.0007588800572139789,
      "loss": 2.3109,
      "step": 2234
    },
    {
      "epoch": 0.3471844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.0007586647662602971,
      "loss": 2.2161,
      "step": 2235
    },
    {
      "epoch": 0.3473398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.0007584494098052517,
      "loss": 2.3774,
      "step": 2236
    },
    {
      "epoch": 0.347495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0007582339879033768,
      "loss": 2.4022,
      "step": 2237
    },
    {
      "epoch": 0.3476504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0007580185006092237,
      "loss": 2.2426,
      "step": 2238
    },
    {
      "epoch": 0.34780582524271847,
      "grad_norm": 0.0,
      "learning_rate": 0.00075780294797736,
      "loss": 2.3763,
      "step": 2239
    },
    {
      "epoch": 0.3479611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0007575873300623697,
      "loss": 2.3079,
      "step": 2240
    },
    {
      "epoch": 0.34811650485436896,
      "grad_norm": 0.0,
      "learning_rate": 0.0007573716469188534,
      "loss": 2.3967,
      "step": 2241
    },
    {
      "epoch": 0.3482718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.0007571558986014286,
      "loss": 2.3235,
      "step": 2242
    },
    {
      "epoch": 0.3484271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.0007569400851647286,
      "loss": 2.3141,
      "step": 2243
    },
    {
      "epoch": 0.34858252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0007567242066634039,
      "loss": 2.3621,
      "step": 2244
    },
    {
      "epoch": 0.3487378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0007565082631521208,
      "loss": 2.4076,
      "step": 2245
    },
    {
      "epoch": 0.34889320388349515,
      "grad_norm": 0.0,
      "learning_rate": 0.0007562922546855628,
      "loss": 2.2544,
      "step": 2246
    },
    {
      "epoch": 0.34904854368932037,
      "grad_norm": 0.0,
      "learning_rate": 0.0007560761813184292,
      "loss": 2.3,
      "step": 2247
    },
    {
      "epoch": 0.34920388349514564,
      "grad_norm": 0.0,
      "learning_rate": 0.000755860043105436,
      "loss": 2.34,
      "step": 2248
    },
    {
      "epoch": 0.34935922330097086,
      "grad_norm": 0.0,
      "learning_rate": 0.0007556438401013157,
      "loss": 2.181,
      "step": 2249
    },
    {
      "epoch": 0.34951456310679613,
      "grad_norm": 0.0,
      "learning_rate": 0.0007554275723608171,
      "loss": 2.1427,
      "step": 2250
    },
    {
      "epoch": 0.34966990291262134,
      "grad_norm": 0.0,
      "learning_rate": 0.0007552112399387054,
      "loss": 2.2194,
      "step": 2251
    },
    {
      "epoch": 0.3498252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0007549948428897624,
      "loss": 2.2566,
      "step": 2252
    },
    {
      "epoch": 0.34998058252427183,
      "grad_norm": 0.0,
      "learning_rate": 0.0007547783812687858,
      "loss": 2.2407,
      "step": 2253
    },
    {
      "epoch": 0.3501359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0007545618551305898,
      "loss": 2.2561,
      "step": 2254
    },
    {
      "epoch": 0.3502912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.0007543452645300053,
      "loss": 2.2579,
      "step": 2255
    },
    {
      "epoch": 0.3504466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.0007541286095218793,
      "loss": 2.2257,
      "step": 2256
    },
    {
      "epoch": 0.3506019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.0007539118901610751,
      "loss": 2.3971,
      "step": 2257
    },
    {
      "epoch": 0.3507572815533981,
      "grad_norm": 0.0,
      "learning_rate": 0.0007536951065024722,
      "loss": 2.3306,
      "step": 2258
    },
    {
      "epoch": 0.3509126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0007534782586009661,
      "loss": 2.3027,
      "step": 2259
    },
    {
      "epoch": 0.35106796116504857,
      "grad_norm": 0.0,
      "learning_rate": 0.0007532613465114694,
      "loss": 2.357,
      "step": 2260
    },
    {
      "epoch": 0.3512233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.0007530443702889106,
      "loss": 2.3114,
      "step": 2261
    },
    {
      "epoch": 0.351378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.000752827329988234,
      "loss": 2.3324,
      "step": 2262
    },
    {
      "epoch": 0.3515339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0007526102256644003,
      "loss": 2.3105,
      "step": 2263
    },
    {
      "epoch": 0.3516893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.000752393057372387,
      "loss": 2.3306,
      "step": 2264
    },
    {
      "epoch": 0.35184466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.0007521758251671871,
      "loss": 2.3279,
      "step": 2265
    },
    {
      "epoch": 0.352,
      "grad_norm": 0.0,
      "learning_rate": 0.0007519585291038101,
      "loss": 2.3463,
      "step": 2266
    },
    {
      "epoch": 0.35215533980582525,
      "grad_norm": 0.0,
      "learning_rate": 0.0007517411692372817,
      "loss": 2.2054,
      "step": 2267
    },
    {
      "epoch": 0.35231067961165047,
      "grad_norm": 0.0,
      "learning_rate": 0.0007515237456226435,
      "loss": 2.2726,
      "step": 2268
    },
    {
      "epoch": 0.35246601941747574,
      "grad_norm": 0.0,
      "learning_rate": 0.0007513062583149535,
      "loss": 2.368,
      "step": 2269
    },
    {
      "epoch": 0.35262135922330096,
      "grad_norm": 0.0,
      "learning_rate": 0.0007510887073692858,
      "loss": 2.304,
      "step": 2270
    },
    {
      "epoch": 0.35277669902912623,
      "grad_norm": 0.0,
      "learning_rate": 0.0007508710928407305,
      "loss": 2.4012,
      "step": 2271
    },
    {
      "epoch": 0.35293203883495144,
      "grad_norm": 0.0,
      "learning_rate": 0.000750653414784394,
      "loss": 2.4102,
      "step": 2272
    },
    {
      "epoch": 0.3530873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0007504356732553983,
      "loss": 2.3323,
      "step": 2273
    },
    {
      "epoch": 0.35324271844660193,
      "grad_norm": 0.0,
      "learning_rate": 0.0007502178683088821,
      "loss": 2.3247,
      "step": 2274
    },
    {
      "epoch": 0.3533980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.00075,
      "loss": 2.2881,
      "step": 2275
    },
    {
      "epoch": 0.3535533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.0007497820683839224,
      "loss": 2.2683,
      "step": 2276
    },
    {
      "epoch": 0.3537087378640777,
      "grad_norm": 0.0,
      "learning_rate": 0.0007495640735158357,
      "loss": 2.3127,
      "step": 2277
    },
    {
      "epoch": 0.3538640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.0007493460154509424,
      "loss": 2.2355,
      "step": 2278
    },
    {
      "epoch": 0.3540194174757282,
      "grad_norm": 0.0,
      "learning_rate": 0.0007491278942444615,
      "loss": 2.3091,
      "step": 2279
    },
    {
      "epoch": 0.3541747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0007489097099516273,
      "loss": 2.3848,
      "step": 2280
    },
    {
      "epoch": 0.3543300970873786,
      "grad_norm": 0.0,
      "learning_rate": 0.0007486914626276904,
      "loss": 2.1963,
      "step": 2281
    },
    {
      "epoch": 0.3544854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.0007484731523279174,
      "loss": 2.3135,
      "step": 2282
    },
    {
      "epoch": 0.3546407766990291,
      "grad_norm": 0.0,
      "learning_rate": 0.0007482547791075904,
      "loss": 2.2827,
      "step": 2283
    },
    {
      "epoch": 0.3547961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.000748036343022008,
      "loss": 2.2883,
      "step": 2284
    },
    {
      "epoch": 0.3549514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0007478178441264847,
      "loss": 2.39,
      "step": 2285
    },
    {
      "epoch": 0.35510679611650486,
      "grad_norm": 0.0,
      "learning_rate": 0.0007475992824763505,
      "loss": 2.3344,
      "step": 2286
    },
    {
      "epoch": 0.3552621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0007473806581269516,
      "loss": 2.3141,
      "step": 2287
    },
    {
      "epoch": 0.35541747572815535,
      "grad_norm": 0.0,
      "learning_rate": 0.0007471619711336497,
      "loss": 2.3905,
      "step": 2288
    },
    {
      "epoch": 0.35557281553398057,
      "grad_norm": 0.0,
      "learning_rate": 0.000746943221551823,
      "loss": 2.3209,
      "step": 2289
    },
    {
      "epoch": 0.35572815533980584,
      "grad_norm": 0.0,
      "learning_rate": 0.000746724409436865,
      "loss": 2.2554,
      "step": 2290
    },
    {
      "epoch": 0.35588349514563106,
      "grad_norm": 0.0,
      "learning_rate": 0.0007465055348441854,
      "loss": 2.339,
      "step": 2291
    },
    {
      "epoch": 0.35603883495145633,
      "grad_norm": 0.0,
      "learning_rate": 0.0007462865978292093,
      "loss": 2.3464,
      "step": 2292
    },
    {
      "epoch": 0.35619417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.0007460675984473779,
      "loss": 2.2476,
      "step": 2293
    },
    {
      "epoch": 0.3563495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0007458485367541483,
      "loss": 2.3485,
      "step": 2294
    },
    {
      "epoch": 0.35650485436893203,
      "grad_norm": 0.0,
      "learning_rate": 0.0007456294128049931,
      "loss": 2.4174,
      "step": 2295
    },
    {
      "epoch": 0.3566601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.0007454102266554009,
      "loss": 2.2717,
      "step": 2296
    },
    {
      "epoch": 0.3568155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.0007451909783608759,
      "loss": 2.2972,
      "step": 2297
    },
    {
      "epoch": 0.3569708737864078,
      "grad_norm": 0.0,
      "learning_rate": 0.0007449716679769381,
      "loss": 2.35,
      "step": 2298
    },
    {
      "epoch": 0.357126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.000744752295559123,
      "loss": 2.3629,
      "step": 2299
    },
    {
      "epoch": 0.3572815533980582,
      "grad_norm": 0.0,
      "learning_rate": 0.0007445328611629823,
      "loss": 2.3539,
      "step": 2300
    },
    {
      "epoch": 0.3574368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0007443133648440827,
      "loss": 2.3842,
      "step": 2301
    },
    {
      "epoch": 0.3575922330097087,
      "grad_norm": 0.0,
      "learning_rate": 0.0007440938066580075,
      "loss": 2.437,
      "step": 2302
    },
    {
      "epoch": 0.357747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0007438741866603549,
      "loss": 2.3127,
      "step": 2303
    },
    {
      "epoch": 0.3579029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.0007436545049067389,
      "loss": 2.2552,
      "step": 2304
    },
    {
      "epoch": 0.3580582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.0007434347614527896,
      "loss": 2.2382,
      "step": 2305
    },
    {
      "epoch": 0.3582135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0007432149563541518,
      "loss": 2.3148,
      "step": 2306
    },
    {
      "epoch": 0.35836893203883496,
      "grad_norm": 0.0,
      "learning_rate": 0.0007429950896664869,
      "loss": 2.3361,
      "step": 2307
    },
    {
      "epoch": 0.3585242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0007427751614454715,
      "loss": 2.4109,
      "step": 2308
    },
    {
      "epoch": 0.35867961165048545,
      "grad_norm": 0.0,
      "learning_rate": 0.0007425551717467976,
      "loss": 2.2871,
      "step": 2309
    },
    {
      "epoch": 0.35883495145631067,
      "grad_norm": 0.0,
      "learning_rate": 0.0007423351206261728,
      "loss": 2.2733,
      "step": 2310
    },
    {
      "epoch": 0.35899029126213594,
      "grad_norm": 0.0,
      "learning_rate": 0.0007421150081393206,
      "loss": 2.2673,
      "step": 2311
    },
    {
      "epoch": 0.35914563106796116,
      "grad_norm": 0.0,
      "learning_rate": 0.0007418948343419798,
      "loss": 2.2764,
      "step": 2312
    },
    {
      "epoch": 0.35930097087378643,
      "grad_norm": 0.0,
      "learning_rate": 0.0007416745992899047,
      "loss": 2.2178,
      "step": 2313
    },
    {
      "epoch": 0.35945631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0007414543030388652,
      "loss": 2.2793,
      "step": 2314
    },
    {
      "epoch": 0.3596116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0007412339456446465,
      "loss": 2.3466,
      "step": 2315
    },
    {
      "epoch": 0.35976699029126213,
      "grad_norm": 0.0,
      "learning_rate": 0.0007410135271630493,
      "loss": 2.4491,
      "step": 2316
    },
    {
      "epoch": 0.3599223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0007407930476498906,
      "loss": 2.2675,
      "step": 2317
    },
    {
      "epoch": 0.3600776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.0007405725071610013,
      "loss": 2.2871,
      "step": 2318
    },
    {
      "epoch": 0.36023300970873784,
      "grad_norm": 0.0,
      "learning_rate": 0.0007403519057522292,
      "loss": 2.2522,
      "step": 2319
    },
    {
      "epoch": 0.3603883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0007401312434794367,
      "loss": 2.2877,
      "step": 2320
    },
    {
      "epoch": 0.3605436893203883,
      "grad_norm": 0.0,
      "learning_rate": 0.0007399105203985018,
      "loss": 2.3079,
      "step": 2321
    },
    {
      "epoch": 0.3606990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.000739689736565318,
      "loss": 2.2819,
      "step": 2322
    },
    {
      "epoch": 0.3608543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.0007394688920357942,
      "loss": 2.3352,
      "step": 2323
    },
    {
      "epoch": 0.3610097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.0007392479868658544,
      "loss": 2.2461,
      "step": 2324
    },
    {
      "epoch": 0.3611650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.0007390270211114383,
      "loss": 2.2765,
      "step": 2325
    },
    {
      "epoch": 0.3613203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.0007388059948285006,
      "loss": 2.3315,
      "step": 2326
    },
    {
      "epoch": 0.3614757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0007385849080730118,
      "loss": 2.2389,
      "step": 2327
    },
    {
      "epoch": 0.36163106796116506,
      "grad_norm": 0.0,
      "learning_rate": 0.0007383637609009574,
      "loss": 2.2679,
      "step": 2328
    },
    {
      "epoch": 0.3617864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0007381425533683379,
      "loss": 2.3435,
      "step": 2329
    },
    {
      "epoch": 0.36194174757281555,
      "grad_norm": 0.0,
      "learning_rate": 0.0007379212855311697,
      "loss": 2.2646,
      "step": 2330
    },
    {
      "epoch": 0.36209708737864077,
      "grad_norm": 0.0,
      "learning_rate": 0.0007376999574454841,
      "loss": 2.2723,
      "step": 2331
    },
    {
      "epoch": 0.36225242718446604,
      "grad_norm": 0.0,
      "learning_rate": 0.0007374785691673279,
      "loss": 2.2619,
      "step": 2332
    },
    {
      "epoch": 0.36240776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.0007372571207527628,
      "loss": 2.3014,
      "step": 2333
    },
    {
      "epoch": 0.36256310679611653,
      "grad_norm": 0.0,
      "learning_rate": 0.0007370356122578659,
      "loss": 2.2776,
      "step": 2334
    },
    {
      "epoch": 0.36271844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.0007368140437387294,
      "loss": 2.3511,
      "step": 2335
    },
    {
      "epoch": 0.36287378640776696,
      "grad_norm": 0.0,
      "learning_rate": 0.0007365924152514611,
      "loss": 2.3651,
      "step": 2336
    },
    {
      "epoch": 0.36302912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.0007363707268521839,
      "loss": 2.2802,
      "step": 2337
    },
    {
      "epoch": 0.36318446601941745,
      "grad_norm": 0.0,
      "learning_rate": 0.0007361489785970353,
      "loss": 2.2555,
      "step": 2338
    },
    {
      "epoch": 0.3633398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.0007359271705421681,
      "loss": 2.3247,
      "step": 2339
    },
    {
      "epoch": 0.36349514563106794,
      "grad_norm": 0.0,
      "learning_rate": 0.000735705302743751,
      "loss": 2.4108,
      "step": 2340
    },
    {
      "epoch": 0.3636504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0007354833752579673,
      "loss": 2.2489,
      "step": 2341
    },
    {
      "epoch": 0.36380582524271843,
      "grad_norm": 0.0,
      "learning_rate": 0.0007352613881410151,
      "loss": 2.3826,
      "step": 2342
    },
    {
      "epoch": 0.3639611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0007350393414491083,
      "loss": 2.3071,
      "step": 2343
    },
    {
      "epoch": 0.3641165048543689,
      "grad_norm": 0.0,
      "learning_rate": 0.000734817235238475,
      "loss": 2.3203,
      "step": 2344
    },
    {
      "epoch": 0.3642718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.0007345950695653593,
      "loss": 2.2429,
      "step": 2345
    },
    {
      "epoch": 0.3644271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.00073437284448602,
      "loss": 2.2214,
      "step": 2346
    },
    {
      "epoch": 0.3645825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.0007341505600567306,
      "loss": 2.3358,
      "step": 2347
    },
    {
      "epoch": 0.3647378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.00073392821633378,
      "loss": 2.3887,
      "step": 2348
    },
    {
      "epoch": 0.36489320388349517,
      "grad_norm": 0.0,
      "learning_rate": 0.0007337058133734722,
      "loss": 2.3944,
      "step": 2349
    },
    {
      "epoch": 0.3650485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0007334833512321257,
      "loss": 2.3618,
      "step": 2350
    },
    {
      "epoch": 0.36520388349514565,
      "grad_norm": 0.0,
      "learning_rate": 0.0007332608299660747,
      "loss": 2.295,
      "step": 2351
    },
    {
      "epoch": 0.36535922330097087,
      "grad_norm": 0.0,
      "learning_rate": 0.0007330382496316676,
      "loss": 2.2476,
      "step": 2352
    },
    {
      "epoch": 0.36551456310679614,
      "grad_norm": 0.0,
      "learning_rate": 0.0007328156102852686,
      "loss": 2.3184,
      "step": 2353
    },
    {
      "epoch": 0.36566990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.000732592911983256,
      "loss": 2.3402,
      "step": 2354
    },
    {
      "epoch": 0.3658252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0007323701547820236,
      "loss": 2.302,
      "step": 2355
    },
    {
      "epoch": 0.36598058252427185,
      "grad_norm": 0.0,
      "learning_rate": 0.0007321473387379798,
      "loss": 2.2666,
      "step": 2356
    },
    {
      "epoch": 0.36613592233009706,
      "grad_norm": 0.0,
      "learning_rate": 0.0007319244639075482,
      "loss": 2.2291,
      "step": 2357
    },
    {
      "epoch": 0.36629126213592234,
      "grad_norm": 0.0,
      "learning_rate": 0.000731701530347167,
      "loss": 2.3567,
      "step": 2358
    },
    {
      "epoch": 0.36644660194174755,
      "grad_norm": 0.0,
      "learning_rate": 0.0007314785381132894,
      "loss": 2.3933,
      "step": 2359
    },
    {
      "epoch": 0.3666019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.0007312554872623838,
      "loss": 2.3066,
      "step": 2360
    },
    {
      "epoch": 0.36675728155339804,
      "grad_norm": 0.0,
      "learning_rate": 0.0007310323778509325,
      "loss": 2.3333,
      "step": 2361
    },
    {
      "epoch": 0.3669126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0007308092099354335,
      "loss": 2.2946,
      "step": 2362
    },
    {
      "epoch": 0.36706796116504853,
      "grad_norm": 0.0,
      "learning_rate": 0.0007305859835723995,
      "loss": 2.3091,
      "step": 2363
    },
    {
      "epoch": 0.3672233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.0007303626988183576,
      "loss": 2.3707,
      "step": 2364
    },
    {
      "epoch": 0.367378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0007301393557298502,
      "loss": 2.3409,
      "step": 2365
    },
    {
      "epoch": 0.3675339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0007299159543634339,
      "loss": 2.3749,
      "step": 2366
    },
    {
      "epoch": 0.3676893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0007296924947756805,
      "loss": 2.3337,
      "step": 2367
    },
    {
      "epoch": 0.3678446601941748,
      "grad_norm": 0.0,
      "learning_rate": 0.0007294689770231763,
      "loss": 2.2567,
      "step": 2368
    },
    {
      "epoch": 0.368,
      "grad_norm": 0.0,
      "learning_rate": 0.0007292454011625225,
      "loss": 2.3578,
      "step": 2369
    },
    {
      "epoch": 0.36815533980582527,
      "grad_norm": 0.0,
      "learning_rate": 0.0007290217672503349,
      "loss": 2.3082,
      "step": 2370
    },
    {
      "epoch": 0.3683106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0007287980753432442,
      "loss": 2.311,
      "step": 2371
    },
    {
      "epoch": 0.36846601941747575,
      "grad_norm": 0.0,
      "learning_rate": 0.0007285743254978956,
      "loss": 2.3746,
      "step": 2372
    },
    {
      "epoch": 0.36862135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0007283505177709488,
      "loss": 2.2368,
      "step": 2373
    },
    {
      "epoch": 0.3687766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0007281266522190786,
      "loss": 2.2207,
      "step": 2374
    },
    {
      "epoch": 0.36893203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.000727902728898974,
      "loss": 2.3931,
      "step": 2375
    },
    {
      "epoch": 0.3690873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0007276787478673389,
      "loss": 2.3258,
      "step": 2376
    },
    {
      "epoch": 0.36924271844660195,
      "grad_norm": 0.0,
      "learning_rate": 0.0007274547091808918,
      "loss": 2.2494,
      "step": 2377
    },
    {
      "epoch": 0.36939805825242716,
      "grad_norm": 0.0,
      "learning_rate": 0.0007272306128963658,
      "loss": 2.3788,
      "step": 2378
    },
    {
      "epoch": 0.36955339805825244,
      "grad_norm": 0.0,
      "learning_rate": 0.0007270064590705082,
      "loss": 2.166,
      "step": 2379
    },
    {
      "epoch": 0.36970873786407765,
      "grad_norm": 0.0,
      "learning_rate": 0.0007267822477600815,
      "loss": 2.2354,
      "step": 2380
    },
    {
      "epoch": 0.3698640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.0007265579790218623,
      "loss": 2.3273,
      "step": 2381
    },
    {
      "epoch": 0.37001941747572814,
      "grad_norm": 0.0,
      "learning_rate": 0.0007263336529126422,
      "loss": 2.3207,
      "step": 2382
    },
    {
      "epoch": 0.3701747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0007261092694892266,
      "loss": 2.241,
      "step": 2383
    },
    {
      "epoch": 0.37033009708737863,
      "grad_norm": 0.0,
      "learning_rate": 0.0007258848288084362,
      "loss": 2.2699,
      "step": 2384
    },
    {
      "epoch": 0.3704854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.0007256603309271057,
      "loss": 2.2959,
      "step": 2385
    },
    {
      "epoch": 0.3706407766990291,
      "grad_norm": 0.0,
      "learning_rate": 0.0007254357759020843,
      "loss": 2.3346,
      "step": 2386
    },
    {
      "epoch": 0.3707961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.0007252111637902358,
      "loss": 2.2614,
      "step": 2387
    },
    {
      "epoch": 0.3709514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0007249864946484388,
      "loss": 2.4204,
      "step": 2388
    },
    {
      "epoch": 0.3711067961165049,
      "grad_norm": 0.0,
      "learning_rate": 0.0007247617685335859,
      "loss": 2.3398,
      "step": 2389
    },
    {
      "epoch": 0.3712621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0007245369855025839,
      "loss": 2.2917,
      "step": 2390
    },
    {
      "epoch": 0.37141747572815537,
      "grad_norm": 0.0,
      "learning_rate": 0.0007243121456123545,
      "loss": 2.4304,
      "step": 2391
    },
    {
      "epoch": 0.3715728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.0007240872489198334,
      "loss": 2.3667,
      "step": 2392
    },
    {
      "epoch": 0.3717281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.0007238622954819715,
      "loss": 2.4011,
      "step": 2393
    },
    {
      "epoch": 0.37188349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.0007236372853557331,
      "loss": 2.3322,
      "step": 2394
    },
    {
      "epoch": 0.3720388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0007234122185980973,
      "loss": 2.4011,
      "step": 2395
    },
    {
      "epoch": 0.37219417475728156,
      "grad_norm": 0.0,
      "learning_rate": 0.0007231870952660572,
      "loss": 2.2494,
      "step": 2396
    },
    {
      "epoch": 0.3723495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0007229619154166209,
      "loss": 2.3414,
      "step": 2397
    },
    {
      "epoch": 0.37250485436893205,
      "grad_norm": 0.0,
      "learning_rate": 0.0007227366791068104,
      "loss": 2.3063,
      "step": 2398
    },
    {
      "epoch": 0.37266019417475726,
      "grad_norm": 0.0,
      "learning_rate": 0.0007225113863936619,
      "loss": 2.3561,
      "step": 2399
    },
    {
      "epoch": 0.37281553398058254,
      "grad_norm": 0.0,
      "learning_rate": 0.000722286037334226,
      "loss": 2.2861,
      "step": 2400
    },
    {
      "epoch": 0.37297087378640775,
      "grad_norm": 0.0,
      "learning_rate": 0.0007220606319855675,
      "loss": 2.353,
      "step": 2401
    },
    {
      "epoch": 0.373126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0007218351704047657,
      "loss": 2.2278,
      "step": 2402
    },
    {
      "epoch": 0.37328155339805824,
      "grad_norm": 0.0,
      "learning_rate": 0.0007216096526489138,
      "loss": 2.2462,
      "step": 2403
    },
    {
      "epoch": 0.3734368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0007213840787751194,
      "loss": 2.1744,
      "step": 2404
    },
    {
      "epoch": 0.37359223300970873,
      "grad_norm": 0.0,
      "learning_rate": 0.0007211584488405044,
      "loss": 2.3207,
      "step": 2405
    },
    {
      "epoch": 0.373747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0007209327629022047,
      "loss": 2.3881,
      "step": 2406
    },
    {
      "epoch": 0.3739029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.0007207070210173705,
      "loss": 2.3856,
      "step": 2407
    },
    {
      "epoch": 0.3740582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.0007204812232431663,
      "loss": 2.3408,
      "step": 2408
    },
    {
      "epoch": 0.3742135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0007202553696367701,
      "loss": 2.2961,
      "step": 2409
    },
    {
      "epoch": 0.374368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0007200294602553753,
      "loss": 2.2223,
      "step": 2410
    },
    {
      "epoch": 0.3745242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.000719803495156188,
      "loss": 2.2638,
      "step": 2411
    },
    {
      "epoch": 0.3746796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.0007195774743964294,
      "loss": 2.2174,
      "step": 2412
    },
    {
      "epoch": 0.3748349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.0007193513980333344,
      "loss": 2.2761,
      "step": 2413
    },
    {
      "epoch": 0.3749902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.0007191252661241518,
      "loss": 2.254,
      "step": 2414
    },
    {
      "epoch": 0.37514563106796117,
      "grad_norm": 0.0,
      "learning_rate": 0.0007188990787261451,
      "loss": 2.2899,
      "step": 2415
    },
    {
      "epoch": 0.3753009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0007186728358965913,
      "loss": 2.2252,
      "step": 2416
    },
    {
      "epoch": 0.37545631067961166,
      "grad_norm": 0.0,
      "learning_rate": 0.0007184465376927815,
      "loss": 2.394,
      "step": 2417
    },
    {
      "epoch": 0.3756116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.000718220184172021,
      "loss": 2.3268,
      "step": 2418
    },
    {
      "epoch": 0.37576699029126215,
      "grad_norm": 0.0,
      "learning_rate": 0.000717993775391629,
      "loss": 2.355,
      "step": 2419
    },
    {
      "epoch": 0.37592233009708736,
      "grad_norm": 0.0,
      "learning_rate": 0.0007177673114089387,
      "loss": 2.2185,
      "step": 2420
    },
    {
      "epoch": 0.37607766990291264,
      "grad_norm": 0.0,
      "learning_rate": 0.0007175407922812974,
      "loss": 2.3739,
      "step": 2421
    },
    {
      "epoch": 0.37623300970873785,
      "grad_norm": 0.0,
      "learning_rate": 0.0007173142180660663,
      "loss": 2.2453,
      "step": 2422
    },
    {
      "epoch": 0.3763883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0007170875888206204,
      "loss": 2.2312,
      "step": 2423
    },
    {
      "epoch": 0.37654368932038834,
      "grad_norm": 0.0,
      "learning_rate": 0.0007168609046023487,
      "loss": 2.3168,
      "step": 2424
    },
    {
      "epoch": 0.3766990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0007166341654686541,
      "loss": 2.2577,
      "step": 2425
    },
    {
      "epoch": 0.37685436893203883,
      "grad_norm": 0.0,
      "learning_rate": 0.0007164073714769536,
      "loss": 2.3219,
      "step": 2426
    },
    {
      "epoch": 0.3770097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.0007161805226846783,
      "loss": 2.2914,
      "step": 2427
    },
    {
      "epoch": 0.3771650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.0007159536191492723,
      "loss": 2.364,
      "step": 2428
    },
    {
      "epoch": 0.3773203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.0007157266609281943,
      "loss": 2.4176,
      "step": 2429
    },
    {
      "epoch": 0.3774757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0007154996480789167,
      "loss": 2.2882,
      "step": 2430
    },
    {
      "epoch": 0.377631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0007152725806589257,
      "loss": 2.3145,
      "step": 2431
    },
    {
      "epoch": 0.3777864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0007150454587257211,
      "loss": 2.2719,
      "step": 2432
    },
    {
      "epoch": 0.3779417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.0007148182823368171,
      "loss": 2.2651,
      "step": 2433
    },
    {
      "epoch": 0.3780970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.0007145910515497409,
      "loss": 2.2943,
      "step": 2434
    },
    {
      "epoch": 0.378252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0007143637664220341,
      "loss": 2.3925,
      "step": 2435
    },
    {
      "epoch": 0.37840776699029127,
      "grad_norm": 0.0,
      "learning_rate": 0.000714136427011252,
      "loss": 2.3474,
      "step": 2436
    },
    {
      "epoch": 0.3785631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.000713909033374963,
      "loss": 2.3246,
      "step": 2437
    },
    {
      "epoch": 0.37871844660194176,
      "grad_norm": 0.0,
      "learning_rate": 0.0007136815855707502,
      "loss": 2.3093,
      "step": 2438
    },
    {
      "epoch": 0.378873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0007134540836562094,
      "loss": 2.3518,
      "step": 2439
    },
    {
      "epoch": 0.37902912621359225,
      "grad_norm": 0.0,
      "learning_rate": 0.0007132265276889513,
      "loss": 2.3289,
      "step": 2440
    },
    {
      "epoch": 0.37918446601941747,
      "grad_norm": 0.0,
      "learning_rate": 0.0007129989177265992,
      "loss": 2.3439,
      "step": 2441
    },
    {
      "epoch": 0.37933980582524274,
      "grad_norm": 0.0,
      "learning_rate": 0.0007127712538267904,
      "loss": 2.3258,
      "step": 2442
    },
    {
      "epoch": 0.37949514563106795,
      "grad_norm": 0.0,
      "learning_rate": 0.0007125435360471762,
      "loss": 2.3884,
      "step": 2443
    },
    {
      "epoch": 0.3796504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0007123157644454212,
      "loss": 2.2726,
      "step": 2444
    },
    {
      "epoch": 0.37980582524271844,
      "grad_norm": 0.0,
      "learning_rate": 0.0007120879390792036,
      "loss": 2.2168,
      "step": 2445
    },
    {
      "epoch": 0.3799611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0007118600600062158,
      "loss": 2.3661,
      "step": 2446
    },
    {
      "epoch": 0.38011650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.0007116321272841628,
      "loss": 2.2969,
      "step": 2447
    },
    {
      "epoch": 0.38027184466019415,
      "grad_norm": 0.0,
      "learning_rate": 0.0007114041409707641,
      "loss": 2.258,
      "step": 2448
    },
    {
      "epoch": 0.3804271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.000711176101123752,
      "loss": 2.2275,
      "step": 2449
    },
    {
      "epoch": 0.38058252427184464,
      "grad_norm": 0.0,
      "learning_rate": 0.0007109480078008731,
      "loss": 2.2415,
      "step": 2450
    },
    {
      "epoch": 0.3807378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0007107198610598872,
      "loss": 2.3628,
      "step": 2451
    },
    {
      "epoch": 0.3808932038834951,
      "grad_norm": 0.0,
      "learning_rate": 0.0007104916609585674,
      "loss": 2.2902,
      "step": 2452
    },
    {
      "epoch": 0.3810485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0007102634075547005,
      "loss": 2.2948,
      "step": 2453
    },
    {
      "epoch": 0.3812038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.0007100351009060872,
      "loss": 2.2887,
      "step": 2454
    },
    {
      "epoch": 0.3813592233009709,
      "grad_norm": 0.0,
      "learning_rate": 0.000709806741070541,
      "loss": 2.385,
      "step": 2455
    },
    {
      "epoch": 0.3815145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.0007095783281058893,
      "loss": 2.3208,
      "step": 2456
    },
    {
      "epoch": 0.3816699029126214,
      "grad_norm": 0.0,
      "learning_rate": 0.0007093498620699729,
      "loss": 2.2577,
      "step": 2457
    },
    {
      "epoch": 0.3818252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0007091213430206457,
      "loss": 2.3158,
      "step": 2458
    },
    {
      "epoch": 0.38198058252427186,
      "grad_norm": 0.0,
      "learning_rate": 0.0007088927710157756,
      "loss": 2.4097,
      "step": 2459
    },
    {
      "epoch": 0.3821359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0007086641461132434,
      "loss": 2.2827,
      "step": 2460
    },
    {
      "epoch": 0.38229126213592235,
      "grad_norm": 0.0,
      "learning_rate": 0.0007084354683709437,
      "loss": 2.2321,
      "step": 2461
    },
    {
      "epoch": 0.38244660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.0007082067378467841,
      "loss": 2.2177,
      "step": 2462
    },
    {
      "epoch": 0.38260194174757284,
      "grad_norm": 0.0,
      "learning_rate": 0.0007079779545986858,
      "loss": 2.2927,
      "step": 2463
    },
    {
      "epoch": 0.38275728155339805,
      "grad_norm": 0.0,
      "learning_rate": 0.0007077491186845835,
      "loss": 2.3677,
      "step": 2464
    },
    {
      "epoch": 0.3829126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0007075202301624247,
      "loss": 2.1405,
      "step": 2465
    },
    {
      "epoch": 0.38306796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.0007072912890901708,
      "loss": 2.3993,
      "step": 2466
    },
    {
      "epoch": 0.38322330097087376,
      "grad_norm": 0.0,
      "learning_rate": 0.0007070622955257964,
      "loss": 2.3844,
      "step": 2467
    },
    {
      "epoch": 0.38337864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0007068332495272888,
      "loss": 2.3254,
      "step": 2468
    },
    {
      "epoch": 0.38353398058252425,
      "grad_norm": 0.0,
      "learning_rate": 0.0007066041511526495,
      "loss": 2.2584,
      "step": 2469
    },
    {
      "epoch": 0.3836893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0007063750004598925,
      "loss": 2.2943,
      "step": 2470
    },
    {
      "epoch": 0.38384466019417474,
      "grad_norm": 0.0,
      "learning_rate": 0.0007061457975070456,
      "loss": 2.3829,
      "step": 2471
    },
    {
      "epoch": 0.384,
      "grad_norm": 0.0,
      "learning_rate": 0.0007059165423521492,
      "loss": 2.3224,
      "step": 2472
    },
    {
      "epoch": 0.3841553398058252,
      "grad_norm": 0.0,
      "learning_rate": 0.0007056872350532578,
      "loss": 2.2866,
      "step": 2473
    },
    {
      "epoch": 0.3843106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0007054578756684383,
      "loss": 2.3228,
      "step": 2474
    },
    {
      "epoch": 0.3844660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.0007052284642557711,
      "loss": 2.2871,
      "step": 2475
    },
    {
      "epoch": 0.384621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.00070499900087335,
      "loss": 2.2192,
      "step": 2476
    },
    {
      "epoch": 0.3847766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0007047694855792813,
      "loss": 2.4443,
      "step": 2477
    },
    {
      "epoch": 0.3849320388349515,
      "grad_norm": 0.0,
      "learning_rate": 0.0007045399184316853,
      "loss": 2.1892,
      "step": 2478
    },
    {
      "epoch": 0.3850873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0007043102994886948,
      "loss": 2.3533,
      "step": 2479
    },
    {
      "epoch": 0.38524271844660196,
      "grad_norm": 0.0,
      "learning_rate": 0.0007040806288084561,
      "loss": 2.2213,
      "step": 2480
    },
    {
      "epoch": 0.3853980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0007038509064491281,
      "loss": 2.2811,
      "step": 2481
    },
    {
      "epoch": 0.38555339805825245,
      "grad_norm": 0.0,
      "learning_rate": 0.0007036211324688835,
      "loss": 2.3358,
      "step": 2482
    },
    {
      "epoch": 0.38570873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0007033913069259075,
      "loss": 2.3452,
      "step": 2483
    },
    {
      "epoch": 0.38586407766990294,
      "grad_norm": 0.0,
      "learning_rate": 0.0007031614298783984,
      "loss": 2.4106,
      "step": 2484
    },
    {
      "epoch": 0.38601941747572815,
      "grad_norm": 0.0,
      "learning_rate": 0.0007029315013845679,
      "loss": 2.3461,
      "step": 2485
    },
    {
      "epoch": 0.38617475728155337,
      "grad_norm": 0.0,
      "learning_rate": 0.0007027015215026405,
      "loss": 2.276,
      "step": 2486
    },
    {
      "epoch": 0.38633009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0007024714902908536,
      "loss": 2.2714,
      "step": 2487
    },
    {
      "epoch": 0.38648543689320386,
      "grad_norm": 0.0,
      "learning_rate": 0.0007022414078074579,
      "loss": 2.2704,
      "step": 2488
    },
    {
      "epoch": 0.38664077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.0007020112741107167,
      "loss": 2.3052,
      "step": 2489
    },
    {
      "epoch": 0.38679611650485435,
      "grad_norm": 0.0,
      "learning_rate": 0.0007017810892589065,
      "loss": 2.2462,
      "step": 2490
    },
    {
      "epoch": 0.3869514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0007015508533103169,
      "loss": 2.3138,
      "step": 2491
    },
    {
      "epoch": 0.38710679611650484,
      "grad_norm": 0.0,
      "learning_rate": 0.0007013205663232503,
      "loss": 2.2958,
      "step": 2492
    },
    {
      "epoch": 0.3872621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0007010902283560216,
      "loss": 2.3217,
      "step": 2493
    },
    {
      "epoch": 0.3874174757281553,
      "grad_norm": 0.0,
      "learning_rate": 0.0007008598394669595,
      "loss": 2.2687,
      "step": 2494
    },
    {
      "epoch": 0.3875728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.0007006293997144047,
      "loss": 2.3094,
      "step": 2495
    },
    {
      "epoch": 0.3877281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.0007003989091567114,
      "loss": 2.1467,
      "step": 2496
    },
    {
      "epoch": 0.3878834951456311,
      "grad_norm": 0.0,
      "learning_rate": 0.0007001683678522464,
      "loss": 2.4106,
      "step": 2497
    },
    {
      "epoch": 0.3880388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0006999377758593894,
      "loss": 2.2461,
      "step": 2498
    },
    {
      "epoch": 0.3881941747572816,
      "grad_norm": 0.0,
      "learning_rate": 0.000699707133236533,
      "loss": 2.3538,
      "step": 2499
    },
    {
      "epoch": 0.3883495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0006994764400420823,
      "loss": 2.2821,
      "step": 2500
    },
    {
      "epoch": 0.38850485436893206,
      "grad_norm": 0.0,
      "learning_rate": 0.0006992456963344555,
      "loss": 2.2699,
      "step": 2501
    },
    {
      "epoch": 0.3886601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.000699014902172084,
      "loss": 2.271,
      "step": 2502
    },
    {
      "epoch": 0.38881553398058255,
      "grad_norm": 0.0,
      "learning_rate": 0.0006987840576134111,
      "loss": 2.3219,
      "step": 2503
    },
    {
      "epoch": 0.38897087378640777,
      "grad_norm": 0.0,
      "learning_rate": 0.0006985531627168933,
      "loss": 2.3214,
      "step": 2504
    },
    {
      "epoch": 0.389126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0006983222175409999,
      "loss": 2.2382,
      "step": 2505
    },
    {
      "epoch": 0.38928155339805826,
      "grad_norm": 0.0,
      "learning_rate": 0.0006980912221442128,
      "loss": 2.2799,
      "step": 2506
    },
    {
      "epoch": 0.38943689320388347,
      "grad_norm": 0.0,
      "learning_rate": 0.000697860176585027,
      "loss": 2.2518,
      "step": 2507
    },
    {
      "epoch": 0.38959223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0006976290809219494,
      "loss": 2.3253,
      "step": 2508
    },
    {
      "epoch": 0.38974757281553396,
      "grad_norm": 0.0,
      "learning_rate": 0.0006973979352135006,
      "loss": 2.2338,
      "step": 2509
    },
    {
      "epoch": 0.38990291262135923,
      "grad_norm": 0.0,
      "learning_rate": 0.0006971667395182129,
      "loss": 2.2916,
      "step": 2510
    },
    {
      "epoch": 0.39005825242718445,
      "grad_norm": 0.0,
      "learning_rate": 0.0006969354938946318,
      "loss": 2.2768,
      "step": 2511
    },
    {
      "epoch": 0.3902135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0006967041984013155,
      "loss": 2.2379,
      "step": 2512
    },
    {
      "epoch": 0.39036893203883494,
      "grad_norm": 0.0,
      "learning_rate": 0.0006964728530968344,
      "loss": 2.2079,
      "step": 2513
    },
    {
      "epoch": 0.3905242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0006962414580397723,
      "loss": 2.1896,
      "step": 2514
    },
    {
      "epoch": 0.3906796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.0006960100132887246,
      "loss": 2.2744,
      "step": 2515
    },
    {
      "epoch": 0.3908349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.0006957785189023,
      "loss": 2.3312,
      "step": 2516
    },
    {
      "epoch": 0.3909902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.0006955469749391195,
      "loss": 2.3015,
      "step": 2517
    },
    {
      "epoch": 0.3911456310679612,
      "grad_norm": 0.0,
      "learning_rate": 0.0006953153814578166,
      "loss": 2.2213,
      "step": 2518
    },
    {
      "epoch": 0.3913009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0006950837385170377,
      "loss": 2.1937,
      "step": 2519
    },
    {
      "epoch": 0.3914563106796117,
      "grad_norm": 0.0,
      "learning_rate": 0.0006948520461754411,
      "loss": 2.3678,
      "step": 2520
    },
    {
      "epoch": 0.3916116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0006946203044916981,
      "loss": 2.4255,
      "step": 2521
    },
    {
      "epoch": 0.39176699029126216,
      "grad_norm": 0.0,
      "learning_rate": 0.0006943885135244925,
      "loss": 2.3064,
      "step": 2522
    },
    {
      "epoch": 0.3919223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0006941566733325204,
      "loss": 2.2626,
      "step": 2523
    },
    {
      "epoch": 0.3920776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.0006939247839744903,
      "loss": 2.389,
      "step": 2524
    },
    {
      "epoch": 0.39223300970873787,
      "grad_norm": 0.0,
      "learning_rate": 0.0006936928455091233,
      "loss": 2.2728,
      "step": 2525
    },
    {
      "epoch": 0.3923883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.000693460857995153,
      "loss": 2.4027,
      "step": 2526
    },
    {
      "epoch": 0.39254368932038836,
      "grad_norm": 0.0,
      "learning_rate": 0.0006932288214913251,
      "loss": 2.3783,
      "step": 2527
    },
    {
      "epoch": 0.3926990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0006929967360563982,
      "loss": 2.2062,
      "step": 2528
    },
    {
      "epoch": 0.39285436893203884,
      "grad_norm": 0.0,
      "learning_rate": 0.0006927646017491428,
      "loss": 2.3163,
      "step": 2529
    },
    {
      "epoch": 0.39300970873786406,
      "grad_norm": 0.0,
      "learning_rate": 0.000692532418628342,
      "loss": 2.3191,
      "step": 2530
    },
    {
      "epoch": 0.39316504854368933,
      "grad_norm": 0.0,
      "learning_rate": 0.0006923001867527915,
      "loss": 2.262,
      "step": 2531
    },
    {
      "epoch": 0.39332038834951455,
      "grad_norm": 0.0,
      "learning_rate": 0.0006920679061812988,
      "loss": 2.3074,
      "step": 2532
    },
    {
      "epoch": 0.3934757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0006918355769726841,
      "loss": 2.2462,
      "step": 2533
    },
    {
      "epoch": 0.39363106796116504,
      "grad_norm": 0.0,
      "learning_rate": 0.0006916031991857798,
      "loss": 2.2694,
      "step": 2534
    },
    {
      "epoch": 0.3937864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0006913707728794308,
      "loss": 2.1904,
      "step": 2535
    },
    {
      "epoch": 0.3939417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.000691138298112494,
      "loss": 2.337,
      "step": 2536
    },
    {
      "epoch": 0.3940970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.0006909057749438389,
      "loss": 2.3071,
      "step": 2537
    },
    {
      "epoch": 0.394252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0006906732034323466,
      "loss": 2.2396,
      "step": 2538
    },
    {
      "epoch": 0.3944077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.0006904405836369114,
      "loss": 2.2219,
      "step": 2539
    },
    {
      "epoch": 0.3945631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0006902079156164392,
      "loss": 2.344,
      "step": 2540
    },
    {
      "epoch": 0.3947184466019418,
      "grad_norm": 0.0,
      "learning_rate": 0.000689975199429848,
      "loss": 2.2988,
      "step": 2541
    },
    {
      "epoch": 0.394873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0006897424351360685,
      "loss": 2.2389,
      "step": 2542
    },
    {
      "epoch": 0.3950291262135922,
      "grad_norm": 0.0,
      "learning_rate": 0.0006895096227940432,
      "loss": 2.2899,
      "step": 2543
    },
    {
      "epoch": 0.3951844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.000689276762462727,
      "loss": 2.3036,
      "step": 2544
    },
    {
      "epoch": 0.3953398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.0006890438542010866,
      "loss": 2.2996,
      "step": 2545
    },
    {
      "epoch": 0.39549514563106797,
      "grad_norm": 0.0,
      "learning_rate": 0.0006888108980681014,
      "loss": 2.2654,
      "step": 2546
    },
    {
      "epoch": 0.3956504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0006885778941227623,
      "loss": 2.2412,
      "step": 2547
    },
    {
      "epoch": 0.39580582524271846,
      "grad_norm": 0.0,
      "learning_rate": 0.0006883448424240728,
      "loss": 2.3138,
      "step": 2548
    },
    {
      "epoch": 0.3959611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0006881117430310482,
      "loss": 2.2274,
      "step": 2549
    },
    {
      "epoch": 0.39611650485436894,
      "grad_norm": 0.0,
      "learning_rate": 0.000687878596002716,
      "loss": 2.3643,
      "step": 2550
    },
    {
      "epoch": 0.39627184466019416,
      "grad_norm": 0.0,
      "learning_rate": 0.0006876454013981157,
      "loss": 2.1719,
      "step": 2551
    },
    {
      "epoch": 0.39642718446601943,
      "grad_norm": 0.0,
      "learning_rate": 0.000687412159276299,
      "loss": 2.4103,
      "step": 2552
    },
    {
      "epoch": 0.39658252427184465,
      "grad_norm": 0.0,
      "learning_rate": 0.0006871788696963293,
      "loss": 2.2942,
      "step": 2553
    },
    {
      "epoch": 0.3967378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0006869455327172825,
      "loss": 2.3005,
      "step": 2554
    },
    {
      "epoch": 0.39689320388349514,
      "grad_norm": 0.0,
      "learning_rate": 0.0006867121483982462,
      "loss": 2.2106,
      "step": 2555
    },
    {
      "epoch": 0.3970485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0006864787167983197,
      "loss": 2.2861,
      "step": 2556
    },
    {
      "epoch": 0.3972038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.0006862452379766151,
      "loss": 2.2665,
      "step": 2557
    },
    {
      "epoch": 0.3973592233009709,
      "grad_norm": 0.0,
      "learning_rate": 0.0006860117119922556,
      "loss": 2.1851,
      "step": 2558
    },
    {
      "epoch": 0.3975145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.0006857781389043767,
      "loss": 2.3164,
      "step": 2559
    },
    {
      "epoch": 0.3976699029126214,
      "grad_norm": 0.0,
      "learning_rate": 0.0006855445187721261,
      "loss": 2.275,
      "step": 2560
    },
    {
      "epoch": 0.3978252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.000685310851654663,
      "loss": 2.3326,
      "step": 2561
    },
    {
      "epoch": 0.3979805825242718,
      "grad_norm": 0.0,
      "learning_rate": 0.0006850771376111584,
      "loss": 2.4205,
      "step": 2562
    },
    {
      "epoch": 0.3981359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0006848433767007957,
      "loss": 2.342,
      "step": 2563
    },
    {
      "epoch": 0.3982912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.0006846095689827701,
      "loss": 2.2416,
      "step": 2564
    },
    {
      "epoch": 0.3984466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.0006843757145162879,
      "loss": 2.4247,
      "step": 2565
    },
    {
      "epoch": 0.3986019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.0006841418133605681,
      "loss": 2.2063,
      "step": 2566
    },
    {
      "epoch": 0.39875728155339807,
      "grad_norm": 0.0,
      "learning_rate": 0.0006839078655748412,
      "loss": 2.3475,
      "step": 2567
    },
    {
      "epoch": 0.3989126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0006836738712183494,
      "loss": 2.2643,
      "step": 2568
    },
    {
      "epoch": 0.39906796116504856,
      "grad_norm": 0.0,
      "learning_rate": 0.000683439830350347,
      "loss": 2.2857,
      "step": 2569
    },
    {
      "epoch": 0.3992233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.0006832057430300997,
      "loss": 2.3263,
      "step": 2570
    },
    {
      "epoch": 0.39937864077669905,
      "grad_norm": 0.0,
      "learning_rate": 0.0006829716093168851,
      "loss": 2.3119,
      "step": 2571
    },
    {
      "epoch": 0.39953398058252426,
      "grad_norm": 0.0,
      "learning_rate": 0.0006827374292699928,
      "loss": 2.2051,
      "step": 2572
    },
    {
      "epoch": 0.39968932038834953,
      "grad_norm": 0.0,
      "learning_rate": 0.0006825032029487239,
      "loss": 2.3713,
      "step": 2573
    },
    {
      "epoch": 0.39984466019417475,
      "grad_norm": 0.0,
      "learning_rate": 0.0006822689304123911,
      "loss": 2.292,
      "step": 2574
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.0,
      "learning_rate": 0.0006820346117203191,
      "loss": 2.2133,
      "step": 2575
    },
    {
      "epoch": 0.40015533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.000681800246931844,
      "loss": 2.3672,
      "step": 2576
    },
    {
      "epoch": 0.4003106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0006815658361063137,
      "loss": 2.2467,
      "step": 2577
    },
    {
      "epoch": 0.4004660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.0006813313793030879,
      "loss": 2.1828,
      "step": 2578
    },
    {
      "epoch": 0.40062135922330094,
      "grad_norm": 0.0,
      "learning_rate": 0.0006810968765815379,
      "loss": 2.3297,
      "step": 2579
    },
    {
      "epoch": 0.4007766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0006808623280010462,
      "loss": 2.3533,
      "step": 2580
    },
    {
      "epoch": 0.40093203883495143,
      "grad_norm": 0.0,
      "learning_rate": 0.0006806277336210075,
      "loss": 2.365,
      "step": 2581
    },
    {
      "epoch": 0.4010873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0006803930935008277,
      "loss": 2.2646,
      "step": 2582
    },
    {
      "epoch": 0.4012427184466019,
      "grad_norm": 0.0,
      "learning_rate": 0.0006801584076999246,
      "loss": 2.2294,
      "step": 2583
    },
    {
      "epoch": 0.4013980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0006799236762777274,
      "loss": 2.1954,
      "step": 2584
    },
    {
      "epoch": 0.4015533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.0006796888992936768,
      "loss": 2.2498,
      "step": 2585
    },
    {
      "epoch": 0.4017087378640777,
      "grad_norm": 0.0,
      "learning_rate": 0.0006794540768072248,
      "loss": 2.2294,
      "step": 2586
    },
    {
      "epoch": 0.4018640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.0006792192088778357,
      "loss": 2.3613,
      "step": 2587
    },
    {
      "epoch": 0.40201941747572817,
      "grad_norm": 0.0,
      "learning_rate": 0.0006789842955649847,
      "loss": 2.2827,
      "step": 2588
    },
    {
      "epoch": 0.4021747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0006787493369281583,
      "loss": 2.3688,
      "step": 2589
    },
    {
      "epoch": 0.40233009708737866,
      "grad_norm": 0.0,
      "learning_rate": 0.0006785143330268551,
      "loss": 2.3214,
      "step": 2590
    },
    {
      "epoch": 0.4024854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.0006782792839205846,
      "loss": 2.3193,
      "step": 2591
    },
    {
      "epoch": 0.40264077669902915,
      "grad_norm": 0.0,
      "learning_rate": 0.0006780441896688682,
      "loss": 2.3207,
      "step": 2592
    },
    {
      "epoch": 0.40279611650485436,
      "grad_norm": 0.0,
      "learning_rate": 0.0006778090503312386,
      "loss": 2.3207,
      "step": 2593
    },
    {
      "epoch": 0.40295145631067963,
      "grad_norm": 0.0,
      "learning_rate": 0.0006775738659672396,
      "loss": 2.3074,
      "step": 2594
    },
    {
      "epoch": 0.40310679611650485,
      "grad_norm": 0.0,
      "learning_rate": 0.0006773386366364269,
      "loss": 2.3401,
      "step": 2595
    },
    {
      "epoch": 0.4032621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0006771033623983671,
      "loss": 2.2817,
      "step": 2596
    },
    {
      "epoch": 0.40341747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0006768680433126386,
      "loss": 2.2467,
      "step": 2597
    },
    {
      "epoch": 0.40357281553398056,
      "grad_norm": 0.0,
      "learning_rate": 0.0006766326794388309,
      "loss": 2.3001,
      "step": 2598
    },
    {
      "epoch": 0.4037281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.0006763972708365449,
      "loss": 2.3337,
      "step": 2599
    },
    {
      "epoch": 0.40388349514563104,
      "grad_norm": 0.0,
      "learning_rate": 0.0006761618175653928,
      "loss": 2.3186,
      "step": 2600
    },
    {
      "epoch": 0.4040388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0006759263196849981,
      "loss": 2.1919,
      "step": 2601
    },
    {
      "epoch": 0.40419417475728153,
      "grad_norm": 0.0,
      "learning_rate": 0.0006756907772549959,
      "loss": 2.3867,
      "step": 2602
    },
    {
      "epoch": 0.4043495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0006754551903350319,
      "loss": 2.189,
      "step": 2603
    },
    {
      "epoch": 0.404504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0006752195589847639,
      "loss": 2.3976,
      "step": 2604
    },
    {
      "epoch": 0.4046601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.0006749838832638601,
      "loss": 2.1631,
      "step": 2605
    },
    {
      "epoch": 0.4048155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.000674748163232001,
      "loss": 2.36,
      "step": 2606
    },
    {
      "epoch": 0.4049708737864078,
      "grad_norm": 0.0,
      "learning_rate": 0.0006745123989488772,
      "loss": 2.2886,
      "step": 2607
    },
    {
      "epoch": 0.405126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0006742765904741912,
      "loss": 2.2611,
      "step": 2608
    },
    {
      "epoch": 0.40528155339805827,
      "grad_norm": 0.0,
      "learning_rate": 0.0006740407378676564,
      "loss": 2.2753,
      "step": 2609
    },
    {
      "epoch": 0.4054368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0006738048411889974,
      "loss": 2.3256,
      "step": 2610
    },
    {
      "epoch": 0.40559223300970876,
      "grad_norm": 0.0,
      "learning_rate": 0.0006735689004979503,
      "loss": 2.2279,
      "step": 2611
    },
    {
      "epoch": 0.405747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0006733329158542618,
      "loss": 2.2008,
      "step": 2612
    },
    {
      "epoch": 0.40590291262135925,
      "grad_norm": 0.0,
      "learning_rate": 0.0006730968873176902,
      "loss": 2.239,
      "step": 2613
    },
    {
      "epoch": 0.40605825242718446,
      "grad_norm": 0.0,
      "learning_rate": 0.0006728608149480045,
      "loss": 2.3041,
      "step": 2614
    },
    {
      "epoch": 0.40621359223300973,
      "grad_norm": 0.0,
      "learning_rate": 0.0006726246988049853,
      "loss": 2.2947,
      "step": 2615
    },
    {
      "epoch": 0.40636893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0006723885389484237,
      "loss": 2.235,
      "step": 2616
    },
    {
      "epoch": 0.40652427184466017,
      "grad_norm": 0.0,
      "learning_rate": 0.0006721523354381225,
      "loss": 2.2004,
      "step": 2617
    },
    {
      "epoch": 0.40667961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.0006719160883338949,
      "loss": 2.3057,
      "step": 2618
    },
    {
      "epoch": 0.40683495145631066,
      "grad_norm": 0.0,
      "learning_rate": 0.0006716797976955655,
      "loss": 2.2413,
      "step": 2619
    },
    {
      "epoch": 0.40699029126213593,
      "grad_norm": 0.0,
      "learning_rate": 0.0006714434635829703,
      "loss": 2.3141,
      "step": 2620
    },
    {
      "epoch": 0.40714563106796114,
      "grad_norm": 0.0,
      "learning_rate": 0.0006712070860559554,
      "loss": 2.278,
      "step": 2621
    },
    {
      "epoch": 0.4073009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0006709706651743788,
      "loss": 2.2744,
      "step": 2622
    },
    {
      "epoch": 0.40745631067961163,
      "grad_norm": 0.0,
      "learning_rate": 0.0006707342009981088,
      "loss": 2.2915,
      "step": 2623
    },
    {
      "epoch": 0.4076116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0006704976935870249,
      "loss": 2.326,
      "step": 2624
    },
    {
      "epoch": 0.4077669902912621,
      "grad_norm": 0.0,
      "learning_rate": 0.0006702611430010177,
      "loss": 2.2625,
      "step": 2625
    },
    {
      "epoch": 0.4079223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0006700245492999886,
      "loss": 2.3271,
      "step": 2626
    },
    {
      "epoch": 0.4080776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.0006697879125438501,
      "loss": 2.2653,
      "step": 2627
    },
    {
      "epoch": 0.4082330097087379,
      "grad_norm": 0.0,
      "learning_rate": 0.0006695512327925252,
      "loss": 2.2511,
      "step": 2628
    },
    {
      "epoch": 0.4083883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.000669314510105948,
      "loss": 2.2024,
      "step": 2629
    },
    {
      "epoch": 0.40854368932038837,
      "grad_norm": 0.0,
      "learning_rate": 0.0006690777445440637,
      "loss": 2.2631,
      "step": 2630
    },
    {
      "epoch": 0.4086990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0006688409361668281,
      "loss": 2.3075,
      "step": 2631
    },
    {
      "epoch": 0.40885436893203886,
      "grad_norm": 0.0,
      "learning_rate": 0.0006686040850342078,
      "loss": 2.2726,
      "step": 2632
    },
    {
      "epoch": 0.4090097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.0006683671912061803,
      "loss": 2.3083,
      "step": 2633
    },
    {
      "epoch": 0.40916504854368935,
      "grad_norm": 0.0,
      "learning_rate": 0.0006681302547427341,
      "loss": 2.2467,
      "step": 2634
    },
    {
      "epoch": 0.40932038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.0006678932757038684,
      "loss": 2.3393,
      "step": 2635
    },
    {
      "epoch": 0.4094757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0006676562541495927,
      "loss": 2.2856,
      "step": 2636
    },
    {
      "epoch": 0.40963106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0006674191901399281,
      "loss": 2.2387,
      "step": 2637
    },
    {
      "epoch": 0.40978640776699027,
      "grad_norm": 0.0,
      "learning_rate": 0.0006671820837349057,
      "loss": 2.3114,
      "step": 2638
    },
    {
      "epoch": 0.40994174757281554,
      "grad_norm": 0.0,
      "learning_rate": 0.000666944934994568,
      "loss": 2.3146,
      "step": 2639
    },
    {
      "epoch": 0.41009708737864076,
      "grad_norm": 0.0,
      "learning_rate": 0.0006667077439789678,
      "loss": 2.244,
      "step": 2640
    },
    {
      "epoch": 0.41025242718446603,
      "grad_norm": 0.0,
      "learning_rate": 0.0006664705107481685,
      "loss": 2.2363,
      "step": 2641
    },
    {
      "epoch": 0.41040776699029125,
      "grad_norm": 0.0,
      "learning_rate": 0.0006662332353622446,
      "loss": 2.3525,
      "step": 2642
    },
    {
      "epoch": 0.4105631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0006659959178812807,
      "loss": 2.3196,
      "step": 2643
    },
    {
      "epoch": 0.41071844660194173,
      "grad_norm": 0.0,
      "learning_rate": 0.0006657585583653728,
      "loss": 2.2782,
      "step": 2644
    },
    {
      "epoch": 0.410873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0006655211568746272,
      "loss": 2.306,
      "step": 2645
    },
    {
      "epoch": 0.4110291262135922,
      "grad_norm": 0.0,
      "learning_rate": 0.0006652837134691605,
      "loss": 2.2338,
      "step": 2646
    },
    {
      "epoch": 0.4111844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.0006650462282091002,
      "loss": 2.2853,
      "step": 2647
    },
    {
      "epoch": 0.4113398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.0006648087011545847,
      "loss": 2.2847,
      "step": 2648
    },
    {
      "epoch": 0.411495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0006645711323657624,
      "loss": 2.2957,
      "step": 2649
    },
    {
      "epoch": 0.4116504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0006643335219027927,
      "loss": 2.3522,
      "step": 2650
    },
    {
      "epoch": 0.41180582524271847,
      "grad_norm": 0.0,
      "learning_rate": 0.0006640958698258455,
      "loss": 2.2328,
      "step": 2651
    },
    {
      "epoch": 0.4119611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.000663858176195101,
      "loss": 2.1445,
      "step": 2652
    },
    {
      "epoch": 0.41211650485436896,
      "grad_norm": 0.0,
      "learning_rate": 0.0006636204410707499,
      "loss": 2.2396,
      "step": 2653
    },
    {
      "epoch": 0.4122718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.0006633826645129939,
      "loss": 2.3992,
      "step": 2654
    },
    {
      "epoch": 0.4124271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.0006631448465820448,
      "loss": 2.2356,
      "step": 2655
    },
    {
      "epoch": 0.41258252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0006629069873381248,
      "loss": 2.4211,
      "step": 2656
    },
    {
      "epoch": 0.4127378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0006626690868414669,
      "loss": 2.3326,
      "step": 2657
    },
    {
      "epoch": 0.41289320388349515,
      "grad_norm": 0.0,
      "learning_rate": 0.0006624311451523142,
      "loss": 2.3273,
      "step": 2658
    },
    {
      "epoch": 0.41304854368932037,
      "grad_norm": 0.0,
      "learning_rate": 0.0006621931623309203,
      "loss": 2.278,
      "step": 2659
    },
    {
      "epoch": 0.41320388349514564,
      "grad_norm": 0.0,
      "learning_rate": 0.0006619551384375495,
      "loss": 2.396,
      "step": 2660
    },
    {
      "epoch": 0.41335922330097086,
      "grad_norm": 0.0,
      "learning_rate": 0.0006617170735324764,
      "loss": 2.3595,
      "step": 2661
    },
    {
      "epoch": 0.41351456310679613,
      "grad_norm": 0.0,
      "learning_rate": 0.0006614789676759854,
      "loss": 2.2859,
      "step": 2662
    },
    {
      "epoch": 0.41366990291262135,
      "grad_norm": 0.0,
      "learning_rate": 0.0006612408209283724,
      "loss": 2.285,
      "step": 2663
    },
    {
      "epoch": 0.4138252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0006610026333499425,
      "loss": 2.4093,
      "step": 2664
    },
    {
      "epoch": 0.41398058252427183,
      "grad_norm": 0.0,
      "learning_rate": 0.0006607644050010119,
      "loss": 2.2676,
      "step": 2665
    },
    {
      "epoch": 0.4141359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0006605261359419066,
      "loss": 2.2457,
      "step": 2666
    },
    {
      "epoch": 0.4142912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.0006602878262329636,
      "loss": 2.2935,
      "step": 2667
    },
    {
      "epoch": 0.4144466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.0006600494759345295,
      "loss": 2.2416,
      "step": 2668
    },
    {
      "epoch": 0.4146019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.0006598110851069617,
      "loss": 2.3098,
      "step": 2669
    },
    {
      "epoch": 0.4147572815533981,
      "grad_norm": 0.0,
      "learning_rate": 0.0006595726538106273,
      "loss": 2.28,
      "step": 2670
    },
    {
      "epoch": 0.4149126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.000659334182105904,
      "loss": 2.3264,
      "step": 2671
    },
    {
      "epoch": 0.41506796116504857,
      "grad_norm": 0.0,
      "learning_rate": 0.0006590956700531799,
      "loss": 2.2926,
      "step": 2672
    },
    {
      "epoch": 0.4152233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.0006588571177128529,
      "loss": 2.3076,
      "step": 2673
    },
    {
      "epoch": 0.415378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0006586185251453319,
      "loss": 2.114,
      "step": 2674
    },
    {
      "epoch": 0.4155339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0006583798924110345,
      "loss": 2.3646,
      "step": 2675
    },
    {
      "epoch": 0.4156893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0006581412195703899,
      "loss": 2.3521,
      "step": 2676
    },
    {
      "epoch": 0.41584466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.0006579025066838368,
      "loss": 2.3013,
      "step": 2677
    },
    {
      "epoch": 0.416,
      "grad_norm": 0.0,
      "learning_rate": 0.0006576637538118244,
      "loss": 2.2257,
      "step": 2678
    },
    {
      "epoch": 0.41615533980582525,
      "grad_norm": 0.0,
      "learning_rate": 0.0006574249610148116,
      "loss": 2.2452,
      "step": 2679
    },
    {
      "epoch": 0.41631067961165047,
      "grad_norm": 0.0,
      "learning_rate": 0.0006571861283532676,
      "loss": 2.2499,
      "step": 2680
    },
    {
      "epoch": 0.41646601941747574,
      "grad_norm": 0.0,
      "learning_rate": 0.0006569472558876718,
      "loss": 2.1652,
      "step": 2681
    },
    {
      "epoch": 0.41662135922330096,
      "grad_norm": 0.0,
      "learning_rate": 0.0006567083436785135,
      "loss": 2.224,
      "step": 2682
    },
    {
      "epoch": 0.41677669902912623,
      "grad_norm": 0.0,
      "learning_rate": 0.0006564693917862921,
      "loss": 2.2699,
      "step": 2683
    },
    {
      "epoch": 0.41693203883495145,
      "grad_norm": 0.0,
      "learning_rate": 0.0006562304002715173,
      "loss": 2.2965,
      "step": 2684
    },
    {
      "epoch": 0.4170873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0006559913691947085,
      "loss": 2.3326,
      "step": 2685
    },
    {
      "epoch": 0.41724271844660193,
      "grad_norm": 0.0,
      "learning_rate": 0.0006557522986163952,
      "loss": 2.1861,
      "step": 2686
    },
    {
      "epoch": 0.4173980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0006555131885971169,
      "loss": 2.3153,
      "step": 2687
    },
    {
      "epoch": 0.4175533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.0006552740391974235,
      "loss": 2.2579,
      "step": 2688
    },
    {
      "epoch": 0.4177087378640777,
      "grad_norm": 0.0,
      "learning_rate": 0.0006550348504778741,
      "loss": 2.3236,
      "step": 2689
    },
    {
      "epoch": 0.4178640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.0006547956224990382,
      "loss": 2.4092,
      "step": 2690
    },
    {
      "epoch": 0.41801941747572813,
      "grad_norm": 0.0,
      "learning_rate": 0.0006545563553214955,
      "loss": 2.1415,
      "step": 2691
    },
    {
      "epoch": 0.4181747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0006543170490058352,
      "loss": 2.3094,
      "step": 2692
    },
    {
      "epoch": 0.4183300970873786,
      "grad_norm": 0.0,
      "learning_rate": 0.0006540777036126563,
      "loss": 2.2582,
      "step": 2693
    },
    {
      "epoch": 0.4184854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.0006538383192025684,
      "loss": 2.2234,
      "step": 2694
    },
    {
      "epoch": 0.4186407766990291,
      "grad_norm": 0.0,
      "learning_rate": 0.0006535988958361901,
      "loss": 2.2699,
      "step": 2695
    },
    {
      "epoch": 0.4187961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.0006533594335741504,
      "loss": 2.2909,
      "step": 2696
    },
    {
      "epoch": 0.4189514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0006531199324770883,
      "loss": 2.2794,
      "step": 2697
    },
    {
      "epoch": 0.41910679611650486,
      "grad_norm": 0.0,
      "learning_rate": 0.0006528803926056521,
      "loss": 2.1807,
      "step": 2698
    },
    {
      "epoch": 0.4192621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0006526408140205003,
      "loss": 2.3901,
      "step": 2699
    },
    {
      "epoch": 0.41941747572815535,
      "grad_norm": 0.0,
      "learning_rate": 0.0006524011967823009,
      "loss": 2.3202,
      "step": 2700
    },
    {
      "epoch": 0.41957281553398057,
      "grad_norm": 0.0,
      "learning_rate": 0.0006521615409517323,
      "loss": 2.2676,
      "step": 2701
    },
    {
      "epoch": 0.41972815533980584,
      "grad_norm": 0.0,
      "learning_rate": 0.0006519218465894819,
      "loss": 2.3597,
      "step": 2702
    },
    {
      "epoch": 0.41988349514563106,
      "grad_norm": 0.0,
      "learning_rate": 0.0006516821137562473,
      "loss": 2.3645,
      "step": 2703
    },
    {
      "epoch": 0.42003883495145633,
      "grad_norm": 0.0,
      "learning_rate": 0.0006514423425127356,
      "loss": 2.26,
      "step": 2704
    },
    {
      "epoch": 0.42019417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.000651202532919664,
      "loss": 2.3148,
      "step": 2705
    },
    {
      "epoch": 0.4203495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0006509626850377592,
      "loss": 2.2655,
      "step": 2706
    },
    {
      "epoch": 0.42050485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0006507227989277576,
      "loss": 2.2354,
      "step": 2707
    },
    {
      "epoch": 0.4206601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.0006504828746504052,
      "loss": 2.2705,
      "step": 2708
    },
    {
      "epoch": 0.4208155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.0006502429122664575,
      "loss": 2.2235,
      "step": 2709
    },
    {
      "epoch": 0.42097087378640774,
      "grad_norm": 0.0,
      "learning_rate": 0.0006500029118366803,
      "loss": 2.3422,
      "step": 2710
    },
    {
      "epoch": 0.421126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0006497628734218482,
      "loss": 2.2502,
      "step": 2711
    },
    {
      "epoch": 0.42128155339805823,
      "grad_norm": 0.0,
      "learning_rate": 0.0006495227970827462,
      "loss": 2.337,
      "step": 2712
    },
    {
      "epoch": 0.4214368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0006492826828801685,
      "loss": 2.2479,
      "step": 2713
    },
    {
      "epoch": 0.4215922330097087,
      "grad_norm": 0.0,
      "learning_rate": 0.0006490425308749186,
      "loss": 2.2501,
      "step": 2714
    },
    {
      "epoch": 0.421747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0006488023411278102,
      "loss": 2.2013,
      "step": 2715
    },
    {
      "epoch": 0.4219029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.0006485621136996662,
      "loss": 2.2172,
      "step": 2716
    },
    {
      "epoch": 0.4220582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.000648321848651319,
      "loss": 2.3398,
      "step": 2717
    },
    {
      "epoch": 0.4222135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0006480815460436108,
      "loss": 2.3035,
      "step": 2718
    },
    {
      "epoch": 0.42236893203883497,
      "grad_norm": 0.0,
      "learning_rate": 0.0006478412059373929,
      "loss": 2.321,
      "step": 2719
    },
    {
      "epoch": 0.4225242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0006476008283935266,
      "loss": 2.4622,
      "step": 2720
    },
    {
      "epoch": 0.42267961165048545,
      "grad_norm": 0.0,
      "learning_rate": 0.0006473604134728824,
      "loss": 2.3474,
      "step": 2721
    },
    {
      "epoch": 0.42283495145631067,
      "grad_norm": 0.0,
      "learning_rate": 0.0006471199612363399,
      "loss": 2.2439,
      "step": 2722
    },
    {
      "epoch": 0.42299029126213594,
      "grad_norm": 0.0,
      "learning_rate": 0.0006468794717447889,
      "loss": 2.2686,
      "step": 2723
    },
    {
      "epoch": 0.42314563106796116,
      "grad_norm": 0.0,
      "learning_rate": 0.000646638945059128,
      "loss": 2.3607,
      "step": 2724
    },
    {
      "epoch": 0.42330097087378643,
      "grad_norm": 0.0,
      "learning_rate": 0.0006463983812402659,
      "loss": 2.3313,
      "step": 2725
    },
    {
      "epoch": 0.42345631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0006461577803491198,
      "loss": 2.1841,
      "step": 2726
    },
    {
      "epoch": 0.4236116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0006459171424466171,
      "loss": 2.2635,
      "step": 2727
    },
    {
      "epoch": 0.42376699029126214,
      "grad_norm": 0.0,
      "learning_rate": 0.0006456764675936939,
      "loss": 2.28,
      "step": 2728
    },
    {
      "epoch": 0.42392233009708735,
      "grad_norm": 0.0,
      "learning_rate": 0.0006454357558512965,
      "loss": 2.1655,
      "step": 2729
    },
    {
      "epoch": 0.4240776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.0006451950072803795,
      "loss": 2.2257,
      "step": 2730
    },
    {
      "epoch": 0.42423300970873784,
      "grad_norm": 0.0,
      "learning_rate": 0.0006449542219419079,
      "loss": 2.3225,
      "step": 2731
    },
    {
      "epoch": 0.4243883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0006447133998968551,
      "loss": 2.2617,
      "step": 2732
    },
    {
      "epoch": 0.42454368932038833,
      "grad_norm": 0.0,
      "learning_rate": 0.0006444725412062041,
      "loss": 2.3114,
      "step": 2733
    },
    {
      "epoch": 0.4246990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0006442316459309476,
      "loss": 2.245,
      "step": 2734
    },
    {
      "epoch": 0.4248543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.0006439907141320871,
      "loss": 2.323,
      "step": 2735
    },
    {
      "epoch": 0.4250097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.0006437497458706334,
      "loss": 2.3191,
      "step": 2736
    },
    {
      "epoch": 0.4251650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.0006435087412076066,
      "loss": 2.3236,
      "step": 2737
    },
    {
      "epoch": 0.4253203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.000643267700204036,
      "loss": 2.2768,
      "step": 2738
    },
    {
      "epoch": 0.4254757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0006430266229209601,
      "loss": 2.3073,
      "step": 2739
    },
    {
      "epoch": 0.42563106796116507,
      "grad_norm": 0.0,
      "learning_rate": 0.0006427855094194268,
      "loss": 2.2228,
      "step": 2740
    },
    {
      "epoch": 0.4257864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0006425443597604928,
      "loss": 2.3276,
      "step": 2741
    },
    {
      "epoch": 0.42594174757281555,
      "grad_norm": 0.0,
      "learning_rate": 0.0006423031740052241,
      "loss": 2.3837,
      "step": 2742
    },
    {
      "epoch": 0.42609708737864077,
      "grad_norm": 0.0,
      "learning_rate": 0.0006420619522146961,
      "loss": 2.291,
      "step": 2743
    },
    {
      "epoch": 0.42625242718446604,
      "grad_norm": 0.0,
      "learning_rate": 0.0006418206944499928,
      "loss": 2.1933,
      "step": 2744
    },
    {
      "epoch": 0.42640776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.0006415794007722078,
      "loss": 2.2791,
      "step": 2745
    },
    {
      "epoch": 0.42656310679611653,
      "grad_norm": 0.0,
      "learning_rate": 0.0006413380712424436,
      "loss": 2.2068,
      "step": 2746
    },
    {
      "epoch": 0.42671844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.0006410967059218117,
      "loss": 2.3013,
      "step": 2747
    },
    {
      "epoch": 0.42687378640776696,
      "grad_norm": 0.0,
      "learning_rate": 0.0006408553048714329,
      "loss": 2.2786,
      "step": 2748
    },
    {
      "epoch": 0.42702912621359224,
      "grad_norm": 0.0,
      "learning_rate": 0.0006406138681524367,
      "loss": 2.2186,
      "step": 2749
    },
    {
      "epoch": 0.42718446601941745,
      "grad_norm": 0.0,
      "learning_rate": 0.0006403723958259618,
      "loss": 2.2922,
      "step": 2750
    },
    {
      "epoch": 0.4273398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.0006401308879531562,
      "loss": 2.2672,
      "step": 2751
    },
    {
      "epoch": 0.42749514563106794,
      "grad_norm": 0.0,
      "learning_rate": 0.0006398893445951764,
      "loss": 2.2866,
      "step": 2752
    },
    {
      "epoch": 0.4276504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0006396477658131883,
      "loss": 2.2657,
      "step": 2753
    },
    {
      "epoch": 0.42780582524271843,
      "grad_norm": 0.0,
      "learning_rate": 0.0006394061516683667,
      "loss": 2.2068,
      "step": 2754
    },
    {
      "epoch": 0.4279611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0006391645022218949,
      "loss": 2.2821,
      "step": 2755
    },
    {
      "epoch": 0.4281165048543689,
      "grad_norm": 0.0,
      "learning_rate": 0.0006389228175349657,
      "loss": 2.2174,
      "step": 2756
    },
    {
      "epoch": 0.4282718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.0006386810976687805,
      "loss": 2.2563,
      "step": 2757
    },
    {
      "epoch": 0.4284271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.00063843934268455,
      "loss": 2.3354,
      "step": 2758
    },
    {
      "epoch": 0.4285825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.0006381975526434933,
      "loss": 2.2968,
      "step": 2759
    },
    {
      "epoch": 0.4287378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0006379557276068388,
      "loss": 2.2491,
      "step": 2760
    },
    {
      "epoch": 0.42889320388349517,
      "grad_norm": 0.0,
      "learning_rate": 0.000637713867635823,
      "loss": 2.3196,
      "step": 2761
    },
    {
      "epoch": 0.4290485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0006374719727916925,
      "loss": 2.3033,
      "step": 2762
    },
    {
      "epoch": 0.42920388349514565,
      "grad_norm": 0.0,
      "learning_rate": 0.0006372300431357016,
      "loss": 2.2125,
      "step": 2763
    },
    {
      "epoch": 0.42935922330097087,
      "grad_norm": 0.0,
      "learning_rate": 0.0006369880787291143,
      "loss": 2.2158,
      "step": 2764
    },
    {
      "epoch": 0.42951456310679614,
      "grad_norm": 0.0,
      "learning_rate": 0.0006367460796332025,
      "loss": 2.2655,
      "step": 2765
    },
    {
      "epoch": 0.42966990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0006365040459092473,
      "loss": 2.3055,
      "step": 2766
    },
    {
      "epoch": 0.4298252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.000636261977618539,
      "loss": 2.3517,
      "step": 2767
    },
    {
      "epoch": 0.42998058252427185,
      "grad_norm": 0.0,
      "learning_rate": 0.000636019874822376,
      "loss": 2.2499,
      "step": 2768
    },
    {
      "epoch": 0.43013592233009706,
      "grad_norm": 0.0,
      "learning_rate": 0.0006357777375820656,
      "loss": 2.2708,
      "step": 2769
    },
    {
      "epoch": 0.43029126213592234,
      "grad_norm": 0.0,
      "learning_rate": 0.0006355355659589244,
      "loss": 2.2333,
      "step": 2770
    },
    {
      "epoch": 0.43044660194174755,
      "grad_norm": 0.0,
      "learning_rate": 0.0006352933600142766,
      "loss": 2.2744,
      "step": 2771
    },
    {
      "epoch": 0.4306019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.0006350511198094562,
      "loss": 2.3656,
      "step": 2772
    },
    {
      "epoch": 0.43075728155339804,
      "grad_norm": 0.0,
      "learning_rate": 0.0006348088454058051,
      "loss": 2.2242,
      "step": 2773
    },
    {
      "epoch": 0.4309126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0006345665368646743,
      "loss": 2.2892,
      "step": 2774
    },
    {
      "epoch": 0.43106796116504853,
      "grad_norm": 0.0,
      "learning_rate": 0.0006343241942474232,
      "loss": 2.2956,
      "step": 2775
    },
    {
      "epoch": 0.4312233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.0006340818176154198,
      "loss": 2.2432,
      "step": 2776
    },
    {
      "epoch": 0.431378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0006338394070300414,
      "loss": 2.3032,
      "step": 2777
    },
    {
      "epoch": 0.4315339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0006335969625526727,
      "loss": 2.268,
      "step": 2778
    },
    {
      "epoch": 0.4316893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.000633354484244708,
      "loss": 2.2689,
      "step": 2779
    },
    {
      "epoch": 0.4318446601941748,
      "grad_norm": 0.0,
      "learning_rate": 0.0006331119721675496,
      "loss": 2.328,
      "step": 2780
    },
    {
      "epoch": 0.432,
      "grad_norm": 0.0,
      "learning_rate": 0.0006328694263826085,
      "loss": 2.3224,
      "step": 2781
    },
    {
      "epoch": 0.43215533980582527,
      "grad_norm": 0.0,
      "learning_rate": 0.0006326268469513046,
      "loss": 2.3065,
      "step": 2782
    },
    {
      "epoch": 0.4323106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0006323842339350656,
      "loss": 2.334,
      "step": 2783
    },
    {
      "epoch": 0.43246601941747576,
      "grad_norm": 0.0,
      "learning_rate": 0.0006321415873953283,
      "loss": 2.2943,
      "step": 2784
    },
    {
      "epoch": 0.43262135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0006318989073935376,
      "loss": 2.236,
      "step": 2785
    },
    {
      "epoch": 0.4327766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0006316561939911473,
      "loss": 2.2202,
      "step": 2786
    },
    {
      "epoch": 0.43293203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.0006314134472496194,
      "loss": 2.197,
      "step": 2787
    },
    {
      "epoch": 0.4330873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0006311706672304245,
      "loss": 2.2255,
      "step": 2788
    },
    {
      "epoch": 0.43324271844660195,
      "grad_norm": 0.0,
      "learning_rate": 0.000630927853995041,
      "loss": 2.3093,
      "step": 2789
    },
    {
      "epoch": 0.43339805825242717,
      "grad_norm": 0.0,
      "learning_rate": 0.0006306850076049565,
      "loss": 2.359,
      "step": 2790
    },
    {
      "epoch": 0.43355339805825244,
      "grad_norm": 0.0,
      "learning_rate": 0.0006304421281216666,
      "loss": 2.2467,
      "step": 2791
    },
    {
      "epoch": 0.43370873786407765,
      "grad_norm": 0.0,
      "learning_rate": 0.0006301992156066756,
      "loss": 2.2677,
      "step": 2792
    },
    {
      "epoch": 0.4338640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.0006299562701214958,
      "loss": 2.1924,
      "step": 2793
    },
    {
      "epoch": 0.43401941747572814,
      "grad_norm": 0.0,
      "learning_rate": 0.0006297132917276481,
      "loss": 2.3453,
      "step": 2794
    },
    {
      "epoch": 0.4341747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0006294702804866614,
      "loss": 2.3427,
      "step": 2795
    },
    {
      "epoch": 0.43433009708737863,
      "grad_norm": 0.0,
      "learning_rate": 0.0006292272364600731,
      "loss": 2.2478,
      "step": 2796
    },
    {
      "epoch": 0.4344854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.0006289841597094292,
      "loss": 2.1981,
      "step": 2797
    },
    {
      "epoch": 0.4346407766990291,
      "grad_norm": 0.0,
      "learning_rate": 0.0006287410502962836,
      "loss": 2.2476,
      "step": 2798
    },
    {
      "epoch": 0.4347961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.0006284979082821987,
      "loss": 2.298,
      "step": 2799
    },
    {
      "epoch": 0.4349514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0006282547337287449,
      "loss": 2.3203,
      "step": 2800
    },
    {
      "epoch": 0.4351067961165049,
      "grad_norm": 0.0,
      "learning_rate": 0.0006280115266975011,
      "loss": 2.2743,
      "step": 2801
    },
    {
      "epoch": 0.4352621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0006277682872500543,
      "loss": 2.2932,
      "step": 2802
    },
    {
      "epoch": 0.4354174757281553,
      "grad_norm": 0.0,
      "learning_rate": 0.0006275250154479996,
      "loss": 2.2007,
      "step": 2803
    },
    {
      "epoch": 0.4355728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.0006272817113529407,
      "loss": 2.3018,
      "step": 2804
    },
    {
      "epoch": 0.4357281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.0006270383750264889,
      "loss": 2.2761,
      "step": 2805
    },
    {
      "epoch": 0.4358834951456311,
      "grad_norm": 0.0,
      "learning_rate": 0.0006267950065302644,
      "loss": 2.2322,
      "step": 2806
    },
    {
      "epoch": 0.4360388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0006265516059258946,
      "loss": 2.2306,
      "step": 2807
    },
    {
      "epoch": 0.43619417475728156,
      "grad_norm": 0.0,
      "learning_rate": 0.0006263081732750157,
      "loss": 2.2056,
      "step": 2808
    },
    {
      "epoch": 0.4363495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0006260647086392719,
      "loss": 2.2235,
      "step": 2809
    },
    {
      "epoch": 0.43650485436893205,
      "grad_norm": 0.0,
      "learning_rate": 0.0006258212120803155,
      "loss": 2.2372,
      "step": 2810
    },
    {
      "epoch": 0.43666019417475727,
      "grad_norm": 0.0,
      "learning_rate": 0.0006255776836598068,
      "loss": 2.1692,
      "step": 2811
    },
    {
      "epoch": 0.43681553398058254,
      "grad_norm": 0.0,
      "learning_rate": 0.0006253341234394144,
      "loss": 2.2723,
      "step": 2812
    },
    {
      "epoch": 0.43697087378640775,
      "grad_norm": 0.0,
      "learning_rate": 0.0006250905314808141,
      "loss": 2.2898,
      "step": 2813
    },
    {
      "epoch": 0.437126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0006248469078456912,
      "loss": 2.282,
      "step": 2814
    },
    {
      "epoch": 0.43728155339805824,
      "grad_norm": 0.0,
      "learning_rate": 0.0006246032525957378,
      "loss": 2.2714,
      "step": 2815
    },
    {
      "epoch": 0.4374368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0006243595657926545,
      "loss": 2.2618,
      "step": 2816
    },
    {
      "epoch": 0.43759223300970873,
      "grad_norm": 0.0,
      "learning_rate": 0.0006241158474981498,
      "loss": 2.3224,
      "step": 2817
    },
    {
      "epoch": 0.437747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0006238720977739401,
      "loss": 2.2812,
      "step": 2818
    },
    {
      "epoch": 0.4379029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.0006236283166817501,
      "loss": 2.3085,
      "step": 2819
    },
    {
      "epoch": 0.4380582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.000623384504283312,
      "loss": 2.1674,
      "step": 2820
    },
    {
      "epoch": 0.4382135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.000623140660640366,
      "loss": 2.2067,
      "step": 2821
    },
    {
      "epoch": 0.4383689320388349,
      "grad_norm": 0.0,
      "learning_rate": 0.0006228967858146607,
      "loss": 2.2715,
      "step": 2822
    },
    {
      "epoch": 0.4385242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0006226528798679518,
      "loss": 2.2305,
      "step": 2823
    },
    {
      "epoch": 0.4386796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.0006224089428620036,
      "loss": 2.3175,
      "step": 2824
    },
    {
      "epoch": 0.4388349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.0006221649748585878,
      "loss": 2.2943,
      "step": 2825
    },
    {
      "epoch": 0.4389902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.0006219209759194843,
      "loss": 2.3076,
      "step": 2826
    },
    {
      "epoch": 0.4391456310679612,
      "grad_norm": 0.0,
      "learning_rate": 0.0006216769461064804,
      "loss": 2.1823,
      "step": 2827
    },
    {
      "epoch": 0.4393009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0006214328854813718,
      "loss": 2.2249,
      "step": 2828
    },
    {
      "epoch": 0.43945631067961166,
      "grad_norm": 0.0,
      "learning_rate": 0.0006211887941059613,
      "loss": 2.2648,
      "step": 2829
    },
    {
      "epoch": 0.4396116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0006209446720420601,
      "loss": 2.3407,
      "step": 2830
    },
    {
      "epoch": 0.43976699029126215,
      "grad_norm": 0.0,
      "learning_rate": 0.000620700519351487,
      "loss": 2.2747,
      "step": 2831
    },
    {
      "epoch": 0.43992233009708737,
      "grad_norm": 0.0,
      "learning_rate": 0.0006204563360960683,
      "loss": 2.2064,
      "step": 2832
    },
    {
      "epoch": 0.44007766990291264,
      "grad_norm": 0.0,
      "learning_rate": 0.0006202121223376383,
      "loss": 2.1924,
      "step": 2833
    },
    {
      "epoch": 0.44023300970873785,
      "grad_norm": 0.0,
      "learning_rate": 0.0006199678781380391,
      "loss": 2.2294,
      "step": 2834
    },
    {
      "epoch": 0.4403883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0006197236035591201,
      "loss": 2.267,
      "step": 2835
    },
    {
      "epoch": 0.44054368932038834,
      "grad_norm": 0.0,
      "learning_rate": 0.0006194792986627388,
      "loss": 2.2214,
      "step": 2836
    },
    {
      "epoch": 0.4406990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0006192349635107601,
      "loss": 2.2546,
      "step": 2837
    },
    {
      "epoch": 0.44085436893203883,
      "grad_norm": 0.0,
      "learning_rate": 0.0006189905981650569,
      "loss": 2.2319,
      "step": 2838
    },
    {
      "epoch": 0.4410097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.0006187462026875093,
      "loss": 2.2417,
      "step": 2839
    },
    {
      "epoch": 0.4411650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.0006185017771400058,
      "loss": 2.2114,
      "step": 2840
    },
    {
      "epoch": 0.44132038834951454,
      "grad_norm": 0.0,
      "learning_rate": 0.0006182573215844412,
      "loss": 2.2238,
      "step": 2841
    },
    {
      "epoch": 0.4414757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0006180128360827191,
      "loss": 2.2649,
      "step": 2842
    },
    {
      "epoch": 0.441631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0006177683206967504,
      "loss": 2.1803,
      "step": 2843
    },
    {
      "epoch": 0.4417864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0006175237754884533,
      "loss": 2.2581,
      "step": 2844
    },
    {
      "epoch": 0.4419417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.0006172792005197536,
      "loss": 2.3442,
      "step": 2845
    },
    {
      "epoch": 0.4420970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.0006170345958525847,
      "loss": 2.3485,
      "step": 2846
    },
    {
      "epoch": 0.442252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0006167899615488877,
      "loss": 2.1808,
      "step": 2847
    },
    {
      "epoch": 0.4424077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.000616545297670611,
      "loss": 2.324,
      "step": 2848
    },
    {
      "epoch": 0.4425631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0006163006042797103,
      "loss": 2.2777,
      "step": 2849
    },
    {
      "epoch": 0.44271844660194176,
      "grad_norm": 0.0,
      "learning_rate": 0.0006160558814381494,
      "loss": 2.1948,
      "step": 2850
    },
    {
      "epoch": 0.442873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0006158111292078989,
      "loss": 2.3661,
      "step": 2851
    },
    {
      "epoch": 0.44302912621359225,
      "grad_norm": 0.0,
      "learning_rate": 0.0006155663476509373,
      "loss": 2.2307,
      "step": 2852
    },
    {
      "epoch": 0.44318446601941747,
      "grad_norm": 0.0,
      "learning_rate": 0.0006153215368292502,
      "loss": 2.0564,
      "step": 2853
    },
    {
      "epoch": 0.44333980582524274,
      "grad_norm": 0.0,
      "learning_rate": 0.0006150766968048307,
      "loss": 2.2596,
      "step": 2854
    },
    {
      "epoch": 0.44349514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0006148318276396795,
      "loss": 2.2689,
      "step": 2855
    },
    {
      "epoch": 0.4436504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0006145869293958044,
      "loss": 2.2754,
      "step": 2856
    },
    {
      "epoch": 0.44380582524271844,
      "grad_norm": 0.0,
      "learning_rate": 0.0006143420021352206,
      "loss": 2.217,
      "step": 2857
    },
    {
      "epoch": 0.4439611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0006140970459199511,
      "loss": 2.2899,
      "step": 2858
    },
    {
      "epoch": 0.44411650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.0006138520608120254,
      "loss": 2.3357,
      "step": 2859
    },
    {
      "epoch": 0.44427184466019415,
      "grad_norm": 0.0,
      "learning_rate": 0.000613607046873481,
      "loss": 2.2896,
      "step": 2860
    },
    {
      "epoch": 0.4444271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.0006133620041663625,
      "loss": 2.2778,
      "step": 2861
    },
    {
      "epoch": 0.44458252427184464,
      "grad_norm": 0.0,
      "learning_rate": 0.0006131169327527219,
      "loss": 2.2704,
      "step": 2862
    },
    {
      "epoch": 0.4447378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.000612871832694618,
      "loss": 2.2581,
      "step": 2863
    },
    {
      "epoch": 0.4448932038834951,
      "grad_norm": 0.0,
      "learning_rate": 0.0006126267040541174,
      "loss": 2.2609,
      "step": 2864
    },
    {
      "epoch": 0.4450485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0006123815468932937,
      "loss": 2.1353,
      "step": 2865
    },
    {
      "epoch": 0.4452038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.0006121363612742277,
      "loss": 2.2106,
      "step": 2866
    },
    {
      "epoch": 0.4453592233009709,
      "grad_norm": 0.0,
      "learning_rate": 0.0006118911472590075,
      "loss": 2.3293,
      "step": 2867
    },
    {
      "epoch": 0.4455145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.0006116459049097286,
      "loss": 2.2665,
      "step": 2868
    },
    {
      "epoch": 0.4456699029126214,
      "grad_norm": 0.0,
      "learning_rate": 0.0006114006342884929,
      "loss": 2.291,
      "step": 2869
    },
    {
      "epoch": 0.4458252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0006111553354574102,
      "loss": 2.3197,
      "step": 2870
    },
    {
      "epoch": 0.44598058252427186,
      "grad_norm": 0.0,
      "learning_rate": 0.0006109100084785975,
      "loss": 2.3473,
      "step": 2871
    },
    {
      "epoch": 0.4461359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0006106646534141783,
      "loss": 2.3186,
      "step": 2872
    },
    {
      "epoch": 0.44629126213592235,
      "grad_norm": 0.0,
      "learning_rate": 0.0006104192703262837,
      "loss": 2.2928,
      "step": 2873
    },
    {
      "epoch": 0.44644660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.0006101738592770518,
      "loss": 2.2457,
      "step": 2874
    },
    {
      "epoch": 0.44660194174757284,
      "grad_norm": 0.0,
      "learning_rate": 0.0006099284203286275,
      "loss": 2.3828,
      "step": 2875
    },
    {
      "epoch": 0.44675728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.0006096829535431634,
      "loss": 2.2134,
      "step": 2876
    },
    {
      "epoch": 0.4469126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0006094374589828184,
      "loss": 2.2706,
      "step": 2877
    },
    {
      "epoch": 0.44706796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.0006091919367097592,
      "loss": 2.2097,
      "step": 2878
    },
    {
      "epoch": 0.44722330097087376,
      "grad_norm": 0.0,
      "learning_rate": 0.0006089463867861585,
      "loss": 2.2169,
      "step": 2879
    },
    {
      "epoch": 0.44737864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0006087008092741972,
      "loss": 2.2823,
      "step": 2880
    },
    {
      "epoch": 0.44753398058252425,
      "grad_norm": 0.0,
      "learning_rate": 0.0006084552042360622,
      "loss": 2.3,
      "step": 2881
    },
    {
      "epoch": 0.4476893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0006082095717339481,
      "loss": 2.3484,
      "step": 2882
    },
    {
      "epoch": 0.44784466019417474,
      "grad_norm": 0.0,
      "learning_rate": 0.0006079639118300557,
      "loss": 2.2565,
      "step": 2883
    },
    {
      "epoch": 0.448,
      "grad_norm": 0.0,
      "learning_rate": 0.0006077182245865935,
      "loss": 2.233,
      "step": 2884
    },
    {
      "epoch": 0.4481553398058252,
      "grad_norm": 0.0,
      "learning_rate": 0.0006074725100657766,
      "loss": 2.3096,
      "step": 2885
    },
    {
      "epoch": 0.4483106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0006072267683298267,
      "loss": 2.3555,
      "step": 2886
    },
    {
      "epoch": 0.4484660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.0006069809994409728,
      "loss": 2.322,
      "step": 2887
    },
    {
      "epoch": 0.448621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0006067352034614509,
      "loss": 2.3139,
      "step": 2888
    },
    {
      "epoch": 0.4487766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0006064893804535032,
      "loss": 2.3071,
      "step": 2889
    },
    {
      "epoch": 0.4489320388349515,
      "grad_norm": 0.0,
      "learning_rate": 0.0006062435304793794,
      "loss": 2.3137,
      "step": 2890
    },
    {
      "epoch": 0.4490873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.000605997653601336,
      "loss": 2.2371,
      "step": 2891
    },
    {
      "epoch": 0.44924271844660196,
      "grad_norm": 0.0,
      "learning_rate": 0.0006057517498816356,
      "loss": 2.2397,
      "step": 2892
    },
    {
      "epoch": 0.4493980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0006055058193825485,
      "loss": 2.192,
      "step": 2893
    },
    {
      "epoch": 0.44955339805825245,
      "grad_norm": 0.0,
      "learning_rate": 0.0006052598621663511,
      "loss": 2.2528,
      "step": 2894
    },
    {
      "epoch": 0.44970873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0006050138782953271,
      "loss": 2.2609,
      "step": 2895
    },
    {
      "epoch": 0.44986407766990294,
      "grad_norm": 0.0,
      "learning_rate": 0.0006047678678317666,
      "loss": 2.2784,
      "step": 2896
    },
    {
      "epoch": 0.45001941747572816,
      "grad_norm": 0.0,
      "learning_rate": 0.0006045218308379667,
      "loss": 2.2404,
      "step": 2897
    },
    {
      "epoch": 0.4501747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0006042757673762306,
      "loss": 2.2659,
      "step": 2898
    },
    {
      "epoch": 0.45033009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.000604029677508869,
      "loss": 2.2026,
      "step": 2899
    },
    {
      "epoch": 0.45048543689320386,
      "grad_norm": 0.0,
      "learning_rate": 0.0006037835612981988,
      "loss": 2.285,
      "step": 2900
    },
    {
      "epoch": 0.45064077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.0006035374188065438,
      "loss": 2.2319,
      "step": 2901
    },
    {
      "epoch": 0.45079611650485435,
      "grad_norm": 0.0,
      "learning_rate": 0.0006032912500962345,
      "loss": 2.3081,
      "step": 2902
    },
    {
      "epoch": 0.4509514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0006030450552296074,
      "loss": 2.2178,
      "step": 2903
    },
    {
      "epoch": 0.45110679611650484,
      "grad_norm": 0.0,
      "learning_rate": 0.0006027988342690067,
      "loss": 2.1776,
      "step": 2904
    },
    {
      "epoch": 0.4512621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0006025525872767823,
      "loss": 2.1967,
      "step": 2905
    },
    {
      "epoch": 0.4514174757281553,
      "grad_norm": 0.0,
      "learning_rate": 0.0006023063143152912,
      "loss": 2.2257,
      "step": 2906
    },
    {
      "epoch": 0.4515728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.0006020600154468965,
      "loss": 2.246,
      "step": 2907
    },
    {
      "epoch": 0.4517281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.0006018136907339686,
      "loss": 2.2219,
      "step": 2908
    },
    {
      "epoch": 0.4518834951456311,
      "grad_norm": 0.0,
      "learning_rate": 0.0006015673402388838,
      "loss": 2.345,
      "step": 2909
    },
    {
      "epoch": 0.4520388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.000601320964024025,
      "loss": 2.3059,
      "step": 2910
    },
    {
      "epoch": 0.4521941747572816,
      "grad_norm": 0.0,
      "learning_rate": 0.0006010745621517819,
      "loss": 2.1448,
      "step": 2911
    },
    {
      "epoch": 0.4523495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0006008281346845507,
      "loss": 2.2009,
      "step": 2912
    },
    {
      "epoch": 0.45250485436893206,
      "grad_norm": 0.0,
      "learning_rate": 0.0006005816816847335,
      "loss": 2.1438,
      "step": 2913
    },
    {
      "epoch": 0.4526601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.0006003352032147397,
      "loss": 2.1488,
      "step": 2914
    },
    {
      "epoch": 0.45281553398058255,
      "grad_norm": 0.0,
      "learning_rate": 0.0006000886993369845,
      "loss": 2.2825,
      "step": 2915
    },
    {
      "epoch": 0.45297087378640777,
      "grad_norm": 0.0,
      "learning_rate": 0.0005998421701138898,
      "loss": 2.2292,
      "step": 2916
    },
    {
      "epoch": 0.453126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0005995956156078841,
      "loss": 2.2096,
      "step": 2917
    },
    {
      "epoch": 0.45328155339805826,
      "grad_norm": 0.0,
      "learning_rate": 0.0005993490358814018,
      "loss": 2.3094,
      "step": 2918
    },
    {
      "epoch": 0.4534368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0005991024309968844,
      "loss": 2.2313,
      "step": 2919
    },
    {
      "epoch": 0.45359223300970875,
      "grad_norm": 0.0,
      "learning_rate": 0.0005988558010167789,
      "loss": 2.2419,
      "step": 2920
    },
    {
      "epoch": 0.45374757281553396,
      "grad_norm": 0.0,
      "learning_rate": 0.0005986091460035392,
      "loss": 2.2366,
      "step": 2921
    },
    {
      "epoch": 0.45390291262135923,
      "grad_norm": 0.0,
      "learning_rate": 0.0005983624660196256,
      "loss": 2.168,
      "step": 2922
    },
    {
      "epoch": 0.45405825242718445,
      "grad_norm": 0.0,
      "learning_rate": 0.0005981157611275045,
      "loss": 2.2725,
      "step": 2923
    },
    {
      "epoch": 0.4542135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0005978690313896488,
      "loss": 2.2678,
      "step": 2924
    },
    {
      "epoch": 0.45436893203883494,
      "grad_norm": 0.0,
      "learning_rate": 0.0005976222768685376,
      "loss": 2.2053,
      "step": 2925
    },
    {
      "epoch": 0.4545242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0005973754976266558,
      "loss": 2.3397,
      "step": 2926
    },
    {
      "epoch": 0.4546796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.0005971286937264955,
      "loss": 2.2438,
      "step": 2927
    },
    {
      "epoch": 0.4548349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.0005968818652305544,
      "loss": 2.2405,
      "step": 2928
    },
    {
      "epoch": 0.4549902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.0005966350122013364,
      "loss": 2.319,
      "step": 2929
    },
    {
      "epoch": 0.4551456310679612,
      "grad_norm": 0.0,
      "learning_rate": 0.0005963881347013521,
      "loss": 2.3134,
      "step": 2930
    },
    {
      "epoch": 0.4553009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0005961412327931179,
      "loss": 2.2817,
      "step": 2931
    },
    {
      "epoch": 0.4554563106796117,
      "grad_norm": 0.0,
      "learning_rate": 0.0005958943065391562,
      "loss": 2.2492,
      "step": 2932
    },
    {
      "epoch": 0.4556116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0005956473560019961,
      "loss": 2.3154,
      "step": 2933
    },
    {
      "epoch": 0.4557669902912621,
      "grad_norm": 0.0,
      "learning_rate": 0.0005954003812441726,
      "loss": 2.3379,
      "step": 2934
    },
    {
      "epoch": 0.4559223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0005951533823282267,
      "loss": 2.262,
      "step": 2935
    },
    {
      "epoch": 0.4560776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.0005949063593167057,
      "loss": 2.2719,
      "step": 2936
    },
    {
      "epoch": 0.45623300970873787,
      "grad_norm": 0.0,
      "learning_rate": 0.0005946593122721629,
      "loss": 2.1482,
      "step": 2937
    },
    {
      "epoch": 0.4563883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0005944122412571578,
      "loss": 2.3579,
      "step": 2938
    },
    {
      "epoch": 0.45654368932038836,
      "grad_norm": 0.0,
      "learning_rate": 0.0005941651463342557,
      "loss": 2.2304,
      "step": 2939
    },
    {
      "epoch": 0.4566990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0005939180275660283,
      "loss": 2.3115,
      "step": 2940
    },
    {
      "epoch": 0.45685436893203885,
      "grad_norm": 0.0,
      "learning_rate": 0.0005936708850150531,
      "loss": 2.1852,
      "step": 2941
    },
    {
      "epoch": 0.45700970873786406,
      "grad_norm": 0.0,
      "learning_rate": 0.0005934237187439139,
      "loss": 2.2326,
      "step": 2942
    },
    {
      "epoch": 0.45716504854368933,
      "grad_norm": 0.0,
      "learning_rate": 0.0005931765288152001,
      "loss": 2.2776,
      "step": 2943
    },
    {
      "epoch": 0.45732038834951455,
      "grad_norm": 0.0,
      "learning_rate": 0.0005929293152915074,
      "loss": 2.2673,
      "step": 2944
    },
    {
      "epoch": 0.4574757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0005926820782354374,
      "loss": 2.2146,
      "step": 2945
    },
    {
      "epoch": 0.45763106796116504,
      "grad_norm": 0.0,
      "learning_rate": 0.0005924348177095975,
      "loss": 2.2601,
      "step": 2946
    },
    {
      "epoch": 0.4577864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0005921875337766012,
      "loss": 2.2444,
      "step": 2947
    },
    {
      "epoch": 0.4579417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.0005919402264990683,
      "loss": 2.2045,
      "step": 2948
    },
    {
      "epoch": 0.4580970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.0005916928959396237,
      "loss": 2.2284,
      "step": 2949
    },
    {
      "epoch": 0.458252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0005914455421608987,
      "loss": 2.3565,
      "step": 2950
    },
    {
      "epoch": 0.4584077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.0005911981652255302,
      "loss": 2.1993,
      "step": 2951
    },
    {
      "epoch": 0.4585631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0005909507651961617,
      "loss": 2.3573,
      "step": 2952
    },
    {
      "epoch": 0.4587184466019417,
      "grad_norm": 0.0,
      "learning_rate": 0.0005907033421354415,
      "loss": 2.2812,
      "step": 2953
    },
    {
      "epoch": 0.458873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0005904558961060247,
      "loss": 2.3603,
      "step": 2954
    },
    {
      "epoch": 0.4590291262135922,
      "grad_norm": 0.0,
      "learning_rate": 0.0005902084271705712,
      "loss": 2.2429,
      "step": 2955
    },
    {
      "epoch": 0.4591844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.0005899609353917476,
      "loss": 2.2925,
      "step": 2956
    },
    {
      "epoch": 0.4593398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.0005897134208322259,
      "loss": 2.3114,
      "step": 2957
    },
    {
      "epoch": 0.45949514563106797,
      "grad_norm": 0.0,
      "learning_rate": 0.000589465883554684,
      "loss": 2.251,
      "step": 2958
    },
    {
      "epoch": 0.4596504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0005892183236218054,
      "loss": 2.2227,
      "step": 2959
    },
    {
      "epoch": 0.45980582524271846,
      "grad_norm": 0.0,
      "learning_rate": 0.0005889707410962791,
      "loss": 2.2156,
      "step": 2960
    },
    {
      "epoch": 0.4599611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0005887231360408007,
      "loss": 2.3215,
      "step": 2961
    },
    {
      "epoch": 0.46011650485436895,
      "grad_norm": 0.0,
      "learning_rate": 0.0005884755085180704,
      "loss": 2.2815,
      "step": 2962
    },
    {
      "epoch": 0.46027184466019416,
      "grad_norm": 0.0,
      "learning_rate": 0.0005882278585907949,
      "loss": 2.4442,
      "step": 2963
    },
    {
      "epoch": 0.46042718446601943,
      "grad_norm": 0.0,
      "learning_rate": 0.0005879801863216862,
      "loss": 2.259,
      "step": 2964
    },
    {
      "epoch": 0.46058252427184465,
      "grad_norm": 0.0,
      "learning_rate": 0.0005877324917734621,
      "loss": 2.168,
      "step": 2965
    },
    {
      "epoch": 0.4607378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0005874847750088459,
      "loss": 2.2523,
      "step": 2966
    },
    {
      "epoch": 0.46089320388349514,
      "grad_norm": 0.0,
      "learning_rate": 0.0005872370360905666,
      "loss": 2.2131,
      "step": 2967
    },
    {
      "epoch": 0.4610485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0005869892750813591,
      "loss": 2.2488,
      "step": 2968
    },
    {
      "epoch": 0.46120388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0005867414920439632,
      "loss": 2.1797,
      "step": 2969
    },
    {
      "epoch": 0.4613592233009709,
      "grad_norm": 0.0,
      "learning_rate": 0.0005864936870411249,
      "loss": 2.2631,
      "step": 2970
    },
    {
      "epoch": 0.4615145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.0005862458601355958,
      "loss": 2.2825,
      "step": 2971
    },
    {
      "epoch": 0.46166990291262133,
      "grad_norm": 0.0,
      "learning_rate": 0.0005859980113901324,
      "loss": 2.2946,
      "step": 2972
    },
    {
      "epoch": 0.4618252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0005857501408674974,
      "loss": 2.1194,
      "step": 2973
    },
    {
      "epoch": 0.4619805825242718,
      "grad_norm": 0.0,
      "learning_rate": 0.0005855022486304587,
      "loss": 2.1993,
      "step": 2974
    },
    {
      "epoch": 0.4621359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0005852543347417893,
      "loss": 2.3853,
      "step": 2975
    },
    {
      "epoch": 0.4622912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.000585006399264269,
      "loss": 2.3284,
      "step": 2976
    },
    {
      "epoch": 0.4624466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.0005847584422606816,
      "loss": 2.3091,
      "step": 2977
    },
    {
      "epoch": 0.4626019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.0005845104637938172,
      "loss": 2.2755,
      "step": 2978
    },
    {
      "epoch": 0.46275728155339807,
      "grad_norm": 0.0,
      "learning_rate": 0.0005842624639264707,
      "loss": 2.1897,
      "step": 2979
    },
    {
      "epoch": 0.4629126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0005840144427214433,
      "loss": 2.1173,
      "step": 2980
    },
    {
      "epoch": 0.46306796116504856,
      "grad_norm": 0.0,
      "learning_rate": 0.0005837664002415408,
      "loss": 2.2891,
      "step": 2981
    },
    {
      "epoch": 0.4632233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.0005835183365495749,
      "loss": 2.3823,
      "step": 2982
    },
    {
      "epoch": 0.46337864077669905,
      "grad_norm": 0.0,
      "learning_rate": 0.0005832702517083622,
      "loss": 2.1635,
      "step": 2983
    },
    {
      "epoch": 0.46353398058252426,
      "grad_norm": 0.0,
      "learning_rate": 0.0005830221457807249,
      "loss": 2.2361,
      "step": 2984
    },
    {
      "epoch": 0.46368932038834954,
      "grad_norm": 0.0,
      "learning_rate": 0.000582774018829491,
      "loss": 2.1104,
      "step": 2985
    },
    {
      "epoch": 0.46384466019417475,
      "grad_norm": 0.0,
      "learning_rate": 0.000582525870917493,
      "loss": 2.2811,
      "step": 2986
    },
    {
      "epoch": 0.464,
      "grad_norm": 0.0,
      "learning_rate": 0.0005822777021075693,
      "loss": 2.27,
      "step": 2987
    },
    {
      "epoch": 0.46415533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.0005820295124625635,
      "loss": 2.2184,
      "step": 2988
    },
    {
      "epoch": 0.4643106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0005817813020453237,
      "loss": 2.2605,
      "step": 2989
    },
    {
      "epoch": 0.46446601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.0005815330709187047,
      "loss": 2.2739,
      "step": 2990
    },
    {
      "epoch": 0.46462135922330094,
      "grad_norm": 0.0,
      "learning_rate": 0.0005812848191455653,
      "loss": 2.3425,
      "step": 2991
    },
    {
      "epoch": 0.4647766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0005810365467887701,
      "loss": 2.2234,
      "step": 2992
    },
    {
      "epoch": 0.46493203883495143,
      "grad_norm": 0.0,
      "learning_rate": 0.0005807882539111891,
      "loss": 2.3322,
      "step": 2993
    },
    {
      "epoch": 0.4650873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0005805399405756967,
      "loss": 2.2504,
      "step": 2994
    },
    {
      "epoch": 0.4652427184466019,
      "grad_norm": 0.0,
      "learning_rate": 0.0005802916068451733,
      "loss": 2.2932,
      "step": 2995
    },
    {
      "epoch": 0.4653980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0005800432527825042,
      "loss": 2.3269,
      "step": 2996
    },
    {
      "epoch": 0.4655533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.0005797948784505795,
      "loss": 2.1567,
      "step": 2997
    },
    {
      "epoch": 0.4657087378640777,
      "grad_norm": 0.0,
      "learning_rate": 0.000579546483912295,
      "loss": 2.2208,
      "step": 2998
    },
    {
      "epoch": 0.4658640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.0005792980692305513,
      "loss": 2.2787,
      "step": 2999
    },
    {
      "epoch": 0.46601941747572817,
      "grad_norm": 0.0,
      "learning_rate": 0.000579049634468254,
      "loss": 2.2961,
      "step": 3000
    },
    {
      "epoch": 0.4661747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0005788011796883141,
      "loss": 2.321,
      "step": 3001
    },
    {
      "epoch": 0.46633009708737866,
      "grad_norm": 0.0,
      "learning_rate": 0.0005785527049536474,
      "loss": 2.3279,
      "step": 3002
    },
    {
      "epoch": 0.4664854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.0005783042103271748,
      "loss": 2.3064,
      "step": 3003
    },
    {
      "epoch": 0.46664077669902915,
      "grad_norm": 0.0,
      "learning_rate": 0.0005780556958718225,
      "loss": 2.2651,
      "step": 3004
    },
    {
      "epoch": 0.46679611650485436,
      "grad_norm": 0.0,
      "learning_rate": 0.0005778071616505215,
      "loss": 2.224,
      "step": 3005
    },
    {
      "epoch": 0.46695145631067964,
      "grad_norm": 0.0,
      "learning_rate": 0.0005775586077262076,
      "loss": 2.1825,
      "step": 3006
    },
    {
      "epoch": 0.46710679611650485,
      "grad_norm": 0.0,
      "learning_rate": 0.0005773100341618218,
      "loss": 2.321,
      "step": 3007
    },
    {
      "epoch": 0.4672621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0005770614410203104,
      "loss": 2.2245,
      "step": 3008
    },
    {
      "epoch": 0.46741747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0005768128283646239,
      "loss": 2.1798,
      "step": 3009
    },
    {
      "epoch": 0.46757281553398056,
      "grad_norm": 0.0,
      "learning_rate": 0.0005765641962577186,
      "loss": 2.2663,
      "step": 3010
    },
    {
      "epoch": 0.46772815533980583,
      "grad_norm": 0.0,
      "learning_rate": 0.0005763155447625553,
      "loss": 2.2889,
      "step": 3011
    },
    {
      "epoch": 0.46788349514563105,
      "grad_norm": 0.0,
      "learning_rate": 0.0005760668739420991,
      "loss": 2.309,
      "step": 3012
    },
    {
      "epoch": 0.4680388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0005758181838593213,
      "loss": 2.2028,
      "step": 3013
    },
    {
      "epoch": 0.46819417475728153,
      "grad_norm": 0.0,
      "learning_rate": 0.0005755694745771972,
      "loss": 2.2582,
      "step": 3014
    },
    {
      "epoch": 0.4683495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0005753207461587069,
      "loss": 2.26,
      "step": 3015
    },
    {
      "epoch": 0.468504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.000575071998666836,
      "loss": 2.1916,
      "step": 3016
    },
    {
      "epoch": 0.4686601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.0005748232321645742,
      "loss": 2.2117,
      "step": 3017
    },
    {
      "epoch": 0.4688155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.0005745744467149164,
      "loss": 2.2581,
      "step": 3018
    },
    {
      "epoch": 0.4689708737864078,
      "grad_norm": 0.0,
      "learning_rate": 0.0005743256423808622,
      "loss": 2.1729,
      "step": 3019
    },
    {
      "epoch": 0.469126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0005740768192254162,
      "loss": 2.2367,
      "step": 3020
    },
    {
      "epoch": 0.46928155339805827,
      "grad_norm": 0.0,
      "learning_rate": 0.0005738279773115875,
      "loss": 2.2308,
      "step": 3021
    },
    {
      "epoch": 0.4694368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.00057357911670239,
      "loss": 2.3206,
      "step": 3022
    },
    {
      "epoch": 0.46959223300970876,
      "grad_norm": 0.0,
      "learning_rate": 0.0005733302374608425,
      "loss": 2.2556,
      "step": 3023
    },
    {
      "epoch": 0.469747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0005730813396499681,
      "loss": 2.199,
      "step": 3024
    },
    {
      "epoch": 0.46990291262135925,
      "grad_norm": 0.0,
      "learning_rate": 0.000572832423332795,
      "loss": 2.2366,
      "step": 3025
    },
    {
      "epoch": 0.47005825242718446,
      "grad_norm": 0.0,
      "learning_rate": 0.0005725834885723562,
      "loss": 2.2432,
      "step": 3026
    },
    {
      "epoch": 0.47021359223300974,
      "grad_norm": 0.0,
      "learning_rate": 0.0005723345354316889,
      "loss": 2.2913,
      "step": 3027
    },
    {
      "epoch": 0.47036893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0005720855639738354,
      "loss": 2.2621,
      "step": 3028
    },
    {
      "epoch": 0.47052427184466017,
      "grad_norm": 0.0,
      "learning_rate": 0.0005718365742618422,
      "loss": 2.1702,
      "step": 3029
    },
    {
      "epoch": 0.47067961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.0005715875663587608,
      "loss": 2.1931,
      "step": 3030
    },
    {
      "epoch": 0.47083495145631066,
      "grad_norm": 0.0,
      "learning_rate": 0.0005713385403276473,
      "loss": 2.1456,
      "step": 3031
    },
    {
      "epoch": 0.47099029126213593,
      "grad_norm": 0.0,
      "learning_rate": 0.0005710894962315618,
      "loss": 2.2848,
      "step": 3032
    },
    {
      "epoch": 0.47114563106796115,
      "grad_norm": 0.0,
      "learning_rate": 0.00057084043413357,
      "loss": 2.2391,
      "step": 3033
    },
    {
      "epoch": 0.4713009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0005705913540967414,
      "loss": 2.2322,
      "step": 3034
    },
    {
      "epoch": 0.47145631067961163,
      "grad_norm": 0.0,
      "learning_rate": 0.00057034225618415,
      "loss": 2.2819,
      "step": 3035
    },
    {
      "epoch": 0.4716116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0005700931404588747,
      "loss": 2.1534,
      "step": 3036
    },
    {
      "epoch": 0.4717669902912621,
      "grad_norm": 0.0,
      "learning_rate": 0.000569844006983999,
      "loss": 2.2891,
      "step": 3037
    },
    {
      "epoch": 0.4719223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0005695948558226102,
      "loss": 2.2272,
      "step": 3038
    },
    {
      "epoch": 0.4720776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.0005693456870378011,
      "loss": 2.3085,
      "step": 3039
    },
    {
      "epoch": 0.4722330097087379,
      "grad_norm": 0.0,
      "learning_rate": 0.000569096500692668,
      "loss": 2.2457,
      "step": 3040
    },
    {
      "epoch": 0.4723883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0005688472968503122,
      "loss": 2.2285,
      "step": 3041
    },
    {
      "epoch": 0.47254368932038837,
      "grad_norm": 0.0,
      "learning_rate": 0.0005685980755738391,
      "loss": 2.2256,
      "step": 3042
    },
    {
      "epoch": 0.4726990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.000568348836926359,
      "loss": 2.2693,
      "step": 3043
    },
    {
      "epoch": 0.47285436893203886,
      "grad_norm": 0.0,
      "learning_rate": 0.0005680995809709862,
      "loss": 2.345,
      "step": 3044
    },
    {
      "epoch": 0.4730097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.0005678503077708395,
      "loss": 2.2699,
      "step": 3045
    },
    {
      "epoch": 0.4731650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.0005676010173890417,
      "loss": 2.3471,
      "step": 3046
    },
    {
      "epoch": 0.47332038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.000567351709888721,
      "loss": 2.2904,
      "step": 3047
    },
    {
      "epoch": 0.4734757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0005671023853330087,
      "loss": 2.2932,
      "step": 3048
    },
    {
      "epoch": 0.47363106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0005668530437850409,
      "loss": 2.209,
      "step": 3049
    },
    {
      "epoch": 0.47378640776699027,
      "grad_norm": 0.0,
      "learning_rate": 0.0005666036853079583,
      "loss": 2.2125,
      "step": 3050
    },
    {
      "epoch": 0.47394174757281554,
      "grad_norm": 0.0,
      "learning_rate": 0.0005663543099649057,
      "loss": 2.2767,
      "step": 3051
    },
    {
      "epoch": 0.47409708737864076,
      "grad_norm": 0.0,
      "learning_rate": 0.0005661049178190319,
      "loss": 2.2479,
      "step": 3052
    },
    {
      "epoch": 0.47425242718446603,
      "grad_norm": 0.0,
      "learning_rate": 0.0005658555089334903,
      "loss": 2.2589,
      "step": 3053
    },
    {
      "epoch": 0.47440776699029125,
      "grad_norm": 0.0,
      "learning_rate": 0.0005656060833714384,
      "loss": 2.2167,
      "step": 3054
    },
    {
      "epoch": 0.4745631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0005653566411960378,
      "loss": 2.2493,
      "step": 3055
    },
    {
      "epoch": 0.47471844660194173,
      "grad_norm": 0.0,
      "learning_rate": 0.0005651071824704547,
      "loss": 2.189,
      "step": 3056
    },
    {
      "epoch": 0.474873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.000564857707257859,
      "loss": 2.2236,
      "step": 3057
    },
    {
      "epoch": 0.4750291262135922,
      "grad_norm": 0.0,
      "learning_rate": 0.0005646082156214251,
      "loss": 2.2755,
      "step": 3058
    },
    {
      "epoch": 0.4751844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.0005643587076243314,
      "loss": 2.4014,
      "step": 3059
    },
    {
      "epoch": 0.4753398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.0005641091833297604,
      "loss": 2.1057,
      "step": 3060
    },
    {
      "epoch": 0.475495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0005638596428008992,
      "loss": 2.1671,
      "step": 3061
    },
    {
      "epoch": 0.4756504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0005636100861009384,
      "loss": 2.2448,
      "step": 3062
    },
    {
      "epoch": 0.47580582524271847,
      "grad_norm": 0.0,
      "learning_rate": 0.0005633605132930729,
      "loss": 2.2556,
      "step": 3063
    },
    {
      "epoch": 0.4759611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0005631109244405018,
      "loss": 2.4139,
      "step": 3064
    },
    {
      "epoch": 0.4761165048543689,
      "grad_norm": 0.0,
      "learning_rate": 0.000562861319606428,
      "loss": 2.2538,
      "step": 3065
    },
    {
      "epoch": 0.4762718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.000562611698854059,
      "loss": 2.1931,
      "step": 3066
    },
    {
      "epoch": 0.4764271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.0005623620622466058,
      "loss": 2.2667,
      "step": 3067
    },
    {
      "epoch": 0.47658252427184467,
      "grad_norm": 0.0,
      "learning_rate": 0.0005621124098472835,
      "loss": 2.3324,
      "step": 3068
    },
    {
      "epoch": 0.4767378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0005618627417193112,
      "loss": 2.2214,
      "step": 3069
    },
    {
      "epoch": 0.47689320388349515,
      "grad_norm": 0.0,
      "learning_rate": 0.0005616130579259121,
      "loss": 2.275,
      "step": 3070
    },
    {
      "epoch": 0.47704854368932037,
      "grad_norm": 0.0,
      "learning_rate": 0.0005613633585303135,
      "loss": 2.2792,
      "step": 3071
    },
    {
      "epoch": 0.47720388349514564,
      "grad_norm": 0.0,
      "learning_rate": 0.0005611136435957464,
      "loss": 2.2227,
      "step": 3072
    },
    {
      "epoch": 0.47735922330097086,
      "grad_norm": 0.0,
      "learning_rate": 0.0005608639131854458,
      "loss": 2.1815,
      "step": 3073
    },
    {
      "epoch": 0.47751456310679613,
      "grad_norm": 0.0,
      "learning_rate": 0.0005606141673626503,
      "loss": 2.231,
      "step": 3074
    },
    {
      "epoch": 0.47766990291262135,
      "grad_norm": 0.0,
      "learning_rate": 0.0005603644061906031,
      "loss": 2.3604,
      "step": 3075
    },
    {
      "epoch": 0.4778252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0005601146297325511,
      "loss": 2.1879,
      "step": 3076
    },
    {
      "epoch": 0.47798058252427184,
      "grad_norm": 0.0,
      "learning_rate": 0.0005598648380517443,
      "loss": 2.2018,
      "step": 3077
    },
    {
      "epoch": 0.4781359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0005596150312114377,
      "loss": 2.3086,
      "step": 3078
    },
    {
      "epoch": 0.4782912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.0005593652092748891,
      "loss": 2.2659,
      "step": 3079
    },
    {
      "epoch": 0.4784466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.000559115372305361,
      "loss": 2.2888,
      "step": 3080
    },
    {
      "epoch": 0.4786019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.0005588655203661193,
      "loss": 2.2858,
      "step": 3081
    },
    {
      "epoch": 0.4787572815533981,
      "grad_norm": 0.0,
      "learning_rate": 0.0005586156535204333,
      "loss": 2.2378,
      "step": 3082
    },
    {
      "epoch": 0.4789126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0005583657718315769,
      "loss": 2.2891,
      "step": 3083
    },
    {
      "epoch": 0.4790679611650485,
      "grad_norm": 0.0,
      "learning_rate": 0.000558115875362827,
      "loss": 2.288,
      "step": 3084
    },
    {
      "epoch": 0.4792233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.000557865964177465,
      "loss": 2.3199,
      "step": 3085
    },
    {
      "epoch": 0.479378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0005576160383387754,
      "loss": 2.3097,
      "step": 3086
    },
    {
      "epoch": 0.4795339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0005573660979100464,
      "loss": 2.3448,
      "step": 3087
    },
    {
      "epoch": 0.4796893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0005571161429545704,
      "loss": 2.2656,
      "step": 3088
    },
    {
      "epoch": 0.47984466019417477,
      "grad_norm": 0.0,
      "learning_rate": 0.0005568661735356428,
      "loss": 2.2672,
      "step": 3089
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.0,
      "learning_rate": 0.0005566161897165637,
      "loss": 2.2258,
      "step": 3090
    },
    {
      "epoch": 0.48015533980582525,
      "grad_norm": 0.0,
      "learning_rate": 0.0005563661915606357,
      "loss": 2.2479,
      "step": 3091
    },
    {
      "epoch": 0.48031067961165047,
      "grad_norm": 0.0,
      "learning_rate": 0.0005561161791311657,
      "loss": 2.2903,
      "step": 3092
    },
    {
      "epoch": 0.48046601941747574,
      "grad_norm": 0.0,
      "learning_rate": 0.0005558661524914638,
      "loss": 2.2456,
      "step": 3093
    },
    {
      "epoch": 0.48062135922330096,
      "grad_norm": 0.0,
      "learning_rate": 0.0005556161117048443,
      "loss": 2.1824,
      "step": 3094
    },
    {
      "epoch": 0.48077669902912623,
      "grad_norm": 0.0,
      "learning_rate": 0.0005553660568346245,
      "loss": 2.2282,
      "step": 3095
    },
    {
      "epoch": 0.48093203883495145,
      "grad_norm": 0.0,
      "learning_rate": 0.0005551159879441257,
      "loss": 2.2695,
      "step": 3096
    },
    {
      "epoch": 0.4810873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0005548659050966723,
      "loss": 2.4572,
      "step": 3097
    },
    {
      "epoch": 0.48124271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.0005546158083555925,
      "loss": 2.1592,
      "step": 3098
    },
    {
      "epoch": 0.4813980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0005543656977842181,
      "loss": 2.273,
      "step": 3099
    },
    {
      "epoch": 0.4815533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.0005541155734458843,
      "loss": 2.3225,
      "step": 3100
    },
    {
      "epoch": 0.4817087378640777,
      "grad_norm": 0.0,
      "learning_rate": 0.0005538654354039298,
      "loss": 2.1962,
      "step": 3101
    },
    {
      "epoch": 0.4818640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.0005536152837216966,
      "loss": 2.2594,
      "step": 3102
    },
    {
      "epoch": 0.48201941747572813,
      "grad_norm": 0.0,
      "learning_rate": 0.0005533651184625304,
      "loss": 2.2967,
      "step": 3103
    },
    {
      "epoch": 0.4821747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0005531149396897803,
      "loss": 2.3209,
      "step": 3104
    },
    {
      "epoch": 0.4823300970873786,
      "grad_norm": 0.0,
      "learning_rate": 0.0005528647474667986,
      "loss": 2.2122,
      "step": 3105
    },
    {
      "epoch": 0.4824854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.0005526145418569413,
      "loss": 2.2246,
      "step": 3106
    },
    {
      "epoch": 0.4826407766990291,
      "grad_norm": 0.0,
      "learning_rate": 0.0005523643229235676,
      "loss": 2.2773,
      "step": 3107
    },
    {
      "epoch": 0.4827961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.0005521140907300404,
      "loss": 2.2255,
      "step": 3108
    },
    {
      "epoch": 0.4829514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0005518638453397253,
      "loss": 2.2242,
      "step": 3109
    },
    {
      "epoch": 0.48310679611650487,
      "grad_norm": 0.0,
      "learning_rate": 0.0005516135868159918,
      "loss": 2.2543,
      "step": 3110
    },
    {
      "epoch": 0.4832621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0005513633152222126,
      "loss": 2.1997,
      "step": 3111
    },
    {
      "epoch": 0.48341747572815535,
      "grad_norm": 0.0,
      "learning_rate": 0.0005511130306217637,
      "loss": 2.2883,
      "step": 3112
    },
    {
      "epoch": 0.48357281553398057,
      "grad_norm": 0.0,
      "learning_rate": 0.0005508627330780242,
      "loss": 2.2296,
      "step": 3113
    },
    {
      "epoch": 0.48372815533980584,
      "grad_norm": 0.0,
      "learning_rate": 0.0005506124226543768,
      "loss": 2.1595,
      "step": 3114
    },
    {
      "epoch": 0.48388349514563106,
      "grad_norm": 0.0,
      "learning_rate": 0.0005503620994142073,
      "loss": 2.2062,
      "step": 3115
    },
    {
      "epoch": 0.48403883495145633,
      "grad_norm": 0.0,
      "learning_rate": 0.0005501117634209047,
      "loss": 2.2218,
      "step": 3116
    },
    {
      "epoch": 0.48419417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.000549861414737861,
      "loss": 2.2157,
      "step": 3117
    },
    {
      "epoch": 0.4843495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0005496110534284722,
      "loss": 2.2642,
      "step": 3118
    },
    {
      "epoch": 0.48450485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0005493606795561369,
      "loss": 2.3143,
      "step": 3119
    },
    {
      "epoch": 0.4846601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.0005491102931842566,
      "loss": 2.1932,
      "step": 3120
    },
    {
      "epoch": 0.4848155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.0005488598943762368,
      "loss": 2.2408,
      "step": 3121
    },
    {
      "epoch": 0.48497087378640774,
      "grad_norm": 0.0,
      "learning_rate": 0.0005486094831954851,
      "loss": 2.2039,
      "step": 3122
    },
    {
      "epoch": 0.485126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0005483590597054134,
      "loss": 2.2207,
      "step": 3123
    },
    {
      "epoch": 0.48528155339805823,
      "grad_norm": 0.0,
      "learning_rate": 0.0005481086239694359,
      "loss": 2.2048,
      "step": 3124
    },
    {
      "epoch": 0.4854368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0005478581760509702,
      "loss": 2.2681,
      "step": 3125
    },
    {
      "epoch": 0.4855922330097087,
      "grad_norm": 0.0,
      "learning_rate": 0.0005476077160134367,
      "loss": 2.1966,
      "step": 3126
    },
    {
      "epoch": 0.485747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0005473572439202593,
      "loss": 2.2788,
      "step": 3127
    },
    {
      "epoch": 0.4859029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.0005471067598348647,
      "loss": 2.24,
      "step": 3128
    },
    {
      "epoch": 0.4860582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.0005468562638206826,
      "loss": 2.2524,
      "step": 3129
    },
    {
      "epoch": 0.4862135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0005466057559411461,
      "loss": 2.2704,
      "step": 3130
    },
    {
      "epoch": 0.48636893203883497,
      "grad_norm": 0.0,
      "learning_rate": 0.0005463552362596906,
      "loss": 2.27,
      "step": 3131
    },
    {
      "epoch": 0.4865242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0005461047048397552,
      "loss": 2.2921,
      "step": 3132
    },
    {
      "epoch": 0.48667961165048546,
      "grad_norm": 0.0,
      "learning_rate": 0.0005458541617447814,
      "loss": 2.2437,
      "step": 3133
    },
    {
      "epoch": 0.48683495145631067,
      "grad_norm": 0.0,
      "learning_rate": 0.000545603607038214,
      "loss": 2.2932,
      "step": 3134
    },
    {
      "epoch": 0.48699029126213594,
      "grad_norm": 0.0,
      "learning_rate": 0.000545353040783501,
      "loss": 2.2953,
      "step": 3135
    },
    {
      "epoch": 0.48714563106796116,
      "grad_norm": 0.0,
      "learning_rate": 0.0005451024630440927,
      "loss": 2.3195,
      "step": 3136
    },
    {
      "epoch": 0.48730097087378643,
      "grad_norm": 0.0,
      "learning_rate": 0.0005448518738834423,
      "loss": 2.3467,
      "step": 3137
    },
    {
      "epoch": 0.48745631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0005446012733650068,
      "loss": 2.3081,
      "step": 3138
    },
    {
      "epoch": 0.4876116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.000544350661552245,
      "loss": 2.2231,
      "step": 3139
    },
    {
      "epoch": 0.48776699029126214,
      "grad_norm": 0.0,
      "learning_rate": 0.0005441000385086192,
      "loss": 2.2666,
      "step": 3140
    },
    {
      "epoch": 0.48792233009708735,
      "grad_norm": 0.0,
      "learning_rate": 0.0005438494042975943,
      "loss": 2.2378,
      "step": 3141
    },
    {
      "epoch": 0.4880776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.0005435987589826384,
      "loss": 2.2043,
      "step": 3142
    },
    {
      "epoch": 0.48823300970873784,
      "grad_norm": 0.0,
      "learning_rate": 0.0005433481026272216,
      "loss": 2.3525,
      "step": 3143
    },
    {
      "epoch": 0.4883883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0005430974352948179,
      "loss": 2.3323,
      "step": 3144
    },
    {
      "epoch": 0.48854368932038833,
      "grad_norm": 0.0,
      "learning_rate": 0.0005428467570489027,
      "loss": 2.321,
      "step": 3145
    },
    {
      "epoch": 0.4886990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0005425960679529556,
      "loss": 2.2257,
      "step": 3146
    },
    {
      "epoch": 0.4888543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.0005423453680704581,
      "loss": 2.2358,
      "step": 3147
    },
    {
      "epoch": 0.4890097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.0005420946574648947,
      "loss": 2.2013,
      "step": 3148
    },
    {
      "epoch": 0.4891650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.0005418439361997523,
      "loss": 2.3258,
      "step": 3149
    },
    {
      "epoch": 0.4893203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.0005415932043385209,
      "loss": 2.1475,
      "step": 3150
    },
    {
      "epoch": 0.4894757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0005413424619446931,
      "loss": 2.3322,
      "step": 3151
    },
    {
      "epoch": 0.48963106796116507,
      "grad_norm": 0.0,
      "learning_rate": 0.000541091709081764,
      "loss": 2.3365,
      "step": 3152
    },
    {
      "epoch": 0.4897864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0005408409458132316,
      "loss": 2.2968,
      "step": 3153
    },
    {
      "epoch": 0.48994174757281556,
      "grad_norm": 0.0,
      "learning_rate": 0.000540590172202596,
      "loss": 2.2593,
      "step": 3154
    },
    {
      "epoch": 0.49009708737864077,
      "grad_norm": 0.0,
      "learning_rate": 0.0005403393883133604,
      "loss": 2.1156,
      "step": 3155
    },
    {
      "epoch": 0.49025242718446604,
      "grad_norm": 0.0,
      "learning_rate": 0.0005400885942090309,
      "loss": 2.2631,
      "step": 3156
    },
    {
      "epoch": 0.49040776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.0005398377899531154,
      "loss": 2.1456,
      "step": 3157
    },
    {
      "epoch": 0.49056310679611653,
      "grad_norm": 0.0,
      "learning_rate": 0.000539586975609125,
      "loss": 2.2853,
      "step": 3158
    },
    {
      "epoch": 0.49071844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.0005393361512405728,
      "loss": 2.2793,
      "step": 3159
    },
    {
      "epoch": 0.49087378640776697,
      "grad_norm": 0.0,
      "learning_rate": 0.0005390853169109749,
      "loss": 2.2802,
      "step": 3160
    },
    {
      "epoch": 0.49102912621359224,
      "grad_norm": 0.0,
      "learning_rate": 0.00053883447268385,
      "loss": 2.2169,
      "step": 3161
    },
    {
      "epoch": 0.49118446601941745,
      "grad_norm": 0.0,
      "learning_rate": 0.0005385836186227187,
      "loss": 2.2233,
      "step": 3162
    },
    {
      "epoch": 0.4913398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.0005383327547911048,
      "loss": 2.2691,
      "step": 3163
    },
    {
      "epoch": 0.49149514563106794,
      "grad_norm": 0.0,
      "learning_rate": 0.000538081881252534,
      "loss": 2.2244,
      "step": 3164
    },
    {
      "epoch": 0.4916504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0005378309980705348,
      "loss": 2.2434,
      "step": 3165
    },
    {
      "epoch": 0.49180582524271843,
      "grad_norm": 0.0,
      "learning_rate": 0.000537580105308638,
      "loss": 2.181,
      "step": 3166
    },
    {
      "epoch": 0.4919611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0005373292030303769,
      "loss": 2.2335,
      "step": 3167
    },
    {
      "epoch": 0.4921165048543689,
      "grad_norm": 0.0,
      "learning_rate": 0.0005370782912992872,
      "loss": 2.2783,
      "step": 3168
    },
    {
      "epoch": 0.4922718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.000536827370178907,
      "loss": 2.3436,
      "step": 3169
    },
    {
      "epoch": 0.4924271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.0005365764397327766,
      "loss": 2.1902,
      "step": 3170
    },
    {
      "epoch": 0.4925825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.000536325500024439,
      "loss": 2.1426,
      "step": 3171
    },
    {
      "epoch": 0.4927378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0005360745511174393,
      "loss": 2.2383,
      "step": 3172
    },
    {
      "epoch": 0.49289320388349517,
      "grad_norm": 0.0,
      "learning_rate": 0.0005358235930753248,
      "loss": 2.198,
      "step": 3173
    },
    {
      "epoch": 0.4930485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0005355726259616455,
      "loss": 2.3073,
      "step": 3174
    },
    {
      "epoch": 0.49320388349514566,
      "grad_norm": 0.0,
      "learning_rate": 0.0005353216498399535,
      "loss": 2.2282,
      "step": 3175
    },
    {
      "epoch": 0.4933592233009709,
      "grad_norm": 0.0,
      "learning_rate": 0.0005350706647738032,
      "loss": 2.1415,
      "step": 3176
    },
    {
      "epoch": 0.4935145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.0005348196708267511,
      "loss": 2.2567,
      "step": 3177
    },
    {
      "epoch": 0.49366990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.000534568668062356,
      "loss": 2.2537,
      "step": 3178
    },
    {
      "epoch": 0.4938252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0005343176565441793,
      "loss": 2.2628,
      "step": 3179
    },
    {
      "epoch": 0.49398058252427185,
      "grad_norm": 0.0,
      "learning_rate": 0.0005340666363357842,
      "loss": 2.2729,
      "step": 3180
    },
    {
      "epoch": 0.49413592233009707,
      "grad_norm": 0.0,
      "learning_rate": 0.0005338156075007364,
      "loss": 2.2427,
      "step": 3181
    },
    {
      "epoch": 0.49429126213592234,
      "grad_norm": 0.0,
      "learning_rate": 0.0005335645701026035,
      "loss": 2.2544,
      "step": 3182
    },
    {
      "epoch": 0.49444660194174755,
      "grad_norm": 0.0,
      "learning_rate": 0.0005333135242049553,
      "loss": 2.2764,
      "step": 3183
    },
    {
      "epoch": 0.4946019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.0005330624698713639,
      "loss": 2.2179,
      "step": 3184
    },
    {
      "epoch": 0.49475728155339804,
      "grad_norm": 0.0,
      "learning_rate": 0.0005328114071654035,
      "loss": 2.1697,
      "step": 3185
    },
    {
      "epoch": 0.4949126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0005325603361506506,
      "loss": 2.2449,
      "step": 3186
    },
    {
      "epoch": 0.49506796116504853,
      "grad_norm": 0.0,
      "learning_rate": 0.0005323092568906833,
      "loss": 2.1951,
      "step": 3187
    },
    {
      "epoch": 0.4952233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.0005320581694490822,
      "loss": 2.2572,
      "step": 3188
    },
    {
      "epoch": 0.495378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.00053180707388943,
      "loss": 2.2478,
      "step": 3189
    },
    {
      "epoch": 0.4955339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0005315559702753111,
      "loss": 2.2061,
      "step": 3190
    },
    {
      "epoch": 0.4956893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0005313048586703123,
      "loss": 2.1783,
      "step": 3191
    },
    {
      "epoch": 0.4958446601941748,
      "grad_norm": 0.0,
      "learning_rate": 0.0005310537391380224,
      "loss": 2.1984,
      "step": 3192
    },
    {
      "epoch": 0.496,
      "grad_norm": 0.0,
      "learning_rate": 0.0005308026117420318,
      "loss": 2.2393,
      "step": 3193
    },
    {
      "epoch": 0.49615533980582527,
      "grad_norm": 0.0,
      "learning_rate": 0.0005305514765459333,
      "loss": 2.2329,
      "step": 3194
    },
    {
      "epoch": 0.4963106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0005303003336133218,
      "loss": 2.2776,
      "step": 3195
    },
    {
      "epoch": 0.4964660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.0005300491830077936,
      "loss": 2.3044,
      "step": 3196
    },
    {
      "epoch": 0.496621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0005297980247929474,
      "loss": 2.2596,
      "step": 3197
    },
    {
      "epoch": 0.4967766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0005295468590323835,
      "loss": 2.1319,
      "step": 3198
    },
    {
      "epoch": 0.49693203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.0005292956857897048,
      "loss": 2.2396,
      "step": 3199
    },
    {
      "epoch": 0.4970873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0005290445051285151,
      "loss": 2.2489,
      "step": 3200
    },
    {
      "epoch": 0.49724271844660195,
      "grad_norm": 0.0,
      "learning_rate": 0.0005287933171124207,
      "loss": 2.1796,
      "step": 3201
    },
    {
      "epoch": 0.49739805825242717,
      "grad_norm": 0.0,
      "learning_rate": 0.0005285421218050296,
      "loss": 2.2332,
      "step": 3202
    },
    {
      "epoch": 0.49755339805825244,
      "grad_norm": 0.0,
      "learning_rate": 0.0005282909192699519,
      "loss": 2.2545,
      "step": 3203
    },
    {
      "epoch": 0.49770873786407765,
      "grad_norm": 0.0,
      "learning_rate": 0.000528039709570799,
      "loss": 2.2273,
      "step": 3204
    },
    {
      "epoch": 0.4978640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.0005277884927711849,
      "loss": 2.2209,
      "step": 3205
    },
    {
      "epoch": 0.49801941747572814,
      "grad_norm": 0.0,
      "learning_rate": 0.0005275372689347244,
      "loss": 2.1897,
      "step": 3206
    },
    {
      "epoch": 0.4981747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0005272860381250349,
      "loss": 2.0967,
      "step": 3207
    },
    {
      "epoch": 0.49833009708737863,
      "grad_norm": 0.0,
      "learning_rate": 0.0005270348004057351,
      "loss": 2.2437,
      "step": 3208
    },
    {
      "epoch": 0.4984854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.0005267835558404458,
      "loss": 2.1705,
      "step": 3209
    },
    {
      "epoch": 0.4986407766990291,
      "grad_norm": 0.0,
      "learning_rate": 0.0005265323044927893,
      "loss": 2.1346,
      "step": 3210
    },
    {
      "epoch": 0.4987961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.0005262810464263895,
      "loss": 2.2144,
      "step": 3211
    },
    {
      "epoch": 0.4989514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0005260297817048725,
      "loss": 2.2114,
      "step": 3212
    },
    {
      "epoch": 0.4991067961165049,
      "grad_norm": 0.0,
      "learning_rate": 0.0005257785103918655,
      "loss": 2.2244,
      "step": 3213
    },
    {
      "epoch": 0.4992621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0005255272325509977,
      "loss": 2.2629,
      "step": 3214
    },
    {
      "epoch": 0.4994174757281553,
      "grad_norm": 0.0,
      "learning_rate": 0.0005252759482458998,
      "loss": 2.1704,
      "step": 3215
    },
    {
      "epoch": 0.4995728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.0005250246575402045,
      "loss": 2.2121,
      "step": 3216
    },
    {
      "epoch": 0.4997281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.0005247733604975455,
      "loss": 2.2595,
      "step": 3217
    },
    {
      "epoch": 0.4998834951456311,
      "grad_norm": 0.0,
      "learning_rate": 0.0005245220571815587,
      "loss": 2.2425,
      "step": 3218
    },
    {
      "epoch": 0.5000388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0005242707476558811,
      "loss": 2.3718,
      "step": 3219
    },
    {
      "epoch": 0.5001941747572816,
      "grad_norm": 0.0,
      "learning_rate": 0.0005240194319841517,
      "loss": 2.3263,
      "step": 3220
    },
    {
      "epoch": 0.5003495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0005237681102300109,
      "loss": 2.2358,
      "step": 3221
    },
    {
      "epoch": 0.500504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0005235167824571002,
      "loss": 2.1027,
      "step": 3222
    },
    {
      "epoch": 0.5006601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.0005232654487290635,
      "loss": 2.2776,
      "step": 3223
    },
    {
      "epoch": 0.5008155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.0005230141091095456,
      "loss": 2.3438,
      "step": 3224
    },
    {
      "epoch": 0.5009708737864078,
      "grad_norm": 0.0,
      "learning_rate": 0.0005227627636621928,
      "loss": 2.2069,
      "step": 3225
    },
    {
      "epoch": 0.501126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0005225114124506531,
      "loss": 2.2384,
      "step": 3226
    },
    {
      "epoch": 0.5012815533980582,
      "grad_norm": 0.0,
      "learning_rate": 0.0005222600555385757,
      "loss": 2.3568,
      "step": 3227
    },
    {
      "epoch": 0.5014368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0005220086929896119,
      "loss": 2.1597,
      "step": 3228
    },
    {
      "epoch": 0.5015922330097088,
      "grad_norm": 0.0,
      "learning_rate": 0.0005217573248674133,
      "loss": 2.2915,
      "step": 3229
    },
    {
      "epoch": 0.501747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0005215059512356339,
      "loss": 2.2512,
      "step": 3230
    },
    {
      "epoch": 0.5019029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.0005212545721579285,
      "loss": 2.2642,
      "step": 3231
    },
    {
      "epoch": 0.5020582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.0005210031876979539,
      "loss": 2.2484,
      "step": 3232
    },
    {
      "epoch": 0.5022135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0005207517979193677,
      "loss": 2.2159,
      "step": 3233
    },
    {
      "epoch": 0.5023689320388349,
      "grad_norm": 0.0,
      "learning_rate": 0.0005205004028858288,
      "loss": 2.2584,
      "step": 3234
    },
    {
      "epoch": 0.5025242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0005202490026609978,
      "loss": 2.2289,
      "step": 3235
    },
    {
      "epoch": 0.5026796116504855,
      "grad_norm": 0.0,
      "learning_rate": 0.0005199975973085366,
      "loss": 2.2932,
      "step": 3236
    },
    {
      "epoch": 0.5028349514563106,
      "grad_norm": 0.0,
      "learning_rate": 0.0005197461868921083,
      "loss": 2.2767,
      "step": 3237
    },
    {
      "epoch": 0.5029902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.0005194947714753771,
      "loss": 2.2332,
      "step": 3238
    },
    {
      "epoch": 0.5031456310679612,
      "grad_norm": 0.0,
      "learning_rate": 0.0005192433511220086,
      "loss": 2.2032,
      "step": 3239
    },
    {
      "epoch": 0.5033009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0005189919258956697,
      "loss": 2.2578,
      "step": 3240
    },
    {
      "epoch": 0.5034563106796116,
      "grad_norm": 0.0,
      "learning_rate": 0.0005187404958600286,
      "loss": 2.1642,
      "step": 3241
    },
    {
      "epoch": 0.5036116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0005184890610787545,
      "loss": 2.2631,
      "step": 3242
    },
    {
      "epoch": 0.5037669902912622,
      "grad_norm": 0.0,
      "learning_rate": 0.000518237621615518,
      "loss": 2.3091,
      "step": 3243
    },
    {
      "epoch": 0.5039223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0005179861775339908,
      "loss": 2.3268,
      "step": 3244
    },
    {
      "epoch": 0.5040776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.0005177347288978458,
      "loss": 2.2554,
      "step": 3245
    },
    {
      "epoch": 0.5042330097087379,
      "grad_norm": 0.0,
      "learning_rate": 0.0005174832757707569,
      "loss": 2.2037,
      "step": 3246
    },
    {
      "epoch": 0.5043883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0005172318182163995,
      "loss": 2.087,
      "step": 3247
    },
    {
      "epoch": 0.5045436893203884,
      "grad_norm": 0.0,
      "learning_rate": 0.0005169803562984497,
      "loss": 2.2181,
      "step": 3248
    },
    {
      "epoch": 0.5046990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.000516728890080585,
      "loss": 2.2935,
      "step": 3249
    },
    {
      "epoch": 0.5048543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.0005164774196264837,
      "loss": 2.2179,
      "step": 3250
    },
    {
      "epoch": 0.5050097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.0005162259449998257,
      "loss": 2.2383,
      "step": 3251
    },
    {
      "epoch": 0.5051650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.0005159744662642913,
      "loss": 2.2125,
      "step": 3252
    },
    {
      "epoch": 0.5053203883495145,
      "grad_norm": 0.0,
      "learning_rate": 0.0005157229834835622,
      "loss": 2.2781,
      "step": 3253
    },
    {
      "epoch": 0.5054757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0005154714967213212,
      "loss": 2.2086,
      "step": 3254
    },
    {
      "epoch": 0.5056310679611651,
      "grad_norm": 0.0,
      "learning_rate": 0.0005152200060412519,
      "loss": 2.3181,
      "step": 3255
    },
    {
      "epoch": 0.5057864077669902,
      "grad_norm": 0.0,
      "learning_rate": 0.0005149685115070391,
      "loss": 2.1634,
      "step": 3256
    },
    {
      "epoch": 0.5059417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.0005147170131823683,
      "loss": 2.2136,
      "step": 3257
    },
    {
      "epoch": 0.5060970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.0005144655111309261,
      "loss": 2.2245,
      "step": 3258
    },
    {
      "epoch": 0.5062524271844661,
      "grad_norm": 0.0,
      "learning_rate": 0.0005142140054164001,
      "loss": 2.2396,
      "step": 3259
    },
    {
      "epoch": 0.5064077669902912,
      "grad_norm": 0.0,
      "learning_rate": 0.0005139624961024789,
      "loss": 2.2621,
      "step": 3260
    },
    {
      "epoch": 0.5065631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0005137109832528518,
      "loss": 2.1723,
      "step": 3261
    },
    {
      "epoch": 0.5067184466019418,
      "grad_norm": 0.0,
      "learning_rate": 0.0005134594669312093,
      "loss": 2.1721,
      "step": 3262
    },
    {
      "epoch": 0.506873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0005132079472012421,
      "loss": 2.2518,
      "step": 3263
    },
    {
      "epoch": 0.5070291262135922,
      "grad_norm": 0.0,
      "learning_rate": 0.0005129564241266426,
      "loss": 2.2571,
      "step": 3264
    },
    {
      "epoch": 0.5071844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.0005127048977711038,
      "loss": 2.1675,
      "step": 3265
    },
    {
      "epoch": 0.5073398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.0005124533681983189,
      "loss": 2.2168,
      "step": 3266
    },
    {
      "epoch": 0.507495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.000512201835471983,
      "loss": 2.1914,
      "step": 3267
    },
    {
      "epoch": 0.5076504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.000511950299655791,
      "loss": 2.2139,
      "step": 3268
    },
    {
      "epoch": 0.5078058252427184,
      "grad_norm": 0.0,
      "learning_rate": 0.0005116987608134393,
      "loss": 2.1721,
      "step": 3269
    },
    {
      "epoch": 0.5079611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0005114472190086244,
      "loss": 2.1856,
      "step": 3270
    },
    {
      "epoch": 0.5081165048543689,
      "grad_norm": 0.0,
      "learning_rate": 0.0005111956743050444,
      "loss": 2.2194,
      "step": 3271
    },
    {
      "epoch": 0.5082718446601941,
      "grad_norm": 0.0,
      "learning_rate": 0.0005109441267663973,
      "loss": 2.1602,
      "step": 3272
    },
    {
      "epoch": 0.5084271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.0005106925764563825,
      "loss": 2.1876,
      "step": 3273
    },
    {
      "epoch": 0.5085825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.0005104410234386994,
      "loss": 2.1941,
      "step": 3274
    },
    {
      "epoch": 0.5087378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0005101894677770486,
      "loss": 2.2872,
      "step": 3275
    },
    {
      "epoch": 0.5088932038834951,
      "grad_norm": 0.0,
      "learning_rate": 0.0005099379095351315,
      "loss": 2.2257,
      "step": 3276
    },
    {
      "epoch": 0.5090485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0005096863487766496,
      "loss": 2.1891,
      "step": 3277
    },
    {
      "epoch": 0.5092038834951457,
      "grad_norm": 0.0,
      "learning_rate": 0.0005094347855653053,
      "loss": 2.1998,
      "step": 3278
    },
    {
      "epoch": 0.5093592233009708,
      "grad_norm": 0.0,
      "learning_rate": 0.0005091832199648019,
      "loss": 2.1023,
      "step": 3279
    },
    {
      "epoch": 0.5095145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.0005089316520388427,
      "loss": 2.2631,
      "step": 3280
    },
    {
      "epoch": 0.5096699029126214,
      "grad_norm": 0.0,
      "learning_rate": 0.0005086800818511323,
      "loss": 2.2863,
      "step": 3281
    },
    {
      "epoch": 0.5098252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0005084285094653753,
      "loss": 2.2197,
      "step": 3282
    },
    {
      "epoch": 0.5099805825242718,
      "grad_norm": 0.0,
      "learning_rate": 0.0005081769349452768,
      "loss": 2.2021,
      "step": 3283
    },
    {
      "epoch": 0.5101359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0005079253583545433,
      "loss": 2.265,
      "step": 3284
    },
    {
      "epoch": 0.5102912621359224,
      "grad_norm": 0.0,
      "learning_rate": 0.0005076737797568811,
      "loss": 2.2484,
      "step": 3285
    },
    {
      "epoch": 0.5104466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.000507422199215997,
      "loss": 2.2358,
      "step": 3286
    },
    {
      "epoch": 0.5106019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.0005071706167955981,
      "loss": 2.2427,
      "step": 3287
    },
    {
      "epoch": 0.5107572815533981,
      "grad_norm": 0.0,
      "learning_rate": 0.0005069190325593929,
      "loss": 2.2102,
      "step": 3288
    },
    {
      "epoch": 0.5109126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0005066674465710893,
      "loss": 2.3899,
      "step": 3289
    },
    {
      "epoch": 0.5110679611650485,
      "grad_norm": 0.0,
      "learning_rate": 0.0005064158588943967,
      "loss": 2.299,
      "step": 3290
    },
    {
      "epoch": 0.5112233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.0005061642695930236,
      "loss": 2.2637,
      "step": 3291
    },
    {
      "epoch": 0.511378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0005059126787306801,
      "loss": 2.2244,
      "step": 3292
    },
    {
      "epoch": 0.5115339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0005056610863710762,
      "loss": 2.1777,
      "step": 3293
    },
    {
      "epoch": 0.5116893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0005054094925779223,
      "loss": 2.1152,
      "step": 3294
    },
    {
      "epoch": 0.5118446601941747,
      "grad_norm": 0.0,
      "learning_rate": 0.0005051578974149293,
      "loss": 2.1944,
      "step": 3295
    },
    {
      "epoch": 0.512,
      "grad_norm": 0.0,
      "learning_rate": 0.0005049063009458082,
      "loss": 2.2391,
      "step": 3296
    },
    {
      "epoch": 0.5121553398058253,
      "grad_norm": 0.0,
      "learning_rate": 0.0005046547032342702,
      "loss": 2.2921,
      "step": 3297
    },
    {
      "epoch": 0.5123106796116504,
      "grad_norm": 0.0,
      "learning_rate": 0.0005044031043440276,
      "loss": 2.2524,
      "step": 3298
    },
    {
      "epoch": 0.5124660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.0005041515043387922,
      "loss": 2.201,
      "step": 3299
    },
    {
      "epoch": 0.512621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0005038999032822764,
      "loss": 2.2939,
      "step": 3300
    },
    {
      "epoch": 0.5127766990291263,
      "grad_norm": 0.0,
      "learning_rate": 0.0005036483012381929,
      "loss": 2.3021,
      "step": 3301
    },
    {
      "epoch": 0.5129320388349514,
      "grad_norm": 0.0,
      "learning_rate": 0.0005033966982702545,
      "loss": 2.2195,
      "step": 3302
    },
    {
      "epoch": 0.5130873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0005031450944421743,
      "loss": 2.2888,
      "step": 3303
    },
    {
      "epoch": 0.513242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0005028934898176657,
      "loss": 2.149,
      "step": 3304
    },
    {
      "epoch": 0.5133980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0005026418844604422,
      "loss": 2.1697,
      "step": 3305
    },
    {
      "epoch": 0.5135533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.0005023902784342174,
      "loss": 2.3151,
      "step": 3306
    },
    {
      "epoch": 0.5137087378640777,
      "grad_norm": 0.0,
      "learning_rate": 0.0005021386718027054,
      "loss": 2.2206,
      "step": 3307
    },
    {
      "epoch": 0.5138640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.0005018870646296204,
      "loss": 2.3463,
      "step": 3308
    },
    {
      "epoch": 0.5140194174757281,
      "grad_norm": 0.0,
      "learning_rate": 0.000501635456978676,
      "loss": 2.1121,
      "step": 3309
    },
    {
      "epoch": 0.5141747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0005013838489135868,
      "loss": 2.3015,
      "step": 3310
    },
    {
      "epoch": 0.5143300970873786,
      "grad_norm": 0.0,
      "learning_rate": 0.0005011322404980672,
      "loss": 2.2136,
      "step": 3311
    },
    {
      "epoch": 0.5144854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.0005008806317958319,
      "loss": 2.1138,
      "step": 3312
    },
    {
      "epoch": 0.5146407766990291,
      "grad_norm": 0.0,
      "learning_rate": 0.0005006290228705953,
      "loss": 2.1528,
      "step": 3313
    },
    {
      "epoch": 0.5147961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.0005003774137860719,
      "loss": 2.2631,
      "step": 3314
    },
    {
      "epoch": 0.5149514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0005001258046059765,
      "loss": 2.2278,
      "step": 3315
    },
    {
      "epoch": 0.5151067961165049,
      "grad_norm": 0.0,
      "learning_rate": 0.0004998741953940237,
      "loss": 2.1694,
      "step": 3316
    },
    {
      "epoch": 0.5152621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0004996225862139283,
      "loss": 2.2025,
      "step": 3317
    },
    {
      "epoch": 0.5154174757281553,
      "grad_norm": 0.0,
      "learning_rate": 0.0004993709771294049,
      "loss": 2.2465,
      "step": 3318
    },
    {
      "epoch": 0.5155728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.0004991193682041682,
      "loss": 2.2255,
      "step": 3319
    },
    {
      "epoch": 0.5157281553398059,
      "grad_norm": 0.0,
      "learning_rate": 0.0004988677595019329,
      "loss": 2.2872,
      "step": 3320
    },
    {
      "epoch": 0.515883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0004986161510864132,
      "loss": 2.2188,
      "step": 3321
    },
    {
      "epoch": 0.5160388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0004983645430213241,
      "loss": 2.2684,
      "step": 3322
    },
    {
      "epoch": 0.5161941747572816,
      "grad_norm": 0.0,
      "learning_rate": 0.0004981129353703798,
      "loss": 2.2223,
      "step": 3323
    },
    {
      "epoch": 0.5163495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0004978613281972946,
      "loss": 2.16,
      "step": 3324
    },
    {
      "epoch": 0.516504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0004976097215657826,
      "loss": 2.3357,
      "step": 3325
    },
    {
      "epoch": 0.5166601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.0004973581155395578,
      "loss": 2.205,
      "step": 3326
    },
    {
      "epoch": 0.5168155339805826,
      "grad_norm": 0.0,
      "learning_rate": 0.0004971065101823345,
      "loss": 2.2991,
      "step": 3327
    },
    {
      "epoch": 0.5169708737864077,
      "grad_norm": 0.0,
      "learning_rate": 0.0004968549055578259,
      "loss": 2.2156,
      "step": 3328
    },
    {
      "epoch": 0.517126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0004966033017297457,
      "loss": 2.1985,
      "step": 3329
    },
    {
      "epoch": 0.5172815533980583,
      "grad_norm": 0.0,
      "learning_rate": 0.0004963516987618072,
      "loss": 2.2194,
      "step": 3330
    },
    {
      "epoch": 0.5174368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0004961000967177236,
      "loss": 2.205,
      "step": 3331
    },
    {
      "epoch": 0.5175922330097087,
      "grad_norm": 0.0,
      "learning_rate": 0.0004958484956612078,
      "loss": 2.1796,
      "step": 3332
    },
    {
      "epoch": 0.517747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0004955968956559725,
      "loss": 2.3501,
      "step": 3333
    },
    {
      "epoch": 0.5179029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.0004953452967657298,
      "loss": 2.1969,
      "step": 3334
    },
    {
      "epoch": 0.5180582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.000495093699054192,
      "loss": 2.269,
      "step": 3335
    },
    {
      "epoch": 0.5182135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0004948421025850708,
      "loss": 2.1298,
      "step": 3336
    },
    {
      "epoch": 0.5183689320388349,
      "grad_norm": 0.0,
      "learning_rate": 0.0004945905074220776,
      "loss": 2.197,
      "step": 3337
    },
    {
      "epoch": 0.5185242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0004943389136289238,
      "loss": 2.2902,
      "step": 3338
    },
    {
      "epoch": 0.5186796116504855,
      "grad_norm": 0.0,
      "learning_rate": 0.00049408732126932,
      "loss": 2.3245,
      "step": 3339
    },
    {
      "epoch": 0.5188349514563106,
      "grad_norm": 0.0,
      "learning_rate": 0.0004938357304069765,
      "loss": 2.1351,
      "step": 3340
    },
    {
      "epoch": 0.5189902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.0004935841411056036,
      "loss": 2.2607,
      "step": 3341
    },
    {
      "epoch": 0.5191456310679612,
      "grad_norm": 0.0,
      "learning_rate": 0.0004933325534289107,
      "loss": 2.1862,
      "step": 3342
    },
    {
      "epoch": 0.5193009708737865,
      "grad_norm": 0.0,
      "learning_rate": 0.0004930809674406072,
      "loss": 2.2644,
      "step": 3343
    },
    {
      "epoch": 0.5194563106796116,
      "grad_norm": 0.0,
      "learning_rate": 0.0004928293832044019,
      "loss": 2.217,
      "step": 3344
    },
    {
      "epoch": 0.5196116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0004925778007840031,
      "loss": 2.118,
      "step": 3345
    },
    {
      "epoch": 0.5197669902912622,
      "grad_norm": 0.0,
      "learning_rate": 0.000492326220243119,
      "loss": 2.2015,
      "step": 3346
    },
    {
      "epoch": 0.5199223300970873,
      "grad_norm": 0.0,
      "learning_rate": 0.0004920746416454566,
      "loss": 2.1465,
      "step": 3347
    },
    {
      "epoch": 0.5200776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.0004918230650547231,
      "loss": 2.2606,
      "step": 3348
    },
    {
      "epoch": 0.5202330097087379,
      "grad_norm": 0.0,
      "learning_rate": 0.0004915714905346249,
      "loss": 2.2065,
      "step": 3349
    },
    {
      "epoch": 0.5203883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.000491319918148868,
      "loss": 2.2888,
      "step": 3350
    },
    {
      "epoch": 0.5205436893203883,
      "grad_norm": 0.0,
      "learning_rate": 0.0004910683479611575,
      "loss": 2.2604,
      "step": 3351
    },
    {
      "epoch": 0.5206990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0004908167800351984,
      "loss": 2.2795,
      "step": 3352
    },
    {
      "epoch": 0.5208543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.0004905652144346947,
      "loss": 2.3458,
      "step": 3353
    },
    {
      "epoch": 0.5210097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.0004903136512233506,
      "loss": 2.285,
      "step": 3354
    },
    {
      "epoch": 0.5211650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.0004900620904648687,
      "loss": 2.2639,
      "step": 3355
    },
    {
      "epoch": 0.5213203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.0004898105322229514,
      "loss": 2.2559,
      "step": 3356
    },
    {
      "epoch": 0.5214757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0004895589765613006,
      "loss": 2.2529,
      "step": 3357
    },
    {
      "epoch": 0.5216310679611651,
      "grad_norm": 0.0,
      "learning_rate": 0.0004893074235436176,
      "loss": 2.1571,
      "step": 3358
    },
    {
      "epoch": 0.5217864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0004890558732336026,
      "loss": 2.2272,
      "step": 3359
    },
    {
      "epoch": 0.5219417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.0004888043256949556,
      "loss": 2.2422,
      "step": 3360
    },
    {
      "epoch": 0.5220970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.0004885527809913755,
      "loss": 2.241,
      "step": 3361
    },
    {
      "epoch": 0.5222524271844661,
      "grad_norm": 0.0,
      "learning_rate": 0.000488301239186561,
      "loss": 2.2808,
      "step": 3362
    },
    {
      "epoch": 0.5224077669902912,
      "grad_norm": 0.0,
      "learning_rate": 0.0004880497003442091,
      "loss": 2.2427,
      "step": 3363
    },
    {
      "epoch": 0.5225631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0004877981645280172,
      "loss": 2.1661,
      "step": 3364
    },
    {
      "epoch": 0.5227184466019418,
      "grad_norm": 0.0,
      "learning_rate": 0.0004875466318016811,
      "loss": 2.243,
      "step": 3365
    },
    {
      "epoch": 0.5228737864077669,
      "grad_norm": 0.0,
      "learning_rate": 0.0004872951022288964,
      "loss": 2.2243,
      "step": 3366
    },
    {
      "epoch": 0.5230291262135922,
      "grad_norm": 0.0,
      "learning_rate": 0.0004870435758733574,
      "loss": 2.2113,
      "step": 3367
    },
    {
      "epoch": 0.5231844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.0004867920527987578,
      "loss": 2.1718,
      "step": 3368
    },
    {
      "epoch": 0.5233398058252428,
      "grad_norm": 0.0,
      "learning_rate": 0.00048654053306879076,
      "loss": 2.3271,
      "step": 3369
    },
    {
      "epoch": 0.5234951456310679,
      "grad_norm": 0.0,
      "learning_rate": 0.0004862890167471481,
      "loss": 2.14,
      "step": 3370
    },
    {
      "epoch": 0.5236504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0004860375038975211,
      "loss": 2.2173,
      "step": 3371
    },
    {
      "epoch": 0.5238058252427185,
      "grad_norm": 0.0,
      "learning_rate": 0.00048578599458359986,
      "loss": 2.2266,
      "step": 3372
    },
    {
      "epoch": 0.5239611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0004855344888690741,
      "loss": 2.2026,
      "step": 3373
    },
    {
      "epoch": 0.5241165048543689,
      "grad_norm": 0.0,
      "learning_rate": 0.00048528298681763197,
      "loss": 2.1676,
      "step": 3374
    },
    {
      "epoch": 0.5242718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.0004850314884929611,
      "loss": 2.3263,
      "step": 3375
    },
    {
      "epoch": 0.5244271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.0004847799939587483,
      "loss": 2.3104,
      "step": 3376
    },
    {
      "epoch": 0.5245825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.00048452850327867884,
      "loss": 2.2632,
      "step": 3377
    },
    {
      "epoch": 0.5247378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0004842770165164379,
      "loss": 2.2707,
      "step": 3378
    },
    {
      "epoch": 0.5248932038834951,
      "grad_norm": 0.0,
      "learning_rate": 0.00048402553373570886,
      "loss": 2.3868,
      "step": 3379
    },
    {
      "epoch": 0.5250485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0004837740550001744,
      "loss": 2.2678,
      "step": 3380
    },
    {
      "epoch": 0.5252038834951457,
      "grad_norm": 0.0,
      "learning_rate": 0.00048352258037351635,
      "loss": 2.1498,
      "step": 3381
    },
    {
      "epoch": 0.5253592233009708,
      "grad_norm": 0.0,
      "learning_rate": 0.0004832711099194151,
      "loss": 2.2083,
      "step": 3382
    },
    {
      "epoch": 0.5255145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.0004830196437015503,
      "loss": 2.2306,
      "step": 3383
    },
    {
      "epoch": 0.5256699029126214,
      "grad_norm": 0.0,
      "learning_rate": 0.00048276818178360066,
      "loss": 2.2588,
      "step": 3384
    },
    {
      "epoch": 0.5258252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0004825167242292432,
      "loss": 2.2307,
      "step": 3385
    },
    {
      "epoch": 0.5259805825242718,
      "grad_norm": 0.0,
      "learning_rate": 0.0004822652711021543,
      "loss": 2.2433,
      "step": 3386
    },
    {
      "epoch": 0.5261359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0004820138224660093,
      "loss": 2.3455,
      "step": 3387
    },
    {
      "epoch": 0.5262912621359224,
      "grad_norm": 0.0,
      "learning_rate": 0.00048176237838448206,
      "loss": 2.119,
      "step": 3388
    },
    {
      "epoch": 0.5264466019417475,
      "grad_norm": 0.0,
      "learning_rate": 0.0004815109389212456,
      "loss": 2.1416,
      "step": 3389
    },
    {
      "epoch": 0.5266019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.0004812595041399715,
      "loss": 2.2369,
      "step": 3390
    },
    {
      "epoch": 0.5267572815533981,
      "grad_norm": 0.0,
      "learning_rate": 0.0004810080741043304,
      "loss": 2.2169,
      "step": 3391
    },
    {
      "epoch": 0.5269126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0004807566488779914,
      "loss": 2.2543,
      "step": 3392
    },
    {
      "epoch": 0.5270679611650485,
      "grad_norm": 0.0,
      "learning_rate": 0.000480505228524623,
      "loss": 2.1743,
      "step": 3393
    },
    {
      "epoch": 0.5272233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.00048025381310789166,
      "loss": 2.2629,
      "step": 3394
    },
    {
      "epoch": 0.527378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.00048000240269146333,
      "loss": 2.2275,
      "step": 3395
    },
    {
      "epoch": 0.5275339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0004797509973390022,
      "loss": 2.2618,
      "step": 3396
    },
    {
      "epoch": 0.5276893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.00047949959711417135,
      "loss": 2.2727,
      "step": 3397
    },
    {
      "epoch": 0.5278446601941748,
      "grad_norm": 0.0,
      "learning_rate": 0.00047924820208063254,
      "loss": 2.1581,
      "step": 3398
    },
    {
      "epoch": 0.528,
      "grad_norm": 0.0,
      "learning_rate": 0.00047899681230204624,
      "loss": 2.2644,
      "step": 3399
    },
    {
      "epoch": 0.5281553398058253,
      "grad_norm": 0.0,
      "learning_rate": 0.00047874542784207154,
      "loss": 2.1989,
      "step": 3400
    },
    {
      "epoch": 0.5283106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0004784940487643662,
      "loss": 2.2573,
      "step": 3401
    },
    {
      "epoch": 0.5284660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.0004782426751325868,
      "loss": 2.228,
      "step": 3402
    },
    {
      "epoch": 0.528621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.00047799130701038823,
      "loss": 2.2695,
      "step": 3403
    },
    {
      "epoch": 0.5287766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0004777399444614242,
      "loss": 2.2643,
      "step": 3404
    },
    {
      "epoch": 0.5289320388349514,
      "grad_norm": 0.0,
      "learning_rate": 0.000477488587549347,
      "loss": 2.2457,
      "step": 3405
    },
    {
      "epoch": 0.5290873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0004772372363378072,
      "loss": 2.2716,
      "step": 3406
    },
    {
      "epoch": 0.529242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0004769858908904546,
      "loss": 2.1859,
      "step": 3407
    },
    {
      "epoch": 0.5293980582524271,
      "grad_norm": 0.0,
      "learning_rate": 0.0004767345512709367,
      "loss": 2.1693,
      "step": 3408
    },
    {
      "epoch": 0.5295533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.0004764832175428999,
      "loss": 2.165,
      "step": 3409
    },
    {
      "epoch": 0.5297087378640777,
      "grad_norm": 0.0,
      "learning_rate": 0.0004762318897699893,
      "loss": 2.2324,
      "step": 3410
    },
    {
      "epoch": 0.529864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.00047598056801584837,
      "loss": 2.2782,
      "step": 3411
    },
    {
      "epoch": 0.5300194174757281,
      "grad_norm": 0.0,
      "learning_rate": 0.000475729252344119,
      "loss": 2.2281,
      "step": 3412
    },
    {
      "epoch": 0.5301747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.00047547794281844146,
      "loss": 2.1678,
      "step": 3413
    },
    {
      "epoch": 0.5303300970873787,
      "grad_norm": 0.0,
      "learning_rate": 0.0004752266395024546,
      "loss": 2.2428,
      "step": 3414
    },
    {
      "epoch": 0.5304854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.00047497534245979564,
      "loss": 2.261,
      "step": 3415
    },
    {
      "epoch": 0.5306407766990291,
      "grad_norm": 0.0,
      "learning_rate": 0.0004747240517541002,
      "loss": 2.3699,
      "step": 3416
    },
    {
      "epoch": 0.5307961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.00047447276744900233,
      "loss": 2.184,
      "step": 3417
    },
    {
      "epoch": 0.5309514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.00047422148960813456,
      "loss": 2.2709,
      "step": 3418
    },
    {
      "epoch": 0.5311067961165049,
      "grad_norm": 0.0,
      "learning_rate": 0.0004739702182951277,
      "loss": 2.1511,
      "step": 3419
    },
    {
      "epoch": 0.5312621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.00047371895357361057,
      "loss": 2.2617,
      "step": 3420
    },
    {
      "epoch": 0.5314174757281553,
      "grad_norm": 0.0,
      "learning_rate": 0.00047346769550721084,
      "loss": 2.183,
      "step": 3421
    },
    {
      "epoch": 0.5315728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.00047321644415955434,
      "loss": 2.2034,
      "step": 3422
    },
    {
      "epoch": 0.5317281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.000472965199594265,
      "loss": 2.2183,
      "step": 3423
    },
    {
      "epoch": 0.531883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.00047271396187496524,
      "loss": 2.1781,
      "step": 3424
    },
    {
      "epoch": 0.5320388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0004724627310652756,
      "loss": 2.2657,
      "step": 3425
    },
    {
      "epoch": 0.5321941747572816,
      "grad_norm": 0.0,
      "learning_rate": 0.00047221150722881515,
      "loss": 2.1976,
      "step": 3426
    },
    {
      "epoch": 0.5323495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.00047196029042920087,
      "loss": 2.1608,
      "step": 3427
    },
    {
      "epoch": 0.532504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.00047170908073004814,
      "loss": 2.1229,
      "step": 3428
    },
    {
      "epoch": 0.5326601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.00047145787819497035,
      "loss": 2.2799,
      "step": 3429
    },
    {
      "epoch": 0.5328155339805826,
      "grad_norm": 0.0,
      "learning_rate": 0.0004712066828875795,
      "loss": 2.2509,
      "step": 3430
    },
    {
      "epoch": 0.5329708737864077,
      "grad_norm": 0.0,
      "learning_rate": 0.00047095549487148516,
      "loss": 2.2394,
      "step": 3431
    },
    {
      "epoch": 0.533126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0004707043142102955,
      "loss": 2.2172,
      "step": 3432
    },
    {
      "epoch": 0.5332815533980583,
      "grad_norm": 0.0,
      "learning_rate": 0.0004704531409676166,
      "loss": 2.0959,
      "step": 3433
    },
    {
      "epoch": 0.5334368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0004702019752070527,
      "loss": 2.1253,
      "step": 3434
    },
    {
      "epoch": 0.5335922330097087,
      "grad_norm": 0.0,
      "learning_rate": 0.00046995081699220654,
      "loss": 2.3582,
      "step": 3435
    },
    {
      "epoch": 0.533747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.00046969966638667833,
      "loss": 2.1464,
      "step": 3436
    },
    {
      "epoch": 0.5339029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.00046944852345406674,
      "loss": 2.4277,
      "step": 3437
    },
    {
      "epoch": 0.5340582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.00046919738825796834,
      "loss": 2.2822,
      "step": 3438
    },
    {
      "epoch": 0.5342135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0004689462608619778,
      "loss": 2.2649,
      "step": 3439
    },
    {
      "epoch": 0.534368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0004686951413296877,
      "loss": 2.2255,
      "step": 3440
    },
    {
      "epoch": 0.5345242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.00046844402972468885,
      "loss": 2.3273,
      "step": 3441
    },
    {
      "epoch": 0.5346796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.00046819292611057026,
      "loss": 2.3044,
      "step": 3442
    },
    {
      "epoch": 0.5348349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.0004679418305509179,
      "loss": 2.2198,
      "step": 3443
    },
    {
      "epoch": 0.5349902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.0004676907431093168,
      "loss": 2.1892,
      "step": 3444
    },
    {
      "epoch": 0.5351456310679612,
      "grad_norm": 0.0,
      "learning_rate": 0.00046743966384934956,
      "loss": 2.1602,
      "step": 3445
    },
    {
      "epoch": 0.5353009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0004671885928345965,
      "loss": 2.2463,
      "step": 3446
    },
    {
      "epoch": 0.5354563106796116,
      "grad_norm": 0.0,
      "learning_rate": 0.0004669375301286362,
      "loss": 2.2481,
      "step": 3447
    },
    {
      "epoch": 0.5356116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.00046668647579504487,
      "loss": 2.2559,
      "step": 3448
    },
    {
      "epoch": 0.5357669902912622,
      "grad_norm": 0.0,
      "learning_rate": 0.0004664354298973965,
      "loss": 2.177,
      "step": 3449
    },
    {
      "epoch": 0.5359223300970873,
      "grad_norm": 0.0,
      "learning_rate": 0.0004661843924992636,
      "loss": 2.2836,
      "step": 3450
    },
    {
      "epoch": 0.5360776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.0004659333636642157,
      "loss": 2.2292,
      "step": 3451
    },
    {
      "epoch": 0.5362330097087379,
      "grad_norm": 0.0,
      "learning_rate": 0.0004656823434558207,
      "loss": 2.235,
      "step": 3452
    },
    {
      "epoch": 0.5363883495145632,
      "grad_norm": 0.0,
      "learning_rate": 0.0004654313319376441,
      "loss": 2.2877,
      "step": 3453
    },
    {
      "epoch": 0.5365436893203883,
      "grad_norm": 0.0,
      "learning_rate": 0.00046518032917324914,
      "loss": 2.2269,
      "step": 3454
    },
    {
      "epoch": 0.5366990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.000464929335226197,
      "loss": 2.104,
      "step": 3455
    },
    {
      "epoch": 0.5368543689320389,
      "grad_norm": 0.0,
      "learning_rate": 0.00046467835016004663,
      "loss": 2.1313,
      "step": 3456
    },
    {
      "epoch": 0.5370097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.0004644273740383546,
      "loss": 2.0663,
      "step": 3457
    },
    {
      "epoch": 0.5371650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.0004641764069246752,
      "loss": 2.0814,
      "step": 3458
    },
    {
      "epoch": 0.5373203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.00046392544888256086,
      "loss": 2.2847,
      "step": 3459
    },
    {
      "epoch": 0.5374757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.000463674499975561,
      "loss": 2.2717,
      "step": 3460
    },
    {
      "epoch": 0.537631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.00046342356026722343,
      "loss": 2.2626,
      "step": 3461
    },
    {
      "epoch": 0.5377864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0004631726298210931,
      "loss": 2.2879,
      "step": 3462
    },
    {
      "epoch": 0.5379417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.00046292170870071284,
      "loss": 2.209,
      "step": 3463
    },
    {
      "epoch": 0.5380970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.0004626707969696232,
      "loss": 2.1615,
      "step": 3464
    },
    {
      "epoch": 0.538252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.00046241989469136217,
      "loss": 2.2385,
      "step": 3465
    },
    {
      "epoch": 0.5384077669902912,
      "grad_norm": 0.0,
      "learning_rate": 0.00046216900192946544,
      "loss": 2.2695,
      "step": 3466
    },
    {
      "epoch": 0.5385631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0004619181187474661,
      "loss": 2.2478,
      "step": 3467
    },
    {
      "epoch": 0.5387184466019418,
      "grad_norm": 0.0,
      "learning_rate": 0.0004616672452088953,
      "loss": 2.2068,
      "step": 3468
    },
    {
      "epoch": 0.538873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0004614163813772813,
      "loss": 2.1888,
      "step": 3469
    },
    {
      "epoch": 0.5390291262135922,
      "grad_norm": 0.0,
      "learning_rate": 0.0004611655273161501,
      "loss": 2.1868,
      "step": 3470
    },
    {
      "epoch": 0.5391844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.00046091468308902505,
      "loss": 2.241,
      "step": 3471
    },
    {
      "epoch": 0.5393398058252428,
      "grad_norm": 0.0,
      "learning_rate": 0.0004606638487594273,
      "loss": 2.2062,
      "step": 3472
    },
    {
      "epoch": 0.5394951456310679,
      "grad_norm": 0.0,
      "learning_rate": 0.0004604130243908751,
      "loss": 2.137,
      "step": 3473
    },
    {
      "epoch": 0.5396504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0004601622100468845,
      "loss": 2.1934,
      "step": 3474
    },
    {
      "epoch": 0.5398058252427185,
      "grad_norm": 0.0,
      "learning_rate": 0.0004599114057909691,
      "loss": 2.2234,
      "step": 3475
    },
    {
      "epoch": 0.5399611650485436,
      "grad_norm": 0.0,
      "learning_rate": 0.00045966061168663957,
      "loss": 2.1272,
      "step": 3476
    },
    {
      "epoch": 0.5401165048543689,
      "grad_norm": 0.0,
      "learning_rate": 0.0004594098277974042,
      "loss": 2.2881,
      "step": 3477
    },
    {
      "epoch": 0.5402718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.0004591590541867687,
      "loss": 2.1774,
      "step": 3478
    },
    {
      "epoch": 0.5404271844660195,
      "grad_norm": 0.0,
      "learning_rate": 0.0004589082909182361,
      "loss": 2.283,
      "step": 3479
    },
    {
      "epoch": 0.5405825242718446,
      "grad_norm": 0.0,
      "learning_rate": 0.000458657538055307,
      "loss": 2.3146,
      "step": 3480
    },
    {
      "epoch": 0.5407378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0004584067956614792,
      "loss": 2.0711,
      "step": 3481
    },
    {
      "epoch": 0.5408932038834952,
      "grad_norm": 0.0,
      "learning_rate": 0.00045815606380024773,
      "loss": 2.3097,
      "step": 3482
    },
    {
      "epoch": 0.5410485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0004579053425351054,
      "loss": 2.2203,
      "step": 3483
    },
    {
      "epoch": 0.5412038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.0004576546319295419,
      "loss": 2.2374,
      "step": 3484
    },
    {
      "epoch": 0.5413592233009709,
      "grad_norm": 0.0,
      "learning_rate": 0.0004574039320470444,
      "loss": 2.2976,
      "step": 3485
    },
    {
      "epoch": 0.5415145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.0004571532429510973,
      "loss": 2.2047,
      "step": 3486
    },
    {
      "epoch": 0.5416699029126214,
      "grad_norm": 0.0,
      "learning_rate": 0.00045690256470518243,
      "loss": 2.2446,
      "step": 3487
    },
    {
      "epoch": 0.5418252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0004566518973727785,
      "loss": 2.1959,
      "step": 3488
    },
    {
      "epoch": 0.5419805825242718,
      "grad_norm": 0.0,
      "learning_rate": 0.0004564012410173618,
      "loss": 2.1422,
      "step": 3489
    },
    {
      "epoch": 0.5421359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.00045615059570240575,
      "loss": 2.1628,
      "step": 3490
    },
    {
      "epoch": 0.5422912621359224,
      "grad_norm": 0.0,
      "learning_rate": 0.0004558999614913808,
      "loss": 2.2222,
      "step": 3491
    },
    {
      "epoch": 0.5424466019417475,
      "grad_norm": 0.0,
      "learning_rate": 0.0004556493384477551,
      "loss": 2.1299,
      "step": 3492
    },
    {
      "epoch": 0.5426019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.0004553987266349933,
      "loss": 2.3022,
      "step": 3493
    },
    {
      "epoch": 0.5427572815533981,
      "grad_norm": 0.0,
      "learning_rate": 0.0004551481261165577,
      "loss": 2.1754,
      "step": 3494
    },
    {
      "epoch": 0.5429126213592232,
      "grad_norm": 0.0,
      "learning_rate": 0.00045489753695590753,
      "loss": 2.1821,
      "step": 3495
    },
    {
      "epoch": 0.5430679611650485,
      "grad_norm": 0.0,
      "learning_rate": 0.00045464695921649905,
      "loss": 2.2159,
      "step": 3496
    },
    {
      "epoch": 0.5432233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.0004543963929617858,
      "loss": 2.27,
      "step": 3497
    },
    {
      "epoch": 0.5433786407766991,
      "grad_norm": 0.0,
      "learning_rate": 0.00045414583825521853,
      "loss": 2.2177,
      "step": 3498
    },
    {
      "epoch": 0.5435339805825242,
      "grad_norm": 0.0,
      "learning_rate": 0.0004538952951602451,
      "loss": 2.1442,
      "step": 3499
    },
    {
      "epoch": 0.5436893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0004536447637403095,
      "loss": 2.2343,
      "step": 3500
    },
    {
      "epoch": 0.5438446601941748,
      "grad_norm": 0.0,
      "learning_rate": 0.00045339424405885405,
      "loss": 2.1234,
      "step": 3501
    },
    {
      "epoch": 0.544,
      "grad_norm": 0.0,
      "learning_rate": 0.0004531437361793174,
      "loss": 2.2307,
      "step": 3502
    },
    {
      "epoch": 0.5441553398058252,
      "grad_norm": 0.0,
      "learning_rate": 0.0004528932401651354,
      "loss": 2.2551,
      "step": 3503
    },
    {
      "epoch": 0.5443106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0004526427560797408,
      "loss": 2.2928,
      "step": 3504
    },
    {
      "epoch": 0.5444660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.00045239228398656346,
      "loss": 2.1658,
      "step": 3505
    },
    {
      "epoch": 0.544621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0004521418239490299,
      "loss": 2.2883,
      "step": 3506
    },
    {
      "epoch": 0.5447766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0004518913760305641,
      "loss": 2.2688,
      "step": 3507
    },
    {
      "epoch": 0.5449320388349514,
      "grad_norm": 0.0,
      "learning_rate": 0.00045164094029458656,
      "loss": 2.1469,
      "step": 3508
    },
    {
      "epoch": 0.5450873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.00045139051680451485,
      "loss": 2.2031,
      "step": 3509
    },
    {
      "epoch": 0.545242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.00045114010562376347,
      "loss": 2.2153,
      "step": 3510
    },
    {
      "epoch": 0.5453980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.00045088970681574347,
      "loss": 2.2319,
      "step": 3511
    },
    {
      "epoch": 0.5455533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.0004506393204438633,
      "loss": 2.1038,
      "step": 3512
    },
    {
      "epoch": 0.5457087378640777,
      "grad_norm": 0.0,
      "learning_rate": 0.00045038894657152784,
      "loss": 2.218,
      "step": 3513
    },
    {
      "epoch": 0.5458640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.00045013858526213904,
      "loss": 2.267,
      "step": 3514
    },
    {
      "epoch": 0.5460194174757281,
      "grad_norm": 0.0,
      "learning_rate": 0.00044988823657909547,
      "loss": 2.1492,
      "step": 3515
    },
    {
      "epoch": 0.5461747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0004496379005857928,
      "loss": 2.218,
      "step": 3516
    },
    {
      "epoch": 0.5463300970873787,
      "grad_norm": 0.0,
      "learning_rate": 0.0004493875773456232,
      "loss": 2.265,
      "step": 3517
    },
    {
      "epoch": 0.5464854368932038,
      "grad_norm": 0.0,
      "learning_rate": 0.0004491372669219758,
      "loss": 2.1795,
      "step": 3518
    },
    {
      "epoch": 0.5466407766990291,
      "grad_norm": 0.0,
      "learning_rate": 0.0004488869693782364,
      "loss": 2.285,
      "step": 3519
    },
    {
      "epoch": 0.5467961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.0004486366847777874,
      "loss": 2.2159,
      "step": 3520
    },
    {
      "epoch": 0.5469514563106797,
      "grad_norm": 0.0,
      "learning_rate": 0.0004483864131840084,
      "loss": 2.2725,
      "step": 3521
    },
    {
      "epoch": 0.5471067961165048,
      "grad_norm": 0.0,
      "learning_rate": 0.0004481361546602749,
      "loss": 2.1167,
      "step": 3522
    },
    {
      "epoch": 0.5472621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0004478859092699598,
      "loss": 2.2817,
      "step": 3523
    },
    {
      "epoch": 0.5474174757281554,
      "grad_norm": 0.0,
      "learning_rate": 0.0004476356770764324,
      "loss": 2.1772,
      "step": 3524
    },
    {
      "epoch": 0.5475728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.00044738545814305877,
      "loss": 2.2433,
      "step": 3525
    },
    {
      "epoch": 0.5477281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.0004471352525332015,
      "loss": 2.2056,
      "step": 3526
    },
    {
      "epoch": 0.5478834951456311,
      "grad_norm": 0.0,
      "learning_rate": 0.00044688506031021993,
      "loss": 2.1666,
      "step": 3527
    },
    {
      "epoch": 0.5480388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0004466348815374697,
      "loss": 2.2233,
      "step": 3528
    },
    {
      "epoch": 0.5481941747572816,
      "grad_norm": 0.0,
      "learning_rate": 0.00044638471627830357,
      "loss": 2.2527,
      "step": 3529
    },
    {
      "epoch": 0.5483495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.00044613456459607025,
      "loss": 2.1957,
      "step": 3530
    },
    {
      "epoch": 0.548504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0004458844265541157,
      "loss": 2.2294,
      "step": 3531
    },
    {
      "epoch": 0.5486601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.0004456343022157819,
      "loss": 2.2377,
      "step": 3532
    },
    {
      "epoch": 0.5488155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.00044538419164440765,
      "loss": 2.262,
      "step": 3533
    },
    {
      "epoch": 0.5489708737864077,
      "grad_norm": 0.0,
      "learning_rate": 0.0004451340949033279,
      "loss": 2.1821,
      "step": 3534
    },
    {
      "epoch": 0.549126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.00044488401205587446,
      "loss": 2.3123,
      "step": 3535
    },
    {
      "epoch": 0.5492815533980583,
      "grad_norm": 0.0,
      "learning_rate": 0.0004446339431653756,
      "loss": 2.2502,
      "step": 3536
    },
    {
      "epoch": 0.5494368932038834,
      "grad_norm": 0.0,
      "learning_rate": 0.0004443838882951558,
      "loss": 2.2052,
      "step": 3537
    },
    {
      "epoch": 0.5495922330097087,
      "grad_norm": 0.0,
      "learning_rate": 0.0004441338475085363,
      "loss": 2.1588,
      "step": 3538
    },
    {
      "epoch": 0.549747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0004438838208688344,
      "loss": 2.1614,
      "step": 3539
    },
    {
      "epoch": 0.5499029126213593,
      "grad_norm": 0.0,
      "learning_rate": 0.00044363380843936436,
      "loss": 2.1883,
      "step": 3540
    },
    {
      "epoch": 0.5500582524271844,
      "grad_norm": 0.0,
      "learning_rate": 0.0004433838102834363,
      "loss": 2.3065,
      "step": 3541
    },
    {
      "epoch": 0.5502135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0004431338264643571,
      "loss": 2.2012,
      "step": 3542
    },
    {
      "epoch": 0.550368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0004428838570454297,
      "loss": 2.2761,
      "step": 3543
    },
    {
      "epoch": 0.5505242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.00044263390208995374,
      "loss": 2.1377,
      "step": 3544
    },
    {
      "epoch": 0.5506796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.0004423839616612248,
      "loss": 2.1444,
      "step": 3545
    },
    {
      "epoch": 0.5508349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.0004421340358225351,
      "loss": 2.2499,
      "step": 3546
    },
    {
      "epoch": 0.550990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.000441884124637173,
      "loss": 2.2307,
      "step": 3547
    },
    {
      "epoch": 0.5511456310679612,
      "grad_norm": 0.0,
      "learning_rate": 0.0004416342281684231,
      "loss": 2.2861,
      "step": 3548
    },
    {
      "epoch": 0.5513009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.00044138434647956673,
      "loss": 2.1995,
      "step": 3549
    },
    {
      "epoch": 0.5514563106796116,
      "grad_norm": 0.0,
      "learning_rate": 0.0004411344796338809,
      "loss": 2.2258,
      "step": 3550
    },
    {
      "epoch": 0.5516116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.00044088462769463905,
      "loss": 2.2638,
      "step": 3551
    },
    {
      "epoch": 0.5517669902912621,
      "grad_norm": 0.0,
      "learning_rate": 0.0004406347907251109,
      "loss": 2.1909,
      "step": 3552
    },
    {
      "epoch": 0.5519223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0004403849687885625,
      "loss": 2.2767,
      "step": 3553
    },
    {
      "epoch": 0.5520776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.0004401351619482557,
      "loss": 2.2802,
      "step": 3554
    },
    {
      "epoch": 0.5522330097087379,
      "grad_norm": 0.0,
      "learning_rate": 0.000439885370267449,
      "loss": 2.1757,
      "step": 3555
    },
    {
      "epoch": 0.5523883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.00043963559380939707,
      "loss": 2.2682,
      "step": 3556
    },
    {
      "epoch": 0.5525436893203883,
      "grad_norm": 0.0,
      "learning_rate": 0.00043938583263734984,
      "loss": 2.1927,
      "step": 3557
    },
    {
      "epoch": 0.5526990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0004391360868145545,
      "loss": 2.2571,
      "step": 3558
    },
    {
      "epoch": 0.5528543689320389,
      "grad_norm": 0.0,
      "learning_rate": 0.00043888635640425375,
      "loss": 2.163,
      "step": 3559
    },
    {
      "epoch": 0.553009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.00043863664146968657,
      "loss": 2.1682,
      "step": 3560
    },
    {
      "epoch": 0.5531650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.00043838694207408793,
      "loss": 2.1986,
      "step": 3561
    },
    {
      "epoch": 0.5533203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.0004381372582806889,
      "loss": 2.1759,
      "step": 3562
    },
    {
      "epoch": 0.5534757281553399,
      "grad_norm": 0.0,
      "learning_rate": 0.00043788759015271655,
      "loss": 2.2389,
      "step": 3563
    },
    {
      "epoch": 0.553631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.00043763793775339414,
      "loss": 2.1589,
      "step": 3564
    },
    {
      "epoch": 0.5537864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0004373883011459409,
      "loss": 2.2179,
      "step": 3565
    },
    {
      "epoch": 0.5539417475728156,
      "grad_norm": 0.0,
      "learning_rate": 0.00043713868039357183,
      "loss": 2.1791,
      "step": 3566
    },
    {
      "epoch": 0.5540970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.00043688907555949837,
      "loss": 2.2001,
      "step": 3567
    },
    {
      "epoch": 0.554252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.00043663948670692726,
      "loss": 2.1761,
      "step": 3568
    },
    {
      "epoch": 0.5544077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.0004363899138990618,
      "loss": 2.2663,
      "step": 3569
    },
    {
      "epoch": 0.5545631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0004361403571991009,
      "loss": 2.1587,
      "step": 3570
    },
    {
      "epoch": 0.5547184466019417,
      "grad_norm": 0.0,
      "learning_rate": 0.00043589081667023966,
      "loss": 2.1705,
      "step": 3571
    },
    {
      "epoch": 0.554873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0004356412923756687,
      "loss": 2.1672,
      "step": 3572
    },
    {
      "epoch": 0.5550291262135922,
      "grad_norm": 0.0,
      "learning_rate": 0.0004353917843785751,
      "loss": 2.2092,
      "step": 3573
    },
    {
      "epoch": 0.5551844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.00043514229274214114,
      "loss": 2.2804,
      "step": 3574
    },
    {
      "epoch": 0.5553398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.0004348928175295454,
      "loss": 2.2596,
      "step": 3575
    },
    {
      "epoch": 0.5554951456310679,
      "grad_norm": 0.0,
      "learning_rate": 0.0004346433588039622,
      "loss": 2.1917,
      "step": 3576
    },
    {
      "epoch": 0.5556504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0004343939166285617,
      "loss": 2.1738,
      "step": 3577
    },
    {
      "epoch": 0.5558058252427185,
      "grad_norm": 0.0,
      "learning_rate": 0.00043414449106650964,
      "loss": 2.2378,
      "step": 3578
    },
    {
      "epoch": 0.5559611650485436,
      "grad_norm": 0.0,
      "learning_rate": 0.0004338950821809683,
      "loss": 2.1695,
      "step": 3579
    },
    {
      "epoch": 0.5561165048543689,
      "grad_norm": 0.0,
      "learning_rate": 0.0004336456900350945,
      "loss": 2.2039,
      "step": 3580
    },
    {
      "epoch": 0.5562718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.0004333963146920417,
      "loss": 2.2778,
      "step": 3581
    },
    {
      "epoch": 0.5564271844660195,
      "grad_norm": 0.0,
      "learning_rate": 0.0004331469562149592,
      "loss": 2.2062,
      "step": 3582
    },
    {
      "epoch": 0.5565825242718446,
      "grad_norm": 0.0,
      "learning_rate": 0.0004328976146669915,
      "loss": 2.2587,
      "step": 3583
    },
    {
      "epoch": 0.5567378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0004326482901112792,
      "loss": 2.2924,
      "step": 3584
    },
    {
      "epoch": 0.5568932038834952,
      "grad_norm": 0.0,
      "learning_rate": 0.0004323989826109582,
      "loss": 2.24,
      "step": 3585
    },
    {
      "epoch": 0.5570485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0004321496922291606,
      "loss": 2.0991,
      "step": 3586
    },
    {
      "epoch": 0.5572038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.0004319004190290137,
      "loss": 2.2479,
      "step": 3587
    },
    {
      "epoch": 0.5573592233009709,
      "grad_norm": 0.0,
      "learning_rate": 0.00043165116307364087,
      "loss": 2.3497,
      "step": 3588
    },
    {
      "epoch": 0.5575145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.00043140192442616077,
      "loss": 2.1737,
      "step": 3589
    },
    {
      "epoch": 0.5576699029126213,
      "grad_norm": 0.0,
      "learning_rate": 0.00043115270314968797,
      "loss": 2.2346,
      "step": 3590
    },
    {
      "epoch": 0.5578252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.00043090349930733214,
      "loss": 2.1251,
      "step": 3591
    },
    {
      "epoch": 0.5579805825242719,
      "grad_norm": 0.0,
      "learning_rate": 0.00043065431296219906,
      "loss": 2.2829,
      "step": 3592
    },
    {
      "epoch": 0.5581359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0004304051441773898,
      "loss": 2.1918,
      "step": 3593
    },
    {
      "epoch": 0.5582912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.0004301559930160012,
      "loss": 2.2059,
      "step": 3594
    },
    {
      "epoch": 0.5584466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.00042990685954112537,
      "loss": 2.1469,
      "step": 3595
    },
    {
      "epoch": 0.5586019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.00042965774381585007,
      "loss": 2.2652,
      "step": 3596
    },
    {
      "epoch": 0.5587572815533981,
      "grad_norm": 0.0,
      "learning_rate": 0.00042940864590325866,
      "loss": 2.264,
      "step": 3597
    },
    {
      "epoch": 0.5589126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.00042915956586642997,
      "loss": 2.1779,
      "step": 3598
    },
    {
      "epoch": 0.5590679611650485,
      "grad_norm": 0.0,
      "learning_rate": 0.0004289105037684381,
      "loss": 2.2467,
      "step": 3599
    },
    {
      "epoch": 0.5592233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.0004286614596723528,
      "loss": 2.2162,
      "step": 3600
    },
    {
      "epoch": 0.5593786407766991,
      "grad_norm": 0.0,
      "learning_rate": 0.0004284124336412393,
      "loss": 2.2563,
      "step": 3601
    },
    {
      "epoch": 0.5595339805825242,
      "grad_norm": 0.0,
      "learning_rate": 0.0004281634257381579,
      "loss": 2.2605,
      "step": 3602
    },
    {
      "epoch": 0.5596893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0004279144360261648,
      "loss": 2.1351,
      "step": 3603
    },
    {
      "epoch": 0.5598446601941748,
      "grad_norm": 0.0,
      "learning_rate": 0.00042766546456831126,
      "loss": 2.1596,
      "step": 3604
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.0,
      "learning_rate": 0.00042741651142764385,
      "loss": 2.3244,
      "step": 3605
    },
    {
      "epoch": 0.5601553398058252,
      "grad_norm": 0.0,
      "learning_rate": 0.00042716757666720503,
      "loss": 2.2968,
      "step": 3606
    },
    {
      "epoch": 0.5603106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.000426918660350032,
      "loss": 2.3008,
      "step": 3607
    },
    {
      "epoch": 0.5604660194174758,
      "grad_norm": 0.0,
      "learning_rate": 0.0004266697625391577,
      "loss": 2.2039,
      "step": 3608
    },
    {
      "epoch": 0.5606213592233009,
      "grad_norm": 0.0,
      "learning_rate": 0.0004264208832976101,
      "loss": 2.2467,
      "step": 3609
    },
    {
      "epoch": 0.5607766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0004261720226884126,
      "loss": 2.0892,
      "step": 3610
    },
    {
      "epoch": 0.5609320388349515,
      "grad_norm": 0.0,
      "learning_rate": 0.00042592318077458376,
      "loss": 2.2528,
      "step": 3611
    },
    {
      "epoch": 0.5610873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.00042567435761913777,
      "loss": 2.2162,
      "step": 3612
    },
    {
      "epoch": 0.5612427184466019,
      "grad_norm": 0.0,
      "learning_rate": 0.00042542555328508387,
      "loss": 2.182,
      "step": 3613
    },
    {
      "epoch": 0.5613980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.000425176767835426,
      "loss": 2.1497,
      "step": 3614
    },
    {
      "epoch": 0.5615533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.0004249280013331642,
      "loss": 2.2305,
      "step": 3615
    },
    {
      "epoch": 0.5617087378640777,
      "grad_norm": 0.0,
      "learning_rate": 0.0004246792538412931,
      "loss": 2.2523,
      "step": 3616
    },
    {
      "epoch": 0.5618640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.000424430525422803,
      "loss": 2.3881,
      "step": 3617
    },
    {
      "epoch": 0.5620194174757281,
      "grad_norm": 0.0,
      "learning_rate": 0.00042418181614067873,
      "loss": 2.2408,
      "step": 3618
    },
    {
      "epoch": 0.5621747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.00042393312605790097,
      "loss": 2.1546,
      "step": 3619
    },
    {
      "epoch": 0.5623300970873787,
      "grad_norm": 0.0,
      "learning_rate": 0.00042368445523744487,
      "loss": 2.1508,
      "step": 3620
    },
    {
      "epoch": 0.5624854368932038,
      "grad_norm": 0.0,
      "learning_rate": 0.00042343580374228134,
      "loss": 2.1744,
      "step": 3621
    },
    {
      "epoch": 0.5626407766990291,
      "grad_norm": 0.0,
      "learning_rate": 0.000423187171635376,
      "loss": 2.2272,
      "step": 3622
    },
    {
      "epoch": 0.5627961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.00042293855897968967,
      "loss": 2.1512,
      "step": 3623
    },
    {
      "epoch": 0.5629514563106797,
      "grad_norm": 0.0,
      "learning_rate": 0.00042268996583817833,
      "loss": 2.2236,
      "step": 3624
    },
    {
      "epoch": 0.5631067961165048,
      "grad_norm": 0.0,
      "learning_rate": 0.0004224413922737926,
      "loss": 2.1188,
      "step": 3625
    },
    {
      "epoch": 0.5632621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0004221928383494787,
      "loss": 2.1459,
      "step": 3626
    },
    {
      "epoch": 0.5634174757281554,
      "grad_norm": 0.0,
      "learning_rate": 0.00042194430412817755,
      "loss": 2.2161,
      "step": 3627
    },
    {
      "epoch": 0.5635728155339805,
      "grad_norm": 0.0,
      "learning_rate": 0.00042169578967282527,
      "loss": 2.2374,
      "step": 3628
    },
    {
      "epoch": 0.5637281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.00042144729504635267,
      "loss": 2.2028,
      "step": 3629
    },
    {
      "epoch": 0.5638834951456311,
      "grad_norm": 0.0,
      "learning_rate": 0.000421198820311686,
      "loss": 2.0928,
      "step": 3630
    },
    {
      "epoch": 0.5640388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.00042095036553174605,
      "loss": 2.1398,
      "step": 3631
    },
    {
      "epoch": 0.5641941747572815,
      "grad_norm": 0.0,
      "learning_rate": 0.0004207019307694488,
      "loss": 2.2113,
      "step": 3632
    },
    {
      "epoch": 0.5643495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.00042045351608770506,
      "loss": 2.2129,
      "step": 3633
    },
    {
      "epoch": 0.564504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0004202051215494204,
      "loss": 2.2472,
      "step": 3634
    },
    {
      "epoch": 0.5646601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.0004199567472174958,
      "loss": 2.2449,
      "step": 3635
    },
    {
      "epoch": 0.5648155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.00041970839315482686,
      "loss": 2.3369,
      "step": 3636
    },
    {
      "epoch": 0.5649708737864078,
      "grad_norm": 0.0,
      "learning_rate": 0.00041946005942430345,
      "loss": 2.2767,
      "step": 3637
    },
    {
      "epoch": 0.565126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.000419211746088811,
      "loss": 2.1998,
      "step": 3638
    },
    {
      "epoch": 0.5652815533980583,
      "grad_norm": 0.0,
      "learning_rate": 0.00041896345321122984,
      "loss": 2.198,
      "step": 3639
    },
    {
      "epoch": 0.5654368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0004187151808544348,
      "loss": 2.2642,
      "step": 3640
    },
    {
      "epoch": 0.5655922330097087,
      "grad_norm": 0.0,
      "learning_rate": 0.0004184669290812954,
      "loss": 2.205,
      "step": 3641
    },
    {
      "epoch": 0.565747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0004182186979546763,
      "loss": 2.1737,
      "step": 3642
    },
    {
      "epoch": 0.5659029126213593,
      "grad_norm": 0.0,
      "learning_rate": 0.0004179704875374367,
      "loss": 2.1492,
      "step": 3643
    },
    {
      "epoch": 0.5660582524271844,
      "grad_norm": 0.0,
      "learning_rate": 0.00041772229789243065,
      "loss": 2.2261,
      "step": 3644
    },
    {
      "epoch": 0.5662135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0004174741290825069,
      "loss": 2.1633,
      "step": 3645
    },
    {
      "epoch": 0.566368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.00041722598117050895,
      "loss": 2.3036,
      "step": 3646
    },
    {
      "epoch": 0.5665242718446601,
      "grad_norm": 0.0,
      "learning_rate": 0.00041697785421927514,
      "loss": 2.3549,
      "step": 3647
    },
    {
      "epoch": 0.5666796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.00041672974829163803,
      "loss": 2.2709,
      "step": 3648
    },
    {
      "epoch": 0.5668349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.00041648166345042536,
      "loss": 2.2077,
      "step": 3649
    },
    {
      "epoch": 0.566990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0004162335997584594,
      "loss": 2.1919,
      "step": 3650
    },
    {
      "epoch": 0.5671456310679611,
      "grad_norm": 0.0,
      "learning_rate": 0.0004159855572785569,
      "loss": 2.1038,
      "step": 3651
    },
    {
      "epoch": 0.5673009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.00041573753607352943,
      "loss": 2.2391,
      "step": 3652
    },
    {
      "epoch": 0.5674563106796117,
      "grad_norm": 0.0,
      "learning_rate": 0.000415489536206183,
      "loss": 2.213,
      "step": 3653
    },
    {
      "epoch": 0.5676116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.00041524155773931844,
      "loss": 2.1565,
      "step": 3654
    },
    {
      "epoch": 0.5677669902912621,
      "grad_norm": 0.0,
      "learning_rate": 0.000414993600735731,
      "loss": 2.1826,
      "step": 3655
    },
    {
      "epoch": 0.5679223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.00041474566525821055,
      "loss": 2.1334,
      "step": 3656
    },
    {
      "epoch": 0.5680776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.0004144977513695415,
      "loss": 2.2818,
      "step": 3657
    },
    {
      "epoch": 0.5682330097087379,
      "grad_norm": 0.0,
      "learning_rate": 0.00041424985913250277,
      "loss": 2.229,
      "step": 3658
    },
    {
      "epoch": 0.5683883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.00041400198860986774,
      "loss": 2.144,
      "step": 3659
    },
    {
      "epoch": 0.5685436893203883,
      "grad_norm": 0.0,
      "learning_rate": 0.00041375413986440436,
      "loss": 2.2528,
      "step": 3660
    },
    {
      "epoch": 0.5686990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.00041350631295887513,
      "loss": 2.241,
      "step": 3661
    },
    {
      "epoch": 0.5688543689320389,
      "grad_norm": 0.0,
      "learning_rate": 0.00041325850795603683,
      "loss": 2.132,
      "step": 3662
    },
    {
      "epoch": 0.569009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.000413010724918641,
      "loss": 2.2288,
      "step": 3663
    },
    {
      "epoch": 0.5691650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.0004127629639094334,
      "loss": 2.2463,
      "step": 3664
    },
    {
      "epoch": 0.5693203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.00041251522499115425,
      "loss": 2.1976,
      "step": 3665
    },
    {
      "epoch": 0.5694757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.000412267508226538,
      "loss": 2.3368,
      "step": 3666
    },
    {
      "epoch": 0.569631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.00041201981367831376,
      "loss": 2.1791,
      "step": 3667
    },
    {
      "epoch": 0.5697864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0004117721414092051,
      "loss": 2.2277,
      "step": 3668
    },
    {
      "epoch": 0.5699417475728156,
      "grad_norm": 0.0,
      "learning_rate": 0.0004115244914819296,
      "loss": 2.1732,
      "step": 3669
    },
    {
      "epoch": 0.5700970873786407,
      "grad_norm": 0.0,
      "learning_rate": 0.00041127686395919956,
      "loss": 2.2865,
      "step": 3670
    },
    {
      "epoch": 0.570252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0004110292589037209,
      "loss": 2.1235,
      "step": 3671
    },
    {
      "epoch": 0.5704077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.00041078167637819475,
      "loss": 2.1266,
      "step": 3672
    },
    {
      "epoch": 0.5705631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.00041053411644531606,
      "loss": 2.2637,
      "step": 3673
    },
    {
      "epoch": 0.5707184466019417,
      "grad_norm": 0.0,
      "learning_rate": 0.0004102865791677741,
      "loss": 2.1111,
      "step": 3674
    },
    {
      "epoch": 0.570873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.00041003906460825246,
      "loss": 2.1022,
      "step": 3675
    },
    {
      "epoch": 0.5710291262135923,
      "grad_norm": 0.0,
      "learning_rate": 0.0004097915728294288,
      "loss": 2.1781,
      "step": 3676
    },
    {
      "epoch": 0.5711844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.0004095441038939754,
      "loss": 2.2168,
      "step": 3677
    },
    {
      "epoch": 0.5713398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.0004092966578645584,
      "loss": 2.0942,
      "step": 3678
    },
    {
      "epoch": 0.571495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0004090492348038383,
      "loss": 2.2129,
      "step": 3679
    },
    {
      "epoch": 0.5716504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0004088018347744697,
      "loss": 2.1516,
      "step": 3680
    },
    {
      "epoch": 0.5718058252427185,
      "grad_norm": 0.0,
      "learning_rate": 0.0004085544578391015,
      "loss": 2.2756,
      "step": 3681
    },
    {
      "epoch": 0.5719611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.00040830710406037653,
      "loss": 2.1627,
      "step": 3682
    },
    {
      "epoch": 0.5721165048543689,
      "grad_norm": 0.0,
      "learning_rate": 0.0004080597735009318,
      "loss": 2.2413,
      "step": 3683
    },
    {
      "epoch": 0.5722718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.00040781246622339877,
      "loss": 2.1617,
      "step": 3684
    },
    {
      "epoch": 0.5724271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.00040756518229040253,
      "loss": 2.1606,
      "step": 3685
    },
    {
      "epoch": 0.5725825242718446,
      "grad_norm": 0.0,
      "learning_rate": 0.00040731792176456266,
      "loss": 2.2824,
      "step": 3686
    },
    {
      "epoch": 0.5727378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.00040707068470849264,
      "loss": 2.1868,
      "step": 3687
    },
    {
      "epoch": 0.5728932038834952,
      "grad_norm": 0.0,
      "learning_rate": 0.00040682347118479995,
      "loss": 2.1578,
      "step": 3688
    },
    {
      "epoch": 0.5730485436893203,
      "grad_norm": 0.0,
      "learning_rate": 0.00040657628125608615,
      "loss": 2.1177,
      "step": 3689
    },
    {
      "epoch": 0.5732038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.0004063291149849469,
      "loss": 2.2561,
      "step": 3690
    },
    {
      "epoch": 0.5733592233009709,
      "grad_norm": 0.0,
      "learning_rate": 0.00040608197243397165,
      "loss": 2.1527,
      "step": 3691
    },
    {
      "epoch": 0.5735145631067962,
      "grad_norm": 0.0,
      "learning_rate": 0.0004058348536657443,
      "loss": 2.1988,
      "step": 3692
    },
    {
      "epoch": 0.5736699029126213,
      "grad_norm": 0.0,
      "learning_rate": 0.0004055877587428425,
      "loss": 2.166,
      "step": 3693
    },
    {
      "epoch": 0.5738252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.00040534068772783726,
      "loss": 2.246,
      "step": 3694
    },
    {
      "epoch": 0.5739805825242719,
      "grad_norm": 0.0,
      "learning_rate": 0.0004050936406832944,
      "loss": 2.2477,
      "step": 3695
    },
    {
      "epoch": 0.5741359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0004048466176717733,
      "loss": 2.2055,
      "step": 3696
    },
    {
      "epoch": 0.5742912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.00040459961875582745,
      "loss": 2.1759,
      "step": 3697
    },
    {
      "epoch": 0.5744466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.0004043526439980039,
      "loss": 2.2944,
      "step": 3698
    },
    {
      "epoch": 0.5746019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.00040410569346084385,
      "loss": 2.207,
      "step": 3699
    },
    {
      "epoch": 0.574757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0004038587672068821,
      "loss": 2.17,
      "step": 3700
    },
    {
      "epoch": 0.5749126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.00040361186529864785,
      "loss": 2.2366,
      "step": 3701
    },
    {
      "epoch": 0.5750679611650485,
      "grad_norm": 0.0,
      "learning_rate": 0.00040336498779866347,
      "loss": 2.0587,
      "step": 3702
    },
    {
      "epoch": 0.5752233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.00040311813476944566,
      "loss": 2.2096,
      "step": 3703
    },
    {
      "epoch": 0.575378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0004028713062735046,
      "loss": 2.2914,
      "step": 3704
    },
    {
      "epoch": 0.5755339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.00040262450237334423,
      "loss": 2.2207,
      "step": 3705
    },
    {
      "epoch": 0.5756893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.00040237772313146263,
      "loss": 2.2014,
      "step": 3706
    },
    {
      "epoch": 0.5758446601941748,
      "grad_norm": 0.0,
      "learning_rate": 0.00040213096861035127,
      "loss": 2.2002,
      "step": 3707
    },
    {
      "epoch": 0.576,
      "grad_norm": 0.0,
      "learning_rate": 0.0004018842388724956,
      "loss": 2.1448,
      "step": 3708
    },
    {
      "epoch": 0.5761553398058252,
      "grad_norm": 0.0,
      "learning_rate": 0.0004016375339803744,
      "loss": 2.1281,
      "step": 3709
    },
    {
      "epoch": 0.5763106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0004013908539964609,
      "loss": 2.1578,
      "step": 3710
    },
    {
      "epoch": 0.5764660194174758,
      "grad_norm": 0.0,
      "learning_rate": 0.0004011441989832213,
      "loss": 2.1893,
      "step": 3711
    },
    {
      "epoch": 0.5766213592233009,
      "grad_norm": 0.0,
      "learning_rate": 0.0004008975690031158,
      "loss": 2.2666,
      "step": 3712
    },
    {
      "epoch": 0.5767766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0004006509641185982,
      "loss": 2.203,
      "step": 3713
    },
    {
      "epoch": 0.5769320388349515,
      "grad_norm": 0.0,
      "learning_rate": 0.000400404384392116,
      "loss": 2.1515,
      "step": 3714
    },
    {
      "epoch": 0.5770873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.00040015782988611035,
      "loss": 2.3016,
      "step": 3715
    },
    {
      "epoch": 0.5772427184466019,
      "grad_norm": 0.0,
      "learning_rate": 0.0003999113006630157,
      "loss": 2.2697,
      "step": 3716
    },
    {
      "epoch": 0.5773980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.00039966479678526053,
      "loss": 2.2315,
      "step": 3717
    },
    {
      "epoch": 0.5775533980582525,
      "grad_norm": 0.0,
      "learning_rate": 0.0003994183183152665,
      "loss": 2.2219,
      "step": 3718
    },
    {
      "epoch": 0.5777087378640776,
      "grad_norm": 0.0,
      "learning_rate": 0.0003991718653154495,
      "loss": 2.2894,
      "step": 3719
    },
    {
      "epoch": 0.5778640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.0003989254378482181,
      "loss": 2.3389,
      "step": 3720
    },
    {
      "epoch": 0.5780194174757282,
      "grad_norm": 0.0,
      "learning_rate": 0.00039867903597597506,
      "loss": 2.1714,
      "step": 3721
    },
    {
      "epoch": 0.5781747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0003984326597611163,
      "loss": 2.3326,
      "step": 3722
    },
    {
      "epoch": 0.5783300970873786,
      "grad_norm": 0.0,
      "learning_rate": 0.00039818630926603143,
      "loss": 2.1451,
      "step": 3723
    },
    {
      "epoch": 0.5784854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.00039793998455310336,
      "loss": 2.194,
      "step": 3724
    },
    {
      "epoch": 0.5786407766990291,
      "grad_norm": 0.0,
      "learning_rate": 0.00039769368568470884,
      "loss": 2.1863,
      "step": 3725
    },
    {
      "epoch": 0.5787961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.00039744741272321774,
      "loss": 2.2997,
      "step": 3726
    },
    {
      "epoch": 0.5789514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0003972011657309935,
      "loss": 2.098,
      "step": 3727
    },
    {
      "epoch": 0.5791067961165048,
      "grad_norm": 0.0,
      "learning_rate": 0.0003969549447703926,
      "loss": 2.0189,
      "step": 3728
    },
    {
      "epoch": 0.5792621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0003967087499037657,
      "loss": 2.2596,
      "step": 3729
    },
    {
      "epoch": 0.5794174757281554,
      "grad_norm": 0.0,
      "learning_rate": 0.0003964625811934563,
      "loss": 2.2413,
      "step": 3730
    },
    {
      "epoch": 0.5795728155339805,
      "grad_norm": 0.0,
      "learning_rate": 0.0003962164387018013,
      "loss": 2.226,
      "step": 3731
    },
    {
      "epoch": 0.5797281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.0003959703224911312,
      "loss": 2.1907,
      "step": 3732
    },
    {
      "epoch": 0.5798834951456311,
      "grad_norm": 0.0,
      "learning_rate": 0.00039572423262376944,
      "loss": 2.1604,
      "step": 3733
    },
    {
      "epoch": 0.5800388349514564,
      "grad_norm": 0.0,
      "learning_rate": 0.00039547816916203344,
      "loss": 2.1739,
      "step": 3734
    },
    {
      "epoch": 0.5801941747572815,
      "grad_norm": 0.0,
      "learning_rate": 0.0003952321321682334,
      "loss": 2.2599,
      "step": 3735
    },
    {
      "epoch": 0.5803495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.00039498612170467287,
      "loss": 2.2366,
      "step": 3736
    },
    {
      "epoch": 0.5805048543689321,
      "grad_norm": 0.0,
      "learning_rate": 0.0003947401378336489,
      "loss": 2.1456,
      "step": 3737
    },
    {
      "epoch": 0.5806601941747572,
      "grad_norm": 0.0,
      "learning_rate": 0.00039449418061745167,
      "loss": 2.1672,
      "step": 3738
    },
    {
      "epoch": 0.5808155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.00039424825011836453,
      "loss": 2.191,
      "step": 3739
    },
    {
      "epoch": 0.5809708737864078,
      "grad_norm": 0.0,
      "learning_rate": 0.00039400234639866427,
      "loss": 2.1693,
      "step": 3740
    },
    {
      "epoch": 0.581126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.00039375646952062067,
      "loss": 2.2455,
      "step": 3741
    },
    {
      "epoch": 0.5812815533980582,
      "grad_norm": 0.0,
      "learning_rate": 0.0003935106195464968,
      "loss": 2.2173,
      "step": 3742
    },
    {
      "epoch": 0.5814368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.00039326479653854925,
      "loss": 2.2235,
      "step": 3743
    },
    {
      "epoch": 0.5815922330097087,
      "grad_norm": 0.0,
      "learning_rate": 0.00039301900055902715,
      "loss": 2.2036,
      "step": 3744
    },
    {
      "epoch": 0.581747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.00039277323167017335,
      "loss": 2.1464,
      "step": 3745
    },
    {
      "epoch": 0.5819029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.00039252748993422355,
      "loss": 2.1689,
      "step": 3746
    },
    {
      "epoch": 0.5820582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.00039228177541340646,
      "loss": 2.228,
      "step": 3747
    },
    {
      "epoch": 0.5822135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0003920360881699442,
      "loss": 2.2006,
      "step": 3748
    },
    {
      "epoch": 0.582368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.00039179042826605195,
      "loss": 2.1433,
      "step": 3749
    },
    {
      "epoch": 0.5825242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.00039154479576393794,
      "loss": 2.1833,
      "step": 3750
    },
    {
      "epoch": 0.5826796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.0003912991907258029,
      "loss": 2.1992,
      "step": 3751
    },
    {
      "epoch": 0.5828349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.0003910536132138415,
      "loss": 2.199,
      "step": 3752
    },
    {
      "epoch": 0.582990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.00039080806329024097,
      "loss": 2.0715,
      "step": 3753
    },
    {
      "epoch": 0.5831456310679611,
      "grad_norm": 0.0,
      "learning_rate": 0.0003905625410171816,
      "loss": 2.1982,
      "step": 3754
    },
    {
      "epoch": 0.5833009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0003903170464568367,
      "loss": 2.0837,
      "step": 3755
    },
    {
      "epoch": 0.5834563106796117,
      "grad_norm": 0.0,
      "learning_rate": 0.0003900715796713725,
      "loss": 2.2921,
      "step": 3756
    },
    {
      "epoch": 0.5836116504854368,
      "grad_norm": 0.0,
      "learning_rate": 0.00038982614072294823,
      "loss": 2.1331,
      "step": 3757
    },
    {
      "epoch": 0.5837669902912621,
      "grad_norm": 0.0,
      "learning_rate": 0.00038958072967371635,
      "loss": 2.2007,
      "step": 3758
    },
    {
      "epoch": 0.5839223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0003893353465858217,
      "loss": 2.2001,
      "step": 3759
    },
    {
      "epoch": 0.5840776699029127,
      "grad_norm": 0.0,
      "learning_rate": 0.0003890899915214025,
      "loss": 2.2228,
      "step": 3760
    },
    {
      "epoch": 0.5842330097087378,
      "grad_norm": 0.0,
      "learning_rate": 0.0003888446645425898,
      "loss": 2.2224,
      "step": 3761
    },
    {
      "epoch": 0.5843883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.00038859936571150724,
      "loss": 2.1378,
      "step": 3762
    },
    {
      "epoch": 0.5845436893203884,
      "grad_norm": 0.0,
      "learning_rate": 0.00038835409509027164,
      "loss": 2.1687,
      "step": 3763
    },
    {
      "epoch": 0.5846990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.00038810885274099253,
      "loss": 2.3353,
      "step": 3764
    },
    {
      "epoch": 0.5848543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.00038786363872577237,
      "loss": 2.077,
      "step": 3765
    },
    {
      "epoch": 0.5850097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.0003876184531067063,
      "loss": 2.2583,
      "step": 3766
    },
    {
      "epoch": 0.5851650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.00038737329594588265,
      "loss": 2.1266,
      "step": 3767
    },
    {
      "epoch": 0.5853203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.0003871281673053821,
      "loss": 2.226,
      "step": 3768
    },
    {
      "epoch": 0.5854757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.00038688306724727825,
      "loss": 2.1501,
      "step": 3769
    },
    {
      "epoch": 0.585631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0003866379958336375,
      "loss": 2.2635,
      "step": 3770
    },
    {
      "epoch": 0.5857864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.000386392953126519,
      "loss": 2.2716,
      "step": 3771
    },
    {
      "epoch": 0.5859417475728156,
      "grad_norm": 0.0,
      "learning_rate": 0.00038614793918797456,
      "loss": 2.1274,
      "step": 3772
    },
    {
      "epoch": 0.5860970873786407,
      "grad_norm": 0.0,
      "learning_rate": 0.00038590295408004916,
      "loss": 2.2395,
      "step": 3773
    },
    {
      "epoch": 0.586252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0003856579978647795,
      "loss": 2.205,
      "step": 3774
    },
    {
      "epoch": 0.5864077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.0003854130706041957,
      "loss": 2.269,
      "step": 3775
    },
    {
      "epoch": 0.5865631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0003851681723603206,
      "loss": 2.2138,
      "step": 3776
    },
    {
      "epoch": 0.5867184466019417,
      "grad_norm": 0.0,
      "learning_rate": 0.00038492330319516934,
      "loss": 2.2911,
      "step": 3777
    },
    {
      "epoch": 0.586873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0003846784631707499,
      "loss": 2.193,
      "step": 3778
    },
    {
      "epoch": 0.5870291262135923,
      "grad_norm": 0.0,
      "learning_rate": 0.00038443365234906273,
      "loss": 2.1173,
      "step": 3779
    },
    {
      "epoch": 0.5871844660194174,
      "grad_norm": 0.0,
      "learning_rate": 0.00038418887079210107,
      "loss": 2.2017,
      "step": 3780
    },
    {
      "epoch": 0.5873398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.00038394411856185055,
      "loss": 2.1847,
      "step": 3781
    },
    {
      "epoch": 0.587495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0003836993957202896,
      "loss": 2.2191,
      "step": 3782
    },
    {
      "epoch": 0.5876504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.00038345470232938907,
      "loss": 2.197,
      "step": 3783
    },
    {
      "epoch": 0.5878058252427184,
      "grad_norm": 0.0,
      "learning_rate": 0.0003832100384511124,
      "loss": 2.2269,
      "step": 3784
    },
    {
      "epoch": 0.5879611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0003829654041474154,
      "loss": 2.119,
      "step": 3785
    },
    {
      "epoch": 0.588116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0003827207994802465,
      "loss": 2.261,
      "step": 3786
    },
    {
      "epoch": 0.5882718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.00038247622451154686,
      "loss": 2.177,
      "step": 3787
    },
    {
      "epoch": 0.5884271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.00038223167930324963,
      "loss": 2.2151,
      "step": 3788
    },
    {
      "epoch": 0.5885825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.00038198716391728087,
      "loss": 2.21,
      "step": 3789
    },
    {
      "epoch": 0.5887378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0003817426784155588,
      "loss": 2.0955,
      "step": 3790
    },
    {
      "epoch": 0.5888932038834952,
      "grad_norm": 0.0,
      "learning_rate": 0.0003814982228599944,
      "loss": 2.203,
      "step": 3791
    },
    {
      "epoch": 0.5890485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.00038125379731249057,
      "loss": 2.2314,
      "step": 3792
    },
    {
      "epoch": 0.5892038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.0003810094018349431,
      "loss": 2.2803,
      "step": 3793
    },
    {
      "epoch": 0.5893592233009709,
      "grad_norm": 0.0,
      "learning_rate": 0.0003807650364892399,
      "loss": 2.1511,
      "step": 3794
    },
    {
      "epoch": 0.5895145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.0003805207013372614,
      "loss": 2.2472,
      "step": 3795
    },
    {
      "epoch": 0.5896699029126213,
      "grad_norm": 0.0,
      "learning_rate": 0.0003802763964408801,
      "loss": 2.3171,
      "step": 3796
    },
    {
      "epoch": 0.5898252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0003800321218619611,
      "loss": 2.2156,
      "step": 3797
    },
    {
      "epoch": 0.5899805825242719,
      "grad_norm": 0.0,
      "learning_rate": 0.00037978787766236176,
      "loss": 2.1584,
      "step": 3798
    },
    {
      "epoch": 0.590135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0003795436639039317,
      "loss": 2.1343,
      "step": 3799
    },
    {
      "epoch": 0.5902912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.00037929948064851306,
      "loss": 2.0527,
      "step": 3800
    },
    {
      "epoch": 0.5904466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.0003790553279579399,
      "loss": 2.2086,
      "step": 3801
    },
    {
      "epoch": 0.5906019417475729,
      "grad_norm": 0.0,
      "learning_rate": 0.00037881120589403875,
      "loss": 2.1555,
      "step": 3802
    },
    {
      "epoch": 0.590757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.00037856711451862837,
      "loss": 2.2016,
      "step": 3803
    },
    {
      "epoch": 0.5909126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.00037832305389351965,
      "loss": 2.1619,
      "step": 3804
    },
    {
      "epoch": 0.5910679611650486,
      "grad_norm": 0.0,
      "learning_rate": 0.00037807902408051576,
      "loss": 2.1693,
      "step": 3805
    },
    {
      "epoch": 0.5912233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.00037783502514141215,
      "loss": 2.2034,
      "step": 3806
    },
    {
      "epoch": 0.591378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.00037759105713799663,
      "loss": 2.2793,
      "step": 3807
    },
    {
      "epoch": 0.5915339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0003773471201320483,
      "loss": 2.2368,
      "step": 3808
    },
    {
      "epoch": 0.5916893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.00037710321418533944,
      "loss": 2.2279,
      "step": 3809
    },
    {
      "epoch": 0.5918446601941748,
      "grad_norm": 0.0,
      "learning_rate": 0.000376859339359634,
      "loss": 2.2017,
      "step": 3810
    },
    {
      "epoch": 0.592,
      "grad_norm": 0.0,
      "learning_rate": 0.00037661549571668814,
      "loss": 2.2285,
      "step": 3811
    },
    {
      "epoch": 0.5921553398058252,
      "grad_norm": 0.0,
      "learning_rate": 0.00037637168331825,
      "loss": 2.2208,
      "step": 3812
    },
    {
      "epoch": 0.5923106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0003761279022260599,
      "loss": 2.251,
      "step": 3813
    },
    {
      "epoch": 0.5924660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.00037588415250185017,
      "loss": 2.253,
      "step": 3814
    },
    {
      "epoch": 0.592621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0003756404342073455,
      "loss": 2.1769,
      "step": 3815
    },
    {
      "epoch": 0.5927766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0003753967474042622,
      "loss": 2.3071,
      "step": 3816
    },
    {
      "epoch": 0.5929320388349515,
      "grad_norm": 0.0,
      "learning_rate": 0.00037515309215430874,
      "loss": 2.0277,
      "step": 3817
    },
    {
      "epoch": 0.5930873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0003749094685191859,
      "loss": 2.1331,
      "step": 3818
    },
    {
      "epoch": 0.5932427184466019,
      "grad_norm": 0.0,
      "learning_rate": 0.0003746658765605859,
      "loss": 2.2345,
      "step": 3819
    },
    {
      "epoch": 0.5933980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0003744223163401933,
      "loss": 2.2205,
      "step": 3820
    },
    {
      "epoch": 0.5935533980582525,
      "grad_norm": 0.0,
      "learning_rate": 0.00037417878791968465,
      "loss": 2.1755,
      "step": 3821
    },
    {
      "epoch": 0.5937087378640776,
      "grad_norm": 0.0,
      "learning_rate": 0.00037393529136072823,
      "loss": 2.1244,
      "step": 3822
    },
    {
      "epoch": 0.5938640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.00037369182672498437,
      "loss": 2.1633,
      "step": 3823
    },
    {
      "epoch": 0.5940194174757282,
      "grad_norm": 0.0,
      "learning_rate": 0.00037344839407410556,
      "loss": 2.2206,
      "step": 3824
    },
    {
      "epoch": 0.5941747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0003732049934697358,
      "loss": 2.1616,
      "step": 3825
    },
    {
      "epoch": 0.5943300970873786,
      "grad_norm": 0.0,
      "learning_rate": 0.00037296162497351104,
      "loss": 2.1509,
      "step": 3826
    },
    {
      "epoch": 0.5944854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.00037271828864705934,
      "loss": 2.1521,
      "step": 3827
    },
    {
      "epoch": 0.5946407766990291,
      "grad_norm": 0.0,
      "learning_rate": 0.0003724749845520004,
      "loss": 2.2852,
      "step": 3828
    },
    {
      "epoch": 0.5947961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.0003722317127499457,
      "loss": 2.1255,
      "step": 3829
    },
    {
      "epoch": 0.5949514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0003719884733024991,
      "loss": 2.2786,
      "step": 3830
    },
    {
      "epoch": 0.5951067961165049,
      "grad_norm": 0.0,
      "learning_rate": 0.00037174526627125525,
      "loss": 2.1331,
      "step": 3831
    },
    {
      "epoch": 0.5952621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0003715020917178014,
      "loss": 2.2712,
      "step": 3832
    },
    {
      "epoch": 0.5954174757281553,
      "grad_norm": 0.0,
      "learning_rate": 0.00037125894970371645,
      "loss": 2.2479,
      "step": 3833
    },
    {
      "epoch": 0.5955728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.0003710158402905709,
      "loss": 2.1483,
      "step": 3834
    },
    {
      "epoch": 0.5957281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.00037077276353992693,
      "loss": 2.2048,
      "step": 3835
    },
    {
      "epoch": 0.5958834951456311,
      "grad_norm": 0.0,
      "learning_rate": 0.0003705297195133388,
      "loss": 2.1554,
      "step": 3836
    },
    {
      "epoch": 0.5960388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.000370286708272352,
      "loss": 2.2404,
      "step": 3837
    },
    {
      "epoch": 0.5961941747572815,
      "grad_norm": 0.0,
      "learning_rate": 0.0003700437298785041,
      "loss": 2.2134,
      "step": 3838
    },
    {
      "epoch": 0.5963495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.00036980078439332426,
      "loss": 2.2122,
      "step": 3839
    },
    {
      "epoch": 0.5965048543689321,
      "grad_norm": 0.0,
      "learning_rate": 0.0003695578718783332,
      "loss": 2.2826,
      "step": 3840
    },
    {
      "epoch": 0.5966601941747572,
      "grad_norm": 0.0,
      "learning_rate": 0.0003693149923950436,
      "loss": 2.149,
      "step": 3841
    },
    {
      "epoch": 0.5968155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.0003690721460049591,
      "loss": 2.2316,
      "step": 3842
    },
    {
      "epoch": 0.5969708737864078,
      "grad_norm": 0.0,
      "learning_rate": 0.00036882933276957577,
      "loss": 2.2277,
      "step": 3843
    },
    {
      "epoch": 0.5971262135922331,
      "grad_norm": 0.0,
      "learning_rate": 0.00036858655275038063,
      "loss": 2.1578,
      "step": 3844
    },
    {
      "epoch": 0.5972815533980582,
      "grad_norm": 0.0,
      "learning_rate": 0.0003683438060088527,
      "loss": 2.1533,
      "step": 3845
    },
    {
      "epoch": 0.5974368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.00036810109260646245,
      "loss": 2.1054,
      "step": 3846
    },
    {
      "epoch": 0.5975922330097088,
      "grad_norm": 0.0,
      "learning_rate": 0.00036785841260467177,
      "loss": 2.2257,
      "step": 3847
    },
    {
      "epoch": 0.597747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.00036761576606493453,
      "loss": 2.082,
      "step": 3848
    },
    {
      "epoch": 0.5979029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.00036737315304869556,
      "loss": 2.3445,
      "step": 3849
    },
    {
      "epoch": 0.5980582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.0003671305736173915,
      "loss": 2.2859,
      "step": 3850
    },
    {
      "epoch": 0.5982135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0003668880278324505,
      "loss": 2.256,
      "step": 3851
    },
    {
      "epoch": 0.5983689320388349,
      "grad_norm": 0.0,
      "learning_rate": 0.0003666455157552922,
      "loss": 2.1786,
      "step": 3852
    },
    {
      "epoch": 0.5985242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0003664030374473275,
      "loss": 2.1403,
      "step": 3853
    },
    {
      "epoch": 0.5986796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.00036616059296995886,
      "loss": 2.2574,
      "step": 3854
    },
    {
      "epoch": 0.5988349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.00036591818238458016,
      "loss": 2.1825,
      "step": 3855
    },
    {
      "epoch": 0.5989902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.0003656758057525769,
      "loss": 2.1256,
      "step": 3856
    },
    {
      "epoch": 0.5991456310679611,
      "grad_norm": 0.0,
      "learning_rate": 0.0003654334631353259,
      "loss": 2.1556,
      "step": 3857
    },
    {
      "epoch": 0.5993009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.00036519115459419504,
      "loss": 2.1342,
      "step": 3858
    },
    {
      "epoch": 0.5994563106796117,
      "grad_norm": 0.0,
      "learning_rate": 0.00036494888019054396,
      "loss": 2.181,
      "step": 3859
    },
    {
      "epoch": 0.5996116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.00036470663998572345,
      "loss": 2.2188,
      "step": 3860
    },
    {
      "epoch": 0.5997669902912621,
      "grad_norm": 0.0,
      "learning_rate": 0.00036446443404107576,
      "loss": 2.2489,
      "step": 3861
    },
    {
      "epoch": 0.5999223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.00036422226241793426,
      "loss": 2.1716,
      "step": 3862
    },
    {
      "epoch": 0.6000776699029127,
      "grad_norm": 0.0,
      "learning_rate": 0.00036398012517762404,
      "loss": 2.226,
      "step": 3863
    },
    {
      "epoch": 0.6002330097087378,
      "grad_norm": 0.0,
      "learning_rate": 0.0003637380223814613,
      "loss": 2.2117,
      "step": 3864
    },
    {
      "epoch": 0.6003883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0003634959540907528,
      "loss": 2.272,
      "step": 3865
    },
    {
      "epoch": 0.6005436893203884,
      "grad_norm": 0.0,
      "learning_rate": 0.0003632539203667977,
      "loss": 2.1625,
      "step": 3866
    },
    {
      "epoch": 0.6006990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0003630119212708859,
      "loss": 2.2629,
      "step": 3867
    },
    {
      "epoch": 0.6008543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.0003627699568642984,
      "loss": 2.1238,
      "step": 3868
    },
    {
      "epoch": 0.6010097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.0003625280272083076,
      "loss": 2.1632,
      "step": 3869
    },
    {
      "epoch": 0.6011650485436894,
      "grad_norm": 0.0,
      "learning_rate": 0.0003622861323641771,
      "loss": 2.178,
      "step": 3870
    },
    {
      "epoch": 0.6013203883495145,
      "grad_norm": 0.0,
      "learning_rate": 0.0003620442723931614,
      "loss": 2.3556,
      "step": 3871
    },
    {
      "epoch": 0.6014757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0003618024473565067,
      "loss": 2.1484,
      "step": 3872
    },
    {
      "epoch": 0.6016310679611651,
      "grad_norm": 0.0,
      "learning_rate": 0.00036156065731545004,
      "loss": 2.2316,
      "step": 3873
    },
    {
      "epoch": 0.6017864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.00036131890233121947,
      "loss": 2.175,
      "step": 3874
    },
    {
      "epoch": 0.6019417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.00036107718246503447,
      "loss": 2.145,
      "step": 3875
    },
    {
      "epoch": 0.6020970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.00036083549777810524,
      "loss": 2.1765,
      "step": 3876
    },
    {
      "epoch": 0.602252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0003605938483316335,
      "loss": 2.0968,
      "step": 3877
    },
    {
      "epoch": 0.6024077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.00036035223418681174,
      "loss": 2.226,
      "step": 3878
    },
    {
      "epoch": 0.6025631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0003601106554048237,
      "loss": 2.2001,
      "step": 3879
    },
    {
      "epoch": 0.6027184466019417,
      "grad_norm": 0.0,
      "learning_rate": 0.00035986911204684383,
      "loss": 2.0742,
      "step": 3880
    },
    {
      "epoch": 0.602873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.00035962760417403817,
      "loss": 2.0865,
      "step": 3881
    },
    {
      "epoch": 0.6030291262135923,
      "grad_norm": 0.0,
      "learning_rate": 0.0003593861318475634,
      "loss": 2.1454,
      "step": 3882
    },
    {
      "epoch": 0.6031844660194174,
      "grad_norm": 0.0,
      "learning_rate": 0.00035914469512856716,
      "loss": 2.1947,
      "step": 3883
    },
    {
      "epoch": 0.6033398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.00035890329407818833,
      "loss": 2.2157,
      "step": 3884
    },
    {
      "epoch": 0.603495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.00035866192875755646,
      "loss": 2.2018,
      "step": 3885
    },
    {
      "epoch": 0.6036504854368933,
      "grad_norm": 0.0,
      "learning_rate": 0.00035842059922779214,
      "loss": 2.121,
      "step": 3886
    },
    {
      "epoch": 0.6038058252427184,
      "grad_norm": 0.0,
      "learning_rate": 0.0003581793055500073,
      "loss": 2.2213,
      "step": 3887
    },
    {
      "epoch": 0.6039611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.00035793804778530417,
      "loss": 2.1483,
      "step": 3888
    },
    {
      "epoch": 0.604116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.00035769682599477595,
      "loss": 2.2184,
      "step": 3889
    },
    {
      "epoch": 0.6042718446601941,
      "grad_norm": 0.0,
      "learning_rate": 0.00035745564023950726,
      "loss": 2.1925,
      "step": 3890
    },
    {
      "epoch": 0.6044271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.00035721449058057325,
      "loss": 2.1843,
      "step": 3891
    },
    {
      "epoch": 0.6045825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.0003569733770790399,
      "loss": 2.0957,
      "step": 3892
    },
    {
      "epoch": 0.6047378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0003567322997959641,
      "loss": 2.0857,
      "step": 3893
    },
    {
      "epoch": 0.6048932038834951,
      "grad_norm": 0.0,
      "learning_rate": 0.0003564912587923935,
      "loss": 2.1996,
      "step": 3894
    },
    {
      "epoch": 0.6050485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0003562502541293666,
      "loss": 2.3848,
      "step": 3895
    },
    {
      "epoch": 0.6052038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.00035600928586791285,
      "loss": 2.3381,
      "step": 3896
    },
    {
      "epoch": 0.6053592233009709,
      "grad_norm": 0.0,
      "learning_rate": 0.00035576835406905236,
      "loss": 2.2194,
      "step": 3897
    },
    {
      "epoch": 0.6055145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.00035552745879379594,
      "loss": 2.0125,
      "step": 3898
    },
    {
      "epoch": 0.6056699029126213,
      "grad_norm": 0.0,
      "learning_rate": 0.0003552866001031451,
      "loss": 2.2611,
      "step": 3899
    },
    {
      "epoch": 0.6058252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0003550457780580923,
      "loss": 2.0703,
      "step": 3900
    },
    {
      "epoch": 0.6059805825242719,
      "grad_norm": 0.0,
      "learning_rate": 0.0003548049927196206,
      "loss": 2.3048,
      "step": 3901
    },
    {
      "epoch": 0.606135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.00035456424414870367,
      "loss": 2.2335,
      "step": 3902
    },
    {
      "epoch": 0.6062912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.0003543235324063062,
      "loss": 2.1583,
      "step": 3903
    },
    {
      "epoch": 0.6064466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.00035408285755338304,
      "loss": 2.0939,
      "step": 3904
    },
    {
      "epoch": 0.6066019417475729,
      "grad_norm": 0.0,
      "learning_rate": 0.00035384221965088027,
      "loss": 2.2562,
      "step": 3905
    },
    {
      "epoch": 0.606757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.00035360161875973425,
      "loss": 2.2795,
      "step": 3906
    },
    {
      "epoch": 0.6069126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.000353361054940872,
      "loss": 2.1183,
      "step": 3907
    },
    {
      "epoch": 0.6070679611650486,
      "grad_norm": 0.0,
      "learning_rate": 0.00035312052825521123,
      "loss": 2.1563,
      "step": 3908
    },
    {
      "epoch": 0.6072233009708737,
      "grad_norm": 0.0,
      "learning_rate": 0.00035288003876366033,
      "loss": 2.1411,
      "step": 3909
    },
    {
      "epoch": 0.607378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0003526395865271179,
      "loss": 2.1567,
      "step": 3910
    },
    {
      "epoch": 0.6075339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0003523991716064735,
      "loss": 2.1545,
      "step": 3911
    },
    {
      "epoch": 0.6076893203883496,
      "grad_norm": 0.0,
      "learning_rate": 0.0003521587940626072,
      "loss": 2.3088,
      "step": 3912
    },
    {
      "epoch": 0.6078446601941747,
      "grad_norm": 0.0,
      "learning_rate": 0.0003519184539563893,
      "loss": 2.1687,
      "step": 3913
    },
    {
      "epoch": 0.608,
      "grad_norm": 0.0,
      "learning_rate": 0.00035167815134868105,
      "loss": 2.2494,
      "step": 3914
    },
    {
      "epoch": 0.6081553398058253,
      "grad_norm": 0.0,
      "learning_rate": 0.0003514378863003339,
      "loss": 2.203,
      "step": 3915
    },
    {
      "epoch": 0.6083106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.00035119765887218986,
      "loss": 2.1361,
      "step": 3916
    },
    {
      "epoch": 0.6084660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.00035095746912508144,
      "loss": 2.1411,
      "step": 3917
    },
    {
      "epoch": 0.608621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0003507173171198317,
      "loss": 2.1565,
      "step": 3918
    },
    {
      "epoch": 0.6087766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0003504772029172537,
      "loss": 2.1433,
      "step": 3919
    },
    {
      "epoch": 0.6089320388349515,
      "grad_norm": 0.0,
      "learning_rate": 0.0003502371265781517,
      "loss": 2.1603,
      "step": 3920
    },
    {
      "epoch": 0.6090873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0003499970881633199,
      "loss": 2.1929,
      "step": 3921
    },
    {
      "epoch": 0.6092427184466019,
      "grad_norm": 0.0,
      "learning_rate": 0.0003497570877335425,
      "loss": 2.1544,
      "step": 3922
    },
    {
      "epoch": 0.6093980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.00034951712534959494,
      "loss": 2.0896,
      "step": 3923
    },
    {
      "epoch": 0.6095533980582525,
      "grad_norm": 0.0,
      "learning_rate": 0.00034927720107224246,
      "loss": 2.0486,
      "step": 3924
    },
    {
      "epoch": 0.6097087378640776,
      "grad_norm": 0.0,
      "learning_rate": 0.00034903731496224076,
      "loss": 2.2446,
      "step": 3925
    },
    {
      "epoch": 0.6098640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.00034879746708033597,
      "loss": 2.0745,
      "step": 3926
    },
    {
      "epoch": 0.6100194174757282,
      "grad_norm": 0.0,
      "learning_rate": 0.00034855765748726444,
      "loss": 2.1904,
      "step": 3927
    },
    {
      "epoch": 0.6101747572815533,
      "grad_norm": 0.0,
      "learning_rate": 0.0003483178862437528,
      "loss": 2.2612,
      "step": 3928
    },
    {
      "epoch": 0.6103300970873786,
      "grad_norm": 0.0,
      "learning_rate": 0.00034807815341051816,
      "loss": 2.1132,
      "step": 3929
    },
    {
      "epoch": 0.6104854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.0003478384590482677,
      "loss": 2.2012,
      "step": 3930
    },
    {
      "epoch": 0.6106407766990292,
      "grad_norm": 0.0,
      "learning_rate": 0.00034759880321769896,
      "loss": 2.2568,
      "step": 3931
    },
    {
      "epoch": 0.6107961165048543,
      "grad_norm": 0.0,
      "learning_rate": 0.00034735918597949986,
      "loss": 2.2241,
      "step": 3932
    },
    {
      "epoch": 0.6109514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.000347119607394348,
      "loss": 2.1859,
      "step": 3933
    },
    {
      "epoch": 0.6111067961165049,
      "grad_norm": 0.0,
      "learning_rate": 0.00034688006752291184,
      "loss": 2.1532,
      "step": 3934
    },
    {
      "epoch": 0.6112621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0003466405664258497,
      "loss": 2.3088,
      "step": 3935
    },
    {
      "epoch": 0.6114174757281553,
      "grad_norm": 0.0,
      "learning_rate": 0.0003464011041638101,
      "loss": 2.199,
      "step": 3936
    },
    {
      "epoch": 0.6115728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.00034616168079743177,
      "loss": 2.2798,
      "step": 3937
    },
    {
      "epoch": 0.6117281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.00034592229638734373,
      "loss": 2.0916,
      "step": 3938
    },
    {
      "epoch": 0.6118834951456311,
      "grad_norm": 0.0,
      "learning_rate": 0.000345682950994165,
      "loss": 2.2318,
      "step": 3939
    },
    {
      "epoch": 0.6120388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.00034544364467850466,
      "loss": 2.1729,
      "step": 3940
    },
    {
      "epoch": 0.6121941747572816,
      "grad_norm": 0.0,
      "learning_rate": 0.0003452043775009619,
      "loss": 2.212,
      "step": 3941
    },
    {
      "epoch": 0.6123495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0003449651495221261,
      "loss": 2.2135,
      "step": 3942
    },
    {
      "epoch": 0.612504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.00034472596080257654,
      "loss": 2.3179,
      "step": 3943
    },
    {
      "epoch": 0.6126601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.0003444868114028832,
      "loss": 2.1255,
      "step": 3944
    },
    {
      "epoch": 0.6128155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.00034424770138360506,
      "loss": 2.2484,
      "step": 3945
    },
    {
      "epoch": 0.6129708737864078,
      "grad_norm": 0.0,
      "learning_rate": 0.00034400863080529166,
      "loss": 2.2689,
      "step": 3946
    },
    {
      "epoch": 0.613126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.00034376959972848276,
      "loss": 2.3187,
      "step": 3947
    },
    {
      "epoch": 0.6132815533980582,
      "grad_norm": 0.0,
      "learning_rate": 0.00034353060821370795,
      "loss": 2.1693,
      "step": 3948
    },
    {
      "epoch": 0.6134368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0003432916563214866,
      "loss": 2.1741,
      "step": 3949
    },
    {
      "epoch": 0.6135922330097088,
      "grad_norm": 0.0,
      "learning_rate": 0.0003430527441123283,
      "loss": 2.1877,
      "step": 3950
    },
    {
      "epoch": 0.6137475728155339,
      "grad_norm": 0.0,
      "learning_rate": 0.0003428138716467325,
      "loss": 2.1484,
      "step": 3951
    },
    {
      "epoch": 0.6139029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.00034257503898518836,
      "loss": 2.1975,
      "step": 3952
    },
    {
      "epoch": 0.6140582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.00034233624618817553,
      "loss": 2.1479,
      "step": 3953
    },
    {
      "epoch": 0.6142135922330098,
      "grad_norm": 0.0,
      "learning_rate": 0.0003420974933161631,
      "loss": 2.1716,
      "step": 3954
    },
    {
      "epoch": 0.6143689320388349,
      "grad_norm": 0.0,
      "learning_rate": 0.00034185878042961027,
      "loss": 2.1637,
      "step": 3955
    },
    {
      "epoch": 0.6145242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0003416201075889657,
      "loss": 2.2267,
      "step": 3956
    },
    {
      "epoch": 0.6146796116504855,
      "grad_norm": 0.0,
      "learning_rate": 0.0003413814748546684,
      "loss": 2.2557,
      "step": 3957
    },
    {
      "epoch": 0.6148349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.00034114288228714705,
      "loss": 2.2408,
      "step": 3958
    },
    {
      "epoch": 0.6149902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.00034090432994682025,
      "loss": 2.1195,
      "step": 3959
    },
    {
      "epoch": 0.6151456310679612,
      "grad_norm": 0.0,
      "learning_rate": 0.0003406658178940961,
      "loss": 1.987,
      "step": 3960
    },
    {
      "epoch": 0.6153009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0003404273461893729,
      "loss": 2.1524,
      "step": 3961
    },
    {
      "epoch": 0.6154563106796116,
      "grad_norm": 0.0,
      "learning_rate": 0.00034018891489303845,
      "loss": 2.3214,
      "step": 3962
    },
    {
      "epoch": 0.6156116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0003399505240654705,
      "loss": 2.3392,
      "step": 3963
    },
    {
      "epoch": 0.6157669902912621,
      "grad_norm": 0.0,
      "learning_rate": 0.00033971217376703645,
      "loss": 2.1956,
      "step": 3964
    },
    {
      "epoch": 0.6159223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0003394738640580934,
      "loss": 2.2775,
      "step": 3965
    },
    {
      "epoch": 0.6160776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.00033923559499898814,
      "loss": 2.1858,
      "step": 3966
    },
    {
      "epoch": 0.6162330097087378,
      "grad_norm": 0.0,
      "learning_rate": 0.0003389973666500577,
      "loss": 2.1941,
      "step": 3967
    },
    {
      "epoch": 0.6163883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.00033875917907162786,
      "loss": 2.2239,
      "step": 3968
    },
    {
      "epoch": 0.6165436893203884,
      "grad_norm": 0.0,
      "learning_rate": 0.00033852103232401466,
      "loss": 2.1414,
      "step": 3969
    },
    {
      "epoch": 0.6166990291262135,
      "grad_norm": 0.0,
      "learning_rate": 0.00033828292646752376,
      "loss": 2.2148,
      "step": 3970
    },
    {
      "epoch": 0.6168543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.0003380448615624505,
      "loss": 2.2209,
      "step": 3971
    },
    {
      "epoch": 0.6170097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.00033780683766907973,
      "loss": 2.1861,
      "step": 3972
    },
    {
      "epoch": 0.6171650485436894,
      "grad_norm": 0.0,
      "learning_rate": 0.00033756885484768597,
      "loss": 2.1731,
      "step": 3973
    },
    {
      "epoch": 0.6173203883495145,
      "grad_norm": 0.0,
      "learning_rate": 0.0003373309131585332,
      "loss": 2.2458,
      "step": 3974
    },
    {
      "epoch": 0.6174757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0003370930126618751,
      "loss": 2.1334,
      "step": 3975
    },
    {
      "epoch": 0.6176310679611651,
      "grad_norm": 0.0,
      "learning_rate": 0.00033685515341795515,
      "loss": 2.2179,
      "step": 3976
    },
    {
      "epoch": 0.6177864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.000336617335487006,
      "loss": 2.1138,
      "step": 3977
    },
    {
      "epoch": 0.6179417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.0003363795589292502,
      "loss": 2.2426,
      "step": 3978
    },
    {
      "epoch": 0.6180970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.00033614182380489917,
      "loss": 2.1263,
      "step": 3979
    },
    {
      "epoch": 0.618252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0003359041301741546,
      "loss": 2.2732,
      "step": 3980
    },
    {
      "epoch": 0.6184077669902912,
      "grad_norm": 0.0,
      "learning_rate": 0.0003356664780972073,
      "loss": 2.235,
      "step": 3981
    },
    {
      "epoch": 0.6185631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.00033542886763423765,
      "loss": 2.1392,
      "step": 3982
    },
    {
      "epoch": 0.6187184466019418,
      "grad_norm": 0.0,
      "learning_rate": 0.0003351912988454154,
      "loss": 2.2126,
      "step": 3983
    },
    {
      "epoch": 0.618873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.00033495377179089994,
      "loss": 2.1937,
      "step": 3984
    },
    {
      "epoch": 0.6190291262135922,
      "grad_norm": 0.0,
      "learning_rate": 0.0003347162865308396,
      "loss": 2.1035,
      "step": 3985
    },
    {
      "epoch": 0.6191844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.00033447884312537294,
      "loss": 2.1809,
      "step": 3986
    },
    {
      "epoch": 0.6193398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.00033424144163462713,
      "loss": 2.0494,
      "step": 3987
    },
    {
      "epoch": 0.619495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.00033400408211871926,
      "loss": 2.1136,
      "step": 3988
    },
    {
      "epoch": 0.6196504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0003337667646377556,
      "loss": 2.1222,
      "step": 3989
    },
    {
      "epoch": 0.6198058252427184,
      "grad_norm": 0.0,
      "learning_rate": 0.0003335294892518317,
      "loss": 2.2062,
      "step": 3990
    },
    {
      "epoch": 0.6199611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.00033329225602103234,
      "loss": 2.123,
      "step": 3991
    },
    {
      "epoch": 0.620116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.00033305506500543204,
      "loss": 2.2766,
      "step": 3992
    },
    {
      "epoch": 0.6202718446601941,
      "grad_norm": 0.0,
      "learning_rate": 0.0003328179162650943,
      "loss": 2.1761,
      "step": 3993
    },
    {
      "epoch": 0.6204271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.0003325808098600719,
      "loss": 2.2588,
      "step": 3994
    },
    {
      "epoch": 0.6205825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.0003323437458504073,
      "loss": 2.1949,
      "step": 3995
    },
    {
      "epoch": 0.62073786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0003321067242961317,
      "loss": 2.2377,
      "step": 3996
    },
    {
      "epoch": 0.6208932038834951,
      "grad_norm": 0.0,
      "learning_rate": 0.0003318697452572659,
      "loss": 2.1265,
      "step": 3997
    },
    {
      "epoch": 0.6210485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0003316328087938197,
      "loss": 2.1318,
      "step": 3998
    },
    {
      "epoch": 0.6212038834951457,
      "grad_norm": 0.0,
      "learning_rate": 0.00033139591496579216,
      "loss": 2.3773,
      "step": 3999
    },
    {
      "epoch": 0.6213592233009708,
      "grad_norm": 0.0,
      "learning_rate": 0.00033115906383317185,
      "loss": 2.0938,
      "step": 4000
    },
    {
      "epoch": 0.6215145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.00033092225545593645,
      "loss": 2.0759,
      "step": 4001
    },
    {
      "epoch": 0.6216699029126214,
      "grad_norm": 0.0,
      "learning_rate": 0.00033068548989405213,
      "loss": 2.2362,
      "step": 4002
    },
    {
      "epoch": 0.6218252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.00033044876720747496,
      "loss": 2.1606,
      "step": 4003
    },
    {
      "epoch": 0.6219805825242718,
      "grad_norm": 0.0,
      "learning_rate": 0.00033021208745615004,
      "loss": 2.1271,
      "step": 4004
    },
    {
      "epoch": 0.6221359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0003299754507000114,
      "loss": 2.2278,
      "step": 4005
    },
    {
      "epoch": 0.6222912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.0003297388569989824,
      "loss": 2.1224,
      "step": 4006
    },
    {
      "epoch": 0.6224466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.0003295023064129753,
      "loss": 2.2805,
      "step": 4007
    },
    {
      "epoch": 0.6226019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.00032926579900189137,
      "loss": 2.14,
      "step": 4008
    },
    {
      "epoch": 0.622757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.00032902933482562127,
      "loss": 2.0948,
      "step": 4009
    },
    {
      "epoch": 0.6229126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.00032879291394404453,
      "loss": 2.2845,
      "step": 4010
    },
    {
      "epoch": 0.6230679611650486,
      "grad_norm": 0.0,
      "learning_rate": 0.00032855653641702975,
      "loss": 2.1796,
      "step": 4011
    },
    {
      "epoch": 0.6232233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.0003283202023044345,
      "loss": 2.2975,
      "step": 4012
    },
    {
      "epoch": 0.623378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0003280839116661053,
      "loss": 2.1655,
      "step": 4013
    },
    {
      "epoch": 0.6235339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0003278476645618777,
      "loss": 2.2081,
      "step": 4014
    },
    {
      "epoch": 0.6236893203883496,
      "grad_norm": 0.0,
      "learning_rate": 0.0003276114610515764,
      "loss": 2.2223,
      "step": 4015
    },
    {
      "epoch": 0.6238446601941747,
      "grad_norm": 0.0,
      "learning_rate": 0.0003273753011950149,
      "loss": 2.1251,
      "step": 4016
    },
    {
      "epoch": 0.624,
      "grad_norm": 0.0,
      "learning_rate": 0.0003271391850519955,
      "loss": 2.1721,
      "step": 4017
    },
    {
      "epoch": 0.6241553398058253,
      "grad_norm": 0.0,
      "learning_rate": 0.0003269031126823099,
      "loss": 2.1377,
      "step": 4018
    },
    {
      "epoch": 0.6243106796116504,
      "grad_norm": 0.0,
      "learning_rate": 0.0003266670841457382,
      "loss": 2.1603,
      "step": 4019
    },
    {
      "epoch": 0.6244660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.00032643109950204974,
      "loss": 2.1188,
      "step": 4020
    },
    {
      "epoch": 0.624621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0003261951588110026,
      "loss": 2.1796,
      "step": 4021
    },
    {
      "epoch": 0.6247766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0003259592621323436,
      "loss": 2.2085,
      "step": 4022
    },
    {
      "epoch": 0.6249320388349514,
      "grad_norm": 0.0,
      "learning_rate": 0.0003257234095258087,
      "loss": 2.1064,
      "step": 4023
    },
    {
      "epoch": 0.6250873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0003254876010511229,
      "loss": 2.274,
      "step": 4024
    },
    {
      "epoch": 0.625242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.00032525183676799905,
      "loss": 2.0944,
      "step": 4025
    },
    {
      "epoch": 0.6253980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0003250161167361398,
      "loss": 2.1844,
      "step": 4026
    },
    {
      "epoch": 0.6255533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.0003247804410152362,
      "loss": 2.1675,
      "step": 4027
    },
    {
      "epoch": 0.6257087378640777,
      "grad_norm": 0.0,
      "learning_rate": 0.0003245448096649681,
      "loss": 2.2086,
      "step": 4028
    },
    {
      "epoch": 0.6258640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.0003243092227450043,
      "loss": 2.1287,
      "step": 4029
    },
    {
      "epoch": 0.6260194174757282,
      "grad_norm": 0.0,
      "learning_rate": 0.00032407368031500195,
      "loss": 2.1305,
      "step": 4030
    },
    {
      "epoch": 0.6261747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0003238381824346073,
      "loss": 2.1923,
      "step": 4031
    },
    {
      "epoch": 0.6263300970873786,
      "grad_norm": 0.0,
      "learning_rate": 0.00032360272916345513,
      "loss": 2.1637,
      "step": 4032
    },
    {
      "epoch": 0.6264854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.00032336732056116916,
      "loss": 2.2904,
      "step": 4033
    },
    {
      "epoch": 0.6266407766990292,
      "grad_norm": 0.0,
      "learning_rate": 0.0003231319566873614,
      "loss": 2.2788,
      "step": 4034
    },
    {
      "epoch": 0.6267961165048543,
      "grad_norm": 0.0,
      "learning_rate": 0.0003228966376016331,
      "loss": 2.1833,
      "step": 4035
    },
    {
      "epoch": 0.6269514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0003226613633635732,
      "loss": 2.2575,
      "step": 4036
    },
    {
      "epoch": 0.6271067961165049,
      "grad_norm": 0.0,
      "learning_rate": 0.00032242613403276043,
      "loss": 2.2167,
      "step": 4037
    },
    {
      "epoch": 0.62726213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0003221909496687615,
      "loss": 2.1017,
      "step": 4038
    },
    {
      "epoch": 0.6274174757281553,
      "grad_norm": 0.0,
      "learning_rate": 0.00032195581033113186,
      "loss": 2.1456,
      "step": 4039
    },
    {
      "epoch": 0.6275728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.00032172071607941555,
      "loss": 2.2633,
      "step": 4040
    },
    {
      "epoch": 0.6277281553398059,
      "grad_norm": 0.0,
      "learning_rate": 0.00032148566697314505,
      "loss": 2.1417,
      "step": 4041
    },
    {
      "epoch": 0.627883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0003212506630718418,
      "loss": 2.1061,
      "step": 4042
    },
    {
      "epoch": 0.6280388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0003210157044350155,
      "loss": 2.183,
      "step": 4043
    },
    {
      "epoch": 0.6281941747572816,
      "grad_norm": 0.0,
      "learning_rate": 0.0003207807911221643,
      "loss": 2.2825,
      "step": 4044
    },
    {
      "epoch": 0.6283495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0003205459231927752,
      "loss": 2.2606,
      "step": 4045
    },
    {
      "epoch": 0.628504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0003203111007063234,
      "loss": 2.1786,
      "step": 4046
    },
    {
      "epoch": 0.6286601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.0003200763237222728,
      "loss": 2.201,
      "step": 4047
    },
    {
      "epoch": 0.6288155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.0003198415923000755,
      "loss": 2.2173,
      "step": 4048
    },
    {
      "epoch": 0.6289708737864078,
      "grad_norm": 0.0,
      "learning_rate": 0.00031960690649917244,
      "loss": 2.1335,
      "step": 4049
    },
    {
      "epoch": 0.629126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.00031937226637899265,
      "loss": 2.1656,
      "step": 4050
    },
    {
      "epoch": 0.6292815533980582,
      "grad_norm": 0.0,
      "learning_rate": 0.0003191376719989539,
      "loss": 2.196,
      "step": 4051
    },
    {
      "epoch": 0.6294368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0003189031234184623,
      "loss": 2.2018,
      "step": 4052
    },
    {
      "epoch": 0.6295922330097088,
      "grad_norm": 0.0,
      "learning_rate": 0.00031866862069691214,
      "loss": 2.157,
      "step": 4053
    },
    {
      "epoch": 0.629747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.00031843416389368636,
      "loss": 2.1629,
      "step": 4054
    },
    {
      "epoch": 0.6299029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.00031819975306815614,
      "loss": 2.228,
      "step": 4055
    },
    {
      "epoch": 0.6300582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.0003179653882796809,
      "loss": 2.092,
      "step": 4056
    },
    {
      "epoch": 0.6302135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.00031773106958760887,
      "loss": 2.1216,
      "step": 4057
    },
    {
      "epoch": 0.6303689320388349,
      "grad_norm": 0.0,
      "learning_rate": 0.00031749679705127634,
      "loss": 2.1389,
      "step": 4058
    },
    {
      "epoch": 0.6305242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0003172625707300073,
      "loss": 2.2026,
      "step": 4059
    },
    {
      "epoch": 0.6306796116504855,
      "grad_norm": 0.0,
      "learning_rate": 0.00031702839068311494,
      "loss": 2.1891,
      "step": 4060
    },
    {
      "epoch": 0.6308349514563106,
      "grad_norm": 0.0,
      "learning_rate": 0.00031679425696990045,
      "loss": 2.1699,
      "step": 4061
    },
    {
      "epoch": 0.6309902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.0003165601696496531,
      "loss": 2.1492,
      "step": 4062
    },
    {
      "epoch": 0.6311456310679612,
      "grad_norm": 0.0,
      "learning_rate": 0.0003163261287816507,
      "loss": 2.267,
      "step": 4063
    },
    {
      "epoch": 0.6313009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.00031609213442515896,
      "loss": 2.1105,
      "step": 4064
    },
    {
      "epoch": 0.6314563106796116,
      "grad_norm": 0.0,
      "learning_rate": 0.0003158581866394319,
      "loss": 2.1426,
      "step": 4065
    },
    {
      "epoch": 0.6316116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.00031562428548371216,
      "loss": 2.1426,
      "step": 4066
    },
    {
      "epoch": 0.6317669902912622,
      "grad_norm": 0.0,
      "learning_rate": 0.00031539043101722996,
      "loss": 2.1473,
      "step": 4067
    },
    {
      "epoch": 0.6319223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0003151566232992042,
      "loss": 2.1835,
      "step": 4068
    },
    {
      "epoch": 0.6320776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.00031492286238884163,
      "loss": 2.2053,
      "step": 4069
    },
    {
      "epoch": 0.6322330097087379,
      "grad_norm": 0.0,
      "learning_rate": 0.00031468914834533726,
      "loss": 2.0415,
      "step": 4070
    },
    {
      "epoch": 0.6323883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.00031445548122787406,
      "loss": 2.165,
      "step": 4071
    },
    {
      "epoch": 0.6325436893203884,
      "grad_norm": 0.0,
      "learning_rate": 0.00031422186109562335,
      "loss": 2.1572,
      "step": 4072
    },
    {
      "epoch": 0.6326990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.00031398828800774457,
      "loss": 2.1659,
      "step": 4073
    },
    {
      "epoch": 0.6328543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.000313754762023385,
      "loss": 2.1587,
      "step": 4074
    },
    {
      "epoch": 0.6330097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.0003135212832016803,
      "loss": 2.2037,
      "step": 4075
    },
    {
      "epoch": 0.6331650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.00031328785160175394,
      "loss": 2.0613,
      "step": 4076
    },
    {
      "epoch": 0.6333203883495145,
      "grad_norm": 0.0,
      "learning_rate": 0.00031305446728271757,
      "loss": 2.1689,
      "step": 4077
    },
    {
      "epoch": 0.6334757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.00031282113030367076,
      "loss": 2.1451,
      "step": 4078
    },
    {
      "epoch": 0.6336310679611651,
      "grad_norm": 0.0,
      "learning_rate": 0.0003125878407237012,
      "loss": 2.0857,
      "step": 4079
    },
    {
      "epoch": 0.6337864077669902,
      "grad_norm": 0.0,
      "learning_rate": 0.0003123545986018843,
      "loss": 2.1917,
      "step": 4080
    },
    {
      "epoch": 0.6339417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.00031212140399728425,
      "loss": 2.1664,
      "step": 4081
    },
    {
      "epoch": 0.6340970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.00031188825696895205,
      "loss": 2.1892,
      "step": 4082
    },
    {
      "epoch": 0.6342524271844661,
      "grad_norm": 0.0,
      "learning_rate": 0.00031165515757592733,
      "loss": 2.0361,
      "step": 4083
    },
    {
      "epoch": 0.6344077669902912,
      "grad_norm": 0.0,
      "learning_rate": 0.0003114221058772378,
      "loss": 2.2285,
      "step": 4084
    },
    {
      "epoch": 0.6345631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.00031118910193189876,
      "loss": 2.1462,
      "step": 4085
    },
    {
      "epoch": 0.6347184466019418,
      "grad_norm": 0.0,
      "learning_rate": 0.0003109561457989134,
      "loss": 2.1153,
      "step": 4086
    },
    {
      "epoch": 0.634873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0003107232375372731,
      "loss": 2.1532,
      "step": 4087
    },
    {
      "epoch": 0.6350291262135922,
      "grad_norm": 0.0,
      "learning_rate": 0.00031049037720595686,
      "loss": 2.2646,
      "step": 4088
    },
    {
      "epoch": 0.6351844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.0003102575648639315,
      "loss": 2.1835,
      "step": 4089
    },
    {
      "epoch": 0.6353398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.00031002480057015194,
      "loss": 2.1271,
      "step": 4090
    },
    {
      "epoch": 0.635495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0003097920843835609,
      "loss": 2.2761,
      "step": 4091
    },
    {
      "epoch": 0.6356504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.00030955941636308864,
      "loss": 2.2382,
      "step": 4092
    },
    {
      "epoch": 0.6358058252427184,
      "grad_norm": 0.0,
      "learning_rate": 0.00030932679656765343,
      "loss": 2.1859,
      "step": 4093
    },
    {
      "epoch": 0.6359611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.00030909422505616136,
      "loss": 2.1907,
      "step": 4094
    },
    {
      "epoch": 0.6361165048543689,
      "grad_norm": 0.0,
      "learning_rate": 0.0003088617018875061,
      "loss": 2.236,
      "step": 4095
    },
    {
      "epoch": 0.6362718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.0003086292271205693,
      "loss": 2.1166,
      "step": 4096
    },
    {
      "epoch": 0.6364271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.0003083968008142203,
      "loss": 2.2321,
      "step": 4097
    },
    {
      "epoch": 0.6365825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.00030816442302731596,
      "loss": 2.0939,
      "step": 4098
    },
    {
      "epoch": 0.6367378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0003079320938187013,
      "loss": 2.1727,
      "step": 4099
    },
    {
      "epoch": 0.6368932038834951,
      "grad_norm": 0.0,
      "learning_rate": 0.00030769981324720854,
      "loss": 2.1553,
      "step": 4100
    },
    {
      "epoch": 0.6370485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.00030746758137165797,
      "loss": 2.2981,
      "step": 4101
    },
    {
      "epoch": 0.6372038834951457,
      "grad_norm": 0.0,
      "learning_rate": 0.0003072353982508572,
      "loss": 2.1383,
      "step": 4102
    },
    {
      "epoch": 0.6373592233009708,
      "grad_norm": 0.0,
      "learning_rate": 0.00030700326394360187,
      "loss": 2.1205,
      "step": 4103
    },
    {
      "epoch": 0.6375145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.00030677117850867497,
      "loss": 2.2003,
      "step": 4104
    },
    {
      "epoch": 0.6376699029126214,
      "grad_norm": 0.0,
      "learning_rate": 0.00030653914200484716,
      "loss": 2.1542,
      "step": 4105
    },
    {
      "epoch": 0.6378252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.00030630715449087687,
      "loss": 2.1373,
      "step": 4106
    },
    {
      "epoch": 0.6379805825242718,
      "grad_norm": 0.0,
      "learning_rate": 0.0003060752160255098,
      "loss": 2.1489,
      "step": 4107
    },
    {
      "epoch": 0.6381359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0003058433266674797,
      "loss": 2.2666,
      "step": 4108
    },
    {
      "epoch": 0.6382912621359224,
      "grad_norm": 0.0,
      "learning_rate": 0.0003056114864755075,
      "loss": 2.2612,
      "step": 4109
    },
    {
      "epoch": 0.6384466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.00030537969550830193,
      "loss": 2.1583,
      "step": 4110
    },
    {
      "epoch": 0.6386019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.0003051479538245591,
      "loss": 2.1656,
      "step": 4111
    },
    {
      "epoch": 0.6387572815533981,
      "grad_norm": 0.0,
      "learning_rate": 0.00030491626148296247,
      "loss": 2.136,
      "step": 4112
    },
    {
      "epoch": 0.6389126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.00030468461854218335,
      "loss": 2.21,
      "step": 4113
    },
    {
      "epoch": 0.6390679611650485,
      "grad_norm": 0.0,
      "learning_rate": 0.00030445302506088045,
      "loss": 2.197,
      "step": 4114
    },
    {
      "epoch": 0.6392233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.00030422148109770017,
      "loss": 2.1963,
      "step": 4115
    },
    {
      "epoch": 0.639378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0003039899867112754,
      "loss": 2.2106,
      "step": 4116
    },
    {
      "epoch": 0.6395339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0003037585419602278,
      "loss": 2.1226,
      "step": 4117
    },
    {
      "epoch": 0.6396893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0003035271469031655,
      "loss": 2.2229,
      "step": 4118
    },
    {
      "epoch": 0.6398446601941747,
      "grad_norm": 0.0,
      "learning_rate": 0.00030329580159868465,
      "loss": 2.1521,
      "step": 4119
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.0,
      "learning_rate": 0.0003030645061053683,
      "loss": 2.1559,
      "step": 4120
    },
    {
      "epoch": 0.6401553398058253,
      "grad_norm": 0.0,
      "learning_rate": 0.00030283326048178726,
      "loss": 2.0819,
      "step": 4121
    },
    {
      "epoch": 0.6403106796116504,
      "grad_norm": 0.0,
      "learning_rate": 0.0003026020647864994,
      "loss": 2.2015,
      "step": 4122
    },
    {
      "epoch": 0.6404660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.0003023709190780505,
      "loss": 2.1658,
      "step": 4123
    },
    {
      "epoch": 0.640621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.000302139823414973,
      "loss": 2.1391,
      "step": 4124
    },
    {
      "epoch": 0.6407766990291263,
      "grad_norm": 0.0,
      "learning_rate": 0.0003019087778557871,
      "loss": 2.1238,
      "step": 4125
    },
    {
      "epoch": 0.6409320388349514,
      "grad_norm": 0.0,
      "learning_rate": 0.00030167778245900024,
      "loss": 2.1498,
      "step": 4126
    },
    {
      "epoch": 0.6410873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.00030144683728310686,
      "loss": 2.1877,
      "step": 4127
    },
    {
      "epoch": 0.641242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.00030121594238658903,
      "loss": 2.1987,
      "step": 4128
    },
    {
      "epoch": 0.6413980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0003009850978279161,
      "loss": 2.1011,
      "step": 4129
    },
    {
      "epoch": 0.6415533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.00030075430366554445,
      "loss": 2.1356,
      "step": 4130
    },
    {
      "epoch": 0.6417087378640777,
      "grad_norm": 0.0,
      "learning_rate": 0.00030052355995791773,
      "loss": 2.1843,
      "step": 4131
    },
    {
      "epoch": 0.6418640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.0003002928667634671,
      "loss": 2.2653,
      "step": 4132
    },
    {
      "epoch": 0.6420194174757281,
      "grad_norm": 0.0,
      "learning_rate": 0.00030006222414061056,
      "loss": 2.2427,
      "step": 4133
    },
    {
      "epoch": 0.6421747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0002998316321477536,
      "loss": 2.229,
      "step": 4134
    },
    {
      "epoch": 0.6423300970873786,
      "grad_norm": 0.0,
      "learning_rate": 0.00029960109084328855,
      "loss": 2.149,
      "step": 4135
    },
    {
      "epoch": 0.6424854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.0002993706002855953,
      "loss": 2.1296,
      "step": 4136
    },
    {
      "epoch": 0.6426407766990291,
      "grad_norm": 0.0,
      "learning_rate": 0.0002991401605330405,
      "loss": 2.254,
      "step": 4137
    },
    {
      "epoch": 0.6427961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.00029890977164397846,
      "loss": 2.0983,
      "step": 4138
    },
    {
      "epoch": 0.6429514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.00029867943367675,
      "loss": 2.2091,
      "step": 4139
    },
    {
      "epoch": 0.6431067961165049,
      "grad_norm": 0.0,
      "learning_rate": 0.00029844914668968313,
      "loss": 2.2435,
      "step": 4140
    },
    {
      "epoch": 0.6432621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.00029821891074109354,
      "loss": 2.1891,
      "step": 4141
    },
    {
      "epoch": 0.6434174757281553,
      "grad_norm": 0.0,
      "learning_rate": 0.00029798872588928337,
      "loss": 2.0612,
      "step": 4142
    },
    {
      "epoch": 0.6435728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.0002977585921925422,
      "loss": 2.1594,
      "step": 4143
    },
    {
      "epoch": 0.6437281553398059,
      "grad_norm": 0.0,
      "learning_rate": 0.00029752850970914645,
      "loss": 2.1486,
      "step": 4144
    },
    {
      "epoch": 0.643883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.00029729847849735956,
      "loss": 2.1125,
      "step": 4145
    },
    {
      "epoch": 0.6440388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.00029706849861543203,
      "loss": 2.1805,
      "step": 4146
    },
    {
      "epoch": 0.6441941747572816,
      "grad_norm": 0.0,
      "learning_rate": 0.0002968385701216015,
      "loss": 2.2522,
      "step": 4147
    },
    {
      "epoch": 0.6443495145631069,
      "grad_norm": 0.0,
      "learning_rate": 0.00029660869307409255,
      "loss": 2.1628,
      "step": 4148
    },
    {
      "epoch": 0.644504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0002963788675311166,
      "loss": 2.1818,
      "step": 4149
    },
    {
      "epoch": 0.6446601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.00029614909355087196,
      "loss": 2.1934,
      "step": 4150
    },
    {
      "epoch": 0.6448155339805826,
      "grad_norm": 0.0,
      "learning_rate": 0.00029591937119154404,
      "loss": 2.149,
      "step": 4151
    },
    {
      "epoch": 0.6449708737864077,
      "grad_norm": 0.0,
      "learning_rate": 0.0002956897005113053,
      "loss": 2.168,
      "step": 4152
    },
    {
      "epoch": 0.645126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0002954600815683148,
      "loss": 2.1584,
      "step": 4153
    },
    {
      "epoch": 0.6452815533980583,
      "grad_norm": 0.0,
      "learning_rate": 0.0002952305144207188,
      "loss": 2.1493,
      "step": 4154
    },
    {
      "epoch": 0.6454368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.00029500099912665014,
      "loss": 2.0634,
      "step": 4155
    },
    {
      "epoch": 0.6455922330097087,
      "grad_norm": 0.0,
      "learning_rate": 0.00029477153574422886,
      "loss": 2.1427,
      "step": 4156
    },
    {
      "epoch": 0.645747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.00029454212433156177,
      "loss": 2.2324,
      "step": 4157
    },
    {
      "epoch": 0.6459029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.0002943127649467422,
      "loss": 2.0874,
      "step": 4158
    },
    {
      "epoch": 0.6460582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.0002940834576478507,
      "loss": 2.1875,
      "step": 4159
    },
    {
      "epoch": 0.6462135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0002938542024929545,
      "loss": 2.1374,
      "step": 4160
    },
    {
      "epoch": 0.6463689320388349,
      "grad_norm": 0.0,
      "learning_rate": 0.0002936249995401076,
      "loss": 2.1647,
      "step": 4161
    },
    {
      "epoch": 0.6465242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0002933958488473507,
      "loss": 2.1477,
      "step": 4162
    },
    {
      "epoch": 0.6466796116504855,
      "grad_norm": 0.0,
      "learning_rate": 0.00029316675047271135,
      "loss": 2.2029,
      "step": 4163
    },
    {
      "epoch": 0.6468349514563106,
      "grad_norm": 0.0,
      "learning_rate": 0.00029293770447420375,
      "loss": 2.2733,
      "step": 4164
    },
    {
      "epoch": 0.6469902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.0002927087109098292,
      "loss": 2.1866,
      "step": 4165
    },
    {
      "epoch": 0.6471456310679612,
      "grad_norm": 0.0,
      "learning_rate": 0.00029247976983757536,
      "loss": 2.1061,
      "step": 4166
    },
    {
      "epoch": 0.6473009708737865,
      "grad_norm": 0.0,
      "learning_rate": 0.00029225088131541667,
      "loss": 2.1653,
      "step": 4167
    },
    {
      "epoch": 0.6474563106796116,
      "grad_norm": 0.0,
      "learning_rate": 0.0002920220454013143,
      "loss": 2.193,
      "step": 4168
    },
    {
      "epoch": 0.6476116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0002917932621532159,
      "loss": 2.1399,
      "step": 4169
    },
    {
      "epoch": 0.6477669902912622,
      "grad_norm": 0.0,
      "learning_rate": 0.0002915645316290565,
      "loss": 2.2225,
      "step": 4170
    },
    {
      "epoch": 0.6479223300970873,
      "grad_norm": 0.0,
      "learning_rate": 0.00029133585388675665,
      "loss": 2.1454,
      "step": 4171
    },
    {
      "epoch": 0.6480776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.00029110722898422457,
      "loss": 2.1253,
      "step": 4172
    },
    {
      "epoch": 0.6482330097087379,
      "grad_norm": 0.0,
      "learning_rate": 0.00029087865697935455,
      "loss": 2.1705,
      "step": 4173
    },
    {
      "epoch": 0.6483883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.00029065013793002725,
      "loss": 2.1912,
      "step": 4174
    },
    {
      "epoch": 0.6485436893203883,
      "grad_norm": 0.0,
      "learning_rate": 0.0002904216718941109,
      "loss": 2.162,
      "step": 4175
    },
    {
      "epoch": 0.6486990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.00029019325892945904,
      "loss": 2.2565,
      "step": 4176
    },
    {
      "epoch": 0.6488543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.0002899648990939128,
      "loss": 2.1353,
      "step": 4177
    },
    {
      "epoch": 0.6490097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.0002897365924452995,
      "loss": 2.1576,
      "step": 4178
    },
    {
      "epoch": 0.6491650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.00028950833904143263,
      "loss": 2.1168,
      "step": 4179
    },
    {
      "epoch": 0.6493203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.00028928013894011294,
      "loss": 2.0992,
      "step": 4180
    },
    {
      "epoch": 0.6494757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.00028905199219912686,
      "loss": 2.1191,
      "step": 4181
    },
    {
      "epoch": 0.6496310679611651,
      "grad_norm": 0.0,
      "learning_rate": 0.00028882389887624786,
      "loss": 2.112,
      "step": 4182
    },
    {
      "epoch": 0.6497864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.00028859585902923625,
      "loss": 2.1743,
      "step": 4183
    },
    {
      "epoch": 0.6499417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.0002883678727158373,
      "loss": 2.1555,
      "step": 4184
    },
    {
      "epoch": 0.6500970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.0002881399399937842,
      "loss": 2.2839,
      "step": 4185
    },
    {
      "epoch": 0.650252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0002879120609207964,
      "loss": 2.0408,
      "step": 4186
    },
    {
      "epoch": 0.6504077669902912,
      "grad_norm": 0.0,
      "learning_rate": 0.00028768423555457885,
      "loss": 2.1917,
      "step": 4187
    },
    {
      "epoch": 0.6505631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.000287456463952824,
      "loss": 2.1573,
      "step": 4188
    },
    {
      "epoch": 0.6507184466019418,
      "grad_norm": 0.0,
      "learning_rate": 0.00028722874617320964,
      "loss": 2.1246,
      "step": 4189
    },
    {
      "epoch": 0.6508737864077669,
      "grad_norm": 0.0,
      "learning_rate": 0.00028700108227340104,
      "loss": 2.0804,
      "step": 4190
    },
    {
      "epoch": 0.6510291262135922,
      "grad_norm": 0.0,
      "learning_rate": 0.00028677347231104877,
      "loss": 2.1791,
      "step": 4191
    },
    {
      "epoch": 0.6511844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.0002865459163437905,
      "loss": 2.2634,
      "step": 4192
    },
    {
      "epoch": 0.6513398058252428,
      "grad_norm": 0.0,
      "learning_rate": 0.00028631841442924996,
      "loss": 2.1756,
      "step": 4193
    },
    {
      "epoch": 0.6514951456310679,
      "grad_norm": 0.0,
      "learning_rate": 0.0002860909666250369,
      "loss": 2.1488,
      "step": 4194
    },
    {
      "epoch": 0.6516504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0002858635729887481,
      "loss": 2.1556,
      "step": 4195
    },
    {
      "epoch": 0.6518058252427185,
      "grad_norm": 0.0,
      "learning_rate": 0.00028563623357796597,
      "loss": 2.2744,
      "step": 4196
    },
    {
      "epoch": 0.6519611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0002854089484502591,
      "loss": 2.1427,
      "step": 4197
    },
    {
      "epoch": 0.6521165048543689,
      "grad_norm": 0.0,
      "learning_rate": 0.00028518171766318313,
      "loss": 2.2289,
      "step": 4198
    },
    {
      "epoch": 0.6522718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.0002849545412742789,
      "loss": 2.2664,
      "step": 4199
    },
    {
      "epoch": 0.6524271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.00028472741934107437,
      "loss": 2.0742,
      "step": 4200
    },
    {
      "epoch": 0.6525825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.00028450035192108334,
      "loss": 2.0949,
      "step": 4201
    },
    {
      "epoch": 0.6527378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0002842733390718057,
      "loss": 2.1508,
      "step": 4202
    },
    {
      "epoch": 0.6528932038834951,
      "grad_norm": 0.0,
      "learning_rate": 0.0002840463808507279,
      "loss": 2.1093,
      "step": 4203
    },
    {
      "epoch": 0.6530485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.00028381947731532175,
      "loss": 2.2401,
      "step": 4204
    },
    {
      "epoch": 0.6532038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.00028359262852304637,
      "loss": 2.0704,
      "step": 4205
    },
    {
      "epoch": 0.6533592233009708,
      "grad_norm": 0.0,
      "learning_rate": 0.00028336583453134604,
      "loss": 2.0979,
      "step": 4206
    },
    {
      "epoch": 0.6535145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.0002831390953976515,
      "loss": 2.2727,
      "step": 4207
    },
    {
      "epoch": 0.6536699029126214,
      "grad_norm": 0.0,
      "learning_rate": 0.0002829124111793799,
      "loss": 2.2507,
      "step": 4208
    },
    {
      "epoch": 0.6538252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0002826857819339338,
      "loss": 2.1455,
      "step": 4209
    },
    {
      "epoch": 0.6539805825242718,
      "grad_norm": 0.0,
      "learning_rate": 0.00028245920771870247,
      "loss": 2.1401,
      "step": 4210
    },
    {
      "epoch": 0.6541359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.00028223268859106133,
      "loss": 2.0674,
      "step": 4211
    },
    {
      "epoch": 0.6542912621359224,
      "grad_norm": 0.0,
      "learning_rate": 0.000282006224608371,
      "loss": 2.2238,
      "step": 4212
    },
    {
      "epoch": 0.6544466019417475,
      "grad_norm": 0.0,
      "learning_rate": 0.00028177981582797913,
      "loss": 2.2476,
      "step": 4213
    },
    {
      "epoch": 0.6546019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.0002815534623072186,
      "loss": 2.2557,
      "step": 4214
    },
    {
      "epoch": 0.6547572815533981,
      "grad_norm": 0.0,
      "learning_rate": 0.00028132716410340865,
      "loss": 2.125,
      "step": 4215
    },
    {
      "epoch": 0.6549126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0002811009212738549,
      "loss": 2.1506,
      "step": 4216
    },
    {
      "epoch": 0.6550679611650485,
      "grad_norm": 0.0,
      "learning_rate": 0.0002808747338758481,
      "loss": 2.147,
      "step": 4217
    },
    {
      "epoch": 0.6552233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.00028064860196666574,
      "loss": 2.077,
      "step": 4218
    },
    {
      "epoch": 0.655378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0002804225256035707,
      "loss": 2.1993,
      "step": 4219
    },
    {
      "epoch": 0.6555339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.000280196504843812,
      "loss": 2.0486,
      "step": 4220
    },
    {
      "epoch": 0.6556893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.00027997053974462496,
      "loss": 2.0697,
      "step": 4221
    },
    {
      "epoch": 0.6558446601941748,
      "grad_norm": 0.0,
      "learning_rate": 0.00027974463036322983,
      "loss": 2.1901,
      "step": 4222
    },
    {
      "epoch": 0.656,
      "grad_norm": 0.0,
      "learning_rate": 0.00027951877675683396,
      "loss": 2.1774,
      "step": 4223
    },
    {
      "epoch": 0.6561553398058252,
      "grad_norm": 0.0,
      "learning_rate": 0.0002792929789826295,
      "loss": 2.0419,
      "step": 4224
    },
    {
      "epoch": 0.6563106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.00027906723709779526,
      "loss": 2.1089,
      "step": 4225
    },
    {
      "epoch": 0.6564660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.0002788415511594957,
      "loss": 2.2174,
      "step": 4226
    },
    {
      "epoch": 0.656621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0002786159212248806,
      "loss": 2.1238,
      "step": 4227
    },
    {
      "epoch": 0.6567766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.00027839034735108637,
      "loss": 2.2725,
      "step": 4228
    },
    {
      "epoch": 0.6569320388349514,
      "grad_norm": 0.0,
      "learning_rate": 0.00027816482959523454,
      "loss": 2.2081,
      "step": 4229
    },
    {
      "epoch": 0.6570873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0002779393680144325,
      "loss": 2.1567,
      "step": 4230
    },
    {
      "epoch": 0.657242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0002777139626657742,
      "loss": 2.1165,
      "step": 4231
    },
    {
      "epoch": 0.6573980582524271,
      "grad_norm": 0.0,
      "learning_rate": 0.00027748861360633815,
      "loss": 2.1202,
      "step": 4232
    },
    {
      "epoch": 0.6575533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.0002772633208931895,
      "loss": 2.174,
      "step": 4233
    },
    {
      "epoch": 0.6577087378640777,
      "grad_norm": 0.0,
      "learning_rate": 0.0002770380845833791,
      "loss": 2.2516,
      "step": 4234
    },
    {
      "epoch": 0.657864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0002768129047339427,
      "loss": 2.22,
      "step": 4235
    },
    {
      "epoch": 0.6580194174757281,
      "grad_norm": 0.0,
      "learning_rate": 0.00027658778140190294,
      "loss": 2.1764,
      "step": 4236
    },
    {
      "epoch": 0.6581747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.000276362714644267,
      "loss": 2.2333,
      "step": 4237
    },
    {
      "epoch": 0.6583300970873787,
      "grad_norm": 0.0,
      "learning_rate": 0.00027613770451802864,
      "loss": 2.1624,
      "step": 4238
    },
    {
      "epoch": 0.6584854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.0002759127510801665,
      "loss": 2.2513,
      "step": 4239
    },
    {
      "epoch": 0.6586407766990291,
      "grad_norm": 0.0,
      "learning_rate": 0.00027568785438764575,
      "loss": 2.2233,
      "step": 4240
    },
    {
      "epoch": 0.6587961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.0002754630144974164,
      "loss": 2.1752,
      "step": 4241
    },
    {
      "epoch": 0.6589514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0002752382314664143,
      "loss": 2.2156,
      "step": 4242
    },
    {
      "epoch": 0.6591067961165048,
      "grad_norm": 0.0,
      "learning_rate": 0.0002750135053515611,
      "loss": 2.1584,
      "step": 4243
    },
    {
      "epoch": 0.6592621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0002747888362097641,
      "loss": 2.1066,
      "step": 4244
    },
    {
      "epoch": 0.6594174757281553,
      "grad_norm": 0.0,
      "learning_rate": 0.0002745642240979157,
      "loss": 2.1418,
      "step": 4245
    },
    {
      "epoch": 0.6595728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.00027433966907289454,
      "loss": 2.1978,
      "step": 4246
    },
    {
      "epoch": 0.6597281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.0002741151711915638,
      "loss": 2.2037,
      "step": 4247
    },
    {
      "epoch": 0.659883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0002738907305107733,
      "loss": 2.0778,
      "step": 4248
    },
    {
      "epoch": 0.6600388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0002736663470873579,
      "loss": 2.1323,
      "step": 4249
    },
    {
      "epoch": 0.6601941747572816,
      "grad_norm": 0.0,
      "learning_rate": 0.0002734420209781376,
      "loss": 2.2697,
      "step": 4250
    },
    {
      "epoch": 0.6603495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0002732177522399186,
      "loss": 2.1436,
      "step": 4251
    },
    {
      "epoch": 0.660504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.00027299354092949203,
      "loss": 2.1667,
      "step": 4252
    },
    {
      "epoch": 0.6606601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.00027276938710363443,
      "loss": 2.0643,
      "step": 4253
    },
    {
      "epoch": 0.6608155339805826,
      "grad_norm": 0.0,
      "learning_rate": 0.0002725452908191084,
      "loss": 2.1909,
      "step": 4254
    },
    {
      "epoch": 0.6609708737864077,
      "grad_norm": 0.0,
      "learning_rate": 0.0002723212521326611,
      "loss": 2.2043,
      "step": 4255
    },
    {
      "epoch": 0.661126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0002720972711010262,
      "loss": 2.1287,
      "step": 4256
    },
    {
      "epoch": 0.6612815533980583,
      "grad_norm": 0.0,
      "learning_rate": 0.00027187334778092147,
      "loss": 2.1063,
      "step": 4257
    },
    {
      "epoch": 0.6614368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.00027164948222905107,
      "loss": 2.2236,
      "step": 4258
    },
    {
      "epoch": 0.6615922330097087,
      "grad_norm": 0.0,
      "learning_rate": 0.00027142567450210445,
      "loss": 2.2201,
      "step": 4259
    },
    {
      "epoch": 0.661747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0002712019246567556,
      "loss": 2.1937,
      "step": 4260
    },
    {
      "epoch": 0.6619029126213593,
      "grad_norm": 0.0,
      "learning_rate": 0.00027097823274966503,
      "loss": 2.2943,
      "step": 4261
    },
    {
      "epoch": 0.6620582524271844,
      "grad_norm": 0.0,
      "learning_rate": 0.00027075459883747744,
      "loss": 2.1878,
      "step": 4262
    },
    {
      "epoch": 0.6622135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.00027053102297682386,
      "loss": 2.1184,
      "step": 4263
    },
    {
      "epoch": 0.662368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0002703075052243198,
      "loss": 2.1708,
      "step": 4264
    },
    {
      "epoch": 0.6625242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0002700840456365663,
      "loss": 2.1841,
      "step": 4265
    },
    {
      "epoch": 0.6626796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.0002698606442701499,
      "loss": 2.2264,
      "step": 4266
    },
    {
      "epoch": 0.6628349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.00026963730118164243,
      "loss": 2.2208,
      "step": 4267
    },
    {
      "epoch": 0.6629902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.00026941401642760054,
      "loss": 2.1467,
      "step": 4268
    },
    {
      "epoch": 0.6631456310679612,
      "grad_norm": 0.0,
      "learning_rate": 0.0002691907900645666,
      "loss": 2.0386,
      "step": 4269
    },
    {
      "epoch": 0.6633009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.00026896762214906756,
      "loss": 2.1266,
      "step": 4270
    },
    {
      "epoch": 0.6634563106796116,
      "grad_norm": 0.0,
      "learning_rate": 0.0002687445127376165,
      "loss": 2.1543,
      "step": 4271
    },
    {
      "epoch": 0.6636116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.00026852146188671055,
      "loss": 2.2375,
      "step": 4272
    },
    {
      "epoch": 0.6637669902912622,
      "grad_norm": 0.0,
      "learning_rate": 0.00026829846965283294,
      "loss": 2.149,
      "step": 4273
    },
    {
      "epoch": 0.6639223300970873,
      "grad_norm": 0.0,
      "learning_rate": 0.0002680755360924519,
      "loss": 2.1279,
      "step": 4274
    },
    {
      "epoch": 0.6640776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.00026785266126202036,
      "loss": 2.1238,
      "step": 4275
    },
    {
      "epoch": 0.6642330097087379,
      "grad_norm": 0.0,
      "learning_rate": 0.0002676298452179765,
      "loss": 2.1926,
      "step": 4276
    },
    {
      "epoch": 0.6643883495145632,
      "grad_norm": 0.0,
      "learning_rate": 0.00026740708801674417,
      "loss": 2.1195,
      "step": 4277
    },
    {
      "epoch": 0.6645436893203883,
      "grad_norm": 0.0,
      "learning_rate": 0.0002671843897147315,
      "loss": 2.0745,
      "step": 4278
    },
    {
      "epoch": 0.6646990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0002669617503683325,
      "loss": 2.1752,
      "step": 4279
    },
    {
      "epoch": 0.6648543689320389,
      "grad_norm": 0.0,
      "learning_rate": 0.00026673917003392547,
      "loss": 2.1616,
      "step": 4280
    },
    {
      "epoch": 0.665009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.00026651664876787423,
      "loss": 2.1221,
      "step": 4281
    },
    {
      "epoch": 0.6651650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.00026629418662652796,
      "loss": 2.0815,
      "step": 4282
    },
    {
      "epoch": 0.6653203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.00026607178366621996,
      "loss": 2.1371,
      "step": 4283
    },
    {
      "epoch": 0.6654757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.0002658494399432695,
      "loss": 2.2078,
      "step": 4284
    },
    {
      "epoch": 0.665631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.00026562715551398,
      "loss": 2.2186,
      "step": 4285
    },
    {
      "epoch": 0.6657864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.00026540493043464064,
      "loss": 2.0972,
      "step": 4286
    },
    {
      "epoch": 0.6659417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.00026518276476152514,
      "loss": 2.0639,
      "step": 4287
    },
    {
      "epoch": 0.6660970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.0002649606585508919,
      "loss": 2.1189,
      "step": 4288
    },
    {
      "epoch": 0.666252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.00026473861185898507,
      "loss": 2.0848,
      "step": 4289
    },
    {
      "epoch": 0.6664077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.0002645166247420328,
      "loss": 2.2213,
      "step": 4290
    },
    {
      "epoch": 0.6665631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0002642946972562489,
      "loss": 2.1802,
      "step": 4291
    },
    {
      "epoch": 0.6667184466019418,
      "grad_norm": 0.0,
      "learning_rate": 0.00026407282945783197,
      "loss": 2.165,
      "step": 4292
    },
    {
      "epoch": 0.666873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0002638510214029649,
      "loss": 2.156,
      "step": 4293
    },
    {
      "epoch": 0.6670291262135922,
      "grad_norm": 0.0,
      "learning_rate": 0.0002636292731478163,
      "loss": 2.166,
      "step": 4294
    },
    {
      "epoch": 0.6671844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.00026340758474853877,
      "loss": 2.194,
      "step": 4295
    },
    {
      "epoch": 0.6673398058252428,
      "grad_norm": 0.0,
      "learning_rate": 0.0002631859562612705,
      "loss": 2.191,
      "step": 4296
    },
    {
      "epoch": 0.6674951456310679,
      "grad_norm": 0.0,
      "learning_rate": 0.00026296438774213427,
      "loss": 2.2005,
      "step": 4297
    },
    {
      "epoch": 0.6676504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.00026274287924723746,
      "loss": 2.1763,
      "step": 4298
    },
    {
      "epoch": 0.6678058252427185,
      "grad_norm": 0.0,
      "learning_rate": 0.00026252143083267215,
      "loss": 2.1333,
      "step": 4299
    },
    {
      "epoch": 0.6679611650485436,
      "grad_norm": 0.0,
      "learning_rate": 0.00026230004255451593,
      "loss": 2.0931,
      "step": 4300
    },
    {
      "epoch": 0.6681165048543689,
      "grad_norm": 0.0,
      "learning_rate": 0.0002620787144688303,
      "loss": 2.1696,
      "step": 4301
    },
    {
      "epoch": 0.6682718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.0002618574466316622,
      "loss": 2.1707,
      "step": 4302
    },
    {
      "epoch": 0.6684271844660195,
      "grad_norm": 0.0,
      "learning_rate": 0.00026163623909904274,
      "loss": 2.1328,
      "step": 4303
    },
    {
      "epoch": 0.6685825242718446,
      "grad_norm": 0.0,
      "learning_rate": 0.0002614150919269883,
      "loss": 2.0906,
      "step": 4304
    },
    {
      "epoch": 0.6687378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.00026119400517149937,
      "loss": 2.1507,
      "step": 4305
    },
    {
      "epoch": 0.6688932038834952,
      "grad_norm": 0.0,
      "learning_rate": 0.0002609729788885617,
      "loss": 2.215,
      "step": 4306
    },
    {
      "epoch": 0.6690485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0002607520131341457,
      "loss": 2.2108,
      "step": 4307
    },
    {
      "epoch": 0.6692038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.00026053110796420587,
      "loss": 2.1688,
      "step": 4308
    },
    {
      "epoch": 0.6693592233009709,
      "grad_norm": 0.0,
      "learning_rate": 0.00026031026343468203,
      "loss": 2.1759,
      "step": 4309
    },
    {
      "epoch": 0.6695145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.00026008947960149835,
      "loss": 2.147,
      "step": 4310
    },
    {
      "epoch": 0.6696699029126214,
      "grad_norm": 0.0,
      "learning_rate": 0.0002598687565205634,
      "loss": 2.1304,
      "step": 4311
    },
    {
      "epoch": 0.6698252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.00025964809424777096,
      "loss": 2.1906,
      "step": 4312
    },
    {
      "epoch": 0.6699805825242718,
      "grad_norm": 0.0,
      "learning_rate": 0.0002594274928389987,
      "loss": 2.2269,
      "step": 4313
    },
    {
      "epoch": 0.6701359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.00025920695235010947,
      "loss": 2.1955,
      "step": 4314
    },
    {
      "epoch": 0.6702912621359224,
      "grad_norm": 0.0,
      "learning_rate": 0.00025898647283695064,
      "loss": 2.0526,
      "step": 4315
    },
    {
      "epoch": 0.6704466019417475,
      "grad_norm": 0.0,
      "learning_rate": 0.0002587660543553536,
      "loss": 2.2169,
      "step": 4316
    },
    {
      "epoch": 0.6706019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.000258545696961135,
      "loss": 2.1842,
      "step": 4317
    },
    {
      "epoch": 0.6707572815533981,
      "grad_norm": 0.0,
      "learning_rate": 0.0002583254007100953,
      "loss": 2.1725,
      "step": 4318
    },
    {
      "epoch": 0.6709126213592232,
      "grad_norm": 0.0,
      "learning_rate": 0.00025810516565802013,
      "loss": 2.156,
      "step": 4319
    },
    {
      "epoch": 0.6710679611650485,
      "grad_norm": 0.0,
      "learning_rate": 0.0002578849918606796,
      "loss": 2.1627,
      "step": 4320
    },
    {
      "epoch": 0.6712233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.00025766487937382734,
      "loss": 2.1612,
      "step": 4321
    },
    {
      "epoch": 0.6713786407766991,
      "grad_norm": 0.0,
      "learning_rate": 0.0002574448282532027,
      "loss": 2.1606,
      "step": 4322
    },
    {
      "epoch": 0.6715339805825242,
      "grad_norm": 0.0,
      "learning_rate": 0.0002572248385545287,
      "loss": 2.0951,
      "step": 4323
    },
    {
      "epoch": 0.6716893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0002570049103335131,
      "loss": 2.2095,
      "step": 4324
    },
    {
      "epoch": 0.6718446601941748,
      "grad_norm": 0.0,
      "learning_rate": 0.0002567850436458483,
      "loss": 2.2095,
      "step": 4325
    },
    {
      "epoch": 0.672,
      "grad_norm": 0.0,
      "learning_rate": 0.00025656523854721057,
      "loss": 2.0428,
      "step": 4326
    },
    {
      "epoch": 0.6721553398058252,
      "grad_norm": 0.0,
      "learning_rate": 0.0002563454950932612,
      "loss": 2.2035,
      "step": 4327
    },
    {
      "epoch": 0.6723106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.00025612581333964515,
      "loss": 2.2075,
      "step": 4328
    },
    {
      "epoch": 0.6724660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.00025590619334199245,
      "loss": 2.077,
      "step": 4329
    },
    {
      "epoch": 0.672621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.00025568663515591726,
      "loss": 2.0135,
      "step": 4330
    },
    {
      "epoch": 0.6727766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0002554671388370178,
      "loss": 2.1254,
      "step": 4331
    },
    {
      "epoch": 0.6729320388349515,
      "grad_norm": 0.0,
      "learning_rate": 0.0002552477044408771,
      "loss": 2.2543,
      "step": 4332
    },
    {
      "epoch": 0.6730873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.00025502833202306217,
      "loss": 2.0946,
      "step": 4333
    },
    {
      "epoch": 0.673242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0002548090216391242,
      "loss": 2.2253,
      "step": 4334
    },
    {
      "epoch": 0.6733980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0002545897733445992,
      "loss": 2.1719,
      "step": 4335
    },
    {
      "epoch": 0.6735533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.0002543705871950069,
      "loss": 2.0874,
      "step": 4336
    },
    {
      "epoch": 0.6737087378640777,
      "grad_norm": 0.0,
      "learning_rate": 0.0002541514632458518,
      "loss": 2.1049,
      "step": 4337
    },
    {
      "epoch": 0.6738640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.00025393240155262216,
      "loss": 2.094,
      "step": 4338
    },
    {
      "epoch": 0.6740194174757281,
      "grad_norm": 0.0,
      "learning_rate": 0.00025371340217079073,
      "loss": 2.2426,
      "step": 4339
    },
    {
      "epoch": 0.6741747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.00025349446515581474,
      "loss": 2.0863,
      "step": 4340
    },
    {
      "epoch": 0.6743300970873787,
      "grad_norm": 0.0,
      "learning_rate": 0.000253275590563135,
      "loss": 2.1639,
      "step": 4341
    },
    {
      "epoch": 0.6744854368932038,
      "grad_norm": 0.0,
      "learning_rate": 0.0002530567784481771,
      "loss": 2.0945,
      "step": 4342
    },
    {
      "epoch": 0.6746407766990291,
      "grad_norm": 0.0,
      "learning_rate": 0.0002528380288663504,
      "loss": 2.0829,
      "step": 4343
    },
    {
      "epoch": 0.6747961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.0002526193418730486,
      "loss": 2.0972,
      "step": 4344
    },
    {
      "epoch": 0.6749514563106797,
      "grad_norm": 0.0,
      "learning_rate": 0.00025240071752364966,
      "loss": 2.1991,
      "step": 4345
    },
    {
      "epoch": 0.6751067961165048,
      "grad_norm": 0.0,
      "learning_rate": 0.0002521821558735154,
      "loss": 2.1428,
      "step": 4346
    },
    {
      "epoch": 0.6752621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.00025196365697799193,
      "loss": 2.1318,
      "step": 4347
    },
    {
      "epoch": 0.6754174757281554,
      "grad_norm": 0.0,
      "learning_rate": 0.0002517452208924098,
      "loss": 2.1562,
      "step": 4348
    },
    {
      "epoch": 0.6755728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.00025152684767208277,
      "loss": 2.1029,
      "step": 4349
    },
    {
      "epoch": 0.6757281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.00025130853737230974,
      "loss": 2.2264,
      "step": 4350
    },
    {
      "epoch": 0.6758834951456311,
      "grad_norm": 0.0,
      "learning_rate": 0.00025109029004837267,
      "loss": 2.0614,
      "step": 4351
    },
    {
      "epoch": 0.6760388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0002508721057555384,
      "loss": 2.2844,
      "step": 4352
    },
    {
      "epoch": 0.6761941747572816,
      "grad_norm": 0.0,
      "learning_rate": 0.0002506539845490575,
      "loss": 2.1657,
      "step": 4353
    },
    {
      "epoch": 0.6763495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.00025043592648416437,
      "loss": 2.2115,
      "step": 4354
    },
    {
      "epoch": 0.676504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0002502179316160779,
      "loss": 2.2015,
      "step": 4355
    },
    {
      "epoch": 0.6766601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.0002500000000000001,
      "loss": 2.056,
      "step": 4356
    },
    {
      "epoch": 0.6768155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.0002497821316911178,
      "loss": 2.1073,
      "step": 4357
    },
    {
      "epoch": 0.6769708737864077,
      "grad_norm": 0.0,
      "learning_rate": 0.0002495643267446018,
      "loss": 2.1821,
      "step": 4358
    },
    {
      "epoch": 0.677126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0002493465852156061,
      "loss": 2.1399,
      "step": 4359
    },
    {
      "epoch": 0.6772815533980583,
      "grad_norm": 0.0,
      "learning_rate": 0.0002491289071592696,
      "loss": 2.1482,
      "step": 4360
    },
    {
      "epoch": 0.6774368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0002489112926307142,
      "loss": 2.1025,
      "step": 4361
    },
    {
      "epoch": 0.6775922330097087,
      "grad_norm": 0.0,
      "learning_rate": 0.0002486937416850464,
      "loss": 2.1632,
      "step": 4362
    },
    {
      "epoch": 0.677747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.00024847625437735655,
      "loss": 2.115,
      "step": 4363
    },
    {
      "epoch": 0.6779029126213593,
      "grad_norm": 0.0,
      "learning_rate": 0.00024825883076271834,
      "loss": 2.0366,
      "step": 4364
    },
    {
      "epoch": 0.6780582524271844,
      "grad_norm": 0.0,
      "learning_rate": 0.00024804147089618994,
      "loss": 2.1512,
      "step": 4365
    },
    {
      "epoch": 0.6782135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.00024782417483281305,
      "loss": 2.192,
      "step": 4366
    },
    {
      "epoch": 0.678368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.00024760694262761305,
      "loss": 2.0609,
      "step": 4367
    },
    {
      "epoch": 0.6785242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0002473897743355998,
      "loss": 2.0553,
      "step": 4368
    },
    {
      "epoch": 0.6786796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.00024717267001176616,
      "loss": 2.1344,
      "step": 4369
    },
    {
      "epoch": 0.6788349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.00024695562971108957,
      "loss": 2.1498,
      "step": 4370
    },
    {
      "epoch": 0.678990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0002467386534885305,
      "loss": 2.1859,
      "step": 4371
    },
    {
      "epoch": 0.6791456310679612,
      "grad_norm": 0.0,
      "learning_rate": 0.00024652174139903383,
      "loss": 2.1708,
      "step": 4372
    },
    {
      "epoch": 0.6793009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.00024630489349752806,
      "loss": 2.1259,
      "step": 4373
    },
    {
      "epoch": 0.6794563106796117,
      "grad_norm": 0.0,
      "learning_rate": 0.00024608810983892495,
      "loss": 2.02,
      "step": 4374
    },
    {
      "epoch": 0.6796116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0002458713904781207,
      "loss": 2.1457,
      "step": 4375
    },
    {
      "epoch": 0.6797669902912621,
      "grad_norm": 0.0,
      "learning_rate": 0.00024565473546999463,
      "loss": 2.1588,
      "step": 4376
    },
    {
      "epoch": 0.6799223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0002454381448694103,
      "loss": 2.1046,
      "step": 4377
    },
    {
      "epoch": 0.6800776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.0002452216187312145,
      "loss": 2.1141,
      "step": 4378
    },
    {
      "epoch": 0.6802330097087379,
      "grad_norm": 0.0,
      "learning_rate": 0.0002450051571102378,
      "loss": 2.1707,
      "step": 4379
    },
    {
      "epoch": 0.6803883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.00024478876006129457,
      "loss": 2.1493,
      "step": 4380
    },
    {
      "epoch": 0.6805436893203883,
      "grad_norm": 0.0,
      "learning_rate": 0.00024457242763918305,
      "loss": 2.1808,
      "step": 4381
    },
    {
      "epoch": 0.6806990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0002443561598986844,
      "loss": 2.1715,
      "step": 4382
    },
    {
      "epoch": 0.6808543689320389,
      "grad_norm": 0.0,
      "learning_rate": 0.0002441399568945642,
      "loss": 2.1003,
      "step": 4383
    },
    {
      "epoch": 0.681009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.00024392381868157094,
      "loss": 2.109,
      "step": 4384
    },
    {
      "epoch": 0.6811650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.00024370774531443725,
      "loss": 2.1376,
      "step": 4385
    },
    {
      "epoch": 0.6813203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.00024349173684787924,
      "loss": 2.0852,
      "step": 4386
    },
    {
      "epoch": 0.6814757281553399,
      "grad_norm": 0.0,
      "learning_rate": 0.00024327579333659617,
      "loss": 2.0776,
      "step": 4387
    },
    {
      "epoch": 0.681631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.00024305991483527145,
      "loss": 2.2418,
      "step": 4388
    },
    {
      "epoch": 0.6817864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0002428441013985716,
      "loss": 2.1006,
      "step": 4389
    },
    {
      "epoch": 0.6819417475728156,
      "grad_norm": 0.0,
      "learning_rate": 0.00024262835308114662,
      "loss": 2.1502,
      "step": 4390
    },
    {
      "epoch": 0.6820970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.0002424126699376305,
      "loss": 2.1137,
      "step": 4391
    },
    {
      "epoch": 0.682252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0002421970520226401,
      "loss": 2.1592,
      "step": 4392
    },
    {
      "epoch": 0.6824077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.0002419814993907764,
      "loss": 2.0979,
      "step": 4393
    },
    {
      "epoch": 0.6825631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.00024176601209662324,
      "loss": 2.1514,
      "step": 4394
    },
    {
      "epoch": 0.6827184466019417,
      "grad_norm": 0.0,
      "learning_rate": 0.00024155059019474834,
      "loss": 2.1696,
      "step": 4395
    },
    {
      "epoch": 0.682873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.00024133523373970296,
      "loss": 1.9604,
      "step": 4396
    },
    {
      "epoch": 0.6830291262135922,
      "grad_norm": 0.0,
      "learning_rate": 0.0002411199427860211,
      "loss": 2.1973,
      "step": 4397
    },
    {
      "epoch": 0.6831844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.00024090471738822105,
      "loss": 2.1367,
      "step": 4398
    },
    {
      "epoch": 0.6833398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.00024068955760080368,
      "loss": 2.1887,
      "step": 4399
    },
    {
      "epoch": 0.683495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.00024047446347825403,
      "loss": 2.1648,
      "step": 4400
    },
    {
      "epoch": 0.6836504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.00024025943507503995,
      "loss": 2.1504,
      "step": 4401
    },
    {
      "epoch": 0.6838058252427185,
      "grad_norm": 0.0,
      "learning_rate": 0.00024004447244561263,
      "loss": 2.2266,
      "step": 4402
    },
    {
      "epoch": 0.6839611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0002398295756444069,
      "loss": 2.1579,
      "step": 4403
    },
    {
      "epoch": 0.6841165048543689,
      "grad_norm": 0.0,
      "learning_rate": 0.0002396147447258411,
      "loss": 2.1935,
      "step": 4404
    },
    {
      "epoch": 0.6842718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.0002393999797443162,
      "loss": 2.1625,
      "step": 4405
    },
    {
      "epoch": 0.6844271844660195,
      "grad_norm": 0.0,
      "learning_rate": 0.0002391852807542173,
      "loss": 2.2054,
      "step": 4406
    },
    {
      "epoch": 0.6845825242718446,
      "grad_norm": 0.0,
      "learning_rate": 0.00023897064780991186,
      "loss": 2.2074,
      "step": 4407
    },
    {
      "epoch": 0.6847378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.00023875608096575164,
      "loss": 2.1742,
      "step": 4408
    },
    {
      "epoch": 0.6848932038834952,
      "grad_norm": 0.0,
      "learning_rate": 0.00023854158027607065,
      "loss": 2.1541,
      "step": 4409
    },
    {
      "epoch": 0.6850485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0002383271457951869,
      "loss": 2.1462,
      "step": 4410
    },
    {
      "epoch": 0.6852038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.00023811277757740146,
      "loss": 2.0904,
      "step": 4411
    },
    {
      "epoch": 0.6853592233009709,
      "grad_norm": 0.0,
      "learning_rate": 0.00023789847567699847,
      "loss": 2.1798,
      "step": 4412
    },
    {
      "epoch": 0.6855145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.0002376842401482451,
      "loss": 2.1646,
      "step": 4413
    },
    {
      "epoch": 0.6856699029126213,
      "grad_norm": 0.0,
      "learning_rate": 0.00023747007104539232,
      "loss": 2.1478,
      "step": 4414
    },
    {
      "epoch": 0.6858252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.00023725596842267356,
      "loss": 2.1513,
      "step": 4415
    },
    {
      "epoch": 0.6859805825242719,
      "grad_norm": 0.0,
      "learning_rate": 0.0002370419323343061,
      "loss": 2.1326,
      "step": 4416
    },
    {
      "epoch": 0.6861359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0002368279628344897,
      "loss": 2.134,
      "step": 4417
    },
    {
      "epoch": 0.6862912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.00023661405997740776,
      "loss": 2.1125,
      "step": 4418
    },
    {
      "epoch": 0.6864466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.00023640022381722682,
      "loss": 2.1723,
      "step": 4419
    },
    {
      "epoch": 0.6866019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.000236186454408096,
      "loss": 2.1805,
      "step": 4420
    },
    {
      "epoch": 0.6867572815533981,
      "grad_norm": 0.0,
      "learning_rate": 0.0002359727518041483,
      "loss": 2.1838,
      "step": 4421
    },
    {
      "epoch": 0.6869126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.00023575911605949883,
      "loss": 2.1299,
      "step": 4422
    },
    {
      "epoch": 0.6870679611650485,
      "grad_norm": 0.0,
      "learning_rate": 0.00023554554722824683,
      "loss": 2.1248,
      "step": 4423
    },
    {
      "epoch": 0.6872233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.00023533204536447383,
      "loss": 2.1184,
      "step": 4424
    },
    {
      "epoch": 0.6873786407766991,
      "grad_norm": 0.0,
      "learning_rate": 0.00023511861052224448,
      "loss": 2.0928,
      "step": 4425
    },
    {
      "epoch": 0.6875339805825242,
      "grad_norm": 0.0,
      "learning_rate": 0.00023490524275560705,
      "loss": 2.194,
      "step": 4426
    },
    {
      "epoch": 0.6876893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.00023469194211859186,
      "loss": 2.2422,
      "step": 4427
    },
    {
      "epoch": 0.6878446601941748,
      "grad_norm": 0.0,
      "learning_rate": 0.00023447870866521314,
      "loss": 2.1238,
      "step": 4428
    },
    {
      "epoch": 0.688,
      "grad_norm": 0.0,
      "learning_rate": 0.0002342655424494678,
      "loss": 2.0361,
      "step": 4429
    },
    {
      "epoch": 0.6881553398058252,
      "grad_norm": 0.0,
      "learning_rate": 0.00023405244352533538,
      "loss": 2.172,
      "step": 4430
    },
    {
      "epoch": 0.6883106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.00023383941194677894,
      "loss": 2.1615,
      "step": 4431
    },
    {
      "epoch": 0.6884660194174758,
      "grad_norm": 0.0,
      "learning_rate": 0.00023362644776774393,
      "loss": 2.1636,
      "step": 4432
    },
    {
      "epoch": 0.6886213592233009,
      "grad_norm": 0.0,
      "learning_rate": 0.0002334135510421591,
      "loss": 2.2548,
      "step": 4433
    },
    {
      "epoch": 0.6887766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.00023320072182393654,
      "loss": 2.0646,
      "step": 4434
    },
    {
      "epoch": 0.6889320388349515,
      "grad_norm": 0.0,
      "learning_rate": 0.00023298796016696981,
      "loss": 2.1287,
      "step": 4435
    },
    {
      "epoch": 0.6890873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.00023277526612513677,
      "loss": 2.3127,
      "step": 4436
    },
    {
      "epoch": 0.6892427184466019,
      "grad_norm": 0.0,
      "learning_rate": 0.00023256263975229792,
      "loss": 2.197,
      "step": 4437
    },
    {
      "epoch": 0.6893980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.00023235008110229584,
      "loss": 2.0935,
      "step": 4438
    },
    {
      "epoch": 0.6895533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.00023213759022895698,
      "loss": 2.1174,
      "step": 4439
    },
    {
      "epoch": 0.6897087378640777,
      "grad_norm": 0.0,
      "learning_rate": 0.0002319251671860897,
      "loss": 2.2174,
      "step": 4440
    },
    {
      "epoch": 0.6898640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.00023171281202748607,
      "loss": 2.1379,
      "step": 4441
    },
    {
      "epoch": 0.6900194174757281,
      "grad_norm": 0.0,
      "learning_rate": 0.0002315005248069202,
      "loss": 2.1373,
      "step": 4442
    },
    {
      "epoch": 0.6901747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.00023128830557814933,
      "loss": 2.0723,
      "step": 4443
    },
    {
      "epoch": 0.6903300970873787,
      "grad_norm": 0.0,
      "learning_rate": 0.00023107615439491385,
      "loss": 2.189,
      "step": 4444
    },
    {
      "epoch": 0.6904854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.00023086407131093606,
      "loss": 2.1453,
      "step": 4445
    },
    {
      "epoch": 0.6906407766990291,
      "grad_norm": 0.0,
      "learning_rate": 0.00023065205637992197,
      "loss": 2.1317,
      "step": 4446
    },
    {
      "epoch": 0.6907961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.00023044010965555956,
      "loss": 1.985,
      "step": 4447
    },
    {
      "epoch": 0.6909514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.00023022823119151975,
      "loss": 2.2422,
      "step": 4448
    },
    {
      "epoch": 0.6911067961165048,
      "grad_norm": 0.0,
      "learning_rate": 0.0002300164210414566,
      "loss": 2.1224,
      "step": 4449
    },
    {
      "epoch": 0.6912621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.00022980467925900618,
      "loss": 2.168,
      "step": 4450
    },
    {
      "epoch": 0.6914174757281554,
      "grad_norm": 0.0,
      "learning_rate": 0.00022959300589778776,
      "loss": 2.2171,
      "step": 4451
    },
    {
      "epoch": 0.6915728155339805,
      "grad_norm": 0.0,
      "learning_rate": 0.0002293814010114033,
      "loss": 2.1956,
      "step": 4452
    },
    {
      "epoch": 0.6917281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.00022916986465343692,
      "loss": 2.1599,
      "step": 4453
    },
    {
      "epoch": 0.6918834951456311,
      "grad_norm": 0.0,
      "learning_rate": 0.0002289583968774561,
      "loss": 2.2233,
      "step": 4454
    },
    {
      "epoch": 0.6920388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.00022874699773701014,
      "loss": 2.1084,
      "step": 4455
    },
    {
      "epoch": 0.6921941747572815,
      "grad_norm": 0.0,
      "learning_rate": 0.00022853566728563179,
      "loss": 2.0685,
      "step": 4456
    },
    {
      "epoch": 0.6923495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0002283244055768358,
      "loss": 2.1581,
      "step": 4457
    },
    {
      "epoch": 0.692504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.0002281132126641196,
      "loss": 2.2986,
      "step": 4458
    },
    {
      "epoch": 0.6926601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.00022790208860096358,
      "loss": 2.1442,
      "step": 4459
    },
    {
      "epoch": 0.6928155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.00022769103344083007,
      "loss": 2.0522,
      "step": 4460
    },
    {
      "epoch": 0.6929708737864078,
      "grad_norm": 0.0,
      "learning_rate": 0.00022748004723716458,
      "loss": 2.2665,
      "step": 4461
    },
    {
      "epoch": 0.693126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.00022726913004339505,
      "loss": 2.1454,
      "step": 4462
    },
    {
      "epoch": 0.6932815533980583,
      "grad_norm": 0.0,
      "learning_rate": 0.00022705828191293137,
      "loss": 2.1255,
      "step": 4463
    },
    {
      "epoch": 0.6934368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.00022684750289916672,
      "loss": 2.0789,
      "step": 4464
    },
    {
      "epoch": 0.6935922330097087,
      "grad_norm": 0.0,
      "learning_rate": 0.00022663679305547614,
      "loss": 2.2065,
      "step": 4465
    },
    {
      "epoch": 0.693747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.00022642615243521747,
      "loss": 2.174,
      "step": 4466
    },
    {
      "epoch": 0.6939029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.00022621558109173123,
      "loss": 2.2534,
      "step": 4467
    },
    {
      "epoch": 0.6940582524271844,
      "grad_norm": 0.0,
      "learning_rate": 0.00022600507907833984,
      "loss": 2.2106,
      "step": 4468
    },
    {
      "epoch": 0.6942135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.00022579464644834868,
      "loss": 2.0789,
      "step": 4469
    },
    {
      "epoch": 0.694368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.00022558428325504537,
      "loss": 2.0687,
      "step": 4470
    },
    {
      "epoch": 0.6945242718446601,
      "grad_norm": 0.0,
      "learning_rate": 0.00022537398955169957,
      "loss": 2.0392,
      "step": 4471
    },
    {
      "epoch": 0.6946796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.00022516376539156413,
      "loss": 2.1302,
      "step": 4472
    },
    {
      "epoch": 0.6948349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.0002249536108278735,
      "loss": 2.0131,
      "step": 4473
    },
    {
      "epoch": 0.694990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.00022474352591384523,
      "loss": 2.0724,
      "step": 4474
    },
    {
      "epoch": 0.6951456310679611,
      "grad_norm": 0.0,
      "learning_rate": 0.00022453351070267846,
      "loss": 2.1832,
      "step": 4475
    },
    {
      "epoch": 0.6953009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.00022432356524755532,
      "loss": 2.1199,
      "step": 4476
    },
    {
      "epoch": 0.6954563106796117,
      "grad_norm": 0.0,
      "learning_rate": 0.00022411368960164024,
      "loss": 2.1973,
      "step": 4477
    },
    {
      "epoch": 0.6956116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.00022390388381807937,
      "loss": 2.131,
      "step": 4478
    },
    {
      "epoch": 0.6957669902912621,
      "grad_norm": 0.0,
      "learning_rate": 0.00022369414795000186,
      "loss": 2.2372,
      "step": 4479
    },
    {
      "epoch": 0.6959223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.00022348448205051886,
      "loss": 2.131,
      "step": 4480
    },
    {
      "epoch": 0.6960776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.00022327488617272345,
      "loss": 2.1517,
      "step": 4481
    },
    {
      "epoch": 0.6962330097087379,
      "grad_norm": 0.0,
      "learning_rate": 0.00022306536036969187,
      "loss": 2.1038,
      "step": 4482
    },
    {
      "epoch": 0.6963883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.0002228559046944816,
      "loss": 2.1343,
      "step": 4483
    },
    {
      "epoch": 0.6965436893203883,
      "grad_norm": 0.0,
      "learning_rate": 0.00022264651920013306,
      "loss": 2.1423,
      "step": 4484
    },
    {
      "epoch": 0.6966990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.00022243720393966886,
      "loss": 2.1317,
      "step": 4485
    },
    {
      "epoch": 0.6968543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.00022222795896609333,
      "loss": 2.1484,
      "step": 4486
    },
    {
      "epoch": 0.697009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.00022201878433239364,
      "loss": 2.0854,
      "step": 4487
    },
    {
      "epoch": 0.6971650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.00022180968009153845,
      "loss": 2.04,
      "step": 4488
    },
    {
      "epoch": 0.6973203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.0002216006462964794,
      "loss": 2.1084,
      "step": 4489
    },
    {
      "epoch": 0.6974757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.00022139168300014945,
      "loss": 2.0991,
      "step": 4490
    },
    {
      "epoch": 0.697631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.00022118279025546434,
      "loss": 2.1711,
      "step": 4491
    },
    {
      "epoch": 0.6977864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.000220973968115322,
      "loss": 2.0562,
      "step": 4492
    },
    {
      "epoch": 0.6979417475728156,
      "grad_norm": 0.0,
      "learning_rate": 0.00022076521663260158,
      "loss": 2.0599,
      "step": 4493
    },
    {
      "epoch": 0.6980970873786407,
      "grad_norm": 0.0,
      "learning_rate": 0.00022055653586016528,
      "loss": 2.0914,
      "step": 4494
    },
    {
      "epoch": 0.698252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.00022034792585085727,
      "loss": 2.0619,
      "step": 4495
    },
    {
      "epoch": 0.6984077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.00022013938665750322,
      "loss": 2.1802,
      "step": 4496
    },
    {
      "epoch": 0.6985631067961166,
      "grad_norm": 0.0,
      "learning_rate": 0.00021993091833291168,
      "loss": 2.0596,
      "step": 4497
    },
    {
      "epoch": 0.6987184466019417,
      "grad_norm": 0.0,
      "learning_rate": 0.00021972252092987245,
      "loss": 2.059,
      "step": 4498
    },
    {
      "epoch": 0.698873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.00021951419450115783,
      "loss": 2.1713,
      "step": 4499
    },
    {
      "epoch": 0.6990291262135923,
      "grad_norm": 0.0,
      "learning_rate": 0.0002193059390995224,
      "loss": 2.1037,
      "step": 4500
    },
    {
      "epoch": 0.6991844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.00021909775477770195,
      "loss": 2.0991,
      "step": 4501
    },
    {
      "epoch": 0.6993398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.00021888964158841512,
      "loss": 2.0812,
      "step": 4502
    },
    {
      "epoch": 0.699495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.00021868159958436206,
      "loss": 2.1448,
      "step": 4503
    },
    {
      "epoch": 0.6996504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.00021847362881822474,
      "loss": 2.0954,
      "step": 4504
    },
    {
      "epoch": 0.6998058252427184,
      "grad_norm": 0.0,
      "learning_rate": 0.0002182657293426677,
      "loss": 2.043,
      "step": 4505
    },
    {
      "epoch": 0.6999611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.00021805790121033676,
      "loss": 2.1562,
      "step": 4506
    },
    {
      "epoch": 0.7001165048543689,
      "grad_norm": 0.0,
      "learning_rate": 0.00021785014447386032,
      "loss": 2.116,
      "step": 4507
    },
    {
      "epoch": 0.7002718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.00021764245918584797,
      "loss": 2.2566,
      "step": 4508
    },
    {
      "epoch": 0.7004271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.0002174348453988918,
      "loss": 2.0123,
      "step": 4509
    },
    {
      "epoch": 0.7005825242718446,
      "grad_norm": 0.0,
      "learning_rate": 0.00021722730316556582,
      "loss": 2.1422,
      "step": 4510
    },
    {
      "epoch": 0.7007378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.00021701983253842534,
      "loss": 2.0413,
      "step": 4511
    },
    {
      "epoch": 0.7008932038834952,
      "grad_norm": 0.0,
      "learning_rate": 0.0002168124335700082,
      "loss": 2.0978,
      "step": 4512
    },
    {
      "epoch": 0.7010485436893203,
      "grad_norm": 0.0,
      "learning_rate": 0.00021660510631283338,
      "loss": 2.1567,
      "step": 4513
    },
    {
      "epoch": 0.7012038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.0002163978508194026,
      "loss": 2.1366,
      "step": 4514
    },
    {
      "epoch": 0.7013592233009709,
      "grad_norm": 0.0,
      "learning_rate": 0.00021619066714219864,
      "loss": 2.1465,
      "step": 4515
    },
    {
      "epoch": 0.7015145631067962,
      "grad_norm": 0.0,
      "learning_rate": 0.0002159835553336863,
      "loss": 2.2036,
      "step": 4516
    },
    {
      "epoch": 0.7016699029126213,
      "grad_norm": 0.0,
      "learning_rate": 0.00021577651544631228,
      "loss": 2.0765,
      "step": 4517
    },
    {
      "epoch": 0.7018252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0002155695475325053,
      "loss": 2.1026,
      "step": 4518
    },
    {
      "epoch": 0.7019805825242719,
      "grad_norm": 0.0,
      "learning_rate": 0.0002153626516446753,
      "loss": 2.1449,
      "step": 4519
    },
    {
      "epoch": 0.7021359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0002151558278352145,
      "loss": 2.2861,
      "step": 4520
    },
    {
      "epoch": 0.7022912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.00021494907615649634,
      "loss": 2.1238,
      "step": 4521
    },
    {
      "epoch": 0.7024466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.00021474239666087664,
      "loss": 2.2126,
      "step": 4522
    },
    {
      "epoch": 0.7026019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.0002145357894006923,
      "loss": 2.0984,
      "step": 4523
    },
    {
      "epoch": 0.702757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.00021432925442826234,
      "loss": 2.1884,
      "step": 4524
    },
    {
      "epoch": 0.7029126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.00021412279179588756,
      "loss": 2.2018,
      "step": 4525
    },
    {
      "epoch": 0.7030679611650485,
      "grad_norm": 0.0,
      "learning_rate": 0.00021391640155585007,
      "loss": 2.2164,
      "step": 4526
    },
    {
      "epoch": 0.7032233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.00021371008376041363,
      "loss": 2.1514,
      "step": 4527
    },
    {
      "epoch": 0.703378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.00021350383846182415,
      "loss": 2.1281,
      "step": 4528
    },
    {
      "epoch": 0.7035339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.00021329766571230864,
      "loss": 2.0934,
      "step": 4529
    },
    {
      "epoch": 0.7036893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.00021309156556407623,
      "loss": 2.1555,
      "step": 4530
    },
    {
      "epoch": 0.7038446601941748,
      "grad_norm": 0.0,
      "learning_rate": 0.0002128855380693171,
      "loss": 2.123,
      "step": 4531
    },
    {
      "epoch": 0.704,
      "grad_norm": 0.0,
      "learning_rate": 0.00021267958328020348,
      "loss": 2.2172,
      "step": 4532
    },
    {
      "epoch": 0.7041553398058252,
      "grad_norm": 0.0,
      "learning_rate": 0.00021247370124888925,
      "loss": 2.1511,
      "step": 4533
    },
    {
      "epoch": 0.7043106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.00021226789202750935,
      "loss": 2.1626,
      "step": 4534
    },
    {
      "epoch": 0.7044660194174758,
      "grad_norm": 0.0,
      "learning_rate": 0.00021206215566818088,
      "loss": 2.1371,
      "step": 4535
    },
    {
      "epoch": 0.7046213592233009,
      "grad_norm": 0.0,
      "learning_rate": 0.0002118564922230019,
      "loss": 2.1393,
      "step": 4536
    },
    {
      "epoch": 0.7047766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.00021165090174405271,
      "loss": 2.2311,
      "step": 4537
    },
    {
      "epoch": 0.7049320388349515,
      "grad_norm": 0.0,
      "learning_rate": 0.00021144538428339443,
      "loss": 2.1837,
      "step": 4538
    },
    {
      "epoch": 0.7050873786407768,
      "grad_norm": 0.0,
      "learning_rate": 0.00021123993989306995,
      "loss": 2.1262,
      "step": 4539
    },
    {
      "epoch": 0.7052427184466019,
      "grad_norm": 0.0,
      "learning_rate": 0.000211034568625104,
      "loss": 2.1962,
      "step": 4540
    },
    {
      "epoch": 0.7053980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.00021082927053150214,
      "loss": 2.059,
      "step": 4541
    },
    {
      "epoch": 0.7055533980582525,
      "grad_norm": 0.0,
      "learning_rate": 0.00021062404566425192,
      "loss": 2.1382,
      "step": 4542
    },
    {
      "epoch": 0.7057087378640776,
      "grad_norm": 0.0,
      "learning_rate": 0.00021041889407532238,
      "loss": 2.1155,
      "step": 4543
    },
    {
      "epoch": 0.7058640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.0002102138158166635,
      "loss": 2.1404,
      "step": 4544
    },
    {
      "epoch": 0.7060194174757282,
      "grad_norm": 0.0,
      "learning_rate": 0.00021000881094020724,
      "loss": 2.1204,
      "step": 4545
    },
    {
      "epoch": 0.7061747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.00020980387949786646,
      "loss": 2.1551,
      "step": 4546
    },
    {
      "epoch": 0.7063300970873786,
      "grad_norm": 0.0,
      "learning_rate": 0.00020959902154153587,
      "loss": 2.1375,
      "step": 4547
    },
    {
      "epoch": 0.7064854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.00020939423712309157,
      "loss": 2.0527,
      "step": 4548
    },
    {
      "epoch": 0.7066407766990291,
      "grad_norm": 0.0,
      "learning_rate": 0.0002091895262943907,
      "loss": 2.0132,
      "step": 4549
    },
    {
      "epoch": 0.7067961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.00020898488910727175,
      "loss": 2.1893,
      "step": 4550
    },
    {
      "epoch": 0.7069514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0002087803256135551,
      "loss": 2.1053,
      "step": 4551
    },
    {
      "epoch": 0.7071067961165048,
      "grad_norm": 0.0,
      "learning_rate": 0.00020857583586504175,
      "loss": 2.1941,
      "step": 4552
    },
    {
      "epoch": 0.7072621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.00020837141991351471,
      "loss": 2.1428,
      "step": 4553
    },
    {
      "epoch": 0.7074174757281554,
      "grad_norm": 0.0,
      "learning_rate": 0.00020816707781073774,
      "loss": 2.1807,
      "step": 4554
    },
    {
      "epoch": 0.7075728155339805,
      "grad_norm": 0.0,
      "learning_rate": 0.00020796280960845638,
      "loss": 2.0866,
      "step": 4555
    },
    {
      "epoch": 0.7077281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.00020775861535839697,
      "loss": 2.0979,
      "step": 4556
    },
    {
      "epoch": 0.7078834951456311,
      "grad_norm": 0.0,
      "learning_rate": 0.00020755449511226744,
      "loss": 2.2236,
      "step": 4557
    },
    {
      "epoch": 0.7080388349514564,
      "grad_norm": 0.0,
      "learning_rate": 0.00020735044892175713,
      "loss": 2.1836,
      "step": 4558
    },
    {
      "epoch": 0.7081941747572815,
      "grad_norm": 0.0,
      "learning_rate": 0.00020714647683853604,
      "loss": 2.0935,
      "step": 4559
    },
    {
      "epoch": 0.7083495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0002069425789142562,
      "loss": 2.0891,
      "step": 4560
    },
    {
      "epoch": 0.7085048543689321,
      "grad_norm": 0.0,
      "learning_rate": 0.0002067387552005502,
      "loss": 2.132,
      "step": 4561
    },
    {
      "epoch": 0.7086601941747572,
      "grad_norm": 0.0,
      "learning_rate": 0.00020653500574903188,
      "loss": 2.0968,
      "step": 4562
    },
    {
      "epoch": 0.7088155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.00020633133061129683,
      "loss": 2.1847,
      "step": 4563
    },
    {
      "epoch": 0.7089708737864078,
      "grad_norm": 0.0,
      "learning_rate": 0.00020612772983892114,
      "loss": 2.1753,
      "step": 4564
    },
    {
      "epoch": 0.709126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.00020592420348346252,
      "loss": 2.075,
      "step": 4565
    },
    {
      "epoch": 0.7092815533980582,
      "grad_norm": 0.0,
      "learning_rate": 0.00020572075159645988,
      "loss": 2.0848,
      "step": 4566
    },
    {
      "epoch": 0.7094368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0002055173742294328,
      "loss": 2.2001,
      "step": 4567
    },
    {
      "epoch": 0.7095922330097088,
      "grad_norm": 0.0,
      "learning_rate": 0.00020531407143388252,
      "loss": 2.1488,
      "step": 4568
    },
    {
      "epoch": 0.709747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.00020511084326129087,
      "loss": 2.1747,
      "step": 4569
    },
    {
      "epoch": 0.7099029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.00020490768976312145,
      "loss": 2.1133,
      "step": 4570
    },
    {
      "epoch": 0.7100582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.00020470461099081837,
      "loss": 2.1236,
      "step": 4571
    },
    {
      "epoch": 0.7102135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0002045016069958069,
      "loss": 2.1621,
      "step": 4572
    },
    {
      "epoch": 0.710368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.00020429867782949373,
      "loss": 2.0941,
      "step": 4573
    },
    {
      "epoch": 0.7105242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0002040958235432661,
      "loss": 2.1127,
      "step": 4574
    },
    {
      "epoch": 0.7106796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.00020389304418849276,
      "loss": 2.192,
      "step": 4575
    },
    {
      "epoch": 0.7108349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.00020369033981652336,
      "loss": 2.1893,
      "step": 4576
    },
    {
      "epoch": 0.710990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.0002034877104786882,
      "loss": 2.0461,
      "step": 4577
    },
    {
      "epoch": 0.7111456310679611,
      "grad_norm": 0.0,
      "learning_rate": 0.0002032851562262991,
      "loss": 2.1424,
      "step": 4578
    },
    {
      "epoch": 0.7113009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0002030826771106485,
      "loss": 2.1249,
      "step": 4579
    },
    {
      "epoch": 0.7114563106796117,
      "grad_norm": 0.0,
      "learning_rate": 0.00020288027318300983,
      "loss": 2.1655,
      "step": 4580
    },
    {
      "epoch": 0.7116116504854368,
      "grad_norm": 0.0,
      "learning_rate": 0.000202677944494638,
      "loss": 2.2569,
      "step": 4581
    },
    {
      "epoch": 0.7117669902912621,
      "grad_norm": 0.0,
      "learning_rate": 0.000202475691096768,
      "loss": 2.0738,
      "step": 4582
    },
    {
      "epoch": 0.7119223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.00020227351304061657,
      "loss": 2.139,
      "step": 4583
    },
    {
      "epoch": 0.7120776699029127,
      "grad_norm": 0.0,
      "learning_rate": 0.0002020714103773809,
      "loss": 2.1108,
      "step": 4584
    },
    {
      "epoch": 0.7122330097087378,
      "grad_norm": 0.0,
      "learning_rate": 0.00020186938315823894,
      "loss": 2.2119,
      "step": 4585
    },
    {
      "epoch": 0.7123883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.00020166743143435018,
      "loss": 2.2214,
      "step": 4586
    },
    {
      "epoch": 0.7125436893203884,
      "grad_norm": 0.0,
      "learning_rate": 0.00020146555525685423,
      "loss": 1.9962,
      "step": 4587
    },
    {
      "epoch": 0.7126990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.00020126375467687235,
      "loss": 2.1126,
      "step": 4588
    },
    {
      "epoch": 0.7128543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.00020106202974550592,
      "loss": 2.1719,
      "step": 4589
    },
    {
      "epoch": 0.7130097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.0002008603805138376,
      "loss": 2.0654,
      "step": 4590
    },
    {
      "epoch": 0.7131650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.00020065880703293106,
      "loss": 2.1494,
      "step": 4591
    },
    {
      "epoch": 0.7133203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.00020045730935383,
      "loss": 2.143,
      "step": 4592
    },
    {
      "epoch": 0.7134757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.00020025588752755992,
      "loss": 2.125,
      "step": 4593
    },
    {
      "epoch": 0.713631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.00020005454160512638,
      "loss": 2.0991,
      "step": 4594
    },
    {
      "epoch": 0.7137864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.0001998532716375158,
      "loss": 2.1735,
      "step": 4595
    },
    {
      "epoch": 0.7139417475728156,
      "grad_norm": 0.0,
      "learning_rate": 0.00019965207767569593,
      "loss": 2.1314,
      "step": 4596
    },
    {
      "epoch": 0.7140970873786407,
      "grad_norm": 0.0,
      "learning_rate": 0.00019945095977061446,
      "loss": 2.2076,
      "step": 4597
    },
    {
      "epoch": 0.714252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.00019924991797320036,
      "loss": 2.1127,
      "step": 4598
    },
    {
      "epoch": 0.7144077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.00019904895233436353,
      "loss": 2.1477,
      "step": 4599
    },
    {
      "epoch": 0.7145631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0001988480629049938,
      "loss": 2.0839,
      "step": 4600
    },
    {
      "epoch": 0.7147184466019417,
      "grad_norm": 0.0,
      "learning_rate": 0.0001986472497359625,
      "loss": 2.1565,
      "step": 4601
    },
    {
      "epoch": 0.714873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.00019844651287812093,
      "loss": 2.1198,
      "step": 4602
    },
    {
      "epoch": 0.7150291262135923,
      "grad_norm": 0.0,
      "learning_rate": 0.00019824585238230192,
      "loss": 1.9886,
      "step": 4603
    },
    {
      "epoch": 0.7151844660194174,
      "grad_norm": 0.0,
      "learning_rate": 0.000198045268299318,
      "loss": 2.3071,
      "step": 4604
    },
    {
      "epoch": 0.7153398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.00019784476067996305,
      "loss": 2.1124,
      "step": 4605
    },
    {
      "epoch": 0.715495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.00019764432957501178,
      "loss": 2.085,
      "step": 4606
    },
    {
      "epoch": 0.7156504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.00019744397503521828,
      "loss": 2.0717,
      "step": 4607
    },
    {
      "epoch": 0.7158058252427184,
      "grad_norm": 0.0,
      "learning_rate": 0.00019724369711131856,
      "loss": 2.0696,
      "step": 4608
    },
    {
      "epoch": 0.7159611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.00019704349585402886,
      "loss": 2.012,
      "step": 4609
    },
    {
      "epoch": 0.716116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0001968433713140456,
      "loss": 2.1411,
      "step": 4610
    },
    {
      "epoch": 0.7162718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.00019664332354204646,
      "loss": 2.1721,
      "step": 4611
    },
    {
      "epoch": 0.7164271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.00019644335258868877,
      "loss": 2.1731,
      "step": 4612
    },
    {
      "epoch": 0.7165825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.00019624345850461128,
      "loss": 2.2064,
      "step": 4613
    },
    {
      "epoch": 0.7167378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.00019604364134043307,
      "loss": 2.051,
      "step": 4614
    },
    {
      "epoch": 0.7168932038834952,
      "grad_norm": 0.0,
      "learning_rate": 0.00019584390114675315,
      "loss": 2.1584,
      "step": 4615
    },
    {
      "epoch": 0.7170485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.00019564423797415194,
      "loss": 2.2137,
      "step": 4616
    },
    {
      "epoch": 0.7172038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.00019544465187318967,
      "loss": 2.1055,
      "step": 4617
    },
    {
      "epoch": 0.7173592233009709,
      "grad_norm": 0.0,
      "learning_rate": 0.00019524514289440714,
      "loss": 2.1019,
      "step": 4618
    },
    {
      "epoch": 0.7175145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.00019504571108832613,
      "loss": 2.0634,
      "step": 4619
    },
    {
      "epoch": 0.7176699029126213,
      "grad_norm": 0.0,
      "learning_rate": 0.00019484635650544814,
      "loss": 2.1951,
      "step": 4620
    },
    {
      "epoch": 0.7178252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0001946470791962558,
      "loss": 2.1378,
      "step": 4621
    },
    {
      "epoch": 0.7179805825242719,
      "grad_norm": 0.0,
      "learning_rate": 0.0001944478792112116,
      "loss": 2.1277,
      "step": 4622
    },
    {
      "epoch": 0.718135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.00019424875660075874,
      "loss": 2.2531,
      "step": 4623
    },
    {
      "epoch": 0.7182912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.0001940497114153212,
      "loss": 2.136,
      "step": 4624
    },
    {
      "epoch": 0.7184466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.00019385074370530236,
      "loss": 2.1736,
      "step": 4625
    },
    {
      "epoch": 0.7186019417475729,
      "grad_norm": 0.0,
      "learning_rate": 0.00019365185352108704,
      "loss": 2.1251,
      "step": 4626
    },
    {
      "epoch": 0.718757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.00019345304091303962,
      "loss": 2.1486,
      "step": 4627
    },
    {
      "epoch": 0.7189126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.00019325430593150532,
      "loss": 2.2007,
      "step": 4628
    },
    {
      "epoch": 0.7190679611650486,
      "grad_norm": 0.0,
      "learning_rate": 0.0001930556486268099,
      "loss": 2.1153,
      "step": 4629
    },
    {
      "epoch": 0.7192233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.00019285706904925842,
      "loss": 2.1189,
      "step": 4630
    },
    {
      "epoch": 0.719378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.00019265856724913727,
      "loss": 2.1324,
      "step": 4631
    },
    {
      "epoch": 0.7195339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.00019246014327671307,
      "loss": 2.1396,
      "step": 4632
    },
    {
      "epoch": 0.7196893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.00019226179718223203,
      "loss": 2.2073,
      "step": 4633
    },
    {
      "epoch": 0.7198446601941748,
      "grad_norm": 0.0,
      "learning_rate": 0.00019206352901592156,
      "loss": 2.1632,
      "step": 4634
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.0,
      "learning_rate": 0.00019186533882798834,
      "loss": 2.0236,
      "step": 4635
    },
    {
      "epoch": 0.7201553398058252,
      "grad_norm": 0.0,
      "learning_rate": 0.00019166722666862036,
      "loss": 2.1754,
      "step": 4636
    },
    {
      "epoch": 0.7203106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.00019146919258798489,
      "loss": 2.1403,
      "step": 4637
    },
    {
      "epoch": 0.7204660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.00019127123663623003,
      "loss": 2.1504,
      "step": 4638
    },
    {
      "epoch": 0.720621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0001910733588634841,
      "loss": 2.1721,
      "step": 4639
    },
    {
      "epoch": 0.7207766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0001908755593198553,
      "loss": 2.1333,
      "step": 4640
    },
    {
      "epoch": 0.7209320388349515,
      "grad_norm": 0.0,
      "learning_rate": 0.00019067783805543193,
      "loss": 2.1624,
      "step": 4641
    },
    {
      "epoch": 0.7210873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.00019048019512028314,
      "loss": 2.028,
      "step": 4642
    },
    {
      "epoch": 0.7212427184466019,
      "grad_norm": 0.0,
      "learning_rate": 0.00019028263056445737,
      "loss": 2.0229,
      "step": 4643
    },
    {
      "epoch": 0.7213980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.00019008514443798408,
      "loss": 2.0899,
      "step": 4644
    },
    {
      "epoch": 0.7215533980582525,
      "grad_norm": 0.0,
      "learning_rate": 0.00018988773679087207,
      "loss": 2.163,
      "step": 4645
    },
    {
      "epoch": 0.7217087378640776,
      "grad_norm": 0.0,
      "learning_rate": 0.0001896904076731108,
      "loss": 2.2021,
      "step": 4646
    },
    {
      "epoch": 0.7218640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.00018949315713466986,
      "loss": 2.1,
      "step": 4647
    },
    {
      "epoch": 0.7220194174757282,
      "grad_norm": 0.0,
      "learning_rate": 0.0001892959852254984,
      "loss": 1.9854,
      "step": 4648
    },
    {
      "epoch": 0.7221747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.00018909889199552644,
      "loss": 2.0317,
      "step": 4649
    },
    {
      "epoch": 0.7223300970873786,
      "grad_norm": 0.0,
      "learning_rate": 0.0001889018774946632,
      "loss": 2.2261,
      "step": 4650
    },
    {
      "epoch": 0.7224854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.00018870494177279884,
      "loss": 2.1218,
      "step": 4651
    },
    {
      "epoch": 0.7226407766990292,
      "grad_norm": 0.0,
      "learning_rate": 0.00018850808487980297,
      "loss": 2.142,
      "step": 4652
    },
    {
      "epoch": 0.7227961165048543,
      "grad_norm": 0.0,
      "learning_rate": 0.00018831130686552527,
      "loss": 2.1125,
      "step": 4653
    },
    {
      "epoch": 0.7229514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0001881146077797959,
      "loss": 2.0876,
      "step": 4654
    },
    {
      "epoch": 0.7231067961165049,
      "grad_norm": 0.0,
      "learning_rate": 0.0001879179876724244,
      "loss": 2.1654,
      "step": 4655
    },
    {
      "epoch": 0.7232621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.00018772144659320083,
      "loss": 2.2339,
      "step": 4656
    },
    {
      "epoch": 0.7234174757281553,
      "grad_norm": 0.0,
      "learning_rate": 0.0001875249845918952,
      "loss": 2.0956,
      "step": 4657
    },
    {
      "epoch": 0.7235728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.000187328601718257,
      "loss": 2.2018,
      "step": 4658
    },
    {
      "epoch": 0.7237281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.00018713229802201636,
      "loss": 2.0726,
      "step": 4659
    },
    {
      "epoch": 0.7238834951456311,
      "grad_norm": 0.0,
      "learning_rate": 0.0001869360735528827,
      "loss": 2.1579,
      "step": 4660
    },
    {
      "epoch": 0.7240388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.00018673992836054588,
      "loss": 2.1538,
      "step": 4661
    },
    {
      "epoch": 0.7241941747572815,
      "grad_norm": 0.0,
      "learning_rate": 0.00018654386249467565,
      "loss": 2.0717,
      "step": 4662
    },
    {
      "epoch": 0.7243495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.00018634787600492142,
      "loss": 2.0492,
      "step": 4663
    },
    {
      "epoch": 0.7245048543689321,
      "grad_norm": 0.0,
      "learning_rate": 0.00018615196894091235,
      "loss": 2.1739,
      "step": 4664
    },
    {
      "epoch": 0.7246601941747572,
      "grad_norm": 0.0,
      "learning_rate": 0.0001859561413522582,
      "loss": 2.1674,
      "step": 4665
    },
    {
      "epoch": 0.7248155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.0001857603932885477,
      "loss": 2.0682,
      "step": 4666
    },
    {
      "epoch": 0.7249708737864078,
      "grad_norm": 0.0,
      "learning_rate": 0.00018556472479935033,
      "loss": 2.0991,
      "step": 4667
    },
    {
      "epoch": 0.7251262135922331,
      "grad_norm": 0.0,
      "learning_rate": 0.00018536913593421455,
      "loss": 2.1535,
      "step": 4668
    },
    {
      "epoch": 0.7252815533980582,
      "grad_norm": 0.0,
      "learning_rate": 0.00018517362674266952,
      "loss": 1.9946,
      "step": 4669
    },
    {
      "epoch": 0.7254368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.00018497819727422338,
      "loss": 2.0888,
      "step": 4670
    },
    {
      "epoch": 0.7255922330097088,
      "grad_norm": 0.0,
      "learning_rate": 0.00018478284757836468,
      "loss": 2.1082,
      "step": 4671
    },
    {
      "epoch": 0.7257475728155339,
      "grad_norm": 0.0,
      "learning_rate": 0.00018458757770456181,
      "loss": 2.0452,
      "step": 4672
    },
    {
      "epoch": 0.7259029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.00018439238770226235,
      "loss": 2.0994,
      "step": 4673
    },
    {
      "epoch": 0.7260582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.00018419727762089433,
      "loss": 2.1363,
      "step": 4674
    },
    {
      "epoch": 0.7262135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.0001840022475098651,
      "loss": 2.117,
      "step": 4675
    },
    {
      "epoch": 0.7263689320388349,
      "grad_norm": 0.0,
      "learning_rate": 0.00018380729741856173,
      "loss": 2.0601,
      "step": 4676
    },
    {
      "epoch": 0.7265242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0001836124273963516,
      "loss": 2.1851,
      "step": 4677
    },
    {
      "epoch": 0.7266796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.0001834176374925809,
      "loss": 2.1414,
      "step": 4678
    },
    {
      "epoch": 0.7268349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.00018322292775657634,
      "loss": 1.9941,
      "step": 4679
    },
    {
      "epoch": 0.7269902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.00018302829823764423,
      "loss": 2.1859,
      "step": 4680
    },
    {
      "epoch": 0.7271456310679612,
      "grad_norm": 0.0,
      "learning_rate": 0.00018283374898506988,
      "loss": 2.1274,
      "step": 4681
    },
    {
      "epoch": 0.7273009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.00018263928004811925,
      "loss": 2.1422,
      "step": 4682
    },
    {
      "epoch": 0.7274563106796117,
      "grad_norm": 0.0,
      "learning_rate": 0.00018244489147603704,
      "loss": 2.1326,
      "step": 4683
    },
    {
      "epoch": 0.7276116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.00018225058331804823,
      "loss": 1.9666,
      "step": 4684
    },
    {
      "epoch": 0.7277669902912621,
      "grad_norm": 0.0,
      "learning_rate": 0.00018205635562335742,
      "loss": 2.0842,
      "step": 4685
    },
    {
      "epoch": 0.7279223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.00018186220844114843,
      "loss": 2.133,
      "step": 4686
    },
    {
      "epoch": 0.7280776699029127,
      "grad_norm": 0.0,
      "learning_rate": 0.000181668141820585,
      "loss": 2.1885,
      "step": 4687
    },
    {
      "epoch": 0.7282330097087378,
      "grad_norm": 0.0,
      "learning_rate": 0.00018147415581081018,
      "loss": 2.1159,
      "step": 4688
    },
    {
      "epoch": 0.7283883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.00018128025046094693,
      "loss": 2.3099,
      "step": 4689
    },
    {
      "epoch": 0.7285436893203884,
      "grad_norm": 0.0,
      "learning_rate": 0.00018108642582009789,
      "loss": 2.1147,
      "step": 4690
    },
    {
      "epoch": 0.7286990291262135,
      "grad_norm": 0.0,
      "learning_rate": 0.00018089268193734472,
      "loss": 2.131,
      "step": 4691
    },
    {
      "epoch": 0.7288543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.00018069901886174917,
      "loss": 2.1934,
      "step": 4692
    },
    {
      "epoch": 0.7290097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.00018050543664235207,
      "loss": 2.0834,
      "step": 4693
    },
    {
      "epoch": 0.7291650485436894,
      "grad_norm": 0.0,
      "learning_rate": 0.00018031193532817418,
      "loss": 2.2264,
      "step": 4694
    },
    {
      "epoch": 0.7293203883495145,
      "grad_norm": 0.0,
      "learning_rate": 0.00018011851496821575,
      "loss": 2.1961,
      "step": 4695
    },
    {
      "epoch": 0.7294757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.00017992517561145604,
      "loss": 2.1505,
      "step": 4696
    },
    {
      "epoch": 0.7296310679611651,
      "grad_norm": 0.0,
      "learning_rate": 0.0001797319173068545,
      "loss": 2.1226,
      "step": 4697
    },
    {
      "epoch": 0.7297864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.00017953874010334948,
      "loss": 2.087,
      "step": 4698
    },
    {
      "epoch": 0.7299417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.00017934564404985883,
      "loss": 2.1747,
      "step": 4699
    },
    {
      "epoch": 0.7300970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.00017915262919528043,
      "loss": 2.1018,
      "step": 4700
    },
    {
      "epoch": 0.730252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.00017895969558849085,
      "loss": 2.0958,
      "step": 4701
    },
    {
      "epoch": 0.7304077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.0001787668432783467,
      "loss": 2.0944,
      "step": 4702
    },
    {
      "epoch": 0.7305631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0001785740723136835,
      "loss": 2.0387,
      "step": 4703
    },
    {
      "epoch": 0.7307184466019417,
      "grad_norm": 0.0,
      "learning_rate": 0.0001783813827433165,
      "loss": 2.1022,
      "step": 4704
    },
    {
      "epoch": 0.730873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0001781887746160404,
      "loss": 2.1445,
      "step": 4705
    },
    {
      "epoch": 0.7310291262135923,
      "grad_norm": 0.0,
      "learning_rate": 0.00017799624798062885,
      "loss": 2.1614,
      "step": 4706
    },
    {
      "epoch": 0.7311844660194174,
      "grad_norm": 0.0,
      "learning_rate": 0.00017780380288583548,
      "loss": 2.1086,
      "step": 4707
    },
    {
      "epoch": 0.7313398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.00017761143938039266,
      "loss": 2.1331,
      "step": 4708
    },
    {
      "epoch": 0.731495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.0001774191575130123,
      "loss": 2.0698,
      "step": 4709
    },
    {
      "epoch": 0.7316504854368931,
      "grad_norm": 0.0,
      "learning_rate": 0.00017722695733238597,
      "loss": 2.1637,
      "step": 4710
    },
    {
      "epoch": 0.7318058252427184,
      "grad_norm": 0.0,
      "learning_rate": 0.00017703483888718396,
      "loss": 2.0441,
      "step": 4711
    },
    {
      "epoch": 0.7319611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.00017684280222605642,
      "loss": 2.0635,
      "step": 4712
    },
    {
      "epoch": 0.732116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0001766508473976327,
      "loss": 2.1744,
      "step": 4713
    },
    {
      "epoch": 0.7322718446601941,
      "grad_norm": 0.0,
      "learning_rate": 0.0001764589744505209,
      "loss": 2.1282,
      "step": 4714
    },
    {
      "epoch": 0.7324271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.00017626718343330918,
      "loss": 2.1488,
      "step": 4715
    },
    {
      "epoch": 0.7325825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.00017607547439456412,
      "loss": 2.1431,
      "step": 4716
    },
    {
      "epoch": 0.7327378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.00017588384738283225,
      "loss": 2.2428,
      "step": 4717
    },
    {
      "epoch": 0.7328932038834951,
      "grad_norm": 0.0,
      "learning_rate": 0.00017569230244663913,
      "loss": 2.134,
      "step": 4718
    },
    {
      "epoch": 0.7330485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.00017550083963448914,
      "loss": 2.1422,
      "step": 4719
    },
    {
      "epoch": 0.7332038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.00017530945899486668,
      "loss": 2.1471,
      "step": 4720
    },
    {
      "epoch": 0.7333592233009709,
      "grad_norm": 0.0,
      "learning_rate": 0.00017511816057623415,
      "loss": 2.1353,
      "step": 4721
    },
    {
      "epoch": 0.7335145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.00017492694442703416,
      "loss": 2.0853,
      "step": 4722
    },
    {
      "epoch": 0.7336699029126214,
      "grad_norm": 0.0,
      "learning_rate": 0.00017473581059568834,
      "loss": 2.0669,
      "step": 4723
    },
    {
      "epoch": 0.7338252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.00017454475913059687,
      "loss": 2.1624,
      "step": 4724
    },
    {
      "epoch": 0.7339805825242719,
      "grad_norm": 0.0,
      "learning_rate": 0.0001743537900801399,
      "loss": 2.0795,
      "step": 4725
    },
    {
      "epoch": 0.7341359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0001741629034926759,
      "loss": 2.1074,
      "step": 4726
    },
    {
      "epoch": 0.7342912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.00017397209941654303,
      "loss": 2.0834,
      "step": 4727
    },
    {
      "epoch": 0.7344466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.0001737813779000585,
      "loss": 2.0602,
      "step": 4728
    },
    {
      "epoch": 0.7346019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.00017359073899151823,
      "loss": 2.2047,
      "step": 4729
    },
    {
      "epoch": 0.734757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.00017340018273919784,
      "loss": 2.1843,
      "step": 4730
    },
    {
      "epoch": 0.7349126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.00017320970919135133,
      "loss": 2.1423,
      "step": 4731
    },
    {
      "epoch": 0.7350679611650486,
      "grad_norm": 0.0,
      "learning_rate": 0.00017301931839621204,
      "loss": 2.1302,
      "step": 4732
    },
    {
      "epoch": 0.7352233009708737,
      "grad_norm": 0.0,
      "learning_rate": 0.00017282901040199274,
      "loss": 2.0897,
      "step": 4733
    },
    {
      "epoch": 0.735378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.00017263878525688452,
      "loss": 2.0536,
      "step": 4734
    },
    {
      "epoch": 0.7355339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.00017244864300905804,
      "loss": 2.0238,
      "step": 4735
    },
    {
      "epoch": 0.7356893203883496,
      "grad_norm": 0.0,
      "learning_rate": 0.00017225858370666292,
      "loss": 2.0936,
      "step": 4736
    },
    {
      "epoch": 0.7358446601941747,
      "grad_norm": 0.0,
      "learning_rate": 0.00017206860739782737,
      "loss": 2.2358,
      "step": 4737
    },
    {
      "epoch": 0.736,
      "grad_norm": 0.0,
      "learning_rate": 0.00017187871413065913,
      "loss": 2.1139,
      "step": 4738
    },
    {
      "epoch": 0.7361553398058253,
      "grad_norm": 0.0,
      "learning_rate": 0.00017168890395324437,
      "loss": 2.1543,
      "step": 4739
    },
    {
      "epoch": 0.7363106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.0001714991769136488,
      "loss": 2.1599,
      "step": 4740
    },
    {
      "epoch": 0.7364660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.0001713095330599165,
      "loss": 2.1416,
      "step": 4741
    },
    {
      "epoch": 0.736621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.00017111997244007073,
      "loss": 2.0624,
      "step": 4742
    },
    {
      "epoch": 0.7367766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.00017093049510211422,
      "loss": 2.1234,
      "step": 4743
    },
    {
      "epoch": 0.7369320388349515,
      "grad_norm": 0.0,
      "learning_rate": 0.0001707411010940273,
      "loss": 2.082,
      "step": 4744
    },
    {
      "epoch": 0.7370873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0001705517904637704,
      "loss": 2.1291,
      "step": 4745
    },
    {
      "epoch": 0.7372427184466019,
      "grad_norm": 0.0,
      "learning_rate": 0.00017036256325928258,
      "loss": 2.1114,
      "step": 4746
    },
    {
      "epoch": 0.7373980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.0001701734195284812,
      "loss": 2.157,
      "step": 4747
    },
    {
      "epoch": 0.7375533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.0001699843593192633,
      "loss": 2.1072,
      "step": 4748
    },
    {
      "epoch": 0.7377087378640776,
      "grad_norm": 0.0,
      "learning_rate": 0.000169795382679504,
      "loss": 2.2446,
      "step": 4749
    },
    {
      "epoch": 0.7378640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.00016960648965705784,
      "loss": 2.0767,
      "step": 4750
    },
    {
      "epoch": 0.7380194174757282,
      "grad_norm": 0.0,
      "learning_rate": 0.00016941768029975806,
      "loss": 2.098,
      "step": 4751
    },
    {
      "epoch": 0.7381747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.00016922895465541638,
      "loss": 2.0394,
      "step": 4752
    },
    {
      "epoch": 0.7383300970873786,
      "grad_norm": 0.0,
      "learning_rate": 0.00016904031277182387,
      "loss": 2.1525,
      "step": 4753
    },
    {
      "epoch": 0.7384854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.00016885175469674984,
      "loss": 2.21,
      "step": 4754
    },
    {
      "epoch": 0.7386407766990292,
      "grad_norm": 0.0,
      "learning_rate": 0.00016866328047794254,
      "loss": 2.2234,
      "step": 4755
    },
    {
      "epoch": 0.7387961165048543,
      "grad_norm": 0.0,
      "learning_rate": 0.00016847489016312932,
      "loss": 2.1377,
      "step": 4756
    },
    {
      "epoch": 0.7389514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.0001682865838000158,
      "loss": 2.0838,
      "step": 4757
    },
    {
      "epoch": 0.7391067961165049,
      "grad_norm": 0.0,
      "learning_rate": 0.00016809836143628692,
      "loss": 2.1483,
      "step": 4758
    },
    {
      "epoch": 0.7392621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0001679102231196055,
      "loss": 2.165,
      "step": 4759
    },
    {
      "epoch": 0.7394174757281553,
      "grad_norm": 0.0,
      "learning_rate": 0.00016772216889761394,
      "loss": 2.041,
      "step": 4760
    },
    {
      "epoch": 0.7395728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.00016753419881793298,
      "loss": 2.2046,
      "step": 4761
    },
    {
      "epoch": 0.7397281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.00016734631292816184,
      "loss": 2.1495,
      "step": 4762
    },
    {
      "epoch": 0.7398834951456311,
      "grad_norm": 0.0,
      "learning_rate": 0.00016715851127587888,
      "loss": 2.177,
      "step": 4763
    },
    {
      "epoch": 0.7400388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.0001669707939086406,
      "loss": 2.1469,
      "step": 4764
    },
    {
      "epoch": 0.7401941747572816,
      "grad_norm": 0.0,
      "learning_rate": 0.00016678316087398283,
      "loss": 2.1727,
      "step": 4765
    },
    {
      "epoch": 0.7403495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.00016659561221941938,
      "loss": 1.9788,
      "step": 4766
    },
    {
      "epoch": 0.740504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.00016640814799244287,
      "loss": 2.0493,
      "step": 4767
    },
    {
      "epoch": 0.7406601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.00016622076824052473,
      "loss": 2.0032,
      "step": 4768
    },
    {
      "epoch": 0.7408155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.00016603347301111517,
      "loss": 2.266,
      "step": 4769
    },
    {
      "epoch": 0.7409708737864078,
      "grad_norm": 0.0,
      "learning_rate": 0.00016584626235164235,
      "loss": 2.194,
      "step": 4770
    },
    {
      "epoch": 0.741126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.00016565913630951375,
      "loss": 2.193,
      "step": 4771
    },
    {
      "epoch": 0.7412815533980582,
      "grad_norm": 0.0,
      "learning_rate": 0.0001654720949321147,
      "loss": 2.1121,
      "step": 4772
    },
    {
      "epoch": 0.7414368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0001652851382668098,
      "loss": 2.0588,
      "step": 4773
    },
    {
      "epoch": 0.7415922330097088,
      "grad_norm": 0.0,
      "learning_rate": 0.00016509826636094155,
      "loss": 2.1231,
      "step": 4774
    },
    {
      "epoch": 0.7417475728155339,
      "grad_norm": 0.0,
      "learning_rate": 0.00016491147926183148,
      "loss": 2.0345,
      "step": 4775
    },
    {
      "epoch": 0.7419029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.00016472477701677958,
      "loss": 2.1444,
      "step": 4776
    },
    {
      "epoch": 0.7420582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.00016453815967306407,
      "loss": 2.0688,
      "step": 4777
    },
    {
      "epoch": 0.7422135922330098,
      "grad_norm": 0.0,
      "learning_rate": 0.00016435162727794167,
      "loss": 2.1217,
      "step": 4778
    },
    {
      "epoch": 0.7423689320388349,
      "grad_norm": 0.0,
      "learning_rate": 0.0001641651798786481,
      "loss": 2.129,
      "step": 4779
    },
    {
      "epoch": 0.7425242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0001639788175223969,
      "loss": 2.0637,
      "step": 4780
    },
    {
      "epoch": 0.7426796116504855,
      "grad_norm": 0.0,
      "learning_rate": 0.00016379254025638068,
      "loss": 2.0503,
      "step": 4781
    },
    {
      "epoch": 0.7428349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.00016360634812776986,
      "loss": 1.9392,
      "step": 4782
    },
    {
      "epoch": 0.7429902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.00016342024118371385,
      "loss": 2.2681,
      "step": 4783
    },
    {
      "epoch": 0.7431456310679612,
      "grad_norm": 0.0,
      "learning_rate": 0.00016323421947134038,
      "loss": 2.1875,
      "step": 4784
    },
    {
      "epoch": 0.7433009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.00016304828303775526,
      "loss": 2.0902,
      "step": 4785
    },
    {
      "epoch": 0.7434563106796116,
      "grad_norm": 0.0,
      "learning_rate": 0.00016286243193004325,
      "loss": 2.1312,
      "step": 4786
    },
    {
      "epoch": 0.7436116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.00016267666619526684,
      "loss": 2.1624,
      "step": 4787
    },
    {
      "epoch": 0.7437669902912621,
      "grad_norm": 0.0,
      "learning_rate": 0.00016249098588046757,
      "loss": 2.1006,
      "step": 4788
    },
    {
      "epoch": 0.7439223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.00016230539103266495,
      "loss": 2.0455,
      "step": 4789
    },
    {
      "epoch": 0.7440776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.00016211988169885673,
      "loss": 2.1083,
      "step": 4790
    },
    {
      "epoch": 0.7442330097087378,
      "grad_norm": 0.0,
      "learning_rate": 0.00016193445792601958,
      "loss": 2.1666,
      "step": 4791
    },
    {
      "epoch": 0.7443883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.00016174911976110773,
      "loss": 2.1489,
      "step": 4792
    },
    {
      "epoch": 0.7445436893203884,
      "grad_norm": 0.0,
      "learning_rate": 0.00016156386725105438,
      "loss": 2.149,
      "step": 4793
    },
    {
      "epoch": 0.7446990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.00016137870044277092,
      "loss": 2.1495,
      "step": 4794
    },
    {
      "epoch": 0.7448543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.0001611936193831466,
      "loss": 2.1176,
      "step": 4795
    },
    {
      "epoch": 0.7450097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.00016100862411904955,
      "loss": 2.0991,
      "step": 4796
    },
    {
      "epoch": 0.7451650485436894,
      "grad_norm": 0.0,
      "learning_rate": 0.00016082371469732564,
      "loss": 2.0765,
      "step": 4797
    },
    {
      "epoch": 0.7453203883495145,
      "grad_norm": 0.0,
      "learning_rate": 0.0001606388911647994,
      "loss": 2.1179,
      "step": 4798
    },
    {
      "epoch": 0.7454757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.00016045415356827364,
      "loss": 2.1483,
      "step": 4799
    },
    {
      "epoch": 0.7456310679611651,
      "grad_norm": 0.0,
      "learning_rate": 0.000160269501954529,
      "loss": 2.1068,
      "step": 4800
    },
    {
      "epoch": 0.7457864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.00016008493637032457,
      "loss": 2.1952,
      "step": 4801
    },
    {
      "epoch": 0.7459417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.00015990045686239786,
      "loss": 2.1548,
      "step": 4802
    },
    {
      "epoch": 0.7460970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.0001597160634774641,
      "loss": 2.0712,
      "step": 4803
    },
    {
      "epoch": 0.746252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.00015953175626221732,
      "loss": 2.1222,
      "step": 4804
    },
    {
      "epoch": 0.7464077669902912,
      "grad_norm": 0.0,
      "learning_rate": 0.00015934753526332913,
      "loss": 2.1219,
      "step": 4805
    },
    {
      "epoch": 0.7465631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.00015916340052744986,
      "loss": 2.0871,
      "step": 4806
    },
    {
      "epoch": 0.7467184466019418,
      "grad_norm": 0.0,
      "learning_rate": 0.00015897935210120744,
      "loss": 2.2278,
      "step": 4807
    },
    {
      "epoch": 0.746873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0001587953900312084,
      "loss": 2.1797,
      "step": 4808
    },
    {
      "epoch": 0.7470291262135922,
      "grad_norm": 0.0,
      "learning_rate": 0.00015861151436403736,
      "loss": 2.1073,
      "step": 4809
    },
    {
      "epoch": 0.7471844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.0001584277251462566,
      "loss": 2.108,
      "step": 4810
    },
    {
      "epoch": 0.7473398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.00015824402242440715,
      "loss": 2.0623,
      "step": 4811
    },
    {
      "epoch": 0.747495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.00015806040624500778,
      "loss": 2.141,
      "step": 4812
    },
    {
      "epoch": 0.7476504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.00015787687665455515,
      "loss": 2.1069,
      "step": 4813
    },
    {
      "epoch": 0.7478058252427184,
      "grad_norm": 0.0,
      "learning_rate": 0.0001576934336995245,
      "loss": 2.0836,
      "step": 4814
    },
    {
      "epoch": 0.7479611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.00015751007742636857,
      "loss": 2.1131,
      "step": 4815
    },
    {
      "epoch": 0.748116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0001573268078815187,
      "loss": 2.0859,
      "step": 4816
    },
    {
      "epoch": 0.7482718446601941,
      "grad_norm": 0.0,
      "learning_rate": 0.0001571436251113841,
      "loss": 2.1921,
      "step": 4817
    },
    {
      "epoch": 0.7484271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.00015696052916235154,
      "loss": 2.2171,
      "step": 4818
    },
    {
      "epoch": 0.7485825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.00015677752008078666,
      "loss": 2.0615,
      "step": 4819
    },
    {
      "epoch": 0.74873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.00015659459791303215,
      "loss": 2.1821,
      "step": 4820
    },
    {
      "epoch": 0.7488932038834951,
      "grad_norm": 0.0,
      "learning_rate": 0.00015641176270540964,
      "loss": 2.097,
      "step": 4821
    },
    {
      "epoch": 0.7490485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.00015622901450421778,
      "loss": 2.1471,
      "step": 4822
    },
    {
      "epoch": 0.7492038834951457,
      "grad_norm": 0.0,
      "learning_rate": 0.00015604635335573415,
      "loss": 2.017,
      "step": 4823
    },
    {
      "epoch": 0.7493592233009708,
      "grad_norm": 0.0,
      "learning_rate": 0.00015586377930621348,
      "loss": 2.0933,
      "step": 4824
    },
    {
      "epoch": 0.7495145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.00015568129240188877,
      "loss": 2.0103,
      "step": 4825
    },
    {
      "epoch": 0.7496699029126214,
      "grad_norm": 0.0,
      "learning_rate": 0.00015549889268897112,
      "loss": 2.249,
      "step": 4826
    },
    {
      "epoch": 0.7498252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0001553165802136494,
      "loss": 2.1378,
      "step": 4827
    },
    {
      "epoch": 0.7499805825242718,
      "grad_norm": 0.0,
      "learning_rate": 0.00015513435502209017,
      "loss": 2.0232,
      "step": 4828
    },
    {
      "epoch": 0.7501359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.0001549522171604384,
      "loss": 2.1626,
      "step": 4829
    },
    {
      "epoch": 0.7502912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.0001547701666748163,
      "loss": 2.1711,
      "step": 4830
    },
    {
      "epoch": 0.7504466019417476,
      "grad_norm": 0.0,
      "learning_rate": 0.0001545882036113244,
      "loss": 2.137,
      "step": 4831
    },
    {
      "epoch": 0.7506019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.00015440632801604128,
      "loss": 2.1469,
      "step": 4832
    },
    {
      "epoch": 0.750757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.00015422453993502267,
      "loss": 2.1376,
      "step": 4833
    },
    {
      "epoch": 0.7509126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.00015404283941430286,
      "loss": 2.1715,
      "step": 4834
    },
    {
      "epoch": 0.7510679611650486,
      "grad_norm": 0.0,
      "learning_rate": 0.00015386122649989348,
      "loss": 2.1296,
      "step": 4835
    },
    {
      "epoch": 0.7512233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.00015367970123778403,
      "loss": 2.121,
      "step": 4836
    },
    {
      "epoch": 0.751378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0001534982636739422,
      "loss": 2.0696,
      "step": 4837
    },
    {
      "epoch": 0.7515339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.00015331691385431296,
      "loss": 2.0994,
      "step": 4838
    },
    {
      "epoch": 0.7516893203883496,
      "grad_norm": 0.0,
      "learning_rate": 0.0001531356518248196,
      "loss": 2.0325,
      "step": 4839
    },
    {
      "epoch": 0.7518446601941747,
      "grad_norm": 0.0,
      "learning_rate": 0.0001529544776313625,
      "loss": 2.1241,
      "step": 4840
    },
    {
      "epoch": 0.752,
      "grad_norm": 0.0,
      "learning_rate": 0.0001527733913198204,
      "loss": 2.0864,
      "step": 4841
    },
    {
      "epoch": 0.7521553398058253,
      "grad_norm": 0.0,
      "learning_rate": 0.00015259239293604964,
      "loss": 2.1473,
      "step": 4842
    },
    {
      "epoch": 0.7523106796116504,
      "grad_norm": 0.0,
      "learning_rate": 0.00015241148252588405,
      "loss": 2.171,
      "step": 4843
    },
    {
      "epoch": 0.7524660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.0001522306601351356,
      "loss": 2.1513,
      "step": 4844
    },
    {
      "epoch": 0.752621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.00015204992580959353,
      "loss": 2.1422,
      "step": 4845
    },
    {
      "epoch": 0.7527766990291263,
      "grad_norm": 0.0,
      "learning_rate": 0.00015186927959502485,
      "loss": 2.1415,
      "step": 4846
    },
    {
      "epoch": 0.7529320388349514,
      "grad_norm": 0.0,
      "learning_rate": 0.00015168872153717466,
      "loss": 2.2189,
      "step": 4847
    },
    {
      "epoch": 0.7530873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.00015150825168176518,
      "loss": 2.1539,
      "step": 4848
    },
    {
      "epoch": 0.753242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.00015132787007449673,
      "loss": 2.1118,
      "step": 4849
    },
    {
      "epoch": 0.7533980582524272,
      "grad_norm": 0.0,
      "learning_rate": 0.00015114757676104723,
      "loss": 2.2075,
      "step": 4850
    },
    {
      "epoch": 0.7535533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.00015096737178707192,
      "loss": 2.1465,
      "step": 4851
    },
    {
      "epoch": 0.7537087378640777,
      "grad_norm": 0.0,
      "learning_rate": 0.00015078725519820414,
      "loss": 2.0806,
      "step": 4852
    },
    {
      "epoch": 0.7538640776699029,
      "grad_norm": 0.0,
      "learning_rate": 0.00015060722704005424,
      "loss": 2.1166,
      "step": 4853
    },
    {
      "epoch": 0.7540194174757282,
      "grad_norm": 0.0,
      "learning_rate": 0.00015042728735821086,
      "loss": 2.1326,
      "step": 4854
    },
    {
      "epoch": 0.7541747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.00015024743619823972,
      "loss": 2.179,
      "step": 4855
    },
    {
      "epoch": 0.7543300970873786,
      "grad_norm": 0.0,
      "learning_rate": 0.0001500676736056843,
      "loss": 2.1749,
      "step": 4856
    },
    {
      "epoch": 0.7544854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.00014988799962606608,
      "loss": 2.0598,
      "step": 4857
    },
    {
      "epoch": 0.7546407766990292,
      "grad_norm": 0.0,
      "learning_rate": 0.00014970841430488296,
      "loss": 2.1643,
      "step": 4858
    },
    {
      "epoch": 0.7547961165048543,
      "grad_norm": 0.0,
      "learning_rate": 0.00014952891768761157,
      "loss": 2.2062,
      "step": 4859
    },
    {
      "epoch": 0.7549514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.00014934950981970564,
      "loss": 2.1528,
      "step": 4860
    },
    {
      "epoch": 0.7551067961165049,
      "grad_norm": 0.0,
      "learning_rate": 0.00014917019074659615,
      "loss": 2.1276,
      "step": 4861
    },
    {
      "epoch": 0.75526213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.00014899096051369215,
      "loss": 2.0994,
      "step": 4862
    },
    {
      "epoch": 0.7554174757281553,
      "grad_norm": 0.0,
      "learning_rate": 0.00014881181916637964,
      "loss": 2.1034,
      "step": 4863
    },
    {
      "epoch": 0.7555728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.0001486327667500224,
      "loss": 2.1733,
      "step": 4864
    },
    {
      "epoch": 0.7557281553398059,
      "grad_norm": 0.0,
      "learning_rate": 0.00014845380330996195,
      "loss": 2.1919,
      "step": 4865
    },
    {
      "epoch": 0.755883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.00014827492889151654,
      "loss": 2.1693,
      "step": 4866
    },
    {
      "epoch": 0.7560388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.00014809614353998273,
      "loss": 2.0774,
      "step": 4867
    },
    {
      "epoch": 0.7561941747572816,
      "grad_norm": 0.0,
      "learning_rate": 0.00014791744730063394,
      "loss": 1.9733,
      "step": 4868
    },
    {
      "epoch": 0.7563495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.00014773884021872103,
      "loss": 2.136,
      "step": 4869
    },
    {
      "epoch": 0.756504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.00014756032233947274,
      "loss": 2.0399,
      "step": 4870
    },
    {
      "epoch": 0.7566601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.00014738189370809468,
      "loss": 2.2335,
      "step": 4871
    },
    {
      "epoch": 0.7568155339805825,
      "grad_norm": 0.0,
      "learning_rate": 0.0001472035543697704,
      "loss": 2.1201,
      "step": 4872
    },
    {
      "epoch": 0.7569708737864078,
      "grad_norm": 0.0,
      "learning_rate": 0.0001470253043696602,
      "loss": 2.0947,
      "step": 4873
    },
    {
      "epoch": 0.757126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.00014684714375290238,
      "loss": 2.1414,
      "step": 4874
    },
    {
      "epoch": 0.7572815533980582,
      "grad_norm": 0.0,
      "learning_rate": 0.00014666907256461237,
      "loss": 2.0478,
      "step": 4875
    },
    {
      "epoch": 0.7574368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.00014649109084988266,
      "loss": 2.0749,
      "step": 4876
    },
    {
      "epoch": 0.7575922330097088,
      "grad_norm": 0.0,
      "learning_rate": 0.00014631319865378368,
      "loss": 2.1505,
      "step": 4877
    },
    {
      "epoch": 0.757747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.00014613539602136244,
      "loss": 2.1027,
      "step": 4878
    },
    {
      "epoch": 0.7579029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.000145957682997644,
      "loss": 2.1038,
      "step": 4879
    },
    {
      "epoch": 0.7580582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.00014578005962763058,
      "loss": 2.2691,
      "step": 4880
    },
    {
      "epoch": 0.7582135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.00014560252595630097,
      "loss": 2.0641,
      "step": 4881
    },
    {
      "epoch": 0.7583689320388349,
      "grad_norm": 0.0,
      "learning_rate": 0.00014542508202861214,
      "loss": 2.1466,
      "step": 4882
    },
    {
      "epoch": 0.7585242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.00014524772788949815,
      "loss": 2.1426,
      "step": 4883
    },
    {
      "epoch": 0.7586796116504855,
      "grad_norm": 0.0,
      "learning_rate": 0.00014507046358386977,
      "loss": 2.0858,
      "step": 4884
    },
    {
      "epoch": 0.7588349514563106,
      "grad_norm": 0.0,
      "learning_rate": 0.00014489328915661586,
      "loss": 2.0897,
      "step": 4885
    },
    {
      "epoch": 0.7589902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.00014471620465260172,
      "loss": 2.0316,
      "step": 4886
    },
    {
      "epoch": 0.7591456310679612,
      "grad_norm": 0.0,
      "learning_rate": 0.0001445392101166705,
      "loss": 2.0934,
      "step": 4887
    },
    {
      "epoch": 0.7593009708737865,
      "grad_norm": 0.0,
      "learning_rate": 0.00014436230559364212,
      "loss": 2.094,
      "step": 4888
    },
    {
      "epoch": 0.7594563106796116,
      "grad_norm": 0.0,
      "learning_rate": 0.00014418549112831392,
      "loss": 2.0474,
      "step": 4889
    },
    {
      "epoch": 0.7596116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.00014400876676546066,
      "loss": 2.1477,
      "step": 4890
    },
    {
      "epoch": 0.7597669902912622,
      "grad_norm": 0.0,
      "learning_rate": 0.00014383213254983386,
      "loss": 2.0425,
      "step": 4891
    },
    {
      "epoch": 0.7599223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.00014365558852616218,
      "loss": 2.0929,
      "step": 4892
    },
    {
      "epoch": 0.7600776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.00014347913473915203,
      "loss": 2.1573,
      "step": 4893
    },
    {
      "epoch": 0.7602330097087379,
      "grad_norm": 0.0,
      "learning_rate": 0.00014330277123348622,
      "loss": 2.1452,
      "step": 4894
    },
    {
      "epoch": 0.7603883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.00014312649805382543,
      "loss": 2.149,
      "step": 4895
    },
    {
      "epoch": 0.7605436893203883,
      "grad_norm": 0.0,
      "learning_rate": 0.00014295031524480677,
      "loss": 2.2152,
      "step": 4896
    },
    {
      "epoch": 0.7606990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.00014277422285104496,
      "loss": 2.1436,
      "step": 4897
    },
    {
      "epoch": 0.7608543689320388,
      "grad_norm": 0.0,
      "learning_rate": 0.0001425982209171318,
      "loss": 2.1943,
      "step": 4898
    },
    {
      "epoch": 0.7610097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.00014242230948763578,
      "loss": 2.0947,
      "step": 4899
    },
    {
      "epoch": 0.7611650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.000142246488607103,
      "loss": 2.1012,
      "step": 4900
    },
    {
      "epoch": 0.7613203883495145,
      "grad_norm": 0.0,
      "learning_rate": 0.000142070758320056,
      "loss": 2.0785,
      "step": 4901
    },
    {
      "epoch": 0.7614757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.00014189511867099519,
      "loss": 2.204,
      "step": 4902
    },
    {
      "epoch": 0.7616310679611651,
      "grad_norm": 0.0,
      "learning_rate": 0.00014171956970439732,
      "loss": 2.1585,
      "step": 4903
    },
    {
      "epoch": 0.7617864077669902,
      "grad_norm": 0.0,
      "learning_rate": 0.00014154411146471635,
      "loss": 2.0817,
      "step": 4904
    },
    {
      "epoch": 0.7619417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.00014136874399638366,
      "loss": 2.1238,
      "step": 4905
    },
    {
      "epoch": 0.7620970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.000141193467343807,
      "loss": 1.9737,
      "step": 4906
    },
    {
      "epoch": 0.7622524271844661,
      "grad_norm": 0.0,
      "learning_rate": 0.00014101828155137165,
      "loss": 2.0584,
      "step": 4907
    },
    {
      "epoch": 0.7624077669902912,
      "grad_norm": 0.0,
      "learning_rate": 0.0001408431866634398,
      "loss": 2.1853,
      "step": 4908
    },
    {
      "epoch": 0.7625631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0001406681827243503,
      "loss": 2.1407,
      "step": 4909
    },
    {
      "epoch": 0.7627184466019418,
      "grad_norm": 0.0,
      "learning_rate": 0.00014049326977841936,
      "loss": 2.1644,
      "step": 4910
    },
    {
      "epoch": 0.762873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.00014031844786993974,
      "loss": 2.054,
      "step": 4911
    },
    {
      "epoch": 0.7630291262135922,
      "grad_norm": 0.0,
      "learning_rate": 0.00014014371704318157,
      "loss": 2.1926,
      "step": 4912
    },
    {
      "epoch": 0.7631844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.00013996907734239184,
      "loss": 2.1585,
      "step": 4913
    },
    {
      "epoch": 0.7633398058252427,
      "grad_norm": 0.0,
      "learning_rate": 0.0001397945288117942,
      "loss": 2.1609,
      "step": 4914
    },
    {
      "epoch": 0.7634951456310679,
      "grad_norm": 0.0,
      "learning_rate": 0.0001396200714955892,
      "loss": 2.1017,
      "step": 4915
    },
    {
      "epoch": 0.7636504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.00013944570543795477,
      "loss": 2.1252,
      "step": 4916
    },
    {
      "epoch": 0.7638058252427185,
      "grad_norm": 0.0,
      "learning_rate": 0.00013927143068304503,
      "loss": 2.1412,
      "step": 4917
    },
    {
      "epoch": 0.7639611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0001390972472749918,
      "loss": 2.1136,
      "step": 4918
    },
    {
      "epoch": 0.7641165048543689,
      "grad_norm": 0.0,
      "learning_rate": 0.000138923155257903,
      "loss": 2.1592,
      "step": 4919
    },
    {
      "epoch": 0.7642718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.00013874915467586396,
      "loss": 1.9907,
      "step": 4920
    },
    {
      "epoch": 0.7644271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.00013857524557293638,
      "loss": 2.1495,
      "step": 4921
    },
    {
      "epoch": 0.7645825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.00013840142799315924,
      "loss": 2.1929,
      "step": 4922
    },
    {
      "epoch": 0.7647378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.00013822770198054822,
      "loss": 2.0676,
      "step": 4923
    },
    {
      "epoch": 0.7648932038834951,
      "grad_norm": 0.0,
      "learning_rate": 0.00013805406757909556,
      "loss": 2.1334,
      "step": 4924
    },
    {
      "epoch": 0.7650485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.0001378805248327707,
      "loss": 2.0849,
      "step": 4925
    },
    {
      "epoch": 0.7652038834951457,
      "grad_norm": 0.0,
      "learning_rate": 0.0001377070737855195,
      "loss": 2.1494,
      "step": 4926
    },
    {
      "epoch": 0.7653592233009708,
      "grad_norm": 0.0,
      "learning_rate": 0.00013753371448126467,
      "loss": 2.0433,
      "step": 4927
    },
    {
      "epoch": 0.7655145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.00013736044696390604,
      "loss": 2.1326,
      "step": 4928
    },
    {
      "epoch": 0.7656699029126214,
      "grad_norm": 0.0,
      "learning_rate": 0.0001371872712773196,
      "loss": 2.0374,
      "step": 4929
    },
    {
      "epoch": 0.7658252427184467,
      "grad_norm": 0.0,
      "learning_rate": 0.00013701418746535867,
      "loss": 2.2051,
      "step": 4930
    },
    {
      "epoch": 0.7659805825242718,
      "grad_norm": 0.0,
      "learning_rate": 0.0001368411955718531,
      "loss": 2.113,
      "step": 4931
    },
    {
      "epoch": 0.7661359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.00013666829564060916,
      "loss": 2.1567,
      "step": 4932
    },
    {
      "epoch": 0.7662912621359224,
      "grad_norm": 0.0,
      "learning_rate": 0.00013649548771541036,
      "loss": 2.0922,
      "step": 4933
    },
    {
      "epoch": 0.7664466019417475,
      "grad_norm": 0.0,
      "learning_rate": 0.0001363227718400164,
      "loss": 2.2065,
      "step": 4934
    },
    {
      "epoch": 0.7666019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.00013615014805816417,
      "loss": 2.1866,
      "step": 4935
    },
    {
      "epoch": 0.7667572815533981,
      "grad_norm": 0.0,
      "learning_rate": 0.00013597761641356672,
      "loss": 2.0478,
      "step": 4936
    },
    {
      "epoch": 0.7669126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.00013580517694991424,
      "loss": 2.1126,
      "step": 4937
    },
    {
      "epoch": 0.7670679611650485,
      "grad_norm": 0.0,
      "learning_rate": 0.0001356328297108733,
      "loss": 2.2478,
      "step": 4938
    },
    {
      "epoch": 0.7672233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.000135460574740087,
      "loss": 2.0953,
      "step": 4939
    },
    {
      "epoch": 0.767378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.00013528841208117537,
      "loss": 2.1519,
      "step": 4940
    },
    {
      "epoch": 0.7675339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.0001351163417777352,
      "loss": 2.1596,
      "step": 4941
    },
    {
      "epoch": 0.7676893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.0001349443638733392,
      "loss": 2.0457,
      "step": 4942
    },
    {
      "epoch": 0.7678446601941747,
      "grad_norm": 0.0,
      "learning_rate": 0.00013477247841153756,
      "loss": 2.1725,
      "step": 4943
    },
    {
      "epoch": 0.768,
      "grad_norm": 0.0,
      "learning_rate": 0.00013460068543585623,
      "loss": 2.1796,
      "step": 4944
    },
    {
      "epoch": 0.7681553398058253,
      "grad_norm": 0.0,
      "learning_rate": 0.00013442898498979838,
      "loss": 2.1453,
      "step": 4945
    },
    {
      "epoch": 0.7683106796116504,
      "grad_norm": 0.0,
      "learning_rate": 0.0001342573771168436,
      "loss": 2.1431,
      "step": 4946
    },
    {
      "epoch": 0.7684660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.00013408586186044763,
      "loss": 2.1581,
      "step": 4947
    },
    {
      "epoch": 0.768621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.00013391443926404338,
      "loss": 2.0515,
      "step": 4948
    },
    {
      "epoch": 0.7687766990291263,
      "grad_norm": 0.0,
      "learning_rate": 0.00013374310937103984,
      "loss": 2.1381,
      "step": 4949
    },
    {
      "epoch": 0.7689320388349514,
      "grad_norm": 0.0,
      "learning_rate": 0.00013357187222482253,
      "loss": 2.0925,
      "step": 4950
    },
    {
      "epoch": 0.7690873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.0001334007278687539,
      "loss": 2.1465,
      "step": 4951
    },
    {
      "epoch": 0.769242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.00013322967634617238,
      "loss": 2.1683,
      "step": 4952
    },
    {
      "epoch": 0.7693980582524271,
      "grad_norm": 0.0,
      "learning_rate": 0.00013305871770039336,
      "loss": 2.054,
      "step": 4953
    },
    {
      "epoch": 0.7695533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.00013288785197470827,
      "loss": 2.2236,
      "step": 4954
    },
    {
      "epoch": 0.7697087378640777,
      "grad_norm": 0.0,
      "learning_rate": 0.0001327170792123853,
      "loss": 2.0823,
      "step": 4955
    },
    {
      "epoch": 0.769864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.00013254639945666925,
      "loss": 2.0929,
      "step": 4956
    },
    {
      "epoch": 0.7700194174757281,
      "grad_norm": 0.0,
      "learning_rate": 0.0001323758127507808,
      "loss": 2.1373,
      "step": 4957
    },
    {
      "epoch": 0.7701747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0001322053191379177,
      "loss": 2.1191,
      "step": 4958
    },
    {
      "epoch": 0.7703300970873787,
      "grad_norm": 0.0,
      "learning_rate": 0.0001320349186612538,
      "loss": 2.2363,
      "step": 4959
    },
    {
      "epoch": 0.7704854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.00013186461136393906,
      "loss": 2.1669,
      "step": 4960
    },
    {
      "epoch": 0.7706407766990291,
      "grad_norm": 0.0,
      "learning_rate": 0.0001316943972891007,
      "loss": 2.0645,
      "step": 4961
    },
    {
      "epoch": 0.7707961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.00013152427647984138,
      "loss": 2.054,
      "step": 4962
    },
    {
      "epoch": 0.7709514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.00013135424897924082,
      "loss": 1.9649,
      "step": 4963
    },
    {
      "epoch": 0.7711067961165049,
      "grad_norm": 0.0,
      "learning_rate": 0.0001311843148303549,
      "loss": 2.1431,
      "step": 4964
    },
    {
      "epoch": 0.7712621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.0001310144740762157,
      "loss": 2.142,
      "step": 4965
    },
    {
      "epoch": 0.7714174757281553,
      "grad_norm": 0.0,
      "learning_rate": 0.000130844726759832,
      "loss": 2.1697,
      "step": 4966
    },
    {
      "epoch": 0.7715728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.00013067507292418846,
      "loss": 2.2088,
      "step": 4967
    },
    {
      "epoch": 0.7717281553398059,
      "grad_norm": 0.0,
      "learning_rate": 0.0001305055126122465,
      "loss": 2.1526,
      "step": 4968
    },
    {
      "epoch": 0.771883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.00013033604586694348,
      "loss": 2.1592,
      "step": 4969
    },
    {
      "epoch": 0.7720388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.00013016667273119337,
      "loss": 2.1209,
      "step": 4970
    },
    {
      "epoch": 0.7721941747572816,
      "grad_norm": 0.0,
      "learning_rate": 0.00012999739324788662,
      "loss": 2.0823,
      "step": 4971
    },
    {
      "epoch": 0.7723495145631067,
      "grad_norm": 0.0,
      "learning_rate": 0.0001298282074598891,
      "loss": 2.1614,
      "step": 4972
    },
    {
      "epoch": 0.772504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.00012965911541004376,
      "loss": 2.0712,
      "step": 4973
    },
    {
      "epoch": 0.7726601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.00012949011714116982,
      "loss": 2.1058,
      "step": 4974
    },
    {
      "epoch": 0.7728155339805826,
      "grad_norm": 0.0,
      "learning_rate": 0.00012932121269606207,
      "loss": 2.1523,
      "step": 4975
    },
    {
      "epoch": 0.7729708737864077,
      "grad_norm": 0.0,
      "learning_rate": 0.0001291524021174924,
      "loss": 2.0916,
      "step": 4976
    },
    {
      "epoch": 0.773126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.0001289836854482081,
      "loss": 2.179,
      "step": 4977
    },
    {
      "epoch": 0.7732815533980583,
      "grad_norm": 0.0,
      "learning_rate": 0.00012881506273093336,
      "loss": 2.1343,
      "step": 4978
    },
    {
      "epoch": 0.7734368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.00012864653400836834,
      "loss": 2.114,
      "step": 4979
    },
    {
      "epoch": 0.7735922330097087,
      "grad_norm": 0.0,
      "learning_rate": 0.00012847809932318914,
      "loss": 2.1024,
      "step": 4980
    },
    {
      "epoch": 0.773747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.00012830975871804857,
      "loss": 2.0739,
      "step": 4981
    },
    {
      "epoch": 0.7739029126213592,
      "grad_norm": 0.0,
      "learning_rate": 0.00012814151223557507,
      "loss": 2.1584,
      "step": 4982
    },
    {
      "epoch": 0.7740582524271845,
      "grad_norm": 0.0,
      "learning_rate": 0.00012797335991837346,
      "loss": 2.0004,
      "step": 4983
    },
    {
      "epoch": 0.7742135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.000127805301809025,
      "loss": 2.111,
      "step": 4984
    },
    {
      "epoch": 0.774368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.00012763733795008653,
      "loss": 2.0607,
      "step": 4985
    },
    {
      "epoch": 0.7745242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.0001274694683840916,
      "loss": 2.0752,
      "step": 4986
    },
    {
      "epoch": 0.7746796116504855,
      "grad_norm": 0.0,
      "learning_rate": 0.0001273016931535494,
      "loss": 2.063,
      "step": 4987
    },
    {
      "epoch": 0.7748349514563106,
      "grad_norm": 0.0,
      "learning_rate": 0.00012713401230094557,
      "loss": 2.056,
      "step": 4988
    },
    {
      "epoch": 0.7749902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.00012696642586874186,
      "loss": 2.0929,
      "step": 4989
    },
    {
      "epoch": 0.7751456310679612,
      "grad_norm": 0.0,
      "learning_rate": 0.00012679893389937563,
      "loss": 2.047,
      "step": 4990
    },
    {
      "epoch": 0.7753009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.0001266315364352611,
      "loss": 2.0833,
      "step": 4991
    },
    {
      "epoch": 0.7754563106796116,
      "grad_norm": 0.0,
      "learning_rate": 0.00012646423351878765,
      "loss": 2.1267,
      "step": 4992
    },
    {
      "epoch": 0.7756116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.0001262970251923215,
      "loss": 2.1624,
      "step": 4993
    },
    {
      "epoch": 0.7757669902912622,
      "grad_norm": 0.0,
      "learning_rate": 0.00012612991149820485,
      "loss": 2.1053,
      "step": 4994
    },
    {
      "epoch": 0.7759223300970873,
      "grad_norm": 0.0,
      "learning_rate": 0.0001259628924787551,
      "loss": 2.1234,
      "step": 4995
    },
    {
      "epoch": 0.7760776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.00012579596817626653,
      "loss": 2.1711,
      "step": 4996
    },
    {
      "epoch": 0.7762330097087379,
      "grad_norm": 0.0,
      "learning_rate": 0.0001256291386330094,
      "loss": 2.1028,
      "step": 4997
    },
    {
      "epoch": 0.7763883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.00012546240389122942,
      "loss": 2.1459,
      "step": 4998
    },
    {
      "epoch": 0.7765436893203883,
      "grad_norm": 0.0,
      "learning_rate": 0.0001252957639931489,
      "loss": 2.2128,
      "step": 4999
    },
    {
      "epoch": 0.7766990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.00012512921898096558,
      "loss": 2.1362,
      "step": 5000
    },
    {
      "epoch": 0.7768543689320389,
      "grad_norm": 0.0,
      "learning_rate": 0.00012496276889685377,
      "loss": 2.0468,
      "step": 5001
    },
    {
      "epoch": 0.7770097087378641,
      "grad_norm": 0.0,
      "learning_rate": 0.00012479641378296312,
      "loss": 2.1268,
      "step": 5002
    },
    {
      "epoch": 0.7771650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.00012463015368141967,
      "loss": 2.0066,
      "step": 5003
    },
    {
      "epoch": 0.7773203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.0001244639886343254,
      "loss": 2.0598,
      "step": 5004
    },
    {
      "epoch": 0.7774757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.00012429791868375794,
      "loss": 1.9894,
      "step": 5005
    },
    {
      "epoch": 0.7776310679611651,
      "grad_norm": 0.0,
      "learning_rate": 0.00012413194387177084,
      "loss": 2.1354,
      "step": 5006
    },
    {
      "epoch": 0.7777864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.00012396606424039397,
      "loss": 2.0408,
      "step": 5007
    },
    {
      "epoch": 0.7779417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.0001238002798316326,
      "loss": 2.0455,
      "step": 5008
    },
    {
      "epoch": 0.7780970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.00012363459068746836,
      "loss": 2.1118,
      "step": 5009
    },
    {
      "epoch": 0.778252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.0001234689968498582,
      "loss": 2.1225,
      "step": 5010
    },
    {
      "epoch": 0.7784077669902912,
      "grad_norm": 0.0,
      "learning_rate": 0.0001233034983607354,
      "loss": 2.0591,
      "step": 5011
    },
    {
      "epoch": 0.7785631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.0001231380952620091,
      "loss": 1.9409,
      "step": 5012
    },
    {
      "epoch": 0.7787184466019418,
      "grad_norm": 0.0,
      "learning_rate": 0.00012297278759556385,
      "loss": 2.1164,
      "step": 5013
    },
    {
      "epoch": 0.7788737864077669,
      "grad_norm": 0.0,
      "learning_rate": 0.0001228075754032606,
      "loss": 2.1042,
      "step": 5014
    },
    {
      "epoch": 0.7790291262135922,
      "grad_norm": 0.0,
      "learning_rate": 0.00012264245872693546,
      "loss": 2.1157,
      "step": 5015
    },
    {
      "epoch": 0.7791844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.00012247743760840103,
      "loss": 2.1798,
      "step": 5016
    },
    {
      "epoch": 0.7793398058252428,
      "grad_norm": 0.0,
      "learning_rate": 0.00012231251208944555,
      "loss": 2.1127,
      "step": 5017
    },
    {
      "epoch": 0.7794951456310679,
      "grad_norm": 0.0,
      "learning_rate": 0.00012214768221183238,
      "loss": 2.0992,
      "step": 5018
    },
    {
      "epoch": 0.7796504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.00012198294801730158,
      "loss": 2.0374,
      "step": 5019
    },
    {
      "epoch": 0.7798058252427185,
      "grad_norm": 0.0,
      "learning_rate": 0.0001218183095475684,
      "loss": 2.1306,
      "step": 5020
    },
    {
      "epoch": 0.7799611650485437,
      "grad_norm": 0.0,
      "learning_rate": 0.0001216537668443241,
      "loss": 2.058,
      "step": 5021
    },
    {
      "epoch": 0.7801165048543689,
      "grad_norm": 0.0,
      "learning_rate": 0.00012148931994923585,
      "loss": 2.1451,
      "step": 5022
    },
    {
      "epoch": 0.7802718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.00012132496890394596,
      "loss": 2.0816,
      "step": 5023
    },
    {
      "epoch": 0.7804271844660194,
      "grad_norm": 0.0,
      "learning_rate": 0.0001211607137500732,
      "loss": 2.0878,
      "step": 5024
    },
    {
      "epoch": 0.7805825242718447,
      "grad_norm": 0.0,
      "learning_rate": 0.00012099655452921132,
      "loss": 2.107,
      "step": 5025
    },
    {
      "epoch": 0.7807378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.00012083249128293045,
      "loss": 2.1708,
      "step": 5026
    },
    {
      "epoch": 0.7808932038834951,
      "grad_norm": 0.0,
      "learning_rate": 0.00012066852405277611,
      "loss": 2.1636,
      "step": 5027
    },
    {
      "epoch": 0.7810485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.00012050465288026951,
      "loss": 2.1676,
      "step": 5028
    },
    {
      "epoch": 0.7812038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.00012034087780690739,
      "loss": 2.0687,
      "step": 5029
    },
    {
      "epoch": 0.7813592233009709,
      "grad_norm": 0.0,
      "learning_rate": 0.00012017719887416251,
      "loss": 2.1128,
      "step": 5030
    },
    {
      "epoch": 0.7815145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.00012001361612348283,
      "loss": 2.1126,
      "step": 5031
    },
    {
      "epoch": 0.7816699029126214,
      "grad_norm": 0.0,
      "learning_rate": 0.00011985012959629255,
      "loss": 2.1943,
      "step": 5032
    },
    {
      "epoch": 0.7818252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.00011968673933399089,
      "loss": 2.1368,
      "step": 5033
    },
    {
      "epoch": 0.7819805825242718,
      "grad_norm": 0.0,
      "learning_rate": 0.00011952344537795312,
      "loss": 2.0492,
      "step": 5034
    },
    {
      "epoch": 0.7821359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.00011936024776952981,
      "loss": 2.123,
      "step": 5035
    },
    {
      "epoch": 0.7822912621359224,
      "grad_norm": 0.0,
      "learning_rate": 0.00011919714655004742,
      "loss": 2.0353,
      "step": 5036
    },
    {
      "epoch": 0.7824466019417475,
      "grad_norm": 0.0,
      "learning_rate": 0.00011903414176080795,
      "loss": 2.1472,
      "step": 5037
    },
    {
      "epoch": 0.7826019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.0001188712334430887,
      "loss": 1.9975,
      "step": 5038
    },
    {
      "epoch": 0.7827572815533981,
      "grad_norm": 0.0,
      "learning_rate": 0.000118708421638143,
      "loss": 2.2692,
      "step": 5039
    },
    {
      "epoch": 0.7829126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.00011854570638719936,
      "loss": 2.1559,
      "step": 5040
    },
    {
      "epoch": 0.7830679611650485,
      "grad_norm": 0.0,
      "learning_rate": 0.00011838308773146172,
      "loss": 2.1765,
      "step": 5041
    },
    {
      "epoch": 0.7832233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.0001182205657121102,
      "loss": 2.1146,
      "step": 5042
    },
    {
      "epoch": 0.783378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.00011805814037029978,
      "loss": 2.0656,
      "step": 5043
    },
    {
      "epoch": 0.7835339805825243,
      "grad_norm": 0.0,
      "learning_rate": 0.00011789581174716129,
      "loss": 2.0762,
      "step": 5044
    },
    {
      "epoch": 0.7836893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.00011773357988380118,
      "loss": 2.1804,
      "step": 5045
    },
    {
      "epoch": 0.7838446601941748,
      "grad_norm": 0.0,
      "learning_rate": 0.00011757144482130094,
      "loss": 2.1339,
      "step": 5046
    },
    {
      "epoch": 0.784,
      "grad_norm": 0.0,
      "learning_rate": 0.0001174094066007182,
      "loss": 2.1091,
      "step": 5047
    },
    {
      "epoch": 0.7841553398058252,
      "grad_norm": 0.0,
      "learning_rate": 0.00011724746526308527,
      "loss": 2.0884,
      "step": 5048
    },
    {
      "epoch": 0.7843106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.00011708562084941065,
      "loss": 2.0776,
      "step": 5049
    },
    {
      "epoch": 0.7844660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.000116923873400678,
      "loss": 2.1741,
      "step": 5050
    },
    {
      "epoch": 0.784621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.00011676222295784644,
      "loss": 2.0947,
      "step": 5051
    },
    {
      "epoch": 0.7847766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0001166006695618504,
      "loss": 2.027,
      "step": 5052
    },
    {
      "epoch": 0.7849320388349514,
      "grad_norm": 0.0,
      "learning_rate": 0.00011643921325359974,
      "loss": 2.1774,
      "step": 5053
    },
    {
      "epoch": 0.7850873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.00011627785407398001,
      "loss": 2.1311,
      "step": 5054
    },
    {
      "epoch": 0.785242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.00011611659206385211,
      "loss": 2.0353,
      "step": 5055
    },
    {
      "epoch": 0.7853980582524271,
      "grad_norm": 0.0,
      "learning_rate": 0.00011595542726405195,
      "loss": 2.2677,
      "step": 5056
    },
    {
      "epoch": 0.7855533980582524,
      "grad_norm": 0.0,
      "learning_rate": 0.00011579435971539137,
      "loss": 2.0574,
      "step": 5057
    },
    {
      "epoch": 0.7857087378640777,
      "grad_norm": 0.0,
      "learning_rate": 0.00011563338945865704,
      "loss": 2.0769,
      "step": 5058
    },
    {
      "epoch": 0.785864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.00011547251653461144,
      "loss": 2.0796,
      "step": 5059
    },
    {
      "epoch": 0.7860194174757281,
      "grad_norm": 0.0,
      "learning_rate": 0.00011531174098399227,
      "loss": 2.1246,
      "step": 5060
    },
    {
      "epoch": 0.7861747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0001151510628475123,
      "loss": 2.161,
      "step": 5061
    },
    {
      "epoch": 0.7863300970873787,
      "grad_norm": 0.0,
      "learning_rate": 0.00011499048216586017,
      "loss": 2.0877,
      "step": 5062
    },
    {
      "epoch": 0.7864854368932039,
      "grad_norm": 0.0,
      "learning_rate": 0.00011482999897969937,
      "loss": 2.1503,
      "step": 5063
    },
    {
      "epoch": 0.7866407766990291,
      "grad_norm": 0.0,
      "learning_rate": 0.00011466961332966869,
      "loss": 2.2214,
      "step": 5064
    },
    {
      "epoch": 0.7867961165048544,
      "grad_norm": 0.0,
      "learning_rate": 0.0001145093252563827,
      "loss": 2.0742,
      "step": 5065
    },
    {
      "epoch": 0.7869514563106796,
      "grad_norm": 0.0,
      "learning_rate": 0.00011434913480043069,
      "loss": 2.0795,
      "step": 5066
    },
    {
      "epoch": 0.7871067961165048,
      "grad_norm": 0.0,
      "learning_rate": 0.00011418904200237757,
      "loss": 2.0511,
      "step": 5067
    },
    {
      "epoch": 0.7872621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.00011402904690276367,
      "loss": 2.1144,
      "step": 5068
    },
    {
      "epoch": 0.7874174757281553,
      "grad_norm": 0.0,
      "learning_rate": 0.00011386914954210398,
      "loss": 2.0192,
      "step": 5069
    },
    {
      "epoch": 0.7875728155339806,
      "grad_norm": 0.0,
      "learning_rate": 0.0001137093499608895,
      "loss": 2.114,
      "step": 5070
    },
    {
      "epoch": 0.7877281553398058,
      "grad_norm": 0.0,
      "learning_rate": 0.00011354964819958569,
      "loss": 2.1431,
      "step": 5071
    },
    {
      "epoch": 0.787883495145631,
      "grad_norm": 0.0,
      "learning_rate": 0.00011339004429863403,
      "loss": 2.1852,
      "step": 5072
    },
    {
      "epoch": 0.7880388349514563,
      "grad_norm": 0.0,
      "learning_rate": 0.00011323053829845042,
      "loss": 2.1292,
      "step": 5073
    },
    {
      "epoch": 0.7881941747572816,
      "grad_norm": 0.0,
      "learning_rate": 0.00011307113023942678,
      "loss": 2.0821,
      "step": 5074
    },
    {
      "epoch": 0.7883495145631068,
      "grad_norm": 0.0,
      "learning_rate": 0.00011291182016192959,
      "loss": 2.1536,
      "step": 5075
    },
    {
      "epoch": 0.788504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.00011275260810630062,
      "loss": 2.1426,
      "step": 5076
    },
    {
      "epoch": 0.7886601941747573,
      "grad_norm": 0.0,
      "learning_rate": 0.00011259349411285707,
      "loss": 2.0645,
      "step": 5077
    },
    {
      "epoch": 0.7888155339805826,
      "grad_norm": 0.0,
      "learning_rate": 0.00011243447822189134,
      "loss": 2.1046,
      "step": 5078
    },
    {
      "epoch": 0.7889708737864077,
      "grad_norm": 0.0,
      "learning_rate": 0.0001122755604736706,
      "loss": 1.9678,
      "step": 5079
    },
    {
      "epoch": 0.789126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.00011211674090843765,
      "loss": 1.9845,
      "step": 5080
    },
    {
      "epoch": 0.7892815533980583,
      "grad_norm": 0.0,
      "learning_rate": 0.00011195801956640983,
      "loss": 2.0801,
      "step": 5081
    },
    {
      "epoch": 0.7894368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.00011179939648778015,
      "loss": 2.003,
      "step": 5082
    },
    {
      "epoch": 0.7895922330097087,
      "grad_norm": 0.0,
      "learning_rate": 0.0001116408717127167,
      "loss": 2.0641,
      "step": 5083
    },
    {
      "epoch": 0.789747572815534,
      "grad_norm": 0.0,
      "learning_rate": 0.0001114824452813622,
      "loss": 2.0967,
      "step": 5084
    },
    {
      "epoch": 0.7899029126213593,
      "grad_norm": 0.0,
      "learning_rate": 0.00011132411723383518,
      "loss": 2.0107,
      "step": 5085
    },
    {
      "epoch": 0.7900582524271844,
      "grad_norm": 0.0,
      "learning_rate": 0.00011116588761022833,
      "loss": 2.1711,
      "step": 5086
    },
    {
      "epoch": 0.7902135922330097,
      "grad_norm": 0.0,
      "learning_rate": 0.00011100775645061023,
      "loss": 1.9615,
      "step": 5087
    },
    {
      "epoch": 0.790368932038835,
      "grad_norm": 0.0,
      "learning_rate": 0.0001108497237950244,
      "loss": 2.0473,
      "step": 5088
    },
    {
      "epoch": 0.7905242718446602,
      "grad_norm": 0.0,
      "learning_rate": 0.00011069178968348892,
      "loss": 2.0516,
      "step": 5089
    },
    {
      "epoch": 0.7906796116504854,
      "grad_norm": 0.0,
      "learning_rate": 0.00011053395415599755,
      "loss": 2.1172,
      "step": 5090
    },
    {
      "epoch": 0.7908349514563107,
      "grad_norm": 0.0,
      "learning_rate": 0.00011037621725251851,
      "loss": 2.1716,
      "step": 5091
    },
    {
      "epoch": 0.7909902912621359,
      "grad_norm": 0.0,
      "learning_rate": 0.00011021857901299553,
      "loss": 2.1797,
      "step": 5092
    },
    {
      "epoch": 0.7911456310679612,
      "grad_norm": 0.0,
      "learning_rate": 0.00011006103947734719,
      "loss": 2.1985,
      "step": 5093
    },
    {
      "epoch": 0.7913009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.00010990359868546684,
      "loss": 2.2339,
      "step": 5094
    },
    {
      "epoch": 0.7914563106796116,
      "grad_norm": 0.0,
      "learning_rate": 0.00010974625667722321,
      "loss": 2.162,
      "step": 5095
    },
    {
      "epoch": 0.7916116504854369,
      "grad_norm": 0.0,
      "learning_rate": 0.00010958901349245975,
      "loss": 2.1223,
      "step": 5096
    },
    {
      "epoch": 0.7917669902912622,
      "grad_norm": 0.0,
      "learning_rate": 0.00010943186917099485,
      "loss": 2.1384,
      "step": 5097
    },
    {
      "epoch": 0.7919223300970873,
      "grad_norm": 0.0,
      "learning_rate": 0.00010927482375262216,
      "loss": 2.1429,
      "step": 5098
    },
    {
      "epoch": 0.7920776699029126,
      "grad_norm": 0.0,
      "learning_rate": 0.00010911787727710986,
      "loss": 2.2081,
      "step": 5099
    },
    {
      "epoch": 0.7922330097087379,
      "grad_norm": 0.0,
      "learning_rate": 0.00010896102978420153,
      "loss": 2.0912,
      "step": 5100
    },
    {
      "epoch": 0.7923883495145632,
      "grad_norm": 0.0,
      "learning_rate": 0.00010880428131361547,
      "loss": 2.1483,
      "step": 5101
    },
    {
      "epoch": 0.7925436893203883,
      "grad_norm": 0.0,
      "learning_rate": 0.00010864763190504473,
      "loss": 1.999,
      "step": 5102
    },
    {
      "epoch": 0.7926990291262136,
      "grad_norm": 0.0,
      "learning_rate": 0.00010849108159815774,
      "loss": 2.1703,
      "step": 5103
    },
    {
      "epoch": 0.7928543689320389,
      "grad_norm": 0.0,
      "learning_rate": 0.00010833463043259717,
      "loss": 2.0467,
      "step": 5104
    },
    {
      "epoch": 0.793009708737864,
      "grad_norm": 0.0,
      "learning_rate": 0.00010817827844798129,
      "loss": 2.1443,
      "step": 5105
    },
    {
      "epoch": 0.7931650485436893,
      "grad_norm": 0.0,
      "learning_rate": 0.00010802202568390263,
      "loss": 2.2367,
      "step": 5106
    },
    {
      "epoch": 0.7933203883495146,
      "grad_norm": 0.0,
      "learning_rate": 0.00010786587217992905,
      "loss": 2.0836,
      "step": 5107
    },
    {
      "epoch": 0.7934757281553398,
      "grad_norm": 0.0,
      "learning_rate": 0.00010770981797560342,
      "loss": 2.0153,
      "step": 5108
    },
    {
      "epoch": 0.793631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.00010755386311044252,
      "loss": 2.1132,
      "step": 5109
    },
    {
      "epoch": 0.7937864077669903,
      "grad_norm": 0.0,
      "learning_rate": 0.00010739800762393892,
      "loss": 2.0785,
      "step": 5110
    },
    {
      "epoch": 0.7939417475728155,
      "grad_norm": 0.0,
      "learning_rate": 0.00010724225155555988,
      "loss": 2.1941,
      "step": 5111
    },
    {
      "epoch": 0.7940970873786408,
      "grad_norm": 0.0,
      "learning_rate": 0.00010708659494474699,
      "loss": 2.0877,
      "step": 5112
    },
    {
      "epoch": 0.794252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.00010693103783091735,
      "loss": 2.1828,
      "step": 5113
    },
    {
      "epoch": 0.7944077669902913,
      "grad_norm": 0.0,
      "learning_rate": 0.00010677558025346212,
      "loss": 2.2637,
      "step": 5114
    },
    {
      "epoch": 0.7945631067961165,
      "grad_norm": 0.0,
      "learning_rate": 0.00010662022225174783,
      "loss": 2.0763,
      "step": 5115
    },
    {
      "epoch": 0.7947184466019418,
      "grad_norm": 0.0,
      "learning_rate": 0.00010646496386511572,
      "loss": 2.1002,
      "step": 5116
    },
    {
      "epoch": 0.794873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.00010630980513288141,
      "loss": 2.1684,
      "step": 5117
    },
    {
      "epoch": 0.7950291262135922,
      "grad_norm": 0.0,
      "learning_rate": 0.00010615474609433584,
      "loss": 2.0038,
      "step": 5118
    },
    {
      "epoch": 0.7951844660194175,
      "grad_norm": 0.0,
      "learning_rate": 0.00010599978678874428,
      "loss": 2.1421,
      "step": 5119
    },
    {
      "epoch": 0.7953398058252428,
      "grad_norm": 0.0,
      "learning_rate": 0.00010584492725534672,
      "loss": 2.0936,
      "step": 5120
    },
    {
      "epoch": 0.7954951456310679,
      "grad_norm": 0.0,
      "learning_rate": 0.00010569016753335841,
      "loss": 2.2111,
      "step": 5121
    },
    {
      "epoch": 0.7956504854368932,
      "grad_norm": 0.0,
      "learning_rate": 0.00010553550766196856,
      "loss": 2.2134,
      "step": 5122
    },
    {
      "epoch": 0.7958058252427185,
      "grad_norm": 0.0,
      "learning_rate": 0.0001053809476803419,
      "loss": 2.1919,
      "step": 5123
    },
    {
      "epoch": 0.7959611650485436,
      "grad_norm": 0.0,
      "learning_rate": 0.0001052264876276171,
      "loss": 2.0239,
      "step": 5124
    },
    {
      "epoch": 0.7961165048543689,
      "grad_norm": 0.0,
      "learning_rate": 0.00010507212754290807,
      "loss": 2.1848,
      "step": 5125
    },
    {
      "epoch": 0.7962718446601942,
      "grad_norm": 0.0,
      "learning_rate": 0.00010491786746530335,
      "loss": 2.1829,
      "step": 5126
    },
    {
      "epoch": 0.7964271844660195,
      "grad_norm": 0.0,
      "learning_rate": 0.00010476370743386581,
      "loss": 2.0473,
      "step": 5127
    },
    {
      "epoch": 0.7965825242718446,
      "grad_norm": 0.0,
      "learning_rate": 0.0001046096474876334,
      "loss": 2.1201,
      "step": 5128
    },
    {
      "epoch": 0.7967378640776699,
      "grad_norm": 0.0,
      "learning_rate": 0.0001044556876656183,
      "loss": 2.1921,
      "step": 5129
    },
    {
      "epoch": 0.7968932038834952,
      "grad_norm": 0.0,
      "learning_rate": 0.00010430182800680771,
      "loss": 2.0426,
      "step": 5130
    },
    {
      "epoch": 0.7970485436893204,
      "grad_norm": 0.0,
      "learning_rate": 0.00010414806855016363,
      "loss": 2.0867,
      "step": 5131
    },
    {
      "epoch": 0.7972038834951456,
      "grad_norm": 0.0,
      "learning_rate": 0.00010399440933462173,
      "loss": 2.0936,
      "step": 5132
    },
    {
      "epoch": 0.7973592233009709,
      "grad_norm": 0.0,
      "learning_rate": 0.0001038408503990933,
      "loss": 1.9955,
      "step": 5133
    },
    {
      "epoch": 0.7975145631067961,
      "grad_norm": 0.0,
      "learning_rate": 0.000103687391782464,
      "loss": 2.1512,
      "step": 5134
    },
    {
      "epoch": 0.7976699029126214,
      "grad_norm": 0.0,
      "learning_rate": 0.00010353403352359375,
      "loss": 2.123,
      "step": 5135
    },
    {
      "epoch": 0.7978252427184466,
      "grad_norm": 0.0,
      "learning_rate": 0.00010338077566131743,
      "loss": 2.0947,
      "step": 5136
    },
    {
      "epoch": 0.7979805825242718,
      "grad_norm": 0.0,
      "learning_rate": 0.00010322761823444415,
      "loss": 2.1917,
      "step": 5137
    },
    {
      "epoch": 0.7981359223300971,
      "grad_norm": 0.0,
      "learning_rate": 0.00010307456128175802,
      "loss": 2.1296,
      "step": 5138
    },
    {
      "epoch": 0.7982912621359223,
      "grad_norm": 0.0,
      "learning_rate": 0.00010292160484201724,
      "loss": 2.0598,
      "step": 5139
    },
    {
      "epoch": 0.7984466019417475,
      "grad_norm": 0.0,
      "learning_rate": 0.00010276874895395489,
      "loss": 2.1414,
      "step": 5140
    },
    {
      "epoch": 0.7986019417475728,
      "grad_norm": 0.0,
      "learning_rate": 0.00010261599365627855,
      "loss": 2.0651,
      "step": 5141
    },
    {
      "epoch": 0.7987572815533981,
      "grad_norm": 0.0,
      "learning_rate": 0.00010246333898767023,
      "loss": 2.1862,
      "step": 5142
    },
    {
      "epoch": 0.7989126213592233,
      "grad_norm": 0.0,
      "learning_rate": 0.00010231078498678631,
      "loss": 2.1702,
      "step": 5143
    },
    {
      "epoch": 0.7990679611650485,
      "grad_norm": 0.0,
      "learning_rate": 0.00010215833169225814,
      "loss": 2.0334,
      "step": 5144
    },
    {
      "epoch": 0.7992233009708738,
      "grad_norm": 0.0,
      "learning_rate": 0.00010200597914269094,
      "loss": 2.1506,
      "step": 5145
    },
    {
      "epoch": 0.7993786407766991,
      "grad_norm": 0.0,
      "learning_rate": 0.0001018537273766651,
      "loss": 2.1458,
      "step": 5146
    },
    {
      "epoch": 0.7995339805825242,
      "grad_norm": 0.0,
      "learning_rate": 0.00010170157643273487,
      "loss": 2.0579,
      "step": 5147
    },
    {
      "epoch": 0.7996893203883495,
      "grad_norm": 0.0,
      "learning_rate": 0.00010154952634942937,
      "loss": 2.0536,
      "step": 5148
    },
    {
      "epoch": 0.7998446601941748,
      "grad_norm": 0.0,
      "learning_rate": 0.00010139757716525221,
      "loss": 2.0664,
      "step": 5149
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.0,
      "learning_rate": 0.00010124572891868105,
      "loss": 2.2161,
      "step": 5150
    },
    {
      "epoch": 0.8001553398058252,
      "grad_norm": 0.0,
      "learning_rate": 0.00010109398164816841,
      "loss": 2.074,
      "step": 5151
    },
    {
      "epoch": 0.8003106796116505,
      "grad_norm": 0.0,
      "learning_rate": 0.00010094233539214087,
      "loss": 2.0502,
      "step": 5152
    },
    {
      "epoch": 0.8004660194174757,
      "grad_norm": 0.0,
      "learning_rate": 0.00010079079018899989,
      "loss": 2.0468,
      "step": 5153
    },
    {
      "epoch": 0.800621359223301,
      "grad_norm": 0.0,
      "learning_rate": 0.00010063934607712094,
      "loss": 2.1543,
      "step": 5154
    },
    {
      "epoch": 0.8007766990291262,
      "grad_norm": 0.0,
      "learning_rate": 0.0001004880030948539,
      "loss": 2.099,
      "step": 5155
    },
    {
      "epoch": 0.8009320388349515,
      "grad_norm": 0.0,
      "learning_rate": 0.00010033676128052344,
      "loss": 2.0521,
      "step": 5156
    },
    {
      "epoch": 0.8010873786407767,
      "grad_norm": 0.0,
      "learning_rate": 0.00010018562067242804,
      "loss": 2.0435,
      "step": 5157
    },
    {
      "epoch": 0.8012427184466019,
      "grad_norm": 0.0,
      "learning_rate": 0.00010003458130884096,
      "loss": 2.1512,
      "step": 5158
    },
    {
      "epoch": 0.8013980582524272,
      "grad_norm": 0.0,
      "learning_rate": 9.988364322800996e-05,
      "loss": 2.0844,
      "step": 5159
    },
    {
      "epoch": 0.8015533980582524,
      "grad_norm": 0.0,
      "learning_rate": 9.973280646815658e-05,
      "loss": 2.1099,
      "step": 5160
    },
    {
      "epoch": 0.8017087378640777,
      "grad_norm": 0.0,
      "learning_rate": 9.958207106747724e-05,
      "loss": 2.0723,
      "step": 5161
    },
    {
      "epoch": 0.8018640776699029,
      "grad_norm": 0.0,
      "learning_rate": 9.943143706414231e-05,
      "loss": 2.1863,
      "step": 5162
    },
    {
      "epoch": 0.8020194174757281,
      "grad_norm": 0.0,
      "learning_rate": 9.928090449629668e-05,
      "loss": 2.106,
      "step": 5163
    },
    {
      "epoch": 0.8021747572815534,
      "grad_norm": 0.0,
      "learning_rate": 9.913047340205978e-05,
      "loss": 2.085,
      "step": 5164
    },
    {
      "epoch": 0.8023300970873787,
      "grad_norm": 0.0,
      "learning_rate": 9.898014381952491e-05,
      "loss": 2.1707,
      "step": 5165
    },
    {
      "epoch": 0.8024854368932038,
      "grad_norm": 0.0,
      "learning_rate": 9.882991578675966e-05,
      "loss": 2.0939,
      "step": 5166
    },
    {
      "epoch": 0.8026407766990291,
      "grad_norm": 0.0,
      "learning_rate": 9.867978934180643e-05,
      "loss": 2.217,
      "step": 5167
    },
    {
      "epoch": 0.8027961165048544,
      "grad_norm": 0.0,
      "learning_rate": 9.85297645226813e-05,
      "loss": 2.1523,
      "step": 5168
    },
    {
      "epoch": 0.8029514563106797,
      "grad_norm": 0.0,
      "learning_rate": 9.837984136737505e-05,
      "loss": 2.0504,
      "step": 5169
    },
    {
      "epoch": 0.8031067961165048,
      "grad_norm": 0.0,
      "learning_rate": 9.823001991385239e-05,
      "loss": 2.0372,
      "step": 5170
    },
    {
      "epoch": 0.8032621359223301,
      "grad_norm": 0.0,
      "learning_rate": 9.80803002000526e-05,
      "loss": 2.1716,
      "step": 5171
    },
    {
      "epoch": 0.8034174757281554,
      "grad_norm": 0.0,
      "learning_rate": 9.793068226388874e-05,
      "loss": 2.1461,
      "step": 5172
    },
    {
      "epoch": 0.8035728155339806,
      "grad_norm": 0.0,
      "learning_rate": 9.778116614324862e-05,
      "loss": 2.1816,
      "step": 5173
    },
    {
      "epoch": 0.8037281553398058,
      "grad_norm": 0.0,
      "learning_rate": 9.763175187599404e-05,
      "loss": 2.0676,
      "step": 5174
    },
    {
      "epoch": 0.8038834951456311,
      "grad_norm": 0.0,
      "learning_rate": 9.748243949996077e-05,
      "loss": 2.0619,
      "step": 5175
    },
    {
      "epoch": 0.8040388349514563,
      "grad_norm": 0.0,
      "learning_rate": 9.733322905295932e-05,
      "loss": 2.2104,
      "step": 5176
    },
    {
      "epoch": 0.8041941747572815,
      "grad_norm": 0.0,
      "learning_rate": 9.718412057277388e-05,
      "loss": 2.0743,
      "step": 5177
    },
    {
      "epoch": 0.8043495145631068,
      "grad_norm": 0.0,
      "learning_rate": 9.703511409716287e-05,
      "loss": 2.0643,
      "step": 5178
    },
    {
      "epoch": 0.804504854368932,
      "grad_norm": 0.0,
      "learning_rate": 9.688620966385936e-05,
      "loss": 2.203,
      "step": 5179
    },
    {
      "epoch": 0.8046601941747573,
      "grad_norm": 0.0,
      "learning_rate": 9.673740731056985e-05,
      "loss": 2.1401,
      "step": 5180
    },
    {
      "epoch": 0.8048155339805825,
      "grad_norm": 0.0,
      "learning_rate": 9.658870707497557e-05,
      "loss": 2.0833,
      "step": 5181
    },
    {
      "epoch": 0.8049708737864077,
      "grad_norm": 0.0,
      "learning_rate": 9.644010899473188e-05,
      "loss": 2.1381,
      "step": 5182
    },
    {
      "epoch": 0.805126213592233,
      "grad_norm": 0.0,
      "learning_rate": 9.629161310746781e-05,
      "loss": 2.1505,
      "step": 5183
    },
    {
      "epoch": 0.8052815533980583,
      "grad_norm": 0.0,
      "learning_rate": 9.614321945078697e-05,
      "loss": 1.9984,
      "step": 5184
    },
    {
      "epoch": 0.8054368932038835,
      "grad_norm": 0.0,
      "learning_rate": 9.599492806226673e-05,
      "loss": 2.1556,
      "step": 5185
    },
    {
      "epoch": 0.8055922330097087,
      "grad_norm": 0.0,
      "learning_rate": 9.584673897945894e-05,
      "loss": 1.9349,
      "step": 5186
    },
    {
      "epoch": 0.805747572815534,
      "grad_norm": 0.0,
      "learning_rate": 9.569865223988916e-05,
      "loss": 2.0434,
      "step": 5187
    },
    {
      "epoch": 0.8059029126213593,
      "grad_norm": 0.0,
      "learning_rate": 9.555066788105743e-05,
      "loss": 2.0515,
      "step": 5188
    },
    {
      "epoch": 0.8060582524271844,
      "grad_norm": 0.0,
      "learning_rate": 9.540278594043749e-05,
      "loss": 2.0526,
      "step": 5189
    },
    {
      "epoch": 0.8062135922330097,
      "grad_norm": 0.0,
      "learning_rate": 9.525500645547725e-05,
      "loss": 2.0963,
      "step": 5190
    },
    {
      "epoch": 0.806368932038835,
      "grad_norm": 0.0,
      "learning_rate": 9.510732946359873e-05,
      "loss": 2.1161,
      "step": 5191
    },
    {
      "epoch": 0.8065242718446602,
      "grad_norm": 0.0,
      "learning_rate": 9.495975500219828e-05,
      "loss": 1.9439,
      "step": 5192
    },
    {
      "epoch": 0.8066796116504854,
      "grad_norm": 0.0,
      "learning_rate": 9.481228310864559e-05,
      "loss": 2.0817,
      "step": 5193
    },
    {
      "epoch": 0.8068349514563107,
      "grad_norm": 0.0,
      "learning_rate": 9.466491382028519e-05,
      "loss": 2.0837,
      "step": 5194
    },
    {
      "epoch": 0.806990291262136,
      "grad_norm": 0.0,
      "learning_rate": 9.451764717443484e-05,
      "loss": 2.1074,
      "step": 5195
    },
    {
      "epoch": 0.8071456310679611,
      "grad_norm": 0.0,
      "learning_rate": 9.437048320838692e-05,
      "loss": 2.0535,
      "step": 5196
    },
    {
      "epoch": 0.8073009708737864,
      "grad_norm": 0.0,
      "learning_rate": 9.422342195940769e-05,
      "loss": 2.1813,
      "step": 5197
    },
    {
      "epoch": 0.8074563106796117,
      "grad_norm": 0.0,
      "learning_rate": 9.407646346473698e-05,
      "loss": 2.2296,
      "step": 5198
    },
    {
      "epoch": 0.8076116504854369,
      "grad_norm": 0.0,
      "learning_rate": 9.392960776158921e-05,
      "loss": 1.9882,
      "step": 5199
    },
    {
      "epoch": 0.8077669902912621,
      "grad_norm": 0.0,
      "learning_rate": 9.378285488715238e-05,
      "loss": 2.1256,
      "step": 5200
    },
    {
      "epoch": 0.8079223300970874,
      "grad_norm": 0.0,
      "learning_rate": 9.363620487858832e-05,
      "loss": 2.1106,
      "step": 5201
    },
    {
      "epoch": 0.8080776699029126,
      "grad_norm": 0.0,
      "learning_rate": 9.348965777303336e-05,
      "loss": 2.0768,
      "step": 5202
    },
    {
      "epoch": 0.8082330097087379,
      "grad_norm": 0.0,
      "learning_rate": 9.334321360759718e-05,
      "loss": 2.0599,
      "step": 5203
    },
    {
      "epoch": 0.8083883495145631,
      "grad_norm": 0.0,
      "learning_rate": 9.319687241936386e-05,
      "loss": 2.0883,
      "step": 5204
    },
    {
      "epoch": 0.8085436893203883,
      "grad_norm": 0.0,
      "learning_rate": 9.305063424539107e-05,
      "loss": 2.1233,
      "step": 5205
    },
    {
      "epoch": 0.8086990291262136,
      "grad_norm": 0.0,
      "learning_rate": 9.290449912271048e-05,
      "loss": 2.05,
      "step": 5206
    },
    {
      "epoch": 0.8088543689320389,
      "grad_norm": 0.0,
      "learning_rate": 9.275846708832797e-05,
      "loss": 2.1588,
      "step": 5207
    },
    {
      "epoch": 0.809009708737864,
      "grad_norm": 0.0,
      "learning_rate": 9.261253817922277e-05,
      "loss": 2.0258,
      "step": 5208
    },
    {
      "epoch": 0.8091650485436893,
      "grad_norm": 0.0,
      "learning_rate": 9.24667124323485e-05,
      "loss": 2.1992,
      "step": 5209
    },
    {
      "epoch": 0.8093203883495146,
      "grad_norm": 0.0,
      "learning_rate": 9.232098988463217e-05,
      "loss": 2.1038,
      "step": 5210
    },
    {
      "epoch": 0.8094757281553399,
      "grad_norm": 0.0,
      "learning_rate": 9.217537057297526e-05,
      "loss": 2.1691,
      "step": 5211
    },
    {
      "epoch": 0.809631067961165,
      "grad_norm": 0.0,
      "learning_rate": 9.202985453425256e-05,
      "loss": 2.0974,
      "step": 5212
    },
    {
      "epoch": 0.8097864077669903,
      "grad_norm": 0.0,
      "learning_rate": 9.188444180531285e-05,
      "loss": 2.1231,
      "step": 5213
    },
    {
      "epoch": 0.8099417475728156,
      "grad_norm": 0.0,
      "learning_rate": 9.173913242297894e-05,
      "loss": 2.0604,
      "step": 5214
    },
    {
      "epoch": 0.8100970873786407,
      "grad_norm": 0.0,
      "learning_rate": 9.15939264240474e-05,
      "loss": 2.0933,
      "step": 5215
    },
    {
      "epoch": 0.810252427184466,
      "grad_norm": 0.0,
      "learning_rate": 9.144882384528841e-05,
      "loss": 2.1362,
      "step": 5216
    },
    {
      "epoch": 0.8104077669902913,
      "grad_norm": 0.0,
      "learning_rate": 9.130382472344634e-05,
      "loss": 2.0884,
      "step": 5217
    },
    {
      "epoch": 0.8105631067961165,
      "grad_norm": 0.0,
      "learning_rate": 9.115892909523882e-05,
      "loss": 2.0416,
      "step": 5218
    },
    {
      "epoch": 0.8107184466019417,
      "grad_norm": 0.0,
      "learning_rate": 9.1014136997358e-05,
      "loss": 2.1326,
      "step": 5219
    },
    {
      "epoch": 0.810873786407767,
      "grad_norm": 0.0,
      "learning_rate": 9.0869448466469e-05,
      "loss": 2.107,
      "step": 5220
    },
    {
      "epoch": 0.8110291262135922,
      "grad_norm": 0.0,
      "learning_rate": 9.072486353921134e-05,
      "loss": 2.0667,
      "step": 5221
    },
    {
      "epoch": 0.8111844660194175,
      "grad_norm": 0.0,
      "learning_rate": 9.058038225219828e-05,
      "loss": 2.1616,
      "step": 5222
    },
    {
      "epoch": 0.8113398058252427,
      "grad_norm": 0.0,
      "learning_rate": 9.043600464201613e-05,
      "loss": 2.1132,
      "step": 5223
    },
    {
      "epoch": 0.811495145631068,
      "grad_norm": 0.0,
      "learning_rate": 9.029173074522578e-05,
      "loss": 2.0441,
      "step": 5224
    },
    {
      "epoch": 0.8116504854368932,
      "grad_norm": 0.0,
      "learning_rate": 9.014756059836165e-05,
      "loss": 2.1377,
      "step": 5225
    },
    {
      "epoch": 0.8118058252427185,
      "grad_norm": 0.0,
      "learning_rate": 9.000349423793148e-05,
      "loss": 2.1581,
      "step": 5226
    },
    {
      "epoch": 0.8119611650485437,
      "grad_norm": 0.0,
      "learning_rate": 8.985953170041722e-05,
      "loss": 2.1154,
      "step": 5227
    },
    {
      "epoch": 0.8121165048543689,
      "grad_norm": 0.0,
      "learning_rate": 8.971567302227419e-05,
      "loss": 2.1885,
      "step": 5228
    },
    {
      "epoch": 0.8122718446601942,
      "grad_norm": 0.0,
      "learning_rate": 8.95719182399316e-05,
      "loss": 2.0544,
      "step": 5229
    },
    {
      "epoch": 0.8124271844660195,
      "grad_norm": 0.0,
      "learning_rate": 8.942826738979243e-05,
      "loss": 2.1086,
      "step": 5230
    },
    {
      "epoch": 0.8125825242718446,
      "grad_norm": 0.0,
      "learning_rate": 8.9284720508233e-05,
      "loss": 1.928,
      "step": 5231
    },
    {
      "epoch": 0.8127378640776699,
      "grad_norm": 0.0,
      "learning_rate": 8.914127763160373e-05,
      "loss": 1.9547,
      "step": 5232
    },
    {
      "epoch": 0.8128932038834952,
      "grad_norm": 0.0,
      "learning_rate": 8.899793879622837e-05,
      "loss": 2.0236,
      "step": 5233
    },
    {
      "epoch": 0.8130485436893203,
      "grad_norm": 0.0,
      "learning_rate": 8.885470403840429e-05,
      "loss": 2.0935,
      "step": 5234
    },
    {
      "epoch": 0.8132038834951456,
      "grad_norm": 0.0,
      "learning_rate": 8.871157339440289e-05,
      "loss": 2.0494,
      "step": 5235
    },
    {
      "epoch": 0.8133592233009709,
      "grad_norm": 0.0,
      "learning_rate": 8.856854690046878e-05,
      "loss": 2.0928,
      "step": 5236
    },
    {
      "epoch": 0.8135145631067962,
      "grad_norm": 0.0,
      "learning_rate": 8.842562459282061e-05,
      "loss": 2.0999,
      "step": 5237
    },
    {
      "epoch": 0.8136699029126213,
      "grad_norm": 0.0,
      "learning_rate": 8.82828065076502e-05,
      "loss": 2.09,
      "step": 5238
    },
    {
      "epoch": 0.8138252427184466,
      "grad_norm": 0.0,
      "learning_rate": 8.814009268112321e-05,
      "loss": 2.1575,
      "step": 5239
    },
    {
      "epoch": 0.8139805825242719,
      "grad_norm": 0.0,
      "learning_rate": 8.799748314937917e-05,
      "loss": 2.0638,
      "step": 5240
    },
    {
      "epoch": 0.8141359223300971,
      "grad_norm": 0.0,
      "learning_rate": 8.785497794853059e-05,
      "loss": 2.0604,
      "step": 5241
    },
    {
      "epoch": 0.8142912621359223,
      "grad_norm": 0.0,
      "learning_rate": 8.771257711466413e-05,
      "loss": 2.1046,
      "step": 5242
    },
    {
      "epoch": 0.8144466019417476,
      "grad_norm": 0.0,
      "learning_rate": 8.757028068383948e-05,
      "loss": 2.1,
      "step": 5243
    },
    {
      "epoch": 0.8146019417475728,
      "grad_norm": 0.0,
      "learning_rate": 8.742808869209046e-05,
      "loss": 2.2394,
      "step": 5244
    },
    {
      "epoch": 0.8147572815533981,
      "grad_norm": 0.0,
      "learning_rate": 8.728600117542429e-05,
      "loss": 2.2102,
      "step": 5245
    },
    {
      "epoch": 0.8149126213592233,
      "grad_norm": 0.0,
      "learning_rate": 8.71440181698212e-05,
      "loss": 2.0749,
      "step": 5246
    },
    {
      "epoch": 0.8150679611650485,
      "grad_norm": 0.0,
      "learning_rate": 8.70021397112356e-05,
      "loss": 2.1122,
      "step": 5247
    },
    {
      "epoch": 0.8152233009708738,
      "grad_norm": 0.0,
      "learning_rate": 8.686036583559532e-05,
      "loss": 2.1322,
      "step": 5248
    },
    {
      "epoch": 0.8153786407766991,
      "grad_norm": 0.0,
      "learning_rate": 8.671869657880133e-05,
      "loss": 2.1366,
      "step": 5249
    },
    {
      "epoch": 0.8155339805825242,
      "grad_norm": 0.0,
      "learning_rate": 8.657713197672856e-05,
      "loss": 2.1348,
      "step": 5250
    },
    {
      "epoch": 0.8156893203883495,
      "grad_norm": 0.0,
      "learning_rate": 8.643567206522512e-05,
      "loss": 2.0396,
      "step": 5251
    },
    {
      "epoch": 0.8158446601941748,
      "grad_norm": 0.0,
      "learning_rate": 8.629431688011291e-05,
      "loss": 2.1939,
      "step": 5252
    },
    {
      "epoch": 0.816,
      "grad_norm": 0.0,
      "learning_rate": 8.615306645718684e-05,
      "loss": 2.0991,
      "step": 5253
    },
    {
      "epoch": 0.8161553398058252,
      "grad_norm": 0.0,
      "learning_rate": 8.601192083221582e-05,
      "loss": 2.0887,
      "step": 5254
    },
    {
      "epoch": 0.8163106796116505,
      "grad_norm": 0.0,
      "learning_rate": 8.587088004094195e-05,
      "loss": 2.1334,
      "step": 5255
    },
    {
      "epoch": 0.8164660194174758,
      "grad_norm": 0.0,
      "learning_rate": 8.572994411908087e-05,
      "loss": 2.0427,
      "step": 5256
    },
    {
      "epoch": 0.8166213592233009,
      "grad_norm": 0.0,
      "learning_rate": 8.558911310232137e-05,
      "loss": 2.0799,
      "step": 5257
    },
    {
      "epoch": 0.8167766990291262,
      "grad_norm": 0.0,
      "learning_rate": 8.544838702632617e-05,
      "loss": 2.1696,
      "step": 5258
    },
    {
      "epoch": 0.8169320388349515,
      "grad_norm": 0.0,
      "learning_rate": 8.530776592673095e-05,
      "loss": 2.0431,
      "step": 5259
    },
    {
      "epoch": 0.8170873786407767,
      "grad_norm": 0.0,
      "learning_rate": 8.516724983914531e-05,
      "loss": 2.0771,
      "step": 5260
    },
    {
      "epoch": 0.8172427184466019,
      "grad_norm": 0.0,
      "learning_rate": 8.502683879915163e-05,
      "loss": 2.0465,
      "step": 5261
    },
    {
      "epoch": 0.8173980582524272,
      "grad_norm": 0.0,
      "learning_rate": 8.488653284230618e-05,
      "loss": 2.0532,
      "step": 5262
    },
    {
      "epoch": 0.8175533980582524,
      "grad_norm": 0.0,
      "learning_rate": 8.474633200413861e-05,
      "loss": 2.0463,
      "step": 5263
    },
    {
      "epoch": 0.8177087378640777,
      "grad_norm": 0.0,
      "learning_rate": 8.460623632015151e-05,
      "loss": 2.1229,
      "step": 5264
    },
    {
      "epoch": 0.8178640776699029,
      "grad_norm": 0.0,
      "learning_rate": 8.446624582582146e-05,
      "loss": 2.0898,
      "step": 5265
    },
    {
      "epoch": 0.8180194174757282,
      "grad_norm": 0.0,
      "learning_rate": 8.432636055659781e-05,
      "loss": 2.1705,
      "step": 5266
    },
    {
      "epoch": 0.8181747572815534,
      "grad_norm": 0.0,
      "learning_rate": 8.418658054790379e-05,
      "loss": 2.0975,
      "step": 5267
    },
    {
      "epoch": 0.8183300970873787,
      "grad_norm": 0.0,
      "learning_rate": 8.404690583513558e-05,
      "loss": 2.1748,
      "step": 5268
    },
    {
      "epoch": 0.8184854368932039,
      "grad_norm": 0.0,
      "learning_rate": 8.390733645366267e-05,
      "loss": 2.0897,
      "step": 5269
    },
    {
      "epoch": 0.8186407766990291,
      "grad_norm": 0.0,
      "learning_rate": 8.376787243882839e-05,
      "loss": 2.1129,
      "step": 5270
    },
    {
      "epoch": 0.8187961165048544,
      "grad_norm": 0.0,
      "learning_rate": 8.362851382594871e-05,
      "loss": 1.9855,
      "step": 5271
    },
    {
      "epoch": 0.8189514563106796,
      "grad_norm": 0.0,
      "learning_rate": 8.348926065031343e-05,
      "loss": 2.0994,
      "step": 5272
    },
    {
      "epoch": 0.8191067961165048,
      "grad_norm": 0.0,
      "learning_rate": 8.335011294718553e-05,
      "loss": 2.208,
      "step": 5273
    },
    {
      "epoch": 0.8192621359223301,
      "grad_norm": 0.0,
      "learning_rate": 8.321107075180096e-05,
      "loss": 2.0957,
      "step": 5274
    },
    {
      "epoch": 0.8194174757281554,
      "grad_norm": 0.0,
      "learning_rate": 8.307213409936953e-05,
      "loss": 2.1636,
      "step": 5275
    },
    {
      "epoch": 0.8195728155339805,
      "grad_norm": 0.0,
      "learning_rate": 8.293330302507368e-05,
      "loss": 2.0235,
      "step": 5276
    },
    {
      "epoch": 0.8197281553398058,
      "grad_norm": 0.0,
      "learning_rate": 8.27945775640696e-05,
      "loss": 2.127,
      "step": 5277
    },
    {
      "epoch": 0.8198834951456311,
      "grad_norm": 0.0,
      "learning_rate": 8.265595775148671e-05,
      "loss": 2.0771,
      "step": 5278
    },
    {
      "epoch": 0.8200388349514564,
      "grad_norm": 0.0,
      "learning_rate": 8.251744362242736e-05,
      "loss": 2.16,
      "step": 5279
    },
    {
      "epoch": 0.8201941747572815,
      "grad_norm": 0.0,
      "learning_rate": 8.237903521196716e-05,
      "loss": 2.1393,
      "step": 5280
    },
    {
      "epoch": 0.8203495145631068,
      "grad_norm": 0.0,
      "learning_rate": 8.224073255515546e-05,
      "loss": 2.1537,
      "step": 5281
    },
    {
      "epoch": 0.8205048543689321,
      "grad_norm": 0.0,
      "learning_rate": 8.210253568701415e-05,
      "loss": 2.1428,
      "step": 5282
    },
    {
      "epoch": 0.8206601941747573,
      "grad_norm": 0.0,
      "learning_rate": 8.196444464253888e-05,
      "loss": 2.1147,
      "step": 5283
    },
    {
      "epoch": 0.8208155339805825,
      "grad_norm": 0.0,
      "learning_rate": 8.182645945669808e-05,
      "loss": 2.0963,
      "step": 5284
    },
    {
      "epoch": 0.8209708737864078,
      "grad_norm": 0.0,
      "learning_rate": 8.168858016443386e-05,
      "loss": 2.1,
      "step": 5285
    },
    {
      "epoch": 0.821126213592233,
      "grad_norm": 0.0,
      "learning_rate": 8.155080680066091e-05,
      "loss": 2.145,
      "step": 5286
    },
    {
      "epoch": 0.8212815533980583,
      "grad_norm": 0.0,
      "learning_rate": 8.141313940026751e-05,
      "loss": 2.1631,
      "step": 5287
    },
    {
      "epoch": 0.8214368932038835,
      "grad_norm": 0.0,
      "learning_rate": 8.127557799811519e-05,
      "loss": 2.1301,
      "step": 5288
    },
    {
      "epoch": 0.8215922330097087,
      "grad_norm": 0.0,
      "learning_rate": 8.113812262903818e-05,
      "loss": 2.1166,
      "step": 5289
    },
    {
      "epoch": 0.821747572815534,
      "grad_norm": 0.0,
      "learning_rate": 8.100077332784439e-05,
      "loss": 2.1656,
      "step": 5290
    },
    {
      "epoch": 0.8219029126213592,
      "grad_norm": 0.0,
      "learning_rate": 8.086353012931453e-05,
      "loss": 2.148,
      "step": 5291
    },
    {
      "epoch": 0.8220582524271844,
      "grad_norm": 0.0,
      "learning_rate": 8.072639306820233e-05,
      "loss": 2.0714,
      "step": 5292
    },
    {
      "epoch": 0.8222135922330097,
      "grad_norm": 0.0,
      "learning_rate": 8.058936217923513e-05,
      "loss": 2.1974,
      "step": 5293
    },
    {
      "epoch": 0.822368932038835,
      "grad_norm": 0.0,
      "learning_rate": 8.04524374971129e-05,
      "loss": 2.0497,
      "step": 5294
    },
    {
      "epoch": 0.8225242718446601,
      "grad_norm": 0.0,
      "learning_rate": 8.031561905650891e-05,
      "loss": 2.0385,
      "step": 5295
    },
    {
      "epoch": 0.8226796116504854,
      "grad_norm": 0.0,
      "learning_rate": 8.017890689206975e-05,
      "loss": 2.0909,
      "step": 5296
    },
    {
      "epoch": 0.8228349514563107,
      "grad_norm": 0.0,
      "learning_rate": 8.004230103841459e-05,
      "loss": 2.0324,
      "step": 5297
    },
    {
      "epoch": 0.822990291262136,
      "grad_norm": 0.0,
      "learning_rate": 7.990580153013626e-05,
      "loss": 2.0663,
      "step": 5298
    },
    {
      "epoch": 0.8231456310679611,
      "grad_norm": 0.0,
      "learning_rate": 7.976940840180002e-05,
      "loss": 2.1082,
      "step": 5299
    },
    {
      "epoch": 0.8233009708737864,
      "grad_norm": 0.0,
      "learning_rate": 7.963312168794489e-05,
      "loss": 1.9524,
      "step": 5300
    },
    {
      "epoch": 0.8234563106796117,
      "grad_norm": 0.0,
      "learning_rate": 7.949694142308228e-05,
      "loss": 2.2112,
      "step": 5301
    },
    {
      "epoch": 0.8236116504854369,
      "grad_norm": 0.0,
      "learning_rate": 7.93608676416972e-05,
      "loss": 2.0798,
      "step": 5302
    },
    {
      "epoch": 0.8237669902912621,
      "grad_norm": 0.0,
      "learning_rate": 7.922490037824731e-05,
      "loss": 2.1052,
      "step": 5303
    },
    {
      "epoch": 0.8239223300970874,
      "grad_norm": 0.0,
      "learning_rate": 7.908903966716335e-05,
      "loss": 2.0933,
      "step": 5304
    },
    {
      "epoch": 0.8240776699029126,
      "grad_norm": 0.0,
      "learning_rate": 7.895328554284931e-05,
      "loss": 2.008,
      "step": 5305
    },
    {
      "epoch": 0.8242330097087379,
      "grad_norm": 0.0,
      "learning_rate": 7.881763803968206e-05,
      "loss": 2.0567,
      "step": 5306
    },
    {
      "epoch": 0.8243883495145631,
      "grad_norm": 0.0,
      "learning_rate": 7.86820971920113e-05,
      "loss": 2.1708,
      "step": 5307
    },
    {
      "epoch": 0.8245436893203884,
      "grad_norm": 0.0,
      "learning_rate": 7.854666303416013e-05,
      "loss": 2.0659,
      "step": 5308
    },
    {
      "epoch": 0.8246990291262136,
      "grad_norm": 0.0,
      "learning_rate": 7.841133560042407e-05,
      "loss": 2.1692,
      "step": 5309
    },
    {
      "epoch": 0.8248543689320388,
      "grad_norm": 0.0,
      "learning_rate": 7.827611492507203e-05,
      "loss": 2.066,
      "step": 5310
    },
    {
      "epoch": 0.8250097087378641,
      "grad_norm": 0.0,
      "learning_rate": 7.8141001042346e-05,
      "loss": 2.0956,
      "step": 5311
    },
    {
      "epoch": 0.8251650485436893,
      "grad_norm": 0.0,
      "learning_rate": 7.800599398646036e-05,
      "loss": 2.0226,
      "step": 5312
    },
    {
      "epoch": 0.8253203883495146,
      "grad_norm": 0.0,
      "learning_rate": 7.787109379160307e-05,
      "loss": 2.0222,
      "step": 5313
    },
    {
      "epoch": 0.8254757281553398,
      "grad_norm": 0.0,
      "learning_rate": 7.773630049193459e-05,
      "loss": 2.0171,
      "step": 5314
    },
    {
      "epoch": 0.825631067961165,
      "grad_norm": 0.0,
      "learning_rate": 7.76016141215884e-05,
      "loss": 2.1035,
      "step": 5315
    },
    {
      "epoch": 0.8257864077669903,
      "grad_norm": 0.0,
      "learning_rate": 7.746703471467114e-05,
      "loss": 2.0568,
      "step": 5316
    },
    {
      "epoch": 0.8259417475728156,
      "grad_norm": 0.0,
      "learning_rate": 7.733256230526203e-05,
      "loss": 2.0634,
      "step": 5317
    },
    {
      "epoch": 0.8260970873786407,
      "grad_norm": 0.0,
      "learning_rate": 7.719819692741353e-05,
      "loss": 2.0361,
      "step": 5318
    },
    {
      "epoch": 0.826252427184466,
      "grad_norm": 0.0,
      "learning_rate": 7.706393861515055e-05,
      "loss": 2.0486,
      "step": 5319
    },
    {
      "epoch": 0.8264077669902913,
      "grad_norm": 0.0,
      "learning_rate": 7.692978740247136e-05,
      "loss": 2.0049,
      "step": 5320
    },
    {
      "epoch": 0.8265631067961166,
      "grad_norm": 0.0,
      "learning_rate": 7.679574332334699e-05,
      "loss": 2.0569,
      "step": 5321
    },
    {
      "epoch": 0.8267184466019417,
      "grad_norm": 0.0,
      "learning_rate": 7.666180641172105e-05,
      "loss": 2.078,
      "step": 5322
    },
    {
      "epoch": 0.826873786407767,
      "grad_norm": 0.0,
      "learning_rate": 7.652797670151041e-05,
      "loss": 2.1104,
      "step": 5323
    },
    {
      "epoch": 0.8270291262135923,
      "grad_norm": 0.0,
      "learning_rate": 7.639425422660434e-05,
      "loss": 2.0374,
      "step": 5324
    },
    {
      "epoch": 0.8271844660194175,
      "grad_norm": 0.0,
      "learning_rate": 7.626063902086555e-05,
      "loss": 1.9829,
      "step": 5325
    },
    {
      "epoch": 0.8273398058252427,
      "grad_norm": 0.0,
      "learning_rate": 7.612713111812907e-05,
      "loss": 2.0054,
      "step": 5326
    },
    {
      "epoch": 0.827495145631068,
      "grad_norm": 0.0,
      "learning_rate": 7.599373055220288e-05,
      "loss": 2.0443,
      "step": 5327
    },
    {
      "epoch": 0.8276504854368932,
      "grad_norm": 0.0,
      "learning_rate": 7.586043735686793e-05,
      "loss": 2.1859,
      "step": 5328
    },
    {
      "epoch": 0.8278058252427184,
      "grad_norm": 0.0,
      "learning_rate": 7.572725156587801e-05,
      "loss": 2.1466,
      "step": 5329
    },
    {
      "epoch": 0.8279611650485437,
      "grad_norm": 0.0,
      "learning_rate": 7.559417321295936e-05,
      "loss": 2.1571,
      "step": 5330
    },
    {
      "epoch": 0.8281165048543689,
      "grad_norm": 0.0,
      "learning_rate": 7.546120233181148e-05,
      "loss": 2.011,
      "step": 5331
    },
    {
      "epoch": 0.8282718446601942,
      "grad_norm": 0.0,
      "learning_rate": 7.532833895610619e-05,
      "loss": 2.1332,
      "step": 5332
    },
    {
      "epoch": 0.8284271844660194,
      "grad_norm": 0.0,
      "learning_rate": 7.519558311948854e-05,
      "loss": 1.9883,
      "step": 5333
    },
    {
      "epoch": 0.8285825242718446,
      "grad_norm": 0.0,
      "learning_rate": 7.506293485557597e-05,
      "loss": 2.1906,
      "step": 5334
    },
    {
      "epoch": 0.8287378640776699,
      "grad_norm": 0.0,
      "learning_rate": 7.493039419795882e-05,
      "loss": 2.0753,
      "step": 5335
    },
    {
      "epoch": 0.8288932038834952,
      "grad_norm": 0.0,
      "learning_rate": 7.47979611802006e-05,
      "loss": 2.0433,
      "step": 5336
    },
    {
      "epoch": 0.8290485436893203,
      "grad_norm": 0.0,
      "learning_rate": 7.466563583583658e-05,
      "loss": 2.1779,
      "step": 5337
    },
    {
      "epoch": 0.8292038834951456,
      "grad_norm": 0.0,
      "learning_rate": 7.453341819837556e-05,
      "loss": 2.1353,
      "step": 5338
    },
    {
      "epoch": 0.8293592233009709,
      "grad_norm": 0.0,
      "learning_rate": 7.44013083012991e-05,
      "loss": 2.1143,
      "step": 5339
    },
    {
      "epoch": 0.8295145631067962,
      "grad_norm": 0.0,
      "learning_rate": 7.426930617806083e-05,
      "loss": 2.1468,
      "step": 5340
    },
    {
      "epoch": 0.8296699029126213,
      "grad_norm": 0.0,
      "learning_rate": 7.413741186208784e-05,
      "loss": 2.0917,
      "step": 5341
    },
    {
      "epoch": 0.8298252427184466,
      "grad_norm": 0.0,
      "learning_rate": 7.400562538677929e-05,
      "loss": 2.1294,
      "step": 5342
    },
    {
      "epoch": 0.8299805825242719,
      "grad_norm": 0.0,
      "learning_rate": 7.38739467855074e-05,
      "loss": 2.1049,
      "step": 5343
    },
    {
      "epoch": 0.8301359223300971,
      "grad_norm": 0.0,
      "learning_rate": 7.374237609161716e-05,
      "loss": 2.1367,
      "step": 5344
    },
    {
      "epoch": 0.8302912621359223,
      "grad_norm": 0.0,
      "learning_rate": 7.361091333842579e-05,
      "loss": 2.0634,
      "step": 5345
    },
    {
      "epoch": 0.8304466019417476,
      "grad_norm": 0.0,
      "learning_rate": 7.347955855922367e-05,
      "loss": 2.1343,
      "step": 5346
    },
    {
      "epoch": 0.8306019417475728,
      "grad_norm": 0.0,
      "learning_rate": 7.334831178727341e-05,
      "loss": 2.0859,
      "step": 5347
    },
    {
      "epoch": 0.830757281553398,
      "grad_norm": 0.0,
      "learning_rate": 7.321717305581066e-05,
      "loss": 2.1404,
      "step": 5348
    },
    {
      "epoch": 0.8309126213592233,
      "grad_norm": 0.0,
      "learning_rate": 7.30861423980434e-05,
      "loss": 1.9592,
      "step": 5349
    },
    {
      "epoch": 0.8310679611650486,
      "grad_norm": 0.0,
      "learning_rate": 7.29552198471523e-05,
      "loss": 2.1171,
      "step": 5350
    },
    {
      "epoch": 0.8312233009708738,
      "grad_norm": 0.0,
      "learning_rate": 7.282440543629092e-05,
      "loss": 2.0915,
      "step": 5351
    },
    {
      "epoch": 0.831378640776699,
      "grad_norm": 0.0,
      "learning_rate": 7.269369919858498e-05,
      "loss": 2.0951,
      "step": 5352
    },
    {
      "epoch": 0.8315339805825243,
      "grad_norm": 0.0,
      "learning_rate": 7.256310116713321e-05,
      "loss": 2.1282,
      "step": 5353
    },
    {
      "epoch": 0.8316893203883495,
      "grad_norm": 0.0,
      "learning_rate": 7.24326113750069e-05,
      "loss": 2.0708,
      "step": 5354
    },
    {
      "epoch": 0.8318446601941748,
      "grad_norm": 0.0,
      "learning_rate": 7.23022298552496e-05,
      "loss": 2.1191,
      "step": 5355
    },
    {
      "epoch": 0.832,
      "grad_norm": 0.0,
      "learning_rate": 7.217195664087784e-05,
      "loss": 2.0217,
      "step": 5356
    },
    {
      "epoch": 0.8321553398058252,
      "grad_norm": 0.0,
      "learning_rate": 7.204179176488041e-05,
      "loss": 2.09,
      "step": 5357
    },
    {
      "epoch": 0.8323106796116505,
      "grad_norm": 0.0,
      "learning_rate": 7.191173526021882e-05,
      "loss": 2.1293,
      "step": 5358
    },
    {
      "epoch": 0.8324660194174758,
      "grad_norm": 0.0,
      "learning_rate": 7.178178715982741e-05,
      "loss": 2.1093,
      "step": 5359
    },
    {
      "epoch": 0.8326213592233009,
      "grad_norm": 0.0,
      "learning_rate": 7.165194749661236e-05,
      "loss": 2.0467,
      "step": 5360
    },
    {
      "epoch": 0.8327766990291262,
      "grad_norm": 0.0,
      "learning_rate": 7.15222163034529e-05,
      "loss": 2.0598,
      "step": 5361
    },
    {
      "epoch": 0.8329320388349515,
      "grad_norm": 0.0,
      "learning_rate": 7.139259361320094e-05,
      "loss": 2.0562,
      "step": 5362
    },
    {
      "epoch": 0.8330873786407768,
      "grad_norm": 0.0,
      "learning_rate": 7.126307945868044e-05,
      "loss": 1.967,
      "step": 5363
    },
    {
      "epoch": 0.8332427184466019,
      "grad_norm": 0.0,
      "learning_rate": 7.113367387268827e-05,
      "loss": 1.9666,
      "step": 5364
    },
    {
      "epoch": 0.8333980582524272,
      "grad_norm": 0.0,
      "learning_rate": 7.100437688799344e-05,
      "loss": 2.0339,
      "step": 5365
    },
    {
      "epoch": 0.8335533980582525,
      "grad_norm": 0.0,
      "learning_rate": 7.087518853733799e-05,
      "loss": 2.1277,
      "step": 5366
    },
    {
      "epoch": 0.8337087378640776,
      "grad_norm": 0.0,
      "learning_rate": 7.074610885343574e-05,
      "loss": 2.1631,
      "step": 5367
    },
    {
      "epoch": 0.8338640776699029,
      "grad_norm": 0.0,
      "learning_rate": 7.061713786897361e-05,
      "loss": 2.131,
      "step": 5368
    },
    {
      "epoch": 0.8340194174757282,
      "grad_norm": 0.0,
      "learning_rate": 7.048827561661087e-05,
      "loss": 2.2117,
      "step": 5369
    },
    {
      "epoch": 0.8341747572815534,
      "grad_norm": 0.0,
      "learning_rate": 7.0359522128979e-05,
      "loss": 2.0883,
      "step": 5370
    },
    {
      "epoch": 0.8343300970873786,
      "grad_norm": 0.0,
      "learning_rate": 7.023087743868207e-05,
      "loss": 2.0272,
      "step": 5371
    },
    {
      "epoch": 0.8344854368932039,
      "grad_norm": 0.0,
      "learning_rate": 7.010234157829664e-05,
      "loss": 1.9823,
      "step": 5372
    },
    {
      "epoch": 0.8346407766990291,
      "grad_norm": 0.0,
      "learning_rate": 6.997391458037167e-05,
      "loss": 2.0636,
      "step": 5373
    },
    {
      "epoch": 0.8347961165048544,
      "grad_norm": 0.0,
      "learning_rate": 6.984559647742872e-05,
      "loss": 2.1026,
      "step": 5374
    },
    {
      "epoch": 0.8349514563106796,
      "grad_norm": 0.0,
      "learning_rate": 6.971738730196136e-05,
      "loss": 2.0736,
      "step": 5375
    },
    {
      "epoch": 0.8351067961165048,
      "grad_norm": 0.0,
      "learning_rate": 6.958928708643603e-05,
      "loss": 2.0154,
      "step": 5376
    },
    {
      "epoch": 0.8352621359223301,
      "grad_norm": 0.0,
      "learning_rate": 6.946129586329147e-05,
      "loss": 2.0282,
      "step": 5377
    },
    {
      "epoch": 0.8354174757281554,
      "grad_norm": 0.0,
      "learning_rate": 6.933341366493851e-05,
      "loss": 2.0312,
      "step": 5378
    },
    {
      "epoch": 0.8355728155339806,
      "grad_norm": 0.0,
      "learning_rate": 6.920564052376082e-05,
      "loss": 2.1036,
      "step": 5379
    },
    {
      "epoch": 0.8357281553398058,
      "grad_norm": 0.0,
      "learning_rate": 6.907797647211405e-05,
      "loss": 2.134,
      "step": 5380
    },
    {
      "epoch": 0.8358834951456311,
      "grad_norm": 0.0,
      "learning_rate": 6.89504215423265e-05,
      "loss": 2.1351,
      "step": 5381
    },
    {
      "epoch": 0.8360388349514563,
      "grad_norm": 0.0,
      "learning_rate": 6.8822975766699e-05,
      "loss": 2.0558,
      "step": 5382
    },
    {
      "epoch": 0.8361941747572815,
      "grad_norm": 0.0,
      "learning_rate": 6.8695639177504e-05,
      "loss": 2.1267,
      "step": 5383
    },
    {
      "epoch": 0.8363495145631068,
      "grad_norm": 0.0,
      "learning_rate": 6.856841180698714e-05,
      "loss": 2.1613,
      "step": 5384
    },
    {
      "epoch": 0.8365048543689321,
      "grad_norm": 0.0,
      "learning_rate": 6.844129368736579e-05,
      "loss": 2.054,
      "step": 5385
    },
    {
      "epoch": 0.8366601941747572,
      "grad_norm": 0.0,
      "learning_rate": 6.831428485083014e-05,
      "loss": 2.0953,
      "step": 5386
    },
    {
      "epoch": 0.8368155339805825,
      "grad_norm": 0.0,
      "learning_rate": 6.81873853295425e-05,
      "loss": 2.0625,
      "step": 5387
    },
    {
      "epoch": 0.8369708737864078,
      "grad_norm": 0.0,
      "learning_rate": 6.806059515563723e-05,
      "loss": 2.1935,
      "step": 5388
    },
    {
      "epoch": 0.837126213592233,
      "grad_norm": 0.0,
      "learning_rate": 6.79339143612216e-05,
      "loss": 2.1203,
      "step": 5389
    },
    {
      "epoch": 0.8372815533980582,
      "grad_norm": 0.0,
      "learning_rate": 6.780734297837448e-05,
      "loss": 2.0316,
      "step": 5390
    },
    {
      "epoch": 0.8374368932038835,
      "grad_norm": 0.0,
      "learning_rate": 6.76808810391476e-05,
      "loss": 2.281,
      "step": 5391
    },
    {
      "epoch": 0.8375922330097088,
      "grad_norm": 0.0,
      "learning_rate": 6.755452857556483e-05,
      "loss": 2.0877,
      "step": 5392
    },
    {
      "epoch": 0.837747572815534,
      "grad_norm": 0.0,
      "learning_rate": 6.742828561962212e-05,
      "loss": 2.08,
      "step": 5393
    },
    {
      "epoch": 0.8379029126213592,
      "grad_norm": 0.0,
      "learning_rate": 6.730215220328767e-05,
      "loss": 2.1136,
      "step": 5394
    },
    {
      "epoch": 0.8380582524271845,
      "grad_norm": 0.0,
      "learning_rate": 6.717612835850245e-05,
      "loss": 2.1279,
      "step": 5395
    },
    {
      "epoch": 0.8382135922330097,
      "grad_norm": 0.0,
      "learning_rate": 6.705021411717899e-05,
      "loss": 2.0694,
      "step": 5396
    },
    {
      "epoch": 0.838368932038835,
      "grad_norm": 0.0,
      "learning_rate": 6.692440951120266e-05,
      "loss": 2.0285,
      "step": 5397
    },
    {
      "epoch": 0.8385242718446602,
      "grad_norm": 0.0,
      "learning_rate": 6.679871457243058e-05,
      "loss": 2.0985,
      "step": 5398
    },
    {
      "epoch": 0.8386796116504854,
      "grad_norm": 0.0,
      "learning_rate": 6.667312933269248e-05,
      "loss": 2.1118,
      "step": 5399
    },
    {
      "epoch": 0.8388349514563107,
      "grad_norm": 0.0,
      "learning_rate": 6.654765382379019e-05,
      "loss": 2.0602,
      "step": 5400
    },
    {
      "epoch": 0.8389902912621359,
      "grad_norm": 0.0,
      "learning_rate": 6.642228807749751e-05,
      "loss": 2.0508,
      "step": 5401
    },
    {
      "epoch": 0.8391456310679611,
      "grad_norm": 0.0,
      "learning_rate": 6.629703212556092e-05,
      "loss": 2.1217,
      "step": 5402
    },
    {
      "epoch": 0.8393009708737864,
      "grad_norm": 0.0,
      "learning_rate": 6.617188599969859e-05,
      "loss": 2.0582,
      "step": 5403
    },
    {
      "epoch": 0.8394563106796117,
      "grad_norm": 0.0,
      "learning_rate": 6.604684973160135e-05,
      "loss": 2.0629,
      "step": 5404
    },
    {
      "epoch": 0.8396116504854368,
      "grad_norm": 0.0,
      "learning_rate": 6.59219233529319e-05,
      "loss": 2.0056,
      "step": 5405
    },
    {
      "epoch": 0.8397669902912621,
      "grad_norm": 0.0,
      "learning_rate": 6.579710689532504e-05,
      "loss": 2.1467,
      "step": 5406
    },
    {
      "epoch": 0.8399223300970874,
      "grad_norm": 0.0,
      "learning_rate": 6.567240039038807e-05,
      "loss": 2.1323,
      "step": 5407
    },
    {
      "epoch": 0.8400776699029127,
      "grad_norm": 0.0,
      "learning_rate": 6.554780386970016e-05,
      "loss": 2.0472,
      "step": 5408
    },
    {
      "epoch": 0.8402330097087378,
      "grad_norm": 0.0,
      "learning_rate": 6.542331736481272e-05,
      "loss": 2.0847,
      "step": 5409
    },
    {
      "epoch": 0.8403883495145631,
      "grad_norm": 0.0,
      "learning_rate": 6.529894090724953e-05,
      "loss": 2.0602,
      "step": 5410
    },
    {
      "epoch": 0.8405436893203884,
      "grad_norm": 0.0,
      "learning_rate": 6.5174674528506e-05,
      "loss": 2.1475,
      "step": 5411
    },
    {
      "epoch": 0.8406990291262136,
      "grad_norm": 0.0,
      "learning_rate": 6.50505182600502e-05,
      "loss": 2.0533,
      "step": 5412
    },
    {
      "epoch": 0.8408543689320388,
      "grad_norm": 0.0,
      "learning_rate": 6.492647213332187e-05,
      "loss": 2.189,
      "step": 5413
    },
    {
      "epoch": 0.8410097087378641,
      "grad_norm": 0.0,
      "learning_rate": 6.480253617973313e-05,
      "loss": 2.177,
      "step": 5414
    },
    {
      "epoch": 0.8411650485436893,
      "grad_norm": 0.0,
      "learning_rate": 6.467871043066831e-05,
      "loss": 2.1287,
      "step": 5415
    },
    {
      "epoch": 0.8413203883495146,
      "grad_norm": 0.0,
      "learning_rate": 6.455499491748346e-05,
      "loss": 2.0256,
      "step": 5416
    },
    {
      "epoch": 0.8414757281553398,
      "grad_norm": 0.0,
      "learning_rate": 6.443138967150697e-05,
      "loss": 2.091,
      "step": 5417
    },
    {
      "epoch": 0.841631067961165,
      "grad_norm": 0.0,
      "learning_rate": 6.430789472403909e-05,
      "loss": 2.0494,
      "step": 5418
    },
    {
      "epoch": 0.8417864077669903,
      "grad_norm": 0.0,
      "learning_rate": 6.41845101063524e-05,
      "loss": 2.1323,
      "step": 5419
    },
    {
      "epoch": 0.8419417475728155,
      "grad_norm": 0.0,
      "learning_rate": 6.406123584969159e-05,
      "loss": 2.0899,
      "step": 5420
    },
    {
      "epoch": 0.8420970873786408,
      "grad_norm": 0.0,
      "learning_rate": 6.393807198527302e-05,
      "loss": 2.0191,
      "step": 5421
    },
    {
      "epoch": 0.842252427184466,
      "grad_norm": 0.0,
      "learning_rate": 6.38150185442855e-05,
      "loss": 2.0779,
      "step": 5422
    },
    {
      "epoch": 0.8424077669902913,
      "grad_norm": 0.0,
      "learning_rate": 6.369207555788959e-05,
      "loss": 2.1006,
      "step": 5423
    },
    {
      "epoch": 0.8425631067961165,
      "grad_norm": 0.0,
      "learning_rate": 6.3569243057218e-05,
      "loss": 2.1079,
      "step": 5424
    },
    {
      "epoch": 0.8427184466019417,
      "grad_norm": 0.0,
      "learning_rate": 6.344652107337556e-05,
      "loss": 2.0691,
      "step": 5425
    },
    {
      "epoch": 0.842873786407767,
      "grad_norm": 0.0,
      "learning_rate": 6.332390963743889e-05,
      "loss": 2.137,
      "step": 5426
    },
    {
      "epoch": 0.8430291262135923,
      "grad_norm": 0.0,
      "learning_rate": 6.320140878045682e-05,
      "loss": 2.0785,
      "step": 5427
    },
    {
      "epoch": 0.8431844660194174,
      "grad_norm": 0.0,
      "learning_rate": 6.307901853345011e-05,
      "loss": 2.038,
      "step": 5428
    },
    {
      "epoch": 0.8433398058252427,
      "grad_norm": 0.0,
      "learning_rate": 6.295673892741132e-05,
      "loss": 2.0047,
      "step": 5429
    },
    {
      "epoch": 0.843495145631068,
      "grad_norm": 0.0,
      "learning_rate": 6.283456999330534e-05,
      "loss": 2.1401,
      "step": 5430
    },
    {
      "epoch": 0.8436504854368932,
      "grad_norm": 0.0,
      "learning_rate": 6.271251176206876e-05,
      "loss": 2.0726,
      "step": 5431
    },
    {
      "epoch": 0.8438058252427184,
      "grad_norm": 0.0,
      "learning_rate": 6.259056426461024e-05,
      "loss": 2.0648,
      "step": 5432
    },
    {
      "epoch": 0.8439611650485437,
      "grad_norm": 0.0,
      "learning_rate": 6.246872753181054e-05,
      "loss": 2.0181,
      "step": 5433
    },
    {
      "epoch": 0.844116504854369,
      "grad_norm": 0.0,
      "learning_rate": 6.234700159452206e-05,
      "loss": 2.1959,
      "step": 5434
    },
    {
      "epoch": 0.8442718446601942,
      "grad_norm": 0.0,
      "learning_rate": 6.22253864835694e-05,
      "loss": 1.9933,
      "step": 5435
    },
    {
      "epoch": 0.8444271844660194,
      "grad_norm": 0.0,
      "learning_rate": 6.210388222974894e-05,
      "loss": 2.0857,
      "step": 5436
    },
    {
      "epoch": 0.8445825242718447,
      "grad_norm": 0.0,
      "learning_rate": 6.198248886382918e-05,
      "loss": 2.1516,
      "step": 5437
    },
    {
      "epoch": 0.8447378640776699,
      "grad_norm": 0.0,
      "learning_rate": 6.186120641655024e-05,
      "loss": 2.1209,
      "step": 5438
    },
    {
      "epoch": 0.8448932038834951,
      "grad_norm": 0.0,
      "learning_rate": 6.174003491862451e-05,
      "loss": 2.005,
      "step": 5439
    },
    {
      "epoch": 0.8450485436893204,
      "grad_norm": 0.0,
      "learning_rate": 6.161897440073599e-05,
      "loss": 2.2058,
      "step": 5440
    },
    {
      "epoch": 0.8452038834951456,
      "grad_norm": 0.0,
      "learning_rate": 6.149802489354061e-05,
      "loss": 2.0735,
      "step": 5441
    },
    {
      "epoch": 0.8453592233009709,
      "grad_norm": 0.0,
      "learning_rate": 6.137718642766637e-05,
      "loss": 2.0297,
      "step": 5442
    },
    {
      "epoch": 0.8455145631067961,
      "grad_norm": 0.0,
      "learning_rate": 6.125645903371312e-05,
      "loss": 2.0746,
      "step": 5443
    },
    {
      "epoch": 0.8456699029126213,
      "grad_norm": 0.0,
      "learning_rate": 6.113584274225237e-05,
      "loss": 2.0453,
      "step": 5444
    },
    {
      "epoch": 0.8458252427184466,
      "grad_norm": 0.0,
      "learning_rate": 6.1015337583827765e-05,
      "loss": 2.0219,
      "step": 5445
    },
    {
      "epoch": 0.8459805825242719,
      "grad_norm": 0.0,
      "learning_rate": 6.089494358895448e-05,
      "loss": 2.0687,
      "step": 5446
    },
    {
      "epoch": 0.846135922330097,
      "grad_norm": 0.0,
      "learning_rate": 6.077466078811994e-05,
      "loss": 2.1557,
      "step": 5447
    },
    {
      "epoch": 0.8462912621359223,
      "grad_norm": 0.0,
      "learning_rate": 6.0654489211783184e-05,
      "loss": 2.1094,
      "step": 5448
    },
    {
      "epoch": 0.8464466019417476,
      "grad_norm": 0.0,
      "learning_rate": 6.053442889037497e-05,
      "loss": 2.0684,
      "step": 5449
    },
    {
      "epoch": 0.8466019417475729,
      "grad_norm": 0.0,
      "learning_rate": 6.0414479854298374e-05,
      "loss": 2.0519,
      "step": 5450
    },
    {
      "epoch": 0.846757281553398,
      "grad_norm": 0.0,
      "learning_rate": 6.029464213392749e-05,
      "loss": 2.0737,
      "step": 5451
    },
    {
      "epoch": 0.8469126213592233,
      "grad_norm": 0.0,
      "learning_rate": 6.017491575960882e-05,
      "loss": 2.0016,
      "step": 5452
    },
    {
      "epoch": 0.8470679611650486,
      "grad_norm": 0.0,
      "learning_rate": 6.005530076166077e-05,
      "loss": 2.1053,
      "step": 5453
    },
    {
      "epoch": 0.8472233009708738,
      "grad_norm": 0.0,
      "learning_rate": 5.993579717037301e-05,
      "loss": 2.0066,
      "step": 5454
    },
    {
      "epoch": 0.847378640776699,
      "grad_norm": 0.0,
      "learning_rate": 5.981640501600749e-05,
      "loss": 1.9855,
      "step": 5455
    },
    {
      "epoch": 0.8475339805825243,
      "grad_norm": 0.0,
      "learning_rate": 5.9697124328797536e-05,
      "loss": 1.9987,
      "step": 5456
    },
    {
      "epoch": 0.8476893203883495,
      "grad_norm": 0.0,
      "learning_rate": 5.9577955138948594e-05,
      "loss": 2.0765,
      "step": 5457
    },
    {
      "epoch": 0.8478446601941747,
      "grad_norm": 0.0,
      "learning_rate": 5.94588974766378e-05,
      "loss": 2.0068,
      "step": 5458
    },
    {
      "epoch": 0.848,
      "grad_norm": 0.0,
      "learning_rate": 5.933995137201381e-05,
      "loss": 2.0861,
      "step": 5459
    },
    {
      "epoch": 0.8481553398058252,
      "grad_norm": 0.0,
      "learning_rate": 5.9221116855197374e-05,
      "loss": 2.1561,
      "step": 5460
    },
    {
      "epoch": 0.8483106796116505,
      "grad_norm": 0.0,
      "learning_rate": 5.9102393956280586e-05,
      "loss": 2.1389,
      "step": 5461
    },
    {
      "epoch": 0.8484660194174757,
      "grad_norm": 0.0,
      "learning_rate": 5.898378270532772e-05,
      "loss": 2.147,
      "step": 5462
    },
    {
      "epoch": 0.848621359223301,
      "grad_norm": 0.0,
      "learning_rate": 5.886528313237455e-05,
      "loss": 2.0864,
      "step": 5463
    },
    {
      "epoch": 0.8487766990291262,
      "grad_norm": 0.0,
      "learning_rate": 5.874689526742832e-05,
      "loss": 2.0937,
      "step": 5464
    },
    {
      "epoch": 0.8489320388349515,
      "grad_norm": 0.0,
      "learning_rate": 5.862861914046846e-05,
      "loss": 2.0494,
      "step": 5465
    },
    {
      "epoch": 0.8490873786407767,
      "grad_norm": 0.0,
      "learning_rate": 5.8510454781445955e-05,
      "loss": 2.1402,
      "step": 5466
    },
    {
      "epoch": 0.8492427184466019,
      "grad_norm": 0.0,
      "learning_rate": 5.839240222028325e-05,
      "loss": 2.1837,
      "step": 5467
    },
    {
      "epoch": 0.8493980582524272,
      "grad_norm": 0.0,
      "learning_rate": 5.827446148687476e-05,
      "loss": 2.0857,
      "step": 5468
    },
    {
      "epoch": 0.8495533980582525,
      "grad_norm": 0.0,
      "learning_rate": 5.815663261108639e-05,
      "loss": 2.1617,
      "step": 5469
    },
    {
      "epoch": 0.8497087378640776,
      "grad_norm": 0.0,
      "learning_rate": 5.803891562275593e-05,
      "loss": 2.1934,
      "step": 5470
    },
    {
      "epoch": 0.8498640776699029,
      "grad_norm": 0.0,
      "learning_rate": 5.792131055169247e-05,
      "loss": 2.0605,
      "step": 5471
    },
    {
      "epoch": 0.8500194174757282,
      "grad_norm": 0.0,
      "learning_rate": 5.7803817427677266e-05,
      "loss": 2.0897,
      "step": 5472
    },
    {
      "epoch": 0.8501747572815535,
      "grad_norm": 0.0,
      "learning_rate": 5.768643628046299e-05,
      "loss": 2.1006,
      "step": 5473
    },
    {
      "epoch": 0.8503300970873786,
      "grad_norm": 0.0,
      "learning_rate": 5.756916713977356e-05,
      "loss": 2.09,
      "step": 5474
    },
    {
      "epoch": 0.8504854368932039,
      "grad_norm": 0.0,
      "learning_rate": 5.7452010035305145e-05,
      "loss": 2.0823,
      "step": 5475
    },
    {
      "epoch": 0.8506407766990292,
      "grad_norm": 0.0,
      "learning_rate": 5.7334964996725415e-05,
      "loss": 2.0358,
      "step": 5476
    },
    {
      "epoch": 0.8507961165048543,
      "grad_norm": 0.0,
      "learning_rate": 5.721803205367327e-05,
      "loss": 2.1456,
      "step": 5477
    },
    {
      "epoch": 0.8509514563106796,
      "grad_norm": 0.0,
      "learning_rate": 5.7101211235759733e-05,
      "loss": 2.0599,
      "step": 5478
    },
    {
      "epoch": 0.8511067961165049,
      "grad_norm": 0.0,
      "learning_rate": 5.6984502572566986e-05,
      "loss": 2.1148,
      "step": 5479
    },
    {
      "epoch": 0.8512621359223301,
      "grad_norm": 0.0,
      "learning_rate": 5.6867906093649116e-05,
      "loss": 2.1312,
      "step": 5480
    },
    {
      "epoch": 0.8514174757281553,
      "grad_norm": 0.0,
      "learning_rate": 5.675142182853188e-05,
      "loss": 2.0833,
      "step": 5481
    },
    {
      "epoch": 0.8515728155339806,
      "grad_norm": 0.0,
      "learning_rate": 5.6635049806712136e-05,
      "loss": 2.1757,
      "step": 5482
    },
    {
      "epoch": 0.8517281553398058,
      "grad_norm": 0.0,
      "learning_rate": 5.6518790057658865e-05,
      "loss": 2.1321,
      "step": 5483
    },
    {
      "epoch": 0.8518834951456311,
      "grad_norm": 0.0,
      "learning_rate": 5.640264261081224e-05,
      "loss": 2.1659,
      "step": 5484
    },
    {
      "epoch": 0.8520388349514563,
      "grad_norm": 0.0,
      "learning_rate": 5.628660749558428e-05,
      "loss": 2.0128,
      "step": 5485
    },
    {
      "epoch": 0.8521941747572815,
      "grad_norm": 0.0,
      "learning_rate": 5.6170684741358333e-05,
      "loss": 2.0216,
      "step": 5486
    },
    {
      "epoch": 0.8523495145631068,
      "grad_norm": 0.0,
      "learning_rate": 5.605487437748929e-05,
      "loss": 2.0622,
      "step": 5487
    },
    {
      "epoch": 0.8525048543689321,
      "grad_norm": 0.0,
      "learning_rate": 5.593917643330382e-05,
      "loss": 2.1297,
      "step": 5488
    },
    {
      "epoch": 0.8526601941747572,
      "grad_norm": 0.0,
      "learning_rate": 5.5823590938099856e-05,
      "loss": 2.031,
      "step": 5489
    },
    {
      "epoch": 0.8528155339805825,
      "grad_norm": 0.0,
      "learning_rate": 5.5708117921147e-05,
      "loss": 2.0787,
      "step": 5490
    },
    {
      "epoch": 0.8529708737864078,
      "grad_norm": 0.0,
      "learning_rate": 5.5592757411686456e-05,
      "loss": 2.0929,
      "step": 5491
    },
    {
      "epoch": 0.8531262135922331,
      "grad_norm": 0.0,
      "learning_rate": 5.547750943893059e-05,
      "loss": 1.9759,
      "step": 5492
    },
    {
      "epoch": 0.8532815533980582,
      "grad_norm": 0.0,
      "learning_rate": 5.536237403206384e-05,
      "loss": 2.0203,
      "step": 5493
    },
    {
      "epoch": 0.8534368932038835,
      "grad_norm": 0.0,
      "learning_rate": 5.5247351220241436e-05,
      "loss": 2.1418,
      "step": 5494
    },
    {
      "epoch": 0.8535922330097088,
      "grad_norm": 0.0,
      "learning_rate": 5.513244103259068e-05,
      "loss": 2.1277,
      "step": 5495
    },
    {
      "epoch": 0.8537475728155339,
      "grad_norm": 0.0,
      "learning_rate": 5.501764349821031e-05,
      "loss": 2.1409,
      "step": 5496
    },
    {
      "epoch": 0.8539029126213592,
      "grad_norm": 0.0,
      "learning_rate": 5.490295864616995e-05,
      "loss": 2.0973,
      "step": 5497
    },
    {
      "epoch": 0.8540582524271845,
      "grad_norm": 0.0,
      "learning_rate": 5.4788386505511323e-05,
      "loss": 2.097,
      "step": 5498
    },
    {
      "epoch": 0.8542135922330097,
      "grad_norm": 0.0,
      "learning_rate": 5.467392710524749e-05,
      "loss": 2.1055,
      "step": 5499
    },
    {
      "epoch": 0.8543689320388349,
      "grad_norm": 0.0,
      "learning_rate": 5.455958047436266e-05,
      "loss": 2.1534,
      "step": 5500
    },
    {
      "epoch": 0.8545242718446602,
      "grad_norm": 0.0,
      "learning_rate": 5.4445346641812834e-05,
      "loss": 1.9637,
      "step": 5501
    },
    {
      "epoch": 0.8546796116504854,
      "grad_norm": 0.0,
      "learning_rate": 5.43312256365252e-05,
      "loss": 2.0381,
      "step": 5502
    },
    {
      "epoch": 0.8548349514563107,
      "grad_norm": 0.0,
      "learning_rate": 5.421721748739866e-05,
      "loss": 1.9766,
      "step": 5503
    },
    {
      "epoch": 0.8549902912621359,
      "grad_norm": 0.0,
      "learning_rate": 5.410332222330316e-05,
      "loss": 2.1084,
      "step": 5504
    },
    {
      "epoch": 0.8551456310679612,
      "grad_norm": 0.0,
      "learning_rate": 5.3989539873080303e-05,
      "loss": 2.0128,
      "step": 5505
    },
    {
      "epoch": 0.8553009708737864,
      "grad_norm": 0.0,
      "learning_rate": 5.387587046554321e-05,
      "loss": 2.0578,
      "step": 5506
    },
    {
      "epoch": 0.8554563106796117,
      "grad_norm": 0.0,
      "learning_rate": 5.376231402947612e-05,
      "loss": 2.1097,
      "step": 5507
    },
    {
      "epoch": 0.8556116504854369,
      "grad_norm": 0.0,
      "learning_rate": 5.364887059363471e-05,
      "loss": 1.9847,
      "step": 5508
    },
    {
      "epoch": 0.8557669902912621,
      "grad_norm": 0.0,
      "learning_rate": 5.353554018674628e-05,
      "loss": 2.0037,
      "step": 5509
    },
    {
      "epoch": 0.8559223300970874,
      "grad_norm": 0.0,
      "learning_rate": 5.34223228375092e-05,
      "loss": 2.0239,
      "step": 5510
    },
    {
      "epoch": 0.8560776699029127,
      "grad_norm": 0.0,
      "learning_rate": 5.3309218574593586e-05,
      "loss": 2.0729,
      "step": 5511
    },
    {
      "epoch": 0.8562330097087378,
      "grad_norm": 0.0,
      "learning_rate": 5.319622742664043e-05,
      "loss": 1.9944,
      "step": 5512
    },
    {
      "epoch": 0.8563883495145631,
      "grad_norm": 0.0,
      "learning_rate": 5.308334942226245e-05,
      "loss": 2.0291,
      "step": 5513
    },
    {
      "epoch": 0.8565436893203884,
      "grad_norm": 0.0,
      "learning_rate": 5.297058459004373e-05,
      "loss": 1.9651,
      "step": 5514
    },
    {
      "epoch": 0.8566990291262135,
      "grad_norm": 0.0,
      "learning_rate": 5.285793295853941e-05,
      "loss": 2.0695,
      "step": 5515
    },
    {
      "epoch": 0.8568543689320388,
      "grad_norm": 0.0,
      "learning_rate": 5.274539455627619e-05,
      "loss": 2.0084,
      "step": 5516
    },
    {
      "epoch": 0.8570097087378641,
      "grad_norm": 0.0,
      "learning_rate": 5.2632969411752e-05,
      "loss": 2.1299,
      "step": 5517
    },
    {
      "epoch": 0.8571650485436894,
      "grad_norm": 0.0,
      "learning_rate": 5.252065755343621e-05,
      "loss": 2.0377,
      "step": 5518
    },
    {
      "epoch": 0.8573203883495145,
      "grad_norm": 0.0,
      "learning_rate": 5.240845900976937e-05,
      "loss": 2.0945,
      "step": 5519
    },
    {
      "epoch": 0.8574757281553398,
      "grad_norm": 0.0,
      "learning_rate": 5.2296373809163245e-05,
      "loss": 2.0201,
      "step": 5520
    },
    {
      "epoch": 0.8576310679611651,
      "grad_norm": 0.0,
      "learning_rate": 5.218440198000124e-05,
      "loss": 2.1523,
      "step": 5521
    },
    {
      "epoch": 0.8577864077669903,
      "grad_norm": 0.0,
      "learning_rate": 5.207254355063762e-05,
      "loss": 2.0615,
      "step": 5522
    },
    {
      "epoch": 0.8579417475728155,
      "grad_norm": 0.0,
      "learning_rate": 5.1960798549398336e-05,
      "loss": 2.089,
      "step": 5523
    },
    {
      "epoch": 0.8580970873786408,
      "grad_norm": 0.0,
      "learning_rate": 5.184916700458048e-05,
      "loss": 2.082,
      "step": 5524
    },
    {
      "epoch": 0.858252427184466,
      "grad_norm": 0.0,
      "learning_rate": 5.173764894445221e-05,
      "loss": 2.0368,
      "step": 5525
    },
    {
      "epoch": 0.8584077669902913,
      "grad_norm": 0.0,
      "learning_rate": 5.162624439725327e-05,
      "loss": 2.04,
      "step": 5526
    },
    {
      "epoch": 0.8585631067961165,
      "grad_norm": 0.0,
      "learning_rate": 5.151495339119433e-05,
      "loss": 2.1656,
      "step": 5527
    },
    {
      "epoch": 0.8587184466019417,
      "grad_norm": 0.0,
      "learning_rate": 5.1403775954457497e-05,
      "loss": 2.1355,
      "step": 5528
    },
    {
      "epoch": 0.858873786407767,
      "grad_norm": 0.0,
      "learning_rate": 5.1292712115196315e-05,
      "loss": 2.1998,
      "step": 5529
    },
    {
      "epoch": 0.8590291262135923,
      "grad_norm": 0.0,
      "learning_rate": 5.118176190153517e-05,
      "loss": 2.0703,
      "step": 5530
    },
    {
      "epoch": 0.8591844660194174,
      "grad_norm": 0.0,
      "learning_rate": 5.107092534156982e-05,
      "loss": 2.1308,
      "step": 5531
    },
    {
      "epoch": 0.8593398058252427,
      "grad_norm": 0.0,
      "learning_rate": 5.0960202463367354e-05,
      "loss": 2.0797,
      "step": 5532
    },
    {
      "epoch": 0.859495145631068,
      "grad_norm": 0.0,
      "learning_rate": 5.084959329496586e-05,
      "loss": 2.0391,
      "step": 5533
    },
    {
      "epoch": 0.8596504854368932,
      "grad_norm": 0.0,
      "learning_rate": 5.073909786437503e-05,
      "loss": 2.0558,
      "step": 5534
    },
    {
      "epoch": 0.8598058252427184,
      "grad_norm": 0.0,
      "learning_rate": 5.0628716199575154e-05,
      "loss": 2.074,
      "step": 5535
    },
    {
      "epoch": 0.8599611650485437,
      "grad_norm": 0.0,
      "learning_rate": 5.051844832851832e-05,
      "loss": 2.137,
      "step": 5536
    },
    {
      "epoch": 0.860116504854369,
      "grad_norm": 0.0,
      "learning_rate": 5.0408294279127355e-05,
      "loss": 2.0813,
      "step": 5537
    },
    {
      "epoch": 0.8602718446601941,
      "grad_norm": 0.0,
      "learning_rate": 5.0298254079296446e-05,
      "loss": 2.0732,
      "step": 5538
    },
    {
      "epoch": 0.8604271844660194,
      "grad_norm": 0.0,
      "learning_rate": 5.018832775689114e-05,
      "loss": 2.0315,
      "step": 5539
    },
    {
      "epoch": 0.8605825242718447,
      "grad_norm": 0.0,
      "learning_rate": 5.0078515339747666e-05,
      "loss": 2.231,
      "step": 5540
    },
    {
      "epoch": 0.86073786407767,
      "grad_norm": 0.0,
      "learning_rate": 4.9968816855673895e-05,
      "loss": 2.0656,
      "step": 5541
    },
    {
      "epoch": 0.8608932038834951,
      "grad_norm": 0.0,
      "learning_rate": 4.98592323324486e-05,
      "loss": 2.1602,
      "step": 5542
    },
    {
      "epoch": 0.8610485436893204,
      "grad_norm": 0.0,
      "learning_rate": 4.974976179782159e-05,
      "loss": 2.2493,
      "step": 5543
    },
    {
      "epoch": 0.8612038834951457,
      "grad_norm": 0.0,
      "learning_rate": 4.964040527951419e-05,
      "loss": 2.053,
      "step": 5544
    },
    {
      "epoch": 0.8613592233009709,
      "grad_norm": 0.0,
      "learning_rate": 4.953116280521835e-05,
      "loss": 2.102,
      "step": 5545
    },
    {
      "epoch": 0.8615145631067961,
      "grad_norm": 0.0,
      "learning_rate": 4.942203440259763e-05,
      "loss": 2.0675,
      "step": 5546
    },
    {
      "epoch": 0.8616699029126214,
      "grad_norm": 0.0,
      "learning_rate": 4.931302009928645e-05,
      "loss": 2.0793,
      "step": 5547
    },
    {
      "epoch": 0.8618252427184466,
      "grad_norm": 0.0,
      "learning_rate": 4.92041199228902e-05,
      "loss": 2.1588,
      "step": 5548
    },
    {
      "epoch": 0.8619805825242719,
      "grad_norm": 0.0,
      "learning_rate": 4.90953339009858e-05,
      "loss": 2.087,
      "step": 5549
    },
    {
      "epoch": 0.8621359223300971,
      "grad_norm": 0.0,
      "learning_rate": 4.898666206112079e-05,
      "loss": 2.0776,
      "step": 5550
    },
    {
      "epoch": 0.8622912621359223,
      "grad_norm": 0.0,
      "learning_rate": 4.887810443081414e-05,
      "loss": 1.9362,
      "step": 5551
    },
    {
      "epoch": 0.8624466019417476,
      "grad_norm": 0.0,
      "learning_rate": 4.876966103755565e-05,
      "loss": 2.1003,
      "step": 5552
    },
    {
      "epoch": 0.8626019417475728,
      "grad_norm": 0.0,
      "learning_rate": 4.8661331908806395e-05,
      "loss": 2.1085,
      "step": 5553
    },
    {
      "epoch": 0.862757281553398,
      "grad_norm": 0.0,
      "learning_rate": 4.8553117071998445e-05,
      "loss": 1.9346,
      "step": 5554
    },
    {
      "epoch": 0.8629126213592233,
      "grad_norm": 0.0,
      "learning_rate": 4.844501655453476e-05,
      "loss": 2.074,
      "step": 5555
    },
    {
      "epoch": 0.8630679611650486,
      "grad_norm": 0.0,
      "learning_rate": 4.833703038378967e-05,
      "loss": 2.1438,
      "step": 5556
    },
    {
      "epoch": 0.8632233009708737,
      "grad_norm": 0.0,
      "learning_rate": 4.822915858710836e-05,
      "loss": 2.2053,
      "step": 5557
    },
    {
      "epoch": 0.863378640776699,
      "grad_norm": 0.0,
      "learning_rate": 4.812140119180691e-05,
      "loss": 2.0616,
      "step": 5558
    },
    {
      "epoch": 0.8635339805825243,
      "grad_norm": 0.0,
      "learning_rate": 4.80137582251729e-05,
      "loss": 2.0013,
      "step": 5559
    },
    {
      "epoch": 0.8636893203883496,
      "grad_norm": 0.0,
      "learning_rate": 4.7906229714464286e-05,
      "loss": 2.2273,
      "step": 5560
    },
    {
      "epoch": 0.8638446601941747,
      "grad_norm": 0.0,
      "learning_rate": 4.779881568691058e-05,
      "loss": 2.1979,
      "step": 5561
    },
    {
      "epoch": 0.864,
      "grad_norm": 0.0,
      "learning_rate": 4.7691516169712124e-05,
      "loss": 2.0362,
      "step": 5562
    },
    {
      "epoch": 0.8641553398058253,
      "grad_norm": 0.0,
      "learning_rate": 4.758433119004013e-05,
      "loss": 2.0651,
      "step": 5563
    },
    {
      "epoch": 0.8643106796116505,
      "grad_norm": 0.0,
      "learning_rate": 4.7477260775037055e-05,
      "loss": 2.1241,
      "step": 5564
    },
    {
      "epoch": 0.8644660194174757,
      "grad_norm": 0.0,
      "learning_rate": 4.7370304951816154e-05,
      "loss": 2.1052,
      "step": 5565
    },
    {
      "epoch": 0.864621359223301,
      "grad_norm": 0.0,
      "learning_rate": 4.7263463747461545e-05,
      "loss": 2.1707,
      "step": 5566
    },
    {
      "epoch": 0.8647766990291262,
      "grad_norm": 0.0,
      "learning_rate": 4.7156737189028795e-05,
      "loss": 2.1068,
      "step": 5567
    },
    {
      "epoch": 0.8649320388349515,
      "grad_norm": 0.0,
      "learning_rate": 4.705012530354391e-05,
      "loss": 1.9802,
      "step": 5568
    },
    {
      "epoch": 0.8650873786407767,
      "grad_norm": 0.0,
      "learning_rate": 4.694362811800423e-05,
      "loss": 2.1564,
      "step": 5569
    },
    {
      "epoch": 0.8652427184466019,
      "grad_norm": 0.0,
      "learning_rate": 4.68372456593778e-05,
      "loss": 2.0837,
      "step": 5570
    },
    {
      "epoch": 0.8653980582524272,
      "grad_norm": 0.0,
      "learning_rate": 4.673097795460374e-05,
      "loss": 2.0824,
      "step": 5571
    },
    {
      "epoch": 0.8655533980582524,
      "grad_norm": 0.0,
      "learning_rate": 4.662482503059223e-05,
      "loss": 2.0355,
      "step": 5572
    },
    {
      "epoch": 0.8657087378640776,
      "grad_norm": 0.0,
      "learning_rate": 4.6518786914224006e-05,
      "loss": 2.0872,
      "step": 5573
    },
    {
      "epoch": 0.8658640776699029,
      "grad_norm": 0.0,
      "learning_rate": 4.641286363235126e-05,
      "loss": 2.0818,
      "step": 5574
    },
    {
      "epoch": 0.8660194174757282,
      "grad_norm": 0.0,
      "learning_rate": 4.630705521179651e-05,
      "loss": 2.1745,
      "step": 5575
    },
    {
      "epoch": 0.8661747572815534,
      "grad_norm": 0.0,
      "learning_rate": 4.620136167935374e-05,
      "loss": 2.066,
      "step": 5576
    },
    {
      "epoch": 0.8663300970873786,
      "grad_norm": 0.0,
      "learning_rate": 4.609578306178752e-05,
      "loss": 2.051,
      "step": 5577
    },
    {
      "epoch": 0.8664854368932039,
      "grad_norm": 0.0,
      "learning_rate": 4.599031938583331e-05,
      "loss": 2.1037,
      "step": 5578
    },
    {
      "epoch": 0.8666407766990292,
      "grad_norm": 0.0,
      "learning_rate": 4.588497067819758e-05,
      "loss": 2.1118,
      "step": 5579
    },
    {
      "epoch": 0.8667961165048543,
      "grad_norm": 0.0,
      "learning_rate": 4.577973696555776e-05,
      "loss": 2.0144,
      "step": 5580
    },
    {
      "epoch": 0.8669514563106796,
      "grad_norm": 0.0,
      "learning_rate": 4.5674618274561875e-05,
      "loss": 2.1616,
      "step": 5581
    },
    {
      "epoch": 0.8671067961165049,
      "grad_norm": 0.0,
      "learning_rate": 4.5569614631829234e-05,
      "loss": 2.1108,
      "step": 5582
    },
    {
      "epoch": 0.8672621359223301,
      "grad_norm": 0.0,
      "learning_rate": 4.546472606394953e-05,
      "loss": 2.1045,
      "step": 5583
    },
    {
      "epoch": 0.8674174757281553,
      "grad_norm": 0.0,
      "learning_rate": 4.535995259748382e-05,
      "loss": 2.151,
      "step": 5584
    },
    {
      "epoch": 0.8675728155339806,
      "grad_norm": 0.0,
      "learning_rate": 4.525529425896352e-05,
      "loss": 2.115,
      "step": 5585
    },
    {
      "epoch": 0.8677281553398059,
      "grad_norm": 0.0,
      "learning_rate": 4.515075107489125e-05,
      "loss": 2.1484,
      "step": 5586
    },
    {
      "epoch": 0.8678834951456311,
      "grad_norm": 0.0,
      "learning_rate": 4.504632307174056e-05,
      "loss": 2.115,
      "step": 5587
    },
    {
      "epoch": 0.8680388349514563,
      "grad_norm": 0.0,
      "learning_rate": 4.4942010275955235e-05,
      "loss": 2.0359,
      "step": 5588
    },
    {
      "epoch": 0.8681941747572816,
      "grad_norm": 0.0,
      "learning_rate": 4.483781271395043e-05,
      "loss": 1.998,
      "step": 5589
    },
    {
      "epoch": 0.8683495145631068,
      "grad_norm": 0.0,
      "learning_rate": 4.4733730412112094e-05,
      "loss": 1.9995,
      "step": 5590
    },
    {
      "epoch": 0.868504854368932,
      "grad_norm": 0.0,
      "learning_rate": 4.462976339679664e-05,
      "loss": 2.1229,
      "step": 5591
    },
    {
      "epoch": 0.8686601941747573,
      "grad_norm": 0.0,
      "learning_rate": 4.452591169433179e-05,
      "loss": 2.0582,
      "step": 5592
    },
    {
      "epoch": 0.8688155339805825,
      "grad_norm": 0.0,
      "learning_rate": 4.44221753310155e-05,
      "loss": 2.0436,
      "step": 5593
    },
    {
      "epoch": 0.8689708737864078,
      "grad_norm": 0.0,
      "learning_rate": 4.431855433311694e-05,
      "loss": 2.0408,
      "step": 5594
    },
    {
      "epoch": 0.869126213592233,
      "grad_norm": 0.0,
      "learning_rate": 4.4215048726875985e-05,
      "loss": 2.0614,
      "step": 5595
    },
    {
      "epoch": 0.8692815533980582,
      "grad_norm": 0.0,
      "learning_rate": 4.411165853850307e-05,
      "loss": 1.9436,
      "step": 5596
    },
    {
      "epoch": 0.8694368932038835,
      "grad_norm": 0.0,
      "learning_rate": 4.4008383794179816e-05,
      "loss": 2.064,
      "step": 5597
    },
    {
      "epoch": 0.8695922330097088,
      "grad_norm": 0.0,
      "learning_rate": 4.390522452005807e-05,
      "loss": 2.0685,
      "step": 5598
    },
    {
      "epoch": 0.8697475728155339,
      "grad_norm": 0.0,
      "learning_rate": 4.3802180742260946e-05,
      "loss": 2.1012,
      "step": 5599
    },
    {
      "epoch": 0.8699029126213592,
      "grad_norm": 0.0,
      "learning_rate": 4.369925248688206e-05,
      "loss": 2.171,
      "step": 5600
    },
    {
      "epoch": 0.8700582524271845,
      "grad_norm": 0.0,
      "learning_rate": 4.359643977998562e-05,
      "loss": 2.0403,
      "step": 5601
    },
    {
      "epoch": 0.8702135922330098,
      "grad_norm": 0.0,
      "learning_rate": 4.3493742647607024e-05,
      "loss": 2.1335,
      "step": 5602
    },
    {
      "epoch": 0.8703689320388349,
      "grad_norm": 0.0,
      "learning_rate": 4.339116111575198e-05,
      "loss": 2.0469,
      "step": 5603
    },
    {
      "epoch": 0.8705242718446602,
      "grad_norm": 0.0,
      "learning_rate": 4.328869521039708e-05,
      "loss": 2.1509,
      "step": 5604
    },
    {
      "epoch": 0.8706796116504855,
      "grad_norm": 0.0,
      "learning_rate": 4.318634495748974e-05,
      "loss": 1.918,
      "step": 5605
    },
    {
      "epoch": 0.8708349514563106,
      "grad_norm": 0.0,
      "learning_rate": 4.30841103829479e-05,
      "loss": 2.0718,
      "step": 5606
    },
    {
      "epoch": 0.8709902912621359,
      "grad_norm": 0.0,
      "learning_rate": 4.2981991512660415e-05,
      "loss": 2.0567,
      "step": 5607
    },
    {
      "epoch": 0.8711456310679612,
      "grad_norm": 0.0,
      "learning_rate": 4.287998837248658e-05,
      "loss": 2.1559,
      "step": 5608
    },
    {
      "epoch": 0.8713009708737864,
      "grad_norm": 0.0,
      "learning_rate": 4.277810098825657e-05,
      "loss": 2.1987,
      "step": 5609
    },
    {
      "epoch": 0.8714563106796116,
      "grad_norm": 0.0,
      "learning_rate": 4.267632938577143e-05,
      "loss": 2.1092,
      "step": 5610
    },
    {
      "epoch": 0.8716116504854369,
      "grad_norm": 0.0,
      "learning_rate": 4.257467359080219e-05,
      "loss": 2.1039,
      "step": 5611
    },
    {
      "epoch": 0.8717669902912621,
      "grad_norm": 0.0,
      "learning_rate": 4.2473133629091385e-05,
      "loss": 2.0451,
      "step": 5612
    },
    {
      "epoch": 0.8719223300970874,
      "grad_norm": 0.0,
      "learning_rate": 4.237170952635177e-05,
      "loss": 2.1516,
      "step": 5613
    },
    {
      "epoch": 0.8720776699029126,
      "grad_norm": 0.0,
      "learning_rate": 4.227040130826676e-05,
      "loss": 2.132,
      "step": 5614
    },
    {
      "epoch": 0.8722330097087378,
      "grad_norm": 0.0,
      "learning_rate": 4.2169209000490704e-05,
      "loss": 2.0836,
      "step": 5615
    },
    {
      "epoch": 0.8723883495145631,
      "grad_norm": 0.0,
      "learning_rate": 4.206813262864817e-05,
      "loss": 2.022,
      "step": 5616
    },
    {
      "epoch": 0.8725436893203884,
      "grad_norm": 0.0,
      "learning_rate": 4.196717221833485e-05,
      "loss": 1.9799,
      "step": 5617
    },
    {
      "epoch": 0.8726990291262136,
      "grad_norm": 0.0,
      "learning_rate": 4.1866327795116566e-05,
      "loss": 1.9673,
      "step": 5618
    },
    {
      "epoch": 0.8728543689320388,
      "grad_norm": 0.0,
      "learning_rate": 4.176559938453017e-05,
      "loss": 1.9896,
      "step": 5619
    },
    {
      "epoch": 0.8730097087378641,
      "grad_norm": 0.0,
      "learning_rate": 4.16649870120831e-05,
      "loss": 2.0941,
      "step": 5620
    },
    {
      "epoch": 0.8731650485436894,
      "grad_norm": 0.0,
      "learning_rate": 4.1564490703253134e-05,
      "loss": 2.1461,
      "step": 5621
    },
    {
      "epoch": 0.8733203883495145,
      "grad_norm": 0.0,
      "learning_rate": 4.146411048348886e-05,
      "loss": 1.9722,
      "step": 5622
    },
    {
      "epoch": 0.8734757281553398,
      "grad_norm": 0.0,
      "learning_rate": 4.13638463782095e-05,
      "loss": 2.0902,
      "step": 5623
    },
    {
      "epoch": 0.8736310679611651,
      "grad_norm": 0.0,
      "learning_rate": 4.1263698412804696e-05,
      "loss": 1.9649,
      "step": 5624
    },
    {
      "epoch": 0.8737864077669902,
      "grad_norm": 0.0,
      "learning_rate": 4.1163666612634984e-05,
      "loss": 2.0385,
      "step": 5625
    },
    {
      "epoch": 0.8739417475728155,
      "grad_norm": 0.0,
      "learning_rate": 4.1063751003031115e-05,
      "loss": 2.0661,
      "step": 5626
    },
    {
      "epoch": 0.8740970873786408,
      "grad_norm": 0.0,
      "learning_rate": 4.096395160929461e-05,
      "loss": 2.1347,
      "step": 5627
    },
    {
      "epoch": 0.874252427184466,
      "grad_norm": 0.0,
      "learning_rate": 4.0864268456697716e-05,
      "loss": 2.1554,
      "step": 5628
    },
    {
      "epoch": 0.8744077669902912,
      "grad_norm": 0.0,
      "learning_rate": 4.07647015704829e-05,
      "loss": 2.1194,
      "step": 5629
    },
    {
      "epoch": 0.8745631067961165,
      "grad_norm": 0.0,
      "learning_rate": 4.066525097586349e-05,
      "loss": 2.013,
      "step": 5630
    },
    {
      "epoch": 0.8747184466019418,
      "grad_norm": 0.0,
      "learning_rate": 4.056591669802312e-05,
      "loss": 2.1157,
      "step": 5631
    },
    {
      "epoch": 0.874873786407767,
      "grad_norm": 0.0,
      "learning_rate": 4.0466698762116225e-05,
      "loss": 2.1948,
      "step": 5632
    },
    {
      "epoch": 0.8750291262135922,
      "grad_norm": 0.0,
      "learning_rate": 4.036759719326755e-05,
      "loss": 2.1191,
      "step": 5633
    },
    {
      "epoch": 0.8751844660194175,
      "grad_norm": 0.0,
      "learning_rate": 4.026861201657239e-05,
      "loss": 1.968,
      "step": 5634
    },
    {
      "epoch": 0.8753398058252427,
      "grad_norm": 0.0,
      "learning_rate": 4.016974325709683e-05,
      "loss": 2.1282,
      "step": 5635
    },
    {
      "epoch": 0.875495145631068,
      "grad_norm": 0.0,
      "learning_rate": 4.007099093987709e-05,
      "loss": 2.065,
      "step": 5636
    },
    {
      "epoch": 0.8756504854368932,
      "grad_norm": 0.0,
      "learning_rate": 3.99723550899202e-05,
      "loss": 2.0472,
      "step": 5637
    },
    {
      "epoch": 0.8758058252427184,
      "grad_norm": 0.0,
      "learning_rate": 3.987383573220366e-05,
      "loss": 1.9985,
      "step": 5638
    },
    {
      "epoch": 0.8759611650485437,
      "grad_norm": 0.0,
      "learning_rate": 3.977543289167529e-05,
      "loss": 2.0205,
      "step": 5639
    },
    {
      "epoch": 0.876116504854369,
      "grad_norm": 0.0,
      "learning_rate": 3.967714659325361e-05,
      "loss": 2.1504,
      "step": 5640
    },
    {
      "epoch": 0.8762718446601941,
      "grad_norm": 0.0,
      "learning_rate": 3.957897686182743e-05,
      "loss": 2.0772,
      "step": 5641
    },
    {
      "epoch": 0.8764271844660194,
      "grad_norm": 0.0,
      "learning_rate": 3.94809237222562e-05,
      "loss": 2.0734,
      "step": 5642
    },
    {
      "epoch": 0.8765825242718447,
      "grad_norm": 0.0,
      "learning_rate": 3.938298719936994e-05,
      "loss": 2.0048,
      "step": 5643
    },
    {
      "epoch": 0.8767378640776698,
      "grad_norm": 0.0,
      "learning_rate": 3.928516731796883e-05,
      "loss": 2.0439,
      "step": 5644
    },
    {
      "epoch": 0.8768932038834951,
      "grad_norm": 0.0,
      "learning_rate": 3.9187464102823646e-05,
      "loss": 2.0112,
      "step": 5645
    },
    {
      "epoch": 0.8770485436893204,
      "grad_norm": 0.0,
      "learning_rate": 3.9089877578675824e-05,
      "loss": 2.1332,
      "step": 5646
    },
    {
      "epoch": 0.8772038834951457,
      "grad_norm": 0.0,
      "learning_rate": 3.899240777023694e-05,
      "loss": 2.1278,
      "step": 5647
    },
    {
      "epoch": 0.8773592233009708,
      "grad_norm": 0.0,
      "learning_rate": 3.8895054702189236e-05,
      "loss": 1.9214,
      "step": 5648
    },
    {
      "epoch": 0.8775145631067961,
      "grad_norm": 0.0,
      "learning_rate": 3.879781839918523e-05,
      "loss": 2.0098,
      "step": 5649
    },
    {
      "epoch": 0.8776699029126214,
      "grad_norm": 0.0,
      "learning_rate": 3.870069888584804e-05,
      "loss": 1.9993,
      "step": 5650
    },
    {
      "epoch": 0.8778252427184466,
      "grad_norm": 0.0,
      "learning_rate": 3.860369618677101e-05,
      "loss": 2.101,
      "step": 5651
    },
    {
      "epoch": 0.8779805825242718,
      "grad_norm": 0.0,
      "learning_rate": 3.850681032651809e-05,
      "loss": 2.0422,
      "step": 5652
    },
    {
      "epoch": 0.8781359223300971,
      "grad_norm": 0.0,
      "learning_rate": 3.841004132962367e-05,
      "loss": 2.0168,
      "step": 5653
    },
    {
      "epoch": 0.8782912621359223,
      "grad_norm": 0.0,
      "learning_rate": 3.831338922059224e-05,
      "loss": 2.1396,
      "step": 5654
    },
    {
      "epoch": 0.8784466019417476,
      "grad_norm": 0.0,
      "learning_rate": 3.821685402389907e-05,
      "loss": 2.0407,
      "step": 5655
    },
    {
      "epoch": 0.8786019417475728,
      "grad_norm": 0.0,
      "learning_rate": 3.8120435763989604e-05,
      "loss": 2.1163,
      "step": 5656
    },
    {
      "epoch": 0.878757281553398,
      "grad_norm": 0.0,
      "learning_rate": 3.802413446527958e-05,
      "loss": 2.115,
      "step": 5657
    },
    {
      "epoch": 0.8789126213592233,
      "grad_norm": 0.0,
      "learning_rate": 3.79279501521555e-05,
      "loss": 2.1199,
      "step": 5658
    },
    {
      "epoch": 0.8790679611650486,
      "grad_norm": 0.0,
      "learning_rate": 3.7831882848973707e-05,
      "loss": 2.0046,
      "step": 5659
    },
    {
      "epoch": 0.8792233009708738,
      "grad_norm": 0.0,
      "learning_rate": 3.773593258006147e-05,
      "loss": 2.1915,
      "step": 5660
    },
    {
      "epoch": 0.879378640776699,
      "grad_norm": 0.0,
      "learning_rate": 3.7640099369716074e-05,
      "loss": 1.9447,
      "step": 5661
    },
    {
      "epoch": 0.8795339805825243,
      "grad_norm": 0.0,
      "learning_rate": 3.754438324220516e-05,
      "loss": 2.0491,
      "step": 5662
    },
    {
      "epoch": 0.8796893203883495,
      "grad_norm": 0.0,
      "learning_rate": 3.7448784221767004e-05,
      "loss": 2.1841,
      "step": 5663
    },
    {
      "epoch": 0.8798446601941747,
      "grad_norm": 0.0,
      "learning_rate": 3.735330233260981e-05,
      "loss": 1.9959,
      "step": 5664
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.0,
      "learning_rate": 3.7257937598912496e-05,
      "loss": 2.1724,
      "step": 5665
    },
    {
      "epoch": 0.8801553398058253,
      "grad_norm": 0.0,
      "learning_rate": 3.7162690044823976e-05,
      "loss": 2.0818,
      "step": 5666
    },
    {
      "epoch": 0.8803106796116504,
      "grad_norm": 0.0,
      "learning_rate": 3.706755969446396e-05,
      "loss": 2.1248,
      "step": 5667
    },
    {
      "epoch": 0.8804660194174757,
      "grad_norm": 0.0,
      "learning_rate": 3.697254657192195e-05,
      "loss": 2.0634,
      "step": 5668
    },
    {
      "epoch": 0.880621359223301,
      "grad_norm": 0.0,
      "learning_rate": 3.687765070125804e-05,
      "loss": 2.0648,
      "step": 5669
    },
    {
      "epoch": 0.8807766990291263,
      "grad_norm": 0.0,
      "learning_rate": 3.678287210650261e-05,
      "loss": 2.116,
      "step": 5670
    },
    {
      "epoch": 0.8809320388349514,
      "grad_norm": 0.0,
      "learning_rate": 3.668821081165647e-05,
      "loss": 2.1426,
      "step": 5671
    },
    {
      "epoch": 0.8810873786407767,
      "grad_norm": 0.0,
      "learning_rate": 3.659366684069037e-05,
      "loss": 2.0563,
      "step": 5672
    },
    {
      "epoch": 0.881242718446602,
      "grad_norm": 0.0,
      "learning_rate": 3.649924021754575e-05,
      "loss": 2.0696,
      "step": 5673
    },
    {
      "epoch": 0.8813980582524272,
      "grad_norm": 0.0,
      "learning_rate": 3.6404930966133955e-05,
      "loss": 2.2192,
      "step": 5674
    },
    {
      "epoch": 0.8815533980582524,
      "grad_norm": 0.0,
      "learning_rate": 3.631073911033689e-05,
      "loss": 2.0356,
      "step": 5675
    },
    {
      "epoch": 0.8817087378640777,
      "grad_norm": 0.0,
      "learning_rate": 3.621666467400675e-05,
      "loss": 2.0602,
      "step": 5676
    },
    {
      "epoch": 0.8818640776699029,
      "grad_norm": 0.0,
      "learning_rate": 3.612270768096576e-05,
      "loss": 1.9874,
      "step": 5677
    },
    {
      "epoch": 0.8820194174757282,
      "grad_norm": 0.0,
      "learning_rate": 3.602886815500661e-05,
      "loss": 2.031,
      "step": 5678
    },
    {
      "epoch": 0.8821747572815534,
      "grad_norm": 0.0,
      "learning_rate": 3.593514611989213e-05,
      "loss": 2.0872,
      "step": 5679
    },
    {
      "epoch": 0.8823300970873786,
      "grad_norm": 0.0,
      "learning_rate": 3.5841541599355375e-05,
      "loss": 1.9986,
      "step": 5680
    },
    {
      "epoch": 0.8824854368932039,
      "grad_norm": 0.0,
      "learning_rate": 3.574805461709985e-05,
      "loss": 2.024,
      "step": 5681
    },
    {
      "epoch": 0.8826407766990291,
      "grad_norm": 0.0,
      "learning_rate": 3.5654685196798986e-05,
      "loss": 2.0178,
      "step": 5682
    },
    {
      "epoch": 0.8827961165048543,
      "grad_norm": 0.0,
      "learning_rate": 3.556143336209683e-05,
      "loss": 2.1294,
      "step": 5683
    },
    {
      "epoch": 0.8829514563106796,
      "grad_norm": 0.0,
      "learning_rate": 3.5468299136607154e-05,
      "loss": 2.0154,
      "step": 5684
    },
    {
      "epoch": 0.8831067961165049,
      "grad_norm": 0.0,
      "learning_rate": 3.5375282543914364e-05,
      "loss": 2.0569,
      "step": 5685
    },
    {
      "epoch": 0.88326213592233,
      "grad_norm": 0.0,
      "learning_rate": 3.528238360757302e-05,
      "loss": 2.047,
      "step": 5686
    },
    {
      "epoch": 0.8834174757281553,
      "grad_norm": 0.0,
      "learning_rate": 3.518960235110763e-05,
      "loss": 2.0468,
      "step": 5687
    },
    {
      "epoch": 0.8835728155339806,
      "grad_norm": 0.0,
      "learning_rate": 3.5096938798013236e-05,
      "loss": 2.1525,
      "step": 5688
    },
    {
      "epoch": 0.8837281553398059,
      "grad_norm": 0.0,
      "learning_rate": 3.5004392971754786e-05,
      "loss": 2.0567,
      "step": 5689
    },
    {
      "epoch": 0.883883495145631,
      "grad_norm": 0.0,
      "learning_rate": 3.49119648957677e-05,
      "loss": 2.0827,
      "step": 5690
    },
    {
      "epoch": 0.8840388349514563,
      "grad_norm": 0.0,
      "learning_rate": 3.4819654593457374e-05,
      "loss": 2.1167,
      "step": 5691
    },
    {
      "epoch": 0.8841941747572816,
      "grad_norm": 0.0,
      "learning_rate": 3.472746208819927e-05,
      "loss": 2.2233,
      "step": 5692
    },
    {
      "epoch": 0.8843495145631068,
      "grad_norm": 0.0,
      "learning_rate": 3.4635387403339324e-05,
      "loss": 2.0497,
      "step": 5693
    },
    {
      "epoch": 0.884504854368932,
      "grad_norm": 0.0,
      "learning_rate": 3.4543430562193556e-05,
      "loss": 2.1897,
      "step": 5694
    },
    {
      "epoch": 0.8846601941747573,
      "grad_norm": 0.0,
      "learning_rate": 3.445159158804795e-05,
      "loss": 2.068,
      "step": 5695
    },
    {
      "epoch": 0.8848155339805825,
      "grad_norm": 0.0,
      "learning_rate": 3.435987050415895e-05,
      "loss": 2.0948,
      "step": 5696
    },
    {
      "epoch": 0.8849708737864078,
      "grad_norm": 0.0,
      "learning_rate": 3.426826733375277e-05,
      "loss": 2.02,
      "step": 5697
    },
    {
      "epoch": 0.885126213592233,
      "grad_norm": 0.0,
      "learning_rate": 3.417678210002611e-05,
      "loss": 2.0585,
      "step": 5698
    },
    {
      "epoch": 0.8852815533980583,
      "grad_norm": 0.0,
      "learning_rate": 3.408541482614558e-05,
      "loss": 2.0354,
      "step": 5699
    },
    {
      "epoch": 0.8854368932038835,
      "grad_norm": 0.0,
      "learning_rate": 3.399416553524803e-05,
      "loss": 2.0865,
      "step": 5700
    },
    {
      "epoch": 0.8855922330097087,
      "grad_norm": 0.0,
      "learning_rate": 3.3903034250440555e-05,
      "loss": 2.1446,
      "step": 5701
    },
    {
      "epoch": 0.885747572815534,
      "grad_norm": 0.0,
      "learning_rate": 3.381202099479996e-05,
      "loss": 2.0878,
      "step": 5702
    },
    {
      "epoch": 0.8859029126213592,
      "grad_norm": 0.0,
      "learning_rate": 3.3721125791373545e-05,
      "loss": 2.0837,
      "step": 5703
    },
    {
      "epoch": 0.8860582524271845,
      "grad_norm": 0.0,
      "learning_rate": 3.363034866317866e-05,
      "loss": 2.0974,
      "step": 5704
    },
    {
      "epoch": 0.8862135922330097,
      "grad_norm": 0.0,
      "learning_rate": 3.353968963320253e-05,
      "loss": 1.9855,
      "step": 5705
    },
    {
      "epoch": 0.8863689320388349,
      "grad_norm": 0.0,
      "learning_rate": 3.34491487244028e-05,
      "loss": 2.0965,
      "step": 5706
    },
    {
      "epoch": 0.8865242718446602,
      "grad_norm": 0.0,
      "learning_rate": 3.335872595970685e-05,
      "loss": 1.9929,
      "step": 5707
    },
    {
      "epoch": 0.8866796116504855,
      "grad_norm": 0.0,
      "learning_rate": 3.326842136201247e-05,
      "loss": 2.1098,
      "step": 5708
    },
    {
      "epoch": 0.8868349514563106,
      "grad_norm": 0.0,
      "learning_rate": 3.3178234954187426e-05,
      "loss": 2.0291,
      "step": 5709
    },
    {
      "epoch": 0.8869902912621359,
      "grad_norm": 0.0,
      "learning_rate": 3.308816675906934e-05,
      "loss": 2.1048,
      "step": 5710
    },
    {
      "epoch": 0.8871456310679612,
      "grad_norm": 0.0,
      "learning_rate": 3.299821679946619e-05,
      "loss": 2.1157,
      "step": 5711
    },
    {
      "epoch": 0.8873009708737865,
      "grad_norm": 0.0,
      "learning_rate": 3.2908385098155856e-05,
      "loss": 2.0173,
      "step": 5712
    },
    {
      "epoch": 0.8874563106796116,
      "grad_norm": 0.0,
      "learning_rate": 3.281867167788638e-05,
      "loss": 2.0416,
      "step": 5713
    },
    {
      "epoch": 0.8876116504854369,
      "grad_norm": 0.0,
      "learning_rate": 3.272907656137569e-05,
      "loss": 2.1287,
      "step": 5714
    },
    {
      "epoch": 0.8877669902912622,
      "grad_norm": 0.0,
      "learning_rate": 3.263959977131187e-05,
      "loss": 2.0784,
      "step": 5715
    },
    {
      "epoch": 0.8879223300970874,
      "grad_norm": 0.0,
      "learning_rate": 3.255024133035306e-05,
      "loss": 2.0457,
      "step": 5716
    },
    {
      "epoch": 0.8880776699029126,
      "grad_norm": 0.0,
      "learning_rate": 3.246100126112728e-05,
      "loss": 2.06,
      "step": 5717
    },
    {
      "epoch": 0.8882330097087379,
      "grad_norm": 0.0,
      "learning_rate": 3.237187958623272e-05,
      "loss": 2.06,
      "step": 5718
    },
    {
      "epoch": 0.8883883495145631,
      "grad_norm": 0.0,
      "learning_rate": 3.228287632823773e-05,
      "loss": 2.0715,
      "step": 5719
    },
    {
      "epoch": 0.8885436893203883,
      "grad_norm": 0.0,
      "learning_rate": 3.2193991509680276e-05,
      "loss": 2.1599,
      "step": 5720
    },
    {
      "epoch": 0.8886990291262136,
      "grad_norm": 0.0,
      "learning_rate": 3.210522515306874e-05,
      "loss": 2.0073,
      "step": 5721
    },
    {
      "epoch": 0.8888543689320388,
      "grad_norm": 0.0,
      "learning_rate": 3.201657728088109e-05,
      "loss": 2.0468,
      "step": 5722
    },
    {
      "epoch": 0.8890097087378641,
      "grad_norm": 0.0,
      "learning_rate": 3.19280479155657e-05,
      "loss": 2.1273,
      "step": 5723
    },
    {
      "epoch": 0.8891650485436893,
      "grad_norm": 0.0,
      "learning_rate": 3.183963707954079e-05,
      "loss": 2.1575,
      "step": 5724
    },
    {
      "epoch": 0.8893203883495145,
      "grad_norm": 0.0,
      "learning_rate": 3.1751344795194317e-05,
      "loss": 2.1512,
      "step": 5725
    },
    {
      "epoch": 0.8894757281553398,
      "grad_norm": 0.0,
      "learning_rate": 3.166317108488453e-05,
      "loss": 2.0592,
      "step": 5726
    },
    {
      "epoch": 0.8896310679611651,
      "grad_norm": 0.0,
      "learning_rate": 3.157511597093971e-05,
      "loss": 2.1394,
      "step": 5727
    },
    {
      "epoch": 0.8897864077669903,
      "grad_norm": 0.0,
      "learning_rate": 3.1487179475657655e-05,
      "loss": 2.024,
      "step": 5728
    },
    {
      "epoch": 0.8899417475728155,
      "grad_norm": 0.0,
      "learning_rate": 3.1399361621306686e-05,
      "loss": 2.0647,
      "step": 5729
    },
    {
      "epoch": 0.8900970873786408,
      "grad_norm": 0.0,
      "learning_rate": 3.131166243012457e-05,
      "loss": 2.1288,
      "step": 5730
    },
    {
      "epoch": 0.8902524271844661,
      "grad_norm": 0.0,
      "learning_rate": 3.122408192431941e-05,
      "loss": 2.0219,
      "step": 5731
    },
    {
      "epoch": 0.8904077669902912,
      "grad_norm": 0.0,
      "learning_rate": 3.113662012606916e-05,
      "loss": 2.0047,
      "step": 5732
    },
    {
      "epoch": 0.8905631067961165,
      "grad_norm": 0.0,
      "learning_rate": 3.104927705752153e-05,
      "loss": 2.0573,
      "step": 5733
    },
    {
      "epoch": 0.8907184466019418,
      "grad_norm": 0.0,
      "learning_rate": 3.0962052740794445e-05,
      "loss": 2.1654,
      "step": 5734
    },
    {
      "epoch": 0.890873786407767,
      "grad_norm": 0.0,
      "learning_rate": 3.087494719797546e-05,
      "loss": 2.152,
      "step": 5735
    },
    {
      "epoch": 0.8910291262135922,
      "grad_norm": 0.0,
      "learning_rate": 3.078796045112236e-05,
      "loss": 2.0471,
      "step": 5736
    },
    {
      "epoch": 0.8911844660194175,
      "grad_norm": 0.0,
      "learning_rate": 3.070109252226255e-05,
      "loss": 2.0543,
      "step": 5737
    },
    {
      "epoch": 0.8913398058252427,
      "grad_norm": 0.0,
      "learning_rate": 3.0614343433393535e-05,
      "loss": 2.0143,
      "step": 5738
    },
    {
      "epoch": 0.8914951456310679,
      "grad_norm": 0.0,
      "learning_rate": 3.052771320648273e-05,
      "loss": 2.0538,
      "step": 5739
    },
    {
      "epoch": 0.8916504854368932,
      "grad_norm": 0.0,
      "learning_rate": 3.0441201863467326e-05,
      "loss": 2.067,
      "step": 5740
    },
    {
      "epoch": 0.8918058252427185,
      "grad_norm": 0.0,
      "learning_rate": 3.0354809426254515e-05,
      "loss": 2.1049,
      "step": 5741
    },
    {
      "epoch": 0.8919611650485437,
      "grad_norm": 0.0,
      "learning_rate": 3.0268535916721406e-05,
      "loss": 2.0443,
      "step": 5742
    },
    {
      "epoch": 0.8921165048543689,
      "grad_norm": 0.0,
      "learning_rate": 3.018238135671486e-05,
      "loss": 2.0077,
      "step": 5743
    },
    {
      "epoch": 0.8922718446601942,
      "grad_norm": 0.0,
      "learning_rate": 3.0096345768051758e-05,
      "loss": 1.9086,
      "step": 5744
    },
    {
      "epoch": 0.8924271844660194,
      "grad_norm": 0.0,
      "learning_rate": 3.0010429172518672e-05,
      "loss": 2.0318,
      "step": 5745
    },
    {
      "epoch": 0.8925825242718447,
      "grad_norm": 0.0,
      "learning_rate": 2.9924631591872254e-05,
      "loss": 2.1459,
      "step": 5746
    },
    {
      "epoch": 0.8927378640776699,
      "grad_norm": 0.0,
      "learning_rate": 2.9838953047839013e-05,
      "loss": 2.0319,
      "step": 5747
    },
    {
      "epoch": 0.8928932038834951,
      "grad_norm": 0.0,
      "learning_rate": 2.9753393562114984e-05,
      "loss": 2.1417,
      "step": 5748
    },
    {
      "epoch": 0.8930485436893204,
      "grad_norm": 0.0,
      "learning_rate": 2.96679531563665e-05,
      "loss": 2.0909,
      "step": 5749
    },
    {
      "epoch": 0.8932038834951457,
      "grad_norm": 0.0,
      "learning_rate": 2.958263185222937e-05,
      "loss": 2.0218,
      "step": 5750
    },
    {
      "epoch": 0.8933592233009708,
      "grad_norm": 0.0,
      "learning_rate": 2.9497429671309527e-05,
      "loss": 2.0515,
      "step": 5751
    },
    {
      "epoch": 0.8935145631067961,
      "grad_norm": 0.0,
      "learning_rate": 2.9412346635182552e-05,
      "loss": 2.1581,
      "step": 5752
    },
    {
      "epoch": 0.8936699029126214,
      "grad_norm": 0.0,
      "learning_rate": 2.9327382765393927e-05,
      "loss": 1.9968,
      "step": 5753
    },
    {
      "epoch": 0.8938252427184467,
      "grad_norm": 0.0,
      "learning_rate": 2.9242538083459004e-05,
      "loss": 2.1494,
      "step": 5754
    },
    {
      "epoch": 0.8939805825242718,
      "grad_norm": 0.0,
      "learning_rate": 2.915781261086281e-05,
      "loss": 2.1182,
      "step": 5755
    },
    {
      "epoch": 0.8941359223300971,
      "grad_norm": 0.0,
      "learning_rate": 2.90732063690603e-05,
      "loss": 2.0864,
      "step": 5756
    },
    {
      "epoch": 0.8942912621359224,
      "grad_norm": 0.0,
      "learning_rate": 2.898871937947628e-05,
      "loss": 1.979,
      "step": 5757
    },
    {
      "epoch": 0.8944466019417475,
      "grad_norm": 0.0,
      "learning_rate": 2.8904351663505302e-05,
      "loss": 1.9692,
      "step": 5758
    },
    {
      "epoch": 0.8946019417475728,
      "grad_norm": 0.0,
      "learning_rate": 2.8820103242511496e-05,
      "loss": 2.0835,
      "step": 5759
    },
    {
      "epoch": 0.8947572815533981,
      "grad_norm": 0.0,
      "learning_rate": 2.873597413782919e-05,
      "loss": 2.1387,
      "step": 5760
    },
    {
      "epoch": 0.8949126213592233,
      "grad_norm": 0.0,
      "learning_rate": 2.8651964370762175e-05,
      "loss": 2.115,
      "step": 5761
    },
    {
      "epoch": 0.8950679611650485,
      "grad_norm": 0.0,
      "learning_rate": 2.856807396258426e-05,
      "loss": 2.0612,
      "step": 5762
    },
    {
      "epoch": 0.8952233009708738,
      "grad_norm": 0.0,
      "learning_rate": 2.8484302934538796e-05,
      "loss": 1.9601,
      "step": 5763
    },
    {
      "epoch": 0.895378640776699,
      "grad_norm": 0.0,
      "learning_rate": 2.8400651307839088e-05,
      "loss": 2.1326,
      "step": 5764
    },
    {
      "epoch": 0.8955339805825243,
      "grad_norm": 0.0,
      "learning_rate": 2.831711910366813e-05,
      "loss": 2.0883,
      "step": 5765
    },
    {
      "epoch": 0.8956893203883495,
      "grad_norm": 0.0,
      "learning_rate": 2.823370634317862e-05,
      "loss": 2.1417,
      "step": 5766
    },
    {
      "epoch": 0.8958446601941747,
      "grad_norm": 0.0,
      "learning_rate": 2.8150413047493218e-05,
      "loss": 2.0272,
      "step": 5767
    },
    {
      "epoch": 0.896,
      "grad_norm": 0.0,
      "learning_rate": 2.8067239237703935e-05,
      "loss": 2.0578,
      "step": 5768
    },
    {
      "epoch": 0.8961553398058253,
      "grad_norm": 0.0,
      "learning_rate": 2.7984184934873037e-05,
      "loss": 2.0545,
      "step": 5769
    },
    {
      "epoch": 0.8963106796116505,
      "grad_norm": 0.0,
      "learning_rate": 2.7901250160032144e-05,
      "loss": 1.965,
      "step": 5770
    },
    {
      "epoch": 0.8964660194174757,
      "grad_norm": 0.0,
      "learning_rate": 2.781843493418268e-05,
      "loss": 2.1268,
      "step": 5771
    },
    {
      "epoch": 0.896621359223301,
      "grad_norm": 0.0,
      "learning_rate": 2.773573927829598e-05,
      "loss": 2.06,
      "step": 5772
    },
    {
      "epoch": 0.8967766990291263,
      "grad_norm": 0.0,
      "learning_rate": 2.765316321331274e-05,
      "loss": 2.113,
      "step": 5773
    },
    {
      "epoch": 0.8969320388349514,
      "grad_norm": 0.0,
      "learning_rate": 2.7570706760143782e-05,
      "loss": 1.9716,
      "step": 5774
    },
    {
      "epoch": 0.8970873786407767,
      "grad_norm": 0.0,
      "learning_rate": 2.7488369939669466e-05,
      "loss": 2.1137,
      "step": 5775
    },
    {
      "epoch": 0.897242718446602,
      "grad_norm": 0.0,
      "learning_rate": 2.7406152772739722e-05,
      "loss": 2.0616,
      "step": 5776
    },
    {
      "epoch": 0.8973980582524271,
      "grad_norm": 0.0,
      "learning_rate": 2.7324055280174455e-05,
      "loss": 2.1008,
      "step": 5777
    },
    {
      "epoch": 0.8975533980582524,
      "grad_norm": 0.0,
      "learning_rate": 2.7242077482762917e-05,
      "loss": 1.9327,
      "step": 5778
    },
    {
      "epoch": 0.8977087378640777,
      "grad_norm": 0.0,
      "learning_rate": 2.7160219401264342e-05,
      "loss": 2.1897,
      "step": 5779
    },
    {
      "epoch": 0.897864077669903,
      "grad_norm": 0.0,
      "learning_rate": 2.70784810564077e-05,
      "loss": 2.1058,
      "step": 5780
    },
    {
      "epoch": 0.8980194174757281,
      "grad_norm": 0.0,
      "learning_rate": 2.699686246889127e-05,
      "loss": 2.0638,
      "step": 5781
    },
    {
      "epoch": 0.8981747572815534,
      "grad_norm": 0.0,
      "learning_rate": 2.6915363659383408e-05,
      "loss": 2.1315,
      "step": 5782
    },
    {
      "epoch": 0.8983300970873787,
      "grad_norm": 0.0,
      "learning_rate": 2.6833984648521713e-05,
      "loss": 2.1228,
      "step": 5783
    },
    {
      "epoch": 0.8984854368932039,
      "grad_norm": 0.0,
      "learning_rate": 2.6752725456913872e-05,
      "loss": 2.1111,
      "step": 5784
    },
    {
      "epoch": 0.8986407766990291,
      "grad_norm": 0.0,
      "learning_rate": 2.6671586105137093e-05,
      "loss": 2.1842,
      "step": 5785
    },
    {
      "epoch": 0.8987961165048544,
      "grad_norm": 0.0,
      "learning_rate": 2.6590566613738098e-05,
      "loss": 2.0953,
      "step": 5786
    },
    {
      "epoch": 0.8989514563106796,
      "grad_norm": 0.0,
      "learning_rate": 2.6509667003233373e-05,
      "loss": 2.0659,
      "step": 5787
    },
    {
      "epoch": 0.8991067961165049,
      "grad_norm": 0.0,
      "learning_rate": 2.6428887294109028e-05,
      "loss": 2.0277,
      "step": 5788
    },
    {
      "epoch": 0.8992621359223301,
      "grad_norm": 0.0,
      "learning_rate": 2.634822750682081e-05,
      "loss": 2.0157,
      "step": 5789
    },
    {
      "epoch": 0.8994174757281553,
      "grad_norm": 0.0,
      "learning_rate": 2.626768766179416e-05,
      "loss": 2.1365,
      "step": 5790
    },
    {
      "epoch": 0.8995728155339806,
      "grad_norm": 0.0,
      "learning_rate": 2.6187267779423986e-05,
      "loss": 2.0521,
      "step": 5791
    },
    {
      "epoch": 0.8997281553398059,
      "grad_norm": 0.0,
      "learning_rate": 2.6106967880074993e-05,
      "loss": 2.0714,
      "step": 5792
    },
    {
      "epoch": 0.899883495145631,
      "grad_norm": 0.0,
      "learning_rate": 2.6026787984081425e-05,
      "loss": 2.1131,
      "step": 5793
    },
    {
      "epoch": 0.9000388349514563,
      "grad_norm": 0.0,
      "learning_rate": 2.594672811174703e-05,
      "loss": 2.1628,
      "step": 5794
    },
    {
      "epoch": 0.9001941747572816,
      "grad_norm": 0.0,
      "learning_rate": 2.586678828334543e-05,
      "loss": 2.105,
      "step": 5795
    },
    {
      "epoch": 0.9003495145631067,
      "grad_norm": 0.0,
      "learning_rate": 2.5786968519119546e-05,
      "loss": 2.1484,
      "step": 5796
    },
    {
      "epoch": 0.900504854368932,
      "grad_norm": 0.0,
      "learning_rate": 2.5707268839282094e-05,
      "loss": 2.056,
      "step": 5797
    },
    {
      "epoch": 0.9006601941747573,
      "grad_norm": 0.0,
      "learning_rate": 2.562768926401543e-05,
      "loss": 1.9918,
      "step": 5798
    },
    {
      "epoch": 0.9008155339805826,
      "grad_norm": 0.0,
      "learning_rate": 2.5548229813471215e-05,
      "loss": 2.1805,
      "step": 5799
    },
    {
      "epoch": 0.9009708737864077,
      "grad_norm": 0.0,
      "learning_rate": 2.546889050777107e-05,
      "loss": 2.0828,
      "step": 5800
    },
    {
      "epoch": 0.901126213592233,
      "grad_norm": 0.0,
      "learning_rate": 2.5389671367005762e-05,
      "loss": 2.0848,
      "step": 5801
    },
    {
      "epoch": 0.9012815533980583,
      "grad_norm": 0.0,
      "learning_rate": 2.5310572411236022e-05,
      "loss": 1.9918,
      "step": 5802
    },
    {
      "epoch": 0.9014368932038835,
      "grad_norm": 0.0,
      "learning_rate": 2.5231593660491937e-05,
      "loss": 2.062,
      "step": 5803
    },
    {
      "epoch": 0.9015922330097087,
      "grad_norm": 0.0,
      "learning_rate": 2.515273513477323e-05,
      "loss": 2.0287,
      "step": 5804
    },
    {
      "epoch": 0.901747572815534,
      "grad_norm": 0.0,
      "learning_rate": 2.5073996854049097e-05,
      "loss": 2.0364,
      "step": 5805
    },
    {
      "epoch": 0.9019029126213592,
      "grad_norm": 0.0,
      "learning_rate": 2.499537883825831e-05,
      "loss": 2.1498,
      "step": 5806
    },
    {
      "epoch": 0.9020582524271845,
      "grad_norm": 0.0,
      "learning_rate": 2.4916881107309218e-05,
      "loss": 2.0635,
      "step": 5807
    },
    {
      "epoch": 0.9022135922330097,
      "grad_norm": 0.0,
      "learning_rate": 2.483850368107987e-05,
      "loss": 2.0406,
      "step": 5808
    },
    {
      "epoch": 0.902368932038835,
      "grad_norm": 0.0,
      "learning_rate": 2.4760246579417444e-05,
      "loss": 2.0406,
      "step": 5809
    },
    {
      "epoch": 0.9025242718446602,
      "grad_norm": 0.0,
      "learning_rate": 2.4682109822139144e-05,
      "loss": 2.0443,
      "step": 5810
    },
    {
      "epoch": 0.9026796116504855,
      "grad_norm": 0.0,
      "learning_rate": 2.4604093429031195e-05,
      "loss": 2.0454,
      "step": 5811
    },
    {
      "epoch": 0.9028349514563107,
      "grad_norm": 0.0,
      "learning_rate": 2.452619741984974e-05,
      "loss": 2.115,
      "step": 5812
    },
    {
      "epoch": 0.9029902912621359,
      "grad_norm": 0.0,
      "learning_rate": 2.4448421814320276e-05,
      "loss": 2.0502,
      "step": 5813
    },
    {
      "epoch": 0.9031456310679612,
      "grad_norm": 0.0,
      "learning_rate": 2.4370766632137776e-05,
      "loss": 2.1571,
      "step": 5814
    },
    {
      "epoch": 0.9033009708737864,
      "grad_norm": 0.0,
      "learning_rate": 2.429323189296695e-05,
      "loss": 2.0639,
      "step": 5815
    },
    {
      "epoch": 0.9034563106796116,
      "grad_norm": 0.0,
      "learning_rate": 2.421581761644154e-05,
      "loss": 2.031,
      "step": 5816
    },
    {
      "epoch": 0.9036116504854369,
      "grad_norm": 0.0,
      "learning_rate": 2.4138523822165194e-05,
      "loss": 2.0442,
      "step": 5817
    },
    {
      "epoch": 0.9037669902912622,
      "grad_norm": 0.0,
      "learning_rate": 2.406135052971098e-05,
      "loss": 2.061,
      "step": 5818
    },
    {
      "epoch": 0.9039223300970873,
      "grad_norm": 0.0,
      "learning_rate": 2.398429775862132e-05,
      "loss": 2.0947,
      "step": 5819
    },
    {
      "epoch": 0.9040776699029126,
      "grad_norm": 0.0,
      "learning_rate": 2.3907365528408275e-05,
      "loss": 2.0135,
      "step": 5820
    },
    {
      "epoch": 0.9042330097087379,
      "grad_norm": 0.0,
      "learning_rate": 2.383055385855315e-05,
      "loss": 2.0505,
      "step": 5821
    },
    {
      "epoch": 0.9043883495145632,
      "grad_norm": 0.0,
      "learning_rate": 2.375386276850705e-05,
      "loss": 2.1488,
      "step": 5822
    },
    {
      "epoch": 0.9045436893203883,
      "grad_norm": 0.0,
      "learning_rate": 2.3677292277690278e-05,
      "loss": 2.1155,
      "step": 5823
    },
    {
      "epoch": 0.9046990291262136,
      "grad_norm": 0.0,
      "learning_rate": 2.3600842405492708e-05,
      "loss": 2.0242,
      "step": 5824
    },
    {
      "epoch": 0.9048543689320389,
      "grad_norm": 0.0,
      "learning_rate": 2.352451317127363e-05,
      "loss": 2.0796,
      "step": 5825
    },
    {
      "epoch": 0.9050097087378641,
      "grad_norm": 0.0,
      "learning_rate": 2.3448304594361814e-05,
      "loss": 2.0754,
      "step": 5826
    },
    {
      "epoch": 0.9051650485436893,
      "grad_norm": 0.0,
      "learning_rate": 2.337221669405548e-05,
      "loss": 2.1158,
      "step": 5827
    },
    {
      "epoch": 0.9053203883495146,
      "grad_norm": 0.0,
      "learning_rate": 2.329624948962228e-05,
      "loss": 2.0881,
      "step": 5828
    },
    {
      "epoch": 0.9054757281553398,
      "grad_norm": 0.0,
      "learning_rate": 2.3220403000299152e-05,
      "loss": 2.1042,
      "step": 5829
    },
    {
      "epoch": 0.9056310679611651,
      "grad_norm": 0.0,
      "learning_rate": 2.314467724529279e-05,
      "loss": 2.1075,
      "step": 5830
    },
    {
      "epoch": 0.9057864077669903,
      "grad_norm": 0.0,
      "learning_rate": 2.306907224377913e-05,
      "loss": 2.1577,
      "step": 5831
    },
    {
      "epoch": 0.9059417475728155,
      "grad_norm": 0.0,
      "learning_rate": 2.2993588014903365e-05,
      "loss": 2.1018,
      "step": 5832
    },
    {
      "epoch": 0.9060970873786408,
      "grad_norm": 0.0,
      "learning_rate": 2.2918224577780532e-05,
      "loss": 2.0146,
      "step": 5833
    },
    {
      "epoch": 0.906252427184466,
      "grad_norm": 0.0,
      "learning_rate": 2.284298195149459e-05,
      "loss": 2.1116,
      "step": 5834
    },
    {
      "epoch": 0.9064077669902912,
      "grad_norm": 0.0,
      "learning_rate": 2.276786015509924e-05,
      "loss": 2.0008,
      "step": 5835
    },
    {
      "epoch": 0.9065631067961165,
      "grad_norm": 0.0,
      "learning_rate": 2.2692859207617434e-05,
      "loss": 2.1631,
      "step": 5836
    },
    {
      "epoch": 0.9067184466019418,
      "grad_norm": 0.0,
      "learning_rate": 2.2617979128041642e-05,
      "loss": 2.1331,
      "step": 5837
    },
    {
      "epoch": 0.906873786407767,
      "grad_norm": 0.0,
      "learning_rate": 2.2543219935333703e-05,
      "loss": 2.0789,
      "step": 5838
    },
    {
      "epoch": 0.9070291262135922,
      "grad_norm": 0.0,
      "learning_rate": 2.246858164842458e-05,
      "loss": 2.1457,
      "step": 5839
    },
    {
      "epoch": 0.9071844660194175,
      "grad_norm": 0.0,
      "learning_rate": 2.239406428621499e-05,
      "loss": 2.0399,
      "step": 5840
    },
    {
      "epoch": 0.9073398058252428,
      "grad_norm": 0.0,
      "learning_rate": 2.2319667867574945e-05,
      "loss": 2.0025,
      "step": 5841
    },
    {
      "epoch": 0.9074951456310679,
      "grad_norm": 0.0,
      "learning_rate": 2.2245392411343547e-05,
      "loss": 2.0898,
      "step": 5842
    },
    {
      "epoch": 0.9076504854368932,
      "grad_norm": 0.0,
      "learning_rate": 2.217123793632969e-05,
      "loss": 2.0415,
      "step": 5843
    },
    {
      "epoch": 0.9078058252427185,
      "grad_norm": 0.0,
      "learning_rate": 2.2097204461311304e-05,
      "loss": 2.1393,
      "step": 5844
    },
    {
      "epoch": 0.9079611650485437,
      "grad_norm": 0.0,
      "learning_rate": 2.2023292005035778e-05,
      "loss": 2.0757,
      "step": 5845
    },
    {
      "epoch": 0.9081165048543689,
      "grad_norm": 0.0,
      "learning_rate": 2.1949500586219973e-05,
      "loss": 2.0405,
      "step": 5846
    },
    {
      "epoch": 0.9082718446601942,
      "grad_norm": 0.0,
      "learning_rate": 2.1875830223549888e-05,
      "loss": 1.9733,
      "step": 5847
    },
    {
      "epoch": 0.9084271844660194,
      "grad_norm": 0.0,
      "learning_rate": 2.1802280935681096e-05,
      "loss": 2.175,
      "step": 5848
    },
    {
      "epoch": 0.9085825242718446,
      "grad_norm": 0.0,
      "learning_rate": 2.1728852741238315e-05,
      "loss": 2.1059,
      "step": 5849
    },
    {
      "epoch": 0.9087378640776699,
      "grad_norm": 0.0,
      "learning_rate": 2.1655545658815724e-05,
      "loss": 2.0205,
      "step": 5850
    },
    {
      "epoch": 0.9088932038834951,
      "grad_norm": 0.0,
      "learning_rate": 2.158235970697675e-05,
      "loss": 2.1501,
      "step": 5851
    },
    {
      "epoch": 0.9090485436893204,
      "grad_norm": 0.0,
      "learning_rate": 2.1509294904254127e-05,
      "loss": 2.0494,
      "step": 5852
    },
    {
      "epoch": 0.9092038834951456,
      "grad_norm": 0.0,
      "learning_rate": 2.1436351269150112e-05,
      "loss": 2.0441,
      "step": 5853
    },
    {
      "epoch": 0.9093592233009709,
      "grad_norm": 0.0,
      "learning_rate": 2.136352882013598e-05,
      "loss": 2.1365,
      "step": 5854
    },
    {
      "epoch": 0.9095145631067961,
      "grad_norm": 0.0,
      "learning_rate": 2.129082757565254e-05,
      "loss": 2.0854,
      "step": 5855
    },
    {
      "epoch": 0.9096699029126214,
      "grad_norm": 0.0,
      "learning_rate": 2.1218247554109894e-05,
      "loss": 2.1184,
      "step": 5856
    },
    {
      "epoch": 0.9098252427184466,
      "grad_norm": 0.0,
      "learning_rate": 2.1145788773887286e-05,
      "loss": 2.1365,
      "step": 5857
    },
    {
      "epoch": 0.9099805825242718,
      "grad_norm": 0.0,
      "learning_rate": 2.1073451253333487e-05,
      "loss": 2.0688,
      "step": 5858
    },
    {
      "epoch": 0.9101359223300971,
      "grad_norm": 0.0,
      "learning_rate": 2.1001235010766284e-05,
      "loss": 2.0822,
      "step": 5859
    },
    {
      "epoch": 0.9102912621359224,
      "grad_norm": 0.0,
      "learning_rate": 2.0929140064472996e-05,
      "loss": 2.0468,
      "step": 5860
    },
    {
      "epoch": 0.9104466019417475,
      "grad_norm": 0.0,
      "learning_rate": 2.0857166432710296e-05,
      "loss": 2.1155,
      "step": 5861
    },
    {
      "epoch": 0.9106019417475728,
      "grad_norm": 0.0,
      "learning_rate": 2.078531413370366e-05,
      "loss": 2.1432,
      "step": 5862
    },
    {
      "epoch": 0.9107572815533981,
      "grad_norm": 0.0,
      "learning_rate": 2.0713583185648312e-05,
      "loss": 2.2011,
      "step": 5863
    },
    {
      "epoch": 0.9109126213592234,
      "grad_norm": 0.0,
      "learning_rate": 2.064197360670872e-05,
      "loss": 2.1652,
      "step": 5864
    },
    {
      "epoch": 0.9110679611650485,
      "grad_norm": 0.0,
      "learning_rate": 2.057048541501827e-05,
      "loss": 2.1647,
      "step": 5865
    },
    {
      "epoch": 0.9112233009708738,
      "grad_norm": 0.0,
      "learning_rate": 2.049911862867998e-05,
      "loss": 2.1011,
      "step": 5866
    },
    {
      "epoch": 0.9113786407766991,
      "grad_norm": 0.0,
      "learning_rate": 2.042787326576584e-05,
      "loss": 2.0475,
      "step": 5867
    },
    {
      "epoch": 0.9115339805825242,
      "grad_norm": 0.0,
      "learning_rate": 2.0356749344317414e-05,
      "loss": 2.1531,
      "step": 5868
    },
    {
      "epoch": 0.9116893203883495,
      "grad_norm": 0.0,
      "learning_rate": 2.0285746882345135e-05,
      "loss": 2.0228,
      "step": 5869
    },
    {
      "epoch": 0.9118446601941748,
      "grad_norm": 0.0,
      "learning_rate": 2.0214865897829004e-05,
      "loss": 2.0208,
      "step": 5870
    },
    {
      "epoch": 0.912,
      "grad_norm": 0.0,
      "learning_rate": 2.014410640871811e-05,
      "loss": 2.0246,
      "step": 5871
    },
    {
      "epoch": 0.9121553398058252,
      "grad_norm": 0.0,
      "learning_rate": 2.0073468432930675e-05,
      "loss": 2.011,
      "step": 5872
    },
    {
      "epoch": 0.9123106796116505,
      "grad_norm": 0.0,
      "learning_rate": 2.0002951988354446e-05,
      "loss": 2.0664,
      "step": 5873
    },
    {
      "epoch": 0.9124660194174757,
      "grad_norm": 0.0,
      "learning_rate": 1.9932557092846082e-05,
      "loss": 2.1122,
      "step": 5874
    },
    {
      "epoch": 0.912621359223301,
      "grad_norm": 0.0,
      "learning_rate": 1.986228376423166e-05,
      "loss": 2.1071,
      "step": 5875
    },
    {
      "epoch": 0.9127766990291262,
      "grad_norm": 0.0,
      "learning_rate": 1.9792132020306386e-05,
      "loss": 2.096,
      "step": 5876
    },
    {
      "epoch": 0.9129320388349514,
      "grad_norm": 0.0,
      "learning_rate": 1.9722101878834665e-05,
      "loss": 1.9939,
      "step": 5877
    },
    {
      "epoch": 0.9130873786407767,
      "grad_norm": 0.0,
      "learning_rate": 1.965219335755014e-05,
      "loss": 2.0656,
      "step": 5878
    },
    {
      "epoch": 0.913242718446602,
      "grad_norm": 0.0,
      "learning_rate": 1.9582406474155824e-05,
      "loss": 1.9497,
      "step": 5879
    },
    {
      "epoch": 0.9133980582524271,
      "grad_norm": 0.0,
      "learning_rate": 1.9512741246323574e-05,
      "loss": 2.1493,
      "step": 5880
    },
    {
      "epoch": 0.9135533980582524,
      "grad_norm": 0.0,
      "learning_rate": 1.9443197691694724e-05,
      "loss": 1.9965,
      "step": 5881
    },
    {
      "epoch": 0.9137087378640777,
      "grad_norm": 0.0,
      "learning_rate": 1.9373775827879637e-05,
      "loss": 2.0792,
      "step": 5882
    },
    {
      "epoch": 0.913864077669903,
      "grad_norm": 0.0,
      "learning_rate": 1.9304475672458023e-05,
      "loss": 2.1507,
      "step": 5883
    },
    {
      "epoch": 0.9140194174757281,
      "grad_norm": 0.0,
      "learning_rate": 1.923529724297862e-05,
      "loss": 2.1859,
      "step": 5884
    },
    {
      "epoch": 0.9141747572815534,
      "grad_norm": 0.0,
      "learning_rate": 1.9166240556959368e-05,
      "loss": 2.0531,
      "step": 5885
    },
    {
      "epoch": 0.9143300970873787,
      "grad_norm": 0.0,
      "learning_rate": 1.9097305631887495e-05,
      "loss": 2.0552,
      "step": 5886
    },
    {
      "epoch": 0.9144854368932038,
      "grad_norm": 0.0,
      "learning_rate": 1.9028492485219206e-05,
      "loss": 2.1432,
      "step": 5887
    },
    {
      "epoch": 0.9146407766990291,
      "grad_norm": 0.0,
      "learning_rate": 1.8959801134380005e-05,
      "loss": 2.1517,
      "step": 5888
    },
    {
      "epoch": 0.9147961165048544,
      "grad_norm": 0.0,
      "learning_rate": 1.8891231596764592e-05,
      "loss": 2.1561,
      "step": 5889
    },
    {
      "epoch": 0.9149514563106796,
      "grad_norm": 0.0,
      "learning_rate": 1.8822783889736573e-05,
      "loss": 1.9848,
      "step": 5890
    },
    {
      "epoch": 0.9151067961165048,
      "grad_norm": 0.0,
      "learning_rate": 1.875445803062914e-05,
      "loss": 2.067,
      "step": 5891
    },
    {
      "epoch": 0.9152621359223301,
      "grad_norm": 0.0,
      "learning_rate": 1.868625403674412e-05,
      "loss": 2.0914,
      "step": 5892
    },
    {
      "epoch": 0.9154174757281553,
      "grad_norm": 0.0,
      "learning_rate": 1.861817192535281e-05,
      "loss": 2.0922,
      "step": 5893
    },
    {
      "epoch": 0.9155728155339806,
      "grad_norm": 0.0,
      "learning_rate": 1.855021171369564e-05,
      "loss": 2.0482,
      "step": 5894
    },
    {
      "epoch": 0.9157281553398058,
      "grad_norm": 0.0,
      "learning_rate": 1.8482373418982012e-05,
      "loss": 2.1826,
      "step": 5895
    },
    {
      "epoch": 0.915883495145631,
      "grad_norm": 0.0,
      "learning_rate": 1.8414657058390515e-05,
      "loss": 1.9245,
      "step": 5896
    },
    {
      "epoch": 0.9160388349514563,
      "grad_norm": 0.0,
      "learning_rate": 1.834706264906899e-05,
      "loss": 2.106,
      "step": 5897
    },
    {
      "epoch": 0.9161941747572816,
      "grad_norm": 0.0,
      "learning_rate": 1.8279590208134135e-05,
      "loss": 2.119,
      "step": 5898
    },
    {
      "epoch": 0.9163495145631068,
      "grad_norm": 0.0,
      "learning_rate": 1.8212239752672056e-05,
      "loss": 2.0419,
      "step": 5899
    },
    {
      "epoch": 0.916504854368932,
      "grad_norm": 0.0,
      "learning_rate": 1.8145011299737723e-05,
      "loss": 2.0996,
      "step": 5900
    },
    {
      "epoch": 0.9166601941747573,
      "grad_norm": 0.0,
      "learning_rate": 1.807790486635541e-05,
      "loss": 2.1305,
      "step": 5901
    },
    {
      "epoch": 0.9168155339805826,
      "grad_norm": 0.0,
      "learning_rate": 1.8010920469518245e-05,
      "loss": 2.0874,
      "step": 5902
    },
    {
      "epoch": 0.9169708737864077,
      "grad_norm": 0.0,
      "learning_rate": 1.7944058126188768e-05,
      "loss": 2.1045,
      "step": 5903
    },
    {
      "epoch": 0.917126213592233,
      "grad_norm": 0.0,
      "learning_rate": 1.787731785329838e-05,
      "loss": 2.1221,
      "step": 5904
    },
    {
      "epoch": 0.9172815533980583,
      "grad_norm": 0.0,
      "learning_rate": 1.7810699667747622e-05,
      "loss": 2.0481,
      "step": 5905
    },
    {
      "epoch": 0.9174368932038834,
      "grad_norm": 0.0,
      "learning_rate": 1.774420358640616e-05,
      "loss": 1.9908,
      "step": 5906
    },
    {
      "epoch": 0.9175922330097087,
      "grad_norm": 0.0,
      "learning_rate": 1.7677829626112753e-05,
      "loss": 2.0262,
      "step": 5907
    },
    {
      "epoch": 0.917747572815534,
      "grad_norm": 0.0,
      "learning_rate": 1.7611577803675116e-05,
      "loss": 2.2319,
      "step": 5908
    },
    {
      "epoch": 0.9179029126213593,
      "grad_norm": 0.0,
      "learning_rate": 1.7545448135870222e-05,
      "loss": 2.0022,
      "step": 5909
    },
    {
      "epoch": 0.9180582524271844,
      "grad_norm": 0.0,
      "learning_rate": 1.7479440639443844e-05,
      "loss": 2.0165,
      "step": 5910
    },
    {
      "epoch": 0.9182135922330097,
      "grad_norm": 0.0,
      "learning_rate": 1.7413555331111107e-05,
      "loss": 2.0087,
      "step": 5911
    },
    {
      "epoch": 0.918368932038835,
      "grad_norm": 0.0,
      "learning_rate": 1.7347792227556057e-05,
      "loss": 2.1224,
      "step": 5912
    },
    {
      "epoch": 0.9185242718446602,
      "grad_norm": 0.0,
      "learning_rate": 1.7282151345431764e-05,
      "loss": 2.1158,
      "step": 5913
    },
    {
      "epoch": 0.9186796116504854,
      "grad_norm": 0.0,
      "learning_rate": 1.721663270136048e-05,
      "loss": 2.0121,
      "step": 5914
    },
    {
      "epoch": 0.9188349514563107,
      "grad_norm": 0.0,
      "learning_rate": 1.715123631193327e-05,
      "loss": 2.0991,
      "step": 5915
    },
    {
      "epoch": 0.9189902912621359,
      "grad_norm": 0.0,
      "learning_rate": 1.7085962193710504e-05,
      "loss": 2.1382,
      "step": 5916
    },
    {
      "epoch": 0.9191456310679612,
      "grad_norm": 0.0,
      "learning_rate": 1.7020810363221394e-05,
      "loss": 2.0607,
      "step": 5917
    },
    {
      "epoch": 0.9193009708737864,
      "grad_norm": 0.0,
      "learning_rate": 1.6955780836964307e-05,
      "loss": 2.0251,
      "step": 5918
    },
    {
      "epoch": 0.9194563106796116,
      "grad_norm": 0.0,
      "learning_rate": 1.6890873631406566e-05,
      "loss": 2.0818,
      "step": 5919
    },
    {
      "epoch": 0.9196116504854369,
      "grad_norm": 0.0,
      "learning_rate": 1.6826088762984526e-05,
      "loss": 2.0715,
      "step": 5920
    },
    {
      "epoch": 0.9197669902912622,
      "grad_norm": 0.0,
      "learning_rate": 1.6761426248103563e-05,
      "loss": 1.9335,
      "step": 5921
    },
    {
      "epoch": 0.9199223300970873,
      "grad_norm": 0.0,
      "learning_rate": 1.6696886103138185e-05,
      "loss": 2.0133,
      "step": 5922
    },
    {
      "epoch": 0.9200776699029126,
      "grad_norm": 0.0,
      "learning_rate": 1.6632468344431717e-05,
      "loss": 2.0973,
      "step": 5923
    },
    {
      "epoch": 0.9202330097087379,
      "grad_norm": 0.0,
      "learning_rate": 1.656817298829666e-05,
      "loss": 2.0551,
      "step": 5924
    },
    {
      "epoch": 0.920388349514563,
      "grad_norm": 0.0,
      "learning_rate": 1.650400005101438e-05,
      "loss": 2.1094,
      "step": 5925
    },
    {
      "epoch": 0.9205436893203883,
      "grad_norm": 0.0,
      "learning_rate": 1.643994954883532e-05,
      "loss": 2.0975,
      "step": 5926
    },
    {
      "epoch": 0.9206990291262136,
      "grad_norm": 0.0,
      "learning_rate": 1.637602149797901e-05,
      "loss": 2.067,
      "step": 5927
    },
    {
      "epoch": 0.9208543689320389,
      "grad_norm": 0.0,
      "learning_rate": 1.6312215914633722e-05,
      "loss": 2.0699,
      "step": 5928
    },
    {
      "epoch": 0.921009708737864,
      "grad_norm": 0.0,
      "learning_rate": 1.6248532814956975e-05,
      "loss": 2.1421,
      "step": 5929
    },
    {
      "epoch": 0.9211650485436893,
      "grad_norm": 0.0,
      "learning_rate": 1.61849722150752e-05,
      "loss": 2.1216,
      "step": 5930
    },
    {
      "epoch": 0.9213203883495146,
      "grad_norm": 0.0,
      "learning_rate": 1.612153413108358e-05,
      "loss": 2.1047,
      "step": 5931
    },
    {
      "epoch": 0.9214757281553398,
      "grad_norm": 0.0,
      "learning_rate": 1.605821857904666e-05,
      "loss": 2.0924,
      "step": 5932
    },
    {
      "epoch": 0.921631067961165,
      "grad_norm": 0.0,
      "learning_rate": 1.5995025574997657e-05,
      "loss": 2.0681,
      "step": 5933
    },
    {
      "epoch": 0.9217864077669903,
      "grad_norm": 0.0,
      "learning_rate": 1.5931955134938948e-05,
      "loss": 2.0141,
      "step": 5934
    },
    {
      "epoch": 0.9219417475728156,
      "grad_norm": 0.0,
      "learning_rate": 1.586900727484164e-05,
      "loss": 1.985,
      "step": 5935
    },
    {
      "epoch": 0.9220970873786408,
      "grad_norm": 0.0,
      "learning_rate": 1.5806182010645986e-05,
      "loss": 2.1133,
      "step": 5936
    },
    {
      "epoch": 0.922252427184466,
      "grad_norm": 0.0,
      "learning_rate": 1.5743479358261315e-05,
      "loss": 2.2109,
      "step": 5937
    },
    {
      "epoch": 0.9224077669902913,
      "grad_norm": 0.0,
      "learning_rate": 1.5680899333565533e-05,
      "loss": 2.121,
      "step": 5938
    },
    {
      "epoch": 0.9225631067961165,
      "grad_norm": 0.0,
      "learning_rate": 1.561844195240586e-05,
      "loss": 1.9642,
      "step": 5939
    },
    {
      "epoch": 0.9227184466019418,
      "grad_norm": 0.0,
      "learning_rate": 1.5556107230598138e-05,
      "loss": 1.9711,
      "step": 5940
    },
    {
      "epoch": 0.922873786407767,
      "grad_norm": 0.0,
      "learning_rate": 1.5493895183927466e-05,
      "loss": 1.9191,
      "step": 5941
    },
    {
      "epoch": 0.9230291262135922,
      "grad_norm": 0.0,
      "learning_rate": 1.5431805828147684e-05,
      "loss": 2.1163,
      "step": 5942
    },
    {
      "epoch": 0.9231844660194175,
      "grad_norm": 0.0,
      "learning_rate": 1.5369839178981548e-05,
      "loss": 2.0142,
      "step": 5943
    },
    {
      "epoch": 0.9233398058252427,
      "grad_norm": 0.0,
      "learning_rate": 1.530799525212079e-05,
      "loss": 2.0179,
      "step": 5944
    },
    {
      "epoch": 0.9234951456310679,
      "grad_norm": 0.0,
      "learning_rate": 1.5246274063226206e-05,
      "loss": 2.1212,
      "step": 5945
    },
    {
      "epoch": 0.9236504854368932,
      "grad_norm": 0.0,
      "learning_rate": 1.5184675627927191e-05,
      "loss": 1.9188,
      "step": 5946
    },
    {
      "epoch": 0.9238058252427185,
      "grad_norm": 0.0,
      "learning_rate": 1.5123199961822432e-05,
      "loss": 2.0685,
      "step": 5947
    },
    {
      "epoch": 0.9239611650485436,
      "grad_norm": 0.0,
      "learning_rate": 1.5061847080479196e-05,
      "loss": 2.1534,
      "step": 5948
    },
    {
      "epoch": 0.9241165048543689,
      "grad_norm": 0.0,
      "learning_rate": 1.5000616999433892e-05,
      "loss": 2.0857,
      "step": 5949
    },
    {
      "epoch": 0.9242718446601942,
      "grad_norm": 0.0,
      "learning_rate": 1.4939509734191614e-05,
      "loss": 2.138,
      "step": 5950
    },
    {
      "epoch": 0.9244271844660195,
      "grad_norm": 0.0,
      "learning_rate": 1.4878525300226598e-05,
      "loss": 2.1126,
      "step": 5951
    },
    {
      "epoch": 0.9245825242718446,
      "grad_norm": 0.0,
      "learning_rate": 1.4817663712981876e-05,
      "loss": 2.0408,
      "step": 5952
    },
    {
      "epoch": 0.9247378640776699,
      "grad_norm": 0.0,
      "learning_rate": 1.4756924987869236e-05,
      "loss": 2.111,
      "step": 5953
    },
    {
      "epoch": 0.9248932038834952,
      "grad_norm": 0.0,
      "learning_rate": 1.4696309140269481e-05,
      "loss": 2.1426,
      "step": 5954
    },
    {
      "epoch": 0.9250485436893204,
      "grad_norm": 0.0,
      "learning_rate": 1.4635816185532447e-05,
      "loss": 1.9835,
      "step": 5955
    },
    {
      "epoch": 0.9252038834951456,
      "grad_norm": 0.0,
      "learning_rate": 1.4575446138976489e-05,
      "loss": 2.1473,
      "step": 5956
    },
    {
      "epoch": 0.9253592233009709,
      "grad_norm": 0.0,
      "learning_rate": 1.4515199015889213e-05,
      "loss": 2.0651,
      "step": 5957
    },
    {
      "epoch": 0.9255145631067961,
      "grad_norm": 0.0,
      "learning_rate": 1.4455074831526805e-05,
      "loss": 1.9882,
      "step": 5958
    },
    {
      "epoch": 0.9256699029126214,
      "grad_norm": 0.0,
      "learning_rate": 1.4395073601114473e-05,
      "loss": 2.0752,
      "step": 5959
    },
    {
      "epoch": 0.9258252427184466,
      "grad_norm": 0.0,
      "learning_rate": 1.4335195339846286e-05,
      "loss": 2.0886,
      "step": 5960
    },
    {
      "epoch": 0.9259805825242718,
      "grad_norm": 0.0,
      "learning_rate": 1.427544006288506e-05,
      "loss": 2.0294,
      "step": 5961
    },
    {
      "epoch": 0.9261359223300971,
      "grad_norm": 0.0,
      "learning_rate": 1.4215807785362689e-05,
      "loss": 2.0643,
      "step": 5962
    },
    {
      "epoch": 0.9262912621359223,
      "grad_norm": 0.0,
      "learning_rate": 1.4156298522379652e-05,
      "loss": 2.0104,
      "step": 5963
    },
    {
      "epoch": 0.9264466019417475,
      "grad_norm": 0.0,
      "learning_rate": 1.409691228900556e-05,
      "loss": 2.1027,
      "step": 5964
    },
    {
      "epoch": 0.9266019417475728,
      "grad_norm": 0.0,
      "learning_rate": 1.4037649100278548e-05,
      "loss": 2.0941,
      "step": 5965
    },
    {
      "epoch": 0.9267572815533981,
      "grad_norm": 0.0,
      "learning_rate": 1.3978508971205784e-05,
      "loss": 2.0846,
      "step": 5966
    },
    {
      "epoch": 0.9269126213592233,
      "grad_norm": 0.0,
      "learning_rate": 1.3919491916763339e-05,
      "loss": 1.9995,
      "step": 5967
    },
    {
      "epoch": 0.9270679611650485,
      "grad_norm": 0.0,
      "learning_rate": 1.3860597951895981e-05,
      "loss": 2.1804,
      "step": 5968
    },
    {
      "epoch": 0.9272233009708738,
      "grad_norm": 0.0,
      "learning_rate": 1.3801827091517339e-05,
      "loss": 2.0096,
      "step": 5969
    },
    {
      "epoch": 0.9273786407766991,
      "grad_norm": 0.0,
      "learning_rate": 1.3743179350509948e-05,
      "loss": 2.0662,
      "step": 5970
    },
    {
      "epoch": 0.9275339805825242,
      "grad_norm": 0.0,
      "learning_rate": 1.3684654743725045e-05,
      "loss": 2.0983,
      "step": 5971
    },
    {
      "epoch": 0.9276893203883495,
      "grad_norm": 0.0,
      "learning_rate": 1.362625328598277e-05,
      "loss": 2.0499,
      "step": 5972
    },
    {
      "epoch": 0.9278446601941748,
      "grad_norm": 0.0,
      "learning_rate": 1.3567974992072018e-05,
      "loss": 2.0106,
      "step": 5973
    },
    {
      "epoch": 0.928,
      "grad_norm": 0.0,
      "learning_rate": 1.3509819876750596e-05,
      "loss": 2.0546,
      "step": 5974
    },
    {
      "epoch": 0.9281553398058252,
      "grad_norm": 0.0,
      "learning_rate": 1.345178795474511e-05,
      "loss": 2.0776,
      "step": 5975
    },
    {
      "epoch": 0.9283106796116505,
      "grad_norm": 0.0,
      "learning_rate": 1.3393879240750696e-05,
      "loss": 2.025,
      "step": 5976
    },
    {
      "epoch": 0.9284660194174758,
      "grad_norm": 0.0,
      "learning_rate": 1.3336093749431621e-05,
      "loss": 2.1504,
      "step": 5977
    },
    {
      "epoch": 0.928621359223301,
      "grad_norm": 0.0,
      "learning_rate": 1.3278431495420907e-05,
      "loss": 2.0001,
      "step": 5978
    },
    {
      "epoch": 0.9287766990291262,
      "grad_norm": 0.0,
      "learning_rate": 1.3220892493320202e-05,
      "loss": 2.0747,
      "step": 5979
    },
    {
      "epoch": 0.9289320388349515,
      "grad_norm": 0.0,
      "learning_rate": 1.3163476757700133e-05,
      "loss": 1.9339,
      "step": 5980
    },
    {
      "epoch": 0.9290873786407767,
      "grad_norm": 0.0,
      "learning_rate": 1.31061843030999e-05,
      "loss": 2.0776,
      "step": 5981
    },
    {
      "epoch": 0.9292427184466019,
      "grad_norm": 0.0,
      "learning_rate": 1.3049015144027676e-05,
      "loss": 2.2456,
      "step": 5982
    },
    {
      "epoch": 0.9293980582524272,
      "grad_norm": 0.0,
      "learning_rate": 1.2991969294960271e-05,
      "loss": 2.0662,
      "step": 5983
    },
    {
      "epoch": 0.9295533980582524,
      "grad_norm": 0.0,
      "learning_rate": 1.2935046770343406e-05,
      "loss": 1.9506,
      "step": 5984
    },
    {
      "epoch": 0.9297087378640777,
      "grad_norm": 0.0,
      "learning_rate": 1.2878247584591496e-05,
      "loss": 2.1597,
      "step": 5985
    },
    {
      "epoch": 0.9298640776699029,
      "grad_norm": 0.0,
      "learning_rate": 1.2821571752087701e-05,
      "loss": 2.1304,
      "step": 5986
    },
    {
      "epoch": 0.9300194174757281,
      "grad_norm": 0.0,
      "learning_rate": 1.2765019287184043e-05,
      "loss": 2.1434,
      "step": 5987
    },
    {
      "epoch": 0.9301747572815534,
      "grad_norm": 0.0,
      "learning_rate": 1.270859020420112e-05,
      "loss": 1.9974,
      "step": 5988
    },
    {
      "epoch": 0.9303300970873787,
      "grad_norm": 0.0,
      "learning_rate": 1.2652284517428447e-05,
      "loss": 2.0735,
      "step": 5989
    },
    {
      "epoch": 0.9304854368932038,
      "grad_norm": 0.0,
      "learning_rate": 1.2596102241124285e-05,
      "loss": 2.0193,
      "step": 5990
    },
    {
      "epoch": 0.9306407766990291,
      "grad_norm": 0.0,
      "learning_rate": 1.2540043389515532e-05,
      "loss": 2.0683,
      "step": 5991
    },
    {
      "epoch": 0.9307961165048544,
      "grad_norm": 0.0,
      "learning_rate": 1.2484107976797943e-05,
      "loss": 2.0267,
      "step": 5992
    },
    {
      "epoch": 0.9309514563106797,
      "grad_norm": 0.0,
      "learning_rate": 1.242829601713602e-05,
      "loss": 2.056,
      "step": 5993
    },
    {
      "epoch": 0.9311067961165048,
      "grad_norm": 0.0,
      "learning_rate": 1.237260752466285e-05,
      "loss": 2.1591,
      "step": 5994
    },
    {
      "epoch": 0.9312621359223301,
      "grad_norm": 0.0,
      "learning_rate": 1.231704251348048e-05,
      "loss": 2.124,
      "step": 5995
    },
    {
      "epoch": 0.9314174757281554,
      "grad_norm": 0.0,
      "learning_rate": 1.2261600997659495e-05,
      "loss": 2.0649,
      "step": 5996
    },
    {
      "epoch": 0.9315728155339806,
      "grad_norm": 0.0,
      "learning_rate": 1.2206282991239325e-05,
      "loss": 2.0825,
      "step": 5997
    },
    {
      "epoch": 0.9317281553398058,
      "grad_norm": 0.0,
      "learning_rate": 1.2151088508228047e-05,
      "loss": 2.132,
      "step": 5998
    },
    {
      "epoch": 0.9318834951456311,
      "grad_norm": 0.0,
      "learning_rate": 1.2096017562602479e-05,
      "loss": 2.1454,
      "step": 5999
    },
    {
      "epoch": 0.9320388349514563,
      "grad_norm": 0.0,
      "learning_rate": 1.2041070168308188e-05,
      "loss": 2.0947,
      "step": 6000
    },
    {
      "epoch": 0.9321941747572815,
      "grad_norm": 0.0,
      "learning_rate": 1.1986246339259433e-05,
      "loss": 1.9726,
      "step": 6001
    },
    {
      "epoch": 0.9323495145631068,
      "grad_norm": 0.0,
      "learning_rate": 1.1931546089339218e-05,
      "loss": 2.1855,
      "step": 6002
    },
    {
      "epoch": 0.932504854368932,
      "grad_norm": 0.0,
      "learning_rate": 1.1876969432399188e-05,
      "loss": 2.0237,
      "step": 6003
    },
    {
      "epoch": 0.9326601941747573,
      "grad_norm": 0.0,
      "learning_rate": 1.182251638225973e-05,
      "loss": 2.0862,
      "step": 6004
    },
    {
      "epoch": 0.9328155339805825,
      "grad_norm": 0.0,
      "learning_rate": 1.176818695270998e-05,
      "loss": 2.1627,
      "step": 6005
    },
    {
      "epoch": 0.9329708737864078,
      "grad_norm": 0.0,
      "learning_rate": 1.1713981157507602e-05,
      "loss": 1.9383,
      "step": 6006
    },
    {
      "epoch": 0.933126213592233,
      "grad_norm": 0.0,
      "learning_rate": 1.165989901037906e-05,
      "loss": 2.0751,
      "step": 6007
    },
    {
      "epoch": 0.9332815533980583,
      "grad_norm": 0.0,
      "learning_rate": 1.1605940525019675e-05,
      "loss": 1.9677,
      "step": 6008
    },
    {
      "epoch": 0.9334368932038835,
      "grad_norm": 0.0,
      "learning_rate": 1.155210571509313e-05,
      "loss": 1.9924,
      "step": 6009
    },
    {
      "epoch": 0.9335922330097087,
      "grad_norm": 0.0,
      "learning_rate": 1.1498394594232076e-05,
      "loss": 2.0638,
      "step": 6010
    },
    {
      "epoch": 0.933747572815534,
      "grad_norm": 0.0,
      "learning_rate": 1.1444807176037575e-05,
      "loss": 2.0735,
      "step": 6011
    },
    {
      "epoch": 0.9339029126213593,
      "grad_norm": 0.0,
      "learning_rate": 1.1391343474079551e-05,
      "loss": 2.0923,
      "step": 6012
    },
    {
      "epoch": 0.9340582524271844,
      "grad_norm": 0.0,
      "learning_rate": 1.1338003501896676e-05,
      "loss": 2.0234,
      "step": 6013
    },
    {
      "epoch": 0.9342135922330097,
      "grad_norm": 0.0,
      "learning_rate": 1.1284787272995922e-05,
      "loss": 2.1014,
      "step": 6014
    },
    {
      "epoch": 0.934368932038835,
      "grad_norm": 0.0,
      "learning_rate": 1.1231694800853399e-05,
      "loss": 2.0884,
      "step": 6015
    },
    {
      "epoch": 0.9345242718446602,
      "grad_norm": 0.0,
      "learning_rate": 1.1178726098913517e-05,
      "loss": 2.0583,
      "step": 6016
    },
    {
      "epoch": 0.9346796116504854,
      "grad_norm": 0.0,
      "learning_rate": 1.1125881180589548e-05,
      "loss": 2.0374,
      "step": 6017
    },
    {
      "epoch": 0.9348349514563107,
      "grad_norm": 0.0,
      "learning_rate": 1.1073160059263287e-05,
      "loss": 2.0978,
      "step": 6018
    },
    {
      "epoch": 0.934990291262136,
      "grad_norm": 0.0,
      "learning_rate": 1.1020562748285279e-05,
      "loss": 2.0199,
      "step": 6019
    },
    {
      "epoch": 0.9351456310679611,
      "grad_norm": 0.0,
      "learning_rate": 1.09680892609747e-05,
      "loss": 2.116,
      "step": 6020
    },
    {
      "epoch": 0.9353009708737864,
      "grad_norm": 0.0,
      "learning_rate": 1.0915739610619312e-05,
      "loss": 2.0197,
      "step": 6021
    },
    {
      "epoch": 0.9354563106796117,
      "grad_norm": 0.0,
      "learning_rate": 1.0863513810475511e-05,
      "loss": 1.9716,
      "step": 6022
    },
    {
      "epoch": 0.9356116504854369,
      "grad_norm": 0.0,
      "learning_rate": 1.0811411873768439e-05,
      "loss": 2.0046,
      "step": 6023
    },
    {
      "epoch": 0.9357669902912621,
      "grad_norm": 0.0,
      "learning_rate": 1.0759433813691766e-05,
      "loss": 2.0328,
      "step": 6024
    },
    {
      "epoch": 0.9359223300970874,
      "grad_norm": 0.0,
      "learning_rate": 1.070757964340785e-05,
      "loss": 1.9857,
      "step": 6025
    },
    {
      "epoch": 0.9360776699029126,
      "grad_norm": 0.0,
      "learning_rate": 1.0655849376047688e-05,
      "loss": 2.0334,
      "step": 6026
    },
    {
      "epoch": 0.9362330097087379,
      "grad_norm": 0.0,
      "learning_rate": 1.0604243024710802e-05,
      "loss": 2.0563,
      "step": 6027
    },
    {
      "epoch": 0.9363883495145631,
      "grad_norm": 0.0,
      "learning_rate": 1.0552760602465517e-05,
      "loss": 1.942,
      "step": 6028
    },
    {
      "epoch": 0.9365436893203883,
      "grad_norm": 0.0,
      "learning_rate": 1.0501402122348513e-05,
      "loss": 2.0178,
      "step": 6029
    },
    {
      "epoch": 0.9366990291262136,
      "grad_norm": 0.0,
      "learning_rate": 1.045016759736539e-05,
      "loss": 2.0819,
      "step": 6030
    },
    {
      "epoch": 0.9368543689320389,
      "grad_norm": 0.0,
      "learning_rate": 1.03990570404901e-05,
      "loss": 2.0947,
      "step": 6031
    },
    {
      "epoch": 0.937009708737864,
      "grad_norm": 0.0,
      "learning_rate": 1.0348070464665349e-05,
      "loss": 2.0711,
      "step": 6032
    },
    {
      "epoch": 0.9371650485436893,
      "grad_norm": 0.0,
      "learning_rate": 1.0297207882802418e-05,
      "loss": 2.0309,
      "step": 6033
    },
    {
      "epoch": 0.9373203883495146,
      "grad_norm": 0.0,
      "learning_rate": 1.0246469307781115e-05,
      "loss": 1.9794,
      "step": 6034
    },
    {
      "epoch": 0.9374757281553399,
      "grad_norm": 0.0,
      "learning_rate": 1.019585475244994e-05,
      "loss": 2.0511,
      "step": 6035
    },
    {
      "epoch": 0.937631067961165,
      "grad_norm": 0.0,
      "learning_rate": 1.0145364229625974e-05,
      "loss": 2.0027,
      "step": 6036
    },
    {
      "epoch": 0.9377864077669903,
      "grad_norm": 0.0,
      "learning_rate": 1.0094997752094881e-05,
      "loss": 2.0566,
      "step": 6037
    },
    {
      "epoch": 0.9379417475728156,
      "grad_norm": 0.0,
      "learning_rate": 1.0044755332610844e-05,
      "loss": 2.0859,
      "step": 6038
    },
    {
      "epoch": 0.9380970873786407,
      "grad_norm": 0.0,
      "learning_rate": 9.994636983896744e-06,
      "loss": 2.0955,
      "step": 6039
    },
    {
      "epoch": 0.938252427184466,
      "grad_norm": 0.0,
      "learning_rate": 9.944642718643926e-06,
      "loss": 2.1054,
      "step": 6040
    },
    {
      "epoch": 0.9384077669902913,
      "grad_norm": 0.0,
      "learning_rate": 9.894772549512488e-06,
      "loss": 2.2119,
      "step": 6041
    },
    {
      "epoch": 0.9385631067961165,
      "grad_norm": 0.0,
      "learning_rate": 9.845026489130882e-06,
      "loss": 2.1296,
      "step": 6042
    },
    {
      "epoch": 0.9387184466019417,
      "grad_norm": 0.0,
      "learning_rate": 9.795404550096364e-06,
      "loss": 2.0101,
      "step": 6043
    },
    {
      "epoch": 0.938873786407767,
      "grad_norm": 0.0,
      "learning_rate": 9.745906744974497e-06,
      "loss": 2.0066,
      "step": 6044
    },
    {
      "epoch": 0.9390291262135922,
      "grad_norm": 0.0,
      "learning_rate": 9.696533086299641e-06,
      "loss": 2.0869,
      "step": 6045
    },
    {
      "epoch": 0.9391844660194175,
      "grad_norm": 0.0,
      "learning_rate": 9.647283586574573e-06,
      "loss": 1.9856,
      "step": 6046
    },
    {
      "epoch": 0.9393398058252427,
      "grad_norm": 0.0,
      "learning_rate": 9.598158258270706e-06,
      "loss": 2.131,
      "step": 6047
    },
    {
      "epoch": 0.939495145631068,
      "grad_norm": 0.0,
      "learning_rate": 9.549157113828034e-06,
      "loss": 2.1629,
      "step": 6048
    },
    {
      "epoch": 0.9396504854368932,
      "grad_norm": 0.0,
      "learning_rate": 9.500280165654962e-06,
      "loss": 2.0016,
      "step": 6049
    },
    {
      "epoch": 0.9398058252427185,
      "grad_norm": 0.0,
      "learning_rate": 9.451527426128593e-06,
      "loss": 2.0815,
      "step": 6050
    },
    {
      "epoch": 0.9399611650485437,
      "grad_norm": 0.0,
      "learning_rate": 9.402898907594548e-06,
      "loss": 1.9949,
      "step": 6051
    },
    {
      "epoch": 0.9401165048543689,
      "grad_norm": 0.0,
      "learning_rate": 9.354394622366924e-06,
      "loss": 1.985,
      "step": 6052
    },
    {
      "epoch": 0.9402718446601942,
      "grad_norm": 0.0,
      "learning_rate": 9.306014582728451e-06,
      "loss": 2.0345,
      "step": 6053
    },
    {
      "epoch": 0.9404271844660195,
      "grad_norm": 0.0,
      "learning_rate": 9.257758800930328e-06,
      "loss": 2.0928,
      "step": 6054
    },
    {
      "epoch": 0.9405825242718446,
      "grad_norm": 0.0,
      "learning_rate": 9.209627289192279e-06,
      "loss": 1.9728,
      "step": 6055
    },
    {
      "epoch": 0.9407378640776699,
      "grad_norm": 0.0,
      "learning_rate": 9.161620059702613e-06,
      "loss": 2.191,
      "step": 6056
    },
    {
      "epoch": 0.9408932038834952,
      "grad_norm": 0.0,
      "learning_rate": 9.113737124618105e-06,
      "loss": 2.051,
      "step": 6057
    },
    {
      "epoch": 0.9410485436893203,
      "grad_norm": 0.0,
      "learning_rate": 9.06597849606411e-06,
      "loss": 2.0039,
      "step": 6058
    },
    {
      "epoch": 0.9412038834951456,
      "grad_norm": 0.0,
      "learning_rate": 9.018344186134564e-06,
      "loss": 2.0626,
      "step": 6059
    },
    {
      "epoch": 0.9413592233009709,
      "grad_norm": 0.0,
      "learning_rate": 8.97083420689171e-06,
      "loss": 2.0854,
      "step": 6060
    },
    {
      "epoch": 0.9415145631067962,
      "grad_norm": 0.0,
      "learning_rate": 8.923448570366587e-06,
      "loss": 2.0633,
      "step": 6061
    },
    {
      "epoch": 0.9416699029126213,
      "grad_norm": 0.0,
      "learning_rate": 8.87618728855849e-06,
      "loss": 2.0435,
      "step": 6062
    },
    {
      "epoch": 0.9418252427184466,
      "grad_norm": 0.0,
      "learning_rate": 8.829050373435344e-06,
      "loss": 2.1282,
      "step": 6063
    },
    {
      "epoch": 0.9419805825242719,
      "grad_norm": 0.0,
      "learning_rate": 8.782037836933654e-06,
      "loss": 2.0886,
      "step": 6064
    },
    {
      "epoch": 0.9421359223300971,
      "grad_norm": 0.0,
      "learning_rate": 8.735149690958288e-06,
      "loss": 2.0289,
      "step": 6065
    },
    {
      "epoch": 0.9422912621359223,
      "grad_norm": 0.0,
      "learning_rate": 8.688385947382693e-06,
      "loss": 1.9256,
      "step": 6066
    },
    {
      "epoch": 0.9424466019417476,
      "grad_norm": 0.0,
      "learning_rate": 8.641746618048784e-06,
      "loss": 2.0806,
      "step": 6067
    },
    {
      "epoch": 0.9426019417475728,
      "grad_norm": 0.0,
      "learning_rate": 8.595231714766948e-06,
      "loss": 2.1671,
      "step": 6068
    },
    {
      "epoch": 0.9427572815533981,
      "grad_norm": 0.0,
      "learning_rate": 8.548841249316153e-06,
      "loss": 2.0393,
      "step": 6069
    },
    {
      "epoch": 0.9429126213592233,
      "grad_norm": 0.0,
      "learning_rate": 8.502575233443777e-06,
      "loss": 1.9746,
      "step": 6070
    },
    {
      "epoch": 0.9430679611650485,
      "grad_norm": 0.0,
      "learning_rate": 8.456433678865727e-06,
      "loss": 2.1057,
      "step": 6071
    },
    {
      "epoch": 0.9432233009708738,
      "grad_norm": 0.0,
      "learning_rate": 8.410416597266269e-06,
      "loss": 1.9739,
      "step": 6072
    },
    {
      "epoch": 0.9433786407766991,
      "grad_norm": 0.0,
      "learning_rate": 8.364524000298413e-06,
      "loss": 2.1058,
      "step": 6073
    },
    {
      "epoch": 0.9435339805825242,
      "grad_norm": 0.0,
      "learning_rate": 8.31875589958342e-06,
      "loss": 1.9961,
      "step": 6074
    },
    {
      "epoch": 0.9436893203883495,
      "grad_norm": 0.0,
      "learning_rate": 8.273112306711016e-06,
      "loss": 1.9969,
      "step": 6075
    },
    {
      "epoch": 0.9438446601941748,
      "grad_norm": 0.0,
      "learning_rate": 8.227593233239627e-06,
      "loss": 2.0027,
      "step": 6076
    },
    {
      "epoch": 0.944,
      "grad_norm": 0.0,
      "learning_rate": 8.182198690695807e-06,
      "loss": 2.185,
      "step": 6077
    },
    {
      "epoch": 0.9441553398058252,
      "grad_norm": 0.0,
      "learning_rate": 8.136928690574918e-06,
      "loss": 2.0659,
      "step": 6078
    },
    {
      "epoch": 0.9443106796116505,
      "grad_norm": 0.0,
      "learning_rate": 8.091783244340622e-06,
      "loss": 1.9934,
      "step": 6079
    },
    {
      "epoch": 0.9444660194174758,
      "grad_norm": 0.0,
      "learning_rate": 8.046762363424887e-06,
      "loss": 2.0511,
      "step": 6080
    },
    {
      "epoch": 0.9446213592233009,
      "grad_norm": 0.0,
      "learning_rate": 8.001866059228424e-06,
      "loss": 1.9781,
      "step": 6081
    },
    {
      "epoch": 0.9447766990291262,
      "grad_norm": 0.0,
      "learning_rate": 7.957094343120252e-06,
      "loss": 2.1271,
      "step": 6082
    },
    {
      "epoch": 0.9449320388349515,
      "grad_norm": 0.0,
      "learning_rate": 7.912447226437802e-06,
      "loss": 2.09,
      "step": 6083
    },
    {
      "epoch": 0.9450873786407767,
      "grad_norm": 0.0,
      "learning_rate": 7.86792472048714e-06,
      "loss": 2.0955,
      "step": 6084
    },
    {
      "epoch": 0.9452427184466019,
      "grad_norm": 0.0,
      "learning_rate": 7.823526836542471e-06,
      "loss": 2.1205,
      "step": 6085
    },
    {
      "epoch": 0.9453980582524272,
      "grad_norm": 0.0,
      "learning_rate": 7.779253585846802e-06,
      "loss": 1.9973,
      "step": 6086
    },
    {
      "epoch": 0.9455533980582524,
      "grad_norm": 0.0,
      "learning_rate": 7.735104979611218e-06,
      "loss": 2.0779,
      "step": 6087
    },
    {
      "epoch": 0.9457087378640777,
      "grad_norm": 0.0,
      "learning_rate": 7.691081029015557e-06,
      "loss": 1.9962,
      "step": 6088
    },
    {
      "epoch": 0.9458640776699029,
      "grad_norm": 0.0,
      "learning_rate": 7.647181745207898e-06,
      "loss": 2.0225,
      "step": 6089
    },
    {
      "epoch": 0.9460194174757282,
      "grad_norm": 0.0,
      "learning_rate": 7.603407139304741e-06,
      "loss": 1.972,
      "step": 6090
    },
    {
      "epoch": 0.9461747572815534,
      "grad_norm": 0.0,
      "learning_rate": 7.5597572223911616e-06,
      "loss": 2.1584,
      "step": 6091
    },
    {
      "epoch": 0.9463300970873786,
      "grad_norm": 0.0,
      "learning_rate": 7.516232005520596e-06,
      "loss": 2.0524,
      "step": 6092
    },
    {
      "epoch": 0.9464854368932039,
      "grad_norm": 0.0,
      "learning_rate": 7.472831499714727e-06,
      "loss": 2.0147,
      "step": 6093
    },
    {
      "epoch": 0.9466407766990291,
      "grad_norm": 0.0,
      "learning_rate": 7.429555715963987e-06,
      "loss": 2.0602,
      "step": 6094
    },
    {
      "epoch": 0.9467961165048544,
      "grad_norm": 0.0,
      "learning_rate": 7.386404665226998e-06,
      "loss": 2.063,
      "step": 6095
    },
    {
      "epoch": 0.9469514563106796,
      "grad_norm": 0.0,
      "learning_rate": 7.343378358430741e-06,
      "loss": 2.0342,
      "step": 6096
    },
    {
      "epoch": 0.9471067961165048,
      "grad_norm": 0.0,
      "learning_rate": 7.300476806470891e-06,
      "loss": 2.0533,
      "step": 6097
    },
    {
      "epoch": 0.9472621359223301,
      "grad_norm": 0.0,
      "learning_rate": 7.257700020211256e-06,
      "loss": 2.0433,
      "step": 6098
    },
    {
      "epoch": 0.9474174757281554,
      "grad_norm": 0.0,
      "learning_rate": 7.215048010484171e-06,
      "loss": 1.9974,
      "step": 6099
    },
    {
      "epoch": 0.9475728155339805,
      "grad_norm": 0.0,
      "learning_rate": 7.172520788090275e-06,
      "loss": 2.0347,
      "step": 6100
    },
    {
      "epoch": 0.9477281553398058,
      "grad_norm": 0.0,
      "learning_rate": 7.130118363798843e-06,
      "loss": 2.1851,
      "step": 6101
    },
    {
      "epoch": 0.9478834951456311,
      "grad_norm": 0.0,
      "learning_rate": 7.0878407483472294e-06,
      "loss": 2.0819,
      "step": 6102
    },
    {
      "epoch": 0.9480388349514564,
      "grad_norm": 0.0,
      "learning_rate": 7.045687952441427e-06,
      "loss": 2.0375,
      "step": 6103
    },
    {
      "epoch": 0.9481941747572815,
      "grad_norm": 0.0,
      "learning_rate": 7.003659986755784e-06,
      "loss": 2.0675,
      "step": 6104
    },
    {
      "epoch": 0.9483495145631068,
      "grad_norm": 0.0,
      "learning_rate": 6.961756861932844e-06,
      "loss": 2.1308,
      "step": 6105
    },
    {
      "epoch": 0.9485048543689321,
      "grad_norm": 0.0,
      "learning_rate": 6.919978588583787e-06,
      "loss": 2.1863,
      "step": 6106
    },
    {
      "epoch": 0.9486601941747573,
      "grad_norm": 0.0,
      "learning_rate": 6.878325177288147e-06,
      "loss": 2.0544,
      "step": 6107
    },
    {
      "epoch": 0.9488155339805825,
      "grad_norm": 0.0,
      "learning_rate": 6.836796638593601e-06,
      "loss": 2.0292,
      "step": 6108
    },
    {
      "epoch": 0.9489708737864078,
      "grad_norm": 0.0,
      "learning_rate": 6.795392983016458e-06,
      "loss": 2.2112,
      "step": 6109
    },
    {
      "epoch": 0.949126213592233,
      "grad_norm": 0.0,
      "learning_rate": 6.754114221041274e-06,
      "loss": 2.1613,
      "step": 6110
    },
    {
      "epoch": 0.9492815533980582,
      "grad_norm": 0.0,
      "learning_rate": 6.7129603631210785e-06,
      "loss": 1.9861,
      "step": 6111
    },
    {
      "epoch": 0.9494368932038835,
      "grad_norm": 0.0,
      "learning_rate": 6.671931419677257e-06,
      "loss": 2.0665,
      "step": 6112
    },
    {
      "epoch": 0.9495922330097087,
      "grad_norm": 0.0,
      "learning_rate": 6.631027401099332e-06,
      "loss": 2.0596,
      "step": 6113
    },
    {
      "epoch": 0.949747572815534,
      "grad_norm": 0.0,
      "learning_rate": 6.590248317745573e-06,
      "loss": 2.0116,
      "step": 6114
    },
    {
      "epoch": 0.9499029126213592,
      "grad_norm": 0.0,
      "learning_rate": 6.549594179942275e-06,
      "loss": 1.9963,
      "step": 6115
    },
    {
      "epoch": 0.9500582524271844,
      "grad_norm": 0.0,
      "learning_rate": 6.509064997984315e-06,
      "loss": 2.0438,
      "step": 6116
    },
    {
      "epoch": 0.9502135922330097,
      "grad_norm": 0.0,
      "learning_rate": 6.468660782134872e-06,
      "loss": 2.0114,
      "step": 6117
    },
    {
      "epoch": 0.950368932038835,
      "grad_norm": 0.0,
      "learning_rate": 6.428381542625372e-06,
      "loss": 2.0227,
      "step": 6118
    },
    {
      "epoch": 0.9505242718446602,
      "grad_norm": 0.0,
      "learning_rate": 6.388227289655768e-06,
      "loss": 2.0405,
      "step": 6119
    },
    {
      "epoch": 0.9506796116504854,
      "grad_norm": 0.0,
      "learning_rate": 6.348198033394148e-06,
      "loss": 2.0308,
      "step": 6120
    },
    {
      "epoch": 0.9508349514563107,
      "grad_norm": 0.0,
      "learning_rate": 6.3082937839771815e-06,
      "loss": 2.0664,
      "step": 6121
    },
    {
      "epoch": 0.950990291262136,
      "grad_norm": 0.0,
      "learning_rate": 6.268514551509785e-06,
      "loss": 2.0826,
      "step": 6122
    },
    {
      "epoch": 0.9511456310679611,
      "grad_norm": 0.0,
      "learning_rate": 6.228860346065124e-06,
      "loss": 2.0648,
      "step": 6123
    },
    {
      "epoch": 0.9513009708737864,
      "grad_norm": 0.0,
      "learning_rate": 6.189331177684887e-06,
      "loss": 2.0488,
      "step": 6124
    },
    {
      "epoch": 0.9514563106796117,
      "grad_norm": 0.0,
      "learning_rate": 6.149927056378901e-06,
      "loss": 2.0672,
      "step": 6125
    },
    {
      "epoch": 0.9516116504854369,
      "grad_norm": 0.0,
      "learning_rate": 6.110647992125407e-06,
      "loss": 2.0338,
      "step": 6126
    },
    {
      "epoch": 0.9517669902912621,
      "grad_norm": 0.0,
      "learning_rate": 6.071493994871058e-06,
      "loss": 2.0632,
      "step": 6127
    },
    {
      "epoch": 0.9519223300970874,
      "grad_norm": 0.0,
      "learning_rate": 6.032465074530757e-06,
      "loss": 2.0779,
      "step": 6128
    },
    {
      "epoch": 0.9520776699029126,
      "grad_norm": 0.0,
      "learning_rate": 5.993561240987766e-06,
      "loss": 2.0572,
      "step": 6129
    },
    {
      "epoch": 0.9522330097087378,
      "grad_norm": 0.0,
      "learning_rate": 5.954782504093647e-06,
      "loss": 2.1177,
      "step": 6130
    },
    {
      "epoch": 0.9523883495145631,
      "grad_norm": 0.0,
      "learning_rate": 5.916128873668269e-06,
      "loss": 2.0703,
      "step": 6131
    },
    {
      "epoch": 0.9525436893203884,
      "grad_norm": 0.0,
      "learning_rate": 5.877600359499857e-06,
      "loss": 2.0009,
      "step": 6132
    },
    {
      "epoch": 0.9526990291262136,
      "grad_norm": 0.0,
      "learning_rate": 5.83919697134494e-06,
      "loss": 2.0696,
      "step": 6133
    },
    {
      "epoch": 0.9528543689320388,
      "grad_norm": 0.0,
      "learning_rate": 5.800918718928405e-06,
      "loss": 2.0786,
      "step": 6134
    },
    {
      "epoch": 0.9530097087378641,
      "grad_norm": 0.0,
      "learning_rate": 5.762765611943332e-06,
      "loss": 2.0047,
      "step": 6135
    },
    {
      "epoch": 0.9531650485436893,
      "grad_norm": 0.0,
      "learning_rate": 5.724737660051216e-06,
      "loss": 1.9654,
      "step": 6136
    },
    {
      "epoch": 0.9533203883495146,
      "grad_norm": 0.0,
      "learning_rate": 5.686834872881852e-06,
      "loss": 2.0601,
      "step": 6137
    },
    {
      "epoch": 0.9534757281553398,
      "grad_norm": 0.0,
      "learning_rate": 5.649057260033286e-06,
      "loss": 2.0905,
      "step": 6138
    },
    {
      "epoch": 0.953631067961165,
      "grad_norm": 0.0,
      "learning_rate": 5.61140483107192e-06,
      "loss": 2.0968,
      "step": 6139
    },
    {
      "epoch": 0.9537864077669903,
      "grad_norm": 0.0,
      "learning_rate": 5.573877595532406e-06,
      "loss": 2.1207,
      "step": 6140
    },
    {
      "epoch": 0.9539417475728156,
      "grad_norm": 0.0,
      "learning_rate": 5.536475562917753e-06,
      "loss": 2.0449,
      "step": 6141
    },
    {
      "epoch": 0.9540970873786407,
      "grad_norm": 0.0,
      "learning_rate": 5.499198742699219e-06,
      "loss": 2.0007,
      "step": 6142
    },
    {
      "epoch": 0.954252427184466,
      "grad_norm": 0.0,
      "learning_rate": 5.462047144316307e-06,
      "loss": 2.0221,
      "step": 6143
    },
    {
      "epoch": 0.9544077669902913,
      "grad_norm": 0.0,
      "learning_rate": 5.425020777176937e-06,
      "loss": 2.1658,
      "step": 6144
    },
    {
      "epoch": 0.9545631067961166,
      "grad_norm": 0.0,
      "learning_rate": 5.388119650657331e-06,
      "loss": 2.0315,
      "step": 6145
    },
    {
      "epoch": 0.9547184466019417,
      "grad_norm": 0.0,
      "learning_rate": 5.351343774101791e-06,
      "loss": 2.114,
      "step": 6146
    },
    {
      "epoch": 0.954873786407767,
      "grad_norm": 0.0,
      "learning_rate": 5.314693156823036e-06,
      "loss": 2.1217,
      "step": 6147
    },
    {
      "epoch": 0.9550291262135923,
      "grad_norm": 0.0,
      "learning_rate": 5.278167808102086e-06,
      "loss": 2.023,
      "step": 6148
    },
    {
      "epoch": 0.9551844660194174,
      "grad_norm": 0.0,
      "learning_rate": 5.241767737188208e-06,
      "loss": 2.0518,
      "step": 6149
    },
    {
      "epoch": 0.9553398058252427,
      "grad_norm": 0.0,
      "learning_rate": 5.2054929532989735e-06,
      "loss": 1.9592,
      "step": 6150
    },
    {
      "epoch": 0.955495145631068,
      "grad_norm": 0.0,
      "learning_rate": 5.169343465620091e-06,
      "loss": 2.1324,
      "step": 6151
    },
    {
      "epoch": 0.9556504854368932,
      "grad_norm": 0.0,
      "learning_rate": 5.1333192833058486e-06,
      "loss": 2.1223,
      "step": 6152
    },
    {
      "epoch": 0.9558058252427184,
      "grad_norm": 0.0,
      "learning_rate": 5.097420415478449e-06,
      "loss": 2.1015,
      "step": 6153
    },
    {
      "epoch": 0.9559611650485437,
      "grad_norm": 0.0,
      "learning_rate": 5.061646871228509e-06,
      "loss": 2.1317,
      "step": 6154
    },
    {
      "epoch": 0.9561165048543689,
      "grad_norm": 0.0,
      "learning_rate": 5.025998659615061e-06,
      "loss": 2.0283,
      "step": 6155
    },
    {
      "epoch": 0.9562718446601942,
      "grad_norm": 0.0,
      "learning_rate": 4.990475789665106e-06,
      "loss": 2.1781,
      "step": 6156
    },
    {
      "epoch": 0.9564271844660194,
      "grad_norm": 0.0,
      "learning_rate": 4.955078270374225e-06,
      "loss": 2.0957,
      "step": 6157
    },
    {
      "epoch": 0.9565825242718446,
      "grad_norm": 0.0,
      "learning_rate": 4.919806110705916e-06,
      "loss": 2.1044,
      "step": 6158
    },
    {
      "epoch": 0.9567378640776699,
      "grad_norm": 0.0,
      "learning_rate": 4.884659319592144e-06,
      "loss": 2.1492,
      "step": 6159
    },
    {
      "epoch": 0.9568932038834952,
      "grad_norm": 0.0,
      "learning_rate": 4.8496379059331795e-06,
      "loss": 2.1819,
      "step": 6160
    },
    {
      "epoch": 0.9570485436893204,
      "grad_norm": 0.0,
      "learning_rate": 4.814741878597373e-06,
      "loss": 2.1324,
      "step": 6161
    },
    {
      "epoch": 0.9572038834951456,
      "grad_norm": 0.0,
      "learning_rate": 4.779971246421378e-06,
      "loss": 2.0273,
      "step": 6162
    },
    {
      "epoch": 0.9573592233009709,
      "grad_norm": 0.0,
      "learning_rate": 4.745326018210261e-06,
      "loss": 2.0665,
      "step": 6163
    },
    {
      "epoch": 0.9575145631067962,
      "grad_norm": 0.0,
      "learning_rate": 4.710806202737006e-06,
      "loss": 2.0944,
      "step": 6164
    },
    {
      "epoch": 0.9576699029126213,
      "grad_norm": 0.0,
      "learning_rate": 4.676411808743175e-06,
      "loss": 2.0271,
      "step": 6165
    },
    {
      "epoch": 0.9578252427184466,
      "grad_norm": 0.0,
      "learning_rate": 4.6421428449383016e-06,
      "loss": 2.0648,
      "step": 6166
    },
    {
      "epoch": 0.9579805825242719,
      "grad_norm": 0.0,
      "learning_rate": 4.607999320000389e-06,
      "loss": 2.0262,
      "step": 6167
    },
    {
      "epoch": 0.958135922330097,
      "grad_norm": 0.0,
      "learning_rate": 4.573981242575465e-06,
      "loss": 2.0549,
      "step": 6168
    },
    {
      "epoch": 0.9582912621359223,
      "grad_norm": 0.0,
      "learning_rate": 4.540088621277916e-06,
      "loss": 2.1331,
      "step": 6169
    },
    {
      "epoch": 0.9584466019417476,
      "grad_norm": 0.0,
      "learning_rate": 4.506321464690322e-06,
      "loss": 2.0789,
      "step": 6170
    },
    {
      "epoch": 0.9586019417475728,
      "grad_norm": 0.0,
      "learning_rate": 4.472679781363564e-06,
      "loss": 2.0948,
      "step": 6171
    },
    {
      "epoch": 0.958757281553398,
      "grad_norm": 0.0,
      "learning_rate": 4.4391635798165515e-06,
      "loss": 1.9657,
      "step": 6172
    },
    {
      "epoch": 0.9589126213592233,
      "grad_norm": 0.0,
      "learning_rate": 4.405772868536717e-06,
      "loss": 1.9963,
      "step": 6173
    },
    {
      "epoch": 0.9590679611650486,
      "grad_norm": 0.0,
      "learning_rate": 4.372507655979352e-06,
      "loss": 2.1389,
      "step": 6174
    },
    {
      "epoch": 0.9592233009708738,
      "grad_norm": 0.0,
      "learning_rate": 4.339367950568385e-06,
      "loss": 2.0902,
      "step": 6175
    },
    {
      "epoch": 0.959378640776699,
      "grad_norm": 0.0,
      "learning_rate": 4.306353760695603e-06,
      "loss": 2.0988,
      "step": 6176
    },
    {
      "epoch": 0.9595339805825243,
      "grad_norm": 0.0,
      "learning_rate": 4.273465094721152e-06,
      "loss": 2.0858,
      "step": 6177
    },
    {
      "epoch": 0.9596893203883495,
      "grad_norm": 0.0,
      "learning_rate": 4.240701960973426e-06,
      "loss": 2.0764,
      "step": 6178
    },
    {
      "epoch": 0.9598446601941748,
      "grad_norm": 0.0,
      "learning_rate": 4.20806436774901e-06,
      "loss": 2.1072,
      "step": 6179
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.0,
      "learning_rate": 4.175552323312682e-06,
      "loss": 2.0647,
      "step": 6180
    },
    {
      "epoch": 0.9601553398058252,
      "grad_norm": 0.0,
      "learning_rate": 4.1431658358973554e-06,
      "loss": 2.0748,
      "step": 6181
    },
    {
      "epoch": 0.9603106796116505,
      "grad_norm": 0.0,
      "learning_rate": 4.11090491370425e-06,
      "loss": 2.0245,
      "step": 6182
    },
    {
      "epoch": 0.9604660194174758,
      "grad_norm": 0.0,
      "learning_rate": 4.078769564902829e-06,
      "loss": 2.0032,
      "step": 6183
    },
    {
      "epoch": 0.9606213592233009,
      "grad_norm": 0.0,
      "learning_rate": 4.0467597976306394e-06,
      "loss": 2.0814,
      "step": 6184
    },
    {
      "epoch": 0.9607766990291262,
      "grad_norm": 0.0,
      "learning_rate": 4.014875619993474e-06,
      "loss": 1.9667,
      "step": 6185
    },
    {
      "epoch": 0.9609320388349515,
      "grad_norm": 0.0,
      "learning_rate": 3.983117040065265e-06,
      "loss": 2.1595,
      "step": 6186
    },
    {
      "epoch": 0.9610873786407766,
      "grad_norm": 0.0,
      "learning_rate": 3.9514840658883e-06,
      "loss": 1.9367,
      "step": 6187
    },
    {
      "epoch": 0.9612427184466019,
      "grad_norm": 0.0,
      "learning_rate": 3.919976705472894e-06,
      "loss": 2.0094,
      "step": 6188
    },
    {
      "epoch": 0.9613980582524272,
      "grad_norm": 0.0,
      "learning_rate": 3.888594966797665e-06,
      "loss": 2.1036,
      "step": 6189
    },
    {
      "epoch": 0.9615533980582525,
      "grad_norm": 0.0,
      "learning_rate": 3.857338857809367e-06,
      "loss": 2.0237,
      "step": 6190
    },
    {
      "epoch": 0.9617087378640776,
      "grad_norm": 0.0,
      "learning_rate": 3.826208386422891e-06,
      "loss": 2.1857,
      "step": 6191
    },
    {
      "epoch": 0.9618640776699029,
      "grad_norm": 0.0,
      "learning_rate": 3.795203560521432e-06,
      "loss": 1.9415,
      "step": 6192
    },
    {
      "epoch": 0.9620194174757282,
      "grad_norm": 0.0,
      "learning_rate": 3.764324387956264e-06,
      "loss": 2.1048,
      "step": 6193
    },
    {
      "epoch": 0.9621747572815534,
      "grad_norm": 0.0,
      "learning_rate": 3.7335708765468546e-06,
      "loss": 2.1534,
      "step": 6194
    },
    {
      "epoch": 0.9623300970873786,
      "grad_norm": 0.0,
      "learning_rate": 3.702943034080919e-06,
      "loss": 1.8836,
      "step": 6195
    },
    {
      "epoch": 0.9624854368932039,
      "grad_norm": 0.0,
      "learning_rate": 3.6724408683143638e-06,
      "loss": 2.1251,
      "step": 6196
    },
    {
      "epoch": 0.9626407766990291,
      "grad_norm": 0.0,
      "learning_rate": 3.6420643869711224e-06,
      "loss": 2.0361,
      "step": 6197
    },
    {
      "epoch": 0.9627961165048544,
      "grad_norm": 0.0,
      "learning_rate": 3.6118135977434853e-06,
      "loss": 2.0961,
      "step": 6198
    },
    {
      "epoch": 0.9629514563106796,
      "grad_norm": 0.0,
      "learning_rate": 3.581688508291714e-06,
      "loss": 2.1248,
      "step": 6199
    },
    {
      "epoch": 0.9631067961165048,
      "grad_norm": 0.0,
      "learning_rate": 3.5516891262444283e-06,
      "loss": 2.1728,
      "step": 6200
    },
    {
      "epoch": 0.9632621359223301,
      "grad_norm": 0.0,
      "learning_rate": 3.5218154591982745e-06,
      "loss": 2.0538,
      "step": 6201
    },
    {
      "epoch": 0.9634174757281554,
      "grad_norm": 0.0,
      "learning_rate": 3.4920675147181447e-06,
      "loss": 2.0885,
      "step": 6202
    },
    {
      "epoch": 0.9635728155339806,
      "grad_norm": 0.0,
      "learning_rate": 3.46244530033718e-06,
      "loss": 2.0068,
      "step": 6203
    },
    {
      "epoch": 0.9637281553398058,
      "grad_norm": 0.0,
      "learning_rate": 3.4329488235564367e-06,
      "loss": 2.1128,
      "step": 6204
    },
    {
      "epoch": 0.9638834951456311,
      "grad_norm": 0.0,
      "learning_rate": 3.4035780918453276e-06,
      "loss": 2.069,
      "step": 6205
    },
    {
      "epoch": 0.9640388349514563,
      "grad_norm": 0.0,
      "learning_rate": 3.374333112641348e-06,
      "loss": 2.1486,
      "step": 6206
    },
    {
      "epoch": 0.9641941747572815,
      "grad_norm": 0.0,
      "learning_rate": 3.345213893350185e-06,
      "loss": 1.9821,
      "step": 6207
    },
    {
      "epoch": 0.9643495145631068,
      "grad_norm": 0.0,
      "learning_rate": 3.316220441345663e-06,
      "loss": 2.0424,
      "step": 6208
    },
    {
      "epoch": 0.9645048543689321,
      "grad_norm": 0.0,
      "learning_rate": 3.287352763969742e-06,
      "loss": 1.9922,
      "step": 6209
    },
    {
      "epoch": 0.9646601941747572,
      "grad_norm": 0.0,
      "learning_rate": 3.2586108685325746e-06,
      "loss": 2.0603,
      "step": 6210
    },
    {
      "epoch": 0.9648155339805825,
      "grad_norm": 0.0,
      "learning_rate": 3.229994762312449e-06,
      "loss": 2.0259,
      "step": 6211
    },
    {
      "epoch": 0.9649708737864078,
      "grad_norm": 0.0,
      "learning_rate": 3.201504452555737e-06,
      "loss": 2.0616,
      "step": 6212
    },
    {
      "epoch": 0.965126213592233,
      "grad_norm": 0.0,
      "learning_rate": 3.173139946477055e-06,
      "loss": 2.0959,
      "step": 6213
    },
    {
      "epoch": 0.9652815533980582,
      "grad_norm": 0.0,
      "learning_rate": 3.144901251259047e-06,
      "loss": 2.0572,
      "step": 6214
    },
    {
      "epoch": 0.9654368932038835,
      "grad_norm": 0.0,
      "learning_rate": 3.116788374052659e-06,
      "loss": 1.8564,
      "step": 6215
    },
    {
      "epoch": 0.9655922330097088,
      "grad_norm": 0.0,
      "learning_rate": 3.0888013219768086e-06,
      "loss": 2.0496,
      "step": 6216
    },
    {
      "epoch": 0.965747572815534,
      "grad_norm": 0.0,
      "learning_rate": 3.0609401021186034e-06,
      "loss": 2.1184,
      "step": 6217
    },
    {
      "epoch": 0.9659029126213592,
      "grad_norm": 0.0,
      "learning_rate": 3.0332047215333446e-06,
      "loss": 2.1454,
      "step": 6218
    },
    {
      "epoch": 0.9660582524271845,
      "grad_norm": 0.0,
      "learning_rate": 3.00559518724447e-06,
      "loss": 2.0581,
      "step": 6219
    },
    {
      "epoch": 0.9662135922330097,
      "grad_norm": 0.0,
      "learning_rate": 2.978111506243386e-06,
      "loss": 2.0281,
      "step": 6220
    },
    {
      "epoch": 0.966368932038835,
      "grad_norm": 0.0,
      "learning_rate": 2.950753685489915e-06,
      "loss": 2.0767,
      "step": 6221
    },
    {
      "epoch": 0.9665242718446602,
      "grad_norm": 0.0,
      "learning_rate": 2.923521731911738e-06,
      "loss": 2.0586,
      "step": 6222
    },
    {
      "epoch": 0.9666796116504854,
      "grad_norm": 0.0,
      "learning_rate": 2.8964156524047826e-06,
      "loss": 2.0357,
      "step": 6223
    },
    {
      "epoch": 0.9668349514563107,
      "grad_norm": 0.0,
      "learning_rate": 2.8694354538331134e-06,
      "loss": 2.0701,
      "step": 6224
    },
    {
      "epoch": 0.9669902912621359,
      "grad_norm": 0.0,
      "learning_rate": 2.8425811430288216e-06,
      "loss": 1.9552,
      "step": 6225
    },
    {
      "epoch": 0.9671456310679611,
      "grad_norm": 0.0,
      "learning_rate": 2.815852726792356e-06,
      "loss": 2.0895,
      "step": 6226
    },
    {
      "epoch": 0.9673009708737864,
      "grad_norm": 0.0,
      "learning_rate": 2.789250211891914e-06,
      "loss": 2.0695,
      "step": 6227
    },
    {
      "epoch": 0.9674563106796117,
      "grad_norm": 0.0,
      "learning_rate": 2.762773605064162e-06,
      "loss": 2.0974,
      "step": 6228
    },
    {
      "epoch": 0.9676116504854368,
      "grad_norm": 0.0,
      "learning_rate": 2.736422913013681e-06,
      "loss": 2.0822,
      "step": 6229
    },
    {
      "epoch": 0.9677669902912621,
      "grad_norm": 0.0,
      "learning_rate": 2.710198142413245e-06,
      "loss": 2.0342,
      "step": 6230
    },
    {
      "epoch": 0.9679223300970874,
      "grad_norm": 0.0,
      "learning_rate": 2.6840992999037083e-06,
      "loss": 2.1385,
      "step": 6231
    },
    {
      "epoch": 0.9680776699029127,
      "grad_norm": 0.0,
      "learning_rate": 2.658126392094062e-06,
      "loss": 1.9919,
      "step": 6232
    },
    {
      "epoch": 0.9682330097087378,
      "grad_norm": 0.0,
      "learning_rate": 2.632279425561379e-06,
      "loss": 2.1687,
      "step": 6233
    },
    {
      "epoch": 0.9683883495145631,
      "grad_norm": 0.0,
      "learning_rate": 2.6065584068508673e-06,
      "loss": 2.0007,
      "step": 6234
    },
    {
      "epoch": 0.9685436893203884,
      "grad_norm": 0.0,
      "learning_rate": 2.580963342475817e-06,
      "loss": 2.0077,
      "step": 6235
    },
    {
      "epoch": 0.9686990291262136,
      "grad_norm": 0.0,
      "learning_rate": 2.5554942389175997e-06,
      "loss": 2.0303,
      "step": 6236
    },
    {
      "epoch": 0.9688543689320388,
      "grad_norm": 0.0,
      "learning_rate": 2.5301511026258326e-06,
      "loss": 2.1052,
      "step": 6237
    },
    {
      "epoch": 0.9690097087378641,
      "grad_norm": 0.0,
      "learning_rate": 2.5049339400180505e-06,
      "loss": 2.0908,
      "step": 6238
    },
    {
      "epoch": 0.9691650485436893,
      "grad_norm": 0.0,
      "learning_rate": 2.4798427574799775e-06,
      "loss": 2.0967,
      "step": 6239
    },
    {
      "epoch": 0.9693203883495146,
      "grad_norm": 0.0,
      "learning_rate": 2.454877561365365e-06,
      "loss": 2.0541,
      "step": 6240
    },
    {
      "epoch": 0.9694757281553398,
      "grad_norm": 0.0,
      "learning_rate": 2.430038357996267e-06,
      "loss": 2.0815,
      "step": 6241
    },
    {
      "epoch": 0.969631067961165,
      "grad_norm": 0.0,
      "learning_rate": 2.405325153662541e-06,
      "loss": 2.0389,
      "step": 6242
    },
    {
      "epoch": 0.9697864077669903,
      "grad_norm": 0.0,
      "learning_rate": 2.380737954622292e-06,
      "loss": 2.0159,
      "step": 6243
    },
    {
      "epoch": 0.9699417475728155,
      "grad_norm": 0.0,
      "learning_rate": 2.356276767101817e-06,
      "loss": 2.1547,
      "step": 6244
    },
    {
      "epoch": 0.9700970873786408,
      "grad_norm": 0.0,
      "learning_rate": 2.3319415972953283e-06,
      "loss": 2.0465,
      "step": 6245
    },
    {
      "epoch": 0.970252427184466,
      "grad_norm": 0.0,
      "learning_rate": 2.307732451365174e-06,
      "loss": 2.1075,
      "step": 6246
    },
    {
      "epoch": 0.9704077669902913,
      "grad_norm": 0.0,
      "learning_rate": 2.2836493354417844e-06,
      "loss": 2.1327,
      "step": 6247
    },
    {
      "epoch": 0.9705631067961165,
      "grad_norm": 0.0,
      "learning_rate": 2.2596922556237797e-06,
      "loss": 2.2159,
      "step": 6248
    },
    {
      "epoch": 0.9707184466019417,
      "grad_norm": 0.0,
      "learning_rate": 2.235861217977697e-06,
      "loss": 2.093,
      "step": 6249
    },
    {
      "epoch": 0.970873786407767,
      "grad_norm": 0.0,
      "learning_rate": 2.21215622853832e-06,
      "loss": 2.0919,
      "step": 6250
    },
    {
      "epoch": 0.9710291262135923,
      "grad_norm": 0.0,
      "learning_rate": 2.1885772933084024e-06,
      "loss": 2.0316,
      "step": 6251
    },
    {
      "epoch": 0.9711844660194174,
      "grad_norm": 0.0,
      "learning_rate": 2.1651244182587238e-06,
      "loss": 2.0231,
      "step": 6252
    },
    {
      "epoch": 0.9713398058252427,
      "grad_norm": 0.0,
      "learning_rate": 2.1417976093283664e-06,
      "loss": 2.0546,
      "step": 6253
    },
    {
      "epoch": 0.971495145631068,
      "grad_norm": 0.0,
      "learning_rate": 2.118596872424272e-06,
      "loss": 2.1738,
      "step": 6254
    },
    {
      "epoch": 0.9716504854368933,
      "grad_norm": 0.0,
      "learning_rate": 2.0955222134215748e-06,
      "loss": 2.0538,
      "step": 6255
    },
    {
      "epoch": 0.9718058252427184,
      "grad_norm": 0.0,
      "learning_rate": 2.0725736381633776e-06,
      "loss": 2.045,
      "step": 6256
    },
    {
      "epoch": 0.9719611650485437,
      "grad_norm": 0.0,
      "learning_rate": 2.049751152460977e-06,
      "loss": 2.1003,
      "step": 6257
    },
    {
      "epoch": 0.972116504854369,
      "grad_norm": 0.0,
      "learning_rate": 2.0270547620936387e-06,
      "loss": 2.1135,
      "step": 6258
    },
    {
      "epoch": 0.9722718446601942,
      "grad_norm": 0.0,
      "learning_rate": 2.0044844728088208e-06,
      "loss": 2.1058,
      "step": 6259
    },
    {
      "epoch": 0.9724271844660194,
      "grad_norm": 0.0,
      "learning_rate": 1.982040290321896e-06,
      "loss": 2.0508,
      "step": 6260
    },
    {
      "epoch": 0.9725825242718447,
      "grad_norm": 0.0,
      "learning_rate": 1.959722220316429e-06,
      "loss": 2.0494,
      "step": 6261
    },
    {
      "epoch": 0.9727378640776699,
      "grad_norm": 0.0,
      "learning_rate": 1.9375302684439543e-06,
      "loss": 2.1287,
      "step": 6262
    },
    {
      "epoch": 0.9728932038834951,
      "grad_norm": 0.0,
      "learning_rate": 1.9154644403241437e-06,
      "loss": 2.1207,
      "step": 6263
    },
    {
      "epoch": 0.9730485436893204,
      "grad_norm": 0.0,
      "learning_rate": 1.8935247415446388e-06,
      "loss": 1.9804,
      "step": 6264
    },
    {
      "epoch": 0.9732038834951456,
      "grad_norm": 0.0,
      "learning_rate": 1.8717111776612727e-06,
      "loss": 2.0151,
      "step": 6265
    },
    {
      "epoch": 0.9733592233009709,
      "grad_norm": 0.0,
      "learning_rate": 1.8500237541978494e-06,
      "loss": 2.0217,
      "step": 6266
    },
    {
      "epoch": 0.9735145631067961,
      "grad_norm": 0.0,
      "learning_rate": 1.8284624766462533e-06,
      "loss": 2.0392,
      "step": 6267
    },
    {
      "epoch": 0.9736699029126213,
      "grad_norm": 0.0,
      "learning_rate": 1.8070273504663392e-06,
      "loss": 2.1056,
      "step": 6268
    },
    {
      "epoch": 0.9738252427184466,
      "grad_norm": 0.0,
      "learning_rate": 1.7857183810862654e-06,
      "loss": 2.0778,
      "step": 6269
    },
    {
      "epoch": 0.9739805825242719,
      "grad_norm": 0.0,
      "learning_rate": 1.7645355739019376e-06,
      "loss": 2.1376,
      "step": 6270
    },
    {
      "epoch": 0.974135922330097,
      "grad_norm": 0.0,
      "learning_rate": 1.7434789342775093e-06,
      "loss": 2.0114,
      "step": 6271
    },
    {
      "epoch": 0.9742912621359223,
      "grad_norm": 0.0,
      "learning_rate": 1.7225484675451042e-06,
      "loss": 2.1047,
      "step": 6272
    },
    {
      "epoch": 0.9744466019417476,
      "grad_norm": 0.0,
      "learning_rate": 1.7017441790048715e-06,
      "loss": 2.1475,
      "step": 6273
    },
    {
      "epoch": 0.9746019417475729,
      "grad_norm": 0.0,
      "learning_rate": 1.6810660739252081e-06,
      "loss": 2.1024,
      "step": 6274
    },
    {
      "epoch": 0.974757281553398,
      "grad_norm": 0.0,
      "learning_rate": 1.6605141575422589e-06,
      "loss": 2.0338,
      "step": 6275
    },
    {
      "epoch": 0.9749126213592233,
      "grad_norm": 0.0,
      "learning_rate": 1.6400884350603607e-06,
      "loss": 1.9999,
      "step": 6276
    },
    {
      "epoch": 0.9750679611650486,
      "grad_norm": 0.0,
      "learning_rate": 1.6197889116520426e-06,
      "loss": 2.1375,
      "step": 6277
    },
    {
      "epoch": 0.9752233009708738,
      "grad_norm": 0.0,
      "learning_rate": 1.5996155924575817e-06,
      "loss": 1.9718,
      "step": 6278
    },
    {
      "epoch": 0.975378640776699,
      "grad_norm": 0.0,
      "learning_rate": 1.5795684825855027e-06,
      "loss": 2.0073,
      "step": 6279
    },
    {
      "epoch": 0.9755339805825243,
      "grad_norm": 0.0,
      "learning_rate": 1.559647587112245e-06,
      "loss": 2.0468,
      "step": 6280
    },
    {
      "epoch": 0.9756893203883495,
      "grad_norm": 0.0,
      "learning_rate": 1.5398529110824955e-06,
      "loss": 2.11,
      "step": 6281
    },
    {
      "epoch": 0.9758446601941747,
      "grad_norm": 0.0,
      "learning_rate": 1.520184459508689e-06,
      "loss": 2.1467,
      "step": 6282
    },
    {
      "epoch": 0.976,
      "grad_norm": 0.0,
      "learning_rate": 1.5006422373715079e-06,
      "loss": 1.9975,
      "step": 6283
    },
    {
      "epoch": 0.9761553398058253,
      "grad_norm": 0.0,
      "learning_rate": 1.4812262496196604e-06,
      "loss": 2.1625,
      "step": 6284
    },
    {
      "epoch": 0.9763106796116505,
      "grad_norm": 0.0,
      "learning_rate": 1.461936501169714e-06,
      "loss": 2.0048,
      "step": 6285
    },
    {
      "epoch": 0.9764660194174757,
      "grad_norm": 0.0,
      "learning_rate": 1.4427729969064274e-06,
      "loss": 2.1003,
      "step": 6286
    },
    {
      "epoch": 0.976621359223301,
      "grad_norm": 0.0,
      "learning_rate": 1.4237357416826414e-06,
      "loss": 2.0483,
      "step": 6287
    },
    {
      "epoch": 0.9767766990291262,
      "grad_norm": 0.0,
      "learning_rate": 1.4048247403189996e-06,
      "loss": 2.0946,
      "step": 6288
    },
    {
      "epoch": 0.9769320388349515,
      "grad_norm": 0.0,
      "learning_rate": 1.3860399976043937e-06,
      "loss": 2.0739,
      "step": 6289
    },
    {
      "epoch": 0.9770873786407767,
      "grad_norm": 0.0,
      "learning_rate": 1.3673815182956294e-06,
      "loss": 2.0434,
      "step": 6290
    },
    {
      "epoch": 0.9772427184466019,
      "grad_norm": 0.0,
      "learning_rate": 1.34884930711765e-06,
      "loss": 2.093,
      "step": 6291
    },
    {
      "epoch": 0.9773980582524272,
      "grad_norm": 0.0,
      "learning_rate": 1.3304433687632567e-06,
      "loss": 2.1479,
      "step": 6292
    },
    {
      "epoch": 0.9775533980582525,
      "grad_norm": 0.0,
      "learning_rate": 1.3121637078933325e-06,
      "loss": 2.0316,
      "step": 6293
    },
    {
      "epoch": 0.9777087378640776,
      "grad_norm": 0.0,
      "learning_rate": 1.294010329136952e-06,
      "loss": 2.0422,
      "step": 6294
    },
    {
      "epoch": 0.9778640776699029,
      "grad_norm": 0.0,
      "learning_rate": 1.2759832370909385e-06,
      "loss": 2.0053,
      "step": 6295
    },
    {
      "epoch": 0.9780194174757282,
      "grad_norm": 0.0,
      "learning_rate": 1.2580824363203624e-06,
      "loss": 2.1399,
      "step": 6296
    },
    {
      "epoch": 0.9781747572815535,
      "grad_norm": 0.0,
      "learning_rate": 1.2403079313581534e-06,
      "loss": 2.2001,
      "step": 6297
    },
    {
      "epoch": 0.9783300970873786,
      "grad_norm": 0.0,
      "learning_rate": 1.2226597267053775e-06,
      "loss": 1.9968,
      "step": 6298
    },
    {
      "epoch": 0.9784854368932039,
      "grad_norm": 0.0,
      "learning_rate": 1.2051378268310155e-06,
      "loss": 2.0682,
      "step": 6299
    },
    {
      "epoch": 0.9786407766990292,
      "grad_norm": 0.0,
      "learning_rate": 1.1877422361721291e-06,
      "loss": 2.0641,
      "step": 6300
    },
    {
      "epoch": 0.9787961165048543,
      "grad_norm": 0.0,
      "learning_rate": 1.1704729591338614e-06,
      "loss": 2.1263,
      "step": 6301
    },
    {
      "epoch": 0.9789514563106796,
      "grad_norm": 0.0,
      "learning_rate": 1.1533300000892144e-06,
      "loss": 2.0665,
      "step": 6302
    },
    {
      "epoch": 0.9791067961165049,
      "grad_norm": 0.0,
      "learning_rate": 1.136313363379271e-06,
      "loss": 2.0505,
      "step": 6303
    },
    {
      "epoch": 0.9792621359223301,
      "grad_norm": 0.0,
      "learning_rate": 1.1194230533131954e-06,
      "loss": 2.0878,
      "step": 6304
    },
    {
      "epoch": 0.9794174757281553,
      "grad_norm": 0.0,
      "learning_rate": 1.102659074168011e-06,
      "loss": 1.9743,
      "step": 6305
    },
    {
      "epoch": 0.9795728155339806,
      "grad_norm": 0.0,
      "learning_rate": 1.0860214301889327e-06,
      "loss": 2.0929,
      "step": 6306
    },
    {
      "epoch": 0.9797281553398058,
      "grad_norm": 0.0,
      "learning_rate": 1.0695101255890348e-06,
      "loss": 2.0881,
      "step": 6307
    },
    {
      "epoch": 0.9798834951456311,
      "grad_norm": 0.0,
      "learning_rate": 1.0531251645494732e-06,
      "loss": 2.0265,
      "step": 6308
    },
    {
      "epoch": 0.9800388349514563,
      "grad_norm": 0.0,
      "learning_rate": 1.0368665512193731e-06,
      "loss": 2.0281,
      "step": 6309
    },
    {
      "epoch": 0.9801941747572815,
      "grad_norm": 0.0,
      "learning_rate": 1.020734289715941e-06,
      "loss": 2.0431,
      "step": 6310
    },
    {
      "epoch": 0.9803495145631068,
      "grad_norm": 0.0,
      "learning_rate": 1.0047283841242427e-06,
      "loss": 2.1377,
      "step": 6311
    },
    {
      "epoch": 0.9805048543689321,
      "grad_norm": 0.0,
      "learning_rate": 9.888488384974803e-07,
      "loss": 2.178,
      "step": 6312
    },
    {
      "epoch": 0.9806601941747572,
      "grad_norm": 0.0,
      "learning_rate": 9.73095656856826e-07,
      "loss": 2.0037,
      "step": 6313
    },
    {
      "epoch": 0.9808155339805825,
      "grad_norm": 0.0,
      "learning_rate": 9.574688431914226e-07,
      "loss": 2.0812,
      "step": 6314
    },
    {
      "epoch": 0.9809708737864078,
      "grad_norm": 0.0,
      "learning_rate": 9.419684014583818e-07,
      "loss": 2.0396,
      "step": 6315
    },
    {
      "epoch": 0.9811262135922331,
      "grad_norm": 0.0,
      "learning_rate": 9.265943355829531e-07,
      "loss": 2.0749,
      "step": 6316
    },
    {
      "epoch": 0.9812815533980582,
      "grad_norm": 0.0,
      "learning_rate": 9.113466494582445e-07,
      "loss": 2.0453,
      "step": 6317
    },
    {
      "epoch": 0.9814368932038835,
      "grad_norm": 0.0,
      "learning_rate": 8.962253469454451e-07,
      "loss": 1.9531,
      "step": 6318
    },
    {
      "epoch": 0.9815922330097088,
      "grad_norm": 0.0,
      "learning_rate": 8.812304318736031e-07,
      "loss": 2.0861,
      "step": 6319
    },
    {
      "epoch": 0.9817475728155339,
      "grad_norm": 0.0,
      "learning_rate": 8.663619080400142e-07,
      "loss": 2.0316,
      "step": 6320
    },
    {
      "epoch": 0.9819029126213592,
      "grad_norm": 0.0,
      "learning_rate": 8.516197792096669e-07,
      "loss": 2.1453,
      "step": 6321
    },
    {
      "epoch": 0.9820582524271845,
      "grad_norm": 0.0,
      "learning_rate": 8.370040491157971e-07,
      "loss": 2.0551,
      "step": 6322
    },
    {
      "epoch": 0.9822135922330097,
      "grad_norm": 0.0,
      "learning_rate": 8.225147214594442e-07,
      "loss": 2.0701,
      "step": 6323
    },
    {
      "epoch": 0.9823689320388349,
      "grad_norm": 0.0,
      "learning_rate": 8.081517999097842e-07,
      "loss": 2.066,
      "step": 6324
    },
    {
      "epoch": 0.9825242718446602,
      "grad_norm": 0.0,
      "learning_rate": 7.939152881039636e-07,
      "loss": 2.0314,
      "step": 6325
    },
    {
      "epoch": 0.9826796116504855,
      "grad_norm": 0.0,
      "learning_rate": 7.798051896469316e-07,
      "loss": 2.0368,
      "step": 6326
    },
    {
      "epoch": 0.9828349514563107,
      "grad_norm": 0.0,
      "learning_rate": 7.658215081119413e-07,
      "loss": 2.0031,
      "step": 6327
    },
    {
      "epoch": 0.9829902912621359,
      "grad_norm": 0.0,
      "learning_rate": 7.519642470399379e-07,
      "loss": 2.019,
      "step": 6328
    },
    {
      "epoch": 0.9831456310679612,
      "grad_norm": 0.0,
      "learning_rate": 7.382334099400589e-07,
      "loss": 2.0378,
      "step": 6329
    },
    {
      "epoch": 0.9833009708737864,
      "grad_norm": 0.0,
      "learning_rate": 7.246290002892453e-07,
      "loss": 2.0916,
      "step": 6330
    },
    {
      "epoch": 0.9834563106796117,
      "grad_norm": 0.0,
      "learning_rate": 7.1115102153263e-07,
      "loss": 2.0451,
      "step": 6331
    },
    {
      "epoch": 0.9836116504854369,
      "grad_norm": 0.0,
      "learning_rate": 6.977994770832052e-07,
      "loss": 1.9832,
      "step": 6332
    },
    {
      "epoch": 0.9837669902912621,
      "grad_norm": 0.0,
      "learning_rate": 6.845743703219332e-07,
      "loss": 2.0994,
      "step": 6333
    },
    {
      "epoch": 0.9839223300970874,
      "grad_norm": 0.0,
      "learning_rate": 6.714757045978015e-07,
      "loss": 2.0628,
      "step": 6334
    },
    {
      "epoch": 0.9840776699029126,
      "grad_norm": 0.0,
      "learning_rate": 6.585034832277681e-07,
      "loss": 2.1279,
      "step": 6335
    },
    {
      "epoch": 0.9842330097087378,
      "grad_norm": 0.0,
      "learning_rate": 6.456577094967608e-07,
      "loss": 2.0247,
      "step": 6336
    },
    {
      "epoch": 0.9843883495145631,
      "grad_norm": 0.0,
      "learning_rate": 6.329383866577887e-07,
      "loss": 2.0507,
      "step": 6337
    },
    {
      "epoch": 0.9845436893203884,
      "grad_norm": 0.0,
      "learning_rate": 6.203455179316086e-07,
      "loss": 2.0894,
      "step": 6338
    },
    {
      "epoch": 0.9846990291262135,
      "grad_norm": 0.0,
      "learning_rate": 6.078791065071698e-07,
      "loss": 2.0983,
      "step": 6339
    },
    {
      "epoch": 0.9848543689320388,
      "grad_norm": 0.0,
      "learning_rate": 5.955391555413914e-07,
      "loss": 2.0507,
      "step": 6340
    },
    {
      "epoch": 0.9850097087378641,
      "grad_norm": 0.0,
      "learning_rate": 5.833256681589405e-07,
      "loss": 2.012,
      "step": 6341
    },
    {
      "epoch": 0.9851650485436894,
      "grad_norm": 0.0,
      "learning_rate": 5.712386474527875e-07,
      "loss": 2.0798,
      "step": 6342
    },
    {
      "epoch": 0.9853203883495145,
      "grad_norm": 0.0,
      "learning_rate": 5.592780964836508e-07,
      "loss": 2.082,
      "step": 6343
    },
    {
      "epoch": 0.9854757281553398,
      "grad_norm": 0.0,
      "learning_rate": 5.474440182802188e-07,
      "loss": 2.0701,
      "step": 6344
    },
    {
      "epoch": 0.9856310679611651,
      "grad_norm": 0.0,
      "learning_rate": 5.357364158393718e-07,
      "loss": 2.0686,
      "step": 6345
    },
    {
      "epoch": 0.9857864077669903,
      "grad_norm": 0.0,
      "learning_rate": 5.241552921256276e-07,
      "loss": 2.0689,
      "step": 6346
    },
    {
      "epoch": 0.9859417475728155,
      "grad_norm": 0.0,
      "learning_rate": 5.127006500718068e-07,
      "loss": 2.2102,
      "step": 6347
    },
    {
      "epoch": 0.9860970873786408,
      "grad_norm": 0.0,
      "learning_rate": 5.013724925785335e-07,
      "loss": 2.179,
      "step": 6348
    },
    {
      "epoch": 0.986252427184466,
      "grad_norm": 0.0,
      "learning_rate": 4.901708225143464e-07,
      "loss": 2.1027,
      "step": 6349
    },
    {
      "epoch": 0.9864077669902913,
      "grad_norm": 0.0,
      "learning_rate": 4.79095642715921e-07,
      "loss": 2.0133,
      "step": 6350
    },
    {
      "epoch": 0.9865631067961165,
      "grad_norm": 0.0,
      "learning_rate": 4.6814695598773606e-07,
      "loss": 2.0629,
      "step": 6351
    },
    {
      "epoch": 0.9867184466019417,
      "grad_norm": 0.0,
      "learning_rate": 4.5732476510235154e-07,
      "loss": 2.0674,
      "step": 6352
    },
    {
      "epoch": 0.986873786407767,
      "grad_norm": 0.0,
      "learning_rate": 4.46629072800242e-07,
      "loss": 2.1144,
      "step": 6353
    },
    {
      "epoch": 0.9870291262135922,
      "grad_norm": 0.0,
      "learning_rate": 4.36059881789852e-07,
      "loss": 2.0776,
      "step": 6354
    },
    {
      "epoch": 0.9871844660194175,
      "grad_norm": 0.0,
      "learning_rate": 4.256171947477072e-07,
      "loss": 2.0901,
      "step": 6355
    },
    {
      "epoch": 0.9873398058252427,
      "grad_norm": 0.0,
      "learning_rate": 4.153010143180813e-07,
      "loss": 2.0558,
      "step": 6356
    },
    {
      "epoch": 0.987495145631068,
      "grad_norm": 0.0,
      "learning_rate": 4.051113431133291e-07,
      "loss": 2.179,
      "step": 6357
    },
    {
      "epoch": 0.9876504854368932,
      "grad_norm": 0.0,
      "learning_rate": 3.950481837138309e-07,
      "loss": 2.2039,
      "step": 6358
    },
    {
      "epoch": 0.9878058252427184,
      "grad_norm": 0.0,
      "learning_rate": 3.851115386678261e-07,
      "loss": 2.0783,
      "step": 6359
    },
    {
      "epoch": 0.9879611650485437,
      "grad_norm": 0.0,
      "learning_rate": 3.753014104915797e-07,
      "loss": 2.0735,
      "step": 6360
    },
    {
      "epoch": 0.988116504854369,
      "grad_norm": 0.0,
      "learning_rate": 3.656178016693268e-07,
      "loss": 2.0777,
      "step": 6361
    },
    {
      "epoch": 0.9882718446601941,
      "grad_norm": 0.0,
      "learning_rate": 3.5606071465316137e-07,
      "loss": 2.0018,
      "step": 6362
    },
    {
      "epoch": 0.9884271844660194,
      "grad_norm": 0.0,
      "learning_rate": 3.4663015186331413e-07,
      "loss": 2.0194,
      "step": 6363
    },
    {
      "epoch": 0.9885825242718447,
      "grad_norm": 0.0,
      "learning_rate": 3.373261156877638e-07,
      "loss": 2.0579,
      "step": 6364
    },
    {
      "epoch": 0.98873786407767,
      "grad_norm": 0.0,
      "learning_rate": 3.281486084825702e-07,
      "loss": 2.1473,
      "step": 6365
    },
    {
      "epoch": 0.9888932038834951,
      "grad_norm": 0.0,
      "learning_rate": 3.1909763257187417e-07,
      "loss": 2.1082,
      "step": 6366
    },
    {
      "epoch": 0.9890485436893204,
      "grad_norm": 0.0,
      "learning_rate": 3.101731902475091e-07,
      "loss": 2.0249,
      "step": 6367
    },
    {
      "epoch": 0.9892038834951457,
      "grad_norm": 0.0,
      "learning_rate": 3.01375283769445e-07,
      "loss": 2.0867,
      "step": 6368
    },
    {
      "epoch": 0.9893592233009709,
      "grad_norm": 0.0,
      "learning_rate": 2.927039153655109e-07,
      "loss": 2.153,
      "step": 6369
    },
    {
      "epoch": 0.9895145631067961,
      "grad_norm": 0.0,
      "learning_rate": 2.841590872316724e-07,
      "loss": 2.1066,
      "step": 6370
    },
    {
      "epoch": 0.9896699029126214,
      "grad_norm": 0.0,
      "learning_rate": 2.7574080153164314e-07,
      "loss": 2.0588,
      "step": 6371
    },
    {
      "epoch": 0.9898252427184466,
      "grad_norm": 0.0,
      "learning_rate": 2.6744906039716243e-07,
      "loss": 2.0258,
      "step": 6372
    },
    {
      "epoch": 0.9899805825242718,
      "grad_norm": 0.0,
      "learning_rate": 2.59283865927995e-07,
      "loss": 2.2047,
      "step": 6373
    },
    {
      "epoch": 0.9901359223300971,
      "grad_norm": 0.0,
      "learning_rate": 2.5124522019170923e-07,
      "loss": 2.11,
      "step": 6374
    },
    {
      "epoch": 0.9902912621359223,
      "grad_norm": 0.0,
      "learning_rate": 2.433331252240101e-07,
      "loss": 2.0426,
      "step": 6375
    },
    {
      "epoch": 0.9904466019417476,
      "grad_norm": 0.0,
      "learning_rate": 2.3554758302846145e-07,
      "loss": 2.1246,
      "step": 6376
    },
    {
      "epoch": 0.9906019417475728,
      "grad_norm": 0.0,
      "learning_rate": 2.2788859557654195e-07,
      "loss": 2.0504,
      "step": 6377
    },
    {
      "epoch": 0.990757281553398,
      "grad_norm": 0.0,
      "learning_rate": 2.2035616480775566e-07,
      "loss": 2.1267,
      "step": 6378
    },
    {
      "epoch": 0.9909126213592233,
      "grad_norm": 0.0,
      "learning_rate": 2.1295029262957676e-07,
      "loss": 2.0447,
      "step": 6379
    },
    {
      "epoch": 0.9910679611650486,
      "grad_norm": 0.0,
      "learning_rate": 2.0567098091728297e-07,
      "loss": 2.1446,
      "step": 6380
    },
    {
      "epoch": 0.9912233009708737,
      "grad_norm": 0.0,
      "learning_rate": 1.985182315142331e-07,
      "loss": 2.056,
      "step": 6381
    },
    {
      "epoch": 0.991378640776699,
      "grad_norm": 0.0,
      "learning_rate": 1.9149204623175598e-07,
      "loss": 1.9742,
      "step": 6382
    },
    {
      "epoch": 0.9915339805825243,
      "grad_norm": 0.0,
      "learning_rate": 1.8459242684903954e-07,
      "loss": 2.082,
      "step": 6383
    },
    {
      "epoch": 0.9916893203883496,
      "grad_norm": 0.0,
      "learning_rate": 1.7781937511335278e-07,
      "loss": 2.0965,
      "step": 6384
    },
    {
      "epoch": 0.9918446601941747,
      "grad_norm": 0.0,
      "learning_rate": 1.711728927397127e-07,
      "loss": 1.935,
      "step": 6385
    },
    {
      "epoch": 0.992,
      "grad_norm": 0.0,
      "learning_rate": 1.6465298141121742e-07,
      "loss": 1.9352,
      "step": 6386
    },
    {
      "epoch": 0.9921553398058253,
      "grad_norm": 0.0,
      "learning_rate": 1.5825964277899062e-07,
      "loss": 2.0707,
      "step": 6387
    },
    {
      "epoch": 0.9923106796116505,
      "grad_norm": 0.0,
      "learning_rate": 1.5199287846190402e-07,
      "loss": 2.1583,
      "step": 6388
    },
    {
      "epoch": 0.9924660194174757,
      "grad_norm": 0.0,
      "learning_rate": 1.4585269004691037e-07,
      "loss": 2.0468,
      "step": 6389
    },
    {
      "epoch": 0.992621359223301,
      "grad_norm": 0.0,
      "learning_rate": 1.3983907908893256e-07,
      "loss": 2.1173,
      "step": 6390
    },
    {
      "epoch": 0.9927766990291262,
      "grad_norm": 0.0,
      "learning_rate": 1.339520471107525e-07,
      "loss": 2.0842,
      "step": 6391
    },
    {
      "epoch": 0.9929320388349514,
      "grad_norm": 0.0,
      "learning_rate": 1.281915956031221e-07,
      "loss": 2.1285,
      "step": 6392
    },
    {
      "epoch": 0.9930873786407767,
      "grad_norm": 0.0,
      "learning_rate": 1.2255772602481896e-07,
      "loss": 2.2521,
      "step": 6393
    },
    {
      "epoch": 0.993242718446602,
      "grad_norm": 0.0,
      "learning_rate": 1.1705043980236862e-07,
      "loss": 2.1246,
      "step": 6394
    },
    {
      "epoch": 0.9933980582524272,
      "grad_norm": 0.0,
      "learning_rate": 1.1166973833048877e-07,
      "loss": 2.0281,
      "step": 6395
    },
    {
      "epoch": 0.9935533980582524,
      "grad_norm": 0.0,
      "learning_rate": 1.0641562297175611e-07,
      "loss": 2.0288,
      "step": 6396
    },
    {
      "epoch": 0.9937087378640777,
      "grad_norm": 0.0,
      "learning_rate": 1.012880950565509e-07,
      "loss": 2.051,
      "step": 6397
    },
    {
      "epoch": 0.9938640776699029,
      "grad_norm": 0.0,
      "learning_rate": 9.628715588333448e-08,
      "loss": 2.0587,
      "step": 6398
    },
    {
      "epoch": 0.9940194174757282,
      "grad_norm": 0.0,
      "learning_rate": 9.141280671853824e-08,
      "loss": 2.042,
      "step": 6399
    },
    {
      "epoch": 0.9941747572815534,
      "grad_norm": 0.0,
      "learning_rate": 8.666504879650816e-08,
      "loss": 2.1229,
      "step": 6400
    },
    {
      "epoch": 0.9943300970873786,
      "grad_norm": 0.0,
      "learning_rate": 8.204388331944923e-08,
      "loss": 2.1145,
      "step": 6401
    },
    {
      "epoch": 0.9944854368932039,
      "grad_norm": 0.0,
      "learning_rate": 7.754931145759203e-08,
      "loss": 2.1226,
      "step": 6402
    },
    {
      "epoch": 0.9946407766990292,
      "grad_norm": 0.0,
      "learning_rate": 7.31813343490817e-08,
      "loss": 1.9767,
      "step": 6403
    },
    {
      "epoch": 0.9947961165048543,
      "grad_norm": 0.0,
      "learning_rate": 6.893995310003342e-08,
      "loss": 2.0776,
      "step": 6404
    },
    {
      "epoch": 0.9949514563106796,
      "grad_norm": 0.0,
      "learning_rate": 6.482516878447697e-08,
      "loss": 2.0756,
      "step": 6405
    },
    {
      "epoch": 0.9951067961165049,
      "grad_norm": 0.0,
      "learning_rate": 6.083698244446767e-08,
      "loss": 1.9843,
      "step": 6406
    },
    {
      "epoch": 0.9952621359223301,
      "grad_norm": 0.0,
      "learning_rate": 5.697539508980887e-08,
      "loss": 2.015,
      "step": 6407
    },
    {
      "epoch": 0.9954174757281553,
      "grad_norm": 0.0,
      "learning_rate": 5.3240407698385004e-08,
      "loss": 2.0153,
      "step": 6408
    },
    {
      "epoch": 0.9955728155339806,
      "grad_norm": 0.0,
      "learning_rate": 4.9632021216106104e-08,
      "loss": 2.0341,
      "step": 6409
    },
    {
      "epoch": 0.9957281553398059,
      "grad_norm": 0.0,
      "learning_rate": 4.6150236556574684e-08,
      "loss": 2.0275,
      "step": 6410
    },
    {
      "epoch": 0.995883495145631,
      "grad_norm": 0.0,
      "learning_rate": 4.27950546016409e-08,
      "loss": 2.1008,
      "step": 6411
    },
    {
      "epoch": 0.9960388349514563,
      "grad_norm": 0.0,
      "learning_rate": 3.9566476200791904e-08,
      "loss": 1.9081,
      "step": 6412
    },
    {
      "epoch": 0.9961941747572816,
      "grad_norm": 0.0,
      "learning_rate": 3.646450217165143e-08,
      "loss": 2.0128,
      "step": 6413
    },
    {
      "epoch": 0.9963495145631068,
      "grad_norm": 0.0,
      "learning_rate": 3.3489133299757776e-08,
      "loss": 2.1056,
      "step": 6414
    },
    {
      "epoch": 0.996504854368932,
      "grad_norm": 0.0,
      "learning_rate": 3.064037033850831e-08,
      "loss": 2.1135,
      "step": 6415
    },
    {
      "epoch": 0.9966601941747573,
      "grad_norm": 0.0,
      "learning_rate": 2.7918214009325926e-08,
      "loss": 1.9894,
      "step": 6416
    },
    {
      "epoch": 0.9968155339805825,
      "grad_norm": 0.0,
      "learning_rate": 2.5322665001492608e-08,
      "loss": 2.0598,
      "step": 6417
    },
    {
      "epoch": 0.9969708737864078,
      "grad_norm": 0.0,
      "learning_rate": 2.2853723972315888e-08,
      "loss": 2.1256,
      "step": 6418
    },
    {
      "epoch": 0.997126213592233,
      "grad_norm": 0.0,
      "learning_rate": 2.0511391547017867e-08,
      "loss": 2.1216,
      "step": 6419
    },
    {
      "epoch": 0.9972815533980582,
      "grad_norm": 0.0,
      "learning_rate": 1.8295668318735193e-08,
      "loss": 2.0258,
      "step": 6420
    },
    {
      "epoch": 0.9974368932038835,
      "grad_norm": 0.0,
      "learning_rate": 1.6206554848574585e-08,
      "loss": 1.9987,
      "step": 6421
    },
    {
      "epoch": 0.9975922330097088,
      "grad_norm": 0.0,
      "learning_rate": 1.4244051665446288e-08,
      "loss": 1.9587,
      "step": 6422
    },
    {
      "epoch": 0.997747572815534,
      "grad_norm": 0.0,
      "learning_rate": 1.2408159266452668e-08,
      "loss": 1.9968,
      "step": 6423
    },
    {
      "epoch": 0.9979029126213592,
      "grad_norm": 0.0,
      "learning_rate": 1.0698878116388588e-08,
      "loss": 2.063,
      "step": 6424
    },
    {
      "epoch": 0.9980582524271845,
      "grad_norm": 0.0,
      "learning_rate": 9.11620864818552e-09,
      "loss": 2.0849,
      "step": 6425
    },
    {
      "epoch": 0.9982135922330098,
      "grad_norm": 0.0,
      "learning_rate": 7.660151262522952e-09,
      "loss": 2.0887,
      "step": 6426
    },
    {
      "epoch": 0.9983689320388349,
      "grad_norm": 0.0,
      "learning_rate": 6.330706328216973e-09,
      "loss": 1.8812,
      "step": 6427
    },
    {
      "epoch": 0.9985242718446602,
      "grad_norm": 0.0,
      "learning_rate": 5.1278741818316934e-09,
      "loss": 1.931,
      "step": 6428
    },
    {
      "epoch": 0.9986796116504855,
      "grad_norm": 0.0,
      "learning_rate": 4.051655128067822e-09,
      "loss": 2.0565,
      "step": 6429
    },
    {
      "epoch": 0.9988349514563106,
      "grad_norm": 0.0,
      "learning_rate": 3.1020494393185773e-09,
      "loss": 2.0615,
      "step": 6430
    },
    {
      "epoch": 0.9989902912621359,
      "grad_norm": 0.0,
      "learning_rate": 2.2790573561692894e-09,
      "loss": 2.0458,
      "step": 6431
    },
    {
      "epoch": 0.9991456310679612,
      "grad_norm": 0.0,
      "learning_rate": 1.5826790869533092e-09,
      "loss": 2.0276,
      "step": 6432
    },
    {
      "epoch": 0.9993009708737864,
      "grad_norm": 0.0,
      "learning_rate": 1.0129148080295636e-09,
      "loss": 2.0639,
      "step": 6433
    },
    {
      "epoch": 0.9994563106796116,
      "grad_norm": 0.0,
      "learning_rate": 5.697646636715348e-10,
      "loss": 2.1142,
      "step": 6434
    },
    {
      "epoch": 0.9996116504854369,
      "grad_norm": 0.0,
      "learning_rate": 2.5322876612277055e-10,
      "loss": 2.0555,
      "step": 6435
    },
    {
      "epoch": 0.9997669902912621,
      "grad_norm": 0.0,
      "learning_rate": 6.330719554137332e-11,
      "loss": 2.1106,
      "step": 6436
    },
    {
      "epoch": 0.9999223300970874,
      "grad_norm": 0.0,
      "learning_rate": 0.0,
      "loss": 2.0714,
      "step": 6437
    },
    {
      "epoch": 0.9999223300970874,
      "step": 6437,
      "total_flos": 3.935122096807477e+18,
      "train_loss": 2.2941150721582138,
      "train_runtime": 26170.1652,
      "train_samples_per_second": 62.971,
      "train_steps_per_second": 0.246
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 6437,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 2000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": false,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.935122096807477e+18,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}