{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 0,
  "global_step": 560,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0017857142857142857,
      "grad_norm": 0.451171875,
      "learning_rate": 9.982142857142858e-06,
      "loss": 1.8264,
      "step": 1
    },
    {
      "epoch": 0.0035714285714285713,
      "grad_norm": 0.44921875,
      "learning_rate": 9.964285714285714e-06,
      "loss": 1.7787,
      "step": 2
    },
    {
      "epoch": 0.005357142857142857,
      "grad_norm": 0.44140625,
      "learning_rate": 9.946428571428572e-06,
      "loss": 1.83,
      "step": 3
    },
    {
      "epoch": 0.007142857142857143,
      "grad_norm": 0.427734375,
      "learning_rate": 9.92857142857143e-06,
      "loss": 1.8276,
      "step": 4
    },
    {
      "epoch": 0.008928571428571428,
      "grad_norm": 0.40234375,
      "learning_rate": 9.910714285714288e-06,
      "loss": 1.8274,
      "step": 5
    },
    {
      "epoch": 0.010714285714285714,
      "grad_norm": 0.376953125,
      "learning_rate": 9.892857142857143e-06,
      "loss": 1.854,
      "step": 6
    },
    {
      "epoch": 0.0125,
      "grad_norm": 0.353515625,
      "learning_rate": 9.875000000000001e-06,
      "loss": 1.7426,
      "step": 7
    },
    {
      "epoch": 0.014285714285714285,
      "grad_norm": 0.345703125,
      "learning_rate": 9.857142857142859e-06,
      "loss": 1.8109,
      "step": 8
    },
    {
      "epoch": 0.01607142857142857,
      "grad_norm": 0.330078125,
      "learning_rate": 9.839285714285715e-06,
      "loss": 1.8108,
      "step": 9
    },
    {
      "epoch": 0.017857142857142856,
      "grad_norm": 0.310546875,
      "learning_rate": 9.821428571428573e-06,
      "loss": 1.8006,
      "step": 10
    },
    {
      "epoch": 0.019642857142857142,
      "grad_norm": 0.31640625,
      "learning_rate": 9.803571428571428e-06,
      "loss": 1.7489,
      "step": 11
    },
    {
      "epoch": 0.02142857142857143,
      "grad_norm": 0.28515625,
      "learning_rate": 9.785714285714286e-06,
      "loss": 1.6758,
      "step": 12
    },
    {
      "epoch": 0.023214285714285715,
      "grad_norm": 0.306640625,
      "learning_rate": 9.767857142857144e-06,
      "loss": 1.6705,
      "step": 13
    },
    {
      "epoch": 0.025,
      "grad_norm": 0.283203125,
      "learning_rate": 9.75e-06,
      "loss": 1.6707,
      "step": 14
    },
    {
      "epoch": 0.026785714285714284,
      "grad_norm": 0.29296875,
      "learning_rate": 9.732142857142858e-06,
      "loss": 1.7135,
      "step": 15
    },
    {
      "epoch": 0.02857142857142857,
      "grad_norm": 0.27734375,
      "learning_rate": 9.714285714285715e-06,
      "loss": 1.653,
      "step": 16
    },
    {
      "epoch": 0.030357142857142857,
      "grad_norm": 0.29296875,
      "learning_rate": 9.696428571428573e-06,
      "loss": 1.6137,
      "step": 17
    },
    {
      "epoch": 0.03214285714285714,
      "grad_norm": 0.287109375,
      "learning_rate": 9.678571428571429e-06,
      "loss": 1.6772,
      "step": 18
    },
    {
      "epoch": 0.033928571428571426,
      "grad_norm": 0.373046875,
      "learning_rate": 9.660714285714287e-06,
      "loss": 1.5118,
      "step": 19
    },
    {
      "epoch": 0.03571428571428571,
      "grad_norm": 0.265625,
      "learning_rate": 9.642857142857144e-06,
      "loss": 1.5881,
      "step": 20
    },
    {
      "epoch": 0.0375,
      "grad_norm": 0.27734375,
      "learning_rate": 9.625e-06,
      "loss": 1.6045,
      "step": 21
    },
    {
      "epoch": 0.039285714285714285,
      "grad_norm": 0.2431640625,
      "learning_rate": 9.607142857142858e-06,
      "loss": 1.5647,
      "step": 22
    },
    {
      "epoch": 0.04107142857142857,
      "grad_norm": 0.279296875,
      "learning_rate": 9.589285714285716e-06,
      "loss": 1.5269,
      "step": 23
    },
    {
      "epoch": 0.04285714285714286,
      "grad_norm": 0.2314453125,
      "learning_rate": 9.571428571428573e-06,
      "loss": 1.5098,
      "step": 24
    },
    {
      "epoch": 0.044642857142857144,
      "grad_norm": 0.23828125,
      "learning_rate": 9.55357142857143e-06,
      "loss": 1.5816,
      "step": 25
    },
    {
      "epoch": 0.04642857142857143,
      "grad_norm": 0.220703125,
      "learning_rate": 9.535714285714287e-06,
      "loss": 1.5997,
      "step": 26
    },
    {
      "epoch": 0.048214285714285716,
      "grad_norm": 0.6875,
      "learning_rate": 9.517857142857143e-06,
      "loss": 1.5325,
      "step": 27
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.337890625,
      "learning_rate": 9.5e-06,
      "loss": 1.5751,
      "step": 28
    },
    {
      "epoch": 0.05178571428571429,
      "grad_norm": 0.2109375,
      "learning_rate": 9.482142857142858e-06,
      "loss": 1.4823,
      "step": 29
    },
    {
      "epoch": 0.05357142857142857,
      "grad_norm": 0.201171875,
      "learning_rate": 9.464285714285714e-06,
      "loss": 1.4963,
      "step": 30
    },
    {
      "epoch": 0.055357142857142855,
      "grad_norm": 0.1845703125,
      "learning_rate": 9.446428571428572e-06,
      "loss": 1.3804,
      "step": 31
    },
    {
      "epoch": 0.05714285714285714,
      "grad_norm": 0.1962890625,
      "learning_rate": 9.42857142857143e-06,
      "loss": 1.4485,
      "step": 32
    },
    {
      "epoch": 0.05892857142857143,
      "grad_norm": 0.1904296875,
      "learning_rate": 9.410714285714286e-06,
      "loss": 1.4925,
      "step": 33
    },
    {
      "epoch": 0.060714285714285714,
      "grad_norm": 0.201171875,
      "learning_rate": 9.392857142857143e-06,
      "loss": 1.5104,
      "step": 34
    },
    {
      "epoch": 0.0625,
      "grad_norm": 0.1962890625,
      "learning_rate": 9.375000000000001e-06,
      "loss": 1.4536,
      "step": 35
    },
    {
      "epoch": 0.06428571428571428,
      "grad_norm": 0.189453125,
      "learning_rate": 9.357142857142859e-06,
      "loss": 1.4875,
      "step": 36
    },
    {
      "epoch": 0.06607142857142857,
      "grad_norm": 0.193359375,
      "learning_rate": 9.339285714285715e-06,
      "loss": 1.5001,
      "step": 37
    },
    {
      "epoch": 0.06785714285714285,
      "grad_norm": 0.18359375,
      "learning_rate": 9.321428571428572e-06,
      "loss": 1.4814,
      "step": 38
    },
    {
      "epoch": 0.06964285714285715,
      "grad_norm": 0.177734375,
      "learning_rate": 9.30357142857143e-06,
      "loss": 1.4407,
      "step": 39
    },
    {
      "epoch": 0.07142857142857142,
      "grad_norm": 0.177734375,
      "learning_rate": 9.285714285714288e-06,
      "loss": 1.4605,
      "step": 40
    },
    {
      "epoch": 0.07321428571428572,
      "grad_norm": 0.205078125,
      "learning_rate": 9.267857142857144e-06,
      "loss": 1.4516,
      "step": 41
    },
    {
      "epoch": 0.075,
      "grad_norm": 0.1669921875,
      "learning_rate": 9.250000000000001e-06,
      "loss": 1.3603,
      "step": 42
    },
    {
      "epoch": 0.07678571428571429,
      "grad_norm": 0.2392578125,
      "learning_rate": 9.232142857142859e-06,
      "loss": 1.5122,
      "step": 43
    },
    {
      "epoch": 0.07857142857142857,
      "grad_norm": 0.18359375,
      "learning_rate": 9.214285714285715e-06,
      "loss": 1.499,
      "step": 44
    },
    {
      "epoch": 0.08035714285714286,
      "grad_norm": 0.1787109375,
      "learning_rate": 9.196428571428571e-06,
      "loss": 1.4101,
      "step": 45
    },
    {
      "epoch": 0.08214285714285714,
      "grad_norm": 0.22265625,
      "learning_rate": 9.178571428571429e-06,
      "loss": 1.4654,
      "step": 46
    },
    {
      "epoch": 0.08392857142857142,
      "grad_norm": 0.1904296875,
      "learning_rate": 9.160714285714286e-06,
      "loss": 1.3772,
      "step": 47
    },
    {
      "epoch": 0.08571428571428572,
      "grad_norm": 0.1689453125,
      "learning_rate": 9.142857142857144e-06,
      "loss": 1.3117,
      "step": 48
    },
    {
      "epoch": 0.0875,
      "grad_norm": 0.177734375,
      "learning_rate": 9.125e-06,
      "loss": 1.4509,
      "step": 49
    },
    {
      "epoch": 0.08928571428571429,
      "grad_norm": 0.1875,
      "learning_rate": 9.107142857142858e-06,
      "loss": 1.4234,
      "step": 50
    },
    {
      "epoch": 0.09107142857142857,
      "grad_norm": 0.16015625,
      "learning_rate": 9.089285714285715e-06,
      "loss": 1.3606,
      "step": 51
    },
    {
      "epoch": 0.09285714285714286,
      "grad_norm": 0.1845703125,
      "learning_rate": 9.071428571428573e-06,
      "loss": 1.3514,
      "step": 52
    },
    {
      "epoch": 0.09464285714285714,
      "grad_norm": 0.1875,
      "learning_rate": 9.053571428571429e-06,
      "loss": 1.4011,
      "step": 53
    },
    {
      "epoch": 0.09642857142857143,
      "grad_norm": 0.16796875,
      "learning_rate": 9.035714285714287e-06,
      "loss": 1.3582,
      "step": 54
    },
    {
      "epoch": 0.09821428571428571,
      "grad_norm": 0.1728515625,
      "learning_rate": 9.017857142857144e-06,
      "loss": 1.3714,
      "step": 55
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1923828125,
      "learning_rate": 9e-06,
      "loss": 1.34,
      "step": 56
    },
    {
      "epoch": 0.10178571428571428,
      "grad_norm": 0.1484375,
      "learning_rate": 8.982142857142858e-06,
      "loss": 1.3457,
      "step": 57
    },
    {
      "epoch": 0.10357142857142858,
      "grad_norm": 0.177734375,
      "learning_rate": 8.964285714285716e-06,
      "loss": 1.3943,
      "step": 58
    },
    {
      "epoch": 0.10535714285714286,
      "grad_norm": 0.1787109375,
      "learning_rate": 8.946428571428573e-06,
      "loss": 1.4559,
      "step": 59
    },
    {
      "epoch": 0.10714285714285714,
      "grad_norm": 0.1552734375,
      "learning_rate": 8.92857142857143e-06,
      "loss": 1.3399,
      "step": 60
    },
    {
      "epoch": 0.10892857142857143,
      "grad_norm": 0.1572265625,
      "learning_rate": 8.910714285714287e-06,
      "loss": 1.3514,
      "step": 61
    },
    {
      "epoch": 0.11071428571428571,
      "grad_norm": 0.154296875,
      "learning_rate": 8.892857142857143e-06,
      "loss": 1.3186,
      "step": 62
    },
    {
      "epoch": 0.1125,
      "grad_norm": 0.2119140625,
      "learning_rate": 8.875e-06,
      "loss": 1.3524,
      "step": 63
    },
    {
      "epoch": 0.11428571428571428,
      "grad_norm": 0.1767578125,
      "learning_rate": 8.857142857142858e-06,
      "loss": 1.3673,
      "step": 64
    },
    {
      "epoch": 0.11607142857142858,
      "grad_norm": 0.2177734375,
      "learning_rate": 8.839285714285714e-06,
      "loss": 1.3364,
      "step": 65
    },
    {
      "epoch": 0.11785714285714285,
      "grad_norm": 0.16015625,
      "learning_rate": 8.821428571428572e-06,
      "loss": 1.3614,
      "step": 66
    },
    {
      "epoch": 0.11964285714285715,
      "grad_norm": 0.1591796875,
      "learning_rate": 8.80357142857143e-06,
      "loss": 1.32,
      "step": 67
    },
    {
      "epoch": 0.12142857142857143,
      "grad_norm": 0.1650390625,
      "learning_rate": 8.785714285714286e-06,
      "loss": 1.3709,
      "step": 68
    },
    {
      "epoch": 0.12321428571428572,
      "grad_norm": 0.16015625,
      "learning_rate": 8.767857142857143e-06,
      "loss": 1.3182,
      "step": 69
    },
    {
      "epoch": 0.125,
      "grad_norm": 0.30078125,
      "learning_rate": 8.750000000000001e-06,
      "loss": 1.3977,
      "step": 70
    },
    {
      "epoch": 0.12678571428571428,
      "grad_norm": 0.302734375,
      "learning_rate": 8.732142857142859e-06,
      "loss": 1.3094,
      "step": 71
    },
    {
      "epoch": 0.12857142857142856,
      "grad_norm": 0.1513671875,
      "learning_rate": 8.714285714285715e-06,
      "loss": 1.3111,
      "step": 72
    },
    {
      "epoch": 0.13035714285714287,
      "grad_norm": 0.154296875,
      "learning_rate": 8.696428571428572e-06,
      "loss": 1.3303,
      "step": 73
    },
    {
      "epoch": 0.13214285714285715,
      "grad_norm": 0.146484375,
      "learning_rate": 8.67857142857143e-06,
      "loss": 1.2938,
      "step": 74
    },
    {
      "epoch": 0.13392857142857142,
      "grad_norm": 0.1728515625,
      "learning_rate": 8.660714285714286e-06,
      "loss": 1.2852,
      "step": 75
    },
    {
      "epoch": 0.1357142857142857,
      "grad_norm": 0.1708984375,
      "learning_rate": 8.642857142857144e-06,
      "loss": 1.2962,
      "step": 76
    },
    {
      "epoch": 0.1375,
      "grad_norm": 0.150390625,
      "learning_rate": 8.625000000000001e-06,
      "loss": 1.2855,
      "step": 77
    },
    {
      "epoch": 0.1392857142857143,
      "grad_norm": 0.1591796875,
      "learning_rate": 8.607142857142859e-06,
      "loss": 1.3886,
      "step": 78
    },
    {
      "epoch": 0.14107142857142857,
      "grad_norm": 0.158203125,
      "learning_rate": 8.589285714285715e-06,
      "loss": 1.3353,
      "step": 79
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 0.2158203125,
      "learning_rate": 8.571428571428571e-06,
      "loss": 1.3206,
      "step": 80
    },
    {
      "epoch": 0.14464285714285716,
      "grad_norm": 0.1533203125,
      "learning_rate": 8.553571428571429e-06,
      "loss": 1.2875,
      "step": 81
    },
    {
      "epoch": 0.14642857142857144,
      "grad_norm": 0.1826171875,
      "learning_rate": 8.535714285714286e-06,
      "loss": 1.263,
      "step": 82
    },
    {
      "epoch": 0.14821428571428572,
      "grad_norm": 0.16796875,
      "learning_rate": 8.517857142857144e-06,
      "loss": 1.3335,
      "step": 83
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1591796875,
      "learning_rate": 8.5e-06,
      "loss": 1.3109,
      "step": 84
    },
    {
      "epoch": 0.15178571428571427,
      "grad_norm": 0.2109375,
      "learning_rate": 8.482142857142858e-06,
      "loss": 1.2587,
      "step": 85
    },
    {
      "epoch": 0.15357142857142858,
      "grad_norm": 0.1513671875,
      "learning_rate": 8.464285714285715e-06,
      "loss": 1.2751,
      "step": 86
    },
    {
      "epoch": 0.15535714285714286,
      "grad_norm": 0.1650390625,
      "learning_rate": 8.446428571428571e-06,
      "loss": 1.3281,
      "step": 87
    },
    {
      "epoch": 0.15714285714285714,
      "grad_norm": 0.63671875,
      "learning_rate": 8.428571428571429e-06,
      "loss": 1.294,
      "step": 88
    },
    {
      "epoch": 0.15892857142857142,
      "grad_norm": 0.1533203125,
      "learning_rate": 8.410714285714287e-06,
      "loss": 1.2808,
      "step": 89
    },
    {
      "epoch": 0.16071428571428573,
      "grad_norm": 0.1611328125,
      "learning_rate": 8.392857142857144e-06,
      "loss": 1.2168,
      "step": 90
    },
    {
      "epoch": 0.1625,
      "grad_norm": 0.173828125,
      "learning_rate": 8.375e-06,
      "loss": 1.2869,
      "step": 91
    },
    {
      "epoch": 0.16428571428571428,
      "grad_norm": 0.162109375,
      "learning_rate": 8.357142857142858e-06,
      "loss": 1.2177,
      "step": 92
    },
    {
      "epoch": 0.16607142857142856,
      "grad_norm": 0.1689453125,
      "learning_rate": 8.339285714285716e-06,
      "loss": 1.336,
      "step": 93
    },
    {
      "epoch": 0.16785714285714284,
      "grad_norm": 0.462890625,
      "learning_rate": 8.321428571428573e-06,
      "loss": 1.2556,
      "step": 94
    },
    {
      "epoch": 0.16964285714285715,
      "grad_norm": 0.2099609375,
      "learning_rate": 8.30357142857143e-06,
      "loss": 1.2662,
      "step": 95
    },
    {
      "epoch": 0.17142857142857143,
      "grad_norm": 0.1787109375,
      "learning_rate": 8.285714285714287e-06,
      "loss": 1.2892,
      "step": 96
    },
    {
      "epoch": 0.1732142857142857,
      "grad_norm": 0.1787109375,
      "learning_rate": 8.267857142857143e-06,
      "loss": 1.3663,
      "step": 97
    },
    {
      "epoch": 0.175,
      "grad_norm": 0.158203125,
      "learning_rate": 8.25e-06,
      "loss": 1.2514,
      "step": 98
    },
    {
      "epoch": 0.1767857142857143,
      "grad_norm": 0.27734375,
      "learning_rate": 8.232142857142857e-06,
      "loss": 1.2026,
      "step": 99
    },
    {
      "epoch": 0.17857142857142858,
      "grad_norm": 0.1533203125,
      "learning_rate": 8.214285714285714e-06,
      "loss": 1.2481,
      "step": 100
    },
    {
      "epoch": 0.18035714285714285,
      "grad_norm": 0.1767578125,
      "learning_rate": 8.196428571428572e-06,
      "loss": 1.3102,
      "step": 101
    },
    {
      "epoch": 0.18214285714285713,
      "grad_norm": 0.16796875,
      "learning_rate": 8.17857142857143e-06,
      "loss": 1.3078,
      "step": 102
    },
    {
      "epoch": 0.18392857142857144,
      "grad_norm": 0.1669921875,
      "learning_rate": 8.160714285714286e-06,
      "loss": 1.2453,
      "step": 103
    },
    {
      "epoch": 0.18571428571428572,
      "grad_norm": 0.205078125,
      "learning_rate": 8.142857142857143e-06,
      "loss": 1.2964,
      "step": 104
    },
    {
      "epoch": 0.1875,
      "grad_norm": 0.2138671875,
      "learning_rate": 8.125000000000001e-06,
      "loss": 1.2259,
      "step": 105
    },
    {
      "epoch": 0.18928571428571428,
      "grad_norm": 0.158203125,
      "learning_rate": 8.107142857142859e-06,
      "loss": 1.2752,
      "step": 106
    },
    {
      "epoch": 0.19107142857142856,
      "grad_norm": 0.17578125,
      "learning_rate": 8.089285714285715e-06,
      "loss": 1.2201,
      "step": 107
    },
    {
      "epoch": 0.19285714285714287,
      "grad_norm": 0.2021484375,
      "learning_rate": 8.071428571428572e-06,
      "loss": 1.2729,
      "step": 108
    },
    {
      "epoch": 0.19464285714285715,
      "grad_norm": 0.181640625,
      "learning_rate": 8.05357142857143e-06,
      "loss": 1.3076,
      "step": 109
    },
    {
      "epoch": 0.19642857142857142,
      "grad_norm": 0.1591796875,
      "learning_rate": 8.035714285714286e-06,
      "loss": 1.2438,
      "step": 110
    },
    {
      "epoch": 0.1982142857142857,
      "grad_norm": 0.177734375,
      "learning_rate": 8.017857142857144e-06,
      "loss": 1.2728,
      "step": 111
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.15625,
      "learning_rate": 8.000000000000001e-06,
      "loss": 1.2819,
      "step": 112
    },
    {
      "epoch": 0.2017857142857143,
      "grad_norm": 0.1708984375,
      "learning_rate": 7.982142857142859e-06,
      "loss": 1.2777,
      "step": 113
    },
    {
      "epoch": 0.20357142857142857,
      "grad_norm": 0.173828125,
      "learning_rate": 7.964285714285715e-06,
      "loss": 1.3234,
      "step": 114
    },
    {
      "epoch": 0.20535714285714285,
      "grad_norm": 0.1630859375,
      "learning_rate": 7.946428571428571e-06,
      "loss": 1.2494,
      "step": 115
    },
    {
      "epoch": 0.20714285714285716,
      "grad_norm": 0.1689453125,
      "learning_rate": 7.928571428571429e-06,
      "loss": 1.2596,
      "step": 116
    },
    {
      "epoch": 0.20892857142857144,
      "grad_norm": 0.1611328125,
      "learning_rate": 7.910714285714286e-06,
      "loss": 1.2633,
      "step": 117
    },
    {
      "epoch": 0.21071428571428572,
      "grad_norm": 0.1669921875,
      "learning_rate": 7.892857142857144e-06,
      "loss": 1.2387,
      "step": 118
    },
    {
      "epoch": 0.2125,
      "grad_norm": 0.158203125,
      "learning_rate": 7.875e-06,
      "loss": 1.2819,
      "step": 119
    },
    {
      "epoch": 0.21428571428571427,
      "grad_norm": 0.162109375,
      "learning_rate": 7.857142857142858e-06,
      "loss": 1.2503,
      "step": 120
    },
    {
      "epoch": 0.21607142857142858,
      "grad_norm": 0.1748046875,
      "learning_rate": 7.839285714285715e-06,
      "loss": 1.2501,
      "step": 121
    },
    {
      "epoch": 0.21785714285714286,
      "grad_norm": 0.1884765625,
      "learning_rate": 7.821428571428571e-06,
      "loss": 1.266,
      "step": 122
    },
    {
      "epoch": 0.21964285714285714,
      "grad_norm": 0.1865234375,
      "learning_rate": 7.803571428571429e-06,
      "loss": 1.2812,
      "step": 123
    },
    {
      "epoch": 0.22142857142857142,
      "grad_norm": 0.171875,
      "learning_rate": 7.785714285714287e-06,
      "loss": 1.259,
      "step": 124
    },
    {
      "epoch": 0.22321428571428573,
      "grad_norm": 0.1767578125,
      "learning_rate": 7.767857142857144e-06,
      "loss": 1.2712,
      "step": 125
    },
    {
      "epoch": 0.225,
      "grad_norm": 0.16796875,
      "learning_rate": 7.75e-06,
      "loss": 1.2627,
      "step": 126
    },
    {
      "epoch": 0.22678571428571428,
      "grad_norm": 0.17578125,
      "learning_rate": 7.732142857142858e-06,
      "loss": 1.2744,
      "step": 127
    },
    {
      "epoch": 0.22857142857142856,
      "grad_norm": 0.169921875,
      "learning_rate": 7.714285714285716e-06,
      "loss": 1.2053,
      "step": 128
    },
    {
      "epoch": 0.23035714285714284,
      "grad_norm": 0.1650390625,
      "learning_rate": 7.696428571428572e-06,
      "loss": 1.2231,
      "step": 129
    },
    {
      "epoch": 0.23214285714285715,
      "grad_norm": 0.19921875,
      "learning_rate": 7.67857142857143e-06,
      "loss": 1.2107,
      "step": 130
    },
    {
      "epoch": 0.23392857142857143,
      "grad_norm": 0.1845703125,
      "learning_rate": 7.660714285714287e-06,
      "loss": 1.1928,
      "step": 131
    },
    {
      "epoch": 0.2357142857142857,
      "grad_norm": 0.1826171875,
      "learning_rate": 7.642857142857143e-06,
      "loss": 1.2871,
      "step": 132
    },
    {
      "epoch": 0.2375,
      "grad_norm": 0.1650390625,
      "learning_rate": 7.625e-06,
      "loss": 1.241,
      "step": 133
    },
    {
      "epoch": 0.2392857142857143,
      "grad_norm": 0.1611328125,
      "learning_rate": 7.6071428571428575e-06,
      "loss": 1.2231,
      "step": 134
    },
    {
      "epoch": 0.24107142857142858,
      "grad_norm": 0.47265625,
      "learning_rate": 7.589285714285714e-06,
      "loss": 1.3065,
      "step": 135
    },
    {
      "epoch": 0.24285714285714285,
      "grad_norm": 0.162109375,
      "learning_rate": 7.571428571428572e-06,
      "loss": 1.1785,
      "step": 136
    },
    {
      "epoch": 0.24464285714285713,
      "grad_norm": 0.1748046875,
      "learning_rate": 7.553571428571429e-06,
      "loss": 1.175,
      "step": 137
    },
    {
      "epoch": 0.24642857142857144,
      "grad_norm": 0.18359375,
      "learning_rate": 7.5357142857142865e-06,
      "loss": 1.2725,
      "step": 138
    },
    {
      "epoch": 0.24821428571428572,
      "grad_norm": 0.166015625,
      "learning_rate": 7.517857142857143e-06,
      "loss": 1.2524,
      "step": 139
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3671875,
      "learning_rate": 7.500000000000001e-06,
      "loss": 1.2102,
      "step": 140
    },
    {
      "epoch": 0.2517857142857143,
      "grad_norm": 0.1767578125,
      "learning_rate": 7.482142857142858e-06,
      "loss": 1.2474,
      "step": 141
    },
    {
      "epoch": 0.25357142857142856,
      "grad_norm": 0.1708984375,
      "learning_rate": 7.464285714285715e-06,
      "loss": 1.2537,
      "step": 142
    },
    {
      "epoch": 0.25535714285714284,
      "grad_norm": 0.2119140625,
      "learning_rate": 7.446428571428572e-06,
      "loss": 1.2221,
      "step": 143
    },
    {
      "epoch": 0.2571428571428571,
      "grad_norm": 0.166015625,
      "learning_rate": 7.428571428571429e-06,
      "loss": 1.1796,
      "step": 144
    },
    {
      "epoch": 0.25892857142857145,
      "grad_norm": 0.1728515625,
      "learning_rate": 7.410714285714287e-06,
      "loss": 1.2549,
      "step": 145
    },
    {
      "epoch": 0.26071428571428573,
      "grad_norm": 0.1689453125,
      "learning_rate": 7.392857142857144e-06,
      "loss": 1.2014,
      "step": 146
    },
    {
      "epoch": 0.2625,
      "grad_norm": 0.1708984375,
      "learning_rate": 7.375000000000001e-06,
      "loss": 1.2295,
      "step": 147
    },
    {
      "epoch": 0.2642857142857143,
      "grad_norm": 0.1640625,
      "learning_rate": 7.357142857142858e-06,
      "loss": 1.2125,
      "step": 148
    },
    {
      "epoch": 0.26607142857142857,
      "grad_norm": 0.1669921875,
      "learning_rate": 7.339285714285714e-06,
      "loss": 1.1909,
      "step": 149
    },
    {
      "epoch": 0.26785714285714285,
      "grad_norm": 0.1650390625,
      "learning_rate": 7.321428571428572e-06,
      "loss": 1.1728,
      "step": 150
    },
    {
      "epoch": 0.26964285714285713,
      "grad_norm": 0.16015625,
      "learning_rate": 7.303571428571429e-06,
      "loss": 1.2091,
      "step": 151
    },
    {
      "epoch": 0.2714285714285714,
      "grad_norm": 0.2060546875,
      "learning_rate": 7.285714285714286e-06,
      "loss": 1.247,
      "step": 152
    },
    {
      "epoch": 0.2732142857142857,
      "grad_norm": 0.2451171875,
      "learning_rate": 7.267857142857143e-06,
      "loss": 1.1899,
      "step": 153
    },
    {
      "epoch": 0.275,
      "grad_norm": 0.1728515625,
      "learning_rate": 7.25e-06,
      "loss": 1.2335,
      "step": 154
    },
    {
      "epoch": 0.2767857142857143,
      "grad_norm": 0.173828125,
      "learning_rate": 7.232142857142858e-06,
      "loss": 1.2507,
      "step": 155
    },
    {
      "epoch": 0.2785714285714286,
      "grad_norm": 0.1875,
      "learning_rate": 7.2142857142857145e-06,
      "loss": 1.2654,
      "step": 156
    },
    {
      "epoch": 0.28035714285714286,
      "grad_norm": 0.8671875,
      "learning_rate": 7.196428571428572e-06,
      "loss": 1.2424,
      "step": 157
    },
    {
      "epoch": 0.28214285714285714,
      "grad_norm": 0.197265625,
      "learning_rate": 7.178571428571429e-06,
      "loss": 1.2825,
      "step": 158
    },
    {
      "epoch": 0.2839285714285714,
      "grad_norm": 0.2080078125,
      "learning_rate": 7.160714285714287e-06,
      "loss": 1.2786,
      "step": 159
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 0.1865234375,
      "learning_rate": 7.1428571428571436e-06,
      "loss": 1.167,
      "step": 160
    },
    {
      "epoch": 0.2875,
      "grad_norm": 0.171875,
      "learning_rate": 7.125e-06,
      "loss": 1.1822,
      "step": 161
    },
    {
      "epoch": 0.2892857142857143,
      "grad_norm": 0.1708984375,
      "learning_rate": 7.107142857142858e-06,
      "loss": 1.252,
      "step": 162
    },
    {
      "epoch": 0.2910714285714286,
      "grad_norm": 0.169921875,
      "learning_rate": 7.089285714285715e-06,
      "loss": 1.187,
      "step": 163
    },
    {
      "epoch": 0.29285714285714287,
      "grad_norm": 0.17578125,
      "learning_rate": 7.0714285714285726e-06,
      "loss": 1.2069,
      "step": 164
    },
    {
      "epoch": 0.29464285714285715,
      "grad_norm": 0.169921875,
      "learning_rate": 7.053571428571429e-06,
      "loss": 1.2166,
      "step": 165
    },
    {
      "epoch": 0.29642857142857143,
      "grad_norm": 0.2177734375,
      "learning_rate": 7.035714285714287e-06,
      "loss": 1.2966,
      "step": 166
    },
    {
      "epoch": 0.2982142857142857,
      "grad_norm": 0.1845703125,
      "learning_rate": 7.017857142857143e-06,
      "loss": 1.2777,
      "step": 167
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1669921875,
      "learning_rate": 7e-06,
      "loss": 1.1945,
      "step": 168
    },
    {
      "epoch": 0.30178571428571427,
      "grad_norm": 0.17578125,
      "learning_rate": 6.9821428571428576e-06,
      "loss": 1.2145,
      "step": 169
    },
    {
      "epoch": 0.30357142857142855,
      "grad_norm": 0.171875,
      "learning_rate": 6.964285714285714e-06,
      "loss": 1.2073,
      "step": 170
    },
    {
      "epoch": 0.3053571428571429,
      "grad_norm": 0.1689453125,
      "learning_rate": 6.946428571428572e-06,
      "loss": 1.1993,
      "step": 171
    },
    {
      "epoch": 0.30714285714285716,
      "grad_norm": 0.173828125,
      "learning_rate": 6.928571428571429e-06,
      "loss": 1.2071,
      "step": 172
    },
    {
      "epoch": 0.30892857142857144,
      "grad_norm": 0.1650390625,
      "learning_rate": 6.910714285714286e-06,
      "loss": 1.2003,
      "step": 173
    },
    {
      "epoch": 0.3107142857142857,
      "grad_norm": 0.177734375,
      "learning_rate": 6.892857142857143e-06,
      "loss": 1.2725,
      "step": 174
    },
    {
      "epoch": 0.3125,
      "grad_norm": 0.18359375,
      "learning_rate": 6.875e-06,
      "loss": 1.1951,
      "step": 175
    },
    {
      "epoch": 0.3142857142857143,
      "grad_norm": 0.2236328125,
      "learning_rate": 6.857142857142858e-06,
      "loss": 1.1859,
      "step": 176
    },
    {
      "epoch": 0.31607142857142856,
      "grad_norm": 0.208984375,
      "learning_rate": 6.839285714285715e-06,
      "loss": 1.2176,
      "step": 177
    },
    {
      "epoch": 0.31785714285714284,
      "grad_norm": 0.177734375,
      "learning_rate": 6.8214285714285724e-06,
      "loss": 1.209,
      "step": 178
    },
    {
      "epoch": 0.3196428571428571,
      "grad_norm": 0.1708984375,
      "learning_rate": 6.803571428571429e-06,
      "loss": 1.1817,
      "step": 179
    },
    {
      "epoch": 0.32142857142857145,
      "grad_norm": 0.203125,
      "learning_rate": 6.785714285714287e-06,
      "loss": 1.2103,
      "step": 180
    },
    {
      "epoch": 0.32321428571428573,
      "grad_norm": 0.234375,
      "learning_rate": 6.767857142857144e-06,
      "loss": 1.1874,
      "step": 181
    },
    {
      "epoch": 0.325,
      "grad_norm": 0.193359375,
      "learning_rate": 6.750000000000001e-06,
      "loss": 1.1934,
      "step": 182
    },
    {
      "epoch": 0.3267857142857143,
      "grad_norm": 0.1962890625,
      "learning_rate": 6.732142857142858e-06,
      "loss": 1.1743,
      "step": 183
    },
    {
      "epoch": 0.32857142857142857,
      "grad_norm": 0.259765625,
      "learning_rate": 6.714285714285714e-06,
      "loss": 1.2188,
      "step": 184
    },
    {
      "epoch": 0.33035714285714285,
      "grad_norm": 0.173828125,
      "learning_rate": 6.696428571428571e-06,
      "loss": 1.175,
      "step": 185
    },
    {
      "epoch": 0.33214285714285713,
      "grad_norm": 0.275390625,
      "learning_rate": 6.678571428571429e-06,
      "loss": 1.2215,
      "step": 186
    },
    {
      "epoch": 0.3339285714285714,
      "grad_norm": 0.1748046875,
      "learning_rate": 6.660714285714286e-06,
      "loss": 1.1251,
      "step": 187
    },
    {
      "epoch": 0.3357142857142857,
      "grad_norm": 0.181640625,
      "learning_rate": 6.642857142857143e-06,
      "loss": 1.1588,
      "step": 188
    },
    {
      "epoch": 0.3375,
      "grad_norm": 0.169921875,
      "learning_rate": 6.625e-06,
      "loss": 1.1694,
      "step": 189
    },
    {
      "epoch": 0.3392857142857143,
      "grad_norm": 0.1806640625,
      "learning_rate": 6.607142857142858e-06,
      "loss": 1.266,
      "step": 190
    },
    {
      "epoch": 0.3410714285714286,
      "grad_norm": 0.1767578125,
      "learning_rate": 6.589285714285715e-06,
      "loss": 1.2323,
      "step": 191
    },
    {
      "epoch": 0.34285714285714286,
      "grad_norm": 0.19921875,
      "learning_rate": 6.571428571428572e-06,
      "loss": 1.1801,
      "step": 192
    },
    {
      "epoch": 0.34464285714285714,
      "grad_norm": 0.224609375,
      "learning_rate": 6.553571428571429e-06,
      "loss": 1.181,
      "step": 193
    },
    {
      "epoch": 0.3464285714285714,
      "grad_norm": 0.1748046875,
      "learning_rate": 6.535714285714286e-06,
      "loss": 1.212,
      "step": 194
    },
    {
      "epoch": 0.3482142857142857,
      "grad_norm": 0.18359375,
      "learning_rate": 6.517857142857144e-06,
      "loss": 1.193,
      "step": 195
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1826171875,
      "learning_rate": 6.5000000000000004e-06,
      "loss": 1.1976,
      "step": 196
    },
    {
      "epoch": 0.3517857142857143,
      "grad_norm": 0.236328125,
      "learning_rate": 6.482142857142858e-06,
      "loss": 1.1992,
      "step": 197
    },
    {
      "epoch": 0.3535714285714286,
      "grad_norm": 0.171875,
      "learning_rate": 6.464285714285715e-06,
      "loss": 1.2138,
      "step": 198
    },
    {
      "epoch": 0.35535714285714287,
      "grad_norm": 0.1943359375,
      "learning_rate": 6.446428571428573e-06,
      "loss": 1.1763,
      "step": 199
    },
    {
      "epoch": 0.35714285714285715,
      "grad_norm": 0.177734375,
      "learning_rate": 6.4285714285714295e-06,
      "loss": 1.1462,
      "step": 200
    },
    {
      "epoch": 0.35892857142857143,
      "grad_norm": 0.18359375,
      "learning_rate": 6.410714285714287e-06,
      "loss": 1.1992,
      "step": 201
    },
    {
      "epoch": 0.3607142857142857,
      "grad_norm": 0.1845703125,
      "learning_rate": 6.392857142857143e-06,
      "loss": 1.1701,
      "step": 202
    },
    {
      "epoch": 0.3625,
      "grad_norm": 0.1826171875,
      "learning_rate": 6.375e-06,
      "loss": 1.1651,
      "step": 203
    },
    {
      "epoch": 0.36428571428571427,
      "grad_norm": 0.2216796875,
      "learning_rate": 6.357142857142858e-06,
      "loss": 1.1889,
      "step": 204
    },
    {
      "epoch": 0.36607142857142855,
      "grad_norm": 0.1826171875,
      "learning_rate": 6.3392857142857145e-06,
      "loss": 1.2059,
      "step": 205
    },
    {
      "epoch": 0.3678571428571429,
      "grad_norm": 0.1796875,
      "learning_rate": 6.321428571428571e-06,
      "loss": 1.1756,
      "step": 206
    },
    {
      "epoch": 0.36964285714285716,
      "grad_norm": 0.302734375,
      "learning_rate": 6.303571428571429e-06,
      "loss": 1.2142,
      "step": 207
    },
    {
      "epoch": 0.37142857142857144,
      "grad_norm": 0.263671875,
      "learning_rate": 6.285714285714286e-06,
      "loss": 1.2188,
      "step": 208
    },
    {
      "epoch": 0.3732142857142857,
      "grad_norm": 0.2431640625,
      "learning_rate": 6.2678571428571435e-06,
      "loss": 1.2105,
      "step": 209
    },
    {
      "epoch": 0.375,
      "grad_norm": 0.271484375,
      "learning_rate": 6.25e-06,
      "loss": 1.1784,
      "step": 210
    },
    {
      "epoch": 0.3767857142857143,
      "grad_norm": 0.255859375,
      "learning_rate": 6.232142857142858e-06,
      "loss": 1.248,
      "step": 211
    },
    {
      "epoch": 0.37857142857142856,
      "grad_norm": 0.18359375,
      "learning_rate": 6.214285714285715e-06,
      "loss": 1.1754,
      "step": 212
    },
    {
      "epoch": 0.38035714285714284,
      "grad_norm": 0.1787109375,
      "learning_rate": 6.1964285714285725e-06,
      "loss": 1.1834,
      "step": 213
    },
    {
      "epoch": 0.3821428571428571,
      "grad_norm": 0.1875,
      "learning_rate": 6.178571428571429e-06,
      "loss": 1.2056,
      "step": 214
    },
    {
      "epoch": 0.38392857142857145,
      "grad_norm": 0.51171875,
      "learning_rate": 6.160714285714286e-06,
      "loss": 1.1652,
      "step": 215
    },
    {
      "epoch": 0.38571428571428573,
      "grad_norm": 0.1796875,
      "learning_rate": 6.142857142857144e-06,
      "loss": 1.1651,
      "step": 216
    },
    {
      "epoch": 0.3875,
      "grad_norm": 0.1826171875,
      "learning_rate": 6.125000000000001e-06,
      "loss": 1.2336,
      "step": 217
    },
    {
      "epoch": 0.3892857142857143,
      "grad_norm": 0.22265625,
      "learning_rate": 6.107142857142858e-06,
      "loss": 1.1685,
      "step": 218
    },
    {
      "epoch": 0.39107142857142857,
      "grad_norm": 0.2158203125,
      "learning_rate": 6.089285714285714e-06,
      "loss": 1.1554,
      "step": 219
    },
    {
      "epoch": 0.39285714285714285,
      "grad_norm": 0.193359375,
      "learning_rate": 6.071428571428571e-06,
      "loss": 1.1878,
      "step": 220
    },
    {
      "epoch": 0.39464285714285713,
      "grad_norm": 0.18359375,
      "learning_rate": 6.053571428571429e-06,
      "loss": 1.1796,
      "step": 221
    },
    {
      "epoch": 0.3964285714285714,
      "grad_norm": 0.2373046875,
      "learning_rate": 6.035714285714286e-06,
      "loss": 1.2006,
      "step": 222
    },
    {
      "epoch": 0.3982142857142857,
      "grad_norm": 0.1923828125,
      "learning_rate": 6.017857142857143e-06,
      "loss": 1.1679,
      "step": 223
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.21875,
      "learning_rate": 6e-06,
      "loss": 1.1688,
      "step": 224
    },
    {
      "epoch": 0.4017857142857143,
      "grad_norm": 0.1748046875,
      "learning_rate": 5.982142857142858e-06,
      "loss": 1.1752,
      "step": 225
    },
    {
      "epoch": 0.4035714285714286,
      "grad_norm": 0.17578125,
      "learning_rate": 5.964285714285715e-06,
      "loss": 1.1462,
      "step": 226
    },
    {
      "epoch": 0.40535714285714286,
      "grad_norm": 0.2197265625,
      "learning_rate": 5.9464285714285715e-06,
      "loss": 1.1739,
      "step": 227
    },
    {
      "epoch": 0.40714285714285714,
      "grad_norm": 0.2041015625,
      "learning_rate": 5.928571428571429e-06,
      "loss": 1.1946,
      "step": 228
    },
    {
      "epoch": 0.4089285714285714,
      "grad_norm": 0.185546875,
      "learning_rate": 5.910714285714286e-06,
      "loss": 1.189,
      "step": 229
    },
    {
      "epoch": 0.4107142857142857,
      "grad_norm": 0.1728515625,
      "learning_rate": 5.892857142857144e-06,
      "loss": 1.1436,
      "step": 230
    },
    {
      "epoch": 0.4125,
      "grad_norm": 0.326171875,
      "learning_rate": 5.8750000000000005e-06,
      "loss": 1.29,
      "step": 231
    },
    {
      "epoch": 0.4142857142857143,
      "grad_norm": 0.1953125,
      "learning_rate": 5.857142857142858e-06,
      "loss": 1.1559,
      "step": 232
    },
    {
      "epoch": 0.4160714285714286,
      "grad_norm": 0.18359375,
      "learning_rate": 5.839285714285715e-06,
      "loss": 1.1212,
      "step": 233
    },
    {
      "epoch": 0.41785714285714287,
      "grad_norm": 0.1953125,
      "learning_rate": 5.821428571428573e-06,
      "loss": 1.1776,
      "step": 234
    },
    {
      "epoch": 0.41964285714285715,
      "grad_norm": 0.1923828125,
      "learning_rate": 5.8035714285714295e-06,
      "loss": 1.2123,
      "step": 235
    },
    {
      "epoch": 0.42142857142857143,
      "grad_norm": 0.2177734375,
      "learning_rate": 5.785714285714286e-06,
      "loss": 1.2795,
      "step": 236
    },
    {
      "epoch": 0.4232142857142857,
      "grad_norm": 0.78125,
      "learning_rate": 5.767857142857143e-06,
      "loss": 1.1719,
      "step": 237
    },
    {
      "epoch": 0.425,
      "grad_norm": 0.1826171875,
      "learning_rate": 5.75e-06,
      "loss": 1.1923,
      "step": 238
    },
    {
      "epoch": 0.42678571428571427,
      "grad_norm": 0.2353515625,
      "learning_rate": 5.732142857142857e-06,
      "loss": 1.2586,
      "step": 239
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 0.1806640625,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 1.1634,
      "step": 240
    },
    {
      "epoch": 0.4303571428571429,
      "grad_norm": 0.2021484375,
      "learning_rate": 5.696428571428571e-06,
      "loss": 1.1752,
      "step": 241
    },
    {
      "epoch": 0.43214285714285716,
      "grad_norm": 0.3046875,
      "learning_rate": 5.678571428571429e-06,
      "loss": 1.1835,
      "step": 242
    },
    {
      "epoch": 0.43392857142857144,
      "grad_norm": 0.1884765625,
      "learning_rate": 5.660714285714286e-06,
      "loss": 1.1811,
      "step": 243
    },
    {
      "epoch": 0.4357142857142857,
      "grad_norm": 0.291015625,
      "learning_rate": 5.6428571428571435e-06,
      "loss": 1.3322,
      "step": 244
    },
    {
      "epoch": 0.4375,
      "grad_norm": 2.234375,
      "learning_rate": 5.625e-06,
      "loss": 1.1047,
      "step": 245
    },
    {
      "epoch": 0.4392857142857143,
      "grad_norm": 0.1875,
      "learning_rate": 5.607142857142858e-06,
      "loss": 1.1832,
      "step": 246
    },
    {
      "epoch": 0.44107142857142856,
      "grad_norm": 0.1826171875,
      "learning_rate": 5.589285714285715e-06,
      "loss": 1.2046,
      "step": 247
    },
    {
      "epoch": 0.44285714285714284,
      "grad_norm": 0.1923828125,
      "learning_rate": 5.571428571428572e-06,
      "loss": 1.0937,
      "step": 248
    },
    {
      "epoch": 0.4446428571428571,
      "grad_norm": 0.2138671875,
      "learning_rate": 5.553571428571429e-06,
      "loss": 1.208,
      "step": 249
    },
    {
      "epoch": 0.44642857142857145,
      "grad_norm": 0.197265625,
      "learning_rate": 5.535714285714286e-06,
      "loss": 1.1323,
      "step": 250
    },
    {
      "epoch": 0.44821428571428573,
      "grad_norm": 0.197265625,
      "learning_rate": 5.517857142857144e-06,
      "loss": 1.1391,
      "step": 251
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.19140625,
      "learning_rate": 5.500000000000001e-06,
      "loss": 1.16,
      "step": 252
    },
    {
      "epoch": 0.4517857142857143,
      "grad_norm": 0.1748046875,
      "learning_rate": 5.482142857142858e-06,
      "loss": 1.1171,
      "step": 253
    },
    {
      "epoch": 0.45357142857142857,
      "grad_norm": 0.189453125,
      "learning_rate": 5.464285714285714e-06,
      "loss": 1.2083,
      "step": 254
    },
    {
      "epoch": 0.45535714285714285,
      "grad_norm": 0.21484375,
      "learning_rate": 5.446428571428571e-06,
      "loss": 1.1761,
      "step": 255
    },
    {
      "epoch": 0.45714285714285713,
      "grad_norm": 0.197265625,
      "learning_rate": 5.428571428571429e-06,
      "loss": 1.1479,
      "step": 256
    },
    {
      "epoch": 0.4589285714285714,
      "grad_norm": 0.1953125,
      "learning_rate": 5.410714285714286e-06,
      "loss": 1.1903,
      "step": 257
    },
    {
      "epoch": 0.4607142857142857,
      "grad_norm": 0.2138671875,
      "learning_rate": 5.392857142857143e-06,
      "loss": 1.1795,
      "step": 258
    },
    {
      "epoch": 0.4625,
      "grad_norm": 0.306640625,
      "learning_rate": 5.375e-06,
      "loss": 1.233,
      "step": 259
    },
    {
      "epoch": 0.4642857142857143,
      "grad_norm": 0.1845703125,
      "learning_rate": 5.357142857142857e-06,
      "loss": 1.1406,
      "step": 260
    },
    {
      "epoch": 0.4660714285714286,
      "grad_norm": 0.18359375,
      "learning_rate": 5.339285714285715e-06,
      "loss": 1.176,
      "step": 261
    },
    {
      "epoch": 0.46785714285714286,
      "grad_norm": 0.189453125,
      "learning_rate": 5.3214285714285715e-06,
      "loss": 1.1739,
      "step": 262
    },
    {
      "epoch": 0.46964285714285714,
      "grad_norm": 0.2041015625,
      "learning_rate": 5.303571428571429e-06,
      "loss": 1.1291,
      "step": 263
    },
    {
      "epoch": 0.4714285714285714,
      "grad_norm": 0.193359375,
      "learning_rate": 5.285714285714286e-06,
      "loss": 1.215,
      "step": 264
    },
    {
      "epoch": 0.4732142857142857,
      "grad_norm": 0.2119140625,
      "learning_rate": 5.267857142857144e-06,
      "loss": 1.1484,
      "step": 265
    },
    {
      "epoch": 0.475,
      "grad_norm": 0.205078125,
      "learning_rate": 5.2500000000000006e-06,
      "loss": 1.1621,
      "step": 266
    },
    {
      "epoch": 0.4767857142857143,
      "grad_norm": 0.181640625,
      "learning_rate": 5.232142857142858e-06,
      "loss": 1.1341,
      "step": 267
    },
    {
      "epoch": 0.4785714285714286,
      "grad_norm": 0.1953125,
      "learning_rate": 5.214285714285715e-06,
      "loss": 1.1793,
      "step": 268
    },
    {
      "epoch": 0.48035714285714287,
      "grad_norm": 0.189453125,
      "learning_rate": 5.196428571428572e-06,
      "loss": 1.1767,
      "step": 269
    },
    {
      "epoch": 0.48214285714285715,
      "grad_norm": 0.1875,
      "learning_rate": 5.1785714285714296e-06,
      "loss": 1.1695,
      "step": 270
    },
    {
      "epoch": 0.48392857142857143,
      "grad_norm": 0.197265625,
      "learning_rate": 5.160714285714286e-06,
      "loss": 1.21,
      "step": 271
    },
    {
      "epoch": 0.4857142857142857,
      "grad_norm": 0.1787109375,
      "learning_rate": 5.142857142857142e-06,
      "loss": 1.1121,
      "step": 272
    },
    {
      "epoch": 0.4875,
      "grad_norm": 0.2060546875,
      "learning_rate": 5.125e-06,
      "loss": 1.1956,
      "step": 273
    },
    {
      "epoch": 0.48928571428571427,
      "grad_norm": 0.2333984375,
      "learning_rate": 5.107142857142857e-06,
      "loss": 1.1052,
      "step": 274
    },
    {
      "epoch": 0.49107142857142855,
      "grad_norm": 0.185546875,
      "learning_rate": 5.0892857142857146e-06,
      "loss": 1.1829,
      "step": 275
    },
    {
      "epoch": 0.4928571428571429,
      "grad_norm": 0.263671875,
      "learning_rate": 5.071428571428571e-06,
      "loss": 1.1666,
      "step": 276
    },
    {
      "epoch": 0.49464285714285716,
      "grad_norm": 0.2314453125,
      "learning_rate": 5.053571428571429e-06,
      "loss": 1.1083,
      "step": 277
    },
    {
      "epoch": 0.49642857142857144,
      "grad_norm": 0.20703125,
      "learning_rate": 5.035714285714286e-06,
      "loss": 1.166,
      "step": 278
    },
    {
      "epoch": 0.4982142857142857,
      "grad_norm": 0.1865234375,
      "learning_rate": 5.017857142857144e-06,
      "loss": 1.168,
      "step": 279
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1923828125,
      "learning_rate": 5e-06,
      "loss": 1.1707,
      "step": 280
    },
    {
      "epoch": 0.5017857142857143,
      "grad_norm": 0.181640625,
      "learning_rate": 4.982142857142857e-06,
      "loss": 1.1735,
      "step": 281
    },
    {
      "epoch": 0.5035714285714286,
      "grad_norm": 0.19140625,
      "learning_rate": 4.964285714285715e-06,
      "loss": 1.1395,
      "step": 282
    },
    {
      "epoch": 0.5053571428571428,
      "grad_norm": 0.1953125,
      "learning_rate": 4.946428571428572e-06,
      "loss": 1.1439,
      "step": 283
    },
    {
      "epoch": 0.5071428571428571,
      "grad_norm": 0.244140625,
      "learning_rate": 4.928571428571429e-06,
      "loss": 1.1547,
      "step": 284
    },
    {
      "epoch": 0.5089285714285714,
      "grad_norm": 0.1953125,
      "learning_rate": 4.910714285714286e-06,
      "loss": 1.2232,
      "step": 285
    },
    {
      "epoch": 0.5107142857142857,
      "grad_norm": 0.19921875,
      "learning_rate": 4.892857142857143e-06,
      "loss": 1.1607,
      "step": 286
    },
    {
      "epoch": 0.5125,
      "grad_norm": 0.1806640625,
      "learning_rate": 4.875e-06,
      "loss": 1.1799,
      "step": 287
    },
    {
      "epoch": 0.5142857142857142,
      "grad_norm": 0.18359375,
      "learning_rate": 4.857142857142858e-06,
      "loss": 1.1534,
      "step": 288
    },
    {
      "epoch": 0.5160714285714286,
      "grad_norm": 0.2060546875,
      "learning_rate": 4.839285714285714e-06,
      "loss": 1.1432,
      "step": 289
    },
    {
      "epoch": 0.5178571428571429,
      "grad_norm": 0.1943359375,
      "learning_rate": 4.821428571428572e-06,
      "loss": 1.1573,
      "step": 290
    },
    {
      "epoch": 0.5196428571428572,
      "grad_norm": 0.1923828125,
      "learning_rate": 4.803571428571429e-06,
      "loss": 1.1567,
      "step": 291
    },
    {
      "epoch": 0.5214285714285715,
      "grad_norm": 0.1884765625,
      "learning_rate": 4.785714285714287e-06,
      "loss": 1.163,
      "step": 292
    },
    {
      "epoch": 0.5232142857142857,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.7678571428571434e-06,
      "loss": 1.1372,
      "step": 293
    },
    {
      "epoch": 0.525,
      "grad_norm": 0.18359375,
      "learning_rate": 4.75e-06,
      "loss": 1.1123,
      "step": 294
    },
    {
      "epoch": 0.5267857142857143,
      "grad_norm": 0.2109375,
      "learning_rate": 4.732142857142857e-06,
      "loss": 1.1416,
      "step": 295
    },
    {
      "epoch": 0.5285714285714286,
      "grad_norm": 0.193359375,
      "learning_rate": 4.714285714285715e-06,
      "loss": 1.1302,
      "step": 296
    },
    {
      "epoch": 0.5303571428571429,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.696428571428572e-06,
      "loss": 1.1657,
      "step": 297
    },
    {
      "epoch": 0.5321428571428571,
      "grad_norm": 0.1796875,
      "learning_rate": 4.678571428571429e-06,
      "loss": 1.1288,
      "step": 298
    },
    {
      "epoch": 0.5339285714285714,
      "grad_norm": 0.185546875,
      "learning_rate": 4.660714285714286e-06,
      "loss": 1.1896,
      "step": 299
    },
    {
      "epoch": 0.5357142857142857,
      "grad_norm": 0.1923828125,
      "learning_rate": 4.642857142857144e-06,
      "loss": 1.1692,
      "step": 300
    },
    {
      "epoch": 0.5375,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.625000000000001e-06,
      "loss": 1.1562,
      "step": 301
    },
    {
      "epoch": 0.5392857142857143,
      "grad_norm": 0.21484375,
      "learning_rate": 4.6071428571428574e-06,
      "loss": 1.1978,
      "step": 302
    },
    {
      "epoch": 0.5410714285714285,
      "grad_norm": 0.2001953125,
      "learning_rate": 4.589285714285714e-06,
      "loss": 1.1697,
      "step": 303
    },
    {
      "epoch": 0.5428571428571428,
      "grad_norm": 0.2177734375,
      "learning_rate": 4.571428571428572e-06,
      "loss": 1.2107,
      "step": 304
    },
    {
      "epoch": 0.5446428571428571,
      "grad_norm": 0.19921875,
      "learning_rate": 4.553571428571429e-06,
      "loss": 1.1483,
      "step": 305
    },
    {
      "epoch": 0.5464285714285714,
      "grad_norm": 0.1904296875,
      "learning_rate": 4.5357142857142865e-06,
      "loss": 1.1341,
      "step": 306
    },
    {
      "epoch": 0.5482142857142858,
      "grad_norm": 0.2001953125,
      "learning_rate": 4.517857142857143e-06,
      "loss": 1.1576,
      "step": 307
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.212890625,
      "learning_rate": 4.5e-06,
      "loss": 1.2077,
      "step": 308
    },
    {
      "epoch": 0.5517857142857143,
      "grad_norm": 0.1875,
      "learning_rate": 4.482142857142858e-06,
      "loss": 1.1227,
      "step": 309
    },
    {
      "epoch": 0.5535714285714286,
      "grad_norm": 0.31640625,
      "learning_rate": 4.464285714285715e-06,
      "loss": 1.1901,
      "step": 310
    },
    {
      "epoch": 0.5553571428571429,
      "grad_norm": 0.1904296875,
      "learning_rate": 4.4464285714285715e-06,
      "loss": 1.1644,
      "step": 311
    },
    {
      "epoch": 0.5571428571428572,
      "grad_norm": 0.18359375,
      "learning_rate": 4.428571428571429e-06,
      "loss": 1.2139,
      "step": 312
    },
    {
      "epoch": 0.5589285714285714,
      "grad_norm": 0.18359375,
      "learning_rate": 4.410714285714286e-06,
      "loss": 1.1444,
      "step": 313
    },
    {
      "epoch": 0.5607142857142857,
      "grad_norm": 0.19140625,
      "learning_rate": 4.392857142857143e-06,
      "loss": 1.1729,
      "step": 314
    },
    {
      "epoch": 0.5625,
      "grad_norm": 0.21875,
      "learning_rate": 4.3750000000000005e-06,
      "loss": 1.1465,
      "step": 315
    },
    {
      "epoch": 0.5642857142857143,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.357142857142857e-06,
      "loss": 1.1259,
      "step": 316
    },
    {
      "epoch": 0.5660714285714286,
      "grad_norm": 0.212890625,
      "learning_rate": 4.339285714285715e-06,
      "loss": 1.1134,
      "step": 317
    },
    {
      "epoch": 0.5678571428571428,
      "grad_norm": 0.181640625,
      "learning_rate": 4.321428571428572e-06,
      "loss": 1.145,
      "step": 318
    },
    {
      "epoch": 0.5696428571428571,
      "grad_norm": 0.2119140625,
      "learning_rate": 4.3035714285714295e-06,
      "loss": 1.1057,
      "step": 319
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 0.189453125,
      "learning_rate": 4.2857142857142855e-06,
      "loss": 1.1909,
      "step": 320
    },
    {
      "epoch": 0.5732142857142857,
      "grad_norm": 0.1953125,
      "learning_rate": 4.267857142857143e-06,
      "loss": 1.1526,
      "step": 321
    },
    {
      "epoch": 0.575,
      "grad_norm": 0.2119140625,
      "learning_rate": 4.25e-06,
      "loss": 1.1345,
      "step": 322
    },
    {
      "epoch": 0.5767857142857142,
      "grad_norm": 0.197265625,
      "learning_rate": 4.232142857142858e-06,
      "loss": 1.1276,
      "step": 323
    },
    {
      "epoch": 0.5785714285714286,
      "grad_norm": 0.208984375,
      "learning_rate": 4.2142857142857145e-06,
      "loss": 1.1772,
      "step": 324
    },
    {
      "epoch": 0.5803571428571429,
      "grad_norm": 0.1953125,
      "learning_rate": 4.196428571428572e-06,
      "loss": 1.1043,
      "step": 325
    },
    {
      "epoch": 0.5821428571428572,
      "grad_norm": 0.43359375,
      "learning_rate": 4.178571428571429e-06,
      "loss": 1.1695,
      "step": 326
    },
    {
      "epoch": 0.5839285714285715,
      "grad_norm": 0.201171875,
      "learning_rate": 4.160714285714287e-06,
      "loss": 1.0878,
      "step": 327
    },
    {
      "epoch": 0.5857142857142857,
      "grad_norm": 0.212890625,
      "learning_rate": 4.1428571428571435e-06,
      "loss": 1.175,
      "step": 328
    },
    {
      "epoch": 0.5875,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.125e-06,
      "loss": 1.1478,
      "step": 329
    },
    {
      "epoch": 0.5892857142857143,
      "grad_norm": 0.1923828125,
      "learning_rate": 4.107142857142857e-06,
      "loss": 1.166,
      "step": 330
    },
    {
      "epoch": 0.5910714285714286,
      "grad_norm": 0.1962890625,
      "learning_rate": 4.089285714285715e-06,
      "loss": 1.1393,
      "step": 331
    },
    {
      "epoch": 0.5928571428571429,
      "grad_norm": 0.2021484375,
      "learning_rate": 4.071428571428572e-06,
      "loss": 1.1259,
      "step": 332
    },
    {
      "epoch": 0.5946428571428571,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.053571428571429e-06,
      "loss": 1.1298,
      "step": 333
    },
    {
      "epoch": 0.5964285714285714,
      "grad_norm": 0.1962890625,
      "learning_rate": 4.035714285714286e-06,
      "loss": 1.1384,
      "step": 334
    },
    {
      "epoch": 0.5982142857142857,
      "grad_norm": 0.2001953125,
      "learning_rate": 4.017857142857143e-06,
      "loss": 1.1776,
      "step": 335
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.193359375,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.1937,
      "step": 336
    },
    {
      "epoch": 0.6017857142857143,
      "grad_norm": 0.25,
      "learning_rate": 3.9821428571428575e-06,
      "loss": 1.1618,
      "step": 337
    },
    {
      "epoch": 0.6035714285714285,
      "grad_norm": 0.189453125,
      "learning_rate": 3.964285714285714e-06,
      "loss": 1.1387,
      "step": 338
    },
    {
      "epoch": 0.6053571428571428,
      "grad_norm": 0.1943359375,
      "learning_rate": 3.946428571428572e-06,
      "loss": 1.1479,
      "step": 339
    },
    {
      "epoch": 0.6071428571428571,
      "grad_norm": 0.1845703125,
      "learning_rate": 3.928571428571429e-06,
      "loss": 1.1185,
      "step": 340
    },
    {
      "epoch": 0.6089285714285714,
      "grad_norm": 0.208984375,
      "learning_rate": 3.910714285714286e-06,
      "loss": 1.1447,
      "step": 341
    },
    {
      "epoch": 0.6107142857142858,
      "grad_norm": 0.2021484375,
      "learning_rate": 3.892857142857143e-06,
      "loss": 1.1693,
      "step": 342
    },
    {
      "epoch": 0.6125,
      "grad_norm": 0.18359375,
      "learning_rate": 3.875e-06,
      "loss": 1.1268,
      "step": 343
    },
    {
      "epoch": 0.6142857142857143,
      "grad_norm": 0.197265625,
      "learning_rate": 3.857142857142858e-06,
      "loss": 1.1435,
      "step": 344
    },
    {
      "epoch": 0.6160714285714286,
      "grad_norm": 0.2099609375,
      "learning_rate": 3.839285714285715e-06,
      "loss": 1.102,
      "step": 345
    },
    {
      "epoch": 0.6178571428571429,
      "grad_norm": 0.2294921875,
      "learning_rate": 3.8214285714285715e-06,
      "loss": 1.0893,
      "step": 346
    },
    {
      "epoch": 0.6196428571428572,
      "grad_norm": 0.185546875,
      "learning_rate": 3.8035714285714288e-06,
      "loss": 1.1115,
      "step": 347
    },
    {
      "epoch": 0.6214285714285714,
      "grad_norm": 0.216796875,
      "learning_rate": 3.785714285714286e-06,
      "loss": 1.1497,
      "step": 348
    },
    {
      "epoch": 0.6232142857142857,
      "grad_norm": 0.185546875,
      "learning_rate": 3.7678571428571433e-06,
      "loss": 1.1288,
      "step": 349
    },
    {
      "epoch": 0.625,
      "grad_norm": 0.1865234375,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 1.1183,
      "step": 350
    },
    {
      "epoch": 0.6267857142857143,
      "grad_norm": 0.189453125,
      "learning_rate": 3.7321428571428573e-06,
      "loss": 1.1357,
      "step": 351
    },
    {
      "epoch": 0.6285714285714286,
      "grad_norm": 0.189453125,
      "learning_rate": 3.7142857142857146e-06,
      "loss": 1.1683,
      "step": 352
    },
    {
      "epoch": 0.6303571428571428,
      "grad_norm": 0.1962890625,
      "learning_rate": 3.696428571428572e-06,
      "loss": 1.1582,
      "step": 353
    },
    {
      "epoch": 0.6321428571428571,
      "grad_norm": 0.2041015625,
      "learning_rate": 3.678571428571429e-06,
      "loss": 1.1508,
      "step": 354
    },
    {
      "epoch": 0.6339285714285714,
      "grad_norm": 0.1884765625,
      "learning_rate": 3.660714285714286e-06,
      "loss": 1.1375,
      "step": 355
    },
    {
      "epoch": 0.6357142857142857,
      "grad_norm": 0.2041015625,
      "learning_rate": 3.642857142857143e-06,
      "loss": 1.1921,
      "step": 356
    },
    {
      "epoch": 0.6375,
      "grad_norm": 0.181640625,
      "learning_rate": 3.625e-06,
      "loss": 1.0964,
      "step": 357
    },
    {
      "epoch": 0.6392857142857142,
      "grad_norm": 0.1904296875,
      "learning_rate": 3.6071428571428573e-06,
      "loss": 1.1554,
      "step": 358
    },
    {
      "epoch": 0.6410714285714286,
      "grad_norm": 0.185546875,
      "learning_rate": 3.5892857142857145e-06,
      "loss": 1.1234,
      "step": 359
    },
    {
      "epoch": 0.6428571428571429,
      "grad_norm": 0.1884765625,
      "learning_rate": 3.5714285714285718e-06,
      "loss": 1.1793,
      "step": 360
    },
    {
      "epoch": 0.6446428571428572,
      "grad_norm": 0.2265625,
      "learning_rate": 3.553571428571429e-06,
      "loss": 1.1054,
      "step": 361
    },
    {
      "epoch": 0.6464285714285715,
      "grad_norm": 0.189453125,
      "learning_rate": 3.5357142857142863e-06,
      "loss": 1.1465,
      "step": 362
    },
    {
      "epoch": 0.6482142857142857,
      "grad_norm": 0.1884765625,
      "learning_rate": 3.5178571428571435e-06,
      "loss": 1.1645,
      "step": 363
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1884765625,
      "learning_rate": 3.5e-06,
      "loss": 1.1173,
      "step": 364
    },
    {
      "epoch": 0.6517857142857143,
      "grad_norm": 0.328125,
      "learning_rate": 3.482142857142857e-06,
      "loss": 1.1352,
      "step": 365
    },
    {
      "epoch": 0.6535714285714286,
      "grad_norm": 0.228515625,
      "learning_rate": 3.4642857142857145e-06,
      "loss": 1.1772,
      "step": 366
    },
    {
      "epoch": 0.6553571428571429,
      "grad_norm": 0.2158203125,
      "learning_rate": 3.4464285714285717e-06,
      "loss": 1.1848,
      "step": 367
    },
    {
      "epoch": 0.6571428571428571,
      "grad_norm": 0.216796875,
      "learning_rate": 3.428571428571429e-06,
      "loss": 1.1608,
      "step": 368
    },
    {
      "epoch": 0.6589285714285714,
      "grad_norm": 0.2119140625,
      "learning_rate": 3.4107142857142862e-06,
      "loss": 1.1825,
      "step": 369
    },
    {
      "epoch": 0.6607142857142857,
      "grad_norm": 0.197265625,
      "learning_rate": 3.3928571428571435e-06,
      "loss": 1.1475,
      "step": 370
    },
    {
      "epoch": 0.6625,
      "grad_norm": 0.236328125,
      "learning_rate": 3.3750000000000003e-06,
      "loss": 1.1106,
      "step": 371
    },
    {
      "epoch": 0.6642857142857143,
      "grad_norm": 0.185546875,
      "learning_rate": 3.357142857142857e-06,
      "loss": 1.1679,
      "step": 372
    },
    {
      "epoch": 0.6660714285714285,
      "grad_norm": 0.189453125,
      "learning_rate": 3.3392857142857144e-06,
      "loss": 1.1394,
      "step": 373
    },
    {
      "epoch": 0.6678571428571428,
      "grad_norm": 0.2001953125,
      "learning_rate": 3.3214285714285716e-06,
      "loss": 1.1166,
      "step": 374
    },
    {
      "epoch": 0.6696428571428571,
      "grad_norm": 0.197265625,
      "learning_rate": 3.303571428571429e-06,
      "loss": 1.2486,
      "step": 375
    },
    {
      "epoch": 0.6714285714285714,
      "grad_norm": 0.1904296875,
      "learning_rate": 3.285714285714286e-06,
      "loss": 1.0829,
      "step": 376
    },
    {
      "epoch": 0.6732142857142858,
      "grad_norm": 0.1884765625,
      "learning_rate": 3.267857142857143e-06,
      "loss": 1.1629,
      "step": 377
    },
    {
      "epoch": 0.675,
      "grad_norm": 0.1943359375,
      "learning_rate": 3.2500000000000002e-06,
      "loss": 1.1803,
      "step": 378
    },
    {
      "epoch": 0.6767857142857143,
      "grad_norm": 0.2138671875,
      "learning_rate": 3.2321428571428575e-06,
      "loss": 1.127,
      "step": 379
    },
    {
      "epoch": 0.6785714285714286,
      "grad_norm": 0.234375,
      "learning_rate": 3.2142857142857147e-06,
      "loss": 1.1411,
      "step": 380
    },
    {
      "epoch": 0.6803571428571429,
      "grad_norm": 0.19140625,
      "learning_rate": 3.1964285714285716e-06,
      "loss": 1.1343,
      "step": 381
    },
    {
      "epoch": 0.6821428571428572,
      "grad_norm": 0.2041015625,
      "learning_rate": 3.178571428571429e-06,
      "loss": 1.1033,
      "step": 382
    },
    {
      "epoch": 0.6839285714285714,
      "grad_norm": 0.20703125,
      "learning_rate": 3.1607142857142856e-06,
      "loss": 1.1471,
      "step": 383
    },
    {
      "epoch": 0.6857142857142857,
      "grad_norm": 0.23046875,
      "learning_rate": 3.142857142857143e-06,
      "loss": 1.1144,
      "step": 384
    },
    {
      "epoch": 0.6875,
      "grad_norm": 0.2236328125,
      "learning_rate": 3.125e-06,
      "loss": 1.1688,
      "step": 385
    },
    {
      "epoch": 0.6892857142857143,
      "grad_norm": 0.55859375,
      "learning_rate": 3.1071428571428574e-06,
      "loss": 1.2375,
      "step": 386
    },
    {
      "epoch": 0.6910714285714286,
      "grad_norm": 0.2041015625,
      "learning_rate": 3.0892857142857147e-06,
      "loss": 1.1086,
      "step": 387
    },
    {
      "epoch": 0.6928571428571428,
      "grad_norm": 0.2021484375,
      "learning_rate": 3.071428571428572e-06,
      "loss": 1.1237,
      "step": 388
    },
    {
      "epoch": 0.6946428571428571,
      "grad_norm": 0.205078125,
      "learning_rate": 3.053571428571429e-06,
      "loss": 1.2193,
      "step": 389
    },
    {
      "epoch": 0.6964285714285714,
      "grad_norm": 0.19921875,
      "learning_rate": 3.0357142857142856e-06,
      "loss": 1.1369,
      "step": 390
    },
    {
      "epoch": 0.6982142857142857,
      "grad_norm": 0.1962890625,
      "learning_rate": 3.017857142857143e-06,
      "loss": 1.1489,
      "step": 391
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2080078125,
      "learning_rate": 3e-06,
      "loss": 1.1448,
      "step": 392
    },
    {
      "epoch": 0.7017857142857142,
      "grad_norm": 0.2021484375,
      "learning_rate": 2.9821428571428573e-06,
      "loss": 1.1144,
      "step": 393
    },
    {
      "epoch": 0.7035714285714286,
      "grad_norm": 0.19140625,
      "learning_rate": 2.9642857142857146e-06,
      "loss": 1.1157,
      "step": 394
    },
    {
      "epoch": 0.7053571428571429,
      "grad_norm": 0.1875,
      "learning_rate": 2.946428571428572e-06,
      "loss": 1.1229,
      "step": 395
    },
    {
      "epoch": 0.7071428571428572,
      "grad_norm": 0.2080078125,
      "learning_rate": 2.928571428571429e-06,
      "loss": 1.1004,
      "step": 396
    },
    {
      "epoch": 0.7089285714285715,
      "grad_norm": 0.2119140625,
      "learning_rate": 2.9107142857142863e-06,
      "loss": 1.1635,
      "step": 397
    },
    {
      "epoch": 0.7107142857142857,
      "grad_norm": 0.189453125,
      "learning_rate": 2.892857142857143e-06,
      "loss": 1.1401,
      "step": 398
    },
    {
      "epoch": 0.7125,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.875e-06,
      "loss": 1.1199,
      "step": 399
    },
    {
      "epoch": 0.7142857142857143,
      "grad_norm": 0.19921875,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 1.1308,
      "step": 400
    },
    {
      "epoch": 0.7160714285714286,
      "grad_norm": 0.19921875,
      "learning_rate": 2.8392857142857145e-06,
      "loss": 1.134,
      "step": 401
    },
    {
      "epoch": 0.7178571428571429,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.8214285714285718e-06,
      "loss": 1.1755,
      "step": 402
    },
    {
      "epoch": 0.7196428571428571,
      "grad_norm": 0.1865234375,
      "learning_rate": 2.803571428571429e-06,
      "loss": 1.1561,
      "step": 403
    },
    {
      "epoch": 0.7214285714285714,
      "grad_norm": 0.203125,
      "learning_rate": 2.785714285714286e-06,
      "loss": 1.0835,
      "step": 404
    },
    {
      "epoch": 0.7232142857142857,
      "grad_norm": 0.1953125,
      "learning_rate": 2.767857142857143e-06,
      "loss": 1.1308,
      "step": 405
    },
    {
      "epoch": 0.725,
      "grad_norm": 0.2255859375,
      "learning_rate": 2.7500000000000004e-06,
      "loss": 1.1434,
      "step": 406
    },
    {
      "epoch": 0.7267857142857143,
      "grad_norm": 0.2734375,
      "learning_rate": 2.732142857142857e-06,
      "loss": 1.1328,
      "step": 407
    },
    {
      "epoch": 0.7285714285714285,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.7142857142857144e-06,
      "loss": 1.1019,
      "step": 408
    },
    {
      "epoch": 0.7303571428571428,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.6964285714285717e-06,
      "loss": 1.2163,
      "step": 409
    },
    {
      "epoch": 0.7321428571428571,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.6785714285714285e-06,
      "loss": 1.1774,
      "step": 410
    },
    {
      "epoch": 0.7339285714285714,
      "grad_norm": 0.21484375,
      "learning_rate": 2.6607142857142858e-06,
      "loss": 1.1145,
      "step": 411
    },
    {
      "epoch": 0.7357142857142858,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.642857142857143e-06,
      "loss": 1.1181,
      "step": 412
    },
    {
      "epoch": 0.7375,
      "grad_norm": 0.2255859375,
      "learning_rate": 2.6250000000000003e-06,
      "loss": 1.1201,
      "step": 413
    },
    {
      "epoch": 0.7392857142857143,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.6071428571428575e-06,
      "loss": 1.1604,
      "step": 414
    },
    {
      "epoch": 0.7410714285714286,
      "grad_norm": 0.193359375,
      "learning_rate": 2.5892857142857148e-06,
      "loss": 1.1095,
      "step": 415
    },
    {
      "epoch": 0.7428571428571429,
      "grad_norm": 0.197265625,
      "learning_rate": 2.571428571428571e-06,
      "loss": 1.1058,
      "step": 416
    },
    {
      "epoch": 0.7446428571428572,
      "grad_norm": 0.24609375,
      "learning_rate": 2.5535714285714284e-06,
      "loss": 1.1625,
      "step": 417
    },
    {
      "epoch": 0.7464285714285714,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.5357142857142857e-06,
      "loss": 1.1778,
      "step": 418
    },
    {
      "epoch": 0.7482142857142857,
      "grad_norm": 0.2255859375,
      "learning_rate": 2.517857142857143e-06,
      "loss": 1.136,
      "step": 419
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.5e-06,
      "loss": 1.1642,
      "step": 420
    },
    {
      "epoch": 0.7517857142857143,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.4821428571428575e-06,
      "loss": 1.1833,
      "step": 421
    },
    {
      "epoch": 0.7535714285714286,
      "grad_norm": 0.1875,
      "learning_rate": 2.4642857142857147e-06,
      "loss": 1.1342,
      "step": 422
    },
    {
      "epoch": 0.7553571428571428,
      "grad_norm": 0.255859375,
      "learning_rate": 2.4464285714285715e-06,
      "loss": 1.2057,
      "step": 423
    },
    {
      "epoch": 0.7571428571428571,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.428571428571429e-06,
      "loss": 1.141,
      "step": 424
    },
    {
      "epoch": 0.7589285714285714,
      "grad_norm": 0.279296875,
      "learning_rate": 2.410714285714286e-06,
      "loss": 1.1426,
      "step": 425
    },
    {
      "epoch": 0.7607142857142857,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.3928571428571433e-06,
      "loss": 1.1725,
      "step": 426
    },
    {
      "epoch": 0.7625,
      "grad_norm": 0.31640625,
      "learning_rate": 2.375e-06,
      "loss": 1.1323,
      "step": 427
    },
    {
      "epoch": 0.7642857142857142,
      "grad_norm": 0.197265625,
      "learning_rate": 2.3571428571428574e-06,
      "loss": 1.1337,
      "step": 428
    },
    {
      "epoch": 0.7660714285714286,
      "grad_norm": 0.279296875,
      "learning_rate": 2.3392857142857146e-06,
      "loss": 1.1141,
      "step": 429
    },
    {
      "epoch": 0.7678571428571429,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.321428571428572e-06,
      "loss": 1.1064,
      "step": 430
    },
    {
      "epoch": 0.7696428571428572,
      "grad_norm": 0.21484375,
      "learning_rate": 2.3035714285714287e-06,
      "loss": 1.174,
      "step": 431
    },
    {
      "epoch": 0.7714285714285715,
      "grad_norm": 0.251953125,
      "learning_rate": 2.285714285714286e-06,
      "loss": 1.1263,
      "step": 432
    },
    {
      "epoch": 0.7732142857142857,
      "grad_norm": 0.2109375,
      "learning_rate": 2.2678571428571432e-06,
      "loss": 1.132,
      "step": 433
    },
    {
      "epoch": 0.775,
      "grad_norm": 0.2470703125,
      "learning_rate": 2.25e-06,
      "loss": 1.1163,
      "step": 434
    },
    {
      "epoch": 0.7767857142857143,
      "grad_norm": 0.189453125,
      "learning_rate": 2.2321428571428573e-06,
      "loss": 1.1375,
      "step": 435
    },
    {
      "epoch": 0.7785714285714286,
      "grad_norm": 0.19921875,
      "learning_rate": 2.2142857142857146e-06,
      "loss": 1.1587,
      "step": 436
    },
    {
      "epoch": 0.7803571428571429,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.1964285714285714e-06,
      "loss": 1.1021,
      "step": 437
    },
    {
      "epoch": 0.7821428571428571,
      "grad_norm": 0.193359375,
      "learning_rate": 2.1785714285714286e-06,
      "loss": 1.0877,
      "step": 438
    },
    {
      "epoch": 0.7839285714285714,
      "grad_norm": 0.28515625,
      "learning_rate": 2.160714285714286e-06,
      "loss": 1.1146,
      "step": 439
    },
    {
      "epoch": 0.7857142857142857,
      "grad_norm": 0.193359375,
      "learning_rate": 2.1428571428571427e-06,
      "loss": 1.1518,
      "step": 440
    },
    {
      "epoch": 0.7875,
      "grad_norm": 0.21875,
      "learning_rate": 2.125e-06,
      "loss": 1.1839,
      "step": 441
    },
    {
      "epoch": 0.7892857142857143,
      "grad_norm": 0.193359375,
      "learning_rate": 2.1071428571428572e-06,
      "loss": 1.1355,
      "step": 442
    },
    {
      "epoch": 0.7910714285714285,
      "grad_norm": 0.1943359375,
      "learning_rate": 2.0892857142857145e-06,
      "loss": 1.0875,
      "step": 443
    },
    {
      "epoch": 0.7928571428571428,
      "grad_norm": 0.197265625,
      "learning_rate": 2.0714285714285717e-06,
      "loss": 1.172,
      "step": 444
    },
    {
      "epoch": 0.7946428571428571,
      "grad_norm": 0.19921875,
      "learning_rate": 2.0535714285714286e-06,
      "loss": 1.1502,
      "step": 445
    },
    {
      "epoch": 0.7964285714285714,
      "grad_norm": 0.2392578125,
      "learning_rate": 2.035714285714286e-06,
      "loss": 1.1237,
      "step": 446
    },
    {
      "epoch": 0.7982142857142858,
      "grad_norm": 0.2021484375,
      "learning_rate": 2.017857142857143e-06,
      "loss": 1.111,
      "step": 447
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 1.1212,
      "step": 448
    },
    {
      "epoch": 0.8017857142857143,
      "grad_norm": 0.2060546875,
      "learning_rate": 1.982142857142857e-06,
      "loss": 1.1525,
      "step": 449
    },
    {
      "epoch": 0.8035714285714286,
      "grad_norm": 0.1865234375,
      "learning_rate": 1.9642857142857144e-06,
      "loss": 1.0929,
      "step": 450
    },
    {
      "epoch": 0.8053571428571429,
      "grad_norm": 0.197265625,
      "learning_rate": 1.9464285714285717e-06,
      "loss": 1.1021,
      "step": 451
    },
    {
      "epoch": 0.8071428571428572,
      "grad_norm": 0.185546875,
      "learning_rate": 1.928571428571429e-06,
      "loss": 1.0899,
      "step": 452
    },
    {
      "epoch": 0.8089285714285714,
      "grad_norm": 0.19921875,
      "learning_rate": 1.9107142857142858e-06,
      "loss": 1.153,
      "step": 453
    },
    {
      "epoch": 0.8107142857142857,
      "grad_norm": 0.263671875,
      "learning_rate": 1.892857142857143e-06,
      "loss": 1.157,
      "step": 454
    },
    {
      "epoch": 0.8125,
      "grad_norm": 0.1865234375,
      "learning_rate": 1.8750000000000003e-06,
      "loss": 1.1144,
      "step": 455
    },
    {
      "epoch": 0.8142857142857143,
      "grad_norm": 0.1923828125,
      "learning_rate": 1.8571428571428573e-06,
      "loss": 1.155,
      "step": 456
    },
    {
      "epoch": 0.8160714285714286,
      "grad_norm": 0.193359375,
      "learning_rate": 1.8392857142857146e-06,
      "loss": 1.1415,
      "step": 457
    },
    {
      "epoch": 0.8178571428571428,
      "grad_norm": 0.1953125,
      "learning_rate": 1.8214285714285716e-06,
      "loss": 1.1344,
      "step": 458
    },
    {
      "epoch": 0.8196428571428571,
      "grad_norm": 0.220703125,
      "learning_rate": 1.8035714285714286e-06,
      "loss": 1.0889,
      "step": 459
    },
    {
      "epoch": 0.8214285714285714,
      "grad_norm": 0.1982421875,
      "learning_rate": 1.7857142857142859e-06,
      "loss": 1.1418,
      "step": 460
    },
    {
      "epoch": 0.8232142857142857,
      "grad_norm": 0.189453125,
      "learning_rate": 1.7678571428571431e-06,
      "loss": 1.1216,
      "step": 461
    },
    {
      "epoch": 0.825,
      "grad_norm": 0.201171875,
      "learning_rate": 1.75e-06,
      "loss": 1.1632,
      "step": 462
    },
    {
      "epoch": 0.8267857142857142,
      "grad_norm": 0.2255859375,
      "learning_rate": 1.7321428571428572e-06,
      "loss": 1.106,
      "step": 463
    },
    {
      "epoch": 0.8285714285714286,
      "grad_norm": 0.22265625,
      "learning_rate": 1.7142857142857145e-06,
      "loss": 1.152,
      "step": 464
    },
    {
      "epoch": 0.8303571428571429,
      "grad_norm": 0.1875,
      "learning_rate": 1.6964285714285717e-06,
      "loss": 1.1252,
      "step": 465
    },
    {
      "epoch": 0.8321428571428572,
      "grad_norm": 0.1904296875,
      "learning_rate": 1.6785714285714286e-06,
      "loss": 1.0623,
      "step": 466
    },
    {
      "epoch": 0.8339285714285715,
      "grad_norm": 0.1923828125,
      "learning_rate": 1.6607142857142858e-06,
      "loss": 1.1411,
      "step": 467
    },
    {
      "epoch": 0.8357142857142857,
      "grad_norm": 0.20703125,
      "learning_rate": 1.642857142857143e-06,
      "loss": 1.186,
      "step": 468
    },
    {
      "epoch": 0.8375,
      "grad_norm": 0.1982421875,
      "learning_rate": 1.6250000000000001e-06,
      "loss": 1.1114,
      "step": 469
    },
    {
      "epoch": 0.8392857142857143,
      "grad_norm": 0.197265625,
      "learning_rate": 1.6071428571428574e-06,
      "loss": 1.1707,
      "step": 470
    },
    {
      "epoch": 0.8410714285714286,
      "grad_norm": 0.1982421875,
      "learning_rate": 1.5892857142857144e-06,
      "loss": 1.1489,
      "step": 471
    },
    {
      "epoch": 0.8428571428571429,
      "grad_norm": 0.2314453125,
      "learning_rate": 1.5714285714285714e-06,
      "loss": 1.1259,
      "step": 472
    },
    {
      "epoch": 0.8446428571428571,
      "grad_norm": 0.1904296875,
      "learning_rate": 1.5535714285714287e-06,
      "loss": 1.1018,
      "step": 473
    },
    {
      "epoch": 0.8464285714285714,
      "grad_norm": 0.193359375,
      "learning_rate": 1.535714285714286e-06,
      "loss": 1.0899,
      "step": 474
    },
    {
      "epoch": 0.8482142857142857,
      "grad_norm": 0.271484375,
      "learning_rate": 1.5178571428571428e-06,
      "loss": 1.1349,
      "step": 475
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.203125,
      "learning_rate": 1.5e-06,
      "loss": 1.1787,
      "step": 476
    },
    {
      "epoch": 0.8517857142857143,
      "grad_norm": 0.216796875,
      "learning_rate": 1.4821428571428573e-06,
      "loss": 1.1398,
      "step": 477
    },
    {
      "epoch": 0.8535714285714285,
      "grad_norm": 0.205078125,
      "learning_rate": 1.4642857142857145e-06,
      "loss": 1.1429,
      "step": 478
    },
    {
      "epoch": 0.8553571428571428,
      "grad_norm": 0.2412109375,
      "learning_rate": 1.4464285714285716e-06,
      "loss": 1.1161,
      "step": 479
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 0.193359375,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 1.1527,
      "step": 480
    },
    {
      "epoch": 0.8589285714285714,
      "grad_norm": 0.1923828125,
      "learning_rate": 1.4107142857142859e-06,
      "loss": 1.1439,
      "step": 481
    },
    {
      "epoch": 0.8607142857142858,
      "grad_norm": 0.1884765625,
      "learning_rate": 1.392857142857143e-06,
      "loss": 1.1271,
      "step": 482
    },
    {
      "epoch": 0.8625,
      "grad_norm": 0.337890625,
      "learning_rate": 1.3750000000000002e-06,
      "loss": 1.1871,
      "step": 483
    },
    {
      "epoch": 0.8642857142857143,
      "grad_norm": 0.265625,
      "learning_rate": 1.3571428571428572e-06,
      "loss": 1.0913,
      "step": 484
    },
    {
      "epoch": 0.8660714285714286,
      "grad_norm": 0.2470703125,
      "learning_rate": 1.3392857142857143e-06,
      "loss": 1.1074,
      "step": 485
    },
    {
      "epoch": 0.8678571428571429,
      "grad_norm": 0.185546875,
      "learning_rate": 1.3214285714285715e-06,
      "loss": 1.1388,
      "step": 486
    },
    {
      "epoch": 0.8696428571428572,
      "grad_norm": 0.2470703125,
      "learning_rate": 1.3035714285714288e-06,
      "loss": 1.0727,
      "step": 487
    },
    {
      "epoch": 0.8714285714285714,
      "grad_norm": 0.19140625,
      "learning_rate": 1.2857142857142856e-06,
      "loss": 1.1419,
      "step": 488
    },
    {
      "epoch": 0.8732142857142857,
      "grad_norm": 0.1865234375,
      "learning_rate": 1.2678571428571428e-06,
      "loss": 1.0769,
      "step": 489
    },
    {
      "epoch": 0.875,
      "grad_norm": 0.2001953125,
      "learning_rate": 1.25e-06,
      "loss": 1.1293,
      "step": 490
    },
    {
      "epoch": 0.8767857142857143,
      "grad_norm": 0.216796875,
      "learning_rate": 1.2321428571428574e-06,
      "loss": 1.1096,
      "step": 491
    },
    {
      "epoch": 0.8785714285714286,
      "grad_norm": 0.294921875,
      "learning_rate": 1.2142857142857144e-06,
      "loss": 1.1627,
      "step": 492
    },
    {
      "epoch": 0.8803571428571428,
      "grad_norm": 0.1904296875,
      "learning_rate": 1.1964285714285717e-06,
      "loss": 1.1395,
      "step": 493
    },
    {
      "epoch": 0.8821428571428571,
      "grad_norm": 0.2060546875,
      "learning_rate": 1.1785714285714287e-06,
      "loss": 1.1657,
      "step": 494
    },
    {
      "epoch": 0.8839285714285714,
      "grad_norm": 0.1923828125,
      "learning_rate": 1.160714285714286e-06,
      "loss": 1.131,
      "step": 495
    },
    {
      "epoch": 0.8857142857142857,
      "grad_norm": 0.2021484375,
      "learning_rate": 1.142857142857143e-06,
      "loss": 1.1686,
      "step": 496
    },
    {
      "epoch": 0.8875,
      "grad_norm": 0.1923828125,
      "learning_rate": 1.125e-06,
      "loss": 1.1463,
      "step": 497
    },
    {
      "epoch": 0.8892857142857142,
      "grad_norm": 0.203125,
      "learning_rate": 1.1071428571428573e-06,
      "loss": 1.2182,
      "step": 498
    },
    {
      "epoch": 0.8910714285714286,
      "grad_norm": 0.1962890625,
      "learning_rate": 1.0892857142857143e-06,
      "loss": 1.1011,
      "step": 499
    },
    {
      "epoch": 0.8928571428571429,
      "grad_norm": 0.2021484375,
      "learning_rate": 1.0714285714285714e-06,
      "loss": 1.1859,
      "step": 500
    },
    {
      "epoch": 0.8946428571428572,
      "grad_norm": 0.2353515625,
      "learning_rate": 1.0535714285714286e-06,
      "loss": 1.1523,
      "step": 501
    },
    {
      "epoch": 0.8964285714285715,
      "grad_norm": 0.2021484375,
      "learning_rate": 1.0357142857142859e-06,
      "loss": 1.1122,
      "step": 502
    },
    {
      "epoch": 0.8982142857142857,
      "grad_norm": 0.19921875,
      "learning_rate": 1.017857142857143e-06,
      "loss": 1.1763,
      "step": 503
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.28125,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 1.0727,
      "step": 504
    },
    {
      "epoch": 0.9017857142857143,
      "grad_norm": 0.1923828125,
      "learning_rate": 9.821428571428572e-07,
      "loss": 1.1076,
      "step": 505
    },
    {
      "epoch": 0.9035714285714286,
      "grad_norm": 0.197265625,
      "learning_rate": 9.642857142857145e-07,
      "loss": 1.1587,
      "step": 506
    },
    {
      "epoch": 0.9053571428571429,
      "grad_norm": 0.2890625,
      "learning_rate": 9.464285714285715e-07,
      "loss": 1.1549,
      "step": 507
    },
    {
      "epoch": 0.9071428571428571,
      "grad_norm": 0.19140625,
      "learning_rate": 9.285714285714287e-07,
      "loss": 1.1607,
      "step": 508
    },
    {
      "epoch": 0.9089285714285714,
      "grad_norm": 0.19140625,
      "learning_rate": 9.107142857142858e-07,
      "loss": 1.1395,
      "step": 509
    },
    {
      "epoch": 0.9107142857142857,
      "grad_norm": 0.1904296875,
      "learning_rate": 8.928571428571429e-07,
      "loss": 1.1434,
      "step": 510
    },
    {
      "epoch": 0.9125,
      "grad_norm": 0.208984375,
      "learning_rate": 8.75e-07,
      "loss": 1.0969,
      "step": 511
    },
    {
      "epoch": 0.9142857142857143,
      "grad_norm": 0.1953125,
      "learning_rate": 8.571428571428572e-07,
      "loss": 1.1581,
      "step": 512
    },
    {
      "epoch": 0.9160714285714285,
      "grad_norm": 0.2001953125,
      "learning_rate": 8.392857142857143e-07,
      "loss": 1.1507,
      "step": 513
    },
    {
      "epoch": 0.9178571428571428,
      "grad_norm": 0.216796875,
      "learning_rate": 8.214285714285715e-07,
      "loss": 1.1276,
      "step": 514
    },
    {
      "epoch": 0.9196428571428571,
      "grad_norm": 0.2373046875,
      "learning_rate": 8.035714285714287e-07,
      "loss": 1.1668,
      "step": 515
    },
    {
      "epoch": 0.9214285714285714,
      "grad_norm": 0.1962890625,
      "learning_rate": 7.857142857142857e-07,
      "loss": 1.1198,
      "step": 516
    },
    {
      "epoch": 0.9232142857142858,
      "grad_norm": 0.2138671875,
      "learning_rate": 7.67857142857143e-07,
      "loss": 1.2005,
      "step": 517
    },
    {
      "epoch": 0.925,
      "grad_norm": 0.294921875,
      "learning_rate": 7.5e-07,
      "loss": 1.1021,
      "step": 518
    },
    {
      "epoch": 0.9267857142857143,
      "grad_norm": 0.1904296875,
      "learning_rate": 7.321428571428573e-07,
      "loss": 1.1009,
      "step": 519
    },
    {
      "epoch": 0.9285714285714286,
      "grad_norm": 0.1962890625,
      "learning_rate": 7.142857142857143e-07,
      "loss": 1.1358,
      "step": 520
    },
    {
      "epoch": 0.9303571428571429,
      "grad_norm": 0.193359375,
      "learning_rate": 6.964285714285715e-07,
      "loss": 1.1524,
      "step": 521
    },
    {
      "epoch": 0.9321428571428572,
      "grad_norm": 0.2060546875,
      "learning_rate": 6.785714285714286e-07,
      "loss": 1.0907,
      "step": 522
    },
    {
      "epoch": 0.9339285714285714,
      "grad_norm": 0.1875,
      "learning_rate": 6.607142857142858e-07,
      "loss": 1.1222,
      "step": 523
    },
    {
      "epoch": 0.9357142857142857,
      "grad_norm": 0.1962890625,
      "learning_rate": 6.428571428571428e-07,
      "loss": 1.1787,
      "step": 524
    },
    {
      "epoch": 0.9375,
      "grad_norm": 0.19140625,
      "learning_rate": 6.25e-07,
      "loss": 1.1474,
      "step": 525
    },
    {
      "epoch": 0.9392857142857143,
      "grad_norm": 0.1953125,
      "learning_rate": 6.071428571428572e-07,
      "loss": 1.1283,
      "step": 526
    },
    {
      "epoch": 0.9410714285714286,
      "grad_norm": 0.1875,
      "learning_rate": 5.892857142857143e-07,
      "loss": 1.1177,
      "step": 527
    },
    {
      "epoch": 0.9428571428571428,
      "grad_norm": 0.2373046875,
      "learning_rate": 5.714285714285715e-07,
      "loss": 1.119,
      "step": 528
    },
    {
      "epoch": 0.9446428571428571,
      "grad_norm": 0.2001953125,
      "learning_rate": 5.535714285714286e-07,
      "loss": 1.1682,
      "step": 529
    },
    {
      "epoch": 0.9464285714285714,
      "grad_norm": 0.1962890625,
      "learning_rate": 5.357142857142857e-07,
      "loss": 1.1034,
      "step": 530
    },
    {
      "epoch": 0.9482142857142857,
      "grad_norm": 0.193359375,
      "learning_rate": 5.178571428571429e-07,
      "loss": 1.1415,
      "step": 531
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2255859375,
      "learning_rate": 5.000000000000001e-07,
      "loss": 1.151,
      "step": 532
    },
    {
      "epoch": 0.9517857142857142,
      "grad_norm": 0.220703125,
      "learning_rate": 4.821428571428572e-07,
      "loss": 1.133,
      "step": 533
    },
    {
      "epoch": 0.9535714285714286,
      "grad_norm": 0.1904296875,
      "learning_rate": 4.642857142857143e-07,
      "loss": 1.1081,
      "step": 534
    },
    {
      "epoch": 0.9553571428571429,
      "grad_norm": 0.22265625,
      "learning_rate": 4.4642857142857147e-07,
      "loss": 1.0945,
      "step": 535
    },
    {
      "epoch": 0.9571428571428572,
      "grad_norm": 0.1884765625,
      "learning_rate": 4.285714285714286e-07,
      "loss": 1.1079,
      "step": 536
    },
    {
      "epoch": 0.9589285714285715,
      "grad_norm": 0.18359375,
      "learning_rate": 4.1071428571428577e-07,
      "loss": 1.1026,
      "step": 537
    },
    {
      "epoch": 0.9607142857142857,
      "grad_norm": 0.1923828125,
      "learning_rate": 3.9285714285714286e-07,
      "loss": 1.1647,
      "step": 538
    },
    {
      "epoch": 0.9625,
      "grad_norm": 0.23046875,
      "learning_rate": 3.75e-07,
      "loss": 1.179,
      "step": 539
    },
    {
      "epoch": 0.9642857142857143,
      "grad_norm": 0.2333984375,
      "learning_rate": 3.5714285714285716e-07,
      "loss": 1.1877,
      "step": 540
    },
    {
      "epoch": 0.9660714285714286,
      "grad_norm": 0.1982421875,
      "learning_rate": 3.392857142857143e-07,
      "loss": 1.102,
      "step": 541
    },
    {
      "epoch": 0.9678571428571429,
      "grad_norm": 0.1884765625,
      "learning_rate": 3.214285714285714e-07,
      "loss": 1.1424,
      "step": 542
    },
    {
      "epoch": 0.9696428571428571,
      "grad_norm": 0.1865234375,
      "learning_rate": 3.035714285714286e-07,
      "loss": 1.1172,
      "step": 543
    },
    {
      "epoch": 0.9714285714285714,
      "grad_norm": 0.1982421875,
      "learning_rate": 2.8571428571428575e-07,
      "loss": 1.1402,
      "step": 544
    },
    {
      "epoch": 0.9732142857142857,
      "grad_norm": 0.2119140625,
      "learning_rate": 2.6785714285714284e-07,
      "loss": 1.1771,
      "step": 545
    },
    {
      "epoch": 0.975,
      "grad_norm": 0.220703125,
      "learning_rate": 2.5000000000000004e-07,
      "loss": 1.1276,
      "step": 546
    },
    {
      "epoch": 0.9767857142857143,
      "grad_norm": 0.1875,
      "learning_rate": 2.3214285714285716e-07,
      "loss": 1.143,
      "step": 547
    },
    {
      "epoch": 0.9785714285714285,
      "grad_norm": 0.197265625,
      "learning_rate": 2.142857142857143e-07,
      "loss": 1.1141,
      "step": 548
    },
    {
      "epoch": 0.9803571428571428,
      "grad_norm": 0.193359375,
      "learning_rate": 1.9642857142857143e-07,
      "loss": 1.1463,
      "step": 549
    },
    {
      "epoch": 0.9821428571428571,
      "grad_norm": 0.1943359375,
      "learning_rate": 1.7857142857142858e-07,
      "loss": 1.1643,
      "step": 550
    },
    {
      "epoch": 0.9839285714285714,
      "grad_norm": 0.1904296875,
      "learning_rate": 1.607142857142857e-07,
      "loss": 1.1075,
      "step": 551
    },
    {
      "epoch": 0.9857142857142858,
      "grad_norm": 0.20703125,
      "learning_rate": 1.4285714285714287e-07,
      "loss": 1.1964,
      "step": 552
    },
    {
      "epoch": 0.9875,
      "grad_norm": 0.1826171875,
      "learning_rate": 1.2500000000000002e-07,
      "loss": 1.0845,
      "step": 553
    },
    {
      "epoch": 0.9892857142857143,
      "grad_norm": 0.1953125,
      "learning_rate": 1.0714285714285716e-07,
      "loss": 1.1831,
      "step": 554
    },
    {
      "epoch": 0.9910714285714286,
      "grad_norm": 0.201171875,
      "learning_rate": 8.928571428571429e-08,
      "loss": 1.1154,
      "step": 555
    },
    {
      "epoch": 0.9928571428571429,
      "grad_norm": 0.1953125,
      "learning_rate": 7.142857142857144e-08,
      "loss": 1.1277,
      "step": 556
    },
    {
      "epoch": 0.9946428571428572,
      "grad_norm": 0.1943359375,
      "learning_rate": 5.357142857142858e-08,
      "loss": 1.0752,
      "step": 557
    },
    {
      "epoch": 0.9964285714285714,
      "grad_norm": 0.2099609375,
      "learning_rate": 3.571428571428572e-08,
      "loss": 1.18,
      "step": 558
    },
    {
      "epoch": 0.9982142857142857,
      "grad_norm": 0.2138671875,
      "learning_rate": 1.785714285714286e-08,
      "loss": 1.1936,
      "step": 559
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0,
      "loss": 1.1262,
      "step": 560
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 560,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 0,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.796277576746926e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}