{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.983922829581994,
  "eval_steps": 500,
  "global_step": 775,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 2.1462105969148225e-06,
      "loss": 1.3918,
      "step": 1
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.292421193829645e-06,
      "loss": 1.5201,
      "step": 2
    },
    {
      "epoch": 0.02,
      "learning_rate": 6.438631790744467e-06,
      "loss": 1.4977,
      "step": 3
    },
    {
      "epoch": 0.03,
      "learning_rate": 8.58484238765929e-06,
      "loss": 1.324,
      "step": 4
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.0731052984574111e-05,
      "loss": 2.3381,
      "step": 5
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.2877263581488935e-05,
      "loss": 1.1859,
      "step": 6
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.5023474178403756e-05,
      "loss": 1.429,
      "step": 7
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.716968477531858e-05,
      "loss": 1.4656,
      "step": 8
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.93158953722334e-05,
      "loss": 1.2783,
      "step": 9
    },
    {
      "epoch": 0.06,
      "learning_rate": 2.1462105969148223e-05,
      "loss": 1.5272,
      "step": 10
    },
    {
      "epoch": 0.07,
      "learning_rate": 2.3608316566063045e-05,
      "loss": 1.6185,
      "step": 11
    },
    {
      "epoch": 0.08,
      "learning_rate": 2.575452716297787e-05,
      "loss": 1.9579,
      "step": 12
    },
    {
      "epoch": 0.08,
      "learning_rate": 2.790073775989269e-05,
      "loss": 1.717,
      "step": 13
    },
    {
      "epoch": 0.09,
      "learning_rate": 3.0046948356807513e-05,
      "loss": 1.5498,
      "step": 14
    },
    {
      "epoch": 0.1,
      "learning_rate": 3.219315895372234e-05,
      "loss": 1.5309,
      "step": 15
    },
    {
      "epoch": 0.1,
      "learning_rate": 3.433936955063716e-05,
      "loss": 1.2718,
      "step": 16
    },
    {
      "epoch": 0.11,
      "learning_rate": 3.648558014755198e-05,
      "loss": 1.4084,
      "step": 17
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.86317907444668e-05,
      "loss": 1.6688,
      "step": 18
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.0778001341381624e-05,
      "loss": 1.39,
      "step": 19
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.2924211938296446e-05,
      "loss": 1.2353,
      "step": 20
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.507042253521127e-05,
      "loss": 1.0272,
      "step": 21
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.721663313212609e-05,
      "loss": 1.3035,
      "step": 22
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.936284372904091e-05,
      "loss": 1.3463,
      "step": 23
    },
    {
      "epoch": 0.15,
      "learning_rate": 5.150905432595574e-05,
      "loss": 1.198,
      "step": 24
    },
    {
      "epoch": 0.16,
      "learning_rate": 5.365526492287056e-05,
      "loss": 1.3113,
      "step": 25
    },
    {
      "epoch": 0.17,
      "learning_rate": 5.580147551978538e-05,
      "loss": 1.3093,
      "step": 26
    },
    {
      "epoch": 0.17,
      "learning_rate": 5.7947686116700204e-05,
      "loss": 1.2044,
      "step": 27
    },
    {
      "epoch": 0.18,
      "learning_rate": 6.0093896713615026e-05,
      "loss": 1.2397,
      "step": 28
    },
    {
      "epoch": 0.19,
      "learning_rate": 6.224010731052985e-05,
      "loss": 1.5441,
      "step": 29
    },
    {
      "epoch": 0.19,
      "learning_rate": 6.438631790744468e-05,
      "loss": 1.2465,
      "step": 30
    },
    {
      "epoch": 0.2,
      "learning_rate": 6.653252850435949e-05,
      "loss": 1.4821,
      "step": 31
    },
    {
      "epoch": 0.21,
      "learning_rate": 6.867873910127432e-05,
      "loss": 1.1955,
      "step": 32
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.082494969818913e-05,
      "loss": 1.4909,
      "step": 33
    },
    {
      "epoch": 0.22,
      "learning_rate": 7.297116029510396e-05,
      "loss": 1.2483,
      "step": 34
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.511737089201878e-05,
      "loss": 1.5193,
      "step": 35
    },
    {
      "epoch": 0.23,
      "learning_rate": 7.72635814889336e-05,
      "loss": 1.4266,
      "step": 36
    },
    {
      "epoch": 0.24,
      "learning_rate": 7.940979208584842e-05,
      "loss": 1.3174,
      "step": 37
    },
    {
      "epoch": 0.24,
      "learning_rate": 8.155600268276325e-05,
      "loss": 1.2199,
      "step": 38
    },
    {
      "epoch": 0.25,
      "learning_rate": 8.370221327967808e-05,
      "loss": 1.1735,
      "step": 39
    },
    {
      "epoch": 0.26,
      "learning_rate": 8.584842387659289e-05,
      "loss": 1.1232,
      "step": 40
    },
    {
      "epoch": 0.26,
      "learning_rate": 8.799463447350772e-05,
      "loss": 1.2988,
      "step": 41
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.014084507042254e-05,
      "loss": 1.4124,
      "step": 42
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.228705566733736e-05,
      "loss": 1.1524,
      "step": 43
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.443326626425218e-05,
      "loss": 1.2417,
      "step": 44
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.657947686116701e-05,
      "loss": 1.1235,
      "step": 45
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.872568745808182e-05,
      "loss": 1.362,
      "step": 46
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.999992357809583e-05,
      "loss": 1.1252,
      "step": 47
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.99990842963134e-05,
      "loss": 1.291,
      "step": 48
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.999731892582272e-05,
      "loss": 1.1857,
      "step": 49
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.999462749932216e-05,
      "loss": 1.3472,
      "step": 50
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.999101006666263e-05,
      "loss": 1.3134,
      "step": 51
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.998646669484661e-05,
      "loss": 1.1774,
      "step": 52
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.998099746802699e-05,
      "loss": 1.2472,
      "step": 53
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.997460248750539e-05,
      "loss": 1.3122,
      "step": 54
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.996728187173034e-05,
      "loss": 1.1674,
      "step": 55
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.995903575629515e-05,
      "loss": 0.9907,
      "step": 56
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.994986429393529e-05,
      "loss": 1.012,
      "step": 57
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.993976765452562e-05,
      "loss": 1.1601,
      "step": 58
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.992874602507724e-05,
      "loss": 1.0428,
      "step": 59
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.991679960973404e-05,
      "loss": 1.1541,
      "step": 60
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.990392862976887e-05,
      "loss": 1.269,
      "step": 61
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.989013332357947e-05,
      "loss": 1.2457,
      "step": 62
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.987541394668407e-05,
      "loss": 1.0939,
      "step": 63
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.985977077171664e-05,
      "loss": 1.2349,
      "step": 64
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.984320408842183e-05,
      "loss": 1.3642,
      "step": 65
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.982571420364966e-05,
      "loss": 1.2793,
      "step": 66
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.980730144134971e-05,
      "loss": 1.2894,
      "step": 67
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.97879661425653e-05,
      "loss": 1.373,
      "step": 68
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.976770866542697e-05,
      "loss": 1.2428,
      "step": 69
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.974652938514607e-05,
      "loss": 1.0956,
      "step": 70
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.972442869400759e-05,
      "loss": 1.1131,
      "step": 71
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.970140700136306e-05,
      "loss": 1.1866,
      "step": 72
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.967746473362286e-05,
      "loss": 1.3707,
      "step": 73
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.965260233424838e-05,
      "loss": 1.3979,
      "step": 74
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.962682026374383e-05,
      "loss": 1.2627,
      "step": 75
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.960011899964766e-05,
      "loss": 1.3645,
      "step": 76
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.957249903652364e-05,
      "loss": 1.2458,
      "step": 77
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.954396088595192e-05,
      "loss": 1.3056,
      "step": 78
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.951450507651934e-05,
      "loss": 1.2052,
      "step": 79
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.948413215380974e-05,
      "loss": 1.2625,
      "step": 80
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.945284268039383e-05,
      "loss": 1.1666,
      "step": 81
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.94206372358188e-05,
      "loss": 1.1478,
      "step": 82
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.938751641659756e-05,
      "loss": 1.2515,
      "step": 83
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.935348083619765e-05,
      "loss": 1.1974,
      "step": 84
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.931853112502996e-05,
      "loss": 1.0865,
      "step": 85
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.9282667930437e-05,
      "loss": 1.1842,
      "step": 86
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.924589191668092e-05,
      "loss": 1.2834,
      "step": 87
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.920820376493123e-05,
      "loss": 1.6785,
      "step": 88
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.916960417325214e-05,
      "loss": 1.369,
      "step": 89
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.913009385658966e-05,
      "loss": 1.4058,
      "step": 90
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.90896735467584e-05,
      "loss": 1.0977,
      "step": 91
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.904834399242786e-05,
      "loss": 1.2097,
      "step": 92
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.900610595910875e-05,
      "loss": 1.3962,
      "step": 93
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.896296022913873e-05,
      "loss": 1.3058,
      "step": 94
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.891890760166788e-05,
      "loss": 1.1604,
      "step": 95
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.887394889264398e-05,
      "loss": 1.09,
      "step": 96
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.882808493479732e-05,
      "loss": 1.1185,
      "step": 97
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.878131657762535e-05,
      "loss": 1.1937,
      "step": 98
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.873364468737685e-05,
      "loss": 0.9899,
      "step": 99
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.8685070147036e-05,
      "loss": 1.1637,
      "step": 100
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.863559385630595e-05,
      "loss": 1.1169,
      "step": 101
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.858521673159215e-05,
      "loss": 1.3953,
      "step": 102
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.853393970598542e-05,
      "loss": 0.9958,
      "step": 103
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.848176372924462e-05,
      "loss": 1.1583,
      "step": 104
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.842868976777918e-05,
      "loss": 1.6086,
      "step": 105
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.837471880463093e-05,
      "loss": 1.1513,
      "step": 106
    },
    {
      "epoch": 0.69,
      "learning_rate": 9.831985183945625e-05,
      "loss": 1.3345,
      "step": 107
    },
    {
      "epoch": 0.69,
      "learning_rate": 9.826408988850725e-05,
      "loss": 1.1834,
      "step": 108
    },
    {
      "epoch": 0.7,
      "learning_rate": 9.820743398461312e-05,
      "loss": 1.1338,
      "step": 109
    },
    {
      "epoch": 0.71,
      "learning_rate": 9.814988517716094e-05,
      "loss": 1.2307,
      "step": 110
    },
    {
      "epoch": 0.71,
      "learning_rate": 9.809144453207624e-05,
      "loss": 1.3879,
      "step": 111
    },
    {
      "epoch": 0.72,
      "learning_rate": 9.803211313180325e-05,
      "loss": 1.0249,
      "step": 112
    },
    {
      "epoch": 0.73,
      "learning_rate": 9.797189207528494e-05,
      "loss": 1.1477,
      "step": 113
    },
    {
      "epoch": 0.73,
      "learning_rate": 9.791078247794254e-05,
      "loss": 1.3769,
      "step": 114
    },
    {
      "epoch": 0.74,
      "learning_rate": 9.78487854716549e-05,
      "loss": 1.2494,
      "step": 115
    },
    {
      "epoch": 0.75,
      "learning_rate": 9.778590220473762e-05,
      "loss": 1.3227,
      "step": 116
    },
    {
      "epoch": 0.75,
      "learning_rate": 9.772213384192173e-05,
      "loss": 1.2374,
      "step": 117
    },
    {
      "epoch": 0.76,
      "learning_rate": 9.765748156433203e-05,
      "loss": 1.3604,
      "step": 118
    },
    {
      "epoch": 0.77,
      "learning_rate": 9.759194656946536e-05,
      "loss": 1.1039,
      "step": 119
    },
    {
      "epoch": 0.77,
      "learning_rate": 9.752553007116831e-05,
      "loss": 1.549,
      "step": 120
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.745823329961483e-05,
      "loss": 1.0187,
      "step": 121
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.73900575012833e-05,
      "loss": 1.2503,
      "step": 122
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.732100393893362e-05,
      "loss": 1.3275,
      "step": 123
    },
    {
      "epoch": 0.8,
      "learning_rate": 9.725107389158371e-05,
      "loss": 1.2233,
      "step": 124
    },
    {
      "epoch": 0.8,
      "learning_rate": 9.718026865448581e-05,
      "loss": 1.146,
      "step": 125
    },
    {
      "epoch": 0.81,
      "learning_rate": 9.710858953910256e-05,
      "loss": 1.3118,
      "step": 126
    },
    {
      "epoch": 0.82,
      "learning_rate": 9.703603787308265e-05,
      "loss": 1.3473,
      "step": 127
    },
    {
      "epoch": 0.82,
      "learning_rate": 9.69626150002363e-05,
      "loss": 1.3639,
      "step": 128
    },
    {
      "epoch": 0.83,
      "learning_rate": 9.688832228051023e-05,
      "loss": 1.1583,
      "step": 129
    },
    {
      "epoch": 0.84,
      "learning_rate": 9.681316108996265e-05,
      "loss": 1.2255,
      "step": 130
    },
    {
      "epoch": 0.84,
      "learning_rate": 9.673713282073769e-05,
      "loss": 1.0625,
      "step": 131
    },
    {
      "epoch": 0.85,
      "learning_rate": 9.66602388810395e-05,
      "loss": 1.5252,
      "step": 132
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.658248069510639e-05,
      "loss": 1.2798,
      "step": 133
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.65038597031843e-05,
      "loss": 1.3908,
      "step": 134
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.642437736150014e-05,
      "loss": 1.2803,
      "step": 135
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.634403514223486e-05,
      "loss": 1.1103,
      "step": 136
    },
    {
      "epoch": 0.88,
      "learning_rate": 9.626283453349616e-05,
      "loss": 1.2024,
      "step": 137
    },
    {
      "epoch": 0.89,
      "learning_rate": 9.618077703929092e-05,
      "loss": 1.1211,
      "step": 138
    },
    {
      "epoch": 0.89,
      "learning_rate": 9.609786417949733e-05,
      "loss": 1.1217,
      "step": 139
    },
    {
      "epoch": 0.9,
      "learning_rate": 9.601409748983683e-05,
      "loss": 1.1492,
      "step": 140
    },
    {
      "epoch": 0.91,
      "learning_rate": 9.592947852184549e-05,
      "loss": 0.9924,
      "step": 141
    },
    {
      "epoch": 0.91,
      "learning_rate": 9.584400884284545e-05,
      "loss": 1.1671,
      "step": 142
    },
    {
      "epoch": 0.92,
      "learning_rate": 9.575769003591582e-05,
      "loss": 1.031,
      "step": 143
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.567052369986327e-05,
      "loss": 1.1853,
      "step": 144
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.558251144919262e-05,
      "loss": 1.4693,
      "step": 145
    },
    {
      "epoch": 0.94,
      "learning_rate": 9.549365491407672e-05,
      "loss": 1.4138,
      "step": 146
    },
    {
      "epoch": 0.95,
      "learning_rate": 9.54039557403264e-05,
      "loss": 1.3903,
      "step": 147
    },
    {
      "epoch": 0.95,
      "learning_rate": 9.531341558935994e-05,
      "loss": 1.1711,
      "step": 148
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.522203613817224e-05,
      "loss": 1.2242,
      "step": 149
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.51298190793039e-05,
      "loss": 1.1137,
      "step": 150
    },
    {
      "epoch": 0.97,
      "learning_rate": 9.503676612080971e-05,
      "loss": 1.1836,
      "step": 151
    },
    {
      "epoch": 0.98,
      "learning_rate": 9.494287898622713e-05,
      "loss": 1.3264,
      "step": 152
    },
    {
      "epoch": 0.98,
      "learning_rate": 9.484815941454432e-05,
      "loss": 1.4135,
      "step": 153
    },
    {
      "epoch": 0.99,
      "learning_rate": 9.475260916016792e-05,
      "loss": 1.1234,
      "step": 154
    },
    {
      "epoch": 1.0,
      "learning_rate": 9.465622999289056e-05,
      "loss": 1.1318,
      "step": 155
    },
    {
      "epoch": 1.0,
      "learning_rate": 9.455902369785814e-05,
      "loss": 0.9531,
      "step": 156
    },
    {
      "epoch": 1.01,
      "learning_rate": 9.446099207553663e-05,
      "loss": 1.1359,
      "step": 157
    },
    {
      "epoch": 1.02,
      "learning_rate": 9.436213694167888e-05,
      "loss": 0.9278,
      "step": 158
    },
    {
      "epoch": 1.02,
      "learning_rate": 9.42624601272909e-05,
      "loss": 1.0735,
      "step": 159
    },
    {
      "epoch": 1.03,
      "learning_rate": 9.416196347859792e-05,
      "loss": 1.1143,
      "step": 160
    },
    {
      "epoch": 1.04,
      "learning_rate": 9.406064885701028e-05,
      "loss": 1.0445,
      "step": 161
    },
    {
      "epoch": 1.04,
      "learning_rate": 9.395851813908888e-05,
      "loss": 1.1858,
      "step": 162
    },
    {
      "epoch": 1.05,
      "learning_rate": 9.385557321651046e-05,
      "loss": 1.142,
      "step": 163
    },
    {
      "epoch": 1.05,
      "learning_rate": 9.375181599603255e-05,
      "loss": 1.1216,
      "step": 164
    },
    {
      "epoch": 1.06,
      "learning_rate": 9.364724839945815e-05,
      "loss": 1.2238,
      "step": 165
    },
    {
      "epoch": 1.07,
      "learning_rate": 9.354187236360014e-05,
      "loss": 1.2372,
      "step": 166
    },
    {
      "epoch": 1.07,
      "learning_rate": 9.34356898402454e-05,
      "loss": 1.3738,
      "step": 167
    },
    {
      "epoch": 1.08,
      "learning_rate": 9.332870279611868e-05,
      "loss": 1.2227,
      "step": 168
    },
    {
      "epoch": 1.09,
      "learning_rate": 9.322091321284611e-05,
      "loss": 1.2487,
      "step": 169
    },
    {
      "epoch": 1.09,
      "learning_rate": 9.31123230869186e-05,
      "loss": 1.1387,
      "step": 170
    },
    {
      "epoch": 1.1,
      "learning_rate": 9.300293442965478e-05,
      "loss": 0.908,
      "step": 171
    },
    {
      "epoch": 1.11,
      "learning_rate": 9.289274926716374e-05,
      "loss": 1.2106,
      "step": 172
    },
    {
      "epoch": 1.11,
      "learning_rate": 9.278176964030756e-05,
      "loss": 1.3412,
      "step": 173
    },
    {
      "epoch": 1.12,
      "learning_rate": 9.266999760466349e-05,
      "loss": 1.2203,
      "step": 174
    },
    {
      "epoch": 1.13,
      "learning_rate": 9.255743523048581e-05,
      "loss": 1.1619,
      "step": 175
    },
    {
      "epoch": 1.13,
      "learning_rate": 9.244408460266763e-05,
      "loss": 0.9922,
      "step": 176
    },
    {
      "epoch": 1.14,
      "learning_rate": 9.232994782070207e-05,
      "loss": 1.032,
      "step": 177
    },
    {
      "epoch": 1.14,
      "learning_rate": 9.221502699864358e-05,
      "loss": 1.1281,
      "step": 178
    },
    {
      "epoch": 1.15,
      "learning_rate": 9.209932426506867e-05,
      "loss": 1.0854,
      "step": 179
    },
    {
      "epoch": 1.16,
      "learning_rate": 9.198284176303645e-05,
      "loss": 1.0293,
      "step": 180
    },
    {
      "epoch": 1.16,
      "learning_rate": 9.186558165004906e-05,
      "loss": 1.2615,
      "step": 181
    },
    {
      "epoch": 1.17,
      "learning_rate": 9.174754609801159e-05,
      "loss": 0.9656,
      "step": 182
    },
    {
      "epoch": 1.18,
      "learning_rate": 9.162873729319192e-05,
      "loss": 1.2068,
      "step": 183
    },
    {
      "epoch": 1.18,
      "learning_rate": 9.150915743618023e-05,
      "loss": 1.446,
      "step": 184
    },
    {
      "epoch": 1.19,
      "learning_rate": 9.138880874184816e-05,
      "loss": 1.2485,
      "step": 185
    },
    {
      "epoch": 1.2,
      "learning_rate": 9.126769343930788e-05,
      "loss": 1.0235,
      "step": 186
    },
    {
      "epoch": 1.2,
      "learning_rate": 9.114581377187075e-05,
      "loss": 1.3166,
      "step": 187
    },
    {
      "epoch": 1.21,
      "learning_rate": 9.10231719970058e-05,
      "loss": 1.3087,
      "step": 188
    },
    {
      "epoch": 1.22,
      "learning_rate": 9.08997703862979e-05,
      "loss": 1.35,
      "step": 189
    },
    {
      "epoch": 1.22,
      "learning_rate": 9.077561122540565e-05,
      "loss": 1.1339,
      "step": 190
    },
    {
      "epoch": 1.23,
      "learning_rate": 9.06506968140191e-05,
      "loss": 1.4947,
      "step": 191
    },
    {
      "epoch": 1.23,
      "learning_rate": 9.052502946581717e-05,
      "loss": 1.1739,
      "step": 192
    },
    {
      "epoch": 1.24,
      "learning_rate": 9.039861150842472e-05,
      "loss": 1.3176,
      "step": 193
    },
    {
      "epoch": 1.25,
      "learning_rate": 9.027144528336949e-05,
      "loss": 0.973,
      "step": 194
    },
    {
      "epoch": 1.25,
      "learning_rate": 9.014353314603872e-05,
      "loss": 1.1226,
      "step": 195
    },
    {
      "epoch": 1.26,
      "learning_rate": 9.001487746563551e-05,
      "loss": 1.1688,
      "step": 196
    },
    {
      "epoch": 1.27,
      "learning_rate": 8.988548062513496e-05,
      "loss": 1.1403,
      "step": 197
    },
    {
      "epoch": 1.27,
      "learning_rate": 8.975534502124001e-05,
      "loss": 1.216,
      "step": 198
    },
    {
      "epoch": 1.28,
      "learning_rate": 8.962447306433713e-05,
      "loss": 1.013,
      "step": 199
    },
    {
      "epoch": 1.29,
      "learning_rate": 8.949286717845147e-05,
      "loss": 0.9083,
      "step": 200
    },
    {
      "epoch": 1.29,
      "learning_rate": 8.93605298012022e-05,
      "loss": 1.2401,
      "step": 201
    },
    {
      "epoch": 1.3,
      "learning_rate": 8.922746338375725e-05,
      "loss": 1.0934,
      "step": 202
    },
    {
      "epoch": 1.31,
      "learning_rate": 8.909367039078784e-05,
      "loss": 1.0935,
      "step": 203
    },
    {
      "epoch": 1.31,
      "learning_rate": 8.8959153300423e-05,
      "loss": 1.1181,
      "step": 204
    },
    {
      "epoch": 1.32,
      "learning_rate": 8.882391460420352e-05,
      "loss": 1.2044,
      "step": 205
    },
    {
      "epoch": 1.32,
      "learning_rate": 8.868795680703585e-05,
      "loss": 1.246,
      "step": 206
    },
    {
      "epoch": 1.33,
      "learning_rate": 8.855128242714572e-05,
      "loss": 1.0787,
      "step": 207
    },
    {
      "epoch": 1.34,
      "learning_rate": 8.841389399603151e-05,
      "loss": 1.056,
      "step": 208
    },
    {
      "epoch": 1.34,
      "learning_rate": 8.827579405841733e-05,
      "loss": 1.0761,
      "step": 209
    },
    {
      "epoch": 1.35,
      "learning_rate": 8.813698517220583e-05,
      "loss": 1.3278,
      "step": 210
    },
    {
      "epoch": 1.36,
      "learning_rate": 8.799746990843099e-05,
      "loss": 0.8407,
      "step": 211
    },
    {
      "epoch": 1.36,
      "learning_rate": 8.785725085121032e-05,
      "loss": 0.9841,
      "step": 212
    },
    {
      "epoch": 1.37,
      "learning_rate": 8.771633059769711e-05,
      "loss": 0.907,
      "step": 213
    },
    {
      "epoch": 1.38,
      "learning_rate": 8.757471175803227e-05,
      "loss": 0.9311,
      "step": 214
    },
    {
      "epoch": 1.38,
      "learning_rate": 8.7432396955296e-05,
      "loss": 1.0969,
      "step": 215
    },
    {
      "epoch": 1.39,
      "learning_rate": 8.72893888254592e-05,
      "loss": 1.0037,
      "step": 216
    },
    {
      "epoch": 1.4,
      "learning_rate": 8.714569001733469e-05,
      "loss": 1.2869,
      "step": 217
    },
    {
      "epoch": 1.4,
      "learning_rate": 8.700130319252805e-05,
      "loss": 0.9911,
      "step": 218
    },
    {
      "epoch": 1.41,
      "learning_rate": 8.685623102538847e-05,
      "loss": 1.0296,
      "step": 219
    },
    {
      "epoch": 1.41,
      "learning_rate": 8.671047620295904e-05,
      "loss": 1.1397,
      "step": 220
    },
    {
      "epoch": 1.42,
      "learning_rate": 8.656404142492711e-05,
      "loss": 1.3602,
      "step": 221
    },
    {
      "epoch": 1.43,
      "learning_rate": 8.641692940357426e-05,
      "loss": 1.0019,
      "step": 222
    },
    {
      "epoch": 1.43,
      "learning_rate": 8.626914286372603e-05,
      "loss": 1.1806,
      "step": 223
    },
    {
      "epoch": 1.44,
      "learning_rate": 8.612068454270144e-05,
      "loss": 1.0076,
      "step": 224
    },
    {
      "epoch": 1.45,
      "learning_rate": 8.597155719026236e-05,
      "loss": 1.1405,
      "step": 225
    },
    {
      "epoch": 1.45,
      "learning_rate": 8.582176356856254e-05,
      "loss": 0.9577,
      "step": 226
    },
    {
      "epoch": 1.46,
      "learning_rate": 8.567130645209638e-05,
      "loss": 0.9784,
      "step": 227
    },
    {
      "epoch": 1.47,
      "learning_rate": 8.55201886276477e-05,
      "loss": 1.0341,
      "step": 228
    },
    {
      "epoch": 1.47,
      "learning_rate": 8.536841289423794e-05,
      "loss": 1.3299,
      "step": 229
    },
    {
      "epoch": 1.48,
      "learning_rate": 8.521598206307446e-05,
      "loss": 1.0974,
      "step": 230
    },
    {
      "epoch": 1.49,
      "learning_rate": 8.50628989574984e-05,
      "loss": 1.319,
      "step": 231
    },
    {
      "epoch": 1.49,
      "learning_rate": 8.490916641293238e-05,
      "loss": 1.0588,
      "step": 232
    },
    {
      "epoch": 1.5,
      "learning_rate": 8.475478727682805e-05,
      "loss": 1.2618,
      "step": 233
    },
    {
      "epoch": 1.5,
      "learning_rate": 8.459976440861325e-05,
      "loss": 1.117,
      "step": 234
    },
    {
      "epoch": 1.51,
      "learning_rate": 8.444410067963913e-05,
      "loss": 0.9891,
      "step": 235
    },
    {
      "epoch": 1.52,
      "learning_rate": 8.428779897312692e-05,
      "loss": 1.1294,
      "step": 236
    },
    {
      "epoch": 1.52,
      "learning_rate": 8.413086218411456e-05,
      "loss": 1.0513,
      "step": 237
    },
    {
      "epoch": 1.53,
      "learning_rate": 8.397329321940304e-05,
      "loss": 0.9374,
      "step": 238
    },
    {
      "epoch": 1.54,
      "learning_rate": 8.381509499750259e-05,
      "loss": 1.1192,
      "step": 239
    },
    {
      "epoch": 1.54,
      "learning_rate": 8.365627044857863e-05,
      "loss": 1.0055,
      "step": 240
    },
    {
      "epoch": 1.55,
      "learning_rate": 8.349682251439743e-05,
      "loss": 1.0019,
      "step": 241
    },
    {
      "epoch": 1.56,
      "learning_rate": 8.333675414827173e-05,
      "loss": 1.0185,
      "step": 242
    },
    {
      "epoch": 1.56,
      "learning_rate": 8.317606831500595e-05,
      "loss": 1.4135,
      "step": 243
    },
    {
      "epoch": 1.57,
      "learning_rate": 8.301476799084133e-05,
      "loss": 1.2964,
      "step": 244
    },
    {
      "epoch": 1.58,
      "learning_rate": 8.285285616340073e-05,
      "loss": 1.3921,
      "step": 245
    },
    {
      "epoch": 1.58,
      "learning_rate": 8.269033583163339e-05,
      "loss": 1.0818,
      "step": 246
    },
    {
      "epoch": 1.59,
      "learning_rate": 8.252721000575934e-05,
      "loss": 1.0422,
      "step": 247
    },
    {
      "epoch": 1.59,
      "learning_rate": 8.236348170721358e-05,
      "loss": 1.1379,
      "step": 248
    },
    {
      "epoch": 1.6,
      "learning_rate": 8.219915396859025e-05,
      "loss": 1.0025,
      "step": 249
    },
    {
      "epoch": 1.61,
      "learning_rate": 8.203422983358637e-05,
      "loss": 1.1899,
      "step": 250
    },
    {
      "epoch": 1.61,
      "learning_rate": 8.186871235694544e-05,
      "loss": 1.0156,
      "step": 251
    },
    {
      "epoch": 1.62,
      "learning_rate": 8.170260460440094e-05,
      "loss": 0.8795,
      "step": 252
    },
    {
      "epoch": 1.63,
      "learning_rate": 8.153590965261948e-05,
      "loss": 0.8831,
      "step": 253
    },
    {
      "epoch": 1.63,
      "learning_rate": 8.136863058914387e-05,
      "loss": 0.9319,
      "step": 254
    },
    {
      "epoch": 1.64,
      "learning_rate": 8.12007705123359e-05,
      "loss": 1.0182,
      "step": 255
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.103233253131895e-05,
      "loss": 0.9404,
      "step": 256
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.086331976592037e-05,
      "loss": 1.0921,
      "step": 257
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.069373534661377e-05,
      "loss": 1.0495,
      "step": 258
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.052358241446101e-05,
      "loss": 0.9609,
      "step": 259
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.035286412105399e-05,
      "loss": 1.3647,
      "step": 260
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.01815836284563e-05,
      "loss": 1.1575,
      "step": 261
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.000974410914461e-05,
      "loss": 1.0428,
      "step": 262
    },
    {
      "epoch": 1.69,
      "learning_rate": 7.983734874595001e-05,
      "loss": 1.0343,
      "step": 263
    },
    {
      "epoch": 1.7,
      "learning_rate": 7.966440073199896e-05,
      "loss": 0.9319,
      "step": 264
    },
    {
      "epoch": 1.7,
      "learning_rate": 7.949090327065418e-05,
      "loss": 1.1453,
      "step": 265
    },
    {
      "epoch": 1.71,
      "learning_rate": 7.931685957545534e-05,
      "loss": 1.2479,
      "step": 266
    },
    {
      "epoch": 1.72,
      "learning_rate": 7.914227287005947e-05,
      "loss": 0.9517,
      "step": 267
    },
    {
      "epoch": 1.72,
      "learning_rate": 7.89671463881813e-05,
      "loss": 0.9126,
      "step": 268
    },
    {
      "epoch": 1.73,
      "learning_rate": 7.879148337353343e-05,
      "loss": 1.1086,
      "step": 269
    },
    {
      "epoch": 1.74,
      "learning_rate": 7.861528707976613e-05,
      "loss": 1.0535,
      "step": 270
    },
    {
      "epoch": 1.74,
      "learning_rate": 7.84385607704071e-05,
      "loss": 1.0794,
      "step": 271
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.826130771880111e-05,
      "loss": 0.9526,
      "step": 272
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.808353120804926e-05,
      "loss": 1.1436,
      "step": 273
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.790523453094823e-05,
      "loss": 0.923,
      "step": 274
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.772642098992931e-05,
      "loss": 1.2082,
      "step": 275
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.754709389699719e-05,
      "loss": 1.0869,
      "step": 276
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.736725657366858e-05,
      "loss": 1.0646,
      "step": 277
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.71869123509108e-05,
      "loss": 1.1303,
      "step": 278
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.700606456908e-05,
      "loss": 1.1093,
      "step": 279
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.682471657785926e-05,
      "loss": 1.0458,
      "step": 280
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.664287173619666e-05,
      "loss": 1.1031,
      "step": 281
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.6460533412243e-05,
      "loss": 0.989,
      "step": 282
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.627770498328934e-05,
      "loss": 1.2809,
      "step": 283
    },
    {
      "epoch": 1.83,
      "learning_rate": 7.60943898357046e-05,
      "loss": 1.1159,
      "step": 284
    },
    {
      "epoch": 1.83,
      "learning_rate": 7.591059136487273e-05,
      "loss": 1.0914,
      "step": 285
    },
    {
      "epoch": 1.84,
      "learning_rate": 7.572631297512985e-05,
      "loss": 0.935,
      "step": 286
    },
    {
      "epoch": 1.85,
      "learning_rate": 7.554155807970115e-05,
      "loss": 1.2702,
      "step": 287
    },
    {
      "epoch": 1.85,
      "learning_rate": 7.535633010063778e-05,
      "loss": 1.0784,
      "step": 288
    },
    {
      "epoch": 1.86,
      "learning_rate": 7.517063246875338e-05,
      "loss": 0.9567,
      "step": 289
    },
    {
      "epoch": 1.86,
      "learning_rate": 7.498446862356053e-05,
      "loss": 1.2246,
      "step": 290
    },
    {
      "epoch": 1.87,
      "learning_rate": 7.479784201320708e-05,
      "loss": 0.8282,
      "step": 291
    },
    {
      "epoch": 1.88,
      "learning_rate": 7.461075609441225e-05,
      "loss": 1.0822,
      "step": 292
    },
    {
      "epoch": 1.88,
      "learning_rate": 7.442321433240267e-05,
      "loss": 0.9024,
      "step": 293
    },
    {
      "epoch": 1.89,
      "learning_rate": 7.423522020084807e-05,
      "loss": 1.0324,
      "step": 294
    },
    {
      "epoch": 1.9,
      "learning_rate": 7.404677718179715e-05,
      "loss": 0.8521,
      "step": 295
    },
    {
      "epoch": 1.9,
      "learning_rate": 7.38578887656128e-05,
      "loss": 1.0843,
      "step": 296
    },
    {
      "epoch": 1.91,
      "learning_rate": 7.366855845090771e-05,
      "loss": 0.93,
      "step": 297
    },
    {
      "epoch": 1.92,
      "learning_rate": 7.347878974447944e-05,
      "loss": 0.8364,
      "step": 298
    },
    {
      "epoch": 1.92,
      "learning_rate": 7.328858616124547e-05,
      "loss": 0.9763,
      "step": 299
    },
    {
      "epoch": 1.93,
      "learning_rate": 7.309795122417817e-05,
      "loss": 1.0786,
      "step": 300
    },
    {
      "epoch": 1.94,
      "learning_rate": 7.29068884642394e-05,
      "loss": 1.222,
      "step": 301
    },
    {
      "epoch": 1.94,
      "learning_rate": 7.271540142031533e-05,
      "loss": 1.2857,
      "step": 302
    },
    {
      "epoch": 1.95,
      "learning_rate": 7.252349363915066e-05,
      "loss": 1.0518,
      "step": 303
    },
    {
      "epoch": 1.95,
      "learning_rate": 7.233116867528306e-05,
      "loss": 0.9627,
      "step": 304
    },
    {
      "epoch": 1.96,
      "learning_rate": 7.213843009097734e-05,
      "loss": 1.1409,
      "step": 305
    },
    {
      "epoch": 1.97,
      "learning_rate": 7.19452814561594e-05,
      "loss": 0.9567,
      "step": 306
    },
    {
      "epoch": 1.97,
      "learning_rate": 7.175172634835016e-05,
      "loss": 1.0637,
      "step": 307
    },
    {
      "epoch": 1.98,
      "learning_rate": 7.155776835259926e-05,
      "loss": 1.1823,
      "step": 308
    },
    {
      "epoch": 1.99,
      "learning_rate": 7.136341106141871e-05,
      "loss": 1.1065,
      "step": 309
    },
    {
      "epoch": 1.99,
      "learning_rate": 7.116865807471628e-05,
      "loss": 0.8034,
      "step": 310
    },
    {
      "epoch": 2.0,
      "learning_rate": 7.097351299972886e-05,
      "loss": 0.9012,
      "step": 311
    },
    {
      "epoch": 2.01,
      "learning_rate": 7.077797945095569e-05,
      "loss": 0.6948,
      "step": 312
    },
    {
      "epoch": 2.01,
      "learning_rate": 7.058206105009129e-05,
      "loss": 0.9688,
      "step": 313
    },
    {
      "epoch": 2.02,
      "learning_rate": 7.038576142595852e-05,
      "loss": 0.7873,
      "step": 314
    },
    {
      "epoch": 2.03,
      "learning_rate": 7.018908421444126e-05,
      "loss": 0.8442,
      "step": 315
    },
    {
      "epoch": 2.03,
      "learning_rate": 6.999203305841713e-05,
      "loss": 0.9943,
      "step": 316
    },
    {
      "epoch": 2.04,
      "learning_rate": 6.979461160768998e-05,
      "loss": 0.8536,
      "step": 317
    },
    {
      "epoch": 2.05,
      "learning_rate": 6.95968235189223e-05,
      "loss": 0.9085,
      "step": 318
    },
    {
      "epoch": 2.05,
      "learning_rate": 6.939867245556751e-05,
      "loss": 1.0105,
      "step": 319
    },
    {
      "epoch": 2.06,
      "learning_rate": 6.920016208780207e-05,
      "loss": 0.9681,
      "step": 320
    },
    {
      "epoch": 2.06,
      "learning_rate": 6.900129609245749e-05,
      "loss": 1.0205,
      "step": 321
    },
    {
      "epoch": 2.07,
      "learning_rate": 6.880207815295233e-05,
      "loss": 1.0998,
      "step": 322
    },
    {
      "epoch": 2.08,
      "learning_rate": 6.86025119592238e-05,
      "loss": 1.1063,
      "step": 323
    },
    {
      "epoch": 2.08,
      "learning_rate": 6.840260120765961e-05,
      "loss": 0.889,
      "step": 324
    },
    {
      "epoch": 2.09,
      "learning_rate": 6.820234960102935e-05,
      "loss": 1.1392,
      "step": 325
    },
    {
      "epoch": 2.1,
      "learning_rate": 6.800176084841598e-05,
      "loss": 0.7726,
      "step": 326
    },
    {
      "epoch": 2.1,
      "learning_rate": 6.780083866514714e-05,
      "loss": 0.856,
      "step": 327
    },
    {
      "epoch": 2.11,
      "learning_rate": 6.759958677272625e-05,
      "loss": 1.0061,
      "step": 328
    },
    {
      "epoch": 2.12,
      "learning_rate": 6.739800889876372e-05,
      "loss": 1.0775,
      "step": 329
    },
    {
      "epoch": 2.12,
      "learning_rate": 6.719610877690775e-05,
      "loss": 0.9266,
      "step": 330
    },
    {
      "epoch": 2.13,
      "learning_rate": 6.699389014677533e-05,
      "loss": 1.0029,
      "step": 331
    },
    {
      "epoch": 2.14,
      "learning_rate": 6.679135675388281e-05,
      "loss": 0.7549,
      "step": 332
    },
    {
      "epoch": 2.14,
      "learning_rate": 6.65885123495767e-05,
      "loss": 0.8186,
      "step": 333
    },
    {
      "epoch": 2.15,
      "learning_rate": 6.638536069096402e-05,
      "loss": 0.9525,
      "step": 334
    },
    {
      "epoch": 2.15,
      "learning_rate": 6.618190554084284e-05,
      "loss": 0.8709,
      "step": 335
    },
    {
      "epoch": 2.16,
      "learning_rate": 6.597815066763252e-05,
      "loss": 0.8624,
      "step": 336
    },
    {
      "epoch": 2.17,
      "learning_rate": 6.577409984530391e-05,
      "loss": 0.918,
      "step": 337
    },
    {
      "epoch": 2.17,
      "learning_rate": 6.55697568533095e-05,
      "loss": 0.8944,
      "step": 338
    },
    {
      "epoch": 2.18,
      "learning_rate": 6.536512547651333e-05,
      "loss": 0.8615,
      "step": 339
    },
    {
      "epoch": 2.19,
      "learning_rate": 6.516020950512098e-05,
      "loss": 1.1556,
      "step": 340
    },
    {
      "epoch": 2.19,
      "learning_rate": 6.495501273460931e-05,
      "loss": 1.0322,
      "step": 341
    },
    {
      "epoch": 2.2,
      "learning_rate": 6.474953896565615e-05,
      "loss": 1.1249,
      "step": 342
    },
    {
      "epoch": 2.21,
      "learning_rate": 6.454379200406995e-05,
      "loss": 0.9073,
      "step": 343
    },
    {
      "epoch": 2.21,
      "learning_rate": 6.433777566071928e-05,
      "loss": 1.1746,
      "step": 344
    },
    {
      "epoch": 2.22,
      "learning_rate": 6.413149375146219e-05,
      "loss": 0.9572,
      "step": 345
    },
    {
      "epoch": 2.23,
      "learning_rate": 6.392495009707558e-05,
      "loss": 1.0916,
      "step": 346
    },
    {
      "epoch": 2.23,
      "learning_rate": 6.371814852318446e-05,
      "loss": 1.0927,
      "step": 347
    },
    {
      "epoch": 2.24,
      "learning_rate": 6.351109286019104e-05,
      "loss": 1.0753,
      "step": 348
    },
    {
      "epoch": 2.24,
      "learning_rate": 6.330378694320373e-05,
      "loss": 0.9198,
      "step": 349
    },
    {
      "epoch": 2.25,
      "learning_rate": 6.30962346119663e-05,
      "loss": 0.8238,
      "step": 350
    },
    {
      "epoch": 2.26,
      "learning_rate": 6.288843971078651e-05,
      "loss": 0.876,
      "step": 351
    },
    {
      "epoch": 2.26,
      "learning_rate": 6.268040608846509e-05,
      "loss": 0.8482,
      "step": 352
    },
    {
      "epoch": 2.27,
      "learning_rate": 6.24721375982244e-05,
      "loss": 1.0925,
      "step": 353
    },
    {
      "epoch": 2.28,
      "learning_rate": 6.226363809763701e-05,
      "loss": 0.7676,
      "step": 354
    },
    {
      "epoch": 2.28,
      "learning_rate": 6.205491144855432e-05,
      "loss": 0.9251,
      "step": 355
    },
    {
      "epoch": 2.29,
      "learning_rate": 6.184596151703498e-05,
      "loss": 0.8167,
      "step": 356
    },
    {
      "epoch": 2.3,
      "learning_rate": 6.163679217327333e-05,
      "loss": 0.9087,
      "step": 357
    },
    {
      "epoch": 2.3,
      "learning_rate": 6.142740729152765e-05,
      "loss": 0.7915,
      "step": 358
    },
    {
      "epoch": 2.31,
      "learning_rate": 6.121781075004846e-05,
      "loss": 0.9897,
      "step": 359
    },
    {
      "epoch": 2.32,
      "learning_rate": 6.1008006431006683e-05,
      "loss": 0.8618,
      "step": 360
    },
    {
      "epoch": 2.32,
      "learning_rate": 6.0797998220421694e-05,
      "loss": 1.019,
      "step": 361
    },
    {
      "epoch": 2.33,
      "learning_rate": 6.058779000808937e-05,
      "loss": 0.9897,
      "step": 362
    },
    {
      "epoch": 2.33,
      "learning_rate": 6.037738568751008e-05,
      "loss": 0.845,
      "step": 363
    },
    {
      "epoch": 2.34,
      "learning_rate": 6.0166789155816486e-05,
      "loss": 0.9398,
      "step": 364
    },
    {
      "epoch": 2.35,
      "learning_rate": 5.995600431370142e-05,
      "loss": 0.9563,
      "step": 365
    },
    {
      "epoch": 2.35,
      "learning_rate": 5.974503506534566e-05,
      "loss": 0.7918,
      "step": 366
    },
    {
      "epoch": 2.36,
      "learning_rate": 5.9533885318345504e-05,
      "loss": 0.6668,
      "step": 367
    },
    {
      "epoch": 2.37,
      "learning_rate": 5.932255898364053e-05,
      "loss": 0.6378,
      "step": 368
    },
    {
      "epoch": 2.37,
      "learning_rate": 5.911105997544105e-05,
      "loss": 0.8012,
      "step": 369
    },
    {
      "epoch": 2.38,
      "learning_rate": 5.8899392211155704e-05,
      "loss": 0.7455,
      "step": 370
    },
    {
      "epoch": 2.39,
      "learning_rate": 5.86875596113188e-05,
      "loss": 0.8352,
      "step": 371
    },
    {
      "epoch": 2.39,
      "learning_rate": 5.8475566099517765e-05,
      "loss": 0.9858,
      "step": 372
    },
    {
      "epoch": 2.4,
      "learning_rate": 5.826341560232047e-05,
      "loss": 0.8427,
      "step": 373
    },
    {
      "epoch": 2.41,
      "learning_rate": 5.805111204920248e-05,
      "loss": 0.7964,
      "step": 374
    },
    {
      "epoch": 2.41,
      "learning_rate": 5.783865937247427e-05,
      "loss": 0.9012,
      "step": 375
    },
    {
      "epoch": 2.42,
      "learning_rate": 5.762606150720842e-05,
      "loss": 0.9623,
      "step": 376
    },
    {
      "epoch": 2.42,
      "learning_rate": 5.741332239116668e-05,
      "loss": 0.9144,
      "step": 377
    },
    {
      "epoch": 2.43,
      "learning_rate": 5.720044596472709e-05,
      "loss": 0.8463,
      "step": 378
    },
    {
      "epoch": 2.44,
      "learning_rate": 5.6987436170810924e-05,
      "loss": 1.0161,
      "step": 379
    },
    {
      "epoch": 2.44,
      "learning_rate": 5.677429695480979e-05,
      "loss": 0.7847,
      "step": 380
    },
    {
      "epoch": 2.45,
      "learning_rate": 5.656103226451237e-05,
      "loss": 0.6766,
      "step": 381
    },
    {
      "epoch": 2.46,
      "learning_rate": 5.634764605003145e-05,
      "loss": 0.7335,
      "step": 382
    },
    {
      "epoch": 2.46,
      "learning_rate": 5.6134142263730705e-05,
      "loss": 0.7281,
      "step": 383
    },
    {
      "epoch": 2.47,
      "learning_rate": 5.592052486015145e-05,
      "loss": 1.0305,
      "step": 384
    },
    {
      "epoch": 2.48,
      "learning_rate": 5.5706797795939444e-05,
      "loss": 1.0515,
      "step": 385
    },
    {
      "epoch": 2.48,
      "learning_rate": 5.5492965029771646e-05,
      "loss": 0.9345,
      "step": 386
    },
    {
      "epoch": 2.49,
      "learning_rate": 5.527903052228277e-05,
      "loss": 0.8928,
      "step": 387
    },
    {
      "epoch": 2.5,
      "learning_rate": 5.5064998235992015e-05,
      "loss": 0.8983,
      "step": 388
    },
    {
      "epoch": 2.5,
      "learning_rate": 5.485087213522966e-05,
      "loss": 0.9771,
      "step": 389
    },
    {
      "epoch": 2.51,
      "learning_rate": 5.463665618606362e-05,
      "loss": 0.7857,
      "step": 390
    },
    {
      "epoch": 2.51,
      "learning_rate": 5.442235435622599e-05,
      "loss": 0.8757,
      "step": 391
    },
    {
      "epoch": 2.52,
      "learning_rate": 5.420797061503956e-05,
      "loss": 0.79,
      "step": 392
    },
    {
      "epoch": 2.53,
      "learning_rate": 5.39935089333443e-05,
      "loss": 0.7216,
      "step": 393
    },
    {
      "epoch": 2.53,
      "learning_rate": 5.37789732834238e-05,
      "loss": 0.855,
      "step": 394
    },
    {
      "epoch": 2.54,
      "learning_rate": 5.3564367638931665e-05,
      "loss": 0.7366,
      "step": 395
    },
    {
      "epoch": 2.55,
      "learning_rate": 5.3349695974818006e-05,
      "loss": 0.7464,
      "step": 396
    },
    {
      "epoch": 2.55,
      "learning_rate": 5.313496226725571e-05,
      "loss": 0.8212,
      "step": 397
    },
    {
      "epoch": 2.56,
      "learning_rate": 5.292017049356684e-05,
      "loss": 1.0271,
      "step": 398
    },
    {
      "epoch": 2.57,
      "learning_rate": 5.2705324632149e-05,
      "loss": 1.2483,
      "step": 399
    },
    {
      "epoch": 2.57,
      "learning_rate": 5.24904286624016e-05,
      "loss": 0.9307,
      "step": 400
    },
    {
      "epoch": 2.58,
      "learning_rate": 5.2275486564652145e-05,
      "loss": 1.0639,
      "step": 401
    },
    {
      "epoch": 2.59,
      "learning_rate": 5.206050232008255e-05,
      "loss": 0.7063,
      "step": 402
    },
    {
      "epoch": 2.59,
      "learning_rate": 5.184547991065536e-05,
      "loss": 0.8401,
      "step": 403
    },
    {
      "epoch": 2.6,
      "learning_rate": 5.1630423319040024e-05,
      "loss": 0.8321,
      "step": 404
    },
    {
      "epoch": 2.6,
      "learning_rate": 5.141533652853909e-05,
      "loss": 0.902,
      "step": 405
    },
    {
      "epoch": 2.61,
      "learning_rate": 5.1200223523014535e-05,
      "loss": 0.8816,
      "step": 406
    },
    {
      "epoch": 2.62,
      "learning_rate": 5.098508828681379e-05,
      "loss": 0.7039,
      "step": 407
    },
    {
      "epoch": 2.62,
      "learning_rate": 5.0769934804696116e-05,
      "loss": 0.6317,
      "step": 408
    },
    {
      "epoch": 2.63,
      "learning_rate": 5.055476706175869e-05,
      "loss": 0.7904,
      "step": 409
    },
    {
      "epoch": 2.64,
      "learning_rate": 5.033958904336285e-05,
      "loss": 0.6681,
      "step": 410
    },
    {
      "epoch": 2.64,
      "learning_rate": 5.012440473506023e-05,
      "loss": 0.7589,
      "step": 411
    },
    {
      "epoch": 2.65,
      "learning_rate": 4.9909218122519e-05,
      "loss": 0.7605,
      "step": 412
    },
    {
      "epoch": 2.66,
      "learning_rate": 4.969403319145002e-05,
      "loss": 0.9653,
      "step": 413
    },
    {
      "epoch": 2.66,
      "learning_rate": 4.947885392753295e-05,
      "loss": 0.6047,
      "step": 414
    },
    {
      "epoch": 2.67,
      "learning_rate": 4.9263684316342516e-05,
      "loss": 0.7779,
      "step": 415
    },
    {
      "epoch": 2.68,
      "learning_rate": 4.9048528343274643e-05,
      "loss": 1.1614,
      "step": 416
    },
    {
      "epoch": 2.68,
      "learning_rate": 4.8833389993472695e-05,
      "loss": 0.6775,
      "step": 417
    },
    {
      "epoch": 2.69,
      "learning_rate": 4.861827325175356e-05,
      "loss": 0.8557,
      "step": 418
    },
    {
      "epoch": 2.69,
      "learning_rate": 4.8403182102533916e-05,
      "loss": 0.8097,
      "step": 419
    },
    {
      "epoch": 2.7,
      "learning_rate": 4.818812052975642e-05,
      "loss": 0.7848,
      "step": 420
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.797309251681591e-05,
      "loss": 0.8386,
      "step": 421
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.77581020464856e-05,
      "loss": 0.856,
      "step": 422
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.7543153100843366e-05,
      "loss": 0.7061,
      "step": 423
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.732824966119798e-05,
      "loss": 0.6694,
      "step": 424
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.7113395708015286e-05,
      "loss": 0.9004,
      "step": 425
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.6898595220844574e-05,
      "loss": 0.6941,
      "step": 426
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.668385217824482e-05,
      "loss": 0.691,
      "step": 427
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.6469170557711006e-05,
      "loss": 0.7019,
      "step": 428
    },
    {
      "epoch": 2.76,
      "learning_rate": 4.6254554335600434e-05,
      "loss": 0.7591,
      "step": 429
    },
    {
      "epoch": 2.77,
      "learning_rate": 4.6040007487059136e-05,
      "loss": 0.6131,
      "step": 430
    },
    {
      "epoch": 2.77,
      "learning_rate": 4.582553398594815e-05,
      "loss": 1.0404,
      "step": 431
    },
    {
      "epoch": 2.78,
      "learning_rate": 4.561113780476998e-05,
      "loss": 0.6439,
      "step": 432
    },
    {
      "epoch": 2.78,
      "learning_rate": 4.5396822914595005e-05,
      "loss": 0.8139,
      "step": 433
    },
    {
      "epoch": 2.79,
      "learning_rate": 4.518259328498792e-05,
      "loss": 0.9081,
      "step": 434
    },
    {
      "epoch": 2.8,
      "learning_rate": 4.4968452883934205e-05,
      "loss": 0.8197,
      "step": 435
    },
    {
      "epoch": 2.8,
      "learning_rate": 4.475440567776663e-05,
      "loss": 0.7448,
      "step": 436
    },
    {
      "epoch": 2.81,
      "learning_rate": 4.4540455631091874e-05,
      "loss": 0.9074,
      "step": 437
    },
    {
      "epoch": 2.82,
      "learning_rate": 4.432660670671692e-05,
      "loss": 0.9331,
      "step": 438
    },
    {
      "epoch": 2.82,
      "learning_rate": 4.411286286557579e-05,
      "loss": 0.9756,
      "step": 439
    },
    {
      "epoch": 2.83,
      "learning_rate": 4.3899228066656165e-05,
      "loss": 0.76,
      "step": 440
    },
    {
      "epoch": 2.84,
      "learning_rate": 4.368570626692599e-05,
      "loss": 0.7538,
      "step": 441
    },
    {
      "epoch": 2.84,
      "learning_rate": 4.347230142126025e-05,
      "loss": 0.7114,
      "step": 442
    },
    {
      "epoch": 2.85,
      "learning_rate": 4.325901748236771e-05,
      "loss": 1.0386,
      "step": 443
    },
    {
      "epoch": 2.86,
      "learning_rate": 4.3045858400717666e-05,
      "loss": 0.7839,
      "step": 444
    },
    {
      "epoch": 2.86,
      "learning_rate": 4.2832828124466786e-05,
      "loss": 0.8511,
      "step": 445
    },
    {
      "epoch": 2.87,
      "learning_rate": 4.2619930599385996e-05,
      "loss": 0.7598,
      "step": 446
    },
    {
      "epoch": 2.87,
      "learning_rate": 4.2407169768787416e-05,
      "loss": 0.7567,
      "step": 447
    },
    {
      "epoch": 2.88,
      "learning_rate": 4.219454957345125e-05,
      "loss": 0.7668,
      "step": 448
    },
    {
      "epoch": 2.89,
      "learning_rate": 4.198207395155285e-05,
      "loss": 0.6951,
      "step": 449
    },
    {
      "epoch": 2.89,
      "learning_rate": 4.176974683858982e-05,
      "loss": 0.7152,
      "step": 450
    },
    {
      "epoch": 2.9,
      "learning_rate": 4.1557572167309e-05,
      "loss": 0.7178,
      "step": 451
    },
    {
      "epoch": 2.91,
      "learning_rate": 4.1345553867633694e-05,
      "loss": 0.6558,
      "step": 452
    },
    {
      "epoch": 2.91,
      "learning_rate": 4.113369586659091e-05,
      "loss": 0.7839,
      "step": 453
    },
    {
      "epoch": 2.92,
      "learning_rate": 4.092200208823855e-05,
      "loss": 0.5941,
      "step": 454
    },
    {
      "epoch": 2.93,
      "learning_rate": 4.071047645359276e-05,
      "loss": 0.7356,
      "step": 455
    },
    {
      "epoch": 2.93,
      "learning_rate": 4.0499122880555374e-05,
      "loss": 0.9164,
      "step": 456
    },
    {
      "epoch": 2.94,
      "learning_rate": 4.028794528384121e-05,
      "loss": 0.9154,
      "step": 457
    },
    {
      "epoch": 2.95,
      "learning_rate": 4.007694757490566e-05,
      "loss": 0.851,
      "step": 458
    },
    {
      "epoch": 2.95,
      "learning_rate": 3.986613366187221e-05,
      "loss": 0.7899,
      "step": 459
    },
    {
      "epoch": 2.96,
      "learning_rate": 3.965550744946005e-05,
      "loss": 0.84,
      "step": 460
    },
    {
      "epoch": 2.96,
      "learning_rate": 3.944507283891179e-05,
      "loss": 0.6701,
      "step": 461
    },
    {
      "epoch": 2.97,
      "learning_rate": 3.923483372792112e-05,
      "loss": 0.7656,
      "step": 462
    },
    {
      "epoch": 2.98,
      "learning_rate": 3.902479401056071e-05,
      "loss": 0.8818,
      "step": 463
    },
    {
      "epoch": 2.98,
      "learning_rate": 3.881495757721001e-05,
      "loss": 0.9264,
      "step": 464
    },
    {
      "epoch": 2.99,
      "learning_rate": 3.860532831448323e-05,
      "loss": 0.6381,
      "step": 465
    },
    {
      "epoch": 3.0,
      "learning_rate": 3.8395910105157355e-05,
      "loss": 0.6726,
      "step": 466
    },
    {
      "epoch": 3.0,
      "learning_rate": 3.81867068281002e-05,
      "loss": 0.6025,
      "step": 467
    },
    {
      "epoch": 3.01,
      "learning_rate": 3.797772235819854e-05,
      "loss": 0.649,
      "step": 468
    },
    {
      "epoch": 3.02,
      "learning_rate": 3.776896056628647e-05,
      "loss": 0.4989,
      "step": 469
    },
    {
      "epoch": 3.02,
      "learning_rate": 3.7560425319073566e-05,
      "loss": 0.7171,
      "step": 470
    },
    {
      "epoch": 3.03,
      "learning_rate": 3.7352120479073316e-05,
      "loss": 0.6515,
      "step": 471
    },
    {
      "epoch": 3.04,
      "learning_rate": 3.7144049904531566e-05,
      "loss": 0.6988,
      "step": 472
    },
    {
      "epoch": 3.04,
      "learning_rate": 3.693621744935511e-05,
      "loss": 0.6913,
      "step": 473
    },
    {
      "epoch": 3.05,
      "learning_rate": 3.672862696304024e-05,
      "loss": 0.6733,
      "step": 474
    },
    {
      "epoch": 3.05,
      "learning_rate": 3.6521282290601456e-05,
      "loss": 0.727,
      "step": 475
    },
    {
      "epoch": 3.06,
      "learning_rate": 3.631418727250033e-05,
      "loss": 0.8016,
      "step": 476
    },
    {
      "epoch": 3.07,
      "learning_rate": 3.610734574457425e-05,
      "loss": 0.6353,
      "step": 477
    },
    {
      "epoch": 3.07,
      "learning_rate": 3.5900761537965445e-05,
      "loss": 1.02,
      "step": 478
    },
    {
      "epoch": 3.08,
      "learning_rate": 3.569443847905002e-05,
      "loss": 0.752,
      "step": 479
    },
    {
      "epoch": 3.09,
      "learning_rate": 3.5488380389367055e-05,
      "loss": 0.8038,
      "step": 480
    },
    {
      "epoch": 3.09,
      "learning_rate": 3.5282591085547836e-05,
      "loss": 0.75,
      "step": 481
    },
    {
      "epoch": 3.1,
      "learning_rate": 3.5077074379245225e-05,
      "loss": 0.539,
      "step": 482
    },
    {
      "epoch": 3.11,
      "learning_rate": 3.487183407706293e-05,
      "loss": 0.739,
      "step": 483
    },
    {
      "epoch": 3.11,
      "learning_rate": 3.466687398048511e-05,
      "loss": 0.7921,
      "step": 484
    },
    {
      "epoch": 3.12,
      "learning_rate": 3.446219788580589e-05,
      "loss": 0.762,
      "step": 485
    },
    {
      "epoch": 3.13,
      "learning_rate": 3.425780958405913e-05,
      "loss": 0.7106,
      "step": 486
    },
    {
      "epoch": 3.13,
      "learning_rate": 3.405371286094811e-05,
      "loss": 0.6761,
      "step": 487
    },
    {
      "epoch": 3.14,
      "learning_rate": 3.384991149677546e-05,
      "loss": 0.563,
      "step": 488
    },
    {
      "epoch": 3.14,
      "learning_rate": 3.364640926637319e-05,
      "loss": 0.6492,
      "step": 489
    },
    {
      "epoch": 3.15,
      "learning_rate": 3.344320993903268e-05,
      "loss": 0.6396,
      "step": 490
    },
    {
      "epoch": 3.16,
      "learning_rate": 3.3240317278434896e-05,
      "loss": 0.5531,
      "step": 491
    },
    {
      "epoch": 3.16,
      "learning_rate": 3.303773504258071e-05,
      "loss": 0.7548,
      "step": 492
    },
    {
      "epoch": 3.17,
      "learning_rate": 3.283546698372125e-05,
      "loss": 0.5752,
      "step": 493
    },
    {
      "epoch": 3.18,
      "learning_rate": 3.2633516848288435e-05,
      "loss": 0.6402,
      "step": 494
    },
    {
      "epoch": 3.18,
      "learning_rate": 3.2431888376825596e-05,
      "loss": 0.9139,
      "step": 495
    },
    {
      "epoch": 3.19,
      "learning_rate": 3.223058530391813e-05,
      "loss": 0.8422,
      "step": 496
    },
    {
      "epoch": 3.2,
      "learning_rate": 3.202961135812437e-05,
      "loss": 0.623,
      "step": 497
    },
    {
      "epoch": 3.2,
      "learning_rate": 3.182897026190652e-05,
      "loss": 0.8291,
      "step": 498
    },
    {
      "epoch": 3.21,
      "learning_rate": 3.162866573156174e-05,
      "loss": 0.855,
      "step": 499
    },
    {
      "epoch": 3.22,
      "learning_rate": 3.1428701477153225e-05,
      "loss": 0.8827,
      "step": 500
    },
    {
      "epoch": 3.22,
      "learning_rate": 3.1229081202441556e-05,
      "loss": 0.6139,
      "step": 501
    },
    {
      "epoch": 3.23,
      "learning_rate": 3.102980860481614e-05,
      "loss": 0.9635,
      "step": 502
    },
    {
      "epoch": 3.23,
      "learning_rate": 3.0830887375226605e-05,
      "loss": 0.6944,
      "step": 503
    },
    {
      "epoch": 3.24,
      "learning_rate": 3.06323211981145e-05,
      "loss": 0.8644,
      "step": 504
    },
    {
      "epoch": 3.25,
      "learning_rate": 3.0434113751345095e-05,
      "loss": 0.5708,
      "step": 505
    },
    {
      "epoch": 3.25,
      "learning_rate": 3.0236268706139177e-05,
      "loss": 0.7344,
      "step": 506
    },
    {
      "epoch": 3.26,
      "learning_rate": 3.003878972700509e-05,
      "loss": 0.645,
      "step": 507
    },
    {
      "epoch": 3.27,
      "learning_rate": 2.9841680471670918e-05,
      "loss": 0.7067,
      "step": 508
    },
    {
      "epoch": 3.27,
      "learning_rate": 2.964494459101661e-05,
      "loss": 0.8038,
      "step": 509
    },
    {
      "epoch": 3.28,
      "learning_rate": 2.9448585729006462e-05,
      "loss": 0.6032,
      "step": 510
    },
    {
      "epoch": 3.29,
      "learning_rate": 2.925260752262158e-05,
      "loss": 0.5504,
      "step": 511
    },
    {
      "epoch": 3.29,
      "learning_rate": 2.9057013601792517e-05,
      "loss": 0.7563,
      "step": 512
    },
    {
      "epoch": 3.3,
      "learning_rate": 2.8861807589332048e-05,
      "loss": 0.7433,
      "step": 513
    },
    {
      "epoch": 3.31,
      "learning_rate": 2.8666993100868044e-05,
      "loss": 0.6599,
      "step": 514
    },
    {
      "epoch": 3.31,
      "learning_rate": 2.847257374477661e-05,
      "loss": 0.7858,
      "step": 515
    },
    {
      "epoch": 3.32,
      "learning_rate": 2.8278553122115072e-05,
      "loss": 0.8191,
      "step": 516
    },
    {
      "epoch": 3.32,
      "learning_rate": 2.8084934826555386e-05,
      "loss": 0.8613,
      "step": 517
    },
    {
      "epoch": 3.33,
      "learning_rate": 2.7891722444317603e-05,
      "loss": 0.6136,
      "step": 518
    },
    {
      "epoch": 3.34,
      "learning_rate": 2.769891955410335e-05,
      "loss": 0.6907,
      "step": 519
    },
    {
      "epoch": 3.34,
      "learning_rate": 2.750652972702963e-05,
      "loss": 0.5671,
      "step": 520
    },
    {
      "epoch": 3.35,
      "learning_rate": 2.731455652656263e-05,
      "loss": 0.8927,
      "step": 521
    },
    {
      "epoch": 3.36,
      "learning_rate": 2.712300350845174e-05,
      "loss": 0.4481,
      "step": 522
    },
    {
      "epoch": 3.36,
      "learning_rate": 2.693187422066367e-05,
      "loss": 0.4987,
      "step": 523
    },
    {
      "epoch": 3.37,
      "learning_rate": 2.674117220331678e-05,
      "loss": 0.5639,
      "step": 524
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.6550900988615423e-05,
      "loss": 0.516,
      "step": 525
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.6361064100784644e-05,
      "loss": 0.6383,
      "step": 526
    },
    {
      "epoch": 3.39,
      "learning_rate": 2.6171665056004768e-05,
      "loss": 0.6605,
      "step": 527
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.5982707362346414e-05,
      "loss": 0.7869,
      "step": 528
    },
    {
      "epoch": 3.4,
      "learning_rate": 2.5794194519705394e-05,
      "loss": 0.6375,
      "step": 529
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.5606130019737956e-05,
      "loss": 0.6123,
      "step": 530
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.5418517345796085e-05,
      "loss": 0.727,
      "step": 531
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.5231359972863e-05,
      "loss": 0.8873,
      "step": 532
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.5044661367488786e-05,
      "loss": 0.5689,
      "step": 533
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.4858424987726155e-05,
      "loss": 0.7461,
      "step": 534
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.4672654283066487e-05,
      "loss": 0.512,
      "step": 535
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.4487352694375825e-05,
      "loss": 0.6023,
      "step": 536
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.43025236538312e-05,
      "loss": 0.5548,
      "step": 537
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.411817058485706e-05,
      "loss": 0.5645,
      "step": 538
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.3934296902061843e-05,
      "loss": 0.5506,
      "step": 539
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.3750906011174733e-05,
      "loss": 0.9507,
      "step": 540
    },
    {
      "epoch": 3.48,
      "learning_rate": 2.3568001308982652e-05,
      "loss": 0.6752,
      "step": 541
    },
    {
      "epoch": 3.49,
      "learning_rate": 2.3385586183267193e-05,
      "loss": 0.7537,
      "step": 542
    },
    {
      "epoch": 3.49,
      "learning_rate": 2.3203664012742023e-05,
      "loss": 0.5397,
      "step": 543
    },
    {
      "epoch": 3.5,
      "learning_rate": 2.3022238166990186e-05,
      "loss": 0.9337,
      "step": 544
    },
    {
      "epoch": 3.5,
      "learning_rate": 2.2841312006401784e-05,
      "loss": 0.6197,
      "step": 545
    },
    {
      "epoch": 3.51,
      "learning_rate": 2.2660888882111664e-05,
      "loss": 0.5248,
      "step": 546
    },
    {
      "epoch": 3.52,
      "learning_rate": 2.2480972135937384e-05,
      "loss": 0.712,
      "step": 547
    },
    {
      "epoch": 3.52,
      "learning_rate": 2.2301565100317345e-05,
      "loss": 0.6827,
      "step": 548
    },
    {
      "epoch": 3.53,
      "learning_rate": 2.2122671098248993e-05,
      "loss": 0.4865,
      "step": 549
    },
    {
      "epoch": 3.54,
      "learning_rate": 2.1944293443227303e-05,
      "loss": 0.5161,
      "step": 550
    },
    {
      "epoch": 3.54,
      "learning_rate": 2.1766435439183437e-05,
      "loss": 0.6253,
      "step": 551
    },
    {
      "epoch": 3.55,
      "learning_rate": 2.158910038042351e-05,
      "loss": 0.6002,
      "step": 552
    },
    {
      "epoch": 3.56,
      "learning_rate": 2.141229155156758e-05,
      "loss": 0.6112,
      "step": 553
    },
    {
      "epoch": 3.56,
      "learning_rate": 2.123601222748883e-05,
      "loss": 1.0001,
      "step": 554
    },
    {
      "epoch": 3.57,
      "learning_rate": 2.1060265673252883e-05,
      "loss": 0.9138,
      "step": 555
    },
    {
      "epoch": 3.58,
      "learning_rate": 2.088505514405734e-05,
      "loss": 0.8408,
      "step": 556
    },
    {
      "epoch": 3.58,
      "learning_rate": 2.0710383885171498e-05,
      "loss": 0.5882,
      "step": 557
    },
    {
      "epoch": 3.59,
      "learning_rate": 2.0536255131876215e-05,
      "loss": 0.6517,
      "step": 558
    },
    {
      "epoch": 3.59,
      "learning_rate": 2.0362672109403976e-05,
      "loss": 0.6982,
      "step": 559
    },
    {
      "epoch": 3.6,
      "learning_rate": 2.0189638032879242e-05,
      "loss": 0.5194,
      "step": 560
    },
    {
      "epoch": 3.61,
      "learning_rate": 2.0017156107258788e-05,
      "loss": 0.8238,
      "step": 561
    },
    {
      "epoch": 3.61,
      "learning_rate": 1.9845229527272396e-05,
      "loss": 0.6565,
      "step": 562
    },
    {
      "epoch": 3.62,
      "learning_rate": 1.9673861477363664e-05,
      "loss": 0.4802,
      "step": 563
    },
    {
      "epoch": 3.63,
      "learning_rate": 1.9503055131631048e-05,
      "loss": 0.4344,
      "step": 564
    },
    {
      "epoch": 3.63,
      "learning_rate": 1.933281365376905e-05,
      "loss": 0.6009,
      "step": 565
    },
    {
      "epoch": 3.64,
      "learning_rate": 1.916314019700961e-05,
      "loss": 0.5778,
      "step": 566
    },
    {
      "epoch": 3.65,
      "learning_rate": 1.899403790406376e-05,
      "loss": 0.5552,
      "step": 567
    },
    {
      "epoch": 3.65,
      "learning_rate": 1.8825509907063327e-05,
      "loss": 0.6126,
      "step": 568
    },
    {
      "epoch": 3.66,
      "learning_rate": 1.8657559327502982e-05,
      "loss": 0.667,
      "step": 569
    },
    {
      "epoch": 3.67,
      "learning_rate": 1.8490189276182405e-05,
      "loss": 0.4902,
      "step": 570
    },
    {
      "epoch": 3.67,
      "learning_rate": 1.8323402853148658e-05,
      "loss": 0.8881,
      "step": 571
    },
    {
      "epoch": 3.68,
      "learning_rate": 1.815720314763878e-05,
      "loss": 0.6486,
      "step": 572
    },
    {
      "epoch": 3.68,
      "learning_rate": 1.7991593238022543e-05,
      "loss": 0.5539,
      "step": 573
    },
    {
      "epoch": 3.69,
      "learning_rate": 1.7826576191745502e-05,
      "loss": 0.5621,
      "step": 574
    },
    {
      "epoch": 3.7,
      "learning_rate": 1.766215506527209e-05,
      "loss": 0.5539,
      "step": 575
    },
    {
      "epoch": 3.7,
      "learning_rate": 1.749833290402904e-05,
      "loss": 0.7192,
      "step": 576
    },
    {
      "epoch": 3.71,
      "learning_rate": 1.7335112742349008e-05,
      "loss": 0.7829,
      "step": 577
    },
    {
      "epoch": 3.72,
      "learning_rate": 1.717249760341433e-05,
      "loss": 0.4852,
      "step": 578
    },
    {
      "epoch": 3.72,
      "learning_rate": 1.7010490499201028e-05,
      "loss": 0.5209,
      "step": 579
    },
    {
      "epoch": 3.73,
      "learning_rate": 1.684909443042309e-05,
      "loss": 0.6508,
      "step": 580
    },
    {
      "epoch": 3.74,
      "learning_rate": 1.668831238647679e-05,
      "loss": 0.4907,
      "step": 581
    },
    {
      "epoch": 3.74,
      "learning_rate": 1.652814734538537e-05,
      "loss": 0.4791,
      "step": 582
    },
    {
      "epoch": 3.75,
      "learning_rate": 1.6368602273743877e-05,
      "loss": 0.4628,
      "step": 583
    },
    {
      "epoch": 3.76,
      "learning_rate": 1.6209680126664223e-05,
      "loss": 0.6125,
      "step": 584
    },
    {
      "epoch": 3.76,
      "learning_rate": 1.6051383847720437e-05,
      "loss": 0.3538,
      "step": 585
    },
    {
      "epoch": 3.77,
      "learning_rate": 1.5893716368894118e-05,
      "loss": 0.6828,
      "step": 586
    },
    {
      "epoch": 3.77,
      "learning_rate": 1.5736680610520233e-05,
      "loss": 0.5868,
      "step": 587
    },
    {
      "epoch": 3.78,
      "learning_rate": 1.5580279481232878e-05,
      "loss": 0.6299,
      "step": 588
    },
    {
      "epoch": 3.79,
      "learning_rate": 1.5424515877911498e-05,
      "loss": 0.6469,
      "step": 589
    },
    {
      "epoch": 3.79,
      "learning_rate": 1.5269392685627232e-05,
      "loss": 0.6934,
      "step": 590
    },
    {
      "epoch": 3.8,
      "learning_rate": 1.5114912777589413e-05,
      "loss": 0.5613,
      "step": 591
    },
    {
      "epoch": 3.81,
      "learning_rate": 1.4961079015092427e-05,
      "loss": 0.6814,
      "step": 592
    },
    {
      "epoch": 3.81,
      "learning_rate": 1.4807894247462667e-05,
      "loss": 0.5436,
      "step": 593
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.4655361312005755e-05,
      "loss": 0.834,
      "step": 594
    },
    {
      "epoch": 3.83,
      "learning_rate": 1.4503483033954023e-05,
      "loss": 0.7239,
      "step": 595
    },
    {
      "epoch": 3.83,
      "learning_rate": 1.4352262226414159e-05,
      "loss": 0.6076,
      "step": 596
    },
    {
      "epoch": 3.84,
      "learning_rate": 1.42017016903151e-05,
      "loss": 0.4552,
      "step": 597
    },
    {
      "epoch": 3.85,
      "learning_rate": 1.4051804214356174e-05,
      "loss": 0.8232,
      "step": 598
    },
    {
      "epoch": 3.85,
      "learning_rate": 1.3902572574955402e-05,
      "loss": 0.698,
      "step": 599
    },
    {
      "epoch": 3.86,
      "learning_rate": 1.3754009536198165e-05,
      "loss": 0.4972,
      "step": 600
    },
    {
      "epoch": 3.86,
      "learning_rate": 1.3606117849785899e-05,
      "loss": 0.7743,
      "step": 601
    },
    {
      "epoch": 3.87,
      "learning_rate": 1.345890025498518e-05,
      "loss": 0.5103,
      "step": 602
    },
    {
      "epoch": 3.88,
      "learning_rate": 1.331235947857699e-05,
      "loss": 0.6167,
      "step": 603
    },
    {
      "epoch": 3.88,
      "learning_rate": 1.3166498234806218e-05,
      "loss": 0.612,
      "step": 604
    },
    {
      "epoch": 3.89,
      "learning_rate": 1.302131922533133e-05,
      "loss": 0.5918,
      "step": 605
    },
    {
      "epoch": 3.9,
      "learning_rate": 1.287682513917443e-05,
      "loss": 0.4225,
      "step": 606
    },
    {
      "epoch": 3.9,
      "learning_rate": 1.2733018652671364e-05,
      "loss": 0.6399,
      "step": 607
    },
    {
      "epoch": 3.91,
      "learning_rate": 1.2589902429422162e-05,
      "loss": 0.5534,
      "step": 608
    },
    {
      "epoch": 3.92,
      "learning_rate": 1.2447479120241751e-05,
      "loss": 0.4786,
      "step": 609
    },
    {
      "epoch": 3.92,
      "learning_rate": 1.2305751363110801e-05,
      "loss": 0.6066,
      "step": 610
    },
    {
      "epoch": 3.93,
      "learning_rate": 1.2164721783126908e-05,
      "loss": 0.5157,
      "step": 611
    },
    {
      "epoch": 3.94,
      "learning_rate": 1.2024392992455924e-05,
      "loss": 0.7284,
      "step": 612
    },
    {
      "epoch": 3.94,
      "learning_rate": 1.1884767590283652e-05,
      "loss": 0.781,
      "step": 613
    },
    {
      "epoch": 3.95,
      "learning_rate": 1.1745848162767603e-05,
      "loss": 0.554,
      "step": 614
    },
    {
      "epoch": 3.95,
      "learning_rate": 1.1607637282989164e-05,
      "loss": 0.6483,
      "step": 615
    },
    {
      "epoch": 3.96,
      "learning_rate": 1.1470137510905927e-05,
      "loss": 0.7229,
      "step": 616
    },
    {
      "epoch": 3.97,
      "learning_rate": 1.1333351393304253e-05,
      "loss": 0.6212,
      "step": 617
    },
    {
      "epoch": 3.97,
      "learning_rate": 1.1197281463752112e-05,
      "loss": 0.588,
      "step": 618
    },
    {
      "epoch": 3.98,
      "learning_rate": 1.1061930242552193e-05,
      "loss": 0.7329,
      "step": 619
    },
    {
      "epoch": 3.99,
      "learning_rate": 1.0927300236695149e-05,
      "loss": 0.5646,
      "step": 620
    },
    {
      "epoch": 3.99,
      "learning_rate": 1.0793393939813206e-05,
      "loss": 0.4729,
      "step": 621
    },
    {
      "epoch": 4.0,
      "learning_rate": 1.0660213832133992e-05,
      "loss": 0.5379,
      "step": 622
    },
    {
      "epoch": 4.01,
      "learning_rate": 1.0527762380434547e-05,
      "loss": 0.4103,
      "step": 623
    },
    {
      "epoch": 4.01,
      "learning_rate": 1.0396042037995679e-05,
      "loss": 0.5288,
      "step": 624
    },
    {
      "epoch": 4.02,
      "learning_rate": 1.0265055244556516e-05,
      "loss": 0.5467,
      "step": 625
    },
    {
      "epoch": 4.03,
      "learning_rate": 1.0134804426269284e-05,
      "loss": 0.4474,
      "step": 626
    },
    {
      "epoch": 4.03,
      "learning_rate": 1.0005291995654415e-05,
      "loss": 0.6113,
      "step": 627
    },
    {
      "epoch": 4.04,
      "learning_rate": 9.87652035155583e-06,
      "loss": 0.5161,
      "step": 628
    },
    {
      "epoch": 4.05,
      "learning_rate": 9.748491879096523e-06,
      "loss": 0.4702,
      "step": 629
    },
    {
      "epoch": 4.05,
      "learning_rate": 9.621208949634376e-06,
      "loss": 0.5705,
      "step": 630
    },
    {
      "epoch": 4.06,
      "learning_rate": 9.494673920718222e-06,
      "loss": 0.6609,
      "step": 631
    },
    {
      "epoch": 4.06,
      "learning_rate": 9.368889136044251e-06,
      "loss": 0.5481,
      "step": 632
    },
    {
      "epoch": 4.07,
      "learning_rate": 9.243856925412481e-06,
      "loss": 0.7805,
      "step": 633
    },
    {
      "epoch": 4.08,
      "learning_rate": 9.119579604683703e-06,
      "loss": 0.6868,
      "step": 634
    },
    {
      "epoch": 4.08,
      "learning_rate": 8.99605947573654e-06,
      "loss": 0.5405,
      "step": 635
    },
    {
      "epoch": 4.09,
      "learning_rate": 8.87329882642482e-06,
      "loss": 0.6776,
      "step": 636
    },
    {
      "epoch": 4.1,
      "learning_rate": 8.751299930535211e-06,
      "loss": 0.4999,
      "step": 637
    },
    {
      "epoch": 4.1,
      "learning_rate": 8.630065047745089e-06,
      "loss": 0.5188,
      "step": 638
    },
    {
      "epoch": 4.11,
      "learning_rate": 8.509596423580712e-06,
      "loss": 0.6006,
      "step": 639
    },
    {
      "epoch": 4.12,
      "learning_rate": 8.389896289375593e-06,
      "loss": 0.6381,
      "step": 640
    },
    {
      "epoch": 4.12,
      "learning_rate": 8.270966862229185e-06,
      "loss": 0.5027,
      "step": 641
    },
    {
      "epoch": 4.13,
      "learning_rate": 8.152810344965816e-06,
      "loss": 0.7125,
      "step": 642
    },
    {
      "epoch": 4.14,
      "learning_rate": 8.035428926093907e-06,
      "loss": 0.4542,
      "step": 643
    },
    {
      "epoch": 4.14,
      "learning_rate": 7.918824779765394e-06,
      "loss": 0.3773,
      "step": 644
    },
    {
      "epoch": 4.15,
      "learning_rate": 7.803000065735511e-06,
      "loss": 0.6276,
      "step": 645
    },
    {
      "epoch": 4.15,
      "learning_rate": 7.687956929322736e-06,
      "loss": 0.4788,
      "step": 646
    },
    {
      "epoch": 4.16,
      "learning_rate": 7.573697501369087e-06,
      "loss": 0.4731,
      "step": 647
    },
    {
      "epoch": 4.17,
      "learning_rate": 7.460223898200641e-06,
      "loss": 0.5609,
      "step": 648
    },
    {
      "epoch": 4.17,
      "learning_rate": 7.347538221588341e-06,
      "loss": 0.5118,
      "step": 649
    },
    {
      "epoch": 4.18,
      "learning_rate": 7.235642558709072e-06,
      "loss": 0.4677,
      "step": 650
    },
    {
      "epoch": 4.19,
      "learning_rate": 7.12453898210697e-06,
      "loss": 0.6534,
      "step": 651
    },
    {
      "epoch": 4.19,
      "learning_rate": 7.014229549655105e-06,
      "loss": 0.748,
      "step": 652
    },
    {
      "epoch": 4.2,
      "learning_rate": 6.904716304517278e-06,
      "loss": 0.7332,
      "step": 653
    },
    {
      "epoch": 4.21,
      "learning_rate": 6.796001275110231e-06,
      "loss": 0.5593,
      "step": 654
    },
    {
      "epoch": 4.21,
      "learning_rate": 6.688086475066069e-06,
      "loss": 0.8044,
      "step": 655
    },
    {
      "epoch": 4.22,
      "learning_rate": 6.580973903194948e-06,
      "loss": 0.578,
      "step": 656
    },
    {
      "epoch": 4.23,
      "learning_rate": 6.474665543448072e-06,
      "loss": 0.5653,
      "step": 657
    },
    {
      "epoch": 4.23,
      "learning_rate": 6.369163364880937e-06,
      "loss": 0.6664,
      "step": 658
    },
    {
      "epoch": 4.24,
      "learning_rate": 6.264469321616845e-06,
      "loss": 0.7393,
      "step": 659
    },
    {
      "epoch": 4.24,
      "learning_rate": 6.16058535281075e-06,
      "loss": 0.5658,
      "step": 660
    },
    {
      "epoch": 4.25,
      "learning_rate": 6.057513382613295e-06,
      "loss": 0.5,
      "step": 661
    },
    {
      "epoch": 4.26,
      "learning_rate": 5.955255320135194e-06,
      "loss": 0.5397,
      "step": 662
    },
    {
      "epoch": 4.26,
      "learning_rate": 5.853813059411883e-06,
      "loss": 0.481,
      "step": 663
    },
    {
      "epoch": 4.27,
      "learning_rate": 5.75318847936841e-06,
      "loss": 0.6997,
      "step": 664
    },
    {
      "epoch": 4.28,
      "learning_rate": 5.65338344378466e-06,
      "loss": 0.4791,
      "step": 665
    },
    {
      "epoch": 4.28,
      "learning_rate": 5.554399801260813e-06,
      "loss": 0.6551,
      "step": 666
    },
    {
      "epoch": 4.29,
      "learning_rate": 5.4562393851831205e-06,
      "loss": 0.4305,
      "step": 667
    },
    {
      "epoch": 4.3,
      "learning_rate": 5.358904013689942e-06,
      "loss": 0.591,
      "step": 668
    },
    {
      "epoch": 4.3,
      "learning_rate": 5.262395489638055e-06,
      "loss": 0.5181,
      "step": 669
    },
    {
      "epoch": 4.31,
      "learning_rate": 5.1667156005693e-06,
      "loss": 0.6898,
      "step": 670
    },
    {
      "epoch": 4.32,
      "learning_rate": 5.071866118677404e-06,
      "loss": 0.6176,
      "step": 671
    },
    {
      "epoch": 4.32,
      "learning_rate": 4.977848800775254e-06,
      "loss": 0.6962,
      "step": 672
    },
    {
      "epoch": 4.33,
      "learning_rate": 4.884665388262255e-06,
      "loss": 0.6571,
      "step": 673
    },
    {
      "epoch": 4.33,
      "learning_rate": 4.79231760709215e-06,
      "loss": 0.5185,
      "step": 674
    },
    {
      "epoch": 4.34,
      "learning_rate": 4.700807167741001e-06,
      "loss": 0.6008,
      "step": 675
    },
    {
      "epoch": 4.35,
      "learning_rate": 4.6101357651755494e-06,
      "loss": 0.5831,
      "step": 676
    },
    {
      "epoch": 4.35,
      "learning_rate": 4.520305078821779e-06,
      "loss": 0.5145,
      "step": 677
    },
    {
      "epoch": 4.36,
      "learning_rate": 4.431316772533872e-06,
      "loss": 0.3711,
      "step": 678
    },
    {
      "epoch": 4.37,
      "learning_rate": 4.343172494563308e-06,
      "loss": 0.3811,
      "step": 679
    },
    {
      "epoch": 4.37,
      "learning_rate": 4.25587387752841e-06,
      "loss": 0.421,
      "step": 680
    },
    {
      "epoch": 4.38,
      "learning_rate": 4.1694225383840495e-06,
      "loss": 0.4413,
      "step": 681
    },
    {
      "epoch": 4.39,
      "learning_rate": 4.083820078391737e-06,
      "loss": 0.5581,
      "step": 682
    },
    {
      "epoch": 4.39,
      "learning_rate": 3.99906808308994e-06,
      "loss": 0.6599,
      "step": 683
    },
    {
      "epoch": 4.4,
      "learning_rate": 3.915168122264723e-06,
      "loss": 0.5316,
      "step": 684
    },
    {
      "epoch": 4.41,
      "learning_rate": 3.832121749920681e-06,
      "loss": 0.5447,
      "step": 685
    },
    {
      "epoch": 4.41,
      "learning_rate": 3.749930504252136e-06,
      "loss": 0.6333,
      "step": 686
    },
    {
      "epoch": 4.42,
      "learning_rate": 3.668595907614658e-06,
      "loss": 0.5937,
      "step": 687
    },
    {
      "epoch": 4.42,
      "learning_rate": 3.588119466496864e-06,
      "loss": 0.5606,
      "step": 688
    },
    {
      "epoch": 4.43,
      "learning_rate": 3.508502671492525e-06,
      "loss": 0.548,
      "step": 689
    },
    {
      "epoch": 4.44,
      "learning_rate": 3.429746997272937e-06,
      "loss": 0.6627,
      "step": 690
    },
    {
      "epoch": 4.44,
      "learning_rate": 3.351853902559643e-06,
      "loss": 0.4303,
      "step": 691
    },
    {
      "epoch": 4.45,
      "learning_rate": 3.274824830097362e-06,
      "loss": 0.3945,
      "step": 692
    },
    {
      "epoch": 4.46,
      "learning_rate": 3.1986612066273127e-06,
      "loss": 0.4596,
      "step": 693
    },
    {
      "epoch": 4.46,
      "learning_rate": 3.1233644428607655e-06,
      "loss": 0.4494,
      "step": 694
    },
    {
      "epoch": 4.47,
      "learning_rate": 3.0489359334529123e-06,
      "loss": 0.7181,
      "step": 695
    },
    {
      "epoch": 4.48,
      "learning_rate": 2.9753770569770467e-06,
      "loss": 0.7249,
      "step": 696
    },
    {
      "epoch": 4.48,
      "learning_rate": 2.902689175899015e-06,
      "loss": 0.6261,
      "step": 697
    },
    {
      "epoch": 4.49,
      "learning_rate": 2.830873636551995e-06,
      "loss": 0.5188,
      "step": 698
    },
    {
      "epoch": 4.5,
      "learning_rate": 2.7599317691115457e-06,
      "loss": 0.5898,
      "step": 699
    },
    {
      "epoch": 4.5,
      "learning_rate": 2.6898648875709863e-06,
      "loss": 0.7492,
      "step": 700
    },
    {
      "epoch": 4.51,
      "learning_rate": 2.6206742897170345e-06,
      "loss": 0.415,
      "step": 701
    },
    {
      "epoch": 4.51,
      "learning_rate": 2.552361257105784e-06,
      "loss": 0.5348,
      "step": 702
    },
    {
      "epoch": 4.52,
      "learning_rate": 2.4849270550389768e-06,
      "loss": 0.5342,
      "step": 703
    },
    {
      "epoch": 4.53,
      "learning_rate": 2.4183729325405524e-06,
      "loss": 0.4648,
      "step": 704
    },
    {
      "epoch": 4.53,
      "learning_rate": 2.352700122333512e-06,
      "loss": 0.5283,
      "step": 705
    },
    {
      "epoch": 4.54,
      "learning_rate": 2.2879098408170927e-06,
      "loss": 0.4068,
      "step": 706
    },
    {
      "epoch": 4.55,
      "learning_rate": 2.22400328804423e-06,
      "loss": 0.4841,
      "step": 707
    },
    {
      "epoch": 4.55,
      "learning_rate": 2.160981647699345e-06,
      "loss": 0.5672,
      "step": 708
    },
    {
      "epoch": 4.56,
      "learning_rate": 2.0988460870764004e-06,
      "loss": 0.768,
      "step": 709
    },
    {
      "epoch": 4.57,
      "learning_rate": 2.0375977570572967e-06,
      "loss": 0.9241,
      "step": 710
    },
    {
      "epoch": 4.57,
      "learning_rate": 1.977237792090547e-06,
      "loss": 0.5822,
      "step": 711
    },
    {
      "epoch": 4.58,
      "learning_rate": 1.917767310170271e-06,
      "loss": 0.7651,
      "step": 712
    },
    {
      "epoch": 4.59,
      "learning_rate": 1.8591874128154774e-06,
      "loss": 0.4456,
      "step": 713
    },
    {
      "epoch": 4.59,
      "learning_rate": 1.8014991850496598e-06,
      "loss": 0.5773,
      "step": 714
    },
    {
      "epoch": 4.6,
      "learning_rate": 1.7447036953807171e-06,
      "loss": 0.5549,
      "step": 715
    },
    {
      "epoch": 4.6,
      "learning_rate": 1.6888019957811418e-06,
      "loss": 0.6203,
      "step": 716
    },
    {
      "epoch": 4.61,
      "learning_rate": 1.6337951216685687e-06,
      "loss": 0.6571,
      "step": 717
    },
    {
      "epoch": 4.62,
      "learning_rate": 1.5796840918865408e-06,
      "loss": 0.4795,
      "step": 718
    },
    {
      "epoch": 4.62,
      "learning_rate": 1.5264699086856958e-06,
      "loss": 0.403,
      "step": 719
    },
    {
      "epoch": 4.63,
      "learning_rate": 1.4741535577051657e-06,
      "loss": 0.4862,
      "step": 720
    },
    {
      "epoch": 4.64,
      "learning_rate": 1.4227360079543338e-06,
      "loss": 0.4817,
      "step": 721
    },
    {
      "epoch": 4.64,
      "learning_rate": 1.37221821179489e-06,
      "loss": 0.4838,
      "step": 722
    },
    {
      "epoch": 4.65,
      "learning_rate": 1.3226011049231768e-06,
      "loss": 0.5492,
      "step": 723
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.273885606352876e-06,
      "loss": 0.6379,
      "step": 724
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.2260726183979776e-06,
      "loss": 0.4004,
      "step": 725
    },
    {
      "epoch": 4.67,
      "learning_rate": 1.179163026656066e-06,
      "loss": 0.5348,
      "step": 726
    },
    {
      "epoch": 4.68,
      "learning_rate": 1.1331576999919158e-06,
      "loss": 0.8229,
      "step": 727
    },
    {
      "epoch": 4.68,
      "learning_rate": 1.0880574905213991e-06,
      "loss": 0.4037,
      "step": 728
    },
    {
      "epoch": 4.69,
      "learning_rate": 1.043863233595721e-06,
      "loss": 0.5704,
      "step": 729
    },
    {
      "epoch": 4.69,
      "learning_rate": 1.000575747785909e-06,
      "loss": 0.5503,
      "step": 730
    },
    {
      "epoch": 4.7,
      "learning_rate": 9.581958348676867e-07,
      "loss": 0.5606,
      "step": 731
    },
    {
      "epoch": 4.71,
      "learning_rate": 9.167242798066023e-07,
      "loss": 0.586,
      "step": 732
    },
    {
      "epoch": 4.71,
      "learning_rate": 8.761618507435065e-07,
      "loss": 0.559,
      "step": 733
    },
    {
      "epoch": 4.72,
      "learning_rate": 8.365092989803091e-07,
      "loss": 0.4891,
      "step": 734
    },
    {
      "epoch": 4.73,
      "learning_rate": 7.977673589660617e-07,
      "loss": 0.4036,
      "step": 735
    },
    {
      "epoch": 4.73,
      "learning_rate": 7.599367482833797e-07,
      "loss": 0.5945,
      "step": 736
    },
    {
      "epoch": 4.74,
      "learning_rate": 7.2301816763512e-07,
      "loss": 0.4435,
      "step": 737
    },
    {
      "epoch": 4.75,
      "learning_rate": 6.870123008314188e-07,
      "loss": 0.377,
      "step": 738
    },
    {
      "epoch": 4.75,
      "learning_rate": 6.519198147770189e-07,
      "loss": 0.4269,
      "step": 739
    },
    {
      "epoch": 4.76,
      "learning_rate": 6.177413594589343e-07,
      "loss": 0.4397,
      "step": 740
    },
    {
      "epoch": 4.77,
      "learning_rate": 5.844775679343883e-07,
      "loss": 0.3839,
      "step": 741
    },
    {
      "epoch": 4.77,
      "learning_rate": 5.521290563190895e-07,
      "loss": 0.6592,
      "step": 742
    },
    {
      "epoch": 4.78,
      "learning_rate": 5.206964237758461e-07,
      "loss": 0.4351,
      "step": 743
    },
    {
      "epoch": 4.78,
      "learning_rate": 4.901802525034471e-07,
      "loss": 0.5488,
      "step": 744
    },
    {
      "epoch": 4.79,
      "learning_rate": 4.6058110772586595e-07,
      "loss": 0.6737,
      "step": 745
    },
    {
      "epoch": 4.8,
      "learning_rate": 4.31899537681818e-07,
      "loss": 0.5452,
      "step": 746
    },
    {
      "epoch": 4.8,
      "learning_rate": 4.041360736146027e-07,
      "loss": 0.5144,
      "step": 747
    },
    {
      "epoch": 4.81,
      "learning_rate": 3.7729122976224995e-07,
      "loss": 0.6366,
      "step": 748
    },
    {
      "epoch": 4.82,
      "learning_rate": 3.5136550334798903e-07,
      "loss": 0.6489,
      "step": 749
    },
    {
      "epoch": 4.82,
      "learning_rate": 3.2635937457108356e-07,
      "loss": 0.7178,
      "step": 750
    },
    {
      "epoch": 4.83,
      "learning_rate": 3.0227330659786666e-07,
      "loss": 0.5018,
      "step": 751
    },
    {
      "epoch": 4.84,
      "learning_rate": 2.7910774555323093e-07,
      "loss": 0.4706,
      "step": 752
    },
    {
      "epoch": 4.84,
      "learning_rate": 2.56863120512324e-07,
      "loss": 0.4963,
      "step": 753
    },
    {
      "epoch": 4.85,
      "learning_rate": 2.3553984349261037e-07,
      "loss": 0.7422,
      "step": 754
    },
    {
      "epoch": 4.86,
      "learning_rate": 2.1513830944624447e-07,
      "loss": 0.5224,
      "step": 755
    },
    {
      "epoch": 4.86,
      "learning_rate": 1.9565889625275946e-07,
      "loss": 0.5809,
      "step": 756
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.7710196471204532e-07,
      "loss": 0.545,
      "step": 757
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.5946785853769852e-07,
      "loss": 0.5405,
      "step": 758
    },
    {
      "epoch": 4.88,
      "learning_rate": 1.4275690435062715e-07,
      "loss": 0.5518,
      "step": 759
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.2696941167302801e-07,
      "loss": 0.5056,
      "step": 760
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.1210567292262442e-07,
      "loss": 0.4891,
      "step": 761
    },
    {
      "epoch": 4.9,
      "learning_rate": 9.816596340727069e-08,
      "loss": 0.4964,
      "step": 762
    },
    {
      "epoch": 4.91,
      "learning_rate": 8.515054131984501e-08,
      "loss": 0.4472,
      "step": 763
    },
    {
      "epoch": 4.91,
      "learning_rate": 7.305964773348106e-08,
      "loss": 0.5581,
      "step": 764
    },
    {
      "epoch": 4.92,
      "learning_rate": 6.18935065970716e-08,
      "loss": 0.3874,
      "step": 765
    },
    {
      "epoch": 4.93,
      "learning_rate": 5.165232473114956e-08,
      "loss": 0.5289,
      "step": 766
    },
    {
      "epoch": 4.93,
      "learning_rate": 4.2336291824046636e-08,
      "loss": 0.63,
      "step": 767
    },
    {
      "epoch": 4.94,
      "learning_rate": 3.3945580428385024e-08,
      "loss": 0.6482,
      "step": 768
    },
    {
      "epoch": 4.95,
      "learning_rate": 2.6480345957868856e-08,
      "loss": 0.5725,
      "step": 769
    },
    {
      "epoch": 4.95,
      "learning_rate": 1.994072668441982e-08,
      "loss": 0.5544,
      "step": 770
    },
    {
      "epoch": 4.96,
      "learning_rate": 1.4326843735606998e-08,
      "loss": 0.6202,
      "step": 771
    },
    {
      "epoch": 4.96,
      "learning_rate": 9.638801092398676e-09,
      "loss": 0.4801,
      "step": 772
    },
    {
      "epoch": 4.97,
      "learning_rate": 5.8766855872638415e-09,
      "loss": 0.5713,
      "step": 773
    },
    {
      "epoch": 4.98,
      "learning_rate": 3.040566902529074e-09,
      "loss": 0.6327,
      "step": 774
    },
    {
      "epoch": 4.98,
      "learning_rate": 1.130497569107325e-09,
      "loss": 0.6337,
      "step": 775
    }
  ],
  "logging_steps": 1,
  "max_steps": 775,
  "num_train_epochs": 5,
  "save_steps": 500,
  "total_flos": 6.424682641227776e+16,
  "trial_name": null,
  "trial_params": null
}