{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.4193548387096775,
  "eval_steps": 550,
  "global_step": 2200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0006451612903225806,
      "grad_norm": 1.8828125,
      "learning_rate": 8.547008547008549e-08,
      "loss": 1.3333,
      "step": 1
    },
    {
      "epoch": 0.0012903225806451613,
      "grad_norm": 2.40625,
      "learning_rate": 1.7094017094017097e-07,
      "loss": 1.557,
      "step": 2
    },
    {
      "epoch": 0.001935483870967742,
      "grad_norm": 2.5625,
      "learning_rate": 2.564102564102564e-07,
      "loss": 1.5582,
      "step": 3
    },
    {
      "epoch": 0.0025806451612903226,
      "grad_norm": 1.8046875,
      "learning_rate": 3.4188034188034194e-07,
      "loss": 1.3006,
      "step": 4
    },
    {
      "epoch": 0.0032258064516129032,
      "grad_norm": 3.15625,
      "learning_rate": 4.273504273504274e-07,
      "loss": 1.2333,
      "step": 5
    },
    {
      "epoch": 0.003870967741935484,
      "grad_norm": 2.484375,
      "learning_rate": 5.128205128205128e-07,
      "loss": 1.3196,
      "step": 6
    },
    {
      "epoch": 0.004516129032258065,
      "grad_norm": 5.53125,
      "learning_rate": 5.982905982905984e-07,
      "loss": 1.813,
      "step": 7
    },
    {
      "epoch": 0.005161290322580645,
      "grad_norm": 1.1796875,
      "learning_rate": 6.837606837606839e-07,
      "loss": 0.8668,
      "step": 8
    },
    {
      "epoch": 0.005806451612903226,
      "grad_norm": 3.3125,
      "learning_rate": 7.692307692307694e-07,
      "loss": 1.738,
      "step": 9
    },
    {
      "epoch": 0.0064516129032258064,
      "grad_norm": 1.6328125,
      "learning_rate": 8.547008547008548e-07,
      "loss": 1.1196,
      "step": 10
    },
    {
      "epoch": 0.0070967741935483875,
      "grad_norm": 2.609375,
      "learning_rate": 9.401709401709402e-07,
      "loss": 1.4503,
      "step": 11
    },
    {
      "epoch": 0.007741935483870968,
      "grad_norm": 2.140625,
      "learning_rate": 1.0256410256410257e-06,
      "loss": 1.4077,
      "step": 12
    },
    {
      "epoch": 0.008387096774193548,
      "grad_norm": 1.2265625,
      "learning_rate": 1.111111111111111e-06,
      "loss": 0.9903,
      "step": 13
    },
    {
      "epoch": 0.00903225806451613,
      "grad_norm": 1.1875,
      "learning_rate": 1.1965811965811968e-06,
      "loss": 0.8388,
      "step": 14
    },
    {
      "epoch": 0.00967741935483871,
      "grad_norm": 1.6875,
      "learning_rate": 1.282051282051282e-06,
      "loss": 1.2088,
      "step": 15
    },
    {
      "epoch": 0.01032258064516129,
      "grad_norm": 1.4921875,
      "learning_rate": 1.3675213675213678e-06,
      "loss": 1.1756,
      "step": 16
    },
    {
      "epoch": 0.01096774193548387,
      "grad_norm": 1.7421875,
      "learning_rate": 1.4529914529914531e-06,
      "loss": 1.1176,
      "step": 17
    },
    {
      "epoch": 0.011612903225806452,
      "grad_norm": 2.28125,
      "learning_rate": 1.5384615384615387e-06,
      "loss": 1.2577,
      "step": 18
    },
    {
      "epoch": 0.012258064516129033,
      "grad_norm": 3.265625,
      "learning_rate": 1.623931623931624e-06,
      "loss": 1.715,
      "step": 19
    },
    {
      "epoch": 0.012903225806451613,
      "grad_norm": 1.4140625,
      "learning_rate": 1.7094017094017097e-06,
      "loss": 0.9563,
      "step": 20
    },
    {
      "epoch": 0.013548387096774193,
      "grad_norm": 1.4296875,
      "learning_rate": 1.794871794871795e-06,
      "loss": 1.1738,
      "step": 21
    },
    {
      "epoch": 0.014193548387096775,
      "grad_norm": 2.0,
      "learning_rate": 1.8803418803418804e-06,
      "loss": 1.5235,
      "step": 22
    },
    {
      "epoch": 0.014838709677419355,
      "grad_norm": 1.9765625,
      "learning_rate": 1.9658119658119658e-06,
      "loss": 1.3773,
      "step": 23
    },
    {
      "epoch": 0.015483870967741935,
      "grad_norm": 2.0,
      "learning_rate": 2.0512820512820513e-06,
      "loss": 1.2297,
      "step": 24
    },
    {
      "epoch": 0.016129032258064516,
      "grad_norm": 3.984375,
      "learning_rate": 2.136752136752137e-06,
      "loss": 1.6168,
      "step": 25
    },
    {
      "epoch": 0.016774193548387096,
      "grad_norm": 2.390625,
      "learning_rate": 2.222222222222222e-06,
      "loss": 1.2648,
      "step": 26
    },
    {
      "epoch": 0.017419354838709676,
      "grad_norm": 2.171875,
      "learning_rate": 2.307692307692308e-06,
      "loss": 1.6374,
      "step": 27
    },
    {
      "epoch": 0.01806451612903226,
      "grad_norm": 2.53125,
      "learning_rate": 2.3931623931623937e-06,
      "loss": 1.6925,
      "step": 28
    },
    {
      "epoch": 0.01870967741935484,
      "grad_norm": 3.21875,
      "learning_rate": 2.478632478632479e-06,
      "loss": 1.5046,
      "step": 29
    },
    {
      "epoch": 0.01935483870967742,
      "grad_norm": 2.109375,
      "learning_rate": 2.564102564102564e-06,
      "loss": 1.2482,
      "step": 30
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.796875,
      "learning_rate": 2.64957264957265e-06,
      "loss": 1.5405,
      "step": 31
    },
    {
      "epoch": 0.02064516129032258,
      "grad_norm": 1.8828125,
      "learning_rate": 2.7350427350427355e-06,
      "loss": 1.1863,
      "step": 32
    },
    {
      "epoch": 0.02129032258064516,
      "grad_norm": 2.171875,
      "learning_rate": 2.8205128205128207e-06,
      "loss": 1.2516,
      "step": 33
    },
    {
      "epoch": 0.02193548387096774,
      "grad_norm": 1.7265625,
      "learning_rate": 2.9059829059829063e-06,
      "loss": 1.273,
      "step": 34
    },
    {
      "epoch": 0.02258064516129032,
      "grad_norm": 3.6875,
      "learning_rate": 2.9914529914529914e-06,
      "loss": 0.9979,
      "step": 35
    },
    {
      "epoch": 0.023225806451612905,
      "grad_norm": 2.296875,
      "learning_rate": 3.0769230769230774e-06,
      "loss": 1.4706,
      "step": 36
    },
    {
      "epoch": 0.023870967741935485,
      "grad_norm": 2.109375,
      "learning_rate": 3.1623931623931626e-06,
      "loss": 1.0358,
      "step": 37
    },
    {
      "epoch": 0.024516129032258065,
      "grad_norm": 2.53125,
      "learning_rate": 3.247863247863248e-06,
      "loss": 1.5697,
      "step": 38
    },
    {
      "epoch": 0.025161290322580646,
      "grad_norm": 1.7421875,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.9868,
      "step": 39
    },
    {
      "epoch": 0.025806451612903226,
      "grad_norm": 3.34375,
      "learning_rate": 3.4188034188034193e-06,
      "loss": 1.5831,
      "step": 40
    },
    {
      "epoch": 0.026451612903225806,
      "grad_norm": 2.03125,
      "learning_rate": 3.5042735042735045e-06,
      "loss": 1.0302,
      "step": 41
    },
    {
      "epoch": 0.027096774193548386,
      "grad_norm": 2.90625,
      "learning_rate": 3.58974358974359e-06,
      "loss": 1.5392,
      "step": 42
    },
    {
      "epoch": 0.027741935483870966,
      "grad_norm": 2.25,
      "learning_rate": 3.6752136752136756e-06,
      "loss": 0.8571,
      "step": 43
    },
    {
      "epoch": 0.02838709677419355,
      "grad_norm": 2.5625,
      "learning_rate": 3.760683760683761e-06,
      "loss": 1.3642,
      "step": 44
    },
    {
      "epoch": 0.02903225806451613,
      "grad_norm": 1.703125,
      "learning_rate": 3.846153846153847e-06,
      "loss": 0.8388,
      "step": 45
    },
    {
      "epoch": 0.02967741935483871,
      "grad_norm": 3.328125,
      "learning_rate": 3.9316239316239315e-06,
      "loss": 1.6217,
      "step": 46
    },
    {
      "epoch": 0.03032258064516129,
      "grad_norm": 2.28125,
      "learning_rate": 4.017094017094018e-06,
      "loss": 1.1064,
      "step": 47
    },
    {
      "epoch": 0.03096774193548387,
      "grad_norm": 2.15625,
      "learning_rate": 4.102564102564103e-06,
      "loss": 1.3299,
      "step": 48
    },
    {
      "epoch": 0.031612903225806455,
      "grad_norm": 1.7265625,
      "learning_rate": 4.188034188034188e-06,
      "loss": 1.1497,
      "step": 49
    },
    {
      "epoch": 0.03225806451612903,
      "grad_norm": 1.2890625,
      "learning_rate": 4.273504273504274e-06,
      "loss": 0.7375,
      "step": 50
    },
    {
      "epoch": 0.032903225806451615,
      "grad_norm": 1.296875,
      "learning_rate": 4.358974358974359e-06,
      "loss": 1.1923,
      "step": 51
    },
    {
      "epoch": 0.03354838709677419,
      "grad_norm": 2.625,
      "learning_rate": 4.444444444444444e-06,
      "loss": 1.2889,
      "step": 52
    },
    {
      "epoch": 0.034193548387096775,
      "grad_norm": 4.90625,
      "learning_rate": 4.5299145299145306e-06,
      "loss": 1.6016,
      "step": 53
    },
    {
      "epoch": 0.03483870967741935,
      "grad_norm": 2.578125,
      "learning_rate": 4.615384615384616e-06,
      "loss": 1.4317,
      "step": 54
    },
    {
      "epoch": 0.035483870967741936,
      "grad_norm": 6.96875,
      "learning_rate": 4.700854700854701e-06,
      "loss": 1.344,
      "step": 55
    },
    {
      "epoch": 0.03612903225806452,
      "grad_norm": 1.2109375,
      "learning_rate": 4.786324786324787e-06,
      "loss": 1.1867,
      "step": 56
    },
    {
      "epoch": 0.036774193548387096,
      "grad_norm": 1.8671875,
      "learning_rate": 4.871794871794872e-06,
      "loss": 0.9824,
      "step": 57
    },
    {
      "epoch": 0.03741935483870968,
      "grad_norm": 3.3125,
      "learning_rate": 4.957264957264958e-06,
      "loss": 1.7697,
      "step": 58
    },
    {
      "epoch": 0.03806451612903226,
      "grad_norm": 2.515625,
      "learning_rate": 5.042735042735043e-06,
      "loss": 1.5357,
      "step": 59
    },
    {
      "epoch": 0.03870967741935484,
      "grad_norm": 1.8515625,
      "learning_rate": 5.128205128205128e-06,
      "loss": 1.1525,
      "step": 60
    },
    {
      "epoch": 0.03935483870967742,
      "grad_norm": 2.3125,
      "learning_rate": 5.213675213675214e-06,
      "loss": 1.5507,
      "step": 61
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.5,
      "learning_rate": 5.2991452991453e-06,
      "loss": 1.5311,
      "step": 62
    },
    {
      "epoch": 0.04064516129032258,
      "grad_norm": 1.484375,
      "learning_rate": 5.384615384615385e-06,
      "loss": 1.0941,
      "step": 63
    },
    {
      "epoch": 0.04129032258064516,
      "grad_norm": 3.515625,
      "learning_rate": 5.470085470085471e-06,
      "loss": 1.0836,
      "step": 64
    },
    {
      "epoch": 0.041935483870967745,
      "grad_norm": 2.0,
      "learning_rate": 5.555555555555557e-06,
      "loss": 1.0708,
      "step": 65
    },
    {
      "epoch": 0.04258064516129032,
      "grad_norm": 3.765625,
      "learning_rate": 5.641025641025641e-06,
      "loss": 1.6307,
      "step": 66
    },
    {
      "epoch": 0.043225806451612905,
      "grad_norm": 1.53125,
      "learning_rate": 5.726495726495727e-06,
      "loss": 1.0807,
      "step": 67
    },
    {
      "epoch": 0.04387096774193548,
      "grad_norm": 3.96875,
      "learning_rate": 5.8119658119658126e-06,
      "loss": 1.6155,
      "step": 68
    },
    {
      "epoch": 0.044516129032258066,
      "grad_norm": 2.71875,
      "learning_rate": 5.897435897435898e-06,
      "loss": 1.9351,
      "step": 69
    },
    {
      "epoch": 0.04516129032258064,
      "grad_norm": 1.21875,
      "learning_rate": 5.982905982905983e-06,
      "loss": 0.7807,
      "step": 70
    },
    {
      "epoch": 0.045806451612903226,
      "grad_norm": 1.640625,
      "learning_rate": 6.0683760683760684e-06,
      "loss": 0.9939,
      "step": 71
    },
    {
      "epoch": 0.04645161290322581,
      "grad_norm": 4.03125,
      "learning_rate": 6.153846153846155e-06,
      "loss": 1.2977,
      "step": 72
    },
    {
      "epoch": 0.04709677419354839,
      "grad_norm": 1.5703125,
      "learning_rate": 6.23931623931624e-06,
      "loss": 0.9444,
      "step": 73
    },
    {
      "epoch": 0.04774193548387097,
      "grad_norm": 3.359375,
      "learning_rate": 6.324786324786325e-06,
      "loss": 1.5163,
      "step": 74
    },
    {
      "epoch": 0.04838709677419355,
      "grad_norm": 1.578125,
      "learning_rate": 6.410256410256412e-06,
      "loss": 1.0426,
      "step": 75
    },
    {
      "epoch": 0.04903225806451613,
      "grad_norm": 2.453125,
      "learning_rate": 6.495726495726496e-06,
      "loss": 1.1125,
      "step": 76
    },
    {
      "epoch": 0.04967741935483871,
      "grad_norm": 1.3671875,
      "learning_rate": 6.581196581196582e-06,
      "loss": 0.8547,
      "step": 77
    },
    {
      "epoch": 0.05032258064516129,
      "grad_norm": 3.0625,
      "learning_rate": 6.666666666666667e-06,
      "loss": 1.6828,
      "step": 78
    },
    {
      "epoch": 0.05096774193548387,
      "grad_norm": 3.640625,
      "learning_rate": 6.752136752136753e-06,
      "loss": 1.8029,
      "step": 79
    },
    {
      "epoch": 0.05161290322580645,
      "grad_norm": 4.0,
      "learning_rate": 6.837606837606839e-06,
      "loss": 1.4521,
      "step": 80
    },
    {
      "epoch": 0.052258064516129035,
      "grad_norm": 2.5,
      "learning_rate": 6.923076923076923e-06,
      "loss": 1.383,
      "step": 81
    },
    {
      "epoch": 0.05290322580645161,
      "grad_norm": 2.421875,
      "learning_rate": 7.008547008547009e-06,
      "loss": 1.3982,
      "step": 82
    },
    {
      "epoch": 0.053548387096774196,
      "grad_norm": 2.5625,
      "learning_rate": 7.0940170940170945e-06,
      "loss": 0.9634,
      "step": 83
    },
    {
      "epoch": 0.05419354838709677,
      "grad_norm": 3.0625,
      "learning_rate": 7.17948717948718e-06,
      "loss": 1.3438,
      "step": 84
    },
    {
      "epoch": 0.054838709677419356,
      "grad_norm": 4.71875,
      "learning_rate": 7.264957264957266e-06,
      "loss": 1.8612,
      "step": 85
    },
    {
      "epoch": 0.05548387096774193,
      "grad_norm": 1.578125,
      "learning_rate": 7.350427350427351e-06,
      "loss": 0.9314,
      "step": 86
    },
    {
      "epoch": 0.056129032258064517,
      "grad_norm": 2.4375,
      "learning_rate": 7.435897435897437e-06,
      "loss": 1.3695,
      "step": 87
    },
    {
      "epoch": 0.0567741935483871,
      "grad_norm": 4.4375,
      "learning_rate": 7.521367521367522e-06,
      "loss": 1.5081,
      "step": 88
    },
    {
      "epoch": 0.05741935483870968,
      "grad_norm": 1.921875,
      "learning_rate": 7.606837606837607e-06,
      "loss": 1.6078,
      "step": 89
    },
    {
      "epoch": 0.05806451612903226,
      "grad_norm": 2.484375,
      "learning_rate": 7.692307692307694e-06,
      "loss": 1.0418,
      "step": 90
    },
    {
      "epoch": 0.05870967741935484,
      "grad_norm": 3.78125,
      "learning_rate": 7.77777777777778e-06,
      "loss": 1.3928,
      "step": 91
    },
    {
      "epoch": 0.05935483870967742,
      "grad_norm": 3.703125,
      "learning_rate": 7.863247863247863e-06,
      "loss": 1.2032,
      "step": 92
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9921875,
      "learning_rate": 7.948717948717949e-06,
      "loss": 0.8774,
      "step": 93
    },
    {
      "epoch": 0.06064516129032258,
      "grad_norm": 4.53125,
      "learning_rate": 8.034188034188036e-06,
      "loss": 1.5385,
      "step": 94
    },
    {
      "epoch": 0.06129032258064516,
      "grad_norm": 2.96875,
      "learning_rate": 8.11965811965812e-06,
      "loss": 0.7813,
      "step": 95
    },
    {
      "epoch": 0.06193548387096774,
      "grad_norm": 2.796875,
      "learning_rate": 8.205128205128205e-06,
      "loss": 1.1728,
      "step": 96
    },
    {
      "epoch": 0.06258064516129032,
      "grad_norm": 1.65625,
      "learning_rate": 8.290598290598293e-06,
      "loss": 1.0294,
      "step": 97
    },
    {
      "epoch": 0.06322580645161291,
      "grad_norm": 3.0625,
      "learning_rate": 8.376068376068377e-06,
      "loss": 1.3683,
      "step": 98
    },
    {
      "epoch": 0.06387096774193549,
      "grad_norm": 2.03125,
      "learning_rate": 8.461538461538462e-06,
      "loss": 0.9245,
      "step": 99
    },
    {
      "epoch": 0.06451612903225806,
      "grad_norm": 2.09375,
      "learning_rate": 8.547008547008548e-06,
      "loss": 1.4071,
      "step": 100
    },
    {
      "epoch": 0.06516129032258064,
      "grad_norm": 1.3984375,
      "learning_rate": 8.632478632478633e-06,
      "loss": 1.1323,
      "step": 101
    },
    {
      "epoch": 0.06580645161290323,
      "grad_norm": 3.734375,
      "learning_rate": 8.717948717948719e-06,
      "loss": 1.3161,
      "step": 102
    },
    {
      "epoch": 0.0664516129032258,
      "grad_norm": 2.734375,
      "learning_rate": 8.803418803418804e-06,
      "loss": 1.5418,
      "step": 103
    },
    {
      "epoch": 0.06709677419354838,
      "grad_norm": 2.359375,
      "learning_rate": 8.888888888888888e-06,
      "loss": 1.1518,
      "step": 104
    },
    {
      "epoch": 0.06774193548387097,
      "grad_norm": 1.3828125,
      "learning_rate": 8.974358974358976e-06,
      "loss": 0.8066,
      "step": 105
    },
    {
      "epoch": 0.06838709677419355,
      "grad_norm": 3.5625,
      "learning_rate": 9.059829059829061e-06,
      "loss": 1.3204,
      "step": 106
    },
    {
      "epoch": 0.06903225806451613,
      "grad_norm": 2.078125,
      "learning_rate": 9.145299145299145e-06,
      "loss": 1.0751,
      "step": 107
    },
    {
      "epoch": 0.0696774193548387,
      "grad_norm": 1.5703125,
      "learning_rate": 9.230769230769232e-06,
      "loss": 1.0113,
      "step": 108
    },
    {
      "epoch": 0.0703225806451613,
      "grad_norm": 2.875,
      "learning_rate": 9.316239316239318e-06,
      "loss": 1.3802,
      "step": 109
    },
    {
      "epoch": 0.07096774193548387,
      "grad_norm": 1.9609375,
      "learning_rate": 9.401709401709402e-06,
      "loss": 0.9584,
      "step": 110
    },
    {
      "epoch": 0.07161290322580645,
      "grad_norm": 3.390625,
      "learning_rate": 9.487179487179487e-06,
      "loss": 1.276,
      "step": 111
    },
    {
      "epoch": 0.07225806451612904,
      "grad_norm": 3.515625,
      "learning_rate": 9.572649572649575e-06,
      "loss": 1.7236,
      "step": 112
    },
    {
      "epoch": 0.07290322580645162,
      "grad_norm": 2.78125,
      "learning_rate": 9.658119658119659e-06,
      "loss": 1.4117,
      "step": 113
    },
    {
      "epoch": 0.07354838709677419,
      "grad_norm": 1.71875,
      "learning_rate": 9.743589743589744e-06,
      "loss": 1.084,
      "step": 114
    },
    {
      "epoch": 0.07419354838709677,
      "grad_norm": 1.859375,
      "learning_rate": 9.82905982905983e-06,
      "loss": 1.1163,
      "step": 115
    },
    {
      "epoch": 0.07483870967741936,
      "grad_norm": 1.71875,
      "learning_rate": 9.914529914529915e-06,
      "loss": 0.9619,
      "step": 116
    },
    {
      "epoch": 0.07548387096774194,
      "grad_norm": 3.953125,
      "learning_rate": 1e-05,
      "loss": 1.5004,
      "step": 117
    },
    {
      "epoch": 0.07612903225806451,
      "grad_norm": 2.234375,
      "learning_rate": 9.999998252866154e-06,
      "loss": 1.1359,
      "step": 118
    },
    {
      "epoch": 0.07677419354838709,
      "grad_norm": 1.53125,
      "learning_rate": 9.999993011465834e-06,
      "loss": 1.1179,
      "step": 119
    },
    {
      "epoch": 0.07741935483870968,
      "grad_norm": 1.046875,
      "learning_rate": 9.999984275802708e-06,
      "loss": 0.9188,
      "step": 120
    },
    {
      "epoch": 0.07806451612903226,
      "grad_norm": 1.296875,
      "learning_rate": 9.999972045882874e-06,
      "loss": 1.0007,
      "step": 121
    },
    {
      "epoch": 0.07870967741935483,
      "grad_norm": 1.0703125,
      "learning_rate": 9.999956321714884e-06,
      "loss": 0.642,
      "step": 122
    },
    {
      "epoch": 0.07935483870967742,
      "grad_norm": 1.53125,
      "learning_rate": 9.999937103309725e-06,
      "loss": 1.0673,
      "step": 123
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8359375,
      "learning_rate": 9.999914390680829e-06,
      "loss": 1.0003,
      "step": 124
    },
    {
      "epoch": 0.08064516129032258,
      "grad_norm": 1.6640625,
      "learning_rate": 9.999888183844069e-06,
      "loss": 1.1015,
      "step": 125
    },
    {
      "epoch": 0.08129032258064516,
      "grad_norm": 0.90234375,
      "learning_rate": 9.999858482817758e-06,
      "loss": 0.8178,
      "step": 126
    },
    {
      "epoch": 0.08193548387096775,
      "grad_norm": 0.9375,
      "learning_rate": 9.999825287622652e-06,
      "loss": 0.848,
      "step": 127
    },
    {
      "epoch": 0.08258064516129032,
      "grad_norm": 1.6171875,
      "learning_rate": 9.999788598281952e-06,
      "loss": 1.1395,
      "step": 128
    },
    {
      "epoch": 0.0832258064516129,
      "grad_norm": 1.2890625,
      "learning_rate": 9.999748414821297e-06,
      "loss": 0.92,
      "step": 129
    },
    {
      "epoch": 0.08387096774193549,
      "grad_norm": 2.171875,
      "learning_rate": 9.999704737268771e-06,
      "loss": 1.2977,
      "step": 130
    },
    {
      "epoch": 0.08451612903225807,
      "grad_norm": 1.2890625,
      "learning_rate": 9.999657565654897e-06,
      "loss": 1.0306,
      "step": 131
    },
    {
      "epoch": 0.08516129032258064,
      "grad_norm": 1.6171875,
      "learning_rate": 9.999606900012642e-06,
      "loss": 1.206,
      "step": 132
    },
    {
      "epoch": 0.08580645161290322,
      "grad_norm": 2.3125,
      "learning_rate": 9.99955274037741e-06,
      "loss": 1.2524,
      "step": 133
    },
    {
      "epoch": 0.08645161290322581,
      "grad_norm": 1.75,
      "learning_rate": 9.999495086787056e-06,
      "loss": 0.9126,
      "step": 134
    },
    {
      "epoch": 0.08709677419354839,
      "grad_norm": 1.6640625,
      "learning_rate": 9.999433939281869e-06,
      "loss": 1.0961,
      "step": 135
    },
    {
      "epoch": 0.08774193548387096,
      "grad_norm": 0.9609375,
      "learning_rate": 9.999369297904582e-06,
      "loss": 1.0313,
      "step": 136
    },
    {
      "epoch": 0.08838709677419355,
      "grad_norm": 1.828125,
      "learning_rate": 9.99930116270037e-06,
      "loss": 1.3059,
      "step": 137
    },
    {
      "epoch": 0.08903225806451613,
      "grad_norm": 2.1875,
      "learning_rate": 9.99922953371685e-06,
      "loss": 1.1759,
      "step": 138
    },
    {
      "epoch": 0.08967741935483871,
      "grad_norm": 1.4296875,
      "learning_rate": 9.999154411004079e-06,
      "loss": 0.9465,
      "step": 139
    },
    {
      "epoch": 0.09032258064516129,
      "grad_norm": 0.8203125,
      "learning_rate": 9.99907579461456e-06,
      "loss": 0.9461,
      "step": 140
    },
    {
      "epoch": 0.09096774193548388,
      "grad_norm": 1.6640625,
      "learning_rate": 9.998993684603229e-06,
      "loss": 1.5636,
      "step": 141
    },
    {
      "epoch": 0.09161290322580645,
      "grad_norm": 1.1015625,
      "learning_rate": 9.998908081027472e-06,
      "loss": 0.9418,
      "step": 142
    },
    {
      "epoch": 0.09225806451612903,
      "grad_norm": 1.3828125,
      "learning_rate": 9.998818983947113e-06,
      "loss": 1.1224,
      "step": 143
    },
    {
      "epoch": 0.09290322580645162,
      "grad_norm": 1.0078125,
      "learning_rate": 9.998726393424421e-06,
      "loss": 0.9993,
      "step": 144
    },
    {
      "epoch": 0.0935483870967742,
      "grad_norm": 1.3984375,
      "learning_rate": 9.998630309524097e-06,
      "loss": 1.2941,
      "step": 145
    },
    {
      "epoch": 0.09419354838709677,
      "grad_norm": 2.734375,
      "learning_rate": 9.998530732313295e-06,
      "loss": 1.4663,
      "step": 146
    },
    {
      "epoch": 0.09483870967741935,
      "grad_norm": 1.2421875,
      "learning_rate": 9.9984276618616e-06,
      "loss": 1.052,
      "step": 147
    },
    {
      "epoch": 0.09548387096774194,
      "grad_norm": 1.546875,
      "learning_rate": 9.998321098241048e-06,
      "loss": 1.4316,
      "step": 148
    },
    {
      "epoch": 0.09612903225806452,
      "grad_norm": 0.8359375,
      "learning_rate": 9.998211041526109e-06,
      "loss": 1.0273,
      "step": 149
    },
    {
      "epoch": 0.0967741935483871,
      "grad_norm": 1.90625,
      "learning_rate": 9.998097491793696e-06,
      "loss": 1.5703,
      "step": 150
    },
    {
      "epoch": 0.09741935483870967,
      "grad_norm": 0.74609375,
      "learning_rate": 9.997980449123164e-06,
      "loss": 0.7252,
      "step": 151
    },
    {
      "epoch": 0.09806451612903226,
      "grad_norm": 0.80859375,
      "learning_rate": 9.99785991359631e-06,
      "loss": 0.8431,
      "step": 152
    },
    {
      "epoch": 0.09870967741935484,
      "grad_norm": 1.5859375,
      "learning_rate": 9.997735885297367e-06,
      "loss": 1.603,
      "step": 153
    },
    {
      "epoch": 0.09935483870967741,
      "grad_norm": 1.7109375,
      "learning_rate": 9.99760836431302e-06,
      "loss": 1.1785,
      "step": 154
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5234375,
      "learning_rate": 9.99747735073238e-06,
      "loss": 1.2651,
      "step": 155
    },
    {
      "epoch": 0.10064516129032258,
      "grad_norm": 1.34375,
      "learning_rate": 9.99734284464701e-06,
      "loss": 1.2288,
      "step": 156
    },
    {
      "epoch": 0.10129032258064516,
      "grad_norm": 0.78515625,
      "learning_rate": 9.997204846150907e-06,
      "loss": 0.8489,
      "step": 157
    },
    {
      "epoch": 0.10193548387096774,
      "grad_norm": 1.0703125,
      "learning_rate": 9.997063355340515e-06,
      "loss": 1.0371,
      "step": 158
    },
    {
      "epoch": 0.10258064516129033,
      "grad_norm": 0.83984375,
      "learning_rate": 9.996918372314714e-06,
      "loss": 0.7911,
      "step": 159
    },
    {
      "epoch": 0.1032258064516129,
      "grad_norm": 0.8515625,
      "learning_rate": 9.996769897174827e-06,
      "loss": 1.1003,
      "step": 160
    },
    {
      "epoch": 0.10387096774193548,
      "grad_norm": 0.65234375,
      "learning_rate": 9.996617930024615e-06,
      "loss": 0.6678,
      "step": 161
    },
    {
      "epoch": 0.10451612903225807,
      "grad_norm": 1.1484375,
      "learning_rate": 9.996462470970283e-06,
      "loss": 1.105,
      "step": 162
    },
    {
      "epoch": 0.10516129032258065,
      "grad_norm": 0.99609375,
      "learning_rate": 9.996303520120469e-06,
      "loss": 1.0399,
      "step": 163
    },
    {
      "epoch": 0.10580645161290322,
      "grad_norm": 1.2265625,
      "learning_rate": 9.996141077586263e-06,
      "loss": 1.0441,
      "step": 164
    },
    {
      "epoch": 0.1064516129032258,
      "grad_norm": 1.0390625,
      "learning_rate": 9.995975143481183e-06,
      "loss": 1.0497,
      "step": 165
    },
    {
      "epoch": 0.10709677419354839,
      "grad_norm": 0.90625,
      "learning_rate": 9.995805717921196e-06,
      "loss": 0.969,
      "step": 166
    },
    {
      "epoch": 0.10774193548387097,
      "grad_norm": 0.80078125,
      "learning_rate": 9.995632801024703e-06,
      "loss": 0.7827,
      "step": 167
    },
    {
      "epoch": 0.10838709677419354,
      "grad_norm": 0.82421875,
      "learning_rate": 9.995456392912552e-06,
      "loss": 0.9293,
      "step": 168
    },
    {
      "epoch": 0.10903225806451614,
      "grad_norm": 6.40625,
      "learning_rate": 9.995276493708022e-06,
      "loss": 2.0858,
      "step": 169
    },
    {
      "epoch": 0.10967741935483871,
      "grad_norm": 0.8984375,
      "learning_rate": 9.99509310353684e-06,
      "loss": 0.9377,
      "step": 170
    },
    {
      "epoch": 0.11032258064516129,
      "grad_norm": 0.859375,
      "learning_rate": 9.994906222527163e-06,
      "loss": 0.9357,
      "step": 171
    },
    {
      "epoch": 0.11096774193548387,
      "grad_norm": 0.76171875,
      "learning_rate": 9.994715850809598e-06,
      "loss": 0.8367,
      "step": 172
    },
    {
      "epoch": 0.11161290322580646,
      "grad_norm": 1.1796875,
      "learning_rate": 9.994521988517187e-06,
      "loss": 1.2082,
      "step": 173
    },
    {
      "epoch": 0.11225806451612903,
      "grad_norm": 0.96484375,
      "learning_rate": 9.994324635785411e-06,
      "loss": 0.847,
      "step": 174
    },
    {
      "epoch": 0.11290322580645161,
      "grad_norm": 0.8203125,
      "learning_rate": 9.994123792752189e-06,
      "loss": 0.9347,
      "step": 175
    },
    {
      "epoch": 0.1135483870967742,
      "grad_norm": 1.1015625,
      "learning_rate": 9.993919459557884e-06,
      "loss": 1.2138,
      "step": 176
    },
    {
      "epoch": 0.11419354838709678,
      "grad_norm": 1.0,
      "learning_rate": 9.99371163634529e-06,
      "loss": 1.0669,
      "step": 177
    },
    {
      "epoch": 0.11483870967741935,
      "grad_norm": 0.79296875,
      "learning_rate": 9.99350032325965e-06,
      "loss": 0.79,
      "step": 178
    },
    {
      "epoch": 0.11548387096774193,
      "grad_norm": 1.40625,
      "learning_rate": 9.993285520448637e-06,
      "loss": 1.4555,
      "step": 179
    },
    {
      "epoch": 0.11612903225806452,
      "grad_norm": 1.359375,
      "learning_rate": 9.99306722806237e-06,
      "loss": 1.2415,
      "step": 180
    },
    {
      "epoch": 0.1167741935483871,
      "grad_norm": 1.1015625,
      "learning_rate": 9.992845446253401e-06,
      "loss": 1.13,
      "step": 181
    },
    {
      "epoch": 0.11741935483870967,
      "grad_norm": 0.953125,
      "learning_rate": 9.992620175176723e-06,
      "loss": 1.0078,
      "step": 182
    },
    {
      "epoch": 0.11806451612903225,
      "grad_norm": 0.9921875,
      "learning_rate": 9.99239141498977e-06,
      "loss": 1.0988,
      "step": 183
    },
    {
      "epoch": 0.11870967741935484,
      "grad_norm": 0.73046875,
      "learning_rate": 9.99215916585241e-06,
      "loss": 0.8427,
      "step": 184
    },
    {
      "epoch": 0.11935483870967742,
      "grad_norm": 0.796875,
      "learning_rate": 9.991923427926953e-06,
      "loss": 0.8431,
      "step": 185
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.76171875,
      "learning_rate": 9.991684201378142e-06,
      "loss": 0.876,
      "step": 186
    },
    {
      "epoch": 0.12064516129032259,
      "grad_norm": 1.5390625,
      "learning_rate": 9.991441486373163e-06,
      "loss": 1.2985,
      "step": 187
    },
    {
      "epoch": 0.12129032258064516,
      "grad_norm": 0.76953125,
      "learning_rate": 9.991195283081638e-06,
      "loss": 0.8569,
      "step": 188
    },
    {
      "epoch": 0.12193548387096774,
      "grad_norm": 1.1171875,
      "learning_rate": 9.990945591675627e-06,
      "loss": 1.3719,
      "step": 189
    },
    {
      "epoch": 0.12258064516129032,
      "grad_norm": 0.83984375,
      "learning_rate": 9.99069241232963e-06,
      "loss": 0.7805,
      "step": 190
    },
    {
      "epoch": 0.12322580645161291,
      "grad_norm": 1.171875,
      "learning_rate": 9.990435745220579e-06,
      "loss": 0.9408,
      "step": 191
    },
    {
      "epoch": 0.12387096774193548,
      "grad_norm": 0.81640625,
      "learning_rate": 9.990175590527847e-06,
      "loss": 0.8475,
      "step": 192
    },
    {
      "epoch": 0.12451612903225806,
      "grad_norm": 1.421875,
      "learning_rate": 9.989911948433246e-06,
      "loss": 1.4866,
      "step": 193
    },
    {
      "epoch": 0.12516129032258064,
      "grad_norm": 0.9453125,
      "learning_rate": 9.989644819121023e-06,
      "loss": 0.9207,
      "step": 194
    },
    {
      "epoch": 0.12580645161290321,
      "grad_norm": 0.90234375,
      "learning_rate": 9.98937420277786e-06,
      "loss": 0.9502,
      "step": 195
    },
    {
      "epoch": 0.12645161290322582,
      "grad_norm": 1.296875,
      "learning_rate": 9.989100099592881e-06,
      "loss": 1.1659,
      "step": 196
    },
    {
      "epoch": 0.1270967741935484,
      "grad_norm": 1.1171875,
      "learning_rate": 9.988822509757643e-06,
      "loss": 1.1396,
      "step": 197
    },
    {
      "epoch": 0.12774193548387097,
      "grad_norm": 0.953125,
      "learning_rate": 9.98854143346614e-06,
      "loss": 0.964,
      "step": 198
    },
    {
      "epoch": 0.12838709677419355,
      "grad_norm": 1.203125,
      "learning_rate": 9.988256870914803e-06,
      "loss": 1.1181,
      "step": 199
    },
    {
      "epoch": 0.12903225806451613,
      "grad_norm": 0.84765625,
      "learning_rate": 9.987968822302498e-06,
      "loss": 0.9404,
      "step": 200
    },
    {
      "epoch": 0.1296774193548387,
      "grad_norm": 0.8828125,
      "learning_rate": 9.987677287830533e-06,
      "loss": 0.8582,
      "step": 201
    },
    {
      "epoch": 0.13032258064516128,
      "grad_norm": 1.609375,
      "learning_rate": 9.987382267702645e-06,
      "loss": 1.258,
      "step": 202
    },
    {
      "epoch": 0.13096774193548388,
      "grad_norm": 0.671875,
      "learning_rate": 9.987083762125011e-06,
      "loss": 0.8378,
      "step": 203
    },
    {
      "epoch": 0.13161290322580646,
      "grad_norm": 0.78125,
      "learning_rate": 9.986781771306244e-06,
      "loss": 0.7092,
      "step": 204
    },
    {
      "epoch": 0.13225806451612904,
      "grad_norm": 0.625,
      "learning_rate": 9.986476295457389e-06,
      "loss": 0.7777,
      "step": 205
    },
    {
      "epoch": 0.1329032258064516,
      "grad_norm": 1.0390625,
      "learning_rate": 9.986167334791928e-06,
      "loss": 1.1904,
      "step": 206
    },
    {
      "epoch": 0.1335483870967742,
      "grad_norm": 0.75390625,
      "learning_rate": 9.985854889525781e-06,
      "loss": 0.6754,
      "step": 207
    },
    {
      "epoch": 0.13419354838709677,
      "grad_norm": 0.7734375,
      "learning_rate": 9.9855389598773e-06,
      "loss": 0.7443,
      "step": 208
    },
    {
      "epoch": 0.13483870967741934,
      "grad_norm": 0.984375,
      "learning_rate": 9.985219546067278e-06,
      "loss": 1.0002,
      "step": 209
    },
    {
      "epoch": 0.13548387096774195,
      "grad_norm": 1.2578125,
      "learning_rate": 9.984896648318934e-06,
      "loss": 1.2731,
      "step": 210
    },
    {
      "epoch": 0.13612903225806453,
      "grad_norm": 1.828125,
      "learning_rate": 9.984570266857925e-06,
      "loss": 1.1603,
      "step": 211
    },
    {
      "epoch": 0.1367741935483871,
      "grad_norm": 2.21875,
      "learning_rate": 9.984240401912349e-06,
      "loss": 0.9211,
      "step": 212
    },
    {
      "epoch": 0.13741935483870968,
      "grad_norm": 0.83984375,
      "learning_rate": 9.983907053712731e-06,
      "loss": 0.9287,
      "step": 213
    },
    {
      "epoch": 0.13806451612903226,
      "grad_norm": 2.046875,
      "learning_rate": 9.98357022249203e-06,
      "loss": 1.2716,
      "step": 214
    },
    {
      "epoch": 0.13870967741935483,
      "grad_norm": 0.7578125,
      "learning_rate": 9.983229908485646e-06,
      "loss": 0.7962,
      "step": 215
    },
    {
      "epoch": 0.1393548387096774,
      "grad_norm": 1.484375,
      "learning_rate": 9.982886111931405e-06,
      "loss": 1.0694,
      "step": 216
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1171875,
      "learning_rate": 9.982538833069572e-06,
      "loss": 0.9685,
      "step": 217
    },
    {
      "epoch": 0.1406451612903226,
      "grad_norm": 0.7109375,
      "learning_rate": 9.982188072142844e-06,
      "loss": 0.812,
      "step": 218
    },
    {
      "epoch": 0.14129032258064517,
      "grad_norm": 1.0,
      "learning_rate": 9.981833829396353e-06,
      "loss": 1.0779,
      "step": 219
    },
    {
      "epoch": 0.14193548387096774,
      "grad_norm": 1.78125,
      "learning_rate": 9.98147610507766e-06,
      "loss": 0.7881,
      "step": 220
    },
    {
      "epoch": 0.14258064516129032,
      "grad_norm": 1.0859375,
      "learning_rate": 9.981114899436763e-06,
      "loss": 1.0002,
      "step": 221
    },
    {
      "epoch": 0.1432258064516129,
      "grad_norm": 0.8515625,
      "learning_rate": 9.980750212726093e-06,
      "loss": 0.8835,
      "step": 222
    },
    {
      "epoch": 0.14387096774193547,
      "grad_norm": 1.125,
      "learning_rate": 9.980382045200512e-06,
      "loss": 1.2264,
      "step": 223
    },
    {
      "epoch": 0.14451612903225808,
      "grad_norm": 1.1484375,
      "learning_rate": 9.980010397117315e-06,
      "loss": 1.068,
      "step": 224
    },
    {
      "epoch": 0.14516129032258066,
      "grad_norm": 0.70703125,
      "learning_rate": 9.979635268736228e-06,
      "loss": 0.7863,
      "step": 225
    },
    {
      "epoch": 0.14580645161290323,
      "grad_norm": 1.1640625,
      "learning_rate": 9.979256660319412e-06,
      "loss": 1.2254,
      "step": 226
    },
    {
      "epoch": 0.1464516129032258,
      "grad_norm": 1.0546875,
      "learning_rate": 9.978874572131461e-06,
      "loss": 1.0456,
      "step": 227
    },
    {
      "epoch": 0.14709677419354839,
      "grad_norm": 0.8203125,
      "learning_rate": 9.978489004439394e-06,
      "loss": 0.8037,
      "step": 228
    },
    {
      "epoch": 0.14774193548387096,
      "grad_norm": 1.046875,
      "learning_rate": 9.978099957512672e-06,
      "loss": 1.1191,
      "step": 229
    },
    {
      "epoch": 0.14838709677419354,
      "grad_norm": 0.85546875,
      "learning_rate": 9.977707431623177e-06,
      "loss": 1.0564,
      "step": 230
    },
    {
      "epoch": 0.14903225806451612,
      "grad_norm": 0.59765625,
      "learning_rate": 9.977311427045228e-06,
      "loss": 0.7734,
      "step": 231
    },
    {
      "epoch": 0.14967741935483872,
      "grad_norm": 1.1796875,
      "learning_rate": 9.976911944055578e-06,
      "loss": 0.8634,
      "step": 232
    },
    {
      "epoch": 0.1503225806451613,
      "grad_norm": 0.8203125,
      "learning_rate": 9.976508982933404e-06,
      "loss": 0.9273,
      "step": 233
    },
    {
      "epoch": 0.15096774193548387,
      "grad_norm": 0.7578125,
      "learning_rate": 9.976102543960315e-06,
      "loss": 0.7367,
      "step": 234
    },
    {
      "epoch": 0.15161290322580645,
      "grad_norm": 0.72265625,
      "learning_rate": 9.975692627420357e-06,
      "loss": 0.8658,
      "step": 235
    },
    {
      "epoch": 0.15225806451612903,
      "grad_norm": 0.93359375,
      "learning_rate": 9.975279233599998e-06,
      "loss": 1.0858,
      "step": 236
    },
    {
      "epoch": 0.1529032258064516,
      "grad_norm": 0.9375,
      "learning_rate": 9.974862362788141e-06,
      "loss": 1.0318,
      "step": 237
    },
    {
      "epoch": 0.15354838709677418,
      "grad_norm": 0.9453125,
      "learning_rate": 9.974442015276116e-06,
      "loss": 1.0005,
      "step": 238
    },
    {
      "epoch": 0.15419354838709678,
      "grad_norm": 0.58984375,
      "learning_rate": 9.974018191357686e-06,
      "loss": 0.5678,
      "step": 239
    },
    {
      "epoch": 0.15483870967741936,
      "grad_norm": 1.1328125,
      "learning_rate": 9.973590891329043e-06,
      "loss": 1.1655,
      "step": 240
    },
    {
      "epoch": 0.15548387096774194,
      "grad_norm": 0.90625,
      "learning_rate": 9.973160115488804e-06,
      "loss": 0.958,
      "step": 241
    },
    {
      "epoch": 0.15612903225806452,
      "grad_norm": 0.96875,
      "learning_rate": 9.972725864138022e-06,
      "loss": 1.0552,
      "step": 242
    },
    {
      "epoch": 0.1567741935483871,
      "grad_norm": 0.6796875,
      "learning_rate": 9.972288137580173e-06,
      "loss": 0.5835,
      "step": 243
    },
    {
      "epoch": 0.15741935483870967,
      "grad_norm": 0.76953125,
      "learning_rate": 9.971846936121163e-06,
      "loss": 0.8384,
      "step": 244
    },
    {
      "epoch": 0.15806451612903225,
      "grad_norm": 0.91015625,
      "learning_rate": 9.971402260069328e-06,
      "loss": 0.89,
      "step": 245
    },
    {
      "epoch": 0.15870967741935485,
      "grad_norm": 0.859375,
      "learning_rate": 9.970954109735431e-06,
      "loss": 1.1842,
      "step": 246
    },
    {
      "epoch": 0.15935483870967743,
      "grad_norm": 1.0078125,
      "learning_rate": 9.970502485432665e-06,
      "loss": 0.9333,
      "step": 247
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0703125,
      "learning_rate": 9.970047387476646e-06,
      "loss": 1.0779,
      "step": 248
    },
    {
      "epoch": 0.16064516129032258,
      "grad_norm": 0.75,
      "learning_rate": 9.969588816185424e-06,
      "loss": 0.9552,
      "step": 249
    },
    {
      "epoch": 0.16129032258064516,
      "grad_norm": 0.69921875,
      "learning_rate": 9.969126771879473e-06,
      "loss": 0.7234,
      "step": 250
    },
    {
      "epoch": 0.16193548387096773,
      "grad_norm": 0.81640625,
      "learning_rate": 9.968661254881695e-06,
      "loss": 1.0216,
      "step": 251
    },
    {
      "epoch": 0.1625806451612903,
      "grad_norm": 0.79296875,
      "learning_rate": 9.968192265517414e-06,
      "loss": 0.8529,
      "step": 252
    },
    {
      "epoch": 0.16322580645161291,
      "grad_norm": 1.1484375,
      "learning_rate": 9.967719804114389e-06,
      "loss": 0.7876,
      "step": 253
    },
    {
      "epoch": 0.1638709677419355,
      "grad_norm": 0.80078125,
      "learning_rate": 9.967243871002798e-06,
      "loss": 0.9804,
      "step": 254
    },
    {
      "epoch": 0.16451612903225807,
      "grad_norm": 1.328125,
      "learning_rate": 9.966764466515253e-06,
      "loss": 1.0655,
      "step": 255
    },
    {
      "epoch": 0.16516129032258065,
      "grad_norm": 0.8359375,
      "learning_rate": 9.966281590986782e-06,
      "loss": 1.0284,
      "step": 256
    },
    {
      "epoch": 0.16580645161290322,
      "grad_norm": 1.078125,
      "learning_rate": 9.96579524475485e-06,
      "loss": 1.1077,
      "step": 257
    },
    {
      "epoch": 0.1664516129032258,
      "grad_norm": 0.67578125,
      "learning_rate": 9.965305428159337e-06,
      "loss": 0.5325,
      "step": 258
    },
    {
      "epoch": 0.16709677419354838,
      "grad_norm": 0.89453125,
      "learning_rate": 9.964812141542554e-06,
      "loss": 0.8753,
      "step": 259
    },
    {
      "epoch": 0.16774193548387098,
      "grad_norm": 1.203125,
      "learning_rate": 9.964315385249241e-06,
      "loss": 1.0298,
      "step": 260
    },
    {
      "epoch": 0.16838709677419356,
      "grad_norm": 1.0859375,
      "learning_rate": 9.963815159626551e-06,
      "loss": 0.7215,
      "step": 261
    },
    {
      "epoch": 0.16903225806451613,
      "grad_norm": 0.75390625,
      "learning_rate": 9.963311465024073e-06,
      "loss": 0.8343,
      "step": 262
    },
    {
      "epoch": 0.1696774193548387,
      "grad_norm": 1.0546875,
      "learning_rate": 9.962804301793812e-06,
      "loss": 1.0548,
      "step": 263
    },
    {
      "epoch": 0.1703225806451613,
      "grad_norm": 1.1796875,
      "learning_rate": 9.962293670290204e-06,
      "loss": 0.7539,
      "step": 264
    },
    {
      "epoch": 0.17096774193548386,
      "grad_norm": 1.109375,
      "learning_rate": 9.961779570870106e-06,
      "loss": 1.1531,
      "step": 265
    },
    {
      "epoch": 0.17161290322580644,
      "grad_norm": 1.03125,
      "learning_rate": 9.961262003892796e-06,
      "loss": 0.8971,
      "step": 266
    },
    {
      "epoch": 0.17225806451612904,
      "grad_norm": 1.140625,
      "learning_rate": 9.960740969719978e-06,
      "loss": 1.3348,
      "step": 267
    },
    {
      "epoch": 0.17290322580645162,
      "grad_norm": 1.234375,
      "learning_rate": 9.96021646871578e-06,
      "loss": 1.2319,
      "step": 268
    },
    {
      "epoch": 0.1735483870967742,
      "grad_norm": 0.96484375,
      "learning_rate": 9.959688501246749e-06,
      "loss": 0.7369,
      "step": 269
    },
    {
      "epoch": 0.17419354838709677,
      "grad_norm": 0.62890625,
      "learning_rate": 9.95915706768186e-06,
      "loss": 0.7005,
      "step": 270
    },
    {
      "epoch": 0.17483870967741935,
      "grad_norm": 0.62890625,
      "learning_rate": 9.958622168392503e-06,
      "loss": 0.7175,
      "step": 271
    },
    {
      "epoch": 0.17548387096774193,
      "grad_norm": 0.74609375,
      "learning_rate": 9.958083803752498e-06,
      "loss": 0.9461,
      "step": 272
    },
    {
      "epoch": 0.1761290322580645,
      "grad_norm": 1.046875,
      "learning_rate": 9.957541974138082e-06,
      "loss": 1.0115,
      "step": 273
    },
    {
      "epoch": 0.1767741935483871,
      "grad_norm": 1.1484375,
      "learning_rate": 9.956996679927914e-06,
      "loss": 0.8242,
      "step": 274
    },
    {
      "epoch": 0.1774193548387097,
      "grad_norm": 1.359375,
      "learning_rate": 9.956447921503076e-06,
      "loss": 0.9995,
      "step": 275
    },
    {
      "epoch": 0.17806451612903226,
      "grad_norm": 0.76953125,
      "learning_rate": 9.955895699247067e-06,
      "loss": 0.8498,
      "step": 276
    },
    {
      "epoch": 0.17870967741935484,
      "grad_norm": 0.546875,
      "learning_rate": 9.955340013545814e-06,
      "loss": 0.5949,
      "step": 277
    },
    {
      "epoch": 0.17935483870967742,
      "grad_norm": 0.703125,
      "learning_rate": 9.954780864787656e-06,
      "loss": 0.7856,
      "step": 278
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.93359375,
      "learning_rate": 9.954218253363356e-06,
      "loss": 1.0907,
      "step": 279
    },
    {
      "epoch": 0.18064516129032257,
      "grad_norm": 0.88671875,
      "learning_rate": 9.9536521796661e-06,
      "loss": 0.8208,
      "step": 280
    },
    {
      "epoch": 0.18129032258064517,
      "grad_norm": 0.53125,
      "learning_rate": 9.95308264409149e-06,
      "loss": 0.5974,
      "step": 281
    },
    {
      "epoch": 0.18193548387096775,
      "grad_norm": 1.2890625,
      "learning_rate": 9.952509647037545e-06,
      "loss": 0.8772,
      "step": 282
    },
    {
      "epoch": 0.18258064516129033,
      "grad_norm": 1.0390625,
      "learning_rate": 9.951933188904707e-06,
      "loss": 0.8342,
      "step": 283
    },
    {
      "epoch": 0.1832258064516129,
      "grad_norm": 0.7890625,
      "learning_rate": 9.95135327009584e-06,
      "loss": 0.8676,
      "step": 284
    },
    {
      "epoch": 0.18387096774193548,
      "grad_norm": 1.015625,
      "learning_rate": 9.950769891016217e-06,
      "loss": 0.9969,
      "step": 285
    },
    {
      "epoch": 0.18451612903225806,
      "grad_norm": 0.6875,
      "learning_rate": 9.950183052073536e-06,
      "loss": 0.8181,
      "step": 286
    },
    {
      "epoch": 0.18516129032258064,
      "grad_norm": 0.515625,
      "learning_rate": 9.949592753677914e-06,
      "loss": 0.5935,
      "step": 287
    },
    {
      "epoch": 0.18580645161290324,
      "grad_norm": 1.328125,
      "learning_rate": 9.948998996241879e-06,
      "loss": 1.5798,
      "step": 288
    },
    {
      "epoch": 0.18645161290322582,
      "grad_norm": 0.77734375,
      "learning_rate": 9.948401780180385e-06,
      "loss": 0.845,
      "step": 289
    },
    {
      "epoch": 0.1870967741935484,
      "grad_norm": 1.03125,
      "learning_rate": 9.947801105910795e-06,
      "loss": 0.7626,
      "step": 290
    },
    {
      "epoch": 0.18774193548387097,
      "grad_norm": 1.3984375,
      "learning_rate": 9.947196973852893e-06,
      "loss": 1.5021,
      "step": 291
    },
    {
      "epoch": 0.18838709677419355,
      "grad_norm": 0.8046875,
      "learning_rate": 9.94658938442888e-06,
      "loss": 0.8437,
      "step": 292
    },
    {
      "epoch": 0.18903225806451612,
      "grad_norm": 0.6328125,
      "learning_rate": 9.945978338063371e-06,
      "loss": 0.7976,
      "step": 293
    },
    {
      "epoch": 0.1896774193548387,
      "grad_norm": 0.6796875,
      "learning_rate": 9.945363835183399e-06,
      "loss": 0.712,
      "step": 294
    },
    {
      "epoch": 0.19032258064516128,
      "grad_norm": 0.7734375,
      "learning_rate": 9.94474587621841e-06,
      "loss": 0.8308,
      "step": 295
    },
    {
      "epoch": 0.19096774193548388,
      "grad_norm": 0.71875,
      "learning_rate": 9.944124461600268e-06,
      "loss": 0.7373,
      "step": 296
    },
    {
      "epoch": 0.19161290322580646,
      "grad_norm": 0.5390625,
      "learning_rate": 9.943499591763251e-06,
      "loss": 0.5174,
      "step": 297
    },
    {
      "epoch": 0.19225806451612903,
      "grad_norm": 1.0078125,
      "learning_rate": 9.942871267144051e-06,
      "loss": 1.0544,
      "step": 298
    },
    {
      "epoch": 0.1929032258064516,
      "grad_norm": 0.97265625,
      "learning_rate": 9.942239488181774e-06,
      "loss": 1.1261,
      "step": 299
    },
    {
      "epoch": 0.1935483870967742,
      "grad_norm": 0.74609375,
      "learning_rate": 9.941604255317942e-06,
      "loss": 0.9205,
      "step": 300
    },
    {
      "epoch": 0.19419354838709676,
      "grad_norm": 0.984375,
      "learning_rate": 9.940965568996491e-06,
      "loss": 1.0551,
      "step": 301
    },
    {
      "epoch": 0.19483870967741934,
      "grad_norm": 1.1640625,
      "learning_rate": 9.940323429663764e-06,
      "loss": 1.3329,
      "step": 302
    },
    {
      "epoch": 0.19548387096774195,
      "grad_norm": 0.8125,
      "learning_rate": 9.93967783776853e-06,
      "loss": 0.9766,
      "step": 303
    },
    {
      "epoch": 0.19612903225806452,
      "grad_norm": 0.640625,
      "learning_rate": 9.939028793761958e-06,
      "loss": 0.5867,
      "step": 304
    },
    {
      "epoch": 0.1967741935483871,
      "grad_norm": 0.6953125,
      "learning_rate": 9.938376298097636e-06,
      "loss": 0.8689,
      "step": 305
    },
    {
      "epoch": 0.19741935483870968,
      "grad_norm": 0.76171875,
      "learning_rate": 9.937720351231562e-06,
      "loss": 0.7051,
      "step": 306
    },
    {
      "epoch": 0.19806451612903225,
      "grad_norm": 0.640625,
      "learning_rate": 9.93706095362215e-06,
      "loss": 0.7506,
      "step": 307
    },
    {
      "epoch": 0.19870967741935483,
      "grad_norm": 1.0078125,
      "learning_rate": 9.936398105730217e-06,
      "loss": 1.0745,
      "step": 308
    },
    {
      "epoch": 0.1993548387096774,
      "grad_norm": 0.78125,
      "learning_rate": 9.935731808019e-06,
      "loss": 0.7732,
      "step": 309
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.96484375,
      "learning_rate": 9.935062060954144e-06,
      "loss": 0.9318,
      "step": 310
    },
    {
      "epoch": 0.2006451612903226,
      "grad_norm": 0.703125,
      "learning_rate": 9.934388865003703e-06,
      "loss": 0.803,
      "step": 311
    },
    {
      "epoch": 0.20129032258064516,
      "grad_norm": 0.67578125,
      "learning_rate": 9.933712220638143e-06,
      "loss": 0.9452,
      "step": 312
    },
    {
      "epoch": 0.20193548387096774,
      "grad_norm": 0.72265625,
      "learning_rate": 9.933032128330339e-06,
      "loss": 0.9548,
      "step": 313
    },
    {
      "epoch": 0.20258064516129032,
      "grad_norm": 1.2109375,
      "learning_rate": 9.932348588555576e-06,
      "loss": 0.8689,
      "step": 314
    },
    {
      "epoch": 0.2032258064516129,
      "grad_norm": 0.6953125,
      "learning_rate": 9.931661601791547e-06,
      "loss": 0.8893,
      "step": 315
    },
    {
      "epoch": 0.20387096774193547,
      "grad_norm": 0.765625,
      "learning_rate": 9.930971168518356e-06,
      "loss": 0.7652,
      "step": 316
    },
    {
      "epoch": 0.20451612903225808,
      "grad_norm": 0.59375,
      "learning_rate": 9.930277289218515e-06,
      "loss": 0.6038,
      "step": 317
    },
    {
      "epoch": 0.20516129032258065,
      "grad_norm": 0.9765625,
      "learning_rate": 9.929579964376946e-06,
      "loss": 0.8517,
      "step": 318
    },
    {
      "epoch": 0.20580645161290323,
      "grad_norm": 0.83984375,
      "learning_rate": 9.928879194480972e-06,
      "loss": 1.0203,
      "step": 319
    },
    {
      "epoch": 0.2064516129032258,
      "grad_norm": 0.78515625,
      "learning_rate": 9.928174980020333e-06,
      "loss": 0.7,
      "step": 320
    },
    {
      "epoch": 0.20709677419354838,
      "grad_norm": 0.515625,
      "learning_rate": 9.92746732148717e-06,
      "loss": 0.6329,
      "step": 321
    },
    {
      "epoch": 0.20774193548387096,
      "grad_norm": 0.62109375,
      "learning_rate": 9.926756219376032e-06,
      "loss": 0.7363,
      "step": 322
    },
    {
      "epoch": 0.20838709677419354,
      "grad_norm": 0.7421875,
      "learning_rate": 9.926041674183875e-06,
      "loss": 0.9654,
      "step": 323
    },
    {
      "epoch": 0.20903225806451614,
      "grad_norm": 1.1015625,
      "learning_rate": 9.925323686410065e-06,
      "loss": 1.4433,
      "step": 324
    },
    {
      "epoch": 0.20967741935483872,
      "grad_norm": 0.83203125,
      "learning_rate": 9.924602256556366e-06,
      "loss": 0.6546,
      "step": 325
    },
    {
      "epoch": 0.2103225806451613,
      "grad_norm": 0.96875,
      "learning_rate": 9.923877385126954e-06,
      "loss": 1.1367,
      "step": 326
    },
    {
      "epoch": 0.21096774193548387,
      "grad_norm": 0.5703125,
      "learning_rate": 9.923149072628407e-06,
      "loss": 0.4816,
      "step": 327
    },
    {
      "epoch": 0.21161290322580645,
      "grad_norm": 0.8125,
      "learning_rate": 9.92241731956971e-06,
      "loss": 1.0069,
      "step": 328
    },
    {
      "epoch": 0.21225806451612902,
      "grad_norm": 1.046875,
      "learning_rate": 9.92168212646225e-06,
      "loss": 1.2396,
      "step": 329
    },
    {
      "epoch": 0.2129032258064516,
      "grad_norm": 1.1796875,
      "learning_rate": 9.92094349381982e-06,
      "loss": 1.2078,
      "step": 330
    },
    {
      "epoch": 0.2135483870967742,
      "grad_norm": 0.91796875,
      "learning_rate": 9.920201422158615e-06,
      "loss": 1.084,
      "step": 331
    },
    {
      "epoch": 0.21419354838709678,
      "grad_norm": 0.66015625,
      "learning_rate": 9.919455911997236e-06,
      "loss": 0.7513,
      "step": 332
    },
    {
      "epoch": 0.21483870967741936,
      "grad_norm": 0.90234375,
      "learning_rate": 9.918706963856685e-06,
      "loss": 1.1725,
      "step": 333
    },
    {
      "epoch": 0.21548387096774194,
      "grad_norm": 0.53125,
      "learning_rate": 9.917954578260367e-06,
      "loss": 0.6475,
      "step": 334
    },
    {
      "epoch": 0.2161290322580645,
      "grad_norm": 0.6484375,
      "learning_rate": 9.917198755734088e-06,
      "loss": 0.8303,
      "step": 335
    },
    {
      "epoch": 0.2167741935483871,
      "grad_norm": 1.1015625,
      "learning_rate": 9.91643949680606e-06,
      "loss": 1.056,
      "step": 336
    },
    {
      "epoch": 0.21741935483870967,
      "grad_norm": 0.984375,
      "learning_rate": 9.915676802006888e-06,
      "loss": 1.0047,
      "step": 337
    },
    {
      "epoch": 0.21806451612903227,
      "grad_norm": 0.96875,
      "learning_rate": 9.914910671869593e-06,
      "loss": 1.0321,
      "step": 338
    },
    {
      "epoch": 0.21870967741935485,
      "grad_norm": 0.58984375,
      "learning_rate": 9.914141106929581e-06,
      "loss": 0.5329,
      "step": 339
    },
    {
      "epoch": 0.21935483870967742,
      "grad_norm": 1.1875,
      "learning_rate": 9.913368107724664e-06,
      "loss": 1.0801,
      "step": 340
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.09375,
      "learning_rate": 9.912591674795061e-06,
      "loss": 0.8832,
      "step": 341
    },
    {
      "epoch": 0.22064516129032258,
      "grad_norm": 0.54296875,
      "learning_rate": 9.911811808683382e-06,
      "loss": 0.6521,
      "step": 342
    },
    {
      "epoch": 0.22129032258064515,
      "grad_norm": 0.875,
      "learning_rate": 9.91102850993464e-06,
      "loss": 0.9194,
      "step": 343
    },
    {
      "epoch": 0.22193548387096773,
      "grad_norm": 0.87890625,
      "learning_rate": 9.910241779096244e-06,
      "loss": 1.2793,
      "step": 344
    },
    {
      "epoch": 0.22258064516129034,
      "grad_norm": 0.8828125,
      "learning_rate": 9.909451616718004e-06,
      "loss": 0.9307,
      "step": 345
    },
    {
      "epoch": 0.2232258064516129,
      "grad_norm": 1.0546875,
      "learning_rate": 9.908658023352131e-06,
      "loss": 1.0233,
      "step": 346
    },
    {
      "epoch": 0.2238709677419355,
      "grad_norm": 0.78515625,
      "learning_rate": 9.907860999553228e-06,
      "loss": 0.9072,
      "step": 347
    },
    {
      "epoch": 0.22451612903225807,
      "grad_norm": 0.81640625,
      "learning_rate": 9.907060545878297e-06,
      "loss": 0.8033,
      "step": 348
    },
    {
      "epoch": 0.22516129032258064,
      "grad_norm": 0.65625,
      "learning_rate": 9.906256662886741e-06,
      "loss": 0.8208,
      "step": 349
    },
    {
      "epoch": 0.22580645161290322,
      "grad_norm": 0.6953125,
      "learning_rate": 9.905449351140353e-06,
      "loss": 0.6743,
      "step": 350
    },
    {
      "epoch": 0.2264516129032258,
      "grad_norm": 0.5390625,
      "learning_rate": 9.904638611203327e-06,
      "loss": 0.6184,
      "step": 351
    },
    {
      "epoch": 0.2270967741935484,
      "grad_norm": 1.140625,
      "learning_rate": 9.903824443642253e-06,
      "loss": 1.0984,
      "step": 352
    },
    {
      "epoch": 0.22774193548387098,
      "grad_norm": 0.7265625,
      "learning_rate": 9.903006849026114e-06,
      "loss": 0.6772,
      "step": 353
    },
    {
      "epoch": 0.22838709677419355,
      "grad_norm": 0.7265625,
      "learning_rate": 9.902185827926287e-06,
      "loss": 1.04,
      "step": 354
    },
    {
      "epoch": 0.22903225806451613,
      "grad_norm": 0.640625,
      "learning_rate": 9.901361380916547e-06,
      "loss": 0.8561,
      "step": 355
    },
    {
      "epoch": 0.2296774193548387,
      "grad_norm": 0.73828125,
      "learning_rate": 9.900533508573062e-06,
      "loss": 0.9174,
      "step": 356
    },
    {
      "epoch": 0.23032258064516128,
      "grad_norm": 0.73046875,
      "learning_rate": 9.899702211474394e-06,
      "loss": 0.919,
      "step": 357
    },
    {
      "epoch": 0.23096774193548386,
      "grad_norm": 0.6484375,
      "learning_rate": 9.898867490201496e-06,
      "loss": 0.6882,
      "step": 358
    },
    {
      "epoch": 0.23161290322580644,
      "grad_norm": 0.7890625,
      "learning_rate": 9.898029345337718e-06,
      "loss": 0.8141,
      "step": 359
    },
    {
      "epoch": 0.23225806451612904,
      "grad_norm": 1.03125,
      "learning_rate": 9.8971877774688e-06,
      "loss": 0.8419,
      "step": 360
    },
    {
      "epoch": 0.23290322580645162,
      "grad_norm": 0.7734375,
      "learning_rate": 9.896342787182873e-06,
      "loss": 1.0275,
      "step": 361
    },
    {
      "epoch": 0.2335483870967742,
      "grad_norm": 1.3046875,
      "learning_rate": 9.895494375070464e-06,
      "loss": 1.0708,
      "step": 362
    },
    {
      "epoch": 0.23419354838709677,
      "grad_norm": 0.91796875,
      "learning_rate": 9.894642541724487e-06,
      "loss": 1.0057,
      "step": 363
    },
    {
      "epoch": 0.23483870967741935,
      "grad_norm": 0.76953125,
      "learning_rate": 9.89378728774025e-06,
      "loss": 0.727,
      "step": 364
    },
    {
      "epoch": 0.23548387096774193,
      "grad_norm": 0.67578125,
      "learning_rate": 9.892928613715446e-06,
      "loss": 0.8698,
      "step": 365
    },
    {
      "epoch": 0.2361290322580645,
      "grad_norm": 0.61328125,
      "learning_rate": 9.89206652025017e-06,
      "loss": 0.5485,
      "step": 366
    },
    {
      "epoch": 0.2367741935483871,
      "grad_norm": 0.69921875,
      "learning_rate": 9.891201007946894e-06,
      "loss": 0.7156,
      "step": 367
    },
    {
      "epoch": 0.23741935483870968,
      "grad_norm": 0.671875,
      "learning_rate": 9.890332077410483e-06,
      "loss": 0.7159,
      "step": 368
    },
    {
      "epoch": 0.23806451612903226,
      "grad_norm": 0.921875,
      "learning_rate": 9.889459729248197e-06,
      "loss": 1.0613,
      "step": 369
    },
    {
      "epoch": 0.23870967741935484,
      "grad_norm": 0.81640625,
      "learning_rate": 9.888583964069676e-06,
      "loss": 0.9706,
      "step": 370
    },
    {
      "epoch": 0.23935483870967741,
      "grad_norm": 0.89453125,
      "learning_rate": 9.887704782486952e-06,
      "loss": 0.9112,
      "step": 371
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.79296875,
      "learning_rate": 9.886822185114446e-06,
      "loss": 0.7018,
      "step": 372
    },
    {
      "epoch": 0.24064516129032257,
      "grad_norm": 0.609375,
      "learning_rate": 9.885936172568965e-06,
      "loss": 0.8329,
      "step": 373
    },
    {
      "epoch": 0.24129032258064517,
      "grad_norm": 0.83984375,
      "learning_rate": 9.885046745469697e-06,
      "loss": 1.0875,
      "step": 374
    },
    {
      "epoch": 0.24193548387096775,
      "grad_norm": 0.578125,
      "learning_rate": 9.884153904438226e-06,
      "loss": 0.5204,
      "step": 375
    },
    {
      "epoch": 0.24258064516129033,
      "grad_norm": 0.76953125,
      "learning_rate": 9.883257650098515e-06,
      "loss": 0.6414,
      "step": 376
    },
    {
      "epoch": 0.2432258064516129,
      "grad_norm": 0.578125,
      "learning_rate": 9.882357983076916e-06,
      "loss": 0.5844,
      "step": 377
    },
    {
      "epoch": 0.24387096774193548,
      "grad_norm": 0.76953125,
      "learning_rate": 9.881454904002162e-06,
      "loss": 0.8589,
      "step": 378
    },
    {
      "epoch": 0.24451612903225806,
      "grad_norm": 0.890625,
      "learning_rate": 9.880548413505375e-06,
      "loss": 0.9513,
      "step": 379
    },
    {
      "epoch": 0.24516129032258063,
      "grad_norm": 0.5,
      "learning_rate": 9.87963851222006e-06,
      "loss": 0.4981,
      "step": 380
    },
    {
      "epoch": 0.24580645161290324,
      "grad_norm": 0.73828125,
      "learning_rate": 9.878725200782102e-06,
      "loss": 1.0275,
      "step": 381
    },
    {
      "epoch": 0.24645161290322581,
      "grad_norm": 0.67578125,
      "learning_rate": 9.877808479829774e-06,
      "loss": 0.6953,
      "step": 382
    },
    {
      "epoch": 0.2470967741935484,
      "grad_norm": 0.87109375,
      "learning_rate": 9.876888350003727e-06,
      "loss": 0.8993,
      "step": 383
    },
    {
      "epoch": 0.24774193548387097,
      "grad_norm": 1.0546875,
      "learning_rate": 9.875964811947001e-06,
      "loss": 1.4666,
      "step": 384
    },
    {
      "epoch": 0.24838709677419354,
      "grad_norm": 0.421875,
      "learning_rate": 9.875037866305011e-06,
      "loss": 0.5286,
      "step": 385
    },
    {
      "epoch": 0.24903225806451612,
      "grad_norm": 0.58984375,
      "learning_rate": 9.874107513725557e-06,
      "loss": 0.769,
      "step": 386
    },
    {
      "epoch": 0.2496774193548387,
      "grad_norm": 0.86328125,
      "learning_rate": 9.87317375485882e-06,
      "loss": 0.8134,
      "step": 387
    },
    {
      "epoch": 0.2503225806451613,
      "grad_norm": 1.125,
      "learning_rate": 9.87223659035736e-06,
      "loss": 1.1605,
      "step": 388
    },
    {
      "epoch": 0.2509677419354839,
      "grad_norm": 0.50390625,
      "learning_rate": 9.871296020876117e-06,
      "loss": 0.6311,
      "step": 389
    },
    {
      "epoch": 0.25161290322580643,
      "grad_norm": 0.875,
      "learning_rate": 9.870352047072413e-06,
      "loss": 1.0147,
      "step": 390
    },
    {
      "epoch": 0.25225806451612903,
      "grad_norm": 0.72265625,
      "learning_rate": 9.869404669605943e-06,
      "loss": 0.8057,
      "step": 391
    },
    {
      "epoch": 0.25290322580645164,
      "grad_norm": 0.93359375,
      "learning_rate": 9.868453889138793e-06,
      "loss": 1.0644,
      "step": 392
    },
    {
      "epoch": 0.2535483870967742,
      "grad_norm": 1.015625,
      "learning_rate": 9.867499706335414e-06,
      "loss": 0.8525,
      "step": 393
    },
    {
      "epoch": 0.2541935483870968,
      "grad_norm": 0.703125,
      "learning_rate": 9.86654212186264e-06,
      "loss": 0.7042,
      "step": 394
    },
    {
      "epoch": 0.25483870967741934,
      "grad_norm": 0.9453125,
      "learning_rate": 9.865581136389681e-06,
      "loss": 0.8494,
      "step": 395
    },
    {
      "epoch": 0.25548387096774194,
      "grad_norm": 0.8359375,
      "learning_rate": 9.86461675058813e-06,
      "loss": 0.6745,
      "step": 396
    },
    {
      "epoch": 0.2561290322580645,
      "grad_norm": 0.82421875,
      "learning_rate": 9.863648965131947e-06,
      "loss": 0.6843,
      "step": 397
    },
    {
      "epoch": 0.2567741935483871,
      "grad_norm": 0.70703125,
      "learning_rate": 9.862677780697474e-06,
      "loss": 0.7466,
      "step": 398
    },
    {
      "epoch": 0.2574193548387097,
      "grad_norm": 0.7265625,
      "learning_rate": 9.861703197963428e-06,
      "loss": 0.7867,
      "step": 399
    },
    {
      "epoch": 0.25806451612903225,
      "grad_norm": 1.265625,
      "learning_rate": 9.860725217610898e-06,
      "loss": 1.1322,
      "step": 400
    },
    {
      "epoch": 0.25870967741935486,
      "grad_norm": 0.5703125,
      "learning_rate": 9.859743840323348e-06,
      "loss": 0.7167,
      "step": 401
    },
    {
      "epoch": 0.2593548387096774,
      "grad_norm": 1.2265625,
      "learning_rate": 9.858759066786619e-06,
      "loss": 1.4095,
      "step": 402
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.72265625,
      "learning_rate": 9.857770897688921e-06,
      "loss": 0.7457,
      "step": 403
    },
    {
      "epoch": 0.26064516129032256,
      "grad_norm": 0.6640625,
      "learning_rate": 9.856779333720843e-06,
      "loss": 0.7987,
      "step": 404
    },
    {
      "epoch": 0.26129032258064516,
      "grad_norm": 0.65625,
      "learning_rate": 9.855784375575342e-06,
      "loss": 0.8297,
      "step": 405
    },
    {
      "epoch": 0.26193548387096777,
      "grad_norm": 0.7578125,
      "learning_rate": 9.854786023947745e-06,
      "loss": 1.0662,
      "step": 406
    },
    {
      "epoch": 0.2625806451612903,
      "grad_norm": 0.81640625,
      "learning_rate": 9.853784279535757e-06,
      "loss": 0.6612,
      "step": 407
    },
    {
      "epoch": 0.2632258064516129,
      "grad_norm": 0.984375,
      "learning_rate": 9.852779143039449e-06,
      "loss": 1.0887,
      "step": 408
    },
    {
      "epoch": 0.26387096774193547,
      "grad_norm": 0.99609375,
      "learning_rate": 9.851770615161265e-06,
      "loss": 1.0999,
      "step": 409
    },
    {
      "epoch": 0.2645161290322581,
      "grad_norm": 0.82421875,
      "learning_rate": 9.850758696606017e-06,
      "loss": 0.8194,
      "step": 410
    },
    {
      "epoch": 0.2651612903225806,
      "grad_norm": 1.0546875,
      "learning_rate": 9.84974338808089e-06,
      "loss": 0.8361,
      "step": 411
    },
    {
      "epoch": 0.2658064516129032,
      "grad_norm": 1.015625,
      "learning_rate": 9.848724690295433e-06,
      "loss": 1.1223,
      "step": 412
    },
    {
      "epoch": 0.26645161290322583,
      "grad_norm": 0.8359375,
      "learning_rate": 9.84770260396157e-06,
      "loss": 0.9891,
      "step": 413
    },
    {
      "epoch": 0.2670967741935484,
      "grad_norm": 0.83203125,
      "learning_rate": 9.846677129793586e-06,
      "loss": 0.7592,
      "step": 414
    },
    {
      "epoch": 0.267741935483871,
      "grad_norm": 0.609375,
      "learning_rate": 9.84564826850814e-06,
      "loss": 0.7124,
      "step": 415
    },
    {
      "epoch": 0.26838709677419353,
      "grad_norm": 1.0,
      "learning_rate": 9.844616020824255e-06,
      "loss": 1.0169,
      "step": 416
    },
    {
      "epoch": 0.26903225806451614,
      "grad_norm": 0.5859375,
      "learning_rate": 9.843580387463321e-06,
      "loss": 0.7051,
      "step": 417
    },
    {
      "epoch": 0.2696774193548387,
      "grad_norm": 0.9296875,
      "learning_rate": 9.842541369149093e-06,
      "loss": 0.9436,
      "step": 418
    },
    {
      "epoch": 0.2703225806451613,
      "grad_norm": 0.54296875,
      "learning_rate": 9.841498966607692e-06,
      "loss": 0.5291,
      "step": 419
    },
    {
      "epoch": 0.2709677419354839,
      "grad_norm": 0.67578125,
      "learning_rate": 9.840453180567607e-06,
      "loss": 0.6977,
      "step": 420
    },
    {
      "epoch": 0.27161290322580645,
      "grad_norm": 0.84765625,
      "learning_rate": 9.839404011759687e-06,
      "loss": 0.8942,
      "step": 421
    },
    {
      "epoch": 0.27225806451612905,
      "grad_norm": 0.66796875,
      "learning_rate": 9.838351460917149e-06,
      "loss": 0.8833,
      "step": 422
    },
    {
      "epoch": 0.2729032258064516,
      "grad_norm": 0.53515625,
      "learning_rate": 9.83729552877557e-06,
      "loss": 0.634,
      "step": 423
    },
    {
      "epoch": 0.2735483870967742,
      "grad_norm": 0.77734375,
      "learning_rate": 9.836236216072895e-06,
      "loss": 0.9677,
      "step": 424
    },
    {
      "epoch": 0.27419354838709675,
      "grad_norm": 0.72265625,
      "learning_rate": 9.835173523549426e-06,
      "loss": 0.7028,
      "step": 425
    },
    {
      "epoch": 0.27483870967741936,
      "grad_norm": 0.486328125,
      "learning_rate": 9.83410745194783e-06,
      "loss": 0.5068,
      "step": 426
    },
    {
      "epoch": 0.27548387096774196,
      "grad_norm": 0.7109375,
      "learning_rate": 9.833038002013134e-06,
      "loss": 0.9544,
      "step": 427
    },
    {
      "epoch": 0.2761290322580645,
      "grad_norm": 0.8828125,
      "learning_rate": 9.831965174492729e-06,
      "loss": 0.858,
      "step": 428
    },
    {
      "epoch": 0.2767741935483871,
      "grad_norm": 0.921875,
      "learning_rate": 9.830888970136364e-06,
      "loss": 0.7994,
      "step": 429
    },
    {
      "epoch": 0.27741935483870966,
      "grad_norm": 1.0234375,
      "learning_rate": 9.829809389696145e-06,
      "loss": 1.028,
      "step": 430
    },
    {
      "epoch": 0.27806451612903227,
      "grad_norm": 0.640625,
      "learning_rate": 9.828726433926545e-06,
      "loss": 0.6908,
      "step": 431
    },
    {
      "epoch": 0.2787096774193548,
      "grad_norm": 0.55078125,
      "learning_rate": 9.827640103584388e-06,
      "loss": 0.673,
      "step": 432
    },
    {
      "epoch": 0.2793548387096774,
      "grad_norm": 0.703125,
      "learning_rate": 9.826550399428862e-06,
      "loss": 0.7268,
      "step": 433
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5703125,
      "learning_rate": 9.82545732222151e-06,
      "loss": 0.6882,
      "step": 434
    },
    {
      "epoch": 0.2806451612903226,
      "grad_norm": 0.59375,
      "learning_rate": 9.824360872726232e-06,
      "loss": 0.8119,
      "step": 435
    },
    {
      "epoch": 0.2812903225806452,
      "grad_norm": 0.86328125,
      "learning_rate": 9.823261051709287e-06,
      "loss": 0.8884,
      "step": 436
    },
    {
      "epoch": 0.28193548387096773,
      "grad_norm": 0.49609375,
      "learning_rate": 9.822157859939288e-06,
      "loss": 0.6587,
      "step": 437
    },
    {
      "epoch": 0.28258064516129033,
      "grad_norm": 0.7734375,
      "learning_rate": 9.821051298187204e-06,
      "loss": 0.9843,
      "step": 438
    },
    {
      "epoch": 0.2832258064516129,
      "grad_norm": 0.89453125,
      "learning_rate": 9.819941367226362e-06,
      "loss": 1.1641,
      "step": 439
    },
    {
      "epoch": 0.2838709677419355,
      "grad_norm": 1.203125,
      "learning_rate": 9.818828067832438e-06,
      "loss": 1.1737,
      "step": 440
    },
    {
      "epoch": 0.2845161290322581,
      "grad_norm": 0.8671875,
      "learning_rate": 9.817711400783466e-06,
      "loss": 1.0663,
      "step": 441
    },
    {
      "epoch": 0.28516129032258064,
      "grad_norm": 0.7734375,
      "learning_rate": 9.816591366859833e-06,
      "loss": 1.0095,
      "step": 442
    },
    {
      "epoch": 0.28580645161290325,
      "grad_norm": 1.265625,
      "learning_rate": 9.815467966844281e-06,
      "loss": 1.3254,
      "step": 443
    },
    {
      "epoch": 0.2864516129032258,
      "grad_norm": 0.8828125,
      "learning_rate": 9.8143412015219e-06,
      "loss": 0.8215,
      "step": 444
    },
    {
      "epoch": 0.2870967741935484,
      "grad_norm": 0.75390625,
      "learning_rate": 9.813211071680134e-06,
      "loss": 0.8444,
      "step": 445
    },
    {
      "epoch": 0.28774193548387095,
      "grad_norm": 0.57421875,
      "learning_rate": 9.812077578108776e-06,
      "loss": 0.7051,
      "step": 446
    },
    {
      "epoch": 0.28838709677419355,
      "grad_norm": 0.99609375,
      "learning_rate": 9.810940721599976e-06,
      "loss": 0.9381,
      "step": 447
    },
    {
      "epoch": 0.28903225806451616,
      "grad_norm": 0.8046875,
      "learning_rate": 9.809800502948227e-06,
      "loss": 0.7525,
      "step": 448
    },
    {
      "epoch": 0.2896774193548387,
      "grad_norm": 0.65234375,
      "learning_rate": 9.808656922950379e-06,
      "loss": 0.6358,
      "step": 449
    },
    {
      "epoch": 0.2903225806451613,
      "grad_norm": 0.671875,
      "learning_rate": 9.807509982405623e-06,
      "loss": 0.6657,
      "step": 450
    },
    {
      "epoch": 0.29096774193548386,
      "grad_norm": 0.86328125,
      "learning_rate": 9.806359682115503e-06,
      "loss": 0.9391,
      "step": 451
    },
    {
      "epoch": 0.29161290322580646,
      "grad_norm": 1.03125,
      "learning_rate": 9.80520602288391e-06,
      "loss": 0.8095,
      "step": 452
    },
    {
      "epoch": 0.292258064516129,
      "grad_norm": 0.80078125,
      "learning_rate": 9.804049005517087e-06,
      "loss": 0.7891,
      "step": 453
    },
    {
      "epoch": 0.2929032258064516,
      "grad_norm": 0.75390625,
      "learning_rate": 9.802888630823614e-06,
      "loss": 0.8245,
      "step": 454
    },
    {
      "epoch": 0.29354838709677417,
      "grad_norm": 0.75390625,
      "learning_rate": 9.801724899614427e-06,
      "loss": 0.9632,
      "step": 455
    },
    {
      "epoch": 0.29419354838709677,
      "grad_norm": 0.63671875,
      "learning_rate": 9.800557812702802e-06,
      "loss": 0.7543,
      "step": 456
    },
    {
      "epoch": 0.2948387096774194,
      "grad_norm": 0.67578125,
      "learning_rate": 9.799387370904361e-06,
      "loss": 0.7044,
      "step": 457
    },
    {
      "epoch": 0.2954838709677419,
      "grad_norm": 1.1875,
      "learning_rate": 9.798213575037074e-06,
      "loss": 1.112,
      "step": 458
    },
    {
      "epoch": 0.29612903225806453,
      "grad_norm": 0.7578125,
      "learning_rate": 9.79703642592125e-06,
      "loss": 0.731,
      "step": 459
    },
    {
      "epoch": 0.2967741935483871,
      "grad_norm": 0.490234375,
      "learning_rate": 9.795855924379545e-06,
      "loss": 0.5741,
      "step": 460
    },
    {
      "epoch": 0.2974193548387097,
      "grad_norm": 0.94140625,
      "learning_rate": 9.794672071236956e-06,
      "loss": 0.8812,
      "step": 461
    },
    {
      "epoch": 0.29806451612903223,
      "grad_norm": 0.91796875,
      "learning_rate": 9.793484867320825e-06,
      "loss": 0.9889,
      "step": 462
    },
    {
      "epoch": 0.29870967741935484,
      "grad_norm": 0.90234375,
      "learning_rate": 9.79229431346083e-06,
      "loss": 1.2495,
      "step": 463
    },
    {
      "epoch": 0.29935483870967744,
      "grad_norm": 0.71875,
      "learning_rate": 9.791100410488997e-06,
      "loss": 0.7376,
      "step": 464
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6015625,
      "learning_rate": 9.789903159239686e-06,
      "loss": 0.6351,
      "step": 465
    },
    {
      "epoch": 0.3006451612903226,
      "grad_norm": 0.67578125,
      "learning_rate": 9.788702560549603e-06,
      "loss": 0.7923,
      "step": 466
    },
    {
      "epoch": 0.30129032258064514,
      "grad_norm": 0.8984375,
      "learning_rate": 9.78749861525779e-06,
      "loss": 1.0203,
      "step": 467
    },
    {
      "epoch": 0.30193548387096775,
      "grad_norm": 0.97265625,
      "learning_rate": 9.78629132420563e-06,
      "loss": 0.8065,
      "step": 468
    },
    {
      "epoch": 0.3025806451612903,
      "grad_norm": 1.234375,
      "learning_rate": 9.78508068823684e-06,
      "loss": 0.5164,
      "step": 469
    },
    {
      "epoch": 0.3032258064516129,
      "grad_norm": 1.1953125,
      "learning_rate": 9.783866708197478e-06,
      "loss": 1.1161,
      "step": 470
    },
    {
      "epoch": 0.3038709677419355,
      "grad_norm": 1.046875,
      "learning_rate": 9.78264938493594e-06,
      "loss": 1.0277,
      "step": 471
    },
    {
      "epoch": 0.30451612903225805,
      "grad_norm": 0.76953125,
      "learning_rate": 9.781428719302955e-06,
      "loss": 0.9157,
      "step": 472
    },
    {
      "epoch": 0.30516129032258066,
      "grad_norm": 0.55859375,
      "learning_rate": 9.780204712151588e-06,
      "loss": 0.7513,
      "step": 473
    },
    {
      "epoch": 0.3058064516129032,
      "grad_norm": 0.984375,
      "learning_rate": 9.778977364337244e-06,
      "loss": 0.9225,
      "step": 474
    },
    {
      "epoch": 0.3064516129032258,
      "grad_norm": 0.70703125,
      "learning_rate": 9.777746676717656e-06,
      "loss": 0.7649,
      "step": 475
    },
    {
      "epoch": 0.30709677419354836,
      "grad_norm": 0.7578125,
      "learning_rate": 9.776512650152896e-06,
      "loss": 0.6668,
      "step": 476
    },
    {
      "epoch": 0.30774193548387097,
      "grad_norm": 0.69140625,
      "learning_rate": 9.77527528550537e-06,
      "loss": 0.6749,
      "step": 477
    },
    {
      "epoch": 0.30838709677419357,
      "grad_norm": 0.89453125,
      "learning_rate": 9.77403458363981e-06,
      "loss": 1.0344,
      "step": 478
    },
    {
      "epoch": 0.3090322580645161,
      "grad_norm": 0.6328125,
      "learning_rate": 9.772790545423292e-06,
      "loss": 0.7023,
      "step": 479
    },
    {
      "epoch": 0.3096774193548387,
      "grad_norm": 0.96875,
      "learning_rate": 9.771543171725209e-06,
      "loss": 0.7033,
      "step": 480
    },
    {
      "epoch": 0.3103225806451613,
      "grad_norm": 1.0859375,
      "learning_rate": 9.770292463417296e-06,
      "loss": 1.0342,
      "step": 481
    },
    {
      "epoch": 0.3109677419354839,
      "grad_norm": 0.51171875,
      "learning_rate": 9.769038421373614e-06,
      "loss": 0.5641,
      "step": 482
    },
    {
      "epoch": 0.3116129032258064,
      "grad_norm": 0.83984375,
      "learning_rate": 9.767781046470556e-06,
      "loss": 0.7493,
      "step": 483
    },
    {
      "epoch": 0.31225806451612903,
      "grad_norm": 1.09375,
      "learning_rate": 9.766520339586842e-06,
      "loss": 1.0693,
      "step": 484
    },
    {
      "epoch": 0.31290322580645163,
      "grad_norm": 1.0625,
      "learning_rate": 9.76525630160352e-06,
      "loss": 1.3271,
      "step": 485
    },
    {
      "epoch": 0.3135483870967742,
      "grad_norm": 0.69140625,
      "learning_rate": 9.763988933403972e-06,
      "loss": 0.8939,
      "step": 486
    },
    {
      "epoch": 0.3141935483870968,
      "grad_norm": 0.96484375,
      "learning_rate": 9.762718235873898e-06,
      "loss": 1.2009,
      "step": 487
    },
    {
      "epoch": 0.31483870967741934,
      "grad_norm": 0.66015625,
      "learning_rate": 9.761444209901332e-06,
      "loss": 0.6916,
      "step": 488
    },
    {
      "epoch": 0.31548387096774194,
      "grad_norm": 0.72265625,
      "learning_rate": 9.76016685637663e-06,
      "loss": 0.8845,
      "step": 489
    },
    {
      "epoch": 0.3161290322580645,
      "grad_norm": 1.1484375,
      "learning_rate": 9.758886176192476e-06,
      "loss": 0.6876,
      "step": 490
    },
    {
      "epoch": 0.3167741935483871,
      "grad_norm": 0.9921875,
      "learning_rate": 9.757602170243878e-06,
      "loss": 0.9185,
      "step": 491
    },
    {
      "epoch": 0.3174193548387097,
      "grad_norm": 0.7109375,
      "learning_rate": 9.756314839428167e-06,
      "loss": 0.6954,
      "step": 492
    },
    {
      "epoch": 0.31806451612903225,
      "grad_norm": 0.9609375,
      "learning_rate": 9.755024184645001e-06,
      "loss": 0.7695,
      "step": 493
    },
    {
      "epoch": 0.31870967741935485,
      "grad_norm": 0.94921875,
      "learning_rate": 9.753730206796355e-06,
      "loss": 0.8329,
      "step": 494
    },
    {
      "epoch": 0.3193548387096774,
      "grad_norm": 0.87890625,
      "learning_rate": 9.752432906786535e-06,
      "loss": 0.8037,
      "step": 495
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.85546875,
      "learning_rate": 9.751132285522158e-06,
      "loss": 0.8784,
      "step": 496
    },
    {
      "epoch": 0.32064516129032256,
      "grad_norm": 1.140625,
      "learning_rate": 9.749828343912173e-06,
      "loss": 1.1982,
      "step": 497
    },
    {
      "epoch": 0.32129032258064516,
      "grad_norm": 0.81640625,
      "learning_rate": 9.74852108286784e-06,
      "loss": 1.0143,
      "step": 498
    },
    {
      "epoch": 0.32193548387096776,
      "grad_norm": 1.0078125,
      "learning_rate": 9.747210503302747e-06,
      "loss": 0.9845,
      "step": 499
    },
    {
      "epoch": 0.3225806451612903,
      "grad_norm": 0.9921875,
      "learning_rate": 9.745896606132793e-06,
      "loss": 1.2303,
      "step": 500
    },
    {
      "epoch": 0.3232258064516129,
      "grad_norm": 0.53515625,
      "learning_rate": 9.744579392276202e-06,
      "loss": 0.6269,
      "step": 501
    },
    {
      "epoch": 0.32387096774193547,
      "grad_norm": 0.8125,
      "learning_rate": 9.743258862653513e-06,
      "loss": 0.7666,
      "step": 502
    },
    {
      "epoch": 0.32451612903225807,
      "grad_norm": 0.7734375,
      "learning_rate": 9.741935018187584e-06,
      "loss": 0.8179,
      "step": 503
    },
    {
      "epoch": 0.3251612903225806,
      "grad_norm": 0.640625,
      "learning_rate": 9.740607859803589e-06,
      "loss": 0.7559,
      "step": 504
    },
    {
      "epoch": 0.3258064516129032,
      "grad_norm": 0.6171875,
      "learning_rate": 9.739277388429012e-06,
      "loss": 0.6907,
      "step": 505
    },
    {
      "epoch": 0.32645161290322583,
      "grad_norm": 0.87109375,
      "learning_rate": 9.737943604993663e-06,
      "loss": 0.9202,
      "step": 506
    },
    {
      "epoch": 0.3270967741935484,
      "grad_norm": 0.64453125,
      "learning_rate": 9.736606510429661e-06,
      "loss": 0.4933,
      "step": 507
    },
    {
      "epoch": 0.327741935483871,
      "grad_norm": 0.81640625,
      "learning_rate": 9.735266105671436e-06,
      "loss": 0.733,
      "step": 508
    },
    {
      "epoch": 0.32838709677419353,
      "grad_norm": 0.86328125,
      "learning_rate": 9.73392239165574e-06,
      "loss": 0.8288,
      "step": 509
    },
    {
      "epoch": 0.32903225806451614,
      "grad_norm": 0.8125,
      "learning_rate": 9.732575369321626e-06,
      "loss": 0.8434,
      "step": 510
    },
    {
      "epoch": 0.3296774193548387,
      "grad_norm": 0.546875,
      "learning_rate": 9.731225039610468e-06,
      "loss": 0.6113,
      "step": 511
    },
    {
      "epoch": 0.3303225806451613,
      "grad_norm": 0.734375,
      "learning_rate": 9.729871403465949e-06,
      "loss": 0.7875,
      "step": 512
    },
    {
      "epoch": 0.3309677419354839,
      "grad_norm": 0.87109375,
      "learning_rate": 9.728514461834064e-06,
      "loss": 0.7513,
      "step": 513
    },
    {
      "epoch": 0.33161290322580644,
      "grad_norm": 0.66015625,
      "learning_rate": 9.727154215663113e-06,
      "loss": 0.8888,
      "step": 514
    },
    {
      "epoch": 0.33225806451612905,
      "grad_norm": 0.65234375,
      "learning_rate": 9.725790665903712e-06,
      "loss": 0.7779,
      "step": 515
    },
    {
      "epoch": 0.3329032258064516,
      "grad_norm": 1.03125,
      "learning_rate": 9.724423813508778e-06,
      "loss": 0.9839,
      "step": 516
    },
    {
      "epoch": 0.3335483870967742,
      "grad_norm": 0.53125,
      "learning_rate": 9.723053659433549e-06,
      "loss": 0.5623,
      "step": 517
    },
    {
      "epoch": 0.33419354838709675,
      "grad_norm": 0.72265625,
      "learning_rate": 9.721680204635553e-06,
      "loss": 0.6681,
      "step": 518
    },
    {
      "epoch": 0.33483870967741935,
      "grad_norm": 0.99609375,
      "learning_rate": 9.720303450074637e-06,
      "loss": 0.9068,
      "step": 519
    },
    {
      "epoch": 0.33548387096774196,
      "grad_norm": 0.6953125,
      "learning_rate": 9.718923396712954e-06,
      "loss": 0.7391,
      "step": 520
    },
    {
      "epoch": 0.3361290322580645,
      "grad_norm": 0.77734375,
      "learning_rate": 9.717540045514955e-06,
      "loss": 0.9019,
      "step": 521
    },
    {
      "epoch": 0.3367741935483871,
      "grad_norm": 1.125,
      "learning_rate": 9.716153397447403e-06,
      "loss": 0.8504,
      "step": 522
    },
    {
      "epoch": 0.33741935483870966,
      "grad_norm": 0.890625,
      "learning_rate": 9.714763453479359e-06,
      "loss": 1.0666,
      "step": 523
    },
    {
      "epoch": 0.33806451612903227,
      "grad_norm": 0.765625,
      "learning_rate": 9.713370214582191e-06,
      "loss": 0.7406,
      "step": 524
    },
    {
      "epoch": 0.3387096774193548,
      "grad_norm": 0.7265625,
      "learning_rate": 9.711973681729572e-06,
      "loss": 0.5598,
      "step": 525
    },
    {
      "epoch": 0.3393548387096774,
      "grad_norm": 0.6875,
      "learning_rate": 9.71057385589747e-06,
      "loss": 0.7332,
      "step": 526
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.82421875,
      "learning_rate": 9.709170738064159e-06,
      "loss": 0.9571,
      "step": 527
    },
    {
      "epoch": 0.3406451612903226,
      "grad_norm": 0.83984375,
      "learning_rate": 9.707764329210214e-06,
      "loss": 0.8353,
      "step": 528
    },
    {
      "epoch": 0.3412903225806452,
      "grad_norm": 0.7578125,
      "learning_rate": 9.706354630318509e-06,
      "loss": 0.9837,
      "step": 529
    },
    {
      "epoch": 0.3419354838709677,
      "grad_norm": 0.76171875,
      "learning_rate": 9.704941642374217e-06,
      "loss": 0.7696,
      "step": 530
    },
    {
      "epoch": 0.34258064516129033,
      "grad_norm": 0.8828125,
      "learning_rate": 9.703525366364807e-06,
      "loss": 1.1388,
      "step": 531
    },
    {
      "epoch": 0.3432258064516129,
      "grad_norm": 1.2734375,
      "learning_rate": 9.702105803280052e-06,
      "loss": 0.7919,
      "step": 532
    },
    {
      "epoch": 0.3438709677419355,
      "grad_norm": 0.796875,
      "learning_rate": 9.700682954112017e-06,
      "loss": 0.9766,
      "step": 533
    },
    {
      "epoch": 0.3445161290322581,
      "grad_norm": 0.53125,
      "learning_rate": 9.699256819855065e-06,
      "loss": 0.5244,
      "step": 534
    },
    {
      "epoch": 0.34516129032258064,
      "grad_norm": 0.77734375,
      "learning_rate": 9.697827401505856e-06,
      "loss": 0.865,
      "step": 535
    },
    {
      "epoch": 0.34580645161290324,
      "grad_norm": 1.3515625,
      "learning_rate": 9.696394700063344e-06,
      "loss": 0.9664,
      "step": 536
    },
    {
      "epoch": 0.3464516129032258,
      "grad_norm": 0.796875,
      "learning_rate": 9.694958716528775e-06,
      "loss": 0.6295,
      "step": 537
    },
    {
      "epoch": 0.3470967741935484,
      "grad_norm": 0.61328125,
      "learning_rate": 9.693519451905694e-06,
      "loss": 0.7024,
      "step": 538
    },
    {
      "epoch": 0.34774193548387095,
      "grad_norm": 0.91796875,
      "learning_rate": 9.692076907199935e-06,
      "loss": 0.8264,
      "step": 539
    },
    {
      "epoch": 0.34838709677419355,
      "grad_norm": 0.84375,
      "learning_rate": 9.690631083419626e-06,
      "loss": 1.017,
      "step": 540
    },
    {
      "epoch": 0.34903225806451615,
      "grad_norm": 0.86328125,
      "learning_rate": 9.689181981575186e-06,
      "loss": 0.9667,
      "step": 541
    },
    {
      "epoch": 0.3496774193548387,
      "grad_norm": 0.609375,
      "learning_rate": 9.687729602679326e-06,
      "loss": 0.7558,
      "step": 542
    },
    {
      "epoch": 0.3503225806451613,
      "grad_norm": 0.86328125,
      "learning_rate": 9.686273947747044e-06,
      "loss": 1.1534,
      "step": 543
    },
    {
      "epoch": 0.35096774193548386,
      "grad_norm": 0.73828125,
      "learning_rate": 9.68481501779563e-06,
      "loss": 0.9574,
      "step": 544
    },
    {
      "epoch": 0.35161290322580646,
      "grad_norm": 0.9609375,
      "learning_rate": 9.683352813844663e-06,
      "loss": 0.8929,
      "step": 545
    },
    {
      "epoch": 0.352258064516129,
      "grad_norm": 0.890625,
      "learning_rate": 9.681887336916009e-06,
      "loss": 0.957,
      "step": 546
    },
    {
      "epoch": 0.3529032258064516,
      "grad_norm": 0.5859375,
      "learning_rate": 9.680418588033822e-06,
      "loss": 0.6895,
      "step": 547
    },
    {
      "epoch": 0.3535483870967742,
      "grad_norm": 0.70703125,
      "learning_rate": 9.678946568224543e-06,
      "loss": 0.9499,
      "step": 548
    },
    {
      "epoch": 0.35419354838709677,
      "grad_norm": 0.921875,
      "learning_rate": 9.677471278516897e-06,
      "loss": 0.731,
      "step": 549
    },
    {
      "epoch": 0.3548387096774194,
      "grad_norm": 0.91796875,
      "learning_rate": 9.675992719941896e-06,
      "loss": 0.7558,
      "step": 550
    },
    {
      "epoch": 0.3548387096774194,
      "eval_loss": 0.954688549041748,
      "eval_model_preparation_time": 0.0144,
      "eval_runtime": 83.1506,
      "eval_samples_per_second": 3.584,
      "eval_steps_per_second": 3.584,
      "step": 550
    },
    {
      "epoch": 0.3554838709677419,
      "grad_norm": 1.3046875,
      "learning_rate": 9.674510893532838e-06,
      "loss": 1.01,
      "step": 551
    },
    {
      "epoch": 0.3561290322580645,
      "grad_norm": 0.53125,
      "learning_rate": 9.673025800325298e-06,
      "loss": 0.6393,
      "step": 552
    },
    {
      "epoch": 0.3567741935483871,
      "grad_norm": 0.8984375,
      "learning_rate": 9.671537441357142e-06,
      "loss": 0.819,
      "step": 553
    },
    {
      "epoch": 0.3574193548387097,
      "grad_norm": 0.6953125,
      "learning_rate": 9.670045817668514e-06,
      "loss": 0.5686,
      "step": 554
    },
    {
      "epoch": 0.3580645161290323,
      "grad_norm": 0.9765625,
      "learning_rate": 9.668550930301842e-06,
      "loss": 1.1996,
      "step": 555
    },
    {
      "epoch": 0.35870967741935483,
      "grad_norm": 0.515625,
      "learning_rate": 9.66705278030183e-06,
      "loss": 0.6023,
      "step": 556
    },
    {
      "epoch": 0.35935483870967744,
      "grad_norm": 0.69140625,
      "learning_rate": 9.665551368715468e-06,
      "loss": 0.6937,
      "step": 557
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1171875,
      "learning_rate": 9.664046696592022e-06,
      "loss": 1.1013,
      "step": 558
    },
    {
      "epoch": 0.3606451612903226,
      "grad_norm": 0.734375,
      "learning_rate": 9.662538764983037e-06,
      "loss": 0.6175,
      "step": 559
    },
    {
      "epoch": 0.36129032258064514,
      "grad_norm": 0.59375,
      "learning_rate": 9.66102757494234e-06,
      "loss": 0.5753,
      "step": 560
    },
    {
      "epoch": 0.36193548387096774,
      "grad_norm": 1.0546875,
      "learning_rate": 9.659513127526026e-06,
      "loss": 0.8438,
      "step": 561
    },
    {
      "epoch": 0.36258064516129035,
      "grad_norm": 0.68359375,
      "learning_rate": 9.657995423792475e-06,
      "loss": 0.6,
      "step": 562
    },
    {
      "epoch": 0.3632258064516129,
      "grad_norm": 0.8125,
      "learning_rate": 9.656474464802339e-06,
      "loss": 0.6481,
      "step": 563
    },
    {
      "epoch": 0.3638709677419355,
      "grad_norm": 0.89453125,
      "learning_rate": 9.654950251618545e-06,
      "loss": 1.1171,
      "step": 564
    },
    {
      "epoch": 0.36451612903225805,
      "grad_norm": 1.015625,
      "learning_rate": 9.653422785306297e-06,
      "loss": 0.8167,
      "step": 565
    },
    {
      "epoch": 0.36516129032258066,
      "grad_norm": 0.703125,
      "learning_rate": 9.651892066933067e-06,
      "loss": 0.6971,
      "step": 566
    },
    {
      "epoch": 0.3658064516129032,
      "grad_norm": 0.62890625,
      "learning_rate": 9.650358097568605e-06,
      "loss": 0.6525,
      "step": 567
    },
    {
      "epoch": 0.3664516129032258,
      "grad_norm": 0.85546875,
      "learning_rate": 9.64882087828493e-06,
      "loss": 1.0335,
      "step": 568
    },
    {
      "epoch": 0.3670967741935484,
      "grad_norm": 0.62109375,
      "learning_rate": 9.647280410156335e-06,
      "loss": 0.5741,
      "step": 569
    },
    {
      "epoch": 0.36774193548387096,
      "grad_norm": 0.64453125,
      "learning_rate": 9.64573669425938e-06,
      "loss": 0.7036,
      "step": 570
    },
    {
      "epoch": 0.36838709677419357,
      "grad_norm": 0.9296875,
      "learning_rate": 9.644189731672896e-06,
      "loss": 0.715,
      "step": 571
    },
    {
      "epoch": 0.3690322580645161,
      "grad_norm": 0.85546875,
      "learning_rate": 9.642639523477984e-06,
      "loss": 0.8444,
      "step": 572
    },
    {
      "epoch": 0.3696774193548387,
      "grad_norm": 0.68359375,
      "learning_rate": 9.641086070758012e-06,
      "loss": 0.773,
      "step": 573
    },
    {
      "epoch": 0.37032258064516127,
      "grad_norm": 1.125,
      "learning_rate": 9.639529374598616e-06,
      "loss": 1.0328,
      "step": 574
    },
    {
      "epoch": 0.3709677419354839,
      "grad_norm": 0.71875,
      "learning_rate": 9.637969436087698e-06,
      "loss": 0.8844,
      "step": 575
    },
    {
      "epoch": 0.3716129032258065,
      "grad_norm": 0.859375,
      "learning_rate": 9.636406256315429e-06,
      "loss": 0.8135,
      "step": 576
    },
    {
      "epoch": 0.37225806451612903,
      "grad_norm": 0.8671875,
      "learning_rate": 9.634839836374242e-06,
      "loss": 0.9956,
      "step": 577
    },
    {
      "epoch": 0.37290322580645163,
      "grad_norm": 0.9140625,
      "learning_rate": 9.633270177358832e-06,
      "loss": 0.9346,
      "step": 578
    },
    {
      "epoch": 0.3735483870967742,
      "grad_norm": 0.765625,
      "learning_rate": 9.631697280366164e-06,
      "loss": 0.7581,
      "step": 579
    },
    {
      "epoch": 0.3741935483870968,
      "grad_norm": 0.76171875,
      "learning_rate": 9.630121146495462e-06,
      "loss": 0.8296,
      "step": 580
    },
    {
      "epoch": 0.37483870967741933,
      "grad_norm": 0.7890625,
      "learning_rate": 9.628541776848212e-06,
      "loss": 0.8996,
      "step": 581
    },
    {
      "epoch": 0.37548387096774194,
      "grad_norm": 0.7578125,
      "learning_rate": 9.626959172528162e-06,
      "loss": 0.7392,
      "step": 582
    },
    {
      "epoch": 0.3761290322580645,
      "grad_norm": 1.234375,
      "learning_rate": 9.625373334641322e-06,
      "loss": 1.2738,
      "step": 583
    },
    {
      "epoch": 0.3767741935483871,
      "grad_norm": 0.625,
      "learning_rate": 9.623784264295959e-06,
      "loss": 0.6079,
      "step": 584
    },
    {
      "epoch": 0.3774193548387097,
      "grad_norm": 1.0390625,
      "learning_rate": 9.6221919626026e-06,
      "loss": 1.1461,
      "step": 585
    },
    {
      "epoch": 0.37806451612903225,
      "grad_norm": 0.6953125,
      "learning_rate": 9.620596430674034e-06,
      "loss": 0.7666,
      "step": 586
    },
    {
      "epoch": 0.37870967741935485,
      "grad_norm": 0.80078125,
      "learning_rate": 9.618997669625299e-06,
      "loss": 0.7128,
      "step": 587
    },
    {
      "epoch": 0.3793548387096774,
      "grad_norm": 0.91015625,
      "learning_rate": 9.617395680573699e-06,
      "loss": 0.8776,
      "step": 588
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2734375,
      "learning_rate": 9.615790464638788e-06,
      "loss": 0.9648,
      "step": 589
    },
    {
      "epoch": 0.38064516129032255,
      "grad_norm": 0.91015625,
      "learning_rate": 9.614182022942378e-06,
      "loss": 0.9646,
      "step": 590
    },
    {
      "epoch": 0.38129032258064516,
      "grad_norm": 0.60546875,
      "learning_rate": 9.612570356608531e-06,
      "loss": 0.5662,
      "step": 591
    },
    {
      "epoch": 0.38193548387096776,
      "grad_norm": 0.796875,
      "learning_rate": 9.610955466763569e-06,
      "loss": 0.929,
      "step": 592
    },
    {
      "epoch": 0.3825806451612903,
      "grad_norm": 0.90625,
      "learning_rate": 9.609337354536063e-06,
      "loss": 1.0623,
      "step": 593
    },
    {
      "epoch": 0.3832258064516129,
      "grad_norm": 0.91015625,
      "learning_rate": 9.607716021056836e-06,
      "loss": 0.8528,
      "step": 594
    },
    {
      "epoch": 0.38387096774193546,
      "grad_norm": 0.70703125,
      "learning_rate": 9.606091467458962e-06,
      "loss": 0.6406,
      "step": 595
    },
    {
      "epoch": 0.38451612903225807,
      "grad_norm": 0.87890625,
      "learning_rate": 9.604463694877767e-06,
      "loss": 0.9046,
      "step": 596
    },
    {
      "epoch": 0.3851612903225806,
      "grad_norm": 1.09375,
      "learning_rate": 9.602832704450825e-06,
      "loss": 1.2178,
      "step": 597
    },
    {
      "epoch": 0.3858064516129032,
      "grad_norm": 0.87890625,
      "learning_rate": 9.601198497317961e-06,
      "loss": 1.0013,
      "step": 598
    },
    {
      "epoch": 0.3864516129032258,
      "grad_norm": 0.91796875,
      "learning_rate": 9.599561074621243e-06,
      "loss": 0.9096,
      "step": 599
    },
    {
      "epoch": 0.3870967741935484,
      "grad_norm": 0.828125,
      "learning_rate": 9.597920437504992e-06,
      "loss": 0.9402,
      "step": 600
    },
    {
      "epoch": 0.387741935483871,
      "grad_norm": 0.74609375,
      "learning_rate": 9.596276587115774e-06,
      "loss": 0.8157,
      "step": 601
    },
    {
      "epoch": 0.38838709677419353,
      "grad_norm": 0.67578125,
      "learning_rate": 9.594629524602398e-06,
      "loss": 0.7493,
      "step": 602
    },
    {
      "epoch": 0.38903225806451613,
      "grad_norm": 0.73828125,
      "learning_rate": 9.59297925111592e-06,
      "loss": 0.6688,
      "step": 603
    },
    {
      "epoch": 0.3896774193548387,
      "grad_norm": 0.9609375,
      "learning_rate": 9.591325767809638e-06,
      "loss": 0.8599,
      "step": 604
    },
    {
      "epoch": 0.3903225806451613,
      "grad_norm": 0.76171875,
      "learning_rate": 9.589669075839096e-06,
      "loss": 0.8065,
      "step": 605
    },
    {
      "epoch": 0.3909677419354839,
      "grad_norm": 1.015625,
      "learning_rate": 9.588009176362082e-06,
      "loss": 0.9435,
      "step": 606
    },
    {
      "epoch": 0.39161290322580644,
      "grad_norm": 0.93359375,
      "learning_rate": 9.586346070538618e-06,
      "loss": 1.0443,
      "step": 607
    },
    {
      "epoch": 0.39225806451612905,
      "grad_norm": 0.75,
      "learning_rate": 9.584679759530973e-06,
      "loss": 0.8195,
      "step": 608
    },
    {
      "epoch": 0.3929032258064516,
      "grad_norm": 0.66015625,
      "learning_rate": 9.583010244503655e-06,
      "loss": 0.4538,
      "step": 609
    },
    {
      "epoch": 0.3935483870967742,
      "grad_norm": 0.74609375,
      "learning_rate": 9.58133752662341e-06,
      "loss": 0.8552,
      "step": 610
    },
    {
      "epoch": 0.39419354838709675,
      "grad_norm": 1.3125,
      "learning_rate": 9.57966160705922e-06,
      "loss": 1.0797,
      "step": 611
    },
    {
      "epoch": 0.39483870967741935,
      "grad_norm": 0.90234375,
      "learning_rate": 9.577982486982312e-06,
      "loss": 1.2124,
      "step": 612
    },
    {
      "epoch": 0.39548387096774196,
      "grad_norm": 0.828125,
      "learning_rate": 9.576300167566145e-06,
      "loss": 1.1845,
      "step": 613
    },
    {
      "epoch": 0.3961290322580645,
      "grad_norm": 0.7578125,
      "learning_rate": 9.574614649986409e-06,
      "loss": 0.6229,
      "step": 614
    },
    {
      "epoch": 0.3967741935483871,
      "grad_norm": 0.69921875,
      "learning_rate": 9.572925935421038e-06,
      "loss": 0.8346,
      "step": 615
    },
    {
      "epoch": 0.39741935483870966,
      "grad_norm": 0.69140625,
      "learning_rate": 9.571234025050195e-06,
      "loss": 0.6788,
      "step": 616
    },
    {
      "epoch": 0.39806451612903226,
      "grad_norm": 0.79296875,
      "learning_rate": 9.569538920056279e-06,
      "loss": 0.866,
      "step": 617
    },
    {
      "epoch": 0.3987096774193548,
      "grad_norm": 0.54296875,
      "learning_rate": 9.567840621623917e-06,
      "loss": 0.6621,
      "step": 618
    },
    {
      "epoch": 0.3993548387096774,
      "grad_norm": 0.69140625,
      "learning_rate": 9.566139130939972e-06,
      "loss": 0.743,
      "step": 619
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.60546875,
      "learning_rate": 9.564434449193538e-06,
      "loss": 0.7754,
      "step": 620
    },
    {
      "epoch": 0.40064516129032257,
      "grad_norm": 0.9140625,
      "learning_rate": 9.562726577575938e-06,
      "loss": 0.8939,
      "step": 621
    },
    {
      "epoch": 0.4012903225806452,
      "grad_norm": 0.7421875,
      "learning_rate": 9.561015517280724e-06,
      "loss": 0.7306,
      "step": 622
    },
    {
      "epoch": 0.4019354838709677,
      "grad_norm": 0.66796875,
      "learning_rate": 9.559301269503674e-06,
      "loss": 0.7956,
      "step": 623
    },
    {
      "epoch": 0.40258064516129033,
      "grad_norm": 0.68359375,
      "learning_rate": 9.557583835442799e-06,
      "loss": 0.7798,
      "step": 624
    },
    {
      "epoch": 0.4032258064516129,
      "grad_norm": 0.83203125,
      "learning_rate": 9.555863216298332e-06,
      "loss": 0.7951,
      "step": 625
    },
    {
      "epoch": 0.4038709677419355,
      "grad_norm": 0.875,
      "learning_rate": 9.554139413272734e-06,
      "loss": 1.0137,
      "step": 626
    },
    {
      "epoch": 0.4045161290322581,
      "grad_norm": 0.57421875,
      "learning_rate": 9.552412427570693e-06,
      "loss": 0.6144,
      "step": 627
    },
    {
      "epoch": 0.40516129032258064,
      "grad_norm": 0.84375,
      "learning_rate": 9.550682260399118e-06,
      "loss": 0.7853,
      "step": 628
    },
    {
      "epoch": 0.40580645161290324,
      "grad_norm": 0.96484375,
      "learning_rate": 9.548948912967141e-06,
      "loss": 0.9626,
      "step": 629
    },
    {
      "epoch": 0.4064516129032258,
      "grad_norm": 0.81640625,
      "learning_rate": 9.54721238648612e-06,
      "loss": 0.7658,
      "step": 630
    },
    {
      "epoch": 0.4070967741935484,
      "grad_norm": 1.03125,
      "learning_rate": 9.54547268216963e-06,
      "loss": 0.883,
      "step": 631
    },
    {
      "epoch": 0.40774193548387094,
      "grad_norm": 0.78515625,
      "learning_rate": 9.543729801233472e-06,
      "loss": 0.9922,
      "step": 632
    },
    {
      "epoch": 0.40838709677419355,
      "grad_norm": 0.953125,
      "learning_rate": 9.541983744895664e-06,
      "loss": 0.8632,
      "step": 633
    },
    {
      "epoch": 0.40903225806451615,
      "grad_norm": 1.1953125,
      "learning_rate": 9.540234514376443e-06,
      "loss": 1.3363,
      "step": 634
    },
    {
      "epoch": 0.4096774193548387,
      "grad_norm": 1.2109375,
      "learning_rate": 9.538482110898266e-06,
      "loss": 1.0869,
      "step": 635
    },
    {
      "epoch": 0.4103225806451613,
      "grad_norm": 0.6015625,
      "learning_rate": 9.536726535685805e-06,
      "loss": 0.5075,
      "step": 636
    },
    {
      "epoch": 0.41096774193548385,
      "grad_norm": 0.7265625,
      "learning_rate": 9.53496778996595e-06,
      "loss": 0.704,
      "step": 637
    },
    {
      "epoch": 0.41161290322580646,
      "grad_norm": 0.6640625,
      "learning_rate": 9.533205874967808e-06,
      "loss": 0.5409,
      "step": 638
    },
    {
      "epoch": 0.412258064516129,
      "grad_norm": 1.078125,
      "learning_rate": 9.5314407919227e-06,
      "loss": 0.977,
      "step": 639
    },
    {
      "epoch": 0.4129032258064516,
      "grad_norm": 1.375,
      "learning_rate": 9.529672542064159e-06,
      "loss": 0.9192,
      "step": 640
    },
    {
      "epoch": 0.4135483870967742,
      "grad_norm": 1.25,
      "learning_rate": 9.527901126627931e-06,
      "loss": 1.2309,
      "step": 641
    },
    {
      "epoch": 0.41419354838709677,
      "grad_norm": 0.71875,
      "learning_rate": 9.526126546851978e-06,
      "loss": 0.6082,
      "step": 642
    },
    {
      "epoch": 0.41483870967741937,
      "grad_norm": 0.5625,
      "learning_rate": 9.524348803976475e-06,
      "loss": 0.6538,
      "step": 643
    },
    {
      "epoch": 0.4154838709677419,
      "grad_norm": 0.78515625,
      "learning_rate": 9.522567899243798e-06,
      "loss": 0.5585,
      "step": 644
    },
    {
      "epoch": 0.4161290322580645,
      "grad_norm": 0.73828125,
      "learning_rate": 9.52078383389854e-06,
      "loss": 0.8876,
      "step": 645
    },
    {
      "epoch": 0.4167741935483871,
      "grad_norm": 0.546875,
      "learning_rate": 9.518996609187502e-06,
      "loss": 0.527,
      "step": 646
    },
    {
      "epoch": 0.4174193548387097,
      "grad_norm": 1.109375,
      "learning_rate": 9.517206226359695e-06,
      "loss": 1.0722,
      "step": 647
    },
    {
      "epoch": 0.4180645161290323,
      "grad_norm": 0.8046875,
      "learning_rate": 9.515412686666329e-06,
      "loss": 0.8651,
      "step": 648
    },
    {
      "epoch": 0.41870967741935483,
      "grad_norm": 0.97265625,
      "learning_rate": 9.51361599136083e-06,
      "loss": 0.8862,
      "step": 649
    },
    {
      "epoch": 0.41935483870967744,
      "grad_norm": 0.95703125,
      "learning_rate": 9.511816141698822e-06,
      "loss": 0.8089,
      "step": 650
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8984375,
      "learning_rate": 9.51001313893814e-06,
      "loss": 0.9055,
      "step": 651
    },
    {
      "epoch": 0.4206451612903226,
      "grad_norm": 0.77734375,
      "learning_rate": 9.508206984338814e-06,
      "loss": 0.8167,
      "step": 652
    },
    {
      "epoch": 0.42129032258064514,
      "grad_norm": 0.59765625,
      "learning_rate": 9.506397679163083e-06,
      "loss": 0.5697,
      "step": 653
    },
    {
      "epoch": 0.42193548387096774,
      "grad_norm": 0.546875,
      "learning_rate": 9.50458522467539e-06,
      "loss": 0.5252,
      "step": 654
    },
    {
      "epoch": 0.42258064516129035,
      "grad_norm": 0.94921875,
      "learning_rate": 9.502769622142371e-06,
      "loss": 0.8498,
      "step": 655
    },
    {
      "epoch": 0.4232258064516129,
      "grad_norm": 0.59765625,
      "learning_rate": 9.50095087283287e-06,
      "loss": 0.641,
      "step": 656
    },
    {
      "epoch": 0.4238709677419355,
      "grad_norm": 1.3125,
      "learning_rate": 9.499128978017922e-06,
      "loss": 1.2642,
      "step": 657
    },
    {
      "epoch": 0.42451612903225805,
      "grad_norm": 0.8125,
      "learning_rate": 9.497303938970769e-06,
      "loss": 0.7217,
      "step": 658
    },
    {
      "epoch": 0.42516129032258065,
      "grad_norm": 0.79296875,
      "learning_rate": 9.495475756966842e-06,
      "loss": 1.0474,
      "step": 659
    },
    {
      "epoch": 0.4258064516129032,
      "grad_norm": 0.9609375,
      "learning_rate": 9.493644433283775e-06,
      "loss": 1.1058,
      "step": 660
    },
    {
      "epoch": 0.4264516129032258,
      "grad_norm": 0.96875,
      "learning_rate": 9.491809969201394e-06,
      "loss": 0.9006,
      "step": 661
    },
    {
      "epoch": 0.4270967741935484,
      "grad_norm": 0.66015625,
      "learning_rate": 9.489972366001721e-06,
      "loss": 0.6511,
      "step": 662
    },
    {
      "epoch": 0.42774193548387096,
      "grad_norm": 0.83984375,
      "learning_rate": 9.488131624968972e-06,
      "loss": 0.8846,
      "step": 663
    },
    {
      "epoch": 0.42838709677419357,
      "grad_norm": 1.0,
      "learning_rate": 9.486287747389554e-06,
      "loss": 0.8125,
      "step": 664
    },
    {
      "epoch": 0.4290322580645161,
      "grad_norm": 0.85546875,
      "learning_rate": 9.484440734552069e-06,
      "loss": 0.9728,
      "step": 665
    },
    {
      "epoch": 0.4296774193548387,
      "grad_norm": 0.84375,
      "learning_rate": 9.48259058774731e-06,
      "loss": 0.9001,
      "step": 666
    },
    {
      "epoch": 0.43032258064516127,
      "grad_norm": 1.3203125,
      "learning_rate": 9.480737308268253e-06,
      "loss": 1.3812,
      "step": 667
    },
    {
      "epoch": 0.4309677419354839,
      "grad_norm": 0.67578125,
      "learning_rate": 9.478880897410073e-06,
      "loss": 0.596,
      "step": 668
    },
    {
      "epoch": 0.4316129032258065,
      "grad_norm": 0.6171875,
      "learning_rate": 9.477021356470128e-06,
      "loss": 0.7469,
      "step": 669
    },
    {
      "epoch": 0.432258064516129,
      "grad_norm": 0.9375,
      "learning_rate": 9.475158686747968e-06,
      "loss": 0.9855,
      "step": 670
    },
    {
      "epoch": 0.43290322580645163,
      "grad_norm": 0.5625,
      "learning_rate": 9.473292889545322e-06,
      "loss": 0.7078,
      "step": 671
    },
    {
      "epoch": 0.4335483870967742,
      "grad_norm": 0.921875,
      "learning_rate": 9.47142396616611e-06,
      "loss": 1.0419,
      "step": 672
    },
    {
      "epoch": 0.4341935483870968,
      "grad_norm": 0.94921875,
      "learning_rate": 9.469551917916438e-06,
      "loss": 0.7722,
      "step": 673
    },
    {
      "epoch": 0.43483870967741933,
      "grad_norm": 0.9921875,
      "learning_rate": 9.467676746104591e-06,
      "loss": 1.0574,
      "step": 674
    },
    {
      "epoch": 0.43548387096774194,
      "grad_norm": 0.67578125,
      "learning_rate": 9.465798452041041e-06,
      "loss": 0.8136,
      "step": 675
    },
    {
      "epoch": 0.43612903225806454,
      "grad_norm": 0.9296875,
      "learning_rate": 9.46391703703844e-06,
      "loss": 0.8003,
      "step": 676
    },
    {
      "epoch": 0.4367741935483871,
      "grad_norm": 0.5859375,
      "learning_rate": 9.46203250241162e-06,
      "loss": 0.6027,
      "step": 677
    },
    {
      "epoch": 0.4374193548387097,
      "grad_norm": 1.125,
      "learning_rate": 9.460144849477598e-06,
      "loss": 1.1299,
      "step": 678
    },
    {
      "epoch": 0.43806451612903224,
      "grad_norm": 1.0546875,
      "learning_rate": 9.458254079555564e-06,
      "loss": 1.0232,
      "step": 679
    },
    {
      "epoch": 0.43870967741935485,
      "grad_norm": 1.3828125,
      "learning_rate": 9.45636019396689e-06,
      "loss": 1.418,
      "step": 680
    },
    {
      "epoch": 0.4393548387096774,
      "grad_norm": 0.71484375,
      "learning_rate": 9.454463194035124e-06,
      "loss": 0.8565,
      "step": 681
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.54296875,
      "learning_rate": 9.452563081085995e-06,
      "loss": 0.5774,
      "step": 682
    },
    {
      "epoch": 0.4406451612903226,
      "grad_norm": 0.921875,
      "learning_rate": 9.450659856447397e-06,
      "loss": 0.9413,
      "step": 683
    },
    {
      "epoch": 0.44129032258064516,
      "grad_norm": 0.59765625,
      "learning_rate": 9.448753521449411e-06,
      "loss": 0.7221,
      "step": 684
    },
    {
      "epoch": 0.44193548387096776,
      "grad_norm": 0.81640625,
      "learning_rate": 9.446844077424284e-06,
      "loss": 0.8214,
      "step": 685
    },
    {
      "epoch": 0.4425806451612903,
      "grad_norm": 1.765625,
      "learning_rate": 9.444931525706438e-06,
      "loss": 1.3456,
      "step": 686
    },
    {
      "epoch": 0.4432258064516129,
      "grad_norm": 0.85546875,
      "learning_rate": 9.443015867632465e-06,
      "loss": 0.7939,
      "step": 687
    },
    {
      "epoch": 0.44387096774193546,
      "grad_norm": 0.87109375,
      "learning_rate": 9.441097104541132e-06,
      "loss": 0.7497,
      "step": 688
    },
    {
      "epoch": 0.44451612903225807,
      "grad_norm": 0.85546875,
      "learning_rate": 9.43917523777337e-06,
      "loss": 0.888,
      "step": 689
    },
    {
      "epoch": 0.44516129032258067,
      "grad_norm": 0.59765625,
      "learning_rate": 9.437250268672287e-06,
      "loss": 0.6529,
      "step": 690
    },
    {
      "epoch": 0.4458064516129032,
      "grad_norm": 1.1953125,
      "learning_rate": 9.43532219858315e-06,
      "loss": 1.1033,
      "step": 691
    },
    {
      "epoch": 0.4464516129032258,
      "grad_norm": 0.82421875,
      "learning_rate": 9.4333910288534e-06,
      "loss": 1.1919,
      "step": 692
    },
    {
      "epoch": 0.4470967741935484,
      "grad_norm": 0.859375,
      "learning_rate": 9.431456760832641e-06,
      "loss": 1.0352,
      "step": 693
    },
    {
      "epoch": 0.447741935483871,
      "grad_norm": 0.8515625,
      "learning_rate": 9.429519395872644e-06,
      "loss": 0.7072,
      "step": 694
    },
    {
      "epoch": 0.4483870967741935,
      "grad_norm": 1.09375,
      "learning_rate": 9.427578935327342e-06,
      "loss": 0.8203,
      "step": 695
    },
    {
      "epoch": 0.44903225806451613,
      "grad_norm": 0.9453125,
      "learning_rate": 9.425635380552834e-06,
      "loss": 1.0274,
      "step": 696
    },
    {
      "epoch": 0.44967741935483874,
      "grad_norm": 0.7421875,
      "learning_rate": 9.423688732907378e-06,
      "loss": 0.784,
      "step": 697
    },
    {
      "epoch": 0.4503225806451613,
      "grad_norm": 0.90625,
      "learning_rate": 9.421738993751397e-06,
      "loss": 0.9006,
      "step": 698
    },
    {
      "epoch": 0.4509677419354839,
      "grad_norm": 1.09375,
      "learning_rate": 9.419786164447475e-06,
      "loss": 0.5862,
      "step": 699
    },
    {
      "epoch": 0.45161290322580644,
      "grad_norm": 0.95703125,
      "learning_rate": 9.417830246360351e-06,
      "loss": 0.9386,
      "step": 700
    },
    {
      "epoch": 0.45225806451612904,
      "grad_norm": 1.1875,
      "learning_rate": 9.415871240856927e-06,
      "loss": 0.9061,
      "step": 701
    },
    {
      "epoch": 0.4529032258064516,
      "grad_norm": 0.6015625,
      "learning_rate": 9.413909149306258e-06,
      "loss": 0.6513,
      "step": 702
    },
    {
      "epoch": 0.4535483870967742,
      "grad_norm": 0.77734375,
      "learning_rate": 9.411943973079562e-06,
      "loss": 0.8155,
      "step": 703
    },
    {
      "epoch": 0.4541935483870968,
      "grad_norm": 0.76171875,
      "learning_rate": 9.409975713550207e-06,
      "loss": 0.6568,
      "step": 704
    },
    {
      "epoch": 0.45483870967741935,
      "grad_norm": 0.66015625,
      "learning_rate": 9.408004372093721e-06,
      "loss": 0.6748,
      "step": 705
    },
    {
      "epoch": 0.45548387096774196,
      "grad_norm": 1.1484375,
      "learning_rate": 9.40602995008778e-06,
      "loss": 1.1019,
      "step": 706
    },
    {
      "epoch": 0.4561290322580645,
      "grad_norm": 1.6015625,
      "learning_rate": 9.404052448912218e-06,
      "loss": 1.0911,
      "step": 707
    },
    {
      "epoch": 0.4567741935483871,
      "grad_norm": 0.8125,
      "learning_rate": 9.402071869949017e-06,
      "loss": 0.7629,
      "step": 708
    },
    {
      "epoch": 0.45741935483870966,
      "grad_norm": 0.69921875,
      "learning_rate": 9.400088214582313e-06,
      "loss": 0.8408,
      "step": 709
    },
    {
      "epoch": 0.45806451612903226,
      "grad_norm": 1.0625,
      "learning_rate": 9.398101484198388e-06,
      "loss": 0.9277,
      "step": 710
    },
    {
      "epoch": 0.4587096774193548,
      "grad_norm": 0.515625,
      "learning_rate": 9.39611168018568e-06,
      "loss": 0.5227,
      "step": 711
    },
    {
      "epoch": 0.4593548387096774,
      "grad_norm": 0.734375,
      "learning_rate": 9.394118803934765e-06,
      "loss": 0.8891,
      "step": 712
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8125,
      "learning_rate": 9.392122856838376e-06,
      "loss": 0.665,
      "step": 713
    },
    {
      "epoch": 0.46064516129032257,
      "grad_norm": 0.70703125,
      "learning_rate": 9.390123840291388e-06,
      "loss": 0.6304,
      "step": 714
    },
    {
      "epoch": 0.4612903225806452,
      "grad_norm": 0.83203125,
      "learning_rate": 9.388121755690815e-06,
      "loss": 0.7555,
      "step": 715
    },
    {
      "epoch": 0.4619354838709677,
      "grad_norm": 0.73046875,
      "learning_rate": 9.386116604435828e-06,
      "loss": 0.8047,
      "step": 716
    },
    {
      "epoch": 0.4625806451612903,
      "grad_norm": 0.96484375,
      "learning_rate": 9.384108387927731e-06,
      "loss": 0.909,
      "step": 717
    },
    {
      "epoch": 0.4632258064516129,
      "grad_norm": 0.8515625,
      "learning_rate": 9.382097107569972e-06,
      "loss": 0.91,
      "step": 718
    },
    {
      "epoch": 0.4638709677419355,
      "grad_norm": 0.859375,
      "learning_rate": 9.380082764768143e-06,
      "loss": 0.8508,
      "step": 719
    },
    {
      "epoch": 0.4645161290322581,
      "grad_norm": 0.890625,
      "learning_rate": 9.378065360929975e-06,
      "loss": 0.977,
      "step": 720
    },
    {
      "epoch": 0.46516129032258063,
      "grad_norm": 0.7109375,
      "learning_rate": 9.376044897465338e-06,
      "loss": 0.6769,
      "step": 721
    },
    {
      "epoch": 0.46580645161290324,
      "grad_norm": 0.55859375,
      "learning_rate": 9.374021375786236e-06,
      "loss": 0.6343,
      "step": 722
    },
    {
      "epoch": 0.4664516129032258,
      "grad_norm": 0.8984375,
      "learning_rate": 9.37199479730682e-06,
      "loss": 0.8626,
      "step": 723
    },
    {
      "epoch": 0.4670967741935484,
      "grad_norm": 0.94140625,
      "learning_rate": 9.369965163443367e-06,
      "loss": 0.8344,
      "step": 724
    },
    {
      "epoch": 0.46774193548387094,
      "grad_norm": 0.578125,
      "learning_rate": 9.367932475614297e-06,
      "loss": 0.5573,
      "step": 725
    },
    {
      "epoch": 0.46838709677419355,
      "grad_norm": 1.09375,
      "learning_rate": 9.36589673524016e-06,
      "loss": 1.2083,
      "step": 726
    },
    {
      "epoch": 0.46903225806451615,
      "grad_norm": 1.03125,
      "learning_rate": 9.36385794374364e-06,
      "loss": 0.9418,
      "step": 727
    },
    {
      "epoch": 0.4696774193548387,
      "grad_norm": 0.78515625,
      "learning_rate": 9.361816102549552e-06,
      "loss": 0.6918,
      "step": 728
    },
    {
      "epoch": 0.4703225806451613,
      "grad_norm": 1.0078125,
      "learning_rate": 9.359771213084847e-06,
      "loss": 1.0126,
      "step": 729
    },
    {
      "epoch": 0.47096774193548385,
      "grad_norm": 0.72265625,
      "learning_rate": 9.357723276778603e-06,
      "loss": 0.7745,
      "step": 730
    },
    {
      "epoch": 0.47161290322580646,
      "grad_norm": 0.6953125,
      "learning_rate": 9.355672295062026e-06,
      "loss": 0.7696,
      "step": 731
    },
    {
      "epoch": 0.472258064516129,
      "grad_norm": 1.2265625,
      "learning_rate": 9.353618269368451e-06,
      "loss": 1.0506,
      "step": 732
    },
    {
      "epoch": 0.4729032258064516,
      "grad_norm": 0.94921875,
      "learning_rate": 9.351561201133344e-06,
      "loss": 0.8161,
      "step": 733
    },
    {
      "epoch": 0.4735483870967742,
      "grad_norm": 0.7578125,
      "learning_rate": 9.349501091794292e-06,
      "loss": 0.7961,
      "step": 734
    },
    {
      "epoch": 0.47419354838709676,
      "grad_norm": 0.66015625,
      "learning_rate": 9.347437942791012e-06,
      "loss": 0.6742,
      "step": 735
    },
    {
      "epoch": 0.47483870967741937,
      "grad_norm": 0.75,
      "learning_rate": 9.34537175556534e-06,
      "loss": 0.795,
      "step": 736
    },
    {
      "epoch": 0.4754838709677419,
      "grad_norm": 0.7734375,
      "learning_rate": 9.343302531561241e-06,
      "loss": 0.8032,
      "step": 737
    },
    {
      "epoch": 0.4761290322580645,
      "grad_norm": 0.69921875,
      "learning_rate": 9.341230272224798e-06,
      "loss": 0.8327,
      "step": 738
    },
    {
      "epoch": 0.47677419354838707,
      "grad_norm": 0.9375,
      "learning_rate": 9.339154979004217e-06,
      "loss": 1.0673,
      "step": 739
    },
    {
      "epoch": 0.4774193548387097,
      "grad_norm": 0.87890625,
      "learning_rate": 9.337076653349824e-06,
      "loss": 1.0497,
      "step": 740
    },
    {
      "epoch": 0.4780645161290323,
      "grad_norm": 0.875,
      "learning_rate": 9.334995296714065e-06,
      "loss": 0.7438,
      "step": 741
    },
    {
      "epoch": 0.47870967741935483,
      "grad_norm": 1.2265625,
      "learning_rate": 9.332910910551503e-06,
      "loss": 0.99,
      "step": 742
    },
    {
      "epoch": 0.47935483870967743,
      "grad_norm": 0.55859375,
      "learning_rate": 9.330823496318817e-06,
      "loss": 0.551,
      "step": 743
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.85546875,
      "learning_rate": 9.328733055474806e-06,
      "loss": 0.7295,
      "step": 744
    },
    {
      "epoch": 0.4806451612903226,
      "grad_norm": 0.890625,
      "learning_rate": 9.326639589480382e-06,
      "loss": 0.9049,
      "step": 745
    },
    {
      "epoch": 0.48129032258064514,
      "grad_norm": 0.9375,
      "learning_rate": 9.32454309979857e-06,
      "loss": 0.9941,
      "step": 746
    },
    {
      "epoch": 0.48193548387096774,
      "grad_norm": 1.0234375,
      "learning_rate": 9.32244358789451e-06,
      "loss": 0.798,
      "step": 747
    },
    {
      "epoch": 0.48258064516129034,
      "grad_norm": 0.58203125,
      "learning_rate": 9.32034105523545e-06,
      "loss": 0.6208,
      "step": 748
    },
    {
      "epoch": 0.4832258064516129,
      "grad_norm": 0.921875,
      "learning_rate": 9.31823550329076e-06,
      "loss": 0.9589,
      "step": 749
    },
    {
      "epoch": 0.4838709677419355,
      "grad_norm": 0.91796875,
      "learning_rate": 9.316126933531904e-06,
      "loss": 0.7783,
      "step": 750
    },
    {
      "epoch": 0.48451612903225805,
      "grad_norm": 1.0625,
      "learning_rate": 9.31401534743247e-06,
      "loss": 0.9839,
      "step": 751
    },
    {
      "epoch": 0.48516129032258065,
      "grad_norm": 0.71484375,
      "learning_rate": 9.311900746468144e-06,
      "loss": 0.5985,
      "step": 752
    },
    {
      "epoch": 0.4858064516129032,
      "grad_norm": 0.85546875,
      "learning_rate": 9.309783132116722e-06,
      "loss": 0.7608,
      "step": 753
    },
    {
      "epoch": 0.4864516129032258,
      "grad_norm": 0.625,
      "learning_rate": 9.307662505858107e-06,
      "loss": 0.5585,
      "step": 754
    },
    {
      "epoch": 0.4870967741935484,
      "grad_norm": 0.85546875,
      "learning_rate": 9.305538869174307e-06,
      "loss": 1.073,
      "step": 755
    },
    {
      "epoch": 0.48774193548387096,
      "grad_norm": 1.0,
      "learning_rate": 9.303412223549433e-06,
      "loss": 0.7723,
      "step": 756
    },
    {
      "epoch": 0.48838709677419356,
      "grad_norm": 0.78125,
      "learning_rate": 9.301282570469698e-06,
      "loss": 0.6958,
      "step": 757
    },
    {
      "epoch": 0.4890322580645161,
      "grad_norm": 0.71875,
      "learning_rate": 9.299149911423417e-06,
      "loss": 0.6148,
      "step": 758
    },
    {
      "epoch": 0.4896774193548387,
      "grad_norm": 0.76953125,
      "learning_rate": 9.297014247901008e-06,
      "loss": 0.642,
      "step": 759
    },
    {
      "epoch": 0.49032258064516127,
      "grad_norm": 0.5859375,
      "learning_rate": 9.294875581394987e-06,
      "loss": 0.6219,
      "step": 760
    },
    {
      "epoch": 0.49096774193548387,
      "grad_norm": 1.0625,
      "learning_rate": 9.292733913399968e-06,
      "loss": 1.0906,
      "step": 761
    },
    {
      "epoch": 0.4916129032258065,
      "grad_norm": 0.75,
      "learning_rate": 9.29058924541266e-06,
      "loss": 0.6174,
      "step": 762
    },
    {
      "epoch": 0.492258064516129,
      "grad_norm": 1.078125,
      "learning_rate": 9.288441578931877e-06,
      "loss": 0.8964,
      "step": 763
    },
    {
      "epoch": 0.49290322580645163,
      "grad_norm": 0.9140625,
      "learning_rate": 9.286290915458522e-06,
      "loss": 0.8117,
      "step": 764
    },
    {
      "epoch": 0.4935483870967742,
      "grad_norm": 0.6328125,
      "learning_rate": 9.28413725649559e-06,
      "loss": 0.6407,
      "step": 765
    },
    {
      "epoch": 0.4941935483870968,
      "grad_norm": 0.703125,
      "learning_rate": 9.281980603548179e-06,
      "loss": 0.6051,
      "step": 766
    },
    {
      "epoch": 0.49483870967741933,
      "grad_norm": 0.8359375,
      "learning_rate": 9.279820958123472e-06,
      "loss": 0.8181,
      "step": 767
    },
    {
      "epoch": 0.49548387096774194,
      "grad_norm": 0.890625,
      "learning_rate": 9.277658321730741e-06,
      "loss": 0.7492,
      "step": 768
    },
    {
      "epoch": 0.49612903225806454,
      "grad_norm": 1.5234375,
      "learning_rate": 9.275492695881357e-06,
      "loss": 1.1069,
      "step": 769
    },
    {
      "epoch": 0.4967741935483871,
      "grad_norm": 0.96484375,
      "learning_rate": 9.27332408208877e-06,
      "loss": 0.9892,
      "step": 770
    },
    {
      "epoch": 0.4974193548387097,
      "grad_norm": 0.86328125,
      "learning_rate": 9.271152481868528e-06,
      "loss": 0.6685,
      "step": 771
    },
    {
      "epoch": 0.49806451612903224,
      "grad_norm": 0.8984375,
      "learning_rate": 9.268977896738261e-06,
      "loss": 1.0137,
      "step": 772
    },
    {
      "epoch": 0.49870967741935485,
      "grad_norm": 0.7734375,
      "learning_rate": 9.266800328217682e-06,
      "loss": 0.7903,
      "step": 773
    },
    {
      "epoch": 0.4993548387096774,
      "grad_norm": 0.83203125,
      "learning_rate": 9.264619777828597e-06,
      "loss": 0.6596,
      "step": 774
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7109375,
      "learning_rate": 9.262436247094888e-06,
      "loss": 0.8137,
      "step": 775
    },
    {
      "epoch": 0.5006451612903225,
      "grad_norm": 2.15625,
      "learning_rate": 9.260249737542523e-06,
      "loss": 1.1471,
      "step": 776
    },
    {
      "epoch": 0.5012903225806452,
      "grad_norm": 0.765625,
      "learning_rate": 9.258060250699556e-06,
      "loss": 0.6483,
      "step": 777
    },
    {
      "epoch": 0.5019354838709678,
      "grad_norm": 0.8984375,
      "learning_rate": 9.255867788096114e-06,
      "loss": 0.7946,
      "step": 778
    },
    {
      "epoch": 0.5025806451612903,
      "grad_norm": 0.67578125,
      "learning_rate": 9.253672351264408e-06,
      "loss": 0.7227,
      "step": 779
    },
    {
      "epoch": 0.5032258064516129,
      "grad_norm": 0.64453125,
      "learning_rate": 9.25147394173873e-06,
      "loss": 0.7185,
      "step": 780
    },
    {
      "epoch": 0.5038709677419355,
      "grad_norm": 0.7890625,
      "learning_rate": 9.249272561055439e-06,
      "loss": 1.0343,
      "step": 781
    },
    {
      "epoch": 0.5045161290322581,
      "grad_norm": 0.98046875,
      "learning_rate": 9.247068210752982e-06,
      "loss": 0.9947,
      "step": 782
    },
    {
      "epoch": 0.5051612903225806,
      "grad_norm": 1.0625,
      "learning_rate": 9.244860892371879e-06,
      "loss": 1.0296,
      "step": 783
    },
    {
      "epoch": 0.5058064516129033,
      "grad_norm": 0.6875,
      "learning_rate": 9.242650607454722e-06,
      "loss": 0.6722,
      "step": 784
    },
    {
      "epoch": 0.5064516129032258,
      "grad_norm": 0.98046875,
      "learning_rate": 9.240437357546171e-06,
      "loss": 0.841,
      "step": 785
    },
    {
      "epoch": 0.5070967741935484,
      "grad_norm": 0.890625,
      "learning_rate": 9.23822114419297e-06,
      "loss": 1.0517,
      "step": 786
    },
    {
      "epoch": 0.5077419354838709,
      "grad_norm": 0.59375,
      "learning_rate": 9.236001968943923e-06,
      "loss": 0.4492,
      "step": 787
    },
    {
      "epoch": 0.5083870967741936,
      "grad_norm": 1.109375,
      "learning_rate": 9.233779833349911e-06,
      "loss": 0.8406,
      "step": 788
    },
    {
      "epoch": 0.5090322580645161,
      "grad_norm": 1.0546875,
      "learning_rate": 9.23155473896388e-06,
      "loss": 1.0326,
      "step": 789
    },
    {
      "epoch": 0.5096774193548387,
      "grad_norm": 0.8671875,
      "learning_rate": 9.229326687340846e-06,
      "loss": 0.9996,
      "step": 790
    },
    {
      "epoch": 0.5103225806451613,
      "grad_norm": 0.67578125,
      "learning_rate": 9.227095680037891e-06,
      "loss": 0.754,
      "step": 791
    },
    {
      "epoch": 0.5109677419354839,
      "grad_norm": 0.65234375,
      "learning_rate": 9.22486171861416e-06,
      "loss": 0.6286,
      "step": 792
    },
    {
      "epoch": 0.5116129032258064,
      "grad_norm": 0.890625,
      "learning_rate": 9.222624804630869e-06,
      "loss": 0.8413,
      "step": 793
    },
    {
      "epoch": 0.512258064516129,
      "grad_norm": 0.890625,
      "learning_rate": 9.22038493965129e-06,
      "loss": 0.8486,
      "step": 794
    },
    {
      "epoch": 0.5129032258064516,
      "grad_norm": 0.96875,
      "learning_rate": 9.21814212524076e-06,
      "loss": 0.9727,
      "step": 795
    },
    {
      "epoch": 0.5135483870967742,
      "grad_norm": 0.486328125,
      "learning_rate": 9.215896362966681e-06,
      "loss": 0.4443,
      "step": 796
    },
    {
      "epoch": 0.5141935483870967,
      "grad_norm": 1.0859375,
      "learning_rate": 9.21364765439851e-06,
      "loss": 0.9652,
      "step": 797
    },
    {
      "epoch": 0.5148387096774194,
      "grad_norm": 0.6796875,
      "learning_rate": 9.211396001107764e-06,
      "loss": 0.5612,
      "step": 798
    },
    {
      "epoch": 0.515483870967742,
      "grad_norm": 0.73046875,
      "learning_rate": 9.20914140466802e-06,
      "loss": 0.6829,
      "step": 799
    },
    {
      "epoch": 0.5161290322580645,
      "grad_norm": 0.87890625,
      "learning_rate": 9.206883866654913e-06,
      "loss": 0.8635,
      "step": 800
    },
    {
      "epoch": 0.516774193548387,
      "grad_norm": 0.875,
      "learning_rate": 9.204623388646126e-06,
      "loss": 0.7904,
      "step": 801
    },
    {
      "epoch": 0.5174193548387097,
      "grad_norm": 0.75,
      "learning_rate": 9.202359972221406e-06,
      "loss": 0.734,
      "step": 802
    },
    {
      "epoch": 0.5180645161290323,
      "grad_norm": 0.91796875,
      "learning_rate": 9.20009361896255e-06,
      "loss": 0.9264,
      "step": 803
    },
    {
      "epoch": 0.5187096774193548,
      "grad_norm": 0.734375,
      "learning_rate": 9.197824330453405e-06,
      "loss": 0.5907,
      "step": 804
    },
    {
      "epoch": 0.5193548387096775,
      "grad_norm": 1.125,
      "learning_rate": 9.19555210827987e-06,
      "loss": 0.9998,
      "step": 805
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7109375,
      "learning_rate": 9.1932769540299e-06,
      "loss": 0.6732,
      "step": 806
    },
    {
      "epoch": 0.5206451612903226,
      "grad_norm": 0.734375,
      "learning_rate": 9.190998869293492e-06,
      "loss": 0.8128,
      "step": 807
    },
    {
      "epoch": 0.5212903225806451,
      "grad_norm": 0.96484375,
      "learning_rate": 9.188717855662692e-06,
      "loss": 0.9037,
      "step": 808
    },
    {
      "epoch": 0.5219354838709678,
      "grad_norm": 0.77734375,
      "learning_rate": 9.186433914731599e-06,
      "loss": 0.9559,
      "step": 809
    },
    {
      "epoch": 0.5225806451612903,
      "grad_norm": 0.68359375,
      "learning_rate": 9.184147048096348e-06,
      "loss": 0.6401,
      "step": 810
    },
    {
      "epoch": 0.5232258064516129,
      "grad_norm": 1.234375,
      "learning_rate": 9.181857257355127e-06,
      "loss": 0.7855,
      "step": 811
    },
    {
      "epoch": 0.5238709677419355,
      "grad_norm": 0.52734375,
      "learning_rate": 9.179564544108162e-06,
      "loss": 0.5617,
      "step": 812
    },
    {
      "epoch": 0.5245161290322581,
      "grad_norm": 0.87109375,
      "learning_rate": 9.177268909957727e-06,
      "loss": 0.5924,
      "step": 813
    },
    {
      "epoch": 0.5251612903225806,
      "grad_norm": 0.8359375,
      "learning_rate": 9.174970356508133e-06,
      "loss": 0.92,
      "step": 814
    },
    {
      "epoch": 0.5258064516129032,
      "grad_norm": 0.77734375,
      "learning_rate": 9.17266888536573e-06,
      "loss": 0.8434,
      "step": 815
    },
    {
      "epoch": 0.5264516129032258,
      "grad_norm": 0.69140625,
      "learning_rate": 9.17036449813891e-06,
      "loss": 0.7619,
      "step": 816
    },
    {
      "epoch": 0.5270967741935484,
      "grad_norm": 0.78515625,
      "learning_rate": 9.168057196438105e-06,
      "loss": 0.6174,
      "step": 817
    },
    {
      "epoch": 0.5277419354838709,
      "grad_norm": 0.78515625,
      "learning_rate": 9.165746981875778e-06,
      "loss": 0.6768,
      "step": 818
    },
    {
      "epoch": 0.5283870967741936,
      "grad_norm": 0.82421875,
      "learning_rate": 9.163433856066433e-06,
      "loss": 0.9663,
      "step": 819
    },
    {
      "epoch": 0.5290322580645161,
      "grad_norm": 0.5390625,
      "learning_rate": 9.161117820626603e-06,
      "loss": 0.5546,
      "step": 820
    },
    {
      "epoch": 0.5296774193548387,
      "grad_norm": 1.0390625,
      "learning_rate": 9.15879887717486e-06,
      "loss": 0.9314,
      "step": 821
    },
    {
      "epoch": 0.5303225806451612,
      "grad_norm": 0.8203125,
      "learning_rate": 9.156477027331805e-06,
      "loss": 0.6448,
      "step": 822
    },
    {
      "epoch": 0.5309677419354839,
      "grad_norm": 0.80078125,
      "learning_rate": 9.15415227272007e-06,
      "loss": 0.721,
      "step": 823
    },
    {
      "epoch": 0.5316129032258065,
      "grad_norm": 0.890625,
      "learning_rate": 9.15182461496432e-06,
      "loss": 0.9803,
      "step": 824
    },
    {
      "epoch": 0.532258064516129,
      "grad_norm": 0.51171875,
      "learning_rate": 9.149494055691245e-06,
      "loss": 0.5695,
      "step": 825
    },
    {
      "epoch": 0.5329032258064517,
      "grad_norm": 0.87109375,
      "learning_rate": 9.147160596529568e-06,
      "loss": 0.7885,
      "step": 826
    },
    {
      "epoch": 0.5335483870967742,
      "grad_norm": 1.3515625,
      "learning_rate": 9.144824239110032e-06,
      "loss": 1.0201,
      "step": 827
    },
    {
      "epoch": 0.5341935483870968,
      "grad_norm": 0.87890625,
      "learning_rate": 9.142484985065408e-06,
      "loss": 0.7132,
      "step": 828
    },
    {
      "epoch": 0.5348387096774193,
      "grad_norm": 0.76171875,
      "learning_rate": 9.140142836030495e-06,
      "loss": 0.6726,
      "step": 829
    },
    {
      "epoch": 0.535483870967742,
      "grad_norm": 0.984375,
      "learning_rate": 9.137797793642109e-06,
      "loss": 1.0072,
      "step": 830
    },
    {
      "epoch": 0.5361290322580645,
      "grad_norm": 0.7109375,
      "learning_rate": 9.135449859539094e-06,
      "loss": 0.6849,
      "step": 831
    },
    {
      "epoch": 0.5367741935483871,
      "grad_norm": 0.75390625,
      "learning_rate": 9.133099035362309e-06,
      "loss": 0.5525,
      "step": 832
    },
    {
      "epoch": 0.5374193548387097,
      "grad_norm": 0.72265625,
      "learning_rate": 9.130745322754639e-06,
      "loss": 0.7076,
      "step": 833
    },
    {
      "epoch": 0.5380645161290323,
      "grad_norm": 0.796875,
      "learning_rate": 9.128388723360983e-06,
      "loss": 1.0331,
      "step": 834
    },
    {
      "epoch": 0.5387096774193548,
      "grad_norm": 0.9609375,
      "learning_rate": 9.126029238828258e-06,
      "loss": 1.0182,
      "step": 835
    },
    {
      "epoch": 0.5393548387096774,
      "grad_norm": 0.76171875,
      "learning_rate": 9.123666870805399e-06,
      "loss": 0.8895,
      "step": 836
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.90234375,
      "learning_rate": 9.121301620943355e-06,
      "loss": 0.7363,
      "step": 837
    },
    {
      "epoch": 0.5406451612903226,
      "grad_norm": 0.69140625,
      "learning_rate": 9.118933490895089e-06,
      "loss": 0.6416,
      "step": 838
    },
    {
      "epoch": 0.5412903225806451,
      "grad_norm": 0.66015625,
      "learning_rate": 9.11656248231558e-06,
      "loss": 0.6604,
      "step": 839
    },
    {
      "epoch": 0.5419354838709678,
      "grad_norm": 0.8515625,
      "learning_rate": 9.114188596861809e-06,
      "loss": 0.8192,
      "step": 840
    },
    {
      "epoch": 0.5425806451612903,
      "grad_norm": 0.76171875,
      "learning_rate": 9.11181183619278e-06,
      "loss": 0.8129,
      "step": 841
    },
    {
      "epoch": 0.5432258064516129,
      "grad_norm": 0.875,
      "learning_rate": 9.1094322019695e-06,
      "loss": 0.6197,
      "step": 842
    },
    {
      "epoch": 0.5438709677419354,
      "grad_norm": 0.8671875,
      "learning_rate": 9.107049695854983e-06,
      "loss": 0.7702,
      "step": 843
    },
    {
      "epoch": 0.5445161290322581,
      "grad_norm": 0.61328125,
      "learning_rate": 9.104664319514252e-06,
      "loss": 0.6382,
      "step": 844
    },
    {
      "epoch": 0.5451612903225806,
      "grad_norm": 0.7109375,
      "learning_rate": 9.102276074614338e-06,
      "loss": 0.7102,
      "step": 845
    },
    {
      "epoch": 0.5458064516129032,
      "grad_norm": 0.80859375,
      "learning_rate": 9.09988496282427e-06,
      "loss": 0.5963,
      "step": 846
    },
    {
      "epoch": 0.5464516129032259,
      "grad_norm": 0.875,
      "learning_rate": 9.097490985815091e-06,
      "loss": 0.7096,
      "step": 847
    },
    {
      "epoch": 0.5470967741935484,
      "grad_norm": 1.1328125,
      "learning_rate": 9.095094145259836e-06,
      "loss": 1.1703,
      "step": 848
    },
    {
      "epoch": 0.547741935483871,
      "grad_norm": 1.359375,
      "learning_rate": 9.092694442833545e-06,
      "loss": 1.0412,
      "step": 849
    },
    {
      "epoch": 0.5483870967741935,
      "grad_norm": 0.60546875,
      "learning_rate": 9.09029188021326e-06,
      "loss": 0.5116,
      "step": 850
    },
    {
      "epoch": 0.5490322580645162,
      "grad_norm": 0.76953125,
      "learning_rate": 9.087886459078022e-06,
      "loss": 0.7311,
      "step": 851
    },
    {
      "epoch": 0.5496774193548387,
      "grad_norm": 0.71875,
      "learning_rate": 9.085478181108864e-06,
      "loss": 0.6423,
      "step": 852
    },
    {
      "epoch": 0.5503225806451613,
      "grad_norm": 0.78125,
      "learning_rate": 9.083067047988825e-06,
      "loss": 0.732,
      "step": 853
    },
    {
      "epoch": 0.5509677419354839,
      "grad_norm": 1.0234375,
      "learning_rate": 9.08065306140293e-06,
      "loss": 0.9961,
      "step": 854
    },
    {
      "epoch": 0.5516129032258065,
      "grad_norm": 1.0546875,
      "learning_rate": 9.0782362230382e-06,
      "loss": 0.7943,
      "step": 855
    },
    {
      "epoch": 0.552258064516129,
      "grad_norm": 0.9921875,
      "learning_rate": 9.07581653458366e-06,
      "loss": 0.9307,
      "step": 856
    },
    {
      "epoch": 0.5529032258064516,
      "grad_norm": 1.171875,
      "learning_rate": 9.073393997730306e-06,
      "loss": 0.9179,
      "step": 857
    },
    {
      "epoch": 0.5535483870967742,
      "grad_norm": 0.71875,
      "learning_rate": 9.070968614171146e-06,
      "loss": 0.6248,
      "step": 858
    },
    {
      "epoch": 0.5541935483870968,
      "grad_norm": 0.859375,
      "learning_rate": 9.068540385601164e-06,
      "loss": 0.6796,
      "step": 859
    },
    {
      "epoch": 0.5548387096774193,
      "grad_norm": 0.59765625,
      "learning_rate": 9.066109313717335e-06,
      "loss": 0.6081,
      "step": 860
    },
    {
      "epoch": 0.555483870967742,
      "grad_norm": 0.8125,
      "learning_rate": 9.063675400218623e-06,
      "loss": 0.8802,
      "step": 861
    },
    {
      "epoch": 0.5561290322580645,
      "grad_norm": 0.44921875,
      "learning_rate": 9.06123864680598e-06,
      "loss": 0.3751,
      "step": 862
    },
    {
      "epoch": 0.5567741935483871,
      "grad_norm": 1.2265625,
      "learning_rate": 9.058799055182334e-06,
      "loss": 0.9232,
      "step": 863
    },
    {
      "epoch": 0.5574193548387096,
      "grad_norm": 0.859375,
      "learning_rate": 9.056356627052609e-06,
      "loss": 0.9186,
      "step": 864
    },
    {
      "epoch": 0.5580645161290323,
      "grad_norm": 1.03125,
      "learning_rate": 9.053911364123699e-06,
      "loss": 0.7765,
      "step": 865
    },
    {
      "epoch": 0.5587096774193548,
      "grad_norm": 0.7734375,
      "learning_rate": 9.051463268104488e-06,
      "loss": 0.8053,
      "step": 866
    },
    {
      "epoch": 0.5593548387096774,
      "grad_norm": 1.4296875,
      "learning_rate": 9.049012340705834e-06,
      "loss": 1.041,
      "step": 867
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.109375,
      "learning_rate": 9.046558583640579e-06,
      "loss": 0.9164,
      "step": 868
    },
    {
      "epoch": 0.5606451612903226,
      "grad_norm": 1.1015625,
      "learning_rate": 9.044101998623538e-06,
      "loss": 0.8413,
      "step": 869
    },
    {
      "epoch": 0.5612903225806452,
      "grad_norm": 1.09375,
      "learning_rate": 9.041642587371504e-06,
      "loss": 1.1971,
      "step": 870
    },
    {
      "epoch": 0.5619354838709677,
      "grad_norm": 0.87890625,
      "learning_rate": 9.039180351603248e-06,
      "loss": 0.7692,
      "step": 871
    },
    {
      "epoch": 0.5625806451612904,
      "grad_norm": 0.67578125,
      "learning_rate": 9.036715293039507e-06,
      "loss": 0.7559,
      "step": 872
    },
    {
      "epoch": 0.5632258064516129,
      "grad_norm": 0.6484375,
      "learning_rate": 9.034247413403e-06,
      "loss": 0.6289,
      "step": 873
    },
    {
      "epoch": 0.5638709677419355,
      "grad_norm": 0.75,
      "learning_rate": 9.031776714418413e-06,
      "loss": 0.5934,
      "step": 874
    },
    {
      "epoch": 0.5645161290322581,
      "grad_norm": 0.6328125,
      "learning_rate": 9.0293031978124e-06,
      "loss": 0.6482,
      "step": 875
    },
    {
      "epoch": 0.5651612903225807,
      "grad_norm": 1.2421875,
      "learning_rate": 9.026826865313591e-06,
      "loss": 1.135,
      "step": 876
    },
    {
      "epoch": 0.5658064516129032,
      "grad_norm": 1.234375,
      "learning_rate": 9.024347718652578e-06,
      "loss": 1.2876,
      "step": 877
    },
    {
      "epoch": 0.5664516129032258,
      "grad_norm": 0.7421875,
      "learning_rate": 9.02186575956192e-06,
      "loss": 0.868,
      "step": 878
    },
    {
      "epoch": 0.5670967741935484,
      "grad_norm": 0.98046875,
      "learning_rate": 9.019380989776143e-06,
      "loss": 1.0267,
      "step": 879
    },
    {
      "epoch": 0.567741935483871,
      "grad_norm": 0.6953125,
      "learning_rate": 9.016893411031739e-06,
      "loss": 0.7283,
      "step": 880
    },
    {
      "epoch": 0.5683870967741935,
      "grad_norm": 1.1015625,
      "learning_rate": 9.01440302506716e-06,
      "loss": 0.9602,
      "step": 881
    },
    {
      "epoch": 0.5690322580645162,
      "grad_norm": 0.8203125,
      "learning_rate": 9.011909833622822e-06,
      "loss": 0.7214,
      "step": 882
    },
    {
      "epoch": 0.5696774193548387,
      "grad_norm": 0.74609375,
      "learning_rate": 9.009413838441102e-06,
      "loss": 0.6266,
      "step": 883
    },
    {
      "epoch": 0.5703225806451613,
      "grad_norm": 1.0859375,
      "learning_rate": 9.00691504126633e-06,
      "loss": 1.01,
      "step": 884
    },
    {
      "epoch": 0.5709677419354838,
      "grad_norm": 0.6328125,
      "learning_rate": 9.004413443844802e-06,
      "loss": 0.6085,
      "step": 885
    },
    {
      "epoch": 0.5716129032258065,
      "grad_norm": 0.69140625,
      "learning_rate": 9.00190904792477e-06,
      "loss": 0.7899,
      "step": 886
    },
    {
      "epoch": 0.572258064516129,
      "grad_norm": 0.6796875,
      "learning_rate": 8.999401855256438e-06,
      "loss": 0.659,
      "step": 887
    },
    {
      "epoch": 0.5729032258064516,
      "grad_norm": 0.86328125,
      "learning_rate": 8.996891867591969e-06,
      "loss": 0.8347,
      "step": 888
    },
    {
      "epoch": 0.5735483870967742,
      "grad_norm": 0.7734375,
      "learning_rate": 8.994379086685471e-06,
      "loss": 0.6228,
      "step": 889
    },
    {
      "epoch": 0.5741935483870968,
      "grad_norm": 0.78515625,
      "learning_rate": 8.991863514293015e-06,
      "loss": 0.7068,
      "step": 890
    },
    {
      "epoch": 0.5748387096774193,
      "grad_norm": 0.80078125,
      "learning_rate": 8.989345152172617e-06,
      "loss": 0.9219,
      "step": 891
    },
    {
      "epoch": 0.5754838709677419,
      "grad_norm": 0.66796875,
      "learning_rate": 8.986824002084242e-06,
      "loss": 0.6325,
      "step": 892
    },
    {
      "epoch": 0.5761290322580646,
      "grad_norm": 0.6328125,
      "learning_rate": 8.984300065789806e-06,
      "loss": 0.6315,
      "step": 893
    },
    {
      "epoch": 0.5767741935483871,
      "grad_norm": 1.015625,
      "learning_rate": 8.981773345053168e-06,
      "loss": 1.0669,
      "step": 894
    },
    {
      "epoch": 0.5774193548387097,
      "grad_norm": 1.0625,
      "learning_rate": 8.979243841640139e-06,
      "loss": 0.9856,
      "step": 895
    },
    {
      "epoch": 0.5780645161290323,
      "grad_norm": 1.03125,
      "learning_rate": 8.976711557318469e-06,
      "loss": 1.0389,
      "step": 896
    },
    {
      "epoch": 0.5787096774193549,
      "grad_norm": 0.85546875,
      "learning_rate": 8.974176493857856e-06,
      "loss": 0.7955,
      "step": 897
    },
    {
      "epoch": 0.5793548387096774,
      "grad_norm": 0.5859375,
      "learning_rate": 8.971638653029937e-06,
      "loss": 0.6987,
      "step": 898
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8828125,
      "learning_rate": 8.969098036608293e-06,
      "loss": 0.7679,
      "step": 899
    },
    {
      "epoch": 0.5806451612903226,
      "grad_norm": 1.203125,
      "learning_rate": 8.966554646368437e-06,
      "loss": 1.4047,
      "step": 900
    },
    {
      "epoch": 0.5812903225806452,
      "grad_norm": 0.93359375,
      "learning_rate": 8.964008484087831e-06,
      "loss": 0.7973,
      "step": 901
    },
    {
      "epoch": 0.5819354838709677,
      "grad_norm": 0.8984375,
      "learning_rate": 8.96145955154587e-06,
      "loss": 0.7199,
      "step": 902
    },
    {
      "epoch": 0.5825806451612904,
      "grad_norm": 0.67578125,
      "learning_rate": 8.958907850523884e-06,
      "loss": 0.7561,
      "step": 903
    },
    {
      "epoch": 0.5832258064516129,
      "grad_norm": 1.5390625,
      "learning_rate": 8.956353382805137e-06,
      "loss": 0.7062,
      "step": 904
    },
    {
      "epoch": 0.5838709677419355,
      "grad_norm": 0.87109375,
      "learning_rate": 8.953796150174826e-06,
      "loss": 0.7911,
      "step": 905
    },
    {
      "epoch": 0.584516129032258,
      "grad_norm": 0.671875,
      "learning_rate": 8.951236154420085e-06,
      "loss": 0.6476,
      "step": 906
    },
    {
      "epoch": 0.5851612903225807,
      "grad_norm": 0.66015625,
      "learning_rate": 8.948673397329976e-06,
      "loss": 0.66,
      "step": 907
    },
    {
      "epoch": 0.5858064516129032,
      "grad_norm": 0.796875,
      "learning_rate": 8.946107880695492e-06,
      "loss": 0.8166,
      "step": 908
    },
    {
      "epoch": 0.5864516129032258,
      "grad_norm": 1.125,
      "learning_rate": 8.943539606309548e-06,
      "loss": 1.0796,
      "step": 909
    },
    {
      "epoch": 0.5870967741935483,
      "grad_norm": 1.03125,
      "learning_rate": 8.940968575966999e-06,
      "loss": 1.0964,
      "step": 910
    },
    {
      "epoch": 0.587741935483871,
      "grad_norm": 1.09375,
      "learning_rate": 8.938394791464614e-06,
      "loss": 1.0459,
      "step": 911
    },
    {
      "epoch": 0.5883870967741935,
      "grad_norm": 0.7890625,
      "learning_rate": 8.93581825460109e-06,
      "loss": 0.6995,
      "step": 912
    },
    {
      "epoch": 0.5890322580645161,
      "grad_norm": 0.8984375,
      "learning_rate": 8.933238967177054e-06,
      "loss": 0.8532,
      "step": 913
    },
    {
      "epoch": 0.5896774193548387,
      "grad_norm": 1.234375,
      "learning_rate": 8.930656930995044e-06,
      "loss": 0.8459,
      "step": 914
    },
    {
      "epoch": 0.5903225806451613,
      "grad_norm": 1.65625,
      "learning_rate": 8.928072147859532e-06,
      "loss": 1.0208,
      "step": 915
    },
    {
      "epoch": 0.5909677419354838,
      "grad_norm": 0.859375,
      "learning_rate": 8.925484619576899e-06,
      "loss": 0.8871,
      "step": 916
    },
    {
      "epoch": 0.5916129032258064,
      "grad_norm": 1.3359375,
      "learning_rate": 8.922894347955447e-06,
      "loss": 1.4349,
      "step": 917
    },
    {
      "epoch": 0.5922580645161291,
      "grad_norm": 0.9765625,
      "learning_rate": 8.9203013348054e-06,
      "loss": 0.7191,
      "step": 918
    },
    {
      "epoch": 0.5929032258064516,
      "grad_norm": 0.984375,
      "learning_rate": 8.917705581938892e-06,
      "loss": 1.0234,
      "step": 919
    },
    {
      "epoch": 0.5935483870967742,
      "grad_norm": 0.80859375,
      "learning_rate": 8.915107091169972e-06,
      "loss": 0.7494,
      "step": 920
    },
    {
      "epoch": 0.5941935483870968,
      "grad_norm": 0.74609375,
      "learning_rate": 8.91250586431461e-06,
      "loss": 0.6842,
      "step": 921
    },
    {
      "epoch": 0.5948387096774194,
      "grad_norm": 1.328125,
      "learning_rate": 8.90990190319068e-06,
      "loss": 0.9538,
      "step": 922
    },
    {
      "epoch": 0.5954838709677419,
      "grad_norm": 1.3515625,
      "learning_rate": 8.907295209617967e-06,
      "loss": 1.0789,
      "step": 923
    },
    {
      "epoch": 0.5961290322580645,
      "grad_norm": 0.69140625,
      "learning_rate": 8.904685785418172e-06,
      "loss": 0.5581,
      "step": 924
    },
    {
      "epoch": 0.5967741935483871,
      "grad_norm": 1.5625,
      "learning_rate": 8.902073632414899e-06,
      "loss": 1.2787,
      "step": 925
    },
    {
      "epoch": 0.5974193548387097,
      "grad_norm": 1.328125,
      "learning_rate": 8.899458752433658e-06,
      "loss": 1.0605,
      "step": 926
    },
    {
      "epoch": 0.5980645161290322,
      "grad_norm": 0.74609375,
      "learning_rate": 8.896841147301869e-06,
      "loss": 0.69,
      "step": 927
    },
    {
      "epoch": 0.5987096774193549,
      "grad_norm": 0.83984375,
      "learning_rate": 8.894220818848856e-06,
      "loss": 0.6928,
      "step": 928
    },
    {
      "epoch": 0.5993548387096774,
      "grad_norm": 0.67578125,
      "learning_rate": 8.891597768905842e-06,
      "loss": 0.5992,
      "step": 929
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.71875,
      "learning_rate": 8.888971999305957e-06,
      "loss": 0.6905,
      "step": 930
    },
    {
      "epoch": 0.6006451612903225,
      "grad_norm": 1.15625,
      "learning_rate": 8.88634351188423e-06,
      "loss": 1.1898,
      "step": 931
    },
    {
      "epoch": 0.6012903225806452,
      "grad_norm": 1.0078125,
      "learning_rate": 8.883712308477585e-06,
      "loss": 0.7584,
      "step": 932
    },
    {
      "epoch": 0.6019354838709677,
      "grad_norm": 1.1875,
      "learning_rate": 8.881078390924852e-06,
      "loss": 1.1425,
      "step": 933
    },
    {
      "epoch": 0.6025806451612903,
      "grad_norm": 0.71875,
      "learning_rate": 8.87844176106675e-06,
      "loss": 0.8032,
      "step": 934
    },
    {
      "epoch": 0.603225806451613,
      "grad_norm": 0.95703125,
      "learning_rate": 8.8758024207459e-06,
      "loss": 0.8734,
      "step": 935
    },
    {
      "epoch": 0.6038709677419355,
      "grad_norm": 0.59375,
      "learning_rate": 8.873160371806812e-06,
      "loss": 0.4432,
      "step": 936
    },
    {
      "epoch": 0.604516129032258,
      "grad_norm": 1.0546875,
      "learning_rate": 8.870515616095893e-06,
      "loss": 1.1441,
      "step": 937
    },
    {
      "epoch": 0.6051612903225806,
      "grad_norm": 1.9375,
      "learning_rate": 8.867868155461438e-06,
      "loss": 1.1776,
      "step": 938
    },
    {
      "epoch": 0.6058064516129033,
      "grad_norm": 0.703125,
      "learning_rate": 8.865217991753638e-06,
      "loss": 0.6944,
      "step": 939
    },
    {
      "epoch": 0.6064516129032258,
      "grad_norm": 0.7109375,
      "learning_rate": 8.862565126824564e-06,
      "loss": 0.6879,
      "step": 940
    },
    {
      "epoch": 0.6070967741935483,
      "grad_norm": 0.90234375,
      "learning_rate": 8.859909562528186e-06,
      "loss": 0.7558,
      "step": 941
    },
    {
      "epoch": 0.607741935483871,
      "grad_norm": 1.125,
      "learning_rate": 8.85725130072035e-06,
      "loss": 0.8517,
      "step": 942
    },
    {
      "epoch": 0.6083870967741936,
      "grad_norm": 1.1953125,
      "learning_rate": 8.854590343258792e-06,
      "loss": 0.9401,
      "step": 943
    },
    {
      "epoch": 0.6090322580645161,
      "grad_norm": 0.80078125,
      "learning_rate": 8.851926692003133e-06,
      "loss": 0.6967,
      "step": 944
    },
    {
      "epoch": 0.6096774193548387,
      "grad_norm": 0.84765625,
      "learning_rate": 8.849260348814875e-06,
      "loss": 0.8256,
      "step": 945
    },
    {
      "epoch": 0.6103225806451613,
      "grad_norm": 1.03125,
      "learning_rate": 8.846591315557401e-06,
      "loss": 0.8459,
      "step": 946
    },
    {
      "epoch": 0.6109677419354839,
      "grad_norm": 1.1328125,
      "learning_rate": 8.843919594095974e-06,
      "loss": 1.0278,
      "step": 947
    },
    {
      "epoch": 0.6116129032258064,
      "grad_norm": 0.99609375,
      "learning_rate": 8.841245186297737e-06,
      "loss": 0.989,
      "step": 948
    },
    {
      "epoch": 0.6122580645161291,
      "grad_norm": 0.92578125,
      "learning_rate": 8.838568094031709e-06,
      "loss": 1.0412,
      "step": 949
    },
    {
      "epoch": 0.6129032258064516,
      "grad_norm": 1.0859375,
      "learning_rate": 8.835888319168784e-06,
      "loss": 0.726,
      "step": 950
    },
    {
      "epoch": 0.6135483870967742,
      "grad_norm": 0.4765625,
      "learning_rate": 8.833205863581732e-06,
      "loss": 0.4178,
      "step": 951
    },
    {
      "epoch": 0.6141935483870967,
      "grad_norm": 1.0390625,
      "learning_rate": 8.830520729145201e-06,
      "loss": 0.9979,
      "step": 952
    },
    {
      "epoch": 0.6148387096774194,
      "grad_norm": 1.5625,
      "learning_rate": 8.827832917735702e-06,
      "loss": 1.0622,
      "step": 953
    },
    {
      "epoch": 0.6154838709677419,
      "grad_norm": 0.89453125,
      "learning_rate": 8.825142431231624e-06,
      "loss": 0.7765,
      "step": 954
    },
    {
      "epoch": 0.6161290322580645,
      "grad_norm": 0.87890625,
      "learning_rate": 8.82244927151322e-06,
      "loss": 0.7275,
      "step": 955
    },
    {
      "epoch": 0.6167741935483871,
      "grad_norm": 0.72265625,
      "learning_rate": 8.819753440462618e-06,
      "loss": 0.6022,
      "step": 956
    },
    {
      "epoch": 0.6174193548387097,
      "grad_norm": 0.70703125,
      "learning_rate": 8.817054939963807e-06,
      "loss": 0.6766,
      "step": 957
    },
    {
      "epoch": 0.6180645161290322,
      "grad_norm": 0.8359375,
      "learning_rate": 8.814353771902644e-06,
      "loss": 0.6868,
      "step": 958
    },
    {
      "epoch": 0.6187096774193548,
      "grad_norm": 0.984375,
      "learning_rate": 8.811649938166848e-06,
      "loss": 1.0122,
      "step": 959
    },
    {
      "epoch": 0.6193548387096774,
      "grad_norm": 1.390625,
      "learning_rate": 8.808943440646008e-06,
      "loss": 0.9405,
      "step": 960
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.859375,
      "learning_rate": 8.806234281231563e-06,
      "loss": 0.5982,
      "step": 961
    },
    {
      "epoch": 0.6206451612903225,
      "grad_norm": 1.359375,
      "learning_rate": 8.803522461816823e-06,
      "loss": 1.0529,
      "step": 962
    },
    {
      "epoch": 0.6212903225806452,
      "grad_norm": 0.93359375,
      "learning_rate": 8.80080798429695e-06,
      "loss": 0.7856,
      "step": 963
    },
    {
      "epoch": 0.6219354838709678,
      "grad_norm": 0.890625,
      "learning_rate": 8.79809085056897e-06,
      "loss": 0.7321,
      "step": 964
    },
    {
      "epoch": 0.6225806451612903,
      "grad_norm": 1.1875,
      "learning_rate": 8.795371062531756e-06,
      "loss": 0.9668,
      "step": 965
    },
    {
      "epoch": 0.6232258064516129,
      "grad_norm": 0.921875,
      "learning_rate": 8.792648622086046e-06,
      "loss": 0.8284,
      "step": 966
    },
    {
      "epoch": 0.6238709677419355,
      "grad_norm": 0.93359375,
      "learning_rate": 8.789923531134425e-06,
      "loss": 0.7522,
      "step": 967
    },
    {
      "epoch": 0.6245161290322581,
      "grad_norm": 0.75,
      "learning_rate": 8.787195791581333e-06,
      "loss": 0.7028,
      "step": 968
    },
    {
      "epoch": 0.6251612903225806,
      "grad_norm": 0.921875,
      "learning_rate": 8.784465405333061e-06,
      "loss": 0.659,
      "step": 969
    },
    {
      "epoch": 0.6258064516129033,
      "grad_norm": 1.25,
      "learning_rate": 8.781732374297749e-06,
      "loss": 1.1778,
      "step": 970
    },
    {
      "epoch": 0.6264516129032258,
      "grad_norm": 0.76953125,
      "learning_rate": 8.778996700385385e-06,
      "loss": 0.8352,
      "step": 971
    },
    {
      "epoch": 0.6270967741935484,
      "grad_norm": 1.8359375,
      "learning_rate": 8.776258385507802e-06,
      "loss": 1.325,
      "step": 972
    },
    {
      "epoch": 0.6277419354838709,
      "grad_norm": 0.796875,
      "learning_rate": 8.773517431578687e-06,
      "loss": 0.9435,
      "step": 973
    },
    {
      "epoch": 0.6283870967741936,
      "grad_norm": 0.78515625,
      "learning_rate": 8.770773840513559e-06,
      "loss": 0.5668,
      "step": 974
    },
    {
      "epoch": 0.6290322580645161,
      "grad_norm": 0.6640625,
      "learning_rate": 8.768027614229791e-06,
      "loss": 0.6064,
      "step": 975
    },
    {
      "epoch": 0.6296774193548387,
      "grad_norm": 0.79296875,
      "learning_rate": 8.76527875464659e-06,
      "loss": 0.9469,
      "step": 976
    },
    {
      "epoch": 0.6303225806451613,
      "grad_norm": 0.6328125,
      "learning_rate": 8.762527263685006e-06,
      "loss": 0.632,
      "step": 977
    },
    {
      "epoch": 0.6309677419354839,
      "grad_norm": 1.375,
      "learning_rate": 8.759773143267932e-06,
      "loss": 1.7264,
      "step": 978
    },
    {
      "epoch": 0.6316129032258064,
      "grad_norm": 0.84375,
      "learning_rate": 8.75701639532009e-06,
      "loss": 0.6887,
      "step": 979
    },
    {
      "epoch": 0.632258064516129,
      "grad_norm": 0.9765625,
      "learning_rate": 8.754257021768046e-06,
      "loss": 0.7302,
      "step": 980
    },
    {
      "epoch": 0.6329032258064516,
      "grad_norm": 0.82421875,
      "learning_rate": 8.751495024540197e-06,
      "loss": 0.7089,
      "step": 981
    },
    {
      "epoch": 0.6335483870967742,
      "grad_norm": 0.8828125,
      "learning_rate": 8.748730405566777e-06,
      "loss": 0.771,
      "step": 982
    },
    {
      "epoch": 0.6341935483870967,
      "grad_norm": 0.90625,
      "learning_rate": 8.745963166779845e-06,
      "loss": 0.8107,
      "step": 983
    },
    {
      "epoch": 0.6348387096774194,
      "grad_norm": 0.490234375,
      "learning_rate": 8.743193310113301e-06,
      "loss": 0.4881,
      "step": 984
    },
    {
      "epoch": 0.635483870967742,
      "grad_norm": 0.85546875,
      "learning_rate": 8.740420837502863e-06,
      "loss": 0.7582,
      "step": 985
    },
    {
      "epoch": 0.6361290322580645,
      "grad_norm": 0.890625,
      "learning_rate": 8.737645750886087e-06,
      "loss": 0.9412,
      "step": 986
    },
    {
      "epoch": 0.636774193548387,
      "grad_norm": 0.6953125,
      "learning_rate": 8.734868052202355e-06,
      "loss": 0.6653,
      "step": 987
    },
    {
      "epoch": 0.6374193548387097,
      "grad_norm": 0.97265625,
      "learning_rate": 8.732087743392866e-06,
      "loss": 1.0951,
      "step": 988
    },
    {
      "epoch": 0.6380645161290323,
      "grad_norm": 1.109375,
      "learning_rate": 8.729304826400652e-06,
      "loss": 0.8734,
      "step": 989
    },
    {
      "epoch": 0.6387096774193548,
      "grad_norm": 1.03125,
      "learning_rate": 8.726519303170562e-06,
      "loss": 0.8316,
      "step": 990
    },
    {
      "epoch": 0.6393548387096775,
      "grad_norm": 0.73046875,
      "learning_rate": 8.72373117564927e-06,
      "loss": 0.7182,
      "step": 991
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.69140625,
      "learning_rate": 8.72094044578527e-06,
      "loss": 0.5725,
      "step": 992
    },
    {
      "epoch": 0.6406451612903226,
      "grad_norm": 1.3203125,
      "learning_rate": 8.718147115528872e-06,
      "loss": 1.1901,
      "step": 993
    },
    {
      "epoch": 0.6412903225806451,
      "grad_norm": 0.99609375,
      "learning_rate": 8.715351186832206e-06,
      "loss": 0.9509,
      "step": 994
    },
    {
      "epoch": 0.6419354838709678,
      "grad_norm": 1.03125,
      "learning_rate": 8.712552661649215e-06,
      "loss": 0.6699,
      "step": 995
    },
    {
      "epoch": 0.6425806451612903,
      "grad_norm": 0.95703125,
      "learning_rate": 8.709751541935657e-06,
      "loss": 1.1529,
      "step": 996
    },
    {
      "epoch": 0.6432258064516129,
      "grad_norm": 0.6640625,
      "learning_rate": 8.706947829649109e-06,
      "loss": 0.7405,
      "step": 997
    },
    {
      "epoch": 0.6438709677419355,
      "grad_norm": 0.62890625,
      "learning_rate": 8.704141526748952e-06,
      "loss": 0.6376,
      "step": 998
    },
    {
      "epoch": 0.6445161290322581,
      "grad_norm": 0.734375,
      "learning_rate": 8.701332635196382e-06,
      "loss": 0.8255,
      "step": 999
    },
    {
      "epoch": 0.6451612903225806,
      "grad_norm": 0.83203125,
      "learning_rate": 8.698521156954401e-06,
      "loss": 0.7898,
      "step": 1000
    },
    {
      "epoch": 0.6458064516129032,
      "grad_norm": 0.8515625,
      "learning_rate": 8.695707093987822e-06,
      "loss": 0.6334,
      "step": 1001
    },
    {
      "epoch": 0.6464516129032258,
      "grad_norm": 0.71484375,
      "learning_rate": 8.692890448263263e-06,
      "loss": 0.6362,
      "step": 1002
    },
    {
      "epoch": 0.6470967741935484,
      "grad_norm": 0.984375,
      "learning_rate": 8.690071221749145e-06,
      "loss": 0.8275,
      "step": 1003
    },
    {
      "epoch": 0.6477419354838709,
      "grad_norm": 1.0,
      "learning_rate": 8.687249416415698e-06,
      "loss": 0.7872,
      "step": 1004
    },
    {
      "epoch": 0.6483870967741936,
      "grad_norm": 0.796875,
      "learning_rate": 8.684425034234945e-06,
      "loss": 0.678,
      "step": 1005
    },
    {
      "epoch": 0.6490322580645161,
      "grad_norm": 0.91796875,
      "learning_rate": 8.681598077180722e-06,
      "loss": 0.823,
      "step": 1006
    },
    {
      "epoch": 0.6496774193548387,
      "grad_norm": 1.3203125,
      "learning_rate": 8.678768547228652e-06,
      "loss": 0.9808,
      "step": 1007
    },
    {
      "epoch": 0.6503225806451612,
      "grad_norm": 0.81640625,
      "learning_rate": 8.675936446356168e-06,
      "loss": 1.0269,
      "step": 1008
    },
    {
      "epoch": 0.6509677419354839,
      "grad_norm": 0.66796875,
      "learning_rate": 8.67310177654249e-06,
      "loss": 0.5568,
      "step": 1009
    },
    {
      "epoch": 0.6516129032258065,
      "grad_norm": 1.109375,
      "learning_rate": 8.670264539768636e-06,
      "loss": 1.0549,
      "step": 1010
    },
    {
      "epoch": 0.652258064516129,
      "grad_norm": 0.7265625,
      "learning_rate": 8.667424738017422e-06,
      "loss": 0.6595,
      "step": 1011
    },
    {
      "epoch": 0.6529032258064517,
      "grad_norm": 0.9765625,
      "learning_rate": 8.664582373273451e-06,
      "loss": 0.851,
      "step": 1012
    },
    {
      "epoch": 0.6535483870967742,
      "grad_norm": 0.87109375,
      "learning_rate": 8.661737447523122e-06,
      "loss": 0.7575,
      "step": 1013
    },
    {
      "epoch": 0.6541935483870968,
      "grad_norm": 1.0078125,
      "learning_rate": 8.658889962754618e-06,
      "loss": 0.8465,
      "step": 1014
    },
    {
      "epoch": 0.6548387096774193,
      "grad_norm": 0.75390625,
      "learning_rate": 8.656039920957919e-06,
      "loss": 0.676,
      "step": 1015
    },
    {
      "epoch": 0.655483870967742,
      "grad_norm": 1.1640625,
      "learning_rate": 8.65318732412478e-06,
      "loss": 1.2576,
      "step": 1016
    },
    {
      "epoch": 0.6561290322580645,
      "grad_norm": 0.67578125,
      "learning_rate": 8.650332174248755e-06,
      "loss": 0.5009,
      "step": 1017
    },
    {
      "epoch": 0.6567741935483871,
      "grad_norm": 0.53125,
      "learning_rate": 8.647474473325172e-06,
      "loss": 0.4783,
      "step": 1018
    },
    {
      "epoch": 0.6574193548387097,
      "grad_norm": 0.80859375,
      "learning_rate": 8.644614223351147e-06,
      "loss": 0.7829,
      "step": 1019
    },
    {
      "epoch": 0.6580645161290323,
      "grad_norm": 0.87109375,
      "learning_rate": 8.641751426325574e-06,
      "loss": 0.8274,
      "step": 1020
    },
    {
      "epoch": 0.6587096774193548,
      "grad_norm": 1.375,
      "learning_rate": 8.638886084249129e-06,
      "loss": 1.1104,
      "step": 1021
    },
    {
      "epoch": 0.6593548387096774,
      "grad_norm": 1.0,
      "learning_rate": 8.63601819912427e-06,
      "loss": 1.1207,
      "step": 1022
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.83203125,
      "learning_rate": 8.633147772955224e-06,
      "loss": 0.7705,
      "step": 1023
    },
    {
      "epoch": 0.6606451612903226,
      "grad_norm": 0.66796875,
      "learning_rate": 8.630274807748e-06,
      "loss": 0.7613,
      "step": 1024
    },
    {
      "epoch": 0.6612903225806451,
      "grad_norm": 0.8984375,
      "learning_rate": 8.627399305510381e-06,
      "loss": 0.7417,
      "step": 1025
    },
    {
      "epoch": 0.6619354838709678,
      "grad_norm": 0.55859375,
      "learning_rate": 8.624521268251922e-06,
      "loss": 0.5236,
      "step": 1026
    },
    {
      "epoch": 0.6625806451612903,
      "grad_norm": 0.89453125,
      "learning_rate": 8.621640697983946e-06,
      "loss": 1.0376,
      "step": 1027
    },
    {
      "epoch": 0.6632258064516129,
      "grad_norm": 1.0625,
      "learning_rate": 8.618757596719556e-06,
      "loss": 1.2428,
      "step": 1028
    },
    {
      "epoch": 0.6638709677419354,
      "grad_norm": 1.234375,
      "learning_rate": 8.615871966473613e-06,
      "loss": 0.8877,
      "step": 1029
    },
    {
      "epoch": 0.6645161290322581,
      "grad_norm": 1.109375,
      "learning_rate": 8.61298380926275e-06,
      "loss": 0.9302,
      "step": 1030
    },
    {
      "epoch": 0.6651612903225806,
      "grad_norm": 0.85546875,
      "learning_rate": 8.610093127105366e-06,
      "loss": 0.8337,
      "step": 1031
    },
    {
      "epoch": 0.6658064516129032,
      "grad_norm": 0.7265625,
      "learning_rate": 8.607199922021626e-06,
      "loss": 0.5889,
      "step": 1032
    },
    {
      "epoch": 0.6664516129032259,
      "grad_norm": 0.953125,
      "learning_rate": 8.604304196033457e-06,
      "loss": 0.9814,
      "step": 1033
    },
    {
      "epoch": 0.6670967741935484,
      "grad_norm": 0.5859375,
      "learning_rate": 8.601405951164546e-06,
      "loss": 0.6436,
      "step": 1034
    },
    {
      "epoch": 0.667741935483871,
      "grad_norm": 1.1328125,
      "learning_rate": 8.598505189440341e-06,
      "loss": 0.8361,
      "step": 1035
    },
    {
      "epoch": 0.6683870967741935,
      "grad_norm": 0.95703125,
      "learning_rate": 8.59560191288805e-06,
      "loss": 0.9134,
      "step": 1036
    },
    {
      "epoch": 0.6690322580645162,
      "grad_norm": 1.171875,
      "learning_rate": 8.592696123536639e-06,
      "loss": 0.9736,
      "step": 1037
    },
    {
      "epoch": 0.6696774193548387,
      "grad_norm": 0.77734375,
      "learning_rate": 8.589787823416828e-06,
      "loss": 0.9089,
      "step": 1038
    },
    {
      "epoch": 0.6703225806451613,
      "grad_norm": 1.0703125,
      "learning_rate": 8.586877014561094e-06,
      "loss": 1.0622,
      "step": 1039
    },
    {
      "epoch": 0.6709677419354839,
      "grad_norm": 1.0,
      "learning_rate": 8.583963699003669e-06,
      "loss": 1.1187,
      "step": 1040
    },
    {
      "epoch": 0.6716129032258065,
      "grad_norm": 1.0859375,
      "learning_rate": 8.581047878780526e-06,
      "loss": 0.983,
      "step": 1041
    },
    {
      "epoch": 0.672258064516129,
      "grad_norm": 0.62109375,
      "learning_rate": 8.578129555929403e-06,
      "loss": 0.5269,
      "step": 1042
    },
    {
      "epoch": 0.6729032258064516,
      "grad_norm": 0.7578125,
      "learning_rate": 8.575208732489779e-06,
      "loss": 0.7105,
      "step": 1043
    },
    {
      "epoch": 0.6735483870967742,
      "grad_norm": 1.1171875,
      "learning_rate": 8.572285410502878e-06,
      "loss": 1.0474,
      "step": 1044
    },
    {
      "epoch": 0.6741935483870968,
      "grad_norm": 0.79296875,
      "learning_rate": 8.569359592011678e-06,
      "loss": 0.6184,
      "step": 1045
    },
    {
      "epoch": 0.6748387096774193,
      "grad_norm": 0.80859375,
      "learning_rate": 8.566431279060898e-06,
      "loss": 1.0573,
      "step": 1046
    },
    {
      "epoch": 0.675483870967742,
      "grad_norm": 0.671875,
      "learning_rate": 8.563500473696996e-06,
      "loss": 0.49,
      "step": 1047
    },
    {
      "epoch": 0.6761290322580645,
      "grad_norm": 0.875,
      "learning_rate": 8.56056717796818e-06,
      "loss": 1.0891,
      "step": 1048
    },
    {
      "epoch": 0.6767741935483871,
      "grad_norm": 0.91796875,
      "learning_rate": 8.55763139392439e-06,
      "loss": 0.8351,
      "step": 1049
    },
    {
      "epoch": 0.6774193548387096,
      "grad_norm": 1.28125,
      "learning_rate": 8.554693123617312e-06,
      "loss": 0.8969,
      "step": 1050
    },
    {
      "epoch": 0.6780645161290323,
      "grad_norm": 0.6640625,
      "learning_rate": 8.551752369100365e-06,
      "loss": 0.499,
      "step": 1051
    },
    {
      "epoch": 0.6787096774193548,
      "grad_norm": 0.8046875,
      "learning_rate": 8.548809132428709e-06,
      "loss": 0.8454,
      "step": 1052
    },
    {
      "epoch": 0.6793548387096774,
      "grad_norm": 0.734375,
      "learning_rate": 8.54586341565923e-06,
      "loss": 0.6919,
      "step": 1053
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1328125,
      "learning_rate": 8.542915220850556e-06,
      "loss": 0.6875,
      "step": 1054
    },
    {
      "epoch": 0.6806451612903226,
      "grad_norm": 0.80078125,
      "learning_rate": 8.539964550063045e-06,
      "loss": 0.6555,
      "step": 1055
    },
    {
      "epoch": 0.6812903225806451,
      "grad_norm": 0.72265625,
      "learning_rate": 8.537011405358779e-06,
      "loss": 0.7627,
      "step": 1056
    },
    {
      "epoch": 0.6819354838709677,
      "grad_norm": 0.71875,
      "learning_rate": 8.534055788801576e-06,
      "loss": 0.5813,
      "step": 1057
    },
    {
      "epoch": 0.6825806451612904,
      "grad_norm": 0.66015625,
      "learning_rate": 8.53109770245698e-06,
      "loss": 0.5612,
      "step": 1058
    },
    {
      "epoch": 0.6832258064516129,
      "grad_norm": 1.2578125,
      "learning_rate": 8.52813714839226e-06,
      "loss": 0.9909,
      "step": 1059
    },
    {
      "epoch": 0.6838709677419355,
      "grad_norm": 0.91796875,
      "learning_rate": 8.525174128676409e-06,
      "loss": 0.6038,
      "step": 1060
    },
    {
      "epoch": 0.6845161290322581,
      "grad_norm": 0.93359375,
      "learning_rate": 8.522208645380142e-06,
      "loss": 0.8907,
      "step": 1061
    },
    {
      "epoch": 0.6851612903225807,
      "grad_norm": 0.64453125,
      "learning_rate": 8.5192407005759e-06,
      "loss": 0.6584,
      "step": 1062
    },
    {
      "epoch": 0.6858064516129032,
      "grad_norm": 0.89453125,
      "learning_rate": 8.51627029633784e-06,
      "loss": 0.8781,
      "step": 1063
    },
    {
      "epoch": 0.6864516129032258,
      "grad_norm": 0.80078125,
      "learning_rate": 8.513297434741843e-06,
      "loss": 0.8219,
      "step": 1064
    },
    {
      "epoch": 0.6870967741935484,
      "grad_norm": 0.7265625,
      "learning_rate": 8.5103221178655e-06,
      "loss": 0.6568,
      "step": 1065
    },
    {
      "epoch": 0.687741935483871,
      "grad_norm": 0.76953125,
      "learning_rate": 8.507344347788123e-06,
      "loss": 0.6885,
      "step": 1066
    },
    {
      "epoch": 0.6883870967741935,
      "grad_norm": 0.66796875,
      "learning_rate": 8.504364126590739e-06,
      "loss": 0.5677,
      "step": 1067
    },
    {
      "epoch": 0.6890322580645162,
      "grad_norm": 0.78515625,
      "learning_rate": 8.501381456356081e-06,
      "loss": 0.7589,
      "step": 1068
    },
    {
      "epoch": 0.6896774193548387,
      "grad_norm": 0.828125,
      "learning_rate": 8.498396339168605e-06,
      "loss": 0.7763,
      "step": 1069
    },
    {
      "epoch": 0.6903225806451613,
      "grad_norm": 0.80078125,
      "learning_rate": 8.495408777114467e-06,
      "loss": 0.812,
      "step": 1070
    },
    {
      "epoch": 0.6909677419354838,
      "grad_norm": 1.21875,
      "learning_rate": 8.492418772281534e-06,
      "loss": 1.0384,
      "step": 1071
    },
    {
      "epoch": 0.6916129032258065,
      "grad_norm": 0.9921875,
      "learning_rate": 8.489426326759387e-06,
      "loss": 0.9307,
      "step": 1072
    },
    {
      "epoch": 0.692258064516129,
      "grad_norm": 0.80078125,
      "learning_rate": 8.4864314426393e-06,
      "loss": 0.8132,
      "step": 1073
    },
    {
      "epoch": 0.6929032258064516,
      "grad_norm": 0.8515625,
      "learning_rate": 8.483434122014265e-06,
      "loss": 0.915,
      "step": 1074
    },
    {
      "epoch": 0.6935483870967742,
      "grad_norm": 0.94921875,
      "learning_rate": 8.480434366978966e-06,
      "loss": 0.9699,
      "step": 1075
    },
    {
      "epoch": 0.6941935483870968,
      "grad_norm": 1.0078125,
      "learning_rate": 8.477432179629795e-06,
      "loss": 0.9157,
      "step": 1076
    },
    {
      "epoch": 0.6948387096774193,
      "grad_norm": 0.7109375,
      "learning_rate": 8.474427562064838e-06,
      "loss": 0.6849,
      "step": 1077
    },
    {
      "epoch": 0.6954838709677419,
      "grad_norm": 1.375,
      "learning_rate": 8.471420516383883e-06,
      "loss": 1.0175,
      "step": 1078
    },
    {
      "epoch": 0.6961290322580646,
      "grad_norm": 1.375,
      "learning_rate": 8.46841104468842e-06,
      "loss": 1.0596,
      "step": 1079
    },
    {
      "epoch": 0.6967741935483871,
      "grad_norm": 0.73828125,
      "learning_rate": 8.465399149081623e-06,
      "loss": 0.4409,
      "step": 1080
    },
    {
      "epoch": 0.6974193548387096,
      "grad_norm": 0.94921875,
      "learning_rate": 8.462384831668366e-06,
      "loss": 1.0136,
      "step": 1081
    },
    {
      "epoch": 0.6980645161290323,
      "grad_norm": 0.79296875,
      "learning_rate": 8.459368094555221e-06,
      "loss": 0.8044,
      "step": 1082
    },
    {
      "epoch": 0.6987096774193549,
      "grad_norm": 1.0390625,
      "learning_rate": 8.456348939850438e-06,
      "loss": 0.9542,
      "step": 1083
    },
    {
      "epoch": 0.6993548387096774,
      "grad_norm": 0.82421875,
      "learning_rate": 8.45332736966397e-06,
      "loss": 0.787,
      "step": 1084
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.921875,
      "learning_rate": 8.450303386107447e-06,
      "loss": 0.8124,
      "step": 1085
    },
    {
      "epoch": 0.7006451612903226,
      "grad_norm": 0.78125,
      "learning_rate": 8.447276991294195e-06,
      "loss": 0.6445,
      "step": 1086
    },
    {
      "epoch": 0.7012903225806452,
      "grad_norm": 1.1171875,
      "learning_rate": 8.44424818733922e-06,
      "loss": 0.8605,
      "step": 1087
    },
    {
      "epoch": 0.7019354838709677,
      "grad_norm": 0.6875,
      "learning_rate": 8.441216976359208e-06,
      "loss": 0.7615,
      "step": 1088
    },
    {
      "epoch": 0.7025806451612904,
      "grad_norm": 1.046875,
      "learning_rate": 8.438183360472538e-06,
      "loss": 1.0108,
      "step": 1089
    },
    {
      "epoch": 0.7032258064516129,
      "grad_norm": 1.0234375,
      "learning_rate": 8.43514734179926e-06,
      "loss": 0.6543,
      "step": 1090
    },
    {
      "epoch": 0.7038709677419355,
      "grad_norm": 0.9453125,
      "learning_rate": 8.432108922461104e-06,
      "loss": 0.5568,
      "step": 1091
    },
    {
      "epoch": 0.704516129032258,
      "grad_norm": 0.98828125,
      "learning_rate": 8.429068104581482e-06,
      "loss": 0.9797,
      "step": 1092
    },
    {
      "epoch": 0.7051612903225807,
      "grad_norm": 0.87109375,
      "learning_rate": 8.426024890285483e-06,
      "loss": 0.8777,
      "step": 1093
    },
    {
      "epoch": 0.7058064516129032,
      "grad_norm": 1.078125,
      "learning_rate": 8.422979281699866e-06,
      "loss": 0.9714,
      "step": 1094
    },
    {
      "epoch": 0.7064516129032258,
      "grad_norm": 0.6796875,
      "learning_rate": 8.419931280953065e-06,
      "loss": 0.6061,
      "step": 1095
    },
    {
      "epoch": 0.7070967741935484,
      "grad_norm": 0.7890625,
      "learning_rate": 8.416880890175186e-06,
      "loss": 0.7569,
      "step": 1096
    },
    {
      "epoch": 0.707741935483871,
      "grad_norm": 0.77734375,
      "learning_rate": 8.413828111498005e-06,
      "loss": 0.7367,
      "step": 1097
    },
    {
      "epoch": 0.7083870967741935,
      "grad_norm": 1.3125,
      "learning_rate": 8.410772947054968e-06,
      "loss": 1.0856,
      "step": 1098
    },
    {
      "epoch": 0.7090322580645161,
      "grad_norm": 0.71875,
      "learning_rate": 8.407715398981187e-06,
      "loss": 0.7676,
      "step": 1099
    },
    {
      "epoch": 0.7096774193548387,
      "grad_norm": 0.91796875,
      "learning_rate": 8.404655469413443e-06,
      "loss": 0.9622,
      "step": 1100
    },
    {
      "epoch": 0.7096774193548387,
      "eval_loss": 0.9177858233451843,
      "eval_model_preparation_time": 0.0144,
      "eval_runtime": 81.8651,
      "eval_samples_per_second": 3.64,
      "eval_steps_per_second": 3.64,
      "step": 1100
    },
    {
      "epoch": 0.7103225806451613,
      "grad_norm": 0.72265625,
      "learning_rate": 8.401593160490174e-06,
      "loss": 0.6701,
      "step": 1101
    },
    {
      "epoch": 0.7109677419354838,
      "grad_norm": 0.87109375,
      "learning_rate": 8.398528474351489e-06,
      "loss": 0.9186,
      "step": 1102
    },
    {
      "epoch": 0.7116129032258065,
      "grad_norm": 1.1015625,
      "learning_rate": 8.395461413139154e-06,
      "loss": 0.7959,
      "step": 1103
    },
    {
      "epoch": 0.712258064516129,
      "grad_norm": 0.76953125,
      "learning_rate": 8.392391978996594e-06,
      "loss": 0.681,
      "step": 1104
    },
    {
      "epoch": 0.7129032258064516,
      "grad_norm": 0.90234375,
      "learning_rate": 8.389320174068895e-06,
      "loss": 1.0732,
      "step": 1105
    },
    {
      "epoch": 0.7135483870967742,
      "grad_norm": 1.1640625,
      "learning_rate": 8.386246000502797e-06,
      "loss": 1.0395,
      "step": 1106
    },
    {
      "epoch": 0.7141935483870968,
      "grad_norm": 1.1640625,
      "learning_rate": 8.383169460446702e-06,
      "loss": 0.9993,
      "step": 1107
    },
    {
      "epoch": 0.7148387096774194,
      "grad_norm": 0.953125,
      "learning_rate": 8.380090556050656e-06,
      "loss": 1.0364,
      "step": 1108
    },
    {
      "epoch": 0.7154838709677419,
      "grad_norm": 0.73046875,
      "learning_rate": 8.377009289466365e-06,
      "loss": 0.5472,
      "step": 1109
    },
    {
      "epoch": 0.7161290322580646,
      "grad_norm": 0.94140625,
      "learning_rate": 8.37392566284718e-06,
      "loss": 0.7968,
      "step": 1110
    },
    {
      "epoch": 0.7167741935483871,
      "grad_norm": 0.84375,
      "learning_rate": 8.370839678348109e-06,
      "loss": 0.6287,
      "step": 1111
    },
    {
      "epoch": 0.7174193548387097,
      "grad_norm": 0.53515625,
      "learning_rate": 8.367751338125799e-06,
      "loss": 0.4448,
      "step": 1112
    },
    {
      "epoch": 0.7180645161290322,
      "grad_norm": 0.62109375,
      "learning_rate": 8.36466064433855e-06,
      "loss": 0.62,
      "step": 1113
    },
    {
      "epoch": 0.7187096774193549,
      "grad_norm": 0.8671875,
      "learning_rate": 8.3615675991463e-06,
      "loss": 0.819,
      "step": 1114
    },
    {
      "epoch": 0.7193548387096774,
      "grad_norm": 1.2421875,
      "learning_rate": 8.358472204710641e-06,
      "loss": 1.0411,
      "step": 1115
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0546875,
      "learning_rate": 8.355374463194797e-06,
      "loss": 0.9455,
      "step": 1116
    },
    {
      "epoch": 0.7206451612903226,
      "grad_norm": 0.8203125,
      "learning_rate": 8.352274376763635e-06,
      "loss": 0.7922,
      "step": 1117
    },
    {
      "epoch": 0.7212903225806452,
      "grad_norm": 0.83203125,
      "learning_rate": 8.349171947583663e-06,
      "loss": 0.9258,
      "step": 1118
    },
    {
      "epoch": 0.7219354838709677,
      "grad_norm": 0.640625,
      "learning_rate": 8.346067177823023e-06,
      "loss": 0.6618,
      "step": 1119
    },
    {
      "epoch": 0.7225806451612903,
      "grad_norm": 0.90625,
      "learning_rate": 8.342960069651496e-06,
      "loss": 0.8665,
      "step": 1120
    },
    {
      "epoch": 0.7232258064516129,
      "grad_norm": 0.671875,
      "learning_rate": 8.339850625240495e-06,
      "loss": 0.6496,
      "step": 1121
    },
    {
      "epoch": 0.7238709677419355,
      "grad_norm": 0.76171875,
      "learning_rate": 8.336738846763066e-06,
      "loss": 0.6473,
      "step": 1122
    },
    {
      "epoch": 0.724516129032258,
      "grad_norm": 0.8046875,
      "learning_rate": 8.333624736393886e-06,
      "loss": 0.7035,
      "step": 1123
    },
    {
      "epoch": 0.7251612903225807,
      "grad_norm": 0.58984375,
      "learning_rate": 8.330508296309262e-06,
      "loss": 0.4797,
      "step": 1124
    },
    {
      "epoch": 0.7258064516129032,
      "grad_norm": 0.75,
      "learning_rate": 8.32738952868713e-06,
      "loss": 0.8225,
      "step": 1125
    },
    {
      "epoch": 0.7264516129032258,
      "grad_norm": 0.828125,
      "learning_rate": 8.32426843570705e-06,
      "loss": 0.7186,
      "step": 1126
    },
    {
      "epoch": 0.7270967741935483,
      "grad_norm": 0.55859375,
      "learning_rate": 8.321145019550212e-06,
      "loss": 0.4653,
      "step": 1127
    },
    {
      "epoch": 0.727741935483871,
      "grad_norm": 1.21875,
      "learning_rate": 8.318019282399424e-06,
      "loss": 1.1105,
      "step": 1128
    },
    {
      "epoch": 0.7283870967741936,
      "grad_norm": 0.60546875,
      "learning_rate": 8.31489122643912e-06,
      "loss": 0.5744,
      "step": 1129
    },
    {
      "epoch": 0.7290322580645161,
      "grad_norm": 0.7109375,
      "learning_rate": 8.31176085385535e-06,
      "loss": 0.7123,
      "step": 1130
    },
    {
      "epoch": 0.7296774193548388,
      "grad_norm": 0.99609375,
      "learning_rate": 8.30862816683579e-06,
      "loss": 0.9148,
      "step": 1131
    },
    {
      "epoch": 0.7303225806451613,
      "grad_norm": 0.8828125,
      "learning_rate": 8.305493167569729e-06,
      "loss": 0.9664,
      "step": 1132
    },
    {
      "epoch": 0.7309677419354839,
      "grad_norm": 0.9375,
      "learning_rate": 8.302355858248067e-06,
      "loss": 0.948,
      "step": 1133
    },
    {
      "epoch": 0.7316129032258064,
      "grad_norm": 0.96875,
      "learning_rate": 8.299216241063331e-06,
      "loss": 0.8187,
      "step": 1134
    },
    {
      "epoch": 0.7322580645161291,
      "grad_norm": 1.2265625,
      "learning_rate": 8.296074318209649e-06,
      "loss": 1.1467,
      "step": 1135
    },
    {
      "epoch": 0.7329032258064516,
      "grad_norm": 1.0859375,
      "learning_rate": 8.292930091882765e-06,
      "loss": 0.8955,
      "step": 1136
    },
    {
      "epoch": 0.7335483870967742,
      "grad_norm": 0.59765625,
      "learning_rate": 8.289783564280034e-06,
      "loss": 0.5555,
      "step": 1137
    },
    {
      "epoch": 0.7341935483870968,
      "grad_norm": 0.7109375,
      "learning_rate": 8.286634737600418e-06,
      "loss": 0.6174,
      "step": 1138
    },
    {
      "epoch": 0.7348387096774194,
      "grad_norm": 1.015625,
      "learning_rate": 8.283483614044487e-06,
      "loss": 0.8065,
      "step": 1139
    },
    {
      "epoch": 0.7354838709677419,
      "grad_norm": 0.83203125,
      "learning_rate": 8.28033019581441e-06,
      "loss": 0.7716,
      "step": 1140
    },
    {
      "epoch": 0.7361290322580645,
      "grad_norm": 0.99609375,
      "learning_rate": 8.27717448511397e-06,
      "loss": 1.0441,
      "step": 1141
    },
    {
      "epoch": 0.7367741935483871,
      "grad_norm": 0.6796875,
      "learning_rate": 8.274016484148543e-06,
      "loss": 0.7146,
      "step": 1142
    },
    {
      "epoch": 0.7374193548387097,
      "grad_norm": 0.94921875,
      "learning_rate": 8.270856195125112e-06,
      "loss": 0.8203,
      "step": 1143
    },
    {
      "epoch": 0.7380645161290322,
      "grad_norm": 0.98046875,
      "learning_rate": 8.267693620252252e-06,
      "loss": 0.8165,
      "step": 1144
    },
    {
      "epoch": 0.7387096774193549,
      "grad_norm": 0.85546875,
      "learning_rate": 8.264528761740145e-06,
      "loss": 0.7415,
      "step": 1145
    },
    {
      "epoch": 0.7393548387096774,
      "grad_norm": 1.1015625,
      "learning_rate": 8.26136162180056e-06,
      "loss": 1.0955,
      "step": 1146
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3984375,
      "learning_rate": 8.258192202646864e-06,
      "loss": 0.9897,
      "step": 1147
    },
    {
      "epoch": 0.7406451612903225,
      "grad_norm": 0.94140625,
      "learning_rate": 8.255020506494018e-06,
      "loss": 0.7306,
      "step": 1148
    },
    {
      "epoch": 0.7412903225806452,
      "grad_norm": 0.6171875,
      "learning_rate": 8.251846535558574e-06,
      "loss": 0.6647,
      "step": 1149
    },
    {
      "epoch": 0.7419354838709677,
      "grad_norm": 0.91015625,
      "learning_rate": 8.248670292058669e-06,
      "loss": 0.6649,
      "step": 1150
    },
    {
      "epoch": 0.7425806451612903,
      "grad_norm": 0.78125,
      "learning_rate": 8.245491778214037e-06,
      "loss": 0.7342,
      "step": 1151
    },
    {
      "epoch": 0.743225806451613,
      "grad_norm": 0.8515625,
      "learning_rate": 8.24231099624599e-06,
      "loss": 0.7653,
      "step": 1152
    },
    {
      "epoch": 0.7438709677419355,
      "grad_norm": 1.2734375,
      "learning_rate": 8.239127948377428e-06,
      "loss": 0.8639,
      "step": 1153
    },
    {
      "epoch": 0.7445161290322581,
      "grad_norm": 1.125,
      "learning_rate": 8.23594263683284e-06,
      "loss": 0.9518,
      "step": 1154
    },
    {
      "epoch": 0.7451612903225806,
      "grad_norm": 0.80078125,
      "learning_rate": 8.232755063838287e-06,
      "loss": 0.5537,
      "step": 1155
    },
    {
      "epoch": 0.7458064516129033,
      "grad_norm": 0.7421875,
      "learning_rate": 8.22956523162142e-06,
      "loss": 0.6985,
      "step": 1156
    },
    {
      "epoch": 0.7464516129032258,
      "grad_norm": 0.79296875,
      "learning_rate": 8.22637314241146e-06,
      "loss": 0.6075,
      "step": 1157
    },
    {
      "epoch": 0.7470967741935484,
      "grad_norm": 0.8984375,
      "learning_rate": 8.223178798439215e-06,
      "loss": 0.6982,
      "step": 1158
    },
    {
      "epoch": 0.747741935483871,
      "grad_norm": 0.76171875,
      "learning_rate": 8.219982201937061e-06,
      "loss": 0.6848,
      "step": 1159
    },
    {
      "epoch": 0.7483870967741936,
      "grad_norm": 0.765625,
      "learning_rate": 8.216783355138949e-06,
      "loss": 0.7217,
      "step": 1160
    },
    {
      "epoch": 0.7490322580645161,
      "grad_norm": 0.90234375,
      "learning_rate": 8.213582260280407e-06,
      "loss": 0.8598,
      "step": 1161
    },
    {
      "epoch": 0.7496774193548387,
      "grad_norm": 0.8359375,
      "learning_rate": 8.210378919598531e-06,
      "loss": 0.6895,
      "step": 1162
    },
    {
      "epoch": 0.7503225806451613,
      "grad_norm": 0.9140625,
      "learning_rate": 8.207173335331986e-06,
      "loss": 0.8455,
      "step": 1163
    },
    {
      "epoch": 0.7509677419354839,
      "grad_norm": 1.1796875,
      "learning_rate": 8.203965509721006e-06,
      "loss": 1.1619,
      "step": 1164
    },
    {
      "epoch": 0.7516129032258064,
      "grad_norm": 0.7734375,
      "learning_rate": 8.200755445007393e-06,
      "loss": 0.7517,
      "step": 1165
    },
    {
      "epoch": 0.752258064516129,
      "grad_norm": 0.796875,
      "learning_rate": 8.19754314343451e-06,
      "loss": 0.7802,
      "step": 1166
    },
    {
      "epoch": 0.7529032258064516,
      "grad_norm": 0.90234375,
      "learning_rate": 8.194328607247287e-06,
      "loss": 0.8473,
      "step": 1167
    },
    {
      "epoch": 0.7535483870967742,
      "grad_norm": 0.86328125,
      "learning_rate": 8.191111838692212e-06,
      "loss": 0.6559,
      "step": 1168
    },
    {
      "epoch": 0.7541935483870967,
      "grad_norm": 0.8828125,
      "learning_rate": 8.187892840017337e-06,
      "loss": 0.7547,
      "step": 1169
    },
    {
      "epoch": 0.7548387096774194,
      "grad_norm": 0.7734375,
      "learning_rate": 8.184671613472271e-06,
      "loss": 0.7176,
      "step": 1170
    },
    {
      "epoch": 0.7554838709677419,
      "grad_norm": 0.90625,
      "learning_rate": 8.181448161308178e-06,
      "loss": 0.7288,
      "step": 1171
    },
    {
      "epoch": 0.7561290322580645,
      "grad_norm": 1.125,
      "learning_rate": 8.178222485777779e-06,
      "loss": 1.3519,
      "step": 1172
    },
    {
      "epoch": 0.756774193548387,
      "grad_norm": 0.98828125,
      "learning_rate": 8.174994589135348e-06,
      "loss": 0.7816,
      "step": 1173
    },
    {
      "epoch": 0.7574193548387097,
      "grad_norm": 0.6875,
      "learning_rate": 8.171764473636714e-06,
      "loss": 0.5047,
      "step": 1174
    },
    {
      "epoch": 0.7580645161290323,
      "grad_norm": 0.75,
      "learning_rate": 8.168532141539254e-06,
      "loss": 0.7663,
      "step": 1175
    },
    {
      "epoch": 0.7587096774193548,
      "grad_norm": 1.109375,
      "learning_rate": 8.165297595101896e-06,
      "loss": 0.8209,
      "step": 1176
    },
    {
      "epoch": 0.7593548387096775,
      "grad_norm": 1.171875,
      "learning_rate": 8.162060836585111e-06,
      "loss": 1.0365,
      "step": 1177
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9921875,
      "learning_rate": 8.15882186825092e-06,
      "loss": 0.8385,
      "step": 1178
    },
    {
      "epoch": 0.7606451612903226,
      "grad_norm": 0.859375,
      "learning_rate": 8.15558069236289e-06,
      "loss": 0.7567,
      "step": 1179
    },
    {
      "epoch": 0.7612903225806451,
      "grad_norm": 0.65234375,
      "learning_rate": 8.152337311186126e-06,
      "loss": 0.6133,
      "step": 1180
    },
    {
      "epoch": 0.7619354838709678,
      "grad_norm": 0.91796875,
      "learning_rate": 8.149091726987277e-06,
      "loss": 0.9114,
      "step": 1181
    },
    {
      "epoch": 0.7625806451612903,
      "grad_norm": 0.84375,
      "learning_rate": 8.145843942034533e-06,
      "loss": 0.955,
      "step": 1182
    },
    {
      "epoch": 0.7632258064516129,
      "grad_norm": 0.828125,
      "learning_rate": 8.142593958597615e-06,
      "loss": 0.8106,
      "step": 1183
    },
    {
      "epoch": 0.7638709677419355,
      "grad_norm": 0.8359375,
      "learning_rate": 8.13934177894779e-06,
      "loss": 0.6443,
      "step": 1184
    },
    {
      "epoch": 0.7645161290322581,
      "grad_norm": 0.89453125,
      "learning_rate": 8.136087405357855e-06,
      "loss": 0.8003,
      "step": 1185
    },
    {
      "epoch": 0.7651612903225806,
      "grad_norm": 1.2109375,
      "learning_rate": 8.132830840102137e-06,
      "loss": 0.97,
      "step": 1186
    },
    {
      "epoch": 0.7658064516129032,
      "grad_norm": 1.0234375,
      "learning_rate": 8.129572085456501e-06,
      "loss": 0.8262,
      "step": 1187
    },
    {
      "epoch": 0.7664516129032258,
      "grad_norm": 1.046875,
      "learning_rate": 8.126311143698339e-06,
      "loss": 1.0617,
      "step": 1188
    },
    {
      "epoch": 0.7670967741935484,
      "grad_norm": 1.0546875,
      "learning_rate": 8.123048017106572e-06,
      "loss": 1.0031,
      "step": 1189
    },
    {
      "epoch": 0.7677419354838709,
      "grad_norm": 0.5703125,
      "learning_rate": 8.119782707961646e-06,
      "loss": 0.5114,
      "step": 1190
    },
    {
      "epoch": 0.7683870967741936,
      "grad_norm": 0.921875,
      "learning_rate": 8.116515218545532e-06,
      "loss": 0.836,
      "step": 1191
    },
    {
      "epoch": 0.7690322580645161,
      "grad_norm": 0.99609375,
      "learning_rate": 8.113245551141733e-06,
      "loss": 0.7886,
      "step": 1192
    },
    {
      "epoch": 0.7696774193548387,
      "grad_norm": 0.8046875,
      "learning_rate": 8.109973708035264e-06,
      "loss": 0.6537,
      "step": 1193
    },
    {
      "epoch": 0.7703225806451612,
      "grad_norm": 0.8203125,
      "learning_rate": 8.106699691512662e-06,
      "loss": 0.6543,
      "step": 1194
    },
    {
      "epoch": 0.7709677419354839,
      "grad_norm": 1.0078125,
      "learning_rate": 8.103423503861988e-06,
      "loss": 0.9168,
      "step": 1195
    },
    {
      "epoch": 0.7716129032258064,
      "grad_norm": 1.0859375,
      "learning_rate": 8.100145147372815e-06,
      "loss": 0.8585,
      "step": 1196
    },
    {
      "epoch": 0.772258064516129,
      "grad_norm": 0.73828125,
      "learning_rate": 8.096864624336237e-06,
      "loss": 0.7533,
      "step": 1197
    },
    {
      "epoch": 0.7729032258064517,
      "grad_norm": 1.0546875,
      "learning_rate": 8.093581937044858e-06,
      "loss": 1.0012,
      "step": 1198
    },
    {
      "epoch": 0.7735483870967742,
      "grad_norm": 0.69921875,
      "learning_rate": 8.090297087792793e-06,
      "loss": 0.6386,
      "step": 1199
    },
    {
      "epoch": 0.7741935483870968,
      "grad_norm": 0.7421875,
      "learning_rate": 8.087010078875674e-06,
      "loss": 0.6272,
      "step": 1200
    },
    {
      "epoch": 0.7748387096774193,
      "grad_norm": 1.6484375,
      "learning_rate": 8.083720912590638e-06,
      "loss": 1.1106,
      "step": 1201
    },
    {
      "epoch": 0.775483870967742,
      "grad_norm": 0.59375,
      "learning_rate": 8.08042959123633e-06,
      "loss": 0.4928,
      "step": 1202
    },
    {
      "epoch": 0.7761290322580645,
      "grad_norm": 0.8359375,
      "learning_rate": 8.0771361171129e-06,
      "loss": 0.8052,
      "step": 1203
    },
    {
      "epoch": 0.7767741935483871,
      "grad_norm": 0.75,
      "learning_rate": 8.073840492522008e-06,
      "loss": 0.6223,
      "step": 1204
    },
    {
      "epoch": 0.7774193548387097,
      "grad_norm": 1.1640625,
      "learning_rate": 8.07054271976681e-06,
      "loss": 0.952,
      "step": 1205
    },
    {
      "epoch": 0.7780645161290323,
      "grad_norm": 0.80078125,
      "learning_rate": 8.067242801151966e-06,
      "loss": 0.7289,
      "step": 1206
    },
    {
      "epoch": 0.7787096774193548,
      "grad_norm": 0.65234375,
      "learning_rate": 8.063940738983636e-06,
      "loss": 0.5697,
      "step": 1207
    },
    {
      "epoch": 0.7793548387096774,
      "grad_norm": 0.9921875,
      "learning_rate": 8.060636535569478e-06,
      "loss": 0.8849,
      "step": 1208
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.09375,
      "learning_rate": 8.057330193218646e-06,
      "loss": 0.8629,
      "step": 1209
    },
    {
      "epoch": 0.7806451612903226,
      "grad_norm": 0.86328125,
      "learning_rate": 8.054021714241792e-06,
      "loss": 0.6289,
      "step": 1210
    },
    {
      "epoch": 0.7812903225806451,
      "grad_norm": 1.140625,
      "learning_rate": 8.050711100951054e-06,
      "loss": 1.0229,
      "step": 1211
    },
    {
      "epoch": 0.7819354838709678,
      "grad_norm": 0.71875,
      "learning_rate": 8.04739835566007e-06,
      "loss": 0.6565,
      "step": 1212
    },
    {
      "epoch": 0.7825806451612903,
      "grad_norm": 1.1171875,
      "learning_rate": 8.04408348068396e-06,
      "loss": 0.785,
      "step": 1213
    },
    {
      "epoch": 0.7832258064516129,
      "grad_norm": 0.55078125,
      "learning_rate": 8.040766478339337e-06,
      "loss": 0.4099,
      "step": 1214
    },
    {
      "epoch": 0.7838709677419354,
      "grad_norm": 1.1875,
      "learning_rate": 8.037447350944302e-06,
      "loss": 1.1377,
      "step": 1215
    },
    {
      "epoch": 0.7845161290322581,
      "grad_norm": 1.1328125,
      "learning_rate": 8.034126100818436e-06,
      "loss": 1.0408,
      "step": 1216
    },
    {
      "epoch": 0.7851612903225806,
      "grad_norm": 0.96484375,
      "learning_rate": 8.030802730282808e-06,
      "loss": 0.8525,
      "step": 1217
    },
    {
      "epoch": 0.7858064516129032,
      "grad_norm": 0.9296875,
      "learning_rate": 8.027477241659966e-06,
      "loss": 0.6342,
      "step": 1218
    },
    {
      "epoch": 0.7864516129032258,
      "grad_norm": 0.828125,
      "learning_rate": 8.024149637273941e-06,
      "loss": 0.7811,
      "step": 1219
    },
    {
      "epoch": 0.7870967741935484,
      "grad_norm": 0.7734375,
      "learning_rate": 8.02081991945024e-06,
      "loss": 0.6607,
      "step": 1220
    },
    {
      "epoch": 0.787741935483871,
      "grad_norm": 0.86328125,
      "learning_rate": 8.01748809051585e-06,
      "loss": 0.8867,
      "step": 1221
    },
    {
      "epoch": 0.7883870967741935,
      "grad_norm": 0.84765625,
      "learning_rate": 8.014154152799232e-06,
      "loss": 0.8379,
      "step": 1222
    },
    {
      "epoch": 0.7890322580645162,
      "grad_norm": 1.21875,
      "learning_rate": 8.010818108630314e-06,
      "loss": 1.2554,
      "step": 1223
    },
    {
      "epoch": 0.7896774193548387,
      "grad_norm": 0.9375,
      "learning_rate": 8.007479960340509e-06,
      "loss": 0.7915,
      "step": 1224
    },
    {
      "epoch": 0.7903225806451613,
      "grad_norm": 1.5,
      "learning_rate": 8.00413971026269e-06,
      "loss": 1.309,
      "step": 1225
    },
    {
      "epoch": 0.7909677419354839,
      "grad_norm": 1.2578125,
      "learning_rate": 8.000797360731205e-06,
      "loss": 0.8736,
      "step": 1226
    },
    {
      "epoch": 0.7916129032258065,
      "grad_norm": 0.68359375,
      "learning_rate": 7.997452914081866e-06,
      "loss": 0.6025,
      "step": 1227
    },
    {
      "epoch": 0.792258064516129,
      "grad_norm": 1.0625,
      "learning_rate": 7.99410637265195e-06,
      "loss": 1.0722,
      "step": 1228
    },
    {
      "epoch": 0.7929032258064516,
      "grad_norm": 1.0703125,
      "learning_rate": 7.9907577387802e-06,
      "loss": 1.0594,
      "step": 1229
    },
    {
      "epoch": 0.7935483870967742,
      "grad_norm": 0.62890625,
      "learning_rate": 7.987407014806822e-06,
      "loss": 0.6677,
      "step": 1230
    },
    {
      "epoch": 0.7941935483870968,
      "grad_norm": 1.1484375,
      "learning_rate": 7.984054203073482e-06,
      "loss": 1.1549,
      "step": 1231
    },
    {
      "epoch": 0.7948387096774193,
      "grad_norm": 0.6796875,
      "learning_rate": 7.9806993059233e-06,
      "loss": 0.5605,
      "step": 1232
    },
    {
      "epoch": 0.795483870967742,
      "grad_norm": 0.8828125,
      "learning_rate": 7.97734232570086e-06,
      "loss": 0.9058,
      "step": 1233
    },
    {
      "epoch": 0.7961290322580645,
      "grad_norm": 1.0078125,
      "learning_rate": 7.973983264752201e-06,
      "loss": 0.771,
      "step": 1234
    },
    {
      "epoch": 0.7967741935483871,
      "grad_norm": 0.87890625,
      "learning_rate": 7.970622125424813e-06,
      "loss": 0.7938,
      "step": 1235
    },
    {
      "epoch": 0.7974193548387096,
      "grad_norm": 1.2734375,
      "learning_rate": 7.96725891006764e-06,
      "loss": 1.177,
      "step": 1236
    },
    {
      "epoch": 0.7980645161290323,
      "grad_norm": 0.98046875,
      "learning_rate": 7.96389362103108e-06,
      "loss": 0.6604,
      "step": 1237
    },
    {
      "epoch": 0.7987096774193548,
      "grad_norm": 1.1875,
      "learning_rate": 7.960526260666973e-06,
      "loss": 1.0715,
      "step": 1238
    },
    {
      "epoch": 0.7993548387096774,
      "grad_norm": 0.6328125,
      "learning_rate": 7.957156831328613e-06,
      "loss": 0.5959,
      "step": 1239
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1796875,
      "learning_rate": 7.953785335370738e-06,
      "loss": 1.0465,
      "step": 1240
    },
    {
      "epoch": 0.8006451612903226,
      "grad_norm": 1.0546875,
      "learning_rate": 7.950411775149526e-06,
      "loss": 0.9773,
      "step": 1241
    },
    {
      "epoch": 0.8012903225806451,
      "grad_norm": 0.95703125,
      "learning_rate": 7.947036153022608e-06,
      "loss": 0.6017,
      "step": 1242
    },
    {
      "epoch": 0.8019354838709677,
      "grad_norm": 0.9609375,
      "learning_rate": 7.943658471349042e-06,
      "loss": 0.814,
      "step": 1243
    },
    {
      "epoch": 0.8025806451612904,
      "grad_norm": 1.21875,
      "learning_rate": 7.94027873248934e-06,
      "loss": 1.0717,
      "step": 1244
    },
    {
      "epoch": 0.8032258064516129,
      "grad_norm": 1.109375,
      "learning_rate": 7.936896938805441e-06,
      "loss": 0.9019,
      "step": 1245
    },
    {
      "epoch": 0.8038709677419354,
      "grad_norm": 0.75,
      "learning_rate": 7.933513092660722e-06,
      "loss": 0.6829,
      "step": 1246
    },
    {
      "epoch": 0.8045161290322581,
      "grad_norm": 1.3359375,
      "learning_rate": 7.930127196419998e-06,
      "loss": 0.9223,
      "step": 1247
    },
    {
      "epoch": 0.8051612903225807,
      "grad_norm": 1.0390625,
      "learning_rate": 7.926739252449513e-06,
      "loss": 0.8711,
      "step": 1248
    },
    {
      "epoch": 0.8058064516129032,
      "grad_norm": 0.7265625,
      "learning_rate": 7.923349263116948e-06,
      "loss": 0.7375,
      "step": 1249
    },
    {
      "epoch": 0.8064516129032258,
      "grad_norm": 0.84375,
      "learning_rate": 7.919957230791404e-06,
      "loss": 0.7892,
      "step": 1250
    },
    {
      "epoch": 0.8070967741935484,
      "grad_norm": 1.2265625,
      "learning_rate": 7.916563157843416e-06,
      "loss": 0.9375,
      "step": 1251
    },
    {
      "epoch": 0.807741935483871,
      "grad_norm": 0.9921875,
      "learning_rate": 7.913167046644943e-06,
      "loss": 0.8971,
      "step": 1252
    },
    {
      "epoch": 0.8083870967741935,
      "grad_norm": 0.87890625,
      "learning_rate": 7.909768899569372e-06,
      "loss": 0.696,
      "step": 1253
    },
    {
      "epoch": 0.8090322580645162,
      "grad_norm": 0.7578125,
      "learning_rate": 7.90636871899151e-06,
      "loss": 0.8328,
      "step": 1254
    },
    {
      "epoch": 0.8096774193548387,
      "grad_norm": 1.40625,
      "learning_rate": 7.902966507287584e-06,
      "loss": 0.8586,
      "step": 1255
    },
    {
      "epoch": 0.8103225806451613,
      "grad_norm": 1.328125,
      "learning_rate": 7.89956226683524e-06,
      "loss": 1.1218,
      "step": 1256
    },
    {
      "epoch": 0.8109677419354838,
      "grad_norm": 1.109375,
      "learning_rate": 7.896156000013546e-06,
      "loss": 0.9598,
      "step": 1257
    },
    {
      "epoch": 0.8116129032258065,
      "grad_norm": 1.0859375,
      "learning_rate": 7.892747709202984e-06,
      "loss": 1.0195,
      "step": 1258
    },
    {
      "epoch": 0.812258064516129,
      "grad_norm": 1.1171875,
      "learning_rate": 7.889337396785447e-06,
      "loss": 1.0038,
      "step": 1259
    },
    {
      "epoch": 0.8129032258064516,
      "grad_norm": 0.859375,
      "learning_rate": 7.885925065144249e-06,
      "loss": 0.9596,
      "step": 1260
    },
    {
      "epoch": 0.8135483870967742,
      "grad_norm": 0.8515625,
      "learning_rate": 7.882510716664104e-06,
      "loss": 0.6352,
      "step": 1261
    },
    {
      "epoch": 0.8141935483870968,
      "grad_norm": 0.6640625,
      "learning_rate": 7.879094353731146e-06,
      "loss": 0.4975,
      "step": 1262
    },
    {
      "epoch": 0.8148387096774193,
      "grad_norm": 0.58203125,
      "learning_rate": 7.875675978732911e-06,
      "loss": 0.5712,
      "step": 1263
    },
    {
      "epoch": 0.8154838709677419,
      "grad_norm": 0.98046875,
      "learning_rate": 7.87225559405834e-06,
      "loss": 0.8827,
      "step": 1264
    },
    {
      "epoch": 0.8161290322580645,
      "grad_norm": 0.984375,
      "learning_rate": 7.868833202097786e-06,
      "loss": 0.9438,
      "step": 1265
    },
    {
      "epoch": 0.8167741935483871,
      "grad_norm": 1.1484375,
      "learning_rate": 7.865408805242996e-06,
      "loss": 1.0337,
      "step": 1266
    },
    {
      "epoch": 0.8174193548387096,
      "grad_norm": 1.1640625,
      "learning_rate": 7.86198240588712e-06,
      "loss": 0.9214,
      "step": 1267
    },
    {
      "epoch": 0.8180645161290323,
      "grad_norm": 1.0703125,
      "learning_rate": 7.858554006424717e-06,
      "loss": 0.9749,
      "step": 1268
    },
    {
      "epoch": 0.8187096774193549,
      "grad_norm": 0.81640625,
      "learning_rate": 7.855123609251727e-06,
      "loss": 0.7219,
      "step": 1269
    },
    {
      "epoch": 0.8193548387096774,
      "grad_norm": 0.9765625,
      "learning_rate": 7.851691216765501e-06,
      "loss": 0.7779,
      "step": 1270
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.84765625,
      "learning_rate": 7.848256831364778e-06,
      "loss": 0.735,
      "step": 1271
    },
    {
      "epoch": 0.8206451612903226,
      "grad_norm": 0.97265625,
      "learning_rate": 7.844820455449688e-06,
      "loss": 0.782,
      "step": 1272
    },
    {
      "epoch": 0.8212903225806452,
      "grad_norm": 1.4453125,
      "learning_rate": 7.841382091421757e-06,
      "loss": 1.1494,
      "step": 1273
    },
    {
      "epoch": 0.8219354838709677,
      "grad_norm": 1.109375,
      "learning_rate": 7.837941741683894e-06,
      "loss": 1.2546,
      "step": 1274
    },
    {
      "epoch": 0.8225806451612904,
      "grad_norm": 0.56640625,
      "learning_rate": 7.834499408640405e-06,
      "loss": 0.4687,
      "step": 1275
    },
    {
      "epoch": 0.8232258064516129,
      "grad_norm": 0.62109375,
      "learning_rate": 7.831055094696973e-06,
      "loss": 0.5984,
      "step": 1276
    },
    {
      "epoch": 0.8238709677419355,
      "grad_norm": 0.8671875,
      "learning_rate": 7.82760880226067e-06,
      "loss": 0.727,
      "step": 1277
    },
    {
      "epoch": 0.824516129032258,
      "grad_norm": 0.87109375,
      "learning_rate": 7.82416053373995e-06,
      "loss": 0.7823,
      "step": 1278
    },
    {
      "epoch": 0.8251612903225807,
      "grad_norm": 0.76171875,
      "learning_rate": 7.820710291544648e-06,
      "loss": 0.6392,
      "step": 1279
    },
    {
      "epoch": 0.8258064516129032,
      "grad_norm": 0.84765625,
      "learning_rate": 7.817258078085976e-06,
      "loss": 0.9261,
      "step": 1280
    },
    {
      "epoch": 0.8264516129032258,
      "grad_norm": 0.62109375,
      "learning_rate": 7.813803895776527e-06,
      "loss": 0.545,
      "step": 1281
    },
    {
      "epoch": 0.8270967741935484,
      "grad_norm": 1.2109375,
      "learning_rate": 7.810347747030268e-06,
      "loss": 1.0684,
      "step": 1282
    },
    {
      "epoch": 0.827741935483871,
      "grad_norm": 1.1171875,
      "learning_rate": 7.806889634262543e-06,
      "loss": 1.0532,
      "step": 1283
    },
    {
      "epoch": 0.8283870967741935,
      "grad_norm": 0.5546875,
      "learning_rate": 7.80342955989006e-06,
      "loss": 0.5308,
      "step": 1284
    },
    {
      "epoch": 0.8290322580645161,
      "grad_norm": 0.828125,
      "learning_rate": 7.799967526330915e-06,
      "loss": 0.7783,
      "step": 1285
    },
    {
      "epoch": 0.8296774193548387,
      "grad_norm": 1.421875,
      "learning_rate": 7.796503536004552e-06,
      "loss": 1.1903,
      "step": 1286
    },
    {
      "epoch": 0.8303225806451613,
      "grad_norm": 0.73046875,
      "learning_rate": 7.793037591331797e-06,
      "loss": 0.7944,
      "step": 1287
    },
    {
      "epoch": 0.8309677419354838,
      "grad_norm": 0.859375,
      "learning_rate": 7.789569694734839e-06,
      "loss": 0.6286,
      "step": 1288
    },
    {
      "epoch": 0.8316129032258065,
      "grad_norm": 1.140625,
      "learning_rate": 7.786099848637228e-06,
      "loss": 0.8629,
      "step": 1289
    },
    {
      "epoch": 0.832258064516129,
      "grad_norm": 0.953125,
      "learning_rate": 7.782628055463879e-06,
      "loss": 0.8347,
      "step": 1290
    },
    {
      "epoch": 0.8329032258064516,
      "grad_norm": 0.69921875,
      "learning_rate": 7.779154317641066e-06,
      "loss": 0.6728,
      "step": 1291
    },
    {
      "epoch": 0.8335483870967741,
      "grad_norm": 1.2265625,
      "learning_rate": 7.775678637596425e-06,
      "loss": 0.7919,
      "step": 1292
    },
    {
      "epoch": 0.8341935483870968,
      "grad_norm": 1.0,
      "learning_rate": 7.772201017758943e-06,
      "loss": 0.788,
      "step": 1293
    },
    {
      "epoch": 0.8348387096774194,
      "grad_norm": 0.88671875,
      "learning_rate": 7.768721460558972e-06,
      "loss": 0.9292,
      "step": 1294
    },
    {
      "epoch": 0.8354838709677419,
      "grad_norm": 0.83203125,
      "learning_rate": 7.765239968428212e-06,
      "loss": 0.7195,
      "step": 1295
    },
    {
      "epoch": 0.8361290322580646,
      "grad_norm": 0.99609375,
      "learning_rate": 7.761756543799713e-06,
      "loss": 1.0239,
      "step": 1296
    },
    {
      "epoch": 0.8367741935483871,
      "grad_norm": 0.90234375,
      "learning_rate": 7.758271189107882e-06,
      "loss": 0.8146,
      "step": 1297
    },
    {
      "epoch": 0.8374193548387097,
      "grad_norm": 1.0703125,
      "learning_rate": 7.754783906788468e-06,
      "loss": 1.172,
      "step": 1298
    },
    {
      "epoch": 0.8380645161290322,
      "grad_norm": 0.92578125,
      "learning_rate": 7.751294699278573e-06,
      "loss": 0.8034,
      "step": 1299
    },
    {
      "epoch": 0.8387096774193549,
      "grad_norm": 0.96484375,
      "learning_rate": 7.747803569016641e-06,
      "loss": 0.8407,
      "step": 1300
    },
    {
      "epoch": 0.8393548387096774,
      "grad_norm": 0.90625,
      "learning_rate": 7.744310518442462e-06,
      "loss": 0.8912,
      "step": 1301
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.75,
      "learning_rate": 7.740815549997168e-06,
      "loss": 0.6109,
      "step": 1302
    },
    {
      "epoch": 0.8406451612903226,
      "grad_norm": 0.640625,
      "learning_rate": 7.737318666123224e-06,
      "loss": 0.7428,
      "step": 1303
    },
    {
      "epoch": 0.8412903225806452,
      "grad_norm": 1.109375,
      "learning_rate": 7.733819869264447e-06,
      "loss": 1.3333,
      "step": 1304
    },
    {
      "epoch": 0.8419354838709677,
      "grad_norm": 0.50390625,
      "learning_rate": 7.73031916186598e-06,
      "loss": 0.4535,
      "step": 1305
    },
    {
      "epoch": 0.8425806451612903,
      "grad_norm": 0.73828125,
      "learning_rate": 7.726816546374306e-06,
      "loss": 0.8284,
      "step": 1306
    },
    {
      "epoch": 0.8432258064516129,
      "grad_norm": 1.046875,
      "learning_rate": 7.723312025237235e-06,
      "loss": 0.7,
      "step": 1307
    },
    {
      "epoch": 0.8438709677419355,
      "grad_norm": 0.8515625,
      "learning_rate": 7.719805600903922e-06,
      "loss": 0.7622,
      "step": 1308
    },
    {
      "epoch": 0.844516129032258,
      "grad_norm": 1.40625,
      "learning_rate": 7.716297275824837e-06,
      "loss": 1.2543,
      "step": 1309
    },
    {
      "epoch": 0.8451612903225807,
      "grad_norm": 0.7265625,
      "learning_rate": 7.712787052451793e-06,
      "loss": 0.6503,
      "step": 1310
    },
    {
      "epoch": 0.8458064516129032,
      "grad_norm": 1.25,
      "learning_rate": 7.709274933237916e-06,
      "loss": 1.0442,
      "step": 1311
    },
    {
      "epoch": 0.8464516129032258,
      "grad_norm": 0.7109375,
      "learning_rate": 7.705760920637663e-06,
      "loss": 0.7648,
      "step": 1312
    },
    {
      "epoch": 0.8470967741935483,
      "grad_norm": 0.76171875,
      "learning_rate": 7.702245017106818e-06,
      "loss": 0.6781,
      "step": 1313
    },
    {
      "epoch": 0.847741935483871,
      "grad_norm": 1.1875,
      "learning_rate": 7.69872722510248e-06,
      "loss": 0.968,
      "step": 1314
    },
    {
      "epoch": 0.8483870967741935,
      "grad_norm": 1.234375,
      "learning_rate": 7.695207547083071e-06,
      "loss": 1.0333,
      "step": 1315
    },
    {
      "epoch": 0.8490322580645161,
      "grad_norm": 1.1171875,
      "learning_rate": 7.69168598550833e-06,
      "loss": 0.9421,
      "step": 1316
    },
    {
      "epoch": 0.8496774193548388,
      "grad_norm": 1.1796875,
      "learning_rate": 7.688162542839312e-06,
      "loss": 0.9483,
      "step": 1317
    },
    {
      "epoch": 0.8503225806451613,
      "grad_norm": 0.97265625,
      "learning_rate": 7.684637221538391e-06,
      "loss": 1.1014,
      "step": 1318
    },
    {
      "epoch": 0.8509677419354839,
      "grad_norm": 1.234375,
      "learning_rate": 7.681110024069245e-06,
      "loss": 1.0173,
      "step": 1319
    },
    {
      "epoch": 0.8516129032258064,
      "grad_norm": 0.921875,
      "learning_rate": 7.677580952896874e-06,
      "loss": 0.7196,
      "step": 1320
    },
    {
      "epoch": 0.8522580645161291,
      "grad_norm": 1.109375,
      "learning_rate": 7.674050010487577e-06,
      "loss": 1.0276,
      "step": 1321
    },
    {
      "epoch": 0.8529032258064516,
      "grad_norm": 0.486328125,
      "learning_rate": 7.670517199308967e-06,
      "loss": 0.4121,
      "step": 1322
    },
    {
      "epoch": 0.8535483870967742,
      "grad_norm": 0.71484375,
      "learning_rate": 7.666982521829964e-06,
      "loss": 0.6791,
      "step": 1323
    },
    {
      "epoch": 0.8541935483870968,
      "grad_norm": 1.2890625,
      "learning_rate": 7.663445980520788e-06,
      "loss": 1.0207,
      "step": 1324
    },
    {
      "epoch": 0.8548387096774194,
      "grad_norm": 1.1015625,
      "learning_rate": 7.659907577852963e-06,
      "loss": 0.6842,
      "step": 1325
    },
    {
      "epoch": 0.8554838709677419,
      "grad_norm": 1.3359375,
      "learning_rate": 7.656367316299314e-06,
      "loss": 0.975,
      "step": 1326
    },
    {
      "epoch": 0.8561290322580645,
      "grad_norm": 1.3828125,
      "learning_rate": 7.652825198333966e-06,
      "loss": 0.9808,
      "step": 1327
    },
    {
      "epoch": 0.8567741935483871,
      "grad_norm": 1.1875,
      "learning_rate": 7.649281226432342e-06,
      "loss": 0.8895,
      "step": 1328
    },
    {
      "epoch": 0.8574193548387097,
      "grad_norm": 0.76953125,
      "learning_rate": 7.645735403071158e-06,
      "loss": 0.6565,
      "step": 1329
    },
    {
      "epoch": 0.8580645161290322,
      "grad_norm": 0.875,
      "learning_rate": 7.642187730728423e-06,
      "loss": 0.7514,
      "step": 1330
    },
    {
      "epoch": 0.8587096774193549,
      "grad_norm": 1.5078125,
      "learning_rate": 7.638638211883444e-06,
      "loss": 1.0283,
      "step": 1331
    },
    {
      "epoch": 0.8593548387096774,
      "grad_norm": 0.55859375,
      "learning_rate": 7.635086849016814e-06,
      "loss": 0.4649,
      "step": 1332
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.625,
      "learning_rate": 7.631533644610414e-06,
      "loss": 0.5876,
      "step": 1333
    },
    {
      "epoch": 0.8606451612903225,
      "grad_norm": 1.1171875,
      "learning_rate": 7.627978601147417e-06,
      "loss": 1.0034,
      "step": 1334
    },
    {
      "epoch": 0.8612903225806452,
      "grad_norm": 0.71875,
      "learning_rate": 7.624421721112271e-06,
      "loss": 0.6801,
      "step": 1335
    },
    {
      "epoch": 0.8619354838709677,
      "grad_norm": 0.56640625,
      "learning_rate": 7.620863006990721e-06,
      "loss": 0.4987,
      "step": 1336
    },
    {
      "epoch": 0.8625806451612903,
      "grad_norm": 0.55078125,
      "learning_rate": 7.617302461269782e-06,
      "loss": 0.5108,
      "step": 1337
    },
    {
      "epoch": 0.863225806451613,
      "grad_norm": 1.0546875,
      "learning_rate": 7.613740086437759e-06,
      "loss": 0.9763,
      "step": 1338
    },
    {
      "epoch": 0.8638709677419355,
      "grad_norm": 0.76171875,
      "learning_rate": 7.610175884984226e-06,
      "loss": 0.7581,
      "step": 1339
    },
    {
      "epoch": 0.864516129032258,
      "grad_norm": 0.921875,
      "learning_rate": 7.606609859400039e-06,
      "loss": 0.994,
      "step": 1340
    },
    {
      "epoch": 0.8651612903225806,
      "grad_norm": 1.046875,
      "learning_rate": 7.603042012177328e-06,
      "loss": 0.6744,
      "step": 1341
    },
    {
      "epoch": 0.8658064516129033,
      "grad_norm": 0.62890625,
      "learning_rate": 7.599472345809496e-06,
      "loss": 0.5216,
      "step": 1342
    },
    {
      "epoch": 0.8664516129032258,
      "grad_norm": 1.078125,
      "learning_rate": 7.595900862791216e-06,
      "loss": 1.0162,
      "step": 1343
    },
    {
      "epoch": 0.8670967741935484,
      "grad_norm": 0.984375,
      "learning_rate": 7.5923275656184315e-06,
      "loss": 0.9656,
      "step": 1344
    },
    {
      "epoch": 0.867741935483871,
      "grad_norm": 1.234375,
      "learning_rate": 7.588752456788355e-06,
      "loss": 0.8508,
      "step": 1345
    },
    {
      "epoch": 0.8683870967741936,
      "grad_norm": 1.0859375,
      "learning_rate": 7.585175538799464e-06,
      "loss": 1.0448,
      "step": 1346
    },
    {
      "epoch": 0.8690322580645161,
      "grad_norm": 0.71484375,
      "learning_rate": 7.5815968141514995e-06,
      "loss": 0.5708,
      "step": 1347
    },
    {
      "epoch": 0.8696774193548387,
      "grad_norm": 0.859375,
      "learning_rate": 7.5780162853454656e-06,
      "loss": 0.8408,
      "step": 1348
    },
    {
      "epoch": 0.8703225806451613,
      "grad_norm": 0.77734375,
      "learning_rate": 7.574433954883629e-06,
      "loss": 0.5687,
      "step": 1349
    },
    {
      "epoch": 0.8709677419354839,
      "grad_norm": 0.96875,
      "learning_rate": 7.570849825269512e-06,
      "loss": 0.9877,
      "step": 1350
    },
    {
      "epoch": 0.8716129032258064,
      "grad_norm": 0.94140625,
      "learning_rate": 7.567263899007897e-06,
      "loss": 0.5904,
      "step": 1351
    },
    {
      "epoch": 0.8722580645161291,
      "grad_norm": 1.21875,
      "learning_rate": 7.563676178604822e-06,
      "loss": 0.9965,
      "step": 1352
    },
    {
      "epoch": 0.8729032258064516,
      "grad_norm": 1.1796875,
      "learning_rate": 7.560086666567577e-06,
      "loss": 1.2068,
      "step": 1353
    },
    {
      "epoch": 0.8735483870967742,
      "grad_norm": 0.515625,
      "learning_rate": 7.556495365404706e-06,
      "loss": 0.4146,
      "step": 1354
    },
    {
      "epoch": 0.8741935483870967,
      "grad_norm": 0.87890625,
      "learning_rate": 7.552902277626002e-06,
      "loss": 0.7075,
      "step": 1355
    },
    {
      "epoch": 0.8748387096774194,
      "grad_norm": 0.98046875,
      "learning_rate": 7.549307405742509e-06,
      "loss": 0.8883,
      "step": 1356
    },
    {
      "epoch": 0.8754838709677419,
      "grad_norm": 1.0390625,
      "learning_rate": 7.545710752266513e-06,
      "loss": 0.8254,
      "step": 1357
    },
    {
      "epoch": 0.8761290322580645,
      "grad_norm": 0.734375,
      "learning_rate": 7.54211231971155e-06,
      "loss": 0.6915,
      "step": 1358
    },
    {
      "epoch": 0.8767741935483871,
      "grad_norm": 0.91015625,
      "learning_rate": 7.538512110592396e-06,
      "loss": 0.752,
      "step": 1359
    },
    {
      "epoch": 0.8774193548387097,
      "grad_norm": 0.5390625,
      "learning_rate": 7.534910127425072e-06,
      "loss": 0.5523,
      "step": 1360
    },
    {
      "epoch": 0.8780645161290322,
      "grad_norm": 0.75390625,
      "learning_rate": 7.531306372726834e-06,
      "loss": 0.6738,
      "step": 1361
    },
    {
      "epoch": 0.8787096774193548,
      "grad_norm": 0.765625,
      "learning_rate": 7.527700849016181e-06,
      "loss": 0.5936,
      "step": 1362
    },
    {
      "epoch": 0.8793548387096775,
      "grad_norm": 0.82421875,
      "learning_rate": 7.5240935588128435e-06,
      "loss": 0.7581,
      "step": 1363
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.234375,
      "learning_rate": 7.52048450463779e-06,
      "loss": 1.1994,
      "step": 1364
    },
    {
      "epoch": 0.8806451612903226,
      "grad_norm": 0.953125,
      "learning_rate": 7.5168736890132244e-06,
      "loss": 0.7987,
      "step": 1365
    },
    {
      "epoch": 0.8812903225806452,
      "grad_norm": 0.7421875,
      "learning_rate": 7.513261114462572e-06,
      "loss": 0.7691,
      "step": 1366
    },
    {
      "epoch": 0.8819354838709678,
      "grad_norm": 1.5234375,
      "learning_rate": 7.509646783510498e-06,
      "loss": 0.8938,
      "step": 1367
    },
    {
      "epoch": 0.8825806451612903,
      "grad_norm": 0.89453125,
      "learning_rate": 7.506030698682886e-06,
      "loss": 0.7618,
      "step": 1368
    },
    {
      "epoch": 0.8832258064516129,
      "grad_norm": 0.8359375,
      "learning_rate": 7.502412862506856e-06,
      "loss": 0.8505,
      "step": 1369
    },
    {
      "epoch": 0.8838709677419355,
      "grad_norm": 1.65625,
      "learning_rate": 7.49879327751074e-06,
      "loss": 1.135,
      "step": 1370
    },
    {
      "epoch": 0.8845161290322581,
      "grad_norm": 0.90625,
      "learning_rate": 7.4951719462241e-06,
      "loss": 0.9167,
      "step": 1371
    },
    {
      "epoch": 0.8851612903225806,
      "grad_norm": 1.328125,
      "learning_rate": 7.491548871177715e-06,
      "loss": 0.9713,
      "step": 1372
    },
    {
      "epoch": 0.8858064516129033,
      "grad_norm": 0.8515625,
      "learning_rate": 7.487924054903586e-06,
      "loss": 0.6889,
      "step": 1373
    },
    {
      "epoch": 0.8864516129032258,
      "grad_norm": 1.34375,
      "learning_rate": 7.484297499934925e-06,
      "loss": 1.041,
      "step": 1374
    },
    {
      "epoch": 0.8870967741935484,
      "grad_norm": 0.84375,
      "learning_rate": 7.4806692088061665e-06,
      "loss": 0.7997,
      "step": 1375
    },
    {
      "epoch": 0.8877419354838709,
      "grad_norm": 1.0625,
      "learning_rate": 7.4770391840529535e-06,
      "loss": 0.9399,
      "step": 1376
    },
    {
      "epoch": 0.8883870967741936,
      "grad_norm": 0.7890625,
      "learning_rate": 7.473407428212141e-06,
      "loss": 0.7314,
      "step": 1377
    },
    {
      "epoch": 0.8890322580645161,
      "grad_norm": 0.93359375,
      "learning_rate": 7.4697739438217965e-06,
      "loss": 0.89,
      "step": 1378
    },
    {
      "epoch": 0.8896774193548387,
      "grad_norm": 1.015625,
      "learning_rate": 7.466138733421189e-06,
      "loss": 0.9448,
      "step": 1379
    },
    {
      "epoch": 0.8903225806451613,
      "grad_norm": 0.96875,
      "learning_rate": 7.462501799550802e-06,
      "loss": 0.7991,
      "step": 1380
    },
    {
      "epoch": 0.8909677419354839,
      "grad_norm": 1.109375,
      "learning_rate": 7.458863144752318e-06,
      "loss": 0.9964,
      "step": 1381
    },
    {
      "epoch": 0.8916129032258064,
      "grad_norm": 1.109375,
      "learning_rate": 7.455222771568624e-06,
      "loss": 0.879,
      "step": 1382
    },
    {
      "epoch": 0.892258064516129,
      "grad_norm": 1.1171875,
      "learning_rate": 7.451580682543809e-06,
      "loss": 1.1229,
      "step": 1383
    },
    {
      "epoch": 0.8929032258064517,
      "grad_norm": 1.21875,
      "learning_rate": 7.447936880223158e-06,
      "loss": 0.9829,
      "step": 1384
    },
    {
      "epoch": 0.8935483870967742,
      "grad_norm": 0.90625,
      "learning_rate": 7.444291367153156e-06,
      "loss": 0.5737,
      "step": 1385
    },
    {
      "epoch": 0.8941935483870967,
      "grad_norm": 0.87890625,
      "learning_rate": 7.440644145881482e-06,
      "loss": 0.8927,
      "step": 1386
    },
    {
      "epoch": 0.8948387096774194,
      "grad_norm": 0.8125,
      "learning_rate": 7.43699521895701e-06,
      "loss": 0.7735,
      "step": 1387
    },
    {
      "epoch": 0.895483870967742,
      "grad_norm": 0.6953125,
      "learning_rate": 7.433344588929806e-06,
      "loss": 0.6923,
      "step": 1388
    },
    {
      "epoch": 0.8961290322580645,
      "grad_norm": 0.8046875,
      "learning_rate": 7.429692258351126e-06,
      "loss": 0.6292,
      "step": 1389
    },
    {
      "epoch": 0.896774193548387,
      "grad_norm": 0.8203125,
      "learning_rate": 7.4260382297734124e-06,
      "loss": 0.7977,
      "step": 1390
    },
    {
      "epoch": 0.8974193548387097,
      "grad_norm": 1.171875,
      "learning_rate": 7.422382505750297e-06,
      "loss": 1.1127,
      "step": 1391
    },
    {
      "epoch": 0.8980645161290323,
      "grad_norm": 0.77734375,
      "learning_rate": 7.418725088836596e-06,
      "loss": 0.6072,
      "step": 1392
    },
    {
      "epoch": 0.8987096774193548,
      "grad_norm": 1.0,
      "learning_rate": 7.415065981588307e-06,
      "loss": 0.8067,
      "step": 1393
    },
    {
      "epoch": 0.8993548387096775,
      "grad_norm": 0.93359375,
      "learning_rate": 7.411405186562611e-06,
      "loss": 0.788,
      "step": 1394
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.61328125,
      "learning_rate": 7.407742706317868e-06,
      "loss": 0.6299,
      "step": 1395
    },
    {
      "epoch": 0.9006451612903226,
      "grad_norm": 2.453125,
      "learning_rate": 7.404078543413614e-06,
      "loss": 1.02,
      "step": 1396
    },
    {
      "epoch": 0.9012903225806451,
      "grad_norm": 0.84375,
      "learning_rate": 7.400412700410562e-06,
      "loss": 0.635,
      "step": 1397
    },
    {
      "epoch": 0.9019354838709678,
      "grad_norm": 0.625,
      "learning_rate": 7.396745179870601e-06,
      "loss": 0.5357,
      "step": 1398
    },
    {
      "epoch": 0.9025806451612903,
      "grad_norm": 1.125,
      "learning_rate": 7.39307598435679e-06,
      "loss": 0.9091,
      "step": 1399
    },
    {
      "epoch": 0.9032258064516129,
      "grad_norm": 0.875,
      "learning_rate": 7.3894051164333584e-06,
      "loss": 0.6102,
      "step": 1400
    },
    {
      "epoch": 0.9038709677419355,
      "grad_norm": 0.6875,
      "learning_rate": 7.385732578665705e-06,
      "loss": 0.5979,
      "step": 1401
    },
    {
      "epoch": 0.9045161290322581,
      "grad_norm": 1.203125,
      "learning_rate": 7.382058373620399e-06,
      "loss": 0.8913,
      "step": 1402
    },
    {
      "epoch": 0.9051612903225806,
      "grad_norm": 0.69921875,
      "learning_rate": 7.378382503865167e-06,
      "loss": 0.5957,
      "step": 1403
    },
    {
      "epoch": 0.9058064516129032,
      "grad_norm": 0.81640625,
      "learning_rate": 7.374704971968907e-06,
      "loss": 0.7869,
      "step": 1404
    },
    {
      "epoch": 0.9064516129032258,
      "grad_norm": 0.9296875,
      "learning_rate": 7.371025780501675e-06,
      "loss": 0.7143,
      "step": 1405
    },
    {
      "epoch": 0.9070967741935484,
      "grad_norm": 1.109375,
      "learning_rate": 7.367344932034685e-06,
      "loss": 0.9206,
      "step": 1406
    },
    {
      "epoch": 0.9077419354838709,
      "grad_norm": 1.125,
      "learning_rate": 7.363662429140311e-06,
      "loss": 0.9726,
      "step": 1407
    },
    {
      "epoch": 0.9083870967741936,
      "grad_norm": 0.91015625,
      "learning_rate": 7.359978274392085e-06,
      "loss": 0.7784,
      "step": 1408
    },
    {
      "epoch": 0.9090322580645162,
      "grad_norm": 0.94921875,
      "learning_rate": 7.356292470364692e-06,
      "loss": 0.832,
      "step": 1409
    },
    {
      "epoch": 0.9096774193548387,
      "grad_norm": 1.078125,
      "learning_rate": 7.352605019633967e-06,
      "loss": 0.7682,
      "step": 1410
    },
    {
      "epoch": 0.9103225806451613,
      "grad_norm": 0.94140625,
      "learning_rate": 7.348915924776898e-06,
      "loss": 0.7118,
      "step": 1411
    },
    {
      "epoch": 0.9109677419354839,
      "grad_norm": 0.82421875,
      "learning_rate": 7.345225188371624e-06,
      "loss": 0.8745,
      "step": 1412
    },
    {
      "epoch": 0.9116129032258065,
      "grad_norm": 0.96484375,
      "learning_rate": 7.3415328129974295e-06,
      "loss": 0.8991,
      "step": 1413
    },
    {
      "epoch": 0.912258064516129,
      "grad_norm": 0.5390625,
      "learning_rate": 7.3378388012347415e-06,
      "loss": 0.3883,
      "step": 1414
    },
    {
      "epoch": 0.9129032258064517,
      "grad_norm": 1.2109375,
      "learning_rate": 7.334143155665133e-06,
      "loss": 0.9436,
      "step": 1415
    },
    {
      "epoch": 0.9135483870967742,
      "grad_norm": 1.1953125,
      "learning_rate": 7.330445878871323e-06,
      "loss": 1.0111,
      "step": 1416
    },
    {
      "epoch": 0.9141935483870968,
      "grad_norm": 0.9140625,
      "learning_rate": 7.326746973437162e-06,
      "loss": 0.8822,
      "step": 1417
    },
    {
      "epoch": 0.9148387096774193,
      "grad_norm": 2.40625,
      "learning_rate": 7.3230464419476455e-06,
      "loss": 0.8766,
      "step": 1418
    },
    {
      "epoch": 0.915483870967742,
      "grad_norm": 1.46875,
      "learning_rate": 7.319344286988903e-06,
      "loss": 1.0241,
      "step": 1419
    },
    {
      "epoch": 0.9161290322580645,
      "grad_norm": 0.67578125,
      "learning_rate": 7.3156405111481986e-06,
      "loss": 0.4629,
      "step": 1420
    },
    {
      "epoch": 0.9167741935483871,
      "grad_norm": 1.1328125,
      "learning_rate": 7.3119351170139265e-06,
      "loss": 1.0473,
      "step": 1421
    },
    {
      "epoch": 0.9174193548387096,
      "grad_norm": 0.765625,
      "learning_rate": 7.308228107175619e-06,
      "loss": 0.6341,
      "step": 1422
    },
    {
      "epoch": 0.9180645161290323,
      "grad_norm": 0.8984375,
      "learning_rate": 7.30451948422393e-06,
      "loss": 0.8128,
      "step": 1423
    },
    {
      "epoch": 0.9187096774193548,
      "grad_norm": 0.63671875,
      "learning_rate": 7.300809250750645e-06,
      "loss": 0.5239,
      "step": 1424
    },
    {
      "epoch": 0.9193548387096774,
      "grad_norm": 1.453125,
      "learning_rate": 7.297097409348674e-06,
      "loss": 1.2549,
      "step": 1425
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6328125,
      "learning_rate": 7.293383962612047e-06,
      "loss": 0.5768,
      "step": 1426
    },
    {
      "epoch": 0.9206451612903226,
      "grad_norm": 0.640625,
      "learning_rate": 7.289668913135925e-06,
      "loss": 0.657,
      "step": 1427
    },
    {
      "epoch": 0.9212903225806451,
      "grad_norm": 0.9296875,
      "learning_rate": 7.2859522635165795e-06,
      "loss": 0.8043,
      "step": 1428
    },
    {
      "epoch": 0.9219354838709677,
      "grad_norm": 0.8125,
      "learning_rate": 7.282234016351406e-06,
      "loss": 0.731,
      "step": 1429
    },
    {
      "epoch": 0.9225806451612903,
      "grad_norm": 0.8125,
      "learning_rate": 7.278514174238913e-06,
      "loss": 0.9233,
      "step": 1430
    },
    {
      "epoch": 0.9232258064516129,
      "grad_norm": 0.95703125,
      "learning_rate": 7.274792739778728e-06,
      "loss": 0.6576,
      "step": 1431
    },
    {
      "epoch": 0.9238709677419354,
      "grad_norm": 1.1640625,
      "learning_rate": 7.2710697155715855e-06,
      "loss": 0.8438,
      "step": 1432
    },
    {
      "epoch": 0.9245161290322581,
      "grad_norm": 1.0390625,
      "learning_rate": 7.2673451042193375e-06,
      "loss": 0.8876,
      "step": 1433
    },
    {
      "epoch": 0.9251612903225807,
      "grad_norm": 1.0625,
      "learning_rate": 7.263618908324939e-06,
      "loss": 1.1433,
      "step": 1434
    },
    {
      "epoch": 0.9258064516129032,
      "grad_norm": 0.89453125,
      "learning_rate": 7.259891130492456e-06,
      "loss": 0.9093,
      "step": 1435
    },
    {
      "epoch": 0.9264516129032258,
      "grad_norm": 0.87109375,
      "learning_rate": 7.256161773327061e-06,
      "loss": 0.8464,
      "step": 1436
    },
    {
      "epoch": 0.9270967741935484,
      "grad_norm": 1.484375,
      "learning_rate": 7.252430839435025e-06,
      "loss": 1.0749,
      "step": 1437
    },
    {
      "epoch": 0.927741935483871,
      "grad_norm": 0.78125,
      "learning_rate": 7.248698331423727e-06,
      "loss": 0.6183,
      "step": 1438
    },
    {
      "epoch": 0.9283870967741935,
      "grad_norm": 1.09375,
      "learning_rate": 7.244964251901643e-06,
      "loss": 0.9577,
      "step": 1439
    },
    {
      "epoch": 0.9290322580645162,
      "grad_norm": 0.76171875,
      "learning_rate": 7.241228603478347e-06,
      "loss": 0.8976,
      "step": 1440
    },
    {
      "epoch": 0.9296774193548387,
      "grad_norm": 0.90234375,
      "learning_rate": 7.23749138876451e-06,
      "loss": 0.8762,
      "step": 1441
    },
    {
      "epoch": 0.9303225806451613,
      "grad_norm": 0.81640625,
      "learning_rate": 7.2337526103719e-06,
      "loss": 0.7048,
      "step": 1442
    },
    {
      "epoch": 0.9309677419354838,
      "grad_norm": 0.87109375,
      "learning_rate": 7.2300122709133735e-06,
      "loss": 0.9759,
      "step": 1443
    },
    {
      "epoch": 0.9316129032258065,
      "grad_norm": 1.0234375,
      "learning_rate": 7.226270373002878e-06,
      "loss": 0.7258,
      "step": 1444
    },
    {
      "epoch": 0.932258064516129,
      "grad_norm": 0.52734375,
      "learning_rate": 7.222526919255457e-06,
      "loss": 0.4974,
      "step": 1445
    },
    {
      "epoch": 0.9329032258064516,
      "grad_norm": 1.0859375,
      "learning_rate": 7.218781912287231e-06,
      "loss": 1.0938,
      "step": 1446
    },
    {
      "epoch": 0.9335483870967742,
      "grad_norm": 1.4375,
      "learning_rate": 7.215035354715417e-06,
      "loss": 0.9177,
      "step": 1447
    },
    {
      "epoch": 0.9341935483870968,
      "grad_norm": 1.015625,
      "learning_rate": 7.211287249158305e-06,
      "loss": 1.0046,
      "step": 1448
    },
    {
      "epoch": 0.9348387096774193,
      "grad_norm": 0.95703125,
      "learning_rate": 7.207537598235275e-06,
      "loss": 0.9482,
      "step": 1449
    },
    {
      "epoch": 0.9354838709677419,
      "grad_norm": 1.3515625,
      "learning_rate": 7.203786404566782e-06,
      "loss": 0.9707,
      "step": 1450
    },
    {
      "epoch": 0.9361290322580645,
      "grad_norm": 0.8671875,
      "learning_rate": 7.2000336707743635e-06,
      "loss": 0.8309,
      "step": 1451
    },
    {
      "epoch": 0.9367741935483871,
      "grad_norm": 0.91015625,
      "learning_rate": 7.196279399480627e-06,
      "loss": 0.7583,
      "step": 1452
    },
    {
      "epoch": 0.9374193548387096,
      "grad_norm": 0.6796875,
      "learning_rate": 7.192523593309261e-06,
      "loss": 0.6042,
      "step": 1453
    },
    {
      "epoch": 0.9380645161290323,
      "grad_norm": 0.71875,
      "learning_rate": 7.1887662548850235e-06,
      "loss": 0.6868,
      "step": 1454
    },
    {
      "epoch": 0.9387096774193548,
      "grad_norm": 0.7578125,
      "learning_rate": 7.185007386833742e-06,
      "loss": 0.6496,
      "step": 1455
    },
    {
      "epoch": 0.9393548387096774,
      "grad_norm": 0.89453125,
      "learning_rate": 7.181246991782318e-06,
      "loss": 0.746,
      "step": 1456
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.55859375,
      "learning_rate": 7.177485072358715e-06,
      "loss": 0.5652,
      "step": 1457
    },
    {
      "epoch": 0.9406451612903226,
      "grad_norm": 1.0390625,
      "learning_rate": 7.173721631191963e-06,
      "loss": 1.0451,
      "step": 1458
    },
    {
      "epoch": 0.9412903225806452,
      "grad_norm": 0.59765625,
      "learning_rate": 7.169956670912157e-06,
      "loss": 0.6381,
      "step": 1459
    },
    {
      "epoch": 0.9419354838709677,
      "grad_norm": 0.71875,
      "learning_rate": 7.166190194150454e-06,
      "loss": 0.6412,
      "step": 1460
    },
    {
      "epoch": 0.9425806451612904,
      "grad_norm": 0.92578125,
      "learning_rate": 7.162422203539067e-06,
      "loss": 0.7159,
      "step": 1461
    },
    {
      "epoch": 0.9432258064516129,
      "grad_norm": 0.9140625,
      "learning_rate": 7.158652701711271e-06,
      "loss": 0.7139,
      "step": 1462
    },
    {
      "epoch": 0.9438709677419355,
      "grad_norm": 0.91015625,
      "learning_rate": 7.154881691301396e-06,
      "loss": 0.7929,
      "step": 1463
    },
    {
      "epoch": 0.944516129032258,
      "grad_norm": 0.8359375,
      "learning_rate": 7.151109174944826e-06,
      "loss": 0.7638,
      "step": 1464
    },
    {
      "epoch": 0.9451612903225807,
      "grad_norm": 0.734375,
      "learning_rate": 7.147335155277997e-06,
      "loss": 0.6789,
      "step": 1465
    },
    {
      "epoch": 0.9458064516129032,
      "grad_norm": 1.0,
      "learning_rate": 7.143559634938395e-06,
      "loss": 0.8461,
      "step": 1466
    },
    {
      "epoch": 0.9464516129032258,
      "grad_norm": 0.80859375,
      "learning_rate": 7.139782616564559e-06,
      "loss": 0.769,
      "step": 1467
    },
    {
      "epoch": 0.9470967741935484,
      "grad_norm": 0.66015625,
      "learning_rate": 7.136004102796067e-06,
      "loss": 0.653,
      "step": 1468
    },
    {
      "epoch": 0.947741935483871,
      "grad_norm": 0.94921875,
      "learning_rate": 7.132224096273552e-06,
      "loss": 0.8684,
      "step": 1469
    },
    {
      "epoch": 0.9483870967741935,
      "grad_norm": 0.70703125,
      "learning_rate": 7.12844259963868e-06,
      "loss": 0.6723,
      "step": 1470
    },
    {
      "epoch": 0.9490322580645161,
      "grad_norm": 0.96484375,
      "learning_rate": 7.124659615534166e-06,
      "loss": 0.7988,
      "step": 1471
    },
    {
      "epoch": 0.9496774193548387,
      "grad_norm": 0.6875,
      "learning_rate": 7.120875146603761e-06,
      "loss": 0.5466,
      "step": 1472
    },
    {
      "epoch": 0.9503225806451613,
      "grad_norm": 1.4375,
      "learning_rate": 7.117089195492255e-06,
      "loss": 1.2411,
      "step": 1473
    },
    {
      "epoch": 0.9509677419354838,
      "grad_norm": 0.98828125,
      "learning_rate": 7.113301764845473e-06,
      "loss": 0.9041,
      "step": 1474
    },
    {
      "epoch": 0.9516129032258065,
      "grad_norm": 0.8671875,
      "learning_rate": 7.109512857310273e-06,
      "loss": 0.9033,
      "step": 1475
    },
    {
      "epoch": 0.952258064516129,
      "grad_norm": 1.15625,
      "learning_rate": 7.10572247553455e-06,
      "loss": 0.8893,
      "step": 1476
    },
    {
      "epoch": 0.9529032258064516,
      "grad_norm": 1.0,
      "learning_rate": 7.101930622167221e-06,
      "loss": 0.8471,
      "step": 1477
    },
    {
      "epoch": 0.9535483870967741,
      "grad_norm": 0.67578125,
      "learning_rate": 7.098137299858241e-06,
      "loss": 0.6111,
      "step": 1478
    },
    {
      "epoch": 0.9541935483870968,
      "grad_norm": 0.875,
      "learning_rate": 7.094342511258585e-06,
      "loss": 0.8197,
      "step": 1479
    },
    {
      "epoch": 0.9548387096774194,
      "grad_norm": 0.85546875,
      "learning_rate": 7.090546259020252e-06,
      "loss": 0.8093,
      "step": 1480
    },
    {
      "epoch": 0.9554838709677419,
      "grad_norm": 0.55859375,
      "learning_rate": 7.086748545796269e-06,
      "loss": 0.4471,
      "step": 1481
    },
    {
      "epoch": 0.9561290322580646,
      "grad_norm": 0.921875,
      "learning_rate": 7.08294937424068e-06,
      "loss": 0.7518,
      "step": 1482
    },
    {
      "epoch": 0.9567741935483871,
      "grad_norm": 0.83984375,
      "learning_rate": 7.07914874700855e-06,
      "loss": 0.8507,
      "step": 1483
    },
    {
      "epoch": 0.9574193548387097,
      "grad_norm": 0.8984375,
      "learning_rate": 7.075346666755962e-06,
      "loss": 0.8663,
      "step": 1484
    },
    {
      "epoch": 0.9580645161290322,
      "grad_norm": 0.7734375,
      "learning_rate": 7.071543136140011e-06,
      "loss": 0.6861,
      "step": 1485
    },
    {
      "epoch": 0.9587096774193549,
      "grad_norm": 1.15625,
      "learning_rate": 7.067738157818808e-06,
      "loss": 1.1045,
      "step": 1486
    },
    {
      "epoch": 0.9593548387096774,
      "grad_norm": 0.76953125,
      "learning_rate": 7.0639317344514766e-06,
      "loss": 0.6635,
      "step": 1487
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.95703125,
      "learning_rate": 7.060123868698151e-06,
      "loss": 0.6899,
      "step": 1488
    },
    {
      "epoch": 0.9606451612903226,
      "grad_norm": 0.76171875,
      "learning_rate": 7.056314563219968e-06,
      "loss": 0.5282,
      "step": 1489
    },
    {
      "epoch": 0.9612903225806452,
      "grad_norm": 0.71875,
      "learning_rate": 7.052503820679076e-06,
      "loss": 0.6283,
      "step": 1490
    },
    {
      "epoch": 0.9619354838709677,
      "grad_norm": 1.28125,
      "learning_rate": 7.0486916437386265e-06,
      "loss": 0.9582,
      "step": 1491
    },
    {
      "epoch": 0.9625806451612903,
      "grad_norm": 1.0703125,
      "learning_rate": 7.044878035062772e-06,
      "loss": 0.8581,
      "step": 1492
    },
    {
      "epoch": 0.9632258064516129,
      "grad_norm": 0.96484375,
      "learning_rate": 7.041062997316666e-06,
      "loss": 0.8027,
      "step": 1493
    },
    {
      "epoch": 0.9638709677419355,
      "grad_norm": 0.89453125,
      "learning_rate": 7.037246533166462e-06,
      "loss": 0.6825,
      "step": 1494
    },
    {
      "epoch": 0.964516129032258,
      "grad_norm": 1.078125,
      "learning_rate": 7.03342864527931e-06,
      "loss": 0.9333,
      "step": 1495
    },
    {
      "epoch": 0.9651612903225807,
      "grad_norm": 0.498046875,
      "learning_rate": 7.029609336323352e-06,
      "loss": 0.4311,
      "step": 1496
    },
    {
      "epoch": 0.9658064516129032,
      "grad_norm": 1.296875,
      "learning_rate": 7.025788608967728e-06,
      "loss": 0.8951,
      "step": 1497
    },
    {
      "epoch": 0.9664516129032258,
      "grad_norm": 0.8203125,
      "learning_rate": 7.021966465882567e-06,
      "loss": 0.7625,
      "step": 1498
    },
    {
      "epoch": 0.9670967741935483,
      "grad_norm": 0.76171875,
      "learning_rate": 7.018142909738984e-06,
      "loss": 0.8547,
      "step": 1499
    },
    {
      "epoch": 0.967741935483871,
      "grad_norm": 1.1484375,
      "learning_rate": 7.014317943209089e-06,
      "loss": 1.0058,
      "step": 1500
    },
    {
      "epoch": 0.9683870967741935,
      "grad_norm": 0.6953125,
      "learning_rate": 7.01049156896597e-06,
      "loss": 0.6018,
      "step": 1501
    },
    {
      "epoch": 0.9690322580645161,
      "grad_norm": 0.91015625,
      "learning_rate": 7.0066637896837055e-06,
      "loss": 0.8159,
      "step": 1502
    },
    {
      "epoch": 0.9696774193548388,
      "grad_norm": 1.4453125,
      "learning_rate": 7.0028346080373485e-06,
      "loss": 1.1053,
      "step": 1503
    },
    {
      "epoch": 0.9703225806451613,
      "grad_norm": 0.79296875,
      "learning_rate": 6.99900402670294e-06,
      "loss": 0.688,
      "step": 1504
    },
    {
      "epoch": 0.9709677419354839,
      "grad_norm": 0.734375,
      "learning_rate": 6.995172048357493e-06,
      "loss": 0.6355,
      "step": 1505
    },
    {
      "epoch": 0.9716129032258064,
      "grad_norm": 0.78125,
      "learning_rate": 6.991338675679e-06,
      "loss": 0.6787,
      "step": 1506
    },
    {
      "epoch": 0.9722580645161291,
      "grad_norm": 1.3125,
      "learning_rate": 6.987503911346427e-06,
      "loss": 0.9207,
      "step": 1507
    },
    {
      "epoch": 0.9729032258064516,
      "grad_norm": 0.89453125,
      "learning_rate": 6.983667758039711e-06,
      "loss": 0.7043,
      "step": 1508
    },
    {
      "epoch": 0.9735483870967742,
      "grad_norm": 1.21875,
      "learning_rate": 6.979830218439765e-06,
      "loss": 0.8895,
      "step": 1509
    },
    {
      "epoch": 0.9741935483870968,
      "grad_norm": 1.15625,
      "learning_rate": 6.9759912952284635e-06,
      "loss": 1.0068,
      "step": 1510
    },
    {
      "epoch": 0.9748387096774194,
      "grad_norm": 0.87890625,
      "learning_rate": 6.972150991088653e-06,
      "loss": 0.5481,
      "step": 1511
    },
    {
      "epoch": 0.9754838709677419,
      "grad_norm": 1.1484375,
      "learning_rate": 6.9683093087041446e-06,
      "loss": 0.8151,
      "step": 1512
    },
    {
      "epoch": 0.9761290322580645,
      "grad_norm": 1.40625,
      "learning_rate": 6.964466250759709e-06,
      "loss": 0.9731,
      "step": 1513
    },
    {
      "epoch": 0.9767741935483871,
      "grad_norm": 1.015625,
      "learning_rate": 6.960621819941084e-06,
      "loss": 0.9209,
      "step": 1514
    },
    {
      "epoch": 0.9774193548387097,
      "grad_norm": 0.64453125,
      "learning_rate": 6.956776018934961e-06,
      "loss": 0.5935,
      "step": 1515
    },
    {
      "epoch": 0.9780645161290322,
      "grad_norm": 0.72265625,
      "learning_rate": 6.952928850428995e-06,
      "loss": 0.8034,
      "step": 1516
    },
    {
      "epoch": 0.9787096774193549,
      "grad_norm": 0.83203125,
      "learning_rate": 6.949080317111789e-06,
      "loss": 1.0141,
      "step": 1517
    },
    {
      "epoch": 0.9793548387096774,
      "grad_norm": 0.78125,
      "learning_rate": 6.945230421672906e-06,
      "loss": 0.7797,
      "step": 1518
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3671875,
      "learning_rate": 6.9413791668028595e-06,
      "loss": 1.0459,
      "step": 1519
    },
    {
      "epoch": 0.9806451612903225,
      "grad_norm": 0.91015625,
      "learning_rate": 6.937526555193113e-06,
      "loss": 0.8758,
      "step": 1520
    },
    {
      "epoch": 0.9812903225806452,
      "grad_norm": 0.94921875,
      "learning_rate": 6.933672589536078e-06,
      "loss": 0.9122,
      "step": 1521
    },
    {
      "epoch": 0.9819354838709677,
      "grad_norm": 0.796875,
      "learning_rate": 6.92981727252511e-06,
      "loss": 0.6295,
      "step": 1522
    },
    {
      "epoch": 0.9825806451612903,
      "grad_norm": 0.79296875,
      "learning_rate": 6.925960606854513e-06,
      "loss": 0.6992,
      "step": 1523
    },
    {
      "epoch": 0.983225806451613,
      "grad_norm": 0.625,
      "learning_rate": 6.92210259521953e-06,
      "loss": 0.6304,
      "step": 1524
    },
    {
      "epoch": 0.9838709677419355,
      "grad_norm": 0.921875,
      "learning_rate": 6.918243240316347e-06,
      "loss": 0.7207,
      "step": 1525
    },
    {
      "epoch": 0.984516129032258,
      "grad_norm": 0.765625,
      "learning_rate": 6.914382544842087e-06,
      "loss": 0.5364,
      "step": 1526
    },
    {
      "epoch": 0.9851612903225806,
      "grad_norm": 1.703125,
      "learning_rate": 6.910520511494812e-06,
      "loss": 1.0053,
      "step": 1527
    },
    {
      "epoch": 0.9858064516129033,
      "grad_norm": 0.8671875,
      "learning_rate": 6.906657142973515e-06,
      "loss": 0.9141,
      "step": 1528
    },
    {
      "epoch": 0.9864516129032258,
      "grad_norm": 1.0859375,
      "learning_rate": 6.902792441978129e-06,
      "loss": 0.7647,
      "step": 1529
    },
    {
      "epoch": 0.9870967741935484,
      "grad_norm": 1.1640625,
      "learning_rate": 6.89892641120951e-06,
      "loss": 1.053,
      "step": 1530
    },
    {
      "epoch": 0.987741935483871,
      "grad_norm": 0.640625,
      "learning_rate": 6.895059053369449e-06,
      "loss": 0.5119,
      "step": 1531
    },
    {
      "epoch": 0.9883870967741936,
      "grad_norm": 0.6171875,
      "learning_rate": 6.891190371160663e-06,
      "loss": 0.513,
      "step": 1532
    },
    {
      "epoch": 0.9890322580645161,
      "grad_norm": 0.8125,
      "learning_rate": 6.8873203672867935e-06,
      "loss": 0.8282,
      "step": 1533
    },
    {
      "epoch": 0.9896774193548387,
      "grad_norm": 0.93359375,
      "learning_rate": 6.883449044452408e-06,
      "loss": 0.7093,
      "step": 1534
    },
    {
      "epoch": 0.9903225806451613,
      "grad_norm": 1.140625,
      "learning_rate": 6.879576405362992e-06,
      "loss": 1.1293,
      "step": 1535
    },
    {
      "epoch": 0.9909677419354839,
      "grad_norm": 1.0,
      "learning_rate": 6.8757024527249536e-06,
      "loss": 0.8079,
      "step": 1536
    },
    {
      "epoch": 0.9916129032258064,
      "grad_norm": 1.21875,
      "learning_rate": 6.871827189245619e-06,
      "loss": 1.1038,
      "step": 1537
    },
    {
      "epoch": 0.9922580645161291,
      "grad_norm": 0.87890625,
      "learning_rate": 6.86795061763323e-06,
      "loss": 0.7725,
      "step": 1538
    },
    {
      "epoch": 0.9929032258064516,
      "grad_norm": 0.59765625,
      "learning_rate": 6.864072740596941e-06,
      "loss": 0.5624,
      "step": 1539
    },
    {
      "epoch": 0.9935483870967742,
      "grad_norm": 1.140625,
      "learning_rate": 6.8601935608468215e-06,
      "loss": 0.9084,
      "step": 1540
    },
    {
      "epoch": 0.9941935483870967,
      "grad_norm": 0.89453125,
      "learning_rate": 6.856313081093848e-06,
      "loss": 0.7702,
      "step": 1541
    },
    {
      "epoch": 0.9948387096774194,
      "grad_norm": 1.0234375,
      "learning_rate": 6.8524313040499104e-06,
      "loss": 0.9632,
      "step": 1542
    },
    {
      "epoch": 0.9954838709677419,
      "grad_norm": 0.72265625,
      "learning_rate": 6.848548232427801e-06,
      "loss": 0.4768,
      "step": 1543
    },
    {
      "epoch": 0.9961290322580645,
      "grad_norm": 1.0234375,
      "learning_rate": 6.844663868941217e-06,
      "loss": 0.8201,
      "step": 1544
    },
    {
      "epoch": 0.9967741935483871,
      "grad_norm": 0.78125,
      "learning_rate": 6.840778216304762e-06,
      "loss": 0.8542,
      "step": 1545
    },
    {
      "epoch": 0.9974193548387097,
      "grad_norm": 0.72265625,
      "learning_rate": 6.8368912772339346e-06,
      "loss": 0.6525,
      "step": 1546
    },
    {
      "epoch": 0.9980645161290322,
      "grad_norm": 0.53515625,
      "learning_rate": 6.8330030544451394e-06,
      "loss": 0.4428,
      "step": 1547
    },
    {
      "epoch": 0.9987096774193548,
      "grad_norm": 1.25,
      "learning_rate": 6.829113550655673e-06,
      "loss": 0.9948,
      "step": 1548
    },
    {
      "epoch": 0.9993548387096775,
      "grad_norm": 1.5,
      "learning_rate": 6.825222768583729e-06,
      "loss": 1.5717,
      "step": 1549
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.498046875,
      "learning_rate": 6.821330710948393e-06,
      "loss": 0.4887,
      "step": 1550
    },
    {
      "epoch": 1.0006451612903227,
      "grad_norm": 0.98828125,
      "learning_rate": 6.817437380469647e-06,
      "loss": 0.5321,
      "step": 1551
    },
    {
      "epoch": 1.001290322580645,
      "grad_norm": 0.546875,
      "learning_rate": 6.8135427798683565e-06,
      "loss": 0.5076,
      "step": 1552
    },
    {
      "epoch": 1.0019354838709678,
      "grad_norm": 0.42578125,
      "learning_rate": 6.809646911866275e-06,
      "loss": 0.4048,
      "step": 1553
    },
    {
      "epoch": 1.0025806451612904,
      "grad_norm": 0.8515625,
      "learning_rate": 6.805749779186046e-06,
      "loss": 0.8909,
      "step": 1554
    },
    {
      "epoch": 1.0032258064516129,
      "grad_norm": 1.578125,
      "learning_rate": 6.801851384551192e-06,
      "loss": 0.9206,
      "step": 1555
    },
    {
      "epoch": 1.0038709677419355,
      "grad_norm": 0.90625,
      "learning_rate": 6.7979517306861255e-06,
      "loss": 0.6323,
      "step": 1556
    },
    {
      "epoch": 1.004516129032258,
      "grad_norm": 0.77734375,
      "learning_rate": 6.794050820316127e-06,
      "loss": 0.7356,
      "step": 1557
    },
    {
      "epoch": 1.0051612903225806,
      "grad_norm": 0.828125,
      "learning_rate": 6.790148656167366e-06,
      "loss": 0.7251,
      "step": 1558
    },
    {
      "epoch": 1.0058064516129033,
      "grad_norm": 1.1328125,
      "learning_rate": 6.78624524096688e-06,
      "loss": 1.0077,
      "step": 1559
    },
    {
      "epoch": 1.0064516129032257,
      "grad_norm": 0.72265625,
      "learning_rate": 6.782340577442588e-06,
      "loss": 0.5304,
      "step": 1560
    },
    {
      "epoch": 1.0070967741935484,
      "grad_norm": 0.671875,
      "learning_rate": 6.7784346683232745e-06,
      "loss": 0.6632,
      "step": 1561
    },
    {
      "epoch": 1.007741935483871,
      "grad_norm": 0.98046875,
      "learning_rate": 6.774527516338603e-06,
      "loss": 0.8707,
      "step": 1562
    },
    {
      "epoch": 1.0083870967741935,
      "grad_norm": 0.85546875,
      "learning_rate": 6.770619124219095e-06,
      "loss": 0.8133,
      "step": 1563
    },
    {
      "epoch": 1.0090322580645161,
      "grad_norm": 1.0078125,
      "learning_rate": 6.766709494696146e-06,
      "loss": 0.8064,
      "step": 1564
    },
    {
      "epoch": 1.0096774193548388,
      "grad_norm": 0.82421875,
      "learning_rate": 6.7627986305020154e-06,
      "loss": 0.8336,
      "step": 1565
    },
    {
      "epoch": 1.0103225806451612,
      "grad_norm": 0.71875,
      "learning_rate": 6.758886534369824e-06,
      "loss": 0.6064,
      "step": 1566
    },
    {
      "epoch": 1.0109677419354839,
      "grad_norm": 0.91796875,
      "learning_rate": 6.754973209033554e-06,
      "loss": 0.6675,
      "step": 1567
    },
    {
      "epoch": 1.0116129032258065,
      "grad_norm": 1.21875,
      "learning_rate": 6.751058657228046e-06,
      "loss": 1.3338,
      "step": 1568
    },
    {
      "epoch": 1.012258064516129,
      "grad_norm": 1.5234375,
      "learning_rate": 6.7471428816890005e-06,
      "loss": 0.978,
      "step": 1569
    },
    {
      "epoch": 1.0129032258064516,
      "grad_norm": 1.0234375,
      "learning_rate": 6.743225885152969e-06,
      "loss": 0.8951,
      "step": 1570
    },
    {
      "epoch": 1.013548387096774,
      "grad_norm": 0.9765625,
      "learning_rate": 6.739307670357359e-06,
      "loss": 0.9329,
      "step": 1571
    },
    {
      "epoch": 1.0141935483870967,
      "grad_norm": 0.79296875,
      "learning_rate": 6.735388240040428e-06,
      "loss": 0.7605,
      "step": 1572
    },
    {
      "epoch": 1.0148387096774194,
      "grad_norm": 0.75,
      "learning_rate": 6.731467596941285e-06,
      "loss": 0.6582,
      "step": 1573
    },
    {
      "epoch": 1.0154838709677418,
      "grad_norm": 0.8984375,
      "learning_rate": 6.7275457437998845e-06,
      "loss": 0.7711,
      "step": 1574
    },
    {
      "epoch": 1.0161290322580645,
      "grad_norm": 0.96875,
      "learning_rate": 6.7236226833570275e-06,
      "loss": 0.8502,
      "step": 1575
    },
    {
      "epoch": 1.0167741935483872,
      "grad_norm": 1.34375,
      "learning_rate": 6.71969841835436e-06,
      "loss": 1.0125,
      "step": 1576
    },
    {
      "epoch": 1.0174193548387096,
      "grad_norm": 0.6953125,
      "learning_rate": 6.715772951534366e-06,
      "loss": 0.6454,
      "step": 1577
    },
    {
      "epoch": 1.0180645161290323,
      "grad_norm": 0.6328125,
      "learning_rate": 6.7118462856403755e-06,
      "loss": 0.4994,
      "step": 1578
    },
    {
      "epoch": 1.018709677419355,
      "grad_norm": 1.4296875,
      "learning_rate": 6.70791842341655e-06,
      "loss": 1.0018,
      "step": 1579
    },
    {
      "epoch": 1.0193548387096774,
      "grad_norm": 0.8203125,
      "learning_rate": 6.703989367607891e-06,
      "loss": 0.6518,
      "step": 1580
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.90625,
      "learning_rate": 6.700059120960229e-06,
      "loss": 0.8562,
      "step": 1581
    },
    {
      "epoch": 1.0206451612903227,
      "grad_norm": 0.9609375,
      "learning_rate": 6.696127686220239e-06,
      "loss": 0.848,
      "step": 1582
    },
    {
      "epoch": 1.0212903225806451,
      "grad_norm": 1.1171875,
      "learning_rate": 6.69219506613541e-06,
      "loss": 0.8549,
      "step": 1583
    },
    {
      "epoch": 1.0219354838709678,
      "grad_norm": 0.87109375,
      "learning_rate": 6.688261263454072e-06,
      "loss": 0.6904,
      "step": 1584
    },
    {
      "epoch": 1.0225806451612902,
      "grad_norm": 1.0390625,
      "learning_rate": 6.684326280925376e-06,
      "loss": 1.1758,
      "step": 1585
    },
    {
      "epoch": 1.0232258064516129,
      "grad_norm": 0.82421875,
      "learning_rate": 6.680390121299298e-06,
      "loss": 0.8479,
      "step": 1586
    },
    {
      "epoch": 1.0238709677419355,
      "grad_norm": 0.71484375,
      "learning_rate": 6.676452787326638e-06,
      "loss": 0.6821,
      "step": 1587
    },
    {
      "epoch": 1.024516129032258,
      "grad_norm": 0.6796875,
      "learning_rate": 6.672514281759015e-06,
      "loss": 0.5801,
      "step": 1588
    },
    {
      "epoch": 1.0251612903225806,
      "grad_norm": 1.34375,
      "learning_rate": 6.66857460734887e-06,
      "loss": 0.9633,
      "step": 1589
    },
    {
      "epoch": 1.0258064516129033,
      "grad_norm": 1.15625,
      "learning_rate": 6.664633766849453e-06,
      "loss": 0.9148,
      "step": 1590
    },
    {
      "epoch": 1.0264516129032257,
      "grad_norm": 1.4453125,
      "learning_rate": 6.660691763014838e-06,
      "loss": 0.8785,
      "step": 1591
    },
    {
      "epoch": 1.0270967741935484,
      "grad_norm": 0.6015625,
      "learning_rate": 6.656748598599908e-06,
      "loss": 0.5959,
      "step": 1592
    },
    {
      "epoch": 1.027741935483871,
      "grad_norm": 1.203125,
      "learning_rate": 6.652804276360358e-06,
      "loss": 0.9135,
      "step": 1593
    },
    {
      "epoch": 1.0283870967741935,
      "grad_norm": 0.89453125,
      "learning_rate": 6.64885879905269e-06,
      "loss": 0.7736,
      "step": 1594
    },
    {
      "epoch": 1.0290322580645161,
      "grad_norm": 0.86328125,
      "learning_rate": 6.6449121694342154e-06,
      "loss": 0.5667,
      "step": 1595
    },
    {
      "epoch": 1.0296774193548388,
      "grad_norm": 0.734375,
      "learning_rate": 6.6409643902630506e-06,
      "loss": 0.8633,
      "step": 1596
    },
    {
      "epoch": 1.0303225806451612,
      "grad_norm": 1.0859375,
      "learning_rate": 6.6370154642981144e-06,
      "loss": 0.6813,
      "step": 1597
    },
    {
      "epoch": 1.030967741935484,
      "grad_norm": 0.9140625,
      "learning_rate": 6.633065394299131e-06,
      "loss": 0.9061,
      "step": 1598
    },
    {
      "epoch": 1.0316129032258066,
      "grad_norm": 1.21875,
      "learning_rate": 6.629114183026615e-06,
      "loss": 0.9703,
      "step": 1599
    },
    {
      "epoch": 1.032258064516129,
      "grad_norm": 1.1171875,
      "learning_rate": 6.625161833241888e-06,
      "loss": 1.1961,
      "step": 1600
    },
    {
      "epoch": 1.0329032258064517,
      "grad_norm": 1.4765625,
      "learning_rate": 6.621208347707062e-06,
      "loss": 1.1997,
      "step": 1601
    },
    {
      "epoch": 1.033548387096774,
      "grad_norm": 1.359375,
      "learning_rate": 6.6172537291850465e-06,
      "loss": 1.1047,
      "step": 1602
    },
    {
      "epoch": 1.0341935483870968,
      "grad_norm": 1.03125,
      "learning_rate": 6.6132979804395395e-06,
      "loss": 0.9013,
      "step": 1603
    },
    {
      "epoch": 1.0348387096774194,
      "grad_norm": 1.21875,
      "learning_rate": 6.60934110423503e-06,
      "loss": 0.9712,
      "step": 1604
    },
    {
      "epoch": 1.0354838709677419,
      "grad_norm": 0.71875,
      "learning_rate": 6.6053831033367935e-06,
      "loss": 0.5252,
      "step": 1605
    },
    {
      "epoch": 1.0361290322580645,
      "grad_norm": 0.984375,
      "learning_rate": 6.601423980510896e-06,
      "loss": 0.9422,
      "step": 1606
    },
    {
      "epoch": 1.0367741935483872,
      "grad_norm": 1.03125,
      "learning_rate": 6.5974637385241824e-06,
      "loss": 0.7257,
      "step": 1607
    },
    {
      "epoch": 1.0374193548387096,
      "grad_norm": 0.95703125,
      "learning_rate": 6.593502380144282e-06,
      "loss": 0.6808,
      "step": 1608
    },
    {
      "epoch": 1.0380645161290323,
      "grad_norm": 0.953125,
      "learning_rate": 6.589539908139605e-06,
      "loss": 0.835,
      "step": 1609
    },
    {
      "epoch": 1.038709677419355,
      "grad_norm": 0.984375,
      "learning_rate": 6.585576325279337e-06,
      "loss": 0.7972,
      "step": 1610
    },
    {
      "epoch": 1.0393548387096774,
      "grad_norm": 0.7578125,
      "learning_rate": 6.581611634333445e-06,
      "loss": 0.5809,
      "step": 1611
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.7421875,
      "learning_rate": 6.5776458380726635e-06,
      "loss": 1.2978,
      "step": 1612
    },
    {
      "epoch": 1.0406451612903225,
      "grad_norm": 0.8515625,
      "learning_rate": 6.573678939268507e-06,
      "loss": 0.6821,
      "step": 1613
    },
    {
      "epoch": 1.0412903225806451,
      "grad_norm": 0.984375,
      "learning_rate": 6.569710940693254e-06,
      "loss": 0.8445,
      "step": 1614
    },
    {
      "epoch": 1.0419354838709678,
      "grad_norm": 0.90625,
      "learning_rate": 6.5657418451199565e-06,
      "loss": 0.6315,
      "step": 1615
    },
    {
      "epoch": 1.0425806451612902,
      "grad_norm": 0.93359375,
      "learning_rate": 6.56177165532243e-06,
      "loss": 0.6133,
      "step": 1616
    },
    {
      "epoch": 1.043225806451613,
      "grad_norm": 0.90234375,
      "learning_rate": 6.557800374075254e-06,
      "loss": 0.6638,
      "step": 1617
    },
    {
      "epoch": 1.0438709677419356,
      "grad_norm": 0.66015625,
      "learning_rate": 6.5538280041537765e-06,
      "loss": 0.4986,
      "step": 1618
    },
    {
      "epoch": 1.044516129032258,
      "grad_norm": 1.1171875,
      "learning_rate": 6.549854548334097e-06,
      "loss": 1.0417,
      "step": 1619
    },
    {
      "epoch": 1.0451612903225806,
      "grad_norm": 0.9375,
      "learning_rate": 6.545880009393084e-06,
      "loss": 0.8261,
      "step": 1620
    },
    {
      "epoch": 1.0458064516129033,
      "grad_norm": 0.87890625,
      "learning_rate": 6.541904390108354e-06,
      "loss": 0.7819,
      "step": 1621
    },
    {
      "epoch": 1.0464516129032257,
      "grad_norm": 0.6953125,
      "learning_rate": 6.537927693258288e-06,
      "loss": 0.5873,
      "step": 1622
    },
    {
      "epoch": 1.0470967741935484,
      "grad_norm": 1.015625,
      "learning_rate": 6.533949921622008e-06,
      "loss": 0.7623,
      "step": 1623
    },
    {
      "epoch": 1.047741935483871,
      "grad_norm": 0.8046875,
      "learning_rate": 6.5299710779794e-06,
      "loss": 0.6574,
      "step": 1624
    },
    {
      "epoch": 1.0483870967741935,
      "grad_norm": 0.734375,
      "learning_rate": 6.525991165111088e-06,
      "loss": 0.6813,
      "step": 1625
    },
    {
      "epoch": 1.0490322580645162,
      "grad_norm": 0.93359375,
      "learning_rate": 6.522010185798451e-06,
      "loss": 0.7288,
      "step": 1626
    },
    {
      "epoch": 1.0496774193548386,
      "grad_norm": 0.890625,
      "learning_rate": 6.5180281428236116e-06,
      "loss": 0.7903,
      "step": 1627
    },
    {
      "epoch": 1.0503225806451613,
      "grad_norm": 0.8515625,
      "learning_rate": 6.51404503896943e-06,
      "loss": 0.5963,
      "step": 1628
    },
    {
      "epoch": 1.050967741935484,
      "grad_norm": 0.68359375,
      "learning_rate": 6.5100608770195165e-06,
      "loss": 0.5075,
      "step": 1629
    },
    {
      "epoch": 1.0516129032258064,
      "grad_norm": 1.0703125,
      "learning_rate": 6.506075659758215e-06,
      "loss": 0.6705,
      "step": 1630
    },
    {
      "epoch": 1.052258064516129,
      "grad_norm": 0.5703125,
      "learning_rate": 6.5020893899706095e-06,
      "loss": 0.646,
      "step": 1631
    },
    {
      "epoch": 1.0529032258064517,
      "grad_norm": 1.0234375,
      "learning_rate": 6.498102070442518e-06,
      "loss": 0.8451,
      "step": 1632
    },
    {
      "epoch": 1.0535483870967741,
      "grad_norm": 0.69921875,
      "learning_rate": 6.494113703960494e-06,
      "loss": 0.8388,
      "step": 1633
    },
    {
      "epoch": 1.0541935483870968,
      "grad_norm": 0.671875,
      "learning_rate": 6.4901242933118215e-06,
      "loss": 0.5315,
      "step": 1634
    },
    {
      "epoch": 1.0548387096774194,
      "grad_norm": 0.6875,
      "learning_rate": 6.486133841284512e-06,
      "loss": 0.7165,
      "step": 1635
    },
    {
      "epoch": 1.0554838709677419,
      "grad_norm": 0.64453125,
      "learning_rate": 6.48214235066731e-06,
      "loss": 0.6686,
      "step": 1636
    },
    {
      "epoch": 1.0561290322580645,
      "grad_norm": 1.015625,
      "learning_rate": 6.47814982424968e-06,
      "loss": 0.6564,
      "step": 1637
    },
    {
      "epoch": 1.0567741935483872,
      "grad_norm": 1.1484375,
      "learning_rate": 6.474156264821816e-06,
      "loss": 0.7675,
      "step": 1638
    },
    {
      "epoch": 1.0574193548387096,
      "grad_norm": 1.1953125,
      "learning_rate": 6.470161675174629e-06,
      "loss": 0.9756,
      "step": 1639
    },
    {
      "epoch": 1.0580645161290323,
      "grad_norm": 1.03125,
      "learning_rate": 6.466166058099753e-06,
      "loss": 0.8374,
      "step": 1640
    },
    {
      "epoch": 1.0587096774193547,
      "grad_norm": 0.83984375,
      "learning_rate": 6.462169416389539e-06,
      "loss": 0.8307,
      "step": 1641
    },
    {
      "epoch": 1.0593548387096774,
      "grad_norm": 1.5703125,
      "learning_rate": 6.458171752837054e-06,
      "loss": 1.4426,
      "step": 1642
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.94921875,
      "learning_rate": 6.454173070236079e-06,
      "loss": 0.5614,
      "step": 1643
    },
    {
      "epoch": 1.0606451612903225,
      "grad_norm": 0.60546875,
      "learning_rate": 6.4501733713811095e-06,
      "loss": 0.6289,
      "step": 1644
    },
    {
      "epoch": 1.0612903225806452,
      "grad_norm": 0.94140625,
      "learning_rate": 6.446172659067349e-06,
      "loss": 0.8972,
      "step": 1645
    },
    {
      "epoch": 1.0619354838709678,
      "grad_norm": 0.86328125,
      "learning_rate": 6.442170936090706e-06,
      "loss": 0.7058,
      "step": 1646
    },
    {
      "epoch": 1.0625806451612902,
      "grad_norm": 1.2578125,
      "learning_rate": 6.438168205247801e-06,
      "loss": 1.0715,
      "step": 1647
    },
    {
      "epoch": 1.063225806451613,
      "grad_norm": 0.796875,
      "learning_rate": 6.434164469335956e-06,
      "loss": 0.769,
      "step": 1648
    },
    {
      "epoch": 1.0638709677419356,
      "grad_norm": 0.9375,
      "learning_rate": 6.430159731153198e-06,
      "loss": 0.6641,
      "step": 1649
    },
    {
      "epoch": 1.064516129032258,
      "grad_norm": 0.8359375,
      "learning_rate": 6.426153993498251e-06,
      "loss": 0.6348,
      "step": 1650
    },
    {
      "epoch": 1.064516129032258,
      "eval_loss": 0.9046770334243774,
      "eval_model_preparation_time": 0.0144,
      "eval_runtime": 82.9448,
      "eval_samples_per_second": 3.593,
      "eval_steps_per_second": 3.593,
      "step": 1650
    },
    {
      "epoch": 1.0651612903225807,
      "grad_norm": 0.9140625,
      "learning_rate": 6.422147259170539e-06,
      "loss": 0.7743,
      "step": 1651
    },
    {
      "epoch": 1.0658064516129033,
      "grad_norm": 0.94921875,
      "learning_rate": 6.418139530970181e-06,
      "loss": 0.7838,
      "step": 1652
    },
    {
      "epoch": 1.0664516129032258,
      "grad_norm": 1.21875,
      "learning_rate": 6.414130811697995e-06,
      "loss": 0.9962,
      "step": 1653
    },
    {
      "epoch": 1.0670967741935484,
      "grad_norm": 0.9296875,
      "learning_rate": 6.410121104155487e-06,
      "loss": 0.8499,
      "step": 1654
    },
    {
      "epoch": 1.067741935483871,
      "grad_norm": 0.828125,
      "learning_rate": 6.406110411144855e-06,
      "loss": 0.6782,
      "step": 1655
    },
    {
      "epoch": 1.0683870967741935,
      "grad_norm": 0.890625,
      "learning_rate": 6.402098735468987e-06,
      "loss": 0.7605,
      "step": 1656
    },
    {
      "epoch": 1.0690322580645162,
      "grad_norm": 0.94921875,
      "learning_rate": 6.398086079931455e-06,
      "loss": 0.67,
      "step": 1657
    },
    {
      "epoch": 1.0696774193548386,
      "grad_norm": 0.98046875,
      "learning_rate": 6.3940724473365186e-06,
      "loss": 0.9154,
      "step": 1658
    },
    {
      "epoch": 1.0703225806451613,
      "grad_norm": 1.515625,
      "learning_rate": 6.3900578404891204e-06,
      "loss": 1.192,
      "step": 1659
    },
    {
      "epoch": 1.070967741935484,
      "grad_norm": 0.640625,
      "learning_rate": 6.386042262194882e-06,
      "loss": 0.6026,
      "step": 1660
    },
    {
      "epoch": 1.0716129032258064,
      "grad_norm": 1.046875,
      "learning_rate": 6.382025715260102e-06,
      "loss": 0.994,
      "step": 1661
    },
    {
      "epoch": 1.072258064516129,
      "grad_norm": 0.7265625,
      "learning_rate": 6.378008202491762e-06,
      "loss": 0.764,
      "step": 1662
    },
    {
      "epoch": 1.0729032258064517,
      "grad_norm": 0.55859375,
      "learning_rate": 6.373989726697513e-06,
      "loss": 0.5027,
      "step": 1663
    },
    {
      "epoch": 1.0735483870967741,
      "grad_norm": 1.0625,
      "learning_rate": 6.3699702906856795e-06,
      "loss": 1.0581,
      "step": 1664
    },
    {
      "epoch": 1.0741935483870968,
      "grad_norm": 0.86328125,
      "learning_rate": 6.365949897265263e-06,
      "loss": 0.7987,
      "step": 1665
    },
    {
      "epoch": 1.0748387096774195,
      "grad_norm": 0.86328125,
      "learning_rate": 6.361928549245926e-06,
      "loss": 0.7104,
      "step": 1666
    },
    {
      "epoch": 1.075483870967742,
      "grad_norm": 0.6484375,
      "learning_rate": 6.357906249438004e-06,
      "loss": 0.5627,
      "step": 1667
    },
    {
      "epoch": 1.0761290322580646,
      "grad_norm": 1.453125,
      "learning_rate": 6.353883000652492e-06,
      "loss": 1.1591,
      "step": 1668
    },
    {
      "epoch": 1.076774193548387,
      "grad_norm": 1.0859375,
      "learning_rate": 6.3498588057010565e-06,
      "loss": 0.8734,
      "step": 1669
    },
    {
      "epoch": 1.0774193548387097,
      "grad_norm": 0.96484375,
      "learning_rate": 6.345833667396017e-06,
      "loss": 0.9187,
      "step": 1670
    },
    {
      "epoch": 1.0780645161290323,
      "grad_norm": 1.3203125,
      "learning_rate": 6.341807588550358e-06,
      "loss": 1.0036,
      "step": 1671
    },
    {
      "epoch": 1.0787096774193548,
      "grad_norm": 1.1484375,
      "learning_rate": 6.337780571977716e-06,
      "loss": 0.9098,
      "step": 1672
    },
    {
      "epoch": 1.0793548387096774,
      "grad_norm": 1.140625,
      "learning_rate": 6.333752620492389e-06,
      "loss": 0.8737,
      "step": 1673
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.92578125,
      "learning_rate": 6.3297237369093215e-06,
      "loss": 0.7378,
      "step": 1674
    },
    {
      "epoch": 1.0806451612903225,
      "grad_norm": 0.9296875,
      "learning_rate": 6.325693924044114e-06,
      "loss": 0.7562,
      "step": 1675
    },
    {
      "epoch": 1.0812903225806452,
      "grad_norm": 1.0703125,
      "learning_rate": 6.321663184713018e-06,
      "loss": 0.9581,
      "step": 1676
    },
    {
      "epoch": 1.0819354838709678,
      "grad_norm": 1.5,
      "learning_rate": 6.317631521732929e-06,
      "loss": 0.9337,
      "step": 1677
    },
    {
      "epoch": 1.0825806451612903,
      "grad_norm": 0.89453125,
      "learning_rate": 6.313598937921388e-06,
      "loss": 0.7326,
      "step": 1678
    },
    {
      "epoch": 1.083225806451613,
      "grad_norm": 0.6953125,
      "learning_rate": 6.3095654360965805e-06,
      "loss": 0.6488,
      "step": 1679
    },
    {
      "epoch": 1.0838709677419356,
      "grad_norm": 1.046875,
      "learning_rate": 6.305531019077336e-06,
      "loss": 1.0725,
      "step": 1680
    },
    {
      "epoch": 1.084516129032258,
      "grad_norm": 1.203125,
      "learning_rate": 6.301495689683117e-06,
      "loss": 1.0149,
      "step": 1681
    },
    {
      "epoch": 1.0851612903225807,
      "grad_norm": 0.66796875,
      "learning_rate": 6.297459450734031e-06,
      "loss": 0.4307,
      "step": 1682
    },
    {
      "epoch": 1.0858064516129031,
      "grad_norm": 0.7421875,
      "learning_rate": 6.293422305050814e-06,
      "loss": 0.6339,
      "step": 1683
    },
    {
      "epoch": 1.0864516129032258,
      "grad_norm": 0.91015625,
      "learning_rate": 6.2893842554548444e-06,
      "loss": 0.7021,
      "step": 1684
    },
    {
      "epoch": 1.0870967741935484,
      "grad_norm": 1.0703125,
      "learning_rate": 6.285345304768123e-06,
      "loss": 0.8515,
      "step": 1685
    },
    {
      "epoch": 1.0877419354838709,
      "grad_norm": 0.92578125,
      "learning_rate": 6.281305455813289e-06,
      "loss": 0.8245,
      "step": 1686
    },
    {
      "epoch": 1.0883870967741935,
      "grad_norm": 0.89453125,
      "learning_rate": 6.277264711413602e-06,
      "loss": 0.8589,
      "step": 1687
    },
    {
      "epoch": 1.0890322580645162,
      "grad_norm": 0.60546875,
      "learning_rate": 6.273223074392952e-06,
      "loss": 0.6741,
      "step": 1688
    },
    {
      "epoch": 1.0896774193548386,
      "grad_norm": 0.87109375,
      "learning_rate": 6.269180547575853e-06,
      "loss": 0.9298,
      "step": 1689
    },
    {
      "epoch": 1.0903225806451613,
      "grad_norm": 0.76953125,
      "learning_rate": 6.265137133787434e-06,
      "loss": 0.7035,
      "step": 1690
    },
    {
      "epoch": 1.090967741935484,
      "grad_norm": 0.70703125,
      "learning_rate": 6.261092835853454e-06,
      "loss": 0.4814,
      "step": 1691
    },
    {
      "epoch": 1.0916129032258064,
      "grad_norm": 0.9453125,
      "learning_rate": 6.2570476566002826e-06,
      "loss": 0.9858,
      "step": 1692
    },
    {
      "epoch": 1.092258064516129,
      "grad_norm": 0.5078125,
      "learning_rate": 6.253001598854909e-06,
      "loss": 0.4648,
      "step": 1693
    },
    {
      "epoch": 1.0929032258064517,
      "grad_norm": 0.9140625,
      "learning_rate": 6.248954665444932e-06,
      "loss": 0.6982,
      "step": 1694
    },
    {
      "epoch": 1.0935483870967742,
      "grad_norm": 1.453125,
      "learning_rate": 6.244906859198569e-06,
      "loss": 1.2677,
      "step": 1695
    },
    {
      "epoch": 1.0941935483870968,
      "grad_norm": 0.7265625,
      "learning_rate": 6.240858182944642e-06,
      "loss": 0.686,
      "step": 1696
    },
    {
      "epoch": 1.0948387096774193,
      "grad_norm": 0.921875,
      "learning_rate": 6.236808639512582e-06,
      "loss": 0.942,
      "step": 1697
    },
    {
      "epoch": 1.095483870967742,
      "grad_norm": 0.66796875,
      "learning_rate": 6.232758231732427e-06,
      "loss": 0.584,
      "step": 1698
    },
    {
      "epoch": 1.0961290322580646,
      "grad_norm": 0.83984375,
      "learning_rate": 6.228706962434822e-06,
      "loss": 0.7769,
      "step": 1699
    },
    {
      "epoch": 1.096774193548387,
      "grad_norm": 0.98046875,
      "learning_rate": 6.224654834451007e-06,
      "loss": 0.972,
      "step": 1700
    },
    {
      "epoch": 1.0974193548387097,
      "grad_norm": 0.91796875,
      "learning_rate": 6.220601850612827e-06,
      "loss": 0.6274,
      "step": 1701
    },
    {
      "epoch": 1.0980645161290323,
      "grad_norm": 0.84375,
      "learning_rate": 6.216548013752725e-06,
      "loss": 0.7409,
      "step": 1702
    },
    {
      "epoch": 1.0987096774193548,
      "grad_norm": 0.76953125,
      "learning_rate": 6.212493326703738e-06,
      "loss": 0.7962,
      "step": 1703
    },
    {
      "epoch": 1.0993548387096774,
      "grad_norm": 0.96484375,
      "learning_rate": 6.2084377922994994e-06,
      "loss": 0.9208,
      "step": 1704
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.3359375,
      "learning_rate": 6.204381413374233e-06,
      "loss": 0.9382,
      "step": 1705
    },
    {
      "epoch": 1.1006451612903225,
      "grad_norm": 0.87109375,
      "learning_rate": 6.200324192762756e-06,
      "loss": 0.7279,
      "step": 1706
    },
    {
      "epoch": 1.1012903225806452,
      "grad_norm": 0.796875,
      "learning_rate": 6.196266133300466e-06,
      "loss": 0.6861,
      "step": 1707
    },
    {
      "epoch": 1.1019354838709678,
      "grad_norm": 1.1796875,
      "learning_rate": 6.192207237823358e-06,
      "loss": 0.9927,
      "step": 1708
    },
    {
      "epoch": 1.1025806451612903,
      "grad_norm": 0.75390625,
      "learning_rate": 6.1881475091680034e-06,
      "loss": 0.6841,
      "step": 1709
    },
    {
      "epoch": 1.103225806451613,
      "grad_norm": 1.1796875,
      "learning_rate": 6.184086950171555e-06,
      "loss": 0.9587,
      "step": 1710
    },
    {
      "epoch": 1.1038709677419354,
      "grad_norm": 1.0390625,
      "learning_rate": 6.1800255636717545e-06,
      "loss": 0.8106,
      "step": 1711
    },
    {
      "epoch": 1.104516129032258,
      "grad_norm": 0.95703125,
      "learning_rate": 6.175963352506911e-06,
      "loss": 0.7515,
      "step": 1712
    },
    {
      "epoch": 1.1051612903225807,
      "grad_norm": 0.6171875,
      "learning_rate": 6.171900319515918e-06,
      "loss": 0.643,
      "step": 1713
    },
    {
      "epoch": 1.1058064516129031,
      "grad_norm": 1.15625,
      "learning_rate": 6.1678364675382385e-06,
      "loss": 0.9063,
      "step": 1714
    },
    {
      "epoch": 1.1064516129032258,
      "grad_norm": 0.80859375,
      "learning_rate": 6.163771799413912e-06,
      "loss": 0.7863,
      "step": 1715
    },
    {
      "epoch": 1.1070967741935485,
      "grad_norm": 0.78515625,
      "learning_rate": 6.159706317983546e-06,
      "loss": 0.8621,
      "step": 1716
    },
    {
      "epoch": 1.107741935483871,
      "grad_norm": 0.63671875,
      "learning_rate": 6.155640026088315e-06,
      "loss": 0.605,
      "step": 1717
    },
    {
      "epoch": 1.1083870967741936,
      "grad_norm": 0.76953125,
      "learning_rate": 6.151572926569963e-06,
      "loss": 0.6757,
      "step": 1718
    },
    {
      "epoch": 1.1090322580645162,
      "grad_norm": 0.87890625,
      "learning_rate": 6.147505022270795e-06,
      "loss": 0.7932,
      "step": 1719
    },
    {
      "epoch": 1.1096774193548387,
      "grad_norm": 1.1484375,
      "learning_rate": 6.143436316033684e-06,
      "loss": 0.711,
      "step": 1720
    },
    {
      "epoch": 1.1103225806451613,
      "grad_norm": 1.265625,
      "learning_rate": 6.139366810702055e-06,
      "loss": 1.0219,
      "step": 1721
    },
    {
      "epoch": 1.1109677419354838,
      "grad_norm": 1.171875,
      "learning_rate": 6.1352965091199e-06,
      "loss": 0.9407,
      "step": 1722
    },
    {
      "epoch": 1.1116129032258064,
      "grad_norm": 0.80078125,
      "learning_rate": 6.131225414131761e-06,
      "loss": 0.7715,
      "step": 1723
    },
    {
      "epoch": 1.112258064516129,
      "grad_norm": 1.046875,
      "learning_rate": 6.127153528582739e-06,
      "loss": 0.9999,
      "step": 1724
    },
    {
      "epoch": 1.1129032258064515,
      "grad_norm": 0.84765625,
      "learning_rate": 6.123080855318484e-06,
      "loss": 0.9778,
      "step": 1725
    },
    {
      "epoch": 1.1135483870967742,
      "grad_norm": 1.0703125,
      "learning_rate": 6.1190073971852e-06,
      "loss": 1.0706,
      "step": 1726
    },
    {
      "epoch": 1.1141935483870968,
      "grad_norm": 1.3671875,
      "learning_rate": 6.114933157029636e-06,
      "loss": 0.9378,
      "step": 1727
    },
    {
      "epoch": 1.1148387096774193,
      "grad_norm": 0.92578125,
      "learning_rate": 6.1108581376990905e-06,
      "loss": 0.8829,
      "step": 1728
    },
    {
      "epoch": 1.115483870967742,
      "grad_norm": 1.1015625,
      "learning_rate": 6.106782342041405e-06,
      "loss": 0.7569,
      "step": 1729
    },
    {
      "epoch": 1.1161290322580646,
      "grad_norm": 1.25,
      "learning_rate": 6.102705772904962e-06,
      "loss": 0.8359,
      "step": 1730
    },
    {
      "epoch": 1.116774193548387,
      "grad_norm": 1.0859375,
      "learning_rate": 6.098628433138689e-06,
      "loss": 0.8685,
      "step": 1731
    },
    {
      "epoch": 1.1174193548387097,
      "grad_norm": 1.0234375,
      "learning_rate": 6.094550325592046e-06,
      "loss": 0.9217,
      "step": 1732
    },
    {
      "epoch": 1.1180645161290323,
      "grad_norm": 1.125,
      "learning_rate": 6.0904714531150365e-06,
      "loss": 1.0258,
      "step": 1733
    },
    {
      "epoch": 1.1187096774193548,
      "grad_norm": 0.89453125,
      "learning_rate": 6.086391818558192e-06,
      "loss": 0.6971,
      "step": 1734
    },
    {
      "epoch": 1.1193548387096774,
      "grad_norm": 0.86328125,
      "learning_rate": 6.082311424772581e-06,
      "loss": 0.6777,
      "step": 1735
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8671875,
      "learning_rate": 6.078230274609802e-06,
      "loss": 0.7822,
      "step": 1736
    },
    {
      "epoch": 1.1206451612903225,
      "grad_norm": 0.69140625,
      "learning_rate": 6.074148370921978e-06,
      "loss": 0.6575,
      "step": 1737
    },
    {
      "epoch": 1.1212903225806452,
      "grad_norm": 0.890625,
      "learning_rate": 6.070065716561766e-06,
      "loss": 0.7337,
      "step": 1738
    },
    {
      "epoch": 1.1219354838709676,
      "grad_norm": 1.09375,
      "learning_rate": 6.06598231438234e-06,
      "loss": 0.9291,
      "step": 1739
    },
    {
      "epoch": 1.1225806451612903,
      "grad_norm": 0.9609375,
      "learning_rate": 6.0618981672374015e-06,
      "loss": 0.8303,
      "step": 1740
    },
    {
      "epoch": 1.123225806451613,
      "grad_norm": 0.91796875,
      "learning_rate": 6.05781327798117e-06,
      "loss": 0.7389,
      "step": 1741
    },
    {
      "epoch": 1.1238709677419354,
      "grad_norm": 1.234375,
      "learning_rate": 6.053727649468388e-06,
      "loss": 1.0521,
      "step": 1742
    },
    {
      "epoch": 1.124516129032258,
      "grad_norm": 0.7734375,
      "learning_rate": 6.049641284554309e-06,
      "loss": 0.5185,
      "step": 1743
    },
    {
      "epoch": 1.1251612903225807,
      "grad_norm": 0.74609375,
      "learning_rate": 6.045554186094704e-06,
      "loss": 0.6872,
      "step": 1744
    },
    {
      "epoch": 1.1258064516129032,
      "grad_norm": 0.7109375,
      "learning_rate": 6.041466356945856e-06,
      "loss": 0.8228,
      "step": 1745
    },
    {
      "epoch": 1.1264516129032258,
      "grad_norm": 1.125,
      "learning_rate": 6.037377799964559e-06,
      "loss": 1.0534,
      "step": 1746
    },
    {
      "epoch": 1.1270967741935485,
      "grad_norm": 0.7421875,
      "learning_rate": 6.033288518008116e-06,
      "loss": 0.6103,
      "step": 1747
    },
    {
      "epoch": 1.127741935483871,
      "grad_norm": 1.28125,
      "learning_rate": 6.029198513934335e-06,
      "loss": 1.0988,
      "step": 1748
    },
    {
      "epoch": 1.1283870967741936,
      "grad_norm": 0.63671875,
      "learning_rate": 6.025107790601531e-06,
      "loss": 0.5299,
      "step": 1749
    },
    {
      "epoch": 1.129032258064516,
      "grad_norm": 0.80078125,
      "learning_rate": 6.0210163508685195e-06,
      "loss": 0.5629,
      "step": 1750
    },
    {
      "epoch": 1.1296774193548387,
      "grad_norm": 0.734375,
      "learning_rate": 6.016924197594619e-06,
      "loss": 0.6952,
      "step": 1751
    },
    {
      "epoch": 1.1303225806451613,
      "grad_norm": 0.85546875,
      "learning_rate": 6.012831333639644e-06,
      "loss": 0.8049,
      "step": 1752
    },
    {
      "epoch": 1.1309677419354838,
      "grad_norm": 0.59765625,
      "learning_rate": 6.008737761863907e-06,
      "loss": 0.3948,
      "step": 1753
    },
    {
      "epoch": 1.1316129032258064,
      "grad_norm": 1.21875,
      "learning_rate": 6.004643485128216e-06,
      "loss": 0.9332,
      "step": 1754
    },
    {
      "epoch": 1.132258064516129,
      "grad_norm": 1.0390625,
      "learning_rate": 6.00054850629387e-06,
      "loss": 0.7541,
      "step": 1755
    },
    {
      "epoch": 1.1329032258064515,
      "grad_norm": 0.66015625,
      "learning_rate": 5.996452828222661e-06,
      "loss": 0.575,
      "step": 1756
    },
    {
      "epoch": 1.1335483870967742,
      "grad_norm": 0.7890625,
      "learning_rate": 5.992356453776866e-06,
      "loss": 0.6959,
      "step": 1757
    },
    {
      "epoch": 1.1341935483870969,
      "grad_norm": 0.75,
      "learning_rate": 5.988259385819251e-06,
      "loss": 0.6951,
      "step": 1758
    },
    {
      "epoch": 1.1348387096774193,
      "grad_norm": 0.890625,
      "learning_rate": 5.984161627213068e-06,
      "loss": 0.7987,
      "step": 1759
    },
    {
      "epoch": 1.135483870967742,
      "grad_norm": 0.8984375,
      "learning_rate": 5.98006318082205e-06,
      "loss": 0.5661,
      "step": 1760
    },
    {
      "epoch": 1.1361290322580646,
      "grad_norm": 0.734375,
      "learning_rate": 5.975964049510408e-06,
      "loss": 0.581,
      "step": 1761
    },
    {
      "epoch": 1.136774193548387,
      "grad_norm": 0.82421875,
      "learning_rate": 5.971864236142838e-06,
      "loss": 0.7798,
      "step": 1762
    },
    {
      "epoch": 1.1374193548387097,
      "grad_norm": 0.93359375,
      "learning_rate": 5.967763743584507e-06,
      "loss": 0.7353,
      "step": 1763
    },
    {
      "epoch": 1.1380645161290324,
      "grad_norm": 1.046875,
      "learning_rate": 5.963662574701059e-06,
      "loss": 0.872,
      "step": 1764
    },
    {
      "epoch": 1.1387096774193548,
      "grad_norm": 1.015625,
      "learning_rate": 5.959560732358612e-06,
      "loss": 0.8733,
      "step": 1765
    },
    {
      "epoch": 1.1393548387096775,
      "grad_norm": 0.91015625,
      "learning_rate": 5.955458219423749e-06,
      "loss": 0.6351,
      "step": 1766
    },
    {
      "epoch": 1.1400000000000001,
      "grad_norm": 0.83203125,
      "learning_rate": 5.95135503876353e-06,
      "loss": 0.7051,
      "step": 1767
    },
    {
      "epoch": 1.1406451612903226,
      "grad_norm": 1.2265625,
      "learning_rate": 5.9472511932454745e-06,
      "loss": 1.1174,
      "step": 1768
    },
    {
      "epoch": 1.1412903225806452,
      "grad_norm": 0.98828125,
      "learning_rate": 5.943146685737572e-06,
      "loss": 0.9679,
      "step": 1769
    },
    {
      "epoch": 1.1419354838709677,
      "grad_norm": 1.3515625,
      "learning_rate": 5.939041519108268e-06,
      "loss": 0.9194,
      "step": 1770
    },
    {
      "epoch": 1.1425806451612903,
      "grad_norm": 0.65625,
      "learning_rate": 5.934935696226479e-06,
      "loss": 0.6097,
      "step": 1771
    },
    {
      "epoch": 1.143225806451613,
      "grad_norm": 0.84375,
      "learning_rate": 5.930829219961569e-06,
      "loss": 0.7892,
      "step": 1772
    },
    {
      "epoch": 1.1438709677419354,
      "grad_norm": 0.97265625,
      "learning_rate": 5.926722093183366e-06,
      "loss": 0.8794,
      "step": 1773
    },
    {
      "epoch": 1.144516129032258,
      "grad_norm": 0.8125,
      "learning_rate": 5.922614318762145e-06,
      "loss": 0.7108,
      "step": 1774
    },
    {
      "epoch": 1.1451612903225807,
      "grad_norm": 0.95703125,
      "learning_rate": 5.918505899568645e-06,
      "loss": 0.7997,
      "step": 1775
    },
    {
      "epoch": 1.1458064516129032,
      "grad_norm": 0.79296875,
      "learning_rate": 5.9143968384740465e-06,
      "loss": 0.5968,
      "step": 1776
    },
    {
      "epoch": 1.1464516129032258,
      "grad_norm": 1.140625,
      "learning_rate": 5.910287138349982e-06,
      "loss": 1.0462,
      "step": 1777
    },
    {
      "epoch": 1.1470967741935483,
      "grad_norm": 1.421875,
      "learning_rate": 5.906176802068531e-06,
      "loss": 1.1737,
      "step": 1778
    },
    {
      "epoch": 1.147741935483871,
      "grad_norm": 1.0546875,
      "learning_rate": 5.902065832502212e-06,
      "loss": 0.7379,
      "step": 1779
    },
    {
      "epoch": 1.1483870967741936,
      "grad_norm": 1.1796875,
      "learning_rate": 5.897954232523997e-06,
      "loss": 0.8638,
      "step": 1780
    },
    {
      "epoch": 1.149032258064516,
      "grad_norm": 0.62109375,
      "learning_rate": 5.893842005007288e-06,
      "loss": 0.6171,
      "step": 1781
    },
    {
      "epoch": 1.1496774193548387,
      "grad_norm": 1.0546875,
      "learning_rate": 5.88972915282593e-06,
      "loss": 0.796,
      "step": 1782
    },
    {
      "epoch": 1.1503225806451614,
      "grad_norm": 0.66796875,
      "learning_rate": 5.885615678854205e-06,
      "loss": 0.5778,
      "step": 1783
    },
    {
      "epoch": 1.1509677419354838,
      "grad_norm": 0.86328125,
      "learning_rate": 5.88150158596683e-06,
      "loss": 0.7173,
      "step": 1784
    },
    {
      "epoch": 1.1516129032258065,
      "grad_norm": 0.7421875,
      "learning_rate": 5.877386877038952e-06,
      "loss": 0.7049,
      "step": 1785
    },
    {
      "epoch": 1.152258064516129,
      "grad_norm": 1.2421875,
      "learning_rate": 5.8732715549461495e-06,
      "loss": 0.9326,
      "step": 1786
    },
    {
      "epoch": 1.1529032258064515,
      "grad_norm": 1.171875,
      "learning_rate": 5.869155622564431e-06,
      "loss": 0.9904,
      "step": 1787
    },
    {
      "epoch": 1.1535483870967742,
      "grad_norm": 0.7734375,
      "learning_rate": 5.86503908277023e-06,
      "loss": 0.5726,
      "step": 1788
    },
    {
      "epoch": 1.1541935483870969,
      "grad_norm": 0.81640625,
      "learning_rate": 5.8609219384404044e-06,
      "loss": 0.711,
      "step": 1789
    },
    {
      "epoch": 1.1548387096774193,
      "grad_norm": 1.0,
      "learning_rate": 5.856804192452238e-06,
      "loss": 0.8438,
      "step": 1790
    },
    {
      "epoch": 1.155483870967742,
      "grad_norm": 1.109375,
      "learning_rate": 5.8526858476834284e-06,
      "loss": 0.7916,
      "step": 1791
    },
    {
      "epoch": 1.1561290322580646,
      "grad_norm": 0.83984375,
      "learning_rate": 5.848566907012095e-06,
      "loss": 0.8882,
      "step": 1792
    },
    {
      "epoch": 1.156774193548387,
      "grad_norm": 0.984375,
      "learning_rate": 5.8444473733167786e-06,
      "loss": 1.0849,
      "step": 1793
    },
    {
      "epoch": 1.1574193548387097,
      "grad_norm": 0.8125,
      "learning_rate": 5.840327249476425e-06,
      "loss": 0.6824,
      "step": 1794
    },
    {
      "epoch": 1.1580645161290322,
      "grad_norm": 0.7734375,
      "learning_rate": 5.8362065383704015e-06,
      "loss": 0.7085,
      "step": 1795
    },
    {
      "epoch": 1.1587096774193548,
      "grad_norm": 0.734375,
      "learning_rate": 5.8320852428784766e-06,
      "loss": 0.6693,
      "step": 1796
    },
    {
      "epoch": 1.1593548387096775,
      "grad_norm": 0.734375,
      "learning_rate": 5.827963365880838e-06,
      "loss": 0.7068,
      "step": 1797
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.99609375,
      "learning_rate": 5.823840910258069e-06,
      "loss": 0.7632,
      "step": 1798
    },
    {
      "epoch": 1.1606451612903226,
      "grad_norm": 1.2578125,
      "learning_rate": 5.819717878891168e-06,
      "loss": 1.1998,
      "step": 1799
    },
    {
      "epoch": 1.1612903225806452,
      "grad_norm": 0.90625,
      "learning_rate": 5.815594274661524e-06,
      "loss": 0.9865,
      "step": 1800
    },
    {
      "epoch": 1.1619354838709677,
      "grad_norm": 0.82421875,
      "learning_rate": 5.8114701004509355e-06,
      "loss": 0.7699,
      "step": 1801
    },
    {
      "epoch": 1.1625806451612903,
      "grad_norm": 0.73046875,
      "learning_rate": 5.807345359141596e-06,
      "loss": 0.647,
      "step": 1802
    },
    {
      "epoch": 1.163225806451613,
      "grad_norm": 1.0234375,
      "learning_rate": 5.803220053616096e-06,
      "loss": 0.9701,
      "step": 1803
    },
    {
      "epoch": 1.1638709677419354,
      "grad_norm": 0.65234375,
      "learning_rate": 5.799094186757418e-06,
      "loss": 0.5922,
      "step": 1804
    },
    {
      "epoch": 1.164516129032258,
      "grad_norm": 0.640625,
      "learning_rate": 5.79496776144894e-06,
      "loss": 0.4689,
      "step": 1805
    },
    {
      "epoch": 1.1651612903225805,
      "grad_norm": 1.125,
      "learning_rate": 5.79084078057443e-06,
      "loss": 0.9831,
      "step": 1806
    },
    {
      "epoch": 1.1658064516129032,
      "grad_norm": 0.77734375,
      "learning_rate": 5.7867132470180396e-06,
      "loss": 0.6311,
      "step": 1807
    },
    {
      "epoch": 1.1664516129032259,
      "grad_norm": 0.69140625,
      "learning_rate": 5.7825851636643135e-06,
      "loss": 0.5736,
      "step": 1808
    },
    {
      "epoch": 1.1670967741935483,
      "grad_norm": 1.09375,
      "learning_rate": 5.778456533398179e-06,
      "loss": 0.8699,
      "step": 1809
    },
    {
      "epoch": 1.167741935483871,
      "grad_norm": 0.73046875,
      "learning_rate": 5.774327359104939e-06,
      "loss": 0.6434,
      "step": 1810
    },
    {
      "epoch": 1.1683870967741936,
      "grad_norm": 0.671875,
      "learning_rate": 5.770197643670285e-06,
      "loss": 0.4946,
      "step": 1811
    },
    {
      "epoch": 1.169032258064516,
      "grad_norm": 0.81640625,
      "learning_rate": 5.766067389980281e-06,
      "loss": 0.6934,
      "step": 1812
    },
    {
      "epoch": 1.1696774193548387,
      "grad_norm": 0.68359375,
      "learning_rate": 5.7619366009213716e-06,
      "loss": 0.6455,
      "step": 1813
    },
    {
      "epoch": 1.1703225806451614,
      "grad_norm": 0.91015625,
      "learning_rate": 5.757805279380373e-06,
      "loss": 0.7482,
      "step": 1814
    },
    {
      "epoch": 1.1709677419354838,
      "grad_norm": 0.921875,
      "learning_rate": 5.753673428244473e-06,
      "loss": 0.7231,
      "step": 1815
    },
    {
      "epoch": 1.1716129032258065,
      "grad_norm": 0.7421875,
      "learning_rate": 5.7495410504012296e-06,
      "loss": 0.6141,
      "step": 1816
    },
    {
      "epoch": 1.1722580645161291,
      "grad_norm": 0.9609375,
      "learning_rate": 5.745408148738572e-06,
      "loss": 0.7515,
      "step": 1817
    },
    {
      "epoch": 1.1729032258064516,
      "grad_norm": 0.76171875,
      "learning_rate": 5.741274726144791e-06,
      "loss": 0.7493,
      "step": 1818
    },
    {
      "epoch": 1.1735483870967742,
      "grad_norm": 1.1796875,
      "learning_rate": 5.737140785508545e-06,
      "loss": 0.931,
      "step": 1819
    },
    {
      "epoch": 1.1741935483870969,
      "grad_norm": 0.875,
      "learning_rate": 5.733006329718852e-06,
      "loss": 0.8382,
      "step": 1820
    },
    {
      "epoch": 1.1748387096774193,
      "grad_norm": 0.95703125,
      "learning_rate": 5.728871361665091e-06,
      "loss": 0.7875,
      "step": 1821
    },
    {
      "epoch": 1.175483870967742,
      "grad_norm": 1.1796875,
      "learning_rate": 5.724735884237001e-06,
      "loss": 1.2244,
      "step": 1822
    },
    {
      "epoch": 1.1761290322580644,
      "grad_norm": 1.1484375,
      "learning_rate": 5.720599900324674e-06,
      "loss": 0.7847,
      "step": 1823
    },
    {
      "epoch": 1.176774193548387,
      "grad_norm": 0.75390625,
      "learning_rate": 5.716463412818556e-06,
      "loss": 0.6525,
      "step": 1824
    },
    {
      "epoch": 1.1774193548387097,
      "grad_norm": 0.61328125,
      "learning_rate": 5.712326424609447e-06,
      "loss": 0.6446,
      "step": 1825
    },
    {
      "epoch": 1.1780645161290322,
      "grad_norm": 0.78515625,
      "learning_rate": 5.708188938588495e-06,
      "loss": 0.6885,
      "step": 1826
    },
    {
      "epoch": 1.1787096774193548,
      "grad_norm": 0.77734375,
      "learning_rate": 5.704050957647198e-06,
      "loss": 0.5752,
      "step": 1827
    },
    {
      "epoch": 1.1793548387096775,
      "grad_norm": 1.171875,
      "learning_rate": 5.699912484677397e-06,
      "loss": 0.831,
      "step": 1828
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.6484375,
      "learning_rate": 5.69577352257128e-06,
      "loss": 0.5845,
      "step": 1829
    },
    {
      "epoch": 1.1806451612903226,
      "grad_norm": 1.1484375,
      "learning_rate": 5.691634074221374e-06,
      "loss": 1.0234,
      "step": 1830
    },
    {
      "epoch": 1.1812903225806453,
      "grad_norm": 0.84375,
      "learning_rate": 5.687494142520549e-06,
      "loss": 0.5759,
      "step": 1831
    },
    {
      "epoch": 1.1819354838709677,
      "grad_norm": 0.953125,
      "learning_rate": 5.683353730362007e-06,
      "loss": 0.8477,
      "step": 1832
    },
    {
      "epoch": 1.1825806451612904,
      "grad_norm": 0.8671875,
      "learning_rate": 5.679212840639295e-06,
      "loss": 0.8313,
      "step": 1833
    },
    {
      "epoch": 1.1832258064516128,
      "grad_norm": 1.09375,
      "learning_rate": 5.6750714762462845e-06,
      "loss": 0.9058,
      "step": 1834
    },
    {
      "epoch": 1.1838709677419355,
      "grad_norm": 1.0078125,
      "learning_rate": 5.670929640077185e-06,
      "loss": 0.7839,
      "step": 1835
    },
    {
      "epoch": 1.1845161290322581,
      "grad_norm": 1.2265625,
      "learning_rate": 5.6667873350265325e-06,
      "loss": 0.9768,
      "step": 1836
    },
    {
      "epoch": 1.1851612903225806,
      "grad_norm": 0.828125,
      "learning_rate": 5.662644563989189e-06,
      "loss": 0.7526,
      "step": 1837
    },
    {
      "epoch": 1.1858064516129032,
      "grad_norm": 0.86328125,
      "learning_rate": 5.658501329860349e-06,
      "loss": 0.7088,
      "step": 1838
    },
    {
      "epoch": 1.1864516129032259,
      "grad_norm": 0.85546875,
      "learning_rate": 5.654357635535525e-06,
      "loss": 0.7237,
      "step": 1839
    },
    {
      "epoch": 1.1870967741935483,
      "grad_norm": 1.140625,
      "learning_rate": 5.650213483910551e-06,
      "loss": 0.9502,
      "step": 1840
    },
    {
      "epoch": 1.187741935483871,
      "grad_norm": 1.2265625,
      "learning_rate": 5.646068877881582e-06,
      "loss": 1.009,
      "step": 1841
    },
    {
      "epoch": 1.1883870967741936,
      "grad_norm": 1.109375,
      "learning_rate": 5.641923820345093e-06,
      "loss": 1.195,
      "step": 1842
    },
    {
      "epoch": 1.189032258064516,
      "grad_norm": 0.703125,
      "learning_rate": 5.637778314197869e-06,
      "loss": 0.6004,
      "step": 1843
    },
    {
      "epoch": 1.1896774193548387,
      "grad_norm": 0.83203125,
      "learning_rate": 5.633632362337015e-06,
      "loss": 0.677,
      "step": 1844
    },
    {
      "epoch": 1.1903225806451614,
      "grad_norm": 0.84765625,
      "learning_rate": 5.6294859676599425e-06,
      "loss": 0.732,
      "step": 1845
    },
    {
      "epoch": 1.1909677419354838,
      "grad_norm": 0.77734375,
      "learning_rate": 5.6253391330643725e-06,
      "loss": 0.6822,
      "step": 1846
    },
    {
      "epoch": 1.1916129032258065,
      "grad_norm": 1.0234375,
      "learning_rate": 5.621191861448338e-06,
      "loss": 1.0931,
      "step": 1847
    },
    {
      "epoch": 1.1922580645161291,
      "grad_norm": 0.9765625,
      "learning_rate": 5.617044155710171e-06,
      "loss": 0.7118,
      "step": 1848
    },
    {
      "epoch": 1.1929032258064516,
      "grad_norm": 0.8359375,
      "learning_rate": 5.6128960187485135e-06,
      "loss": 0.8838,
      "step": 1849
    },
    {
      "epoch": 1.1935483870967742,
      "grad_norm": 1.2578125,
      "learning_rate": 5.608747453462303e-06,
      "loss": 0.9572,
      "step": 1850
    },
    {
      "epoch": 1.1941935483870967,
      "grad_norm": 0.9296875,
      "learning_rate": 5.604598462750783e-06,
      "loss": 0.9194,
      "step": 1851
    },
    {
      "epoch": 1.1948387096774193,
      "grad_norm": 1.3125,
      "learning_rate": 5.600449049513486e-06,
      "loss": 1.2097,
      "step": 1852
    },
    {
      "epoch": 1.195483870967742,
      "grad_norm": 0.82421875,
      "learning_rate": 5.5962992166502475e-06,
      "loss": 0.5504,
      "step": 1853
    },
    {
      "epoch": 1.1961290322580644,
      "grad_norm": 0.92578125,
      "learning_rate": 5.592148967061191e-06,
      "loss": 0.7879,
      "step": 1854
    },
    {
      "epoch": 1.196774193548387,
      "grad_norm": 0.87890625,
      "learning_rate": 5.587998303646731e-06,
      "loss": 0.8203,
      "step": 1855
    },
    {
      "epoch": 1.1974193548387098,
      "grad_norm": 1.4453125,
      "learning_rate": 5.583847229307579e-06,
      "loss": 1.1056,
      "step": 1856
    },
    {
      "epoch": 1.1980645161290322,
      "grad_norm": 1.1328125,
      "learning_rate": 5.579695746944722e-06,
      "loss": 0.9911,
      "step": 1857
    },
    {
      "epoch": 1.1987096774193549,
      "grad_norm": 0.91015625,
      "learning_rate": 5.5755438594594436e-06,
      "loss": 0.6675,
      "step": 1858
    },
    {
      "epoch": 1.1993548387096773,
      "grad_norm": 0.671875,
      "learning_rate": 5.5713915697533e-06,
      "loss": 0.4881,
      "step": 1859
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7734375,
      "learning_rate": 5.567238880728138e-06,
      "loss": 0.6122,
      "step": 1860
    },
    {
      "epoch": 1.2006451612903226,
      "grad_norm": 0.82421875,
      "learning_rate": 5.563085795286075e-06,
      "loss": 0.4923,
      "step": 1861
    },
    {
      "epoch": 1.201290322580645,
      "grad_norm": 1.1328125,
      "learning_rate": 5.558932316329513e-06,
      "loss": 0.7835,
      "step": 1862
    },
    {
      "epoch": 1.2019354838709677,
      "grad_norm": 0.7890625,
      "learning_rate": 5.554778446761122e-06,
      "loss": 0.7008,
      "step": 1863
    },
    {
      "epoch": 1.2025806451612904,
      "grad_norm": 0.69921875,
      "learning_rate": 5.550624189483852e-06,
      "loss": 0.5834,
      "step": 1864
    },
    {
      "epoch": 1.2032258064516128,
      "grad_norm": 1.21875,
      "learning_rate": 5.546469547400919e-06,
      "loss": 1.0554,
      "step": 1865
    },
    {
      "epoch": 1.2038709677419355,
      "grad_norm": 0.66796875,
      "learning_rate": 5.542314523415808e-06,
      "loss": 0.4574,
      "step": 1866
    },
    {
      "epoch": 1.2045161290322581,
      "grad_norm": 1.34375,
      "learning_rate": 5.538159120432274e-06,
      "loss": 1.1015,
      "step": 1867
    },
    {
      "epoch": 1.2051612903225806,
      "grad_norm": 0.61328125,
      "learning_rate": 5.5340033413543325e-06,
      "loss": 0.5954,
      "step": 1868
    },
    {
      "epoch": 1.2058064516129032,
      "grad_norm": 0.83984375,
      "learning_rate": 5.5298471890862686e-06,
      "loss": 0.7492,
      "step": 1869
    },
    {
      "epoch": 1.206451612903226,
      "grad_norm": 1.1796875,
      "learning_rate": 5.52569066653262e-06,
      "loss": 1.0677,
      "step": 1870
    },
    {
      "epoch": 1.2070967741935483,
      "grad_norm": 1.140625,
      "learning_rate": 5.52153377659819e-06,
      "loss": 1.0068,
      "step": 1871
    },
    {
      "epoch": 1.207741935483871,
      "grad_norm": 1.015625,
      "learning_rate": 5.517376522188034e-06,
      "loss": 1.0477,
      "step": 1872
    },
    {
      "epoch": 1.2083870967741936,
      "grad_norm": 1.078125,
      "learning_rate": 5.513218906207466e-06,
      "loss": 1.0218,
      "step": 1873
    },
    {
      "epoch": 1.209032258064516,
      "grad_norm": 0.8359375,
      "learning_rate": 5.509060931562047e-06,
      "loss": 0.5684,
      "step": 1874
    },
    {
      "epoch": 1.2096774193548387,
      "grad_norm": 1.078125,
      "learning_rate": 5.504902601157596e-06,
      "loss": 1.1558,
      "step": 1875
    },
    {
      "epoch": 1.2103225806451614,
      "grad_norm": 1.0078125,
      "learning_rate": 5.5007439179001755e-06,
      "loss": 0.858,
      "step": 1876
    },
    {
      "epoch": 1.2109677419354838,
      "grad_norm": 0.77734375,
      "learning_rate": 5.496584884696095e-06,
      "loss": 0.6798,
      "step": 1877
    },
    {
      "epoch": 1.2116129032258065,
      "grad_norm": 0.66796875,
      "learning_rate": 5.492425504451912e-06,
      "loss": 0.5629,
      "step": 1878
    },
    {
      "epoch": 1.212258064516129,
      "grad_norm": 0.78125,
      "learning_rate": 5.488265780074421e-06,
      "loss": 0.756,
      "step": 1879
    },
    {
      "epoch": 1.2129032258064516,
      "grad_norm": 0.80859375,
      "learning_rate": 5.484105714470663e-06,
      "loss": 0.7716,
      "step": 1880
    },
    {
      "epoch": 1.2135483870967743,
      "grad_norm": 1.2890625,
      "learning_rate": 5.479945310547913e-06,
      "loss": 1.042,
      "step": 1881
    },
    {
      "epoch": 1.2141935483870967,
      "grad_norm": 1.03125,
      "learning_rate": 5.475784571213686e-06,
      "loss": 0.8148,
      "step": 1882
    },
    {
      "epoch": 1.2148387096774194,
      "grad_norm": 0.8984375,
      "learning_rate": 5.471623499375725e-06,
      "loss": 0.7102,
      "step": 1883
    },
    {
      "epoch": 1.215483870967742,
      "grad_norm": 0.79296875,
      "learning_rate": 5.467462097942015e-06,
      "loss": 0.653,
      "step": 1884
    },
    {
      "epoch": 1.2161290322580645,
      "grad_norm": 0.9921875,
      "learning_rate": 5.463300369820762e-06,
      "loss": 0.8423,
      "step": 1885
    },
    {
      "epoch": 1.2167741935483871,
      "grad_norm": 0.8515625,
      "learning_rate": 5.459138317920406e-06,
      "loss": 0.815,
      "step": 1886
    },
    {
      "epoch": 1.2174193548387096,
      "grad_norm": 1.046875,
      "learning_rate": 5.454975945149612e-06,
      "loss": 0.8241,
      "step": 1887
    },
    {
      "epoch": 1.2180645161290322,
      "grad_norm": 1.0859375,
      "learning_rate": 5.450813254417268e-06,
      "loss": 0.7619,
      "step": 1888
    },
    {
      "epoch": 1.2187096774193549,
      "grad_norm": 1.140625,
      "learning_rate": 5.446650248632488e-06,
      "loss": 0.9651,
      "step": 1889
    },
    {
      "epoch": 1.2193548387096773,
      "grad_norm": 0.96875,
      "learning_rate": 5.4424869307046e-06,
      "loss": 0.9255,
      "step": 1890
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.015625,
      "learning_rate": 5.438323303543156e-06,
      "loss": 0.7208,
      "step": 1891
    },
    {
      "epoch": 1.2206451612903226,
      "grad_norm": 0.9296875,
      "learning_rate": 5.434159370057919e-06,
      "loss": 0.5993,
      "step": 1892
    },
    {
      "epoch": 1.221290322580645,
      "grad_norm": 1.1484375,
      "learning_rate": 5.429995133158871e-06,
      "loss": 0.6458,
      "step": 1893
    },
    {
      "epoch": 1.2219354838709677,
      "grad_norm": 1.0546875,
      "learning_rate": 5.4258305957562004e-06,
      "loss": 0.9632,
      "step": 1894
    },
    {
      "epoch": 1.2225806451612904,
      "grad_norm": 0.78515625,
      "learning_rate": 5.421665760760313e-06,
      "loss": 0.7494,
      "step": 1895
    },
    {
      "epoch": 1.2232258064516128,
      "grad_norm": 0.86328125,
      "learning_rate": 5.4175006310818146e-06,
      "loss": 0.6996,
      "step": 1896
    },
    {
      "epoch": 1.2238709677419355,
      "grad_norm": 0.76171875,
      "learning_rate": 5.413335209631524e-06,
      "loss": 0.6401,
      "step": 1897
    },
    {
      "epoch": 1.2245161290322581,
      "grad_norm": 0.72265625,
      "learning_rate": 5.409169499320459e-06,
      "loss": 0.6033,
      "step": 1898
    },
    {
      "epoch": 1.2251612903225806,
      "grad_norm": 1.328125,
      "learning_rate": 5.4050035030598425e-06,
      "loss": 0.8814,
      "step": 1899
    },
    {
      "epoch": 1.2258064516129032,
      "grad_norm": 0.96484375,
      "learning_rate": 5.400837223761095e-06,
      "loss": 0.7554,
      "step": 1900
    },
    {
      "epoch": 1.226451612903226,
      "grad_norm": 1.140625,
      "learning_rate": 5.396670664335833e-06,
      "loss": 1.0853,
      "step": 1901
    },
    {
      "epoch": 1.2270967741935483,
      "grad_norm": 0.59375,
      "learning_rate": 5.392503827695877e-06,
      "loss": 0.5001,
      "step": 1902
    },
    {
      "epoch": 1.227741935483871,
      "grad_norm": 1.2578125,
      "learning_rate": 5.388336716753229e-06,
      "loss": 1.1427,
      "step": 1903
    },
    {
      "epoch": 1.2283870967741937,
      "grad_norm": 0.796875,
      "learning_rate": 5.3841693344200944e-06,
      "loss": 0.6737,
      "step": 1904
    },
    {
      "epoch": 1.229032258064516,
      "grad_norm": 0.875,
      "learning_rate": 5.380001683608859e-06,
      "loss": 0.7152,
      "step": 1905
    },
    {
      "epoch": 1.2296774193548388,
      "grad_norm": 0.95703125,
      "learning_rate": 5.375833767232104e-06,
      "loss": 0.6675,
      "step": 1906
    },
    {
      "epoch": 1.2303225806451612,
      "grad_norm": 0.85546875,
      "learning_rate": 5.371665588202589e-06,
      "loss": 0.6367,
      "step": 1907
    },
    {
      "epoch": 1.2309677419354839,
      "grad_norm": 0.90234375,
      "learning_rate": 5.367497149433263e-06,
      "loss": 0.8377,
      "step": 1908
    },
    {
      "epoch": 1.2316129032258065,
      "grad_norm": 0.796875,
      "learning_rate": 5.363328453837255e-06,
      "loss": 0.6779,
      "step": 1909
    },
    {
      "epoch": 1.232258064516129,
      "grad_norm": 0.63671875,
      "learning_rate": 5.359159504327871e-06,
      "loss": 0.5478,
      "step": 1910
    },
    {
      "epoch": 1.2329032258064516,
      "grad_norm": 0.921875,
      "learning_rate": 5.354990303818597e-06,
      "loss": 0.833,
      "step": 1911
    },
    {
      "epoch": 1.2335483870967743,
      "grad_norm": 1.140625,
      "learning_rate": 5.350820855223091e-06,
      "loss": 0.9632,
      "step": 1912
    },
    {
      "epoch": 1.2341935483870967,
      "grad_norm": 1.0,
      "learning_rate": 5.346651161455191e-06,
      "loss": 0.8499,
      "step": 1913
    },
    {
      "epoch": 1.2348387096774194,
      "grad_norm": 0.734375,
      "learning_rate": 5.342481225428899e-06,
      "loss": 0.7195,
      "step": 1914
    },
    {
      "epoch": 1.2354838709677418,
      "grad_norm": 1.171875,
      "learning_rate": 5.3383110500583905e-06,
      "loss": 1.0171,
      "step": 1915
    },
    {
      "epoch": 1.2361290322580645,
      "grad_norm": 0.9609375,
      "learning_rate": 5.334140638258006e-06,
      "loss": 0.734,
      "step": 1916
    },
    {
      "epoch": 1.2367741935483871,
      "grad_norm": 0.984375,
      "learning_rate": 5.329969992942257e-06,
      "loss": 0.8472,
      "step": 1917
    },
    {
      "epoch": 1.2374193548387096,
      "grad_norm": 1.0390625,
      "learning_rate": 5.325799117025809e-06,
      "loss": 1.0135,
      "step": 1918
    },
    {
      "epoch": 1.2380645161290322,
      "grad_norm": 0.73046875,
      "learning_rate": 5.321628013423497e-06,
      "loss": 0.6573,
      "step": 1919
    },
    {
      "epoch": 1.238709677419355,
      "grad_norm": 0.70703125,
      "learning_rate": 5.3174566850503085e-06,
      "loss": 0.6699,
      "step": 1920
    },
    {
      "epoch": 1.2393548387096773,
      "grad_norm": 0.91796875,
      "learning_rate": 5.313285134821391e-06,
      "loss": 0.8351,
      "step": 1921
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.03125,
      "learning_rate": 5.30911336565205e-06,
      "loss": 0.8485,
      "step": 1922
    },
    {
      "epoch": 1.2406451612903227,
      "grad_norm": 1.296875,
      "learning_rate": 5.304941380457739e-06,
      "loss": 1.2298,
      "step": 1923
    },
    {
      "epoch": 1.241290322580645,
      "grad_norm": 0.8671875,
      "learning_rate": 5.300769182154066e-06,
      "loss": 0.778,
      "step": 1924
    },
    {
      "epoch": 1.2419354838709677,
      "grad_norm": 1.1953125,
      "learning_rate": 5.2965967736567846e-06,
      "loss": 0.9239,
      "step": 1925
    },
    {
      "epoch": 1.2425806451612904,
      "grad_norm": 1.2109375,
      "learning_rate": 5.292424157881799e-06,
      "loss": 0.7674,
      "step": 1926
    },
    {
      "epoch": 1.2432258064516128,
      "grad_norm": 0.92578125,
      "learning_rate": 5.288251337745156e-06,
      "loss": 0.6331,
      "step": 1927
    },
    {
      "epoch": 1.2438709677419355,
      "grad_norm": 0.83203125,
      "learning_rate": 5.284078316163045e-06,
      "loss": 0.6731,
      "step": 1928
    },
    {
      "epoch": 1.2445161290322582,
      "grad_norm": 1.0234375,
      "learning_rate": 5.2799050960517985e-06,
      "loss": 0.7558,
      "step": 1929
    },
    {
      "epoch": 1.2451612903225806,
      "grad_norm": 0.84375,
      "learning_rate": 5.275731680327885e-06,
      "loss": 1.0442,
      "step": 1930
    },
    {
      "epoch": 1.2458064516129033,
      "grad_norm": 1.21875,
      "learning_rate": 5.271558071907911e-06,
      "loss": 0.8646,
      "step": 1931
    },
    {
      "epoch": 1.246451612903226,
      "grad_norm": 1.0,
      "learning_rate": 5.267384273708616e-06,
      "loss": 1.0033,
      "step": 1932
    },
    {
      "epoch": 1.2470967741935484,
      "grad_norm": 0.98046875,
      "learning_rate": 5.2632102886468764e-06,
      "loss": 0.8493,
      "step": 1933
    },
    {
      "epoch": 1.247741935483871,
      "grad_norm": 0.93359375,
      "learning_rate": 5.259036119639695e-06,
      "loss": 0.7151,
      "step": 1934
    },
    {
      "epoch": 1.2483870967741935,
      "grad_norm": 1.3359375,
      "learning_rate": 5.254861769604205e-06,
      "loss": 1.057,
      "step": 1935
    },
    {
      "epoch": 1.2490322580645161,
      "grad_norm": 0.84765625,
      "learning_rate": 5.250687241457664e-06,
      "loss": 0.6149,
      "step": 1936
    },
    {
      "epoch": 1.2496774193548388,
      "grad_norm": 0.76953125,
      "learning_rate": 5.246512538117459e-06,
      "loss": 0.5913,
      "step": 1937
    },
    {
      "epoch": 1.2503225806451612,
      "grad_norm": 0.99609375,
      "learning_rate": 5.242337662501094e-06,
      "loss": 0.7061,
      "step": 1938
    },
    {
      "epoch": 1.2509677419354839,
      "grad_norm": 0.9140625,
      "learning_rate": 5.238162617526195e-06,
      "loss": 0.7578,
      "step": 1939
    },
    {
      "epoch": 1.2516129032258063,
      "grad_norm": 0.8359375,
      "learning_rate": 5.233987406110509e-06,
      "loss": 0.6932,
      "step": 1940
    },
    {
      "epoch": 1.252258064516129,
      "grad_norm": 1.1171875,
      "learning_rate": 5.229812031171896e-06,
      "loss": 0.8556,
      "step": 1941
    },
    {
      "epoch": 1.2529032258064516,
      "grad_norm": 1.2734375,
      "learning_rate": 5.225636495628331e-06,
      "loss": 0.9423,
      "step": 1942
    },
    {
      "epoch": 1.253548387096774,
      "grad_norm": 0.8359375,
      "learning_rate": 5.221460802397903e-06,
      "loss": 0.7461,
      "step": 1943
    },
    {
      "epoch": 1.2541935483870967,
      "grad_norm": 0.80078125,
      "learning_rate": 5.217284954398811e-06,
      "loss": 0.6609,
      "step": 1944
    },
    {
      "epoch": 1.2548387096774194,
      "grad_norm": 1.375,
      "learning_rate": 5.213108954549358e-06,
      "loss": 1.1539,
      "step": 1945
    },
    {
      "epoch": 1.2554838709677418,
      "grad_norm": 1.0703125,
      "learning_rate": 5.208932805767959e-06,
      "loss": 0.8766,
      "step": 1946
    },
    {
      "epoch": 1.2561290322580645,
      "grad_norm": 1.21875,
      "learning_rate": 5.204756510973129e-06,
      "loss": 1.203,
      "step": 1947
    },
    {
      "epoch": 1.2567741935483872,
      "grad_norm": 0.87890625,
      "learning_rate": 5.200580073083487e-06,
      "loss": 0.605,
      "step": 1948
    },
    {
      "epoch": 1.2574193548387096,
      "grad_norm": 0.95703125,
      "learning_rate": 5.196403495017751e-06,
      "loss": 0.7979,
      "step": 1949
    },
    {
      "epoch": 1.2580645161290323,
      "grad_norm": 0.6796875,
      "learning_rate": 5.192226779694737e-06,
      "loss": 0.4506,
      "step": 1950
    },
    {
      "epoch": 1.258709677419355,
      "grad_norm": 0.87890625,
      "learning_rate": 5.188049930033358e-06,
      "loss": 0.6402,
      "step": 1951
    },
    {
      "epoch": 1.2593548387096773,
      "grad_norm": 0.88671875,
      "learning_rate": 5.18387294895262e-06,
      "loss": 0.9255,
      "step": 1952
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.1328125,
      "learning_rate": 5.179695839371621e-06,
      "loss": 1.0461,
      "step": 1953
    },
    {
      "epoch": 1.2606451612903227,
      "grad_norm": 1.3671875,
      "learning_rate": 5.175518604209548e-06,
      "loss": 1.2156,
      "step": 1954
    },
    {
      "epoch": 1.261290322580645,
      "grad_norm": 0.86328125,
      "learning_rate": 5.171341246385679e-06,
      "loss": 0.6224,
      "step": 1955
    },
    {
      "epoch": 1.2619354838709678,
      "grad_norm": 0.9921875,
      "learning_rate": 5.167163768819373e-06,
      "loss": 0.9285,
      "step": 1956
    },
    {
      "epoch": 1.2625806451612904,
      "grad_norm": 1.1875,
      "learning_rate": 5.162986174430075e-06,
      "loss": 0.9715,
      "step": 1957
    },
    {
      "epoch": 1.2632258064516129,
      "grad_norm": 1.5390625,
      "learning_rate": 5.158808466137313e-06,
      "loss": 1.0171,
      "step": 1958
    },
    {
      "epoch": 1.2638709677419355,
      "grad_norm": 1.2890625,
      "learning_rate": 5.154630646860692e-06,
      "loss": 0.9551,
      "step": 1959
    },
    {
      "epoch": 1.2645161290322582,
      "grad_norm": 0.828125,
      "learning_rate": 5.150452719519896e-06,
      "loss": 0.7408,
      "step": 1960
    },
    {
      "epoch": 1.2651612903225806,
      "grad_norm": 0.72265625,
      "learning_rate": 5.146274687034685e-06,
      "loss": 0.6913,
      "step": 1961
    },
    {
      "epoch": 1.2658064516129033,
      "grad_norm": 1.1640625,
      "learning_rate": 5.142096552324892e-06,
      "loss": 0.9136,
      "step": 1962
    },
    {
      "epoch": 1.266451612903226,
      "grad_norm": 0.84375,
      "learning_rate": 5.137918318310418e-06,
      "loss": 0.8502,
      "step": 1963
    },
    {
      "epoch": 1.2670967741935484,
      "grad_norm": 0.859375,
      "learning_rate": 5.133739987911242e-06,
      "loss": 0.5902,
      "step": 1964
    },
    {
      "epoch": 1.267741935483871,
      "grad_norm": 1.15625,
      "learning_rate": 5.129561564047402e-06,
      "loss": 1.0237,
      "step": 1965
    },
    {
      "epoch": 1.2683870967741935,
      "grad_norm": 1.1953125,
      "learning_rate": 5.125383049639003e-06,
      "loss": 0.9826,
      "step": 1966
    },
    {
      "epoch": 1.2690322580645161,
      "grad_norm": 0.875,
      "learning_rate": 5.121204447606218e-06,
      "loss": 0.8399,
      "step": 1967
    },
    {
      "epoch": 1.2696774193548386,
      "grad_norm": 1.2109375,
      "learning_rate": 5.117025760869274e-06,
      "loss": 1.1659,
      "step": 1968
    },
    {
      "epoch": 1.2703225806451612,
      "grad_norm": 0.67578125,
      "learning_rate": 5.112846992348463e-06,
      "loss": 0.6262,
      "step": 1969
    },
    {
      "epoch": 1.270967741935484,
      "grad_norm": 0.423828125,
      "learning_rate": 5.108668144964133e-06,
      "loss": 0.3447,
      "step": 1970
    },
    {
      "epoch": 1.2716129032258063,
      "grad_norm": 0.82421875,
      "learning_rate": 5.104489221636686e-06,
      "loss": 0.6335,
      "step": 1971
    },
    {
      "epoch": 1.272258064516129,
      "grad_norm": 0.890625,
      "learning_rate": 5.100310225286574e-06,
      "loss": 0.6577,
      "step": 1972
    },
    {
      "epoch": 1.2729032258064517,
      "grad_norm": 1.046875,
      "learning_rate": 5.096131158834309e-06,
      "loss": 0.7634,
      "step": 1973
    },
    {
      "epoch": 1.273548387096774,
      "grad_norm": 0.97265625,
      "learning_rate": 5.091952025200443e-06,
      "loss": 0.8369,
      "step": 1974
    },
    {
      "epoch": 1.2741935483870968,
      "grad_norm": 0.765625,
      "learning_rate": 5.087772827305577e-06,
      "loss": 0.7491,
      "step": 1975
    },
    {
      "epoch": 1.2748387096774194,
      "grad_norm": 0.9453125,
      "learning_rate": 5.083593568070362e-06,
      "loss": 0.74,
      "step": 1976
    },
    {
      "epoch": 1.2754838709677419,
      "grad_norm": 0.859375,
      "learning_rate": 5.079414250415486e-06,
      "loss": 0.7207,
      "step": 1977
    },
    {
      "epoch": 1.2761290322580645,
      "grad_norm": 0.80859375,
      "learning_rate": 5.0752348772616785e-06,
      "loss": 0.6087,
      "step": 1978
    },
    {
      "epoch": 1.2767741935483872,
      "grad_norm": 1.2265625,
      "learning_rate": 5.0710554515297114e-06,
      "loss": 1.0362,
      "step": 1979
    },
    {
      "epoch": 1.2774193548387096,
      "grad_norm": 0.6875,
      "learning_rate": 5.066875976140392e-06,
      "loss": 0.6127,
      "step": 1980
    },
    {
      "epoch": 1.2780645161290323,
      "grad_norm": 0.74609375,
      "learning_rate": 5.062696454014557e-06,
      "loss": 0.6798,
      "step": 1981
    },
    {
      "epoch": 1.278709677419355,
      "grad_norm": 0.984375,
      "learning_rate": 5.0585168880730864e-06,
      "loss": 0.7926,
      "step": 1982
    },
    {
      "epoch": 1.2793548387096774,
      "grad_norm": 1.1796875,
      "learning_rate": 5.0543372812368805e-06,
      "loss": 1.2971,
      "step": 1983
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.78125,
      "learning_rate": 5.050157636426874e-06,
      "loss": 0.7361,
      "step": 1984
    },
    {
      "epoch": 1.2806451612903227,
      "grad_norm": 1.0546875,
      "learning_rate": 5.0459779565640245e-06,
      "loss": 0.8104,
      "step": 1985
    },
    {
      "epoch": 1.2812903225806451,
      "grad_norm": 0.78125,
      "learning_rate": 5.041798244569317e-06,
      "loss": 0.6697,
      "step": 1986
    },
    {
      "epoch": 1.2819354838709678,
      "grad_norm": 0.55078125,
      "learning_rate": 5.037618503363759e-06,
      "loss": 0.448,
      "step": 1987
    },
    {
      "epoch": 1.2825806451612904,
      "grad_norm": 0.78515625,
      "learning_rate": 5.033438735868377e-06,
      "loss": 0.6304,
      "step": 1988
    },
    {
      "epoch": 1.2832258064516129,
      "grad_norm": 0.73828125,
      "learning_rate": 5.0292589450042164e-06,
      "loss": 0.6346,
      "step": 1989
    },
    {
      "epoch": 1.2838709677419355,
      "grad_norm": 0.85546875,
      "learning_rate": 5.025079133692337e-06,
      "loss": 0.6843,
      "step": 1990
    },
    {
      "epoch": 1.2845161290322582,
      "grad_norm": 0.83984375,
      "learning_rate": 5.020899304853818e-06,
      "loss": 0.6323,
      "step": 1991
    },
    {
      "epoch": 1.2851612903225806,
      "grad_norm": 0.89453125,
      "learning_rate": 5.016719461409746e-06,
      "loss": 0.8819,
      "step": 1992
    },
    {
      "epoch": 1.2858064516129033,
      "grad_norm": 1.1796875,
      "learning_rate": 5.012539606281218e-06,
      "loss": 1.1017,
      "step": 1993
    },
    {
      "epoch": 1.2864516129032257,
      "grad_norm": 0.73046875,
      "learning_rate": 5.008359742389343e-06,
      "loss": 0.6895,
      "step": 1994
    },
    {
      "epoch": 1.2870967741935484,
      "grad_norm": 1.03125,
      "learning_rate": 5.004179872655232e-06,
      "loss": 1.1135,
      "step": 1995
    },
    {
      "epoch": 1.2877419354838708,
      "grad_norm": 0.80078125,
      "learning_rate": 5e-06,
      "loss": 0.5761,
      "step": 1996
    },
    {
      "epoch": 1.2883870967741935,
      "grad_norm": 0.94921875,
      "learning_rate": 4.995820127344771e-06,
      "loss": 0.8227,
      "step": 1997
    },
    {
      "epoch": 1.2890322580645162,
      "grad_norm": 0.81640625,
      "learning_rate": 4.9916402576106595e-06,
      "loss": 0.6532,
      "step": 1998
    },
    {
      "epoch": 1.2896774193548386,
      "grad_norm": 0.98046875,
      "learning_rate": 4.987460393718784e-06,
      "loss": 0.8261,
      "step": 1999
    },
    {
      "epoch": 1.2903225806451613,
      "grad_norm": 1.125,
      "learning_rate": 4.983280538590256e-06,
      "loss": 1.0762,
      "step": 2000
    },
    {
      "epoch": 1.290967741935484,
      "grad_norm": 1.1484375,
      "learning_rate": 4.9791006951461835e-06,
      "loss": 0.8527,
      "step": 2001
    },
    {
      "epoch": 1.2916129032258064,
      "grad_norm": 1.2421875,
      "learning_rate": 4.974920866307663e-06,
      "loss": 0.7887,
      "step": 2002
    },
    {
      "epoch": 1.292258064516129,
      "grad_norm": 0.68359375,
      "learning_rate": 4.970741054995786e-06,
      "loss": 0.6585,
      "step": 2003
    },
    {
      "epoch": 1.2929032258064517,
      "grad_norm": 1.2265625,
      "learning_rate": 4.966561264131624e-06,
      "loss": 0.7782,
      "step": 2004
    },
    {
      "epoch": 1.293548387096774,
      "grad_norm": 0.81640625,
      "learning_rate": 4.962381496636242e-06,
      "loss": 0.5781,
      "step": 2005
    },
    {
      "epoch": 1.2941935483870968,
      "grad_norm": 0.8515625,
      "learning_rate": 4.958201755430685e-06,
      "loss": 0.7005,
      "step": 2006
    },
    {
      "epoch": 1.2948387096774194,
      "grad_norm": 1.1015625,
      "learning_rate": 4.954022043435978e-06,
      "loss": 1.0073,
      "step": 2007
    },
    {
      "epoch": 1.2954838709677419,
      "grad_norm": 0.93359375,
      "learning_rate": 4.949842363573129e-06,
      "loss": 0.8998,
      "step": 2008
    },
    {
      "epoch": 1.2961290322580645,
      "grad_norm": 1.015625,
      "learning_rate": 4.94566271876312e-06,
      "loss": 1.0131,
      "step": 2009
    },
    {
      "epoch": 1.2967741935483872,
      "grad_norm": 0.90234375,
      "learning_rate": 4.941483111926915e-06,
      "loss": 0.8495,
      "step": 2010
    },
    {
      "epoch": 1.2974193548387096,
      "grad_norm": 0.93359375,
      "learning_rate": 4.937303545985442e-06,
      "loss": 0.6962,
      "step": 2011
    },
    {
      "epoch": 1.2980645161290323,
      "grad_norm": 0.99609375,
      "learning_rate": 4.93312402385961e-06,
      "loss": 0.9081,
      "step": 2012
    },
    {
      "epoch": 1.298709677419355,
      "grad_norm": 0.84375,
      "learning_rate": 4.9289445484702885e-06,
      "loss": 0.7779,
      "step": 2013
    },
    {
      "epoch": 1.2993548387096774,
      "grad_norm": 1.2421875,
      "learning_rate": 4.924765122738322e-06,
      "loss": 1.0122,
      "step": 2014
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.1640625,
      "learning_rate": 4.920585749584517e-06,
      "loss": 0.9587,
      "step": 2015
    },
    {
      "epoch": 1.3006451612903227,
      "grad_norm": 0.5390625,
      "learning_rate": 4.916406431929639e-06,
      "loss": 0.4,
      "step": 2016
    },
    {
      "epoch": 1.3012903225806451,
      "grad_norm": 1.015625,
      "learning_rate": 4.912227172694424e-06,
      "loss": 0.7446,
      "step": 2017
    },
    {
      "epoch": 1.3019354838709678,
      "grad_norm": 1.4375,
      "learning_rate": 4.9080479747995585e-06,
      "loss": 1.0452,
      "step": 2018
    },
    {
      "epoch": 1.3025806451612902,
      "grad_norm": 1.3046875,
      "learning_rate": 4.903868841165692e-06,
      "loss": 0.8904,
      "step": 2019
    },
    {
      "epoch": 1.303225806451613,
      "grad_norm": 1.1953125,
      "learning_rate": 4.899689774713426e-06,
      "loss": 0.9409,
      "step": 2020
    },
    {
      "epoch": 1.3038709677419356,
      "grad_norm": 0.859375,
      "learning_rate": 4.895510778363315e-06,
      "loss": 0.6598,
      "step": 2021
    },
    {
      "epoch": 1.304516129032258,
      "grad_norm": 0.76171875,
      "learning_rate": 4.891331855035868e-06,
      "loss": 0.5054,
      "step": 2022
    },
    {
      "epoch": 1.3051612903225807,
      "grad_norm": 0.71484375,
      "learning_rate": 4.887153007651538e-06,
      "loss": 0.6125,
      "step": 2023
    },
    {
      "epoch": 1.305806451612903,
      "grad_norm": 1.015625,
      "learning_rate": 4.882974239130728e-06,
      "loss": 0.7645,
      "step": 2024
    },
    {
      "epoch": 1.3064516129032258,
      "grad_norm": 0.6640625,
      "learning_rate": 4.878795552393784e-06,
      "loss": 0.618,
      "step": 2025
    },
    {
      "epoch": 1.3070967741935484,
      "grad_norm": 0.98046875,
      "learning_rate": 4.874616950360998e-06,
      "loss": 0.9043,
      "step": 2026
    },
    {
      "epoch": 1.3077419354838709,
      "grad_norm": 1.109375,
      "learning_rate": 4.870438435952599e-06,
      "loss": 1.3959,
      "step": 2027
    },
    {
      "epoch": 1.3083870967741935,
      "grad_norm": 0.90625,
      "learning_rate": 4.8662600120887586e-06,
      "loss": 0.9311,
      "step": 2028
    },
    {
      "epoch": 1.3090322580645162,
      "grad_norm": 1.2734375,
      "learning_rate": 4.862081681689581e-06,
      "loss": 1.0039,
      "step": 2029
    },
    {
      "epoch": 1.3096774193548386,
      "grad_norm": 0.53125,
      "learning_rate": 4.85790344767511e-06,
      "loss": 0.3835,
      "step": 2030
    },
    {
      "epoch": 1.3103225806451613,
      "grad_norm": 0.875,
      "learning_rate": 4.853725312965315e-06,
      "loss": 0.6657,
      "step": 2031
    },
    {
      "epoch": 1.310967741935484,
      "grad_norm": 0.69921875,
      "learning_rate": 4.8495472804801045e-06,
      "loss": 0.704,
      "step": 2032
    },
    {
      "epoch": 1.3116129032258064,
      "grad_norm": 1.265625,
      "learning_rate": 4.84536935313931e-06,
      "loss": 0.8559,
      "step": 2033
    },
    {
      "epoch": 1.312258064516129,
      "grad_norm": 1.3984375,
      "learning_rate": 4.841191533862688e-06,
      "loss": 0.9131,
      "step": 2034
    },
    {
      "epoch": 1.3129032258064517,
      "grad_norm": 0.80078125,
      "learning_rate": 4.837013825569927e-06,
      "loss": 0.6028,
      "step": 2035
    },
    {
      "epoch": 1.3135483870967741,
      "grad_norm": 1.0859375,
      "learning_rate": 4.8328362311806285e-06,
      "loss": 0.7621,
      "step": 2036
    },
    {
      "epoch": 1.3141935483870968,
      "grad_norm": 0.65234375,
      "learning_rate": 4.828658753614322e-06,
      "loss": 0.54,
      "step": 2037
    },
    {
      "epoch": 1.3148387096774194,
      "grad_norm": 1.34375,
      "learning_rate": 4.8244813957904525e-06,
      "loss": 0.9111,
      "step": 2038
    },
    {
      "epoch": 1.3154838709677419,
      "grad_norm": 0.8203125,
      "learning_rate": 4.820304160628381e-06,
      "loss": 0.6009,
      "step": 2039
    },
    {
      "epoch": 1.3161290322580645,
      "grad_norm": 0.9765625,
      "learning_rate": 4.816127051047381e-06,
      "loss": 0.8834,
      "step": 2040
    },
    {
      "epoch": 1.3167741935483872,
      "grad_norm": 0.765625,
      "learning_rate": 4.811950069966643e-06,
      "loss": 0.6381,
      "step": 2041
    },
    {
      "epoch": 1.3174193548387096,
      "grad_norm": 1.2109375,
      "learning_rate": 4.807773220305266e-06,
      "loss": 0.8986,
      "step": 2042
    },
    {
      "epoch": 1.3180645161290323,
      "grad_norm": 1.28125,
      "learning_rate": 4.803596504982251e-06,
      "loss": 0.9494,
      "step": 2043
    },
    {
      "epoch": 1.318709677419355,
      "grad_norm": 0.79296875,
      "learning_rate": 4.799419926916515e-06,
      "loss": 0.7625,
      "step": 2044
    },
    {
      "epoch": 1.3193548387096774,
      "grad_norm": 0.5390625,
      "learning_rate": 4.795243489026872e-06,
      "loss": 0.4658,
      "step": 2045
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.66015625,
      "learning_rate": 4.791067194232042e-06,
      "loss": 0.522,
      "step": 2046
    },
    {
      "epoch": 1.3206451612903225,
      "grad_norm": 0.8984375,
      "learning_rate": 4.786891045450642e-06,
      "loss": 0.8675,
      "step": 2047
    },
    {
      "epoch": 1.3212903225806452,
      "grad_norm": 0.6796875,
      "learning_rate": 4.78271504560119e-06,
      "loss": 0.5775,
      "step": 2048
    },
    {
      "epoch": 1.3219354838709678,
      "grad_norm": 1.0078125,
      "learning_rate": 4.778539197602097e-06,
      "loss": 0.9174,
      "step": 2049
    },
    {
      "epoch": 1.3225806451612903,
      "grad_norm": 0.8125,
      "learning_rate": 4.77436350437167e-06,
      "loss": 0.5845,
      "step": 2050
    },
    {
      "epoch": 1.323225806451613,
      "grad_norm": 0.60546875,
      "learning_rate": 4.770187968828107e-06,
      "loss": 0.5039,
      "step": 2051
    },
    {
      "epoch": 1.3238709677419354,
      "grad_norm": 0.6953125,
      "learning_rate": 4.766012593889492e-06,
      "loss": 0.6563,
      "step": 2052
    },
    {
      "epoch": 1.324516129032258,
      "grad_norm": 0.8671875,
      "learning_rate": 4.761837382473806e-06,
      "loss": 0.6924,
      "step": 2053
    },
    {
      "epoch": 1.3251612903225807,
      "grad_norm": 1.1484375,
      "learning_rate": 4.757662337498908e-06,
      "loss": 0.8638,
      "step": 2054
    },
    {
      "epoch": 1.3258064516129031,
      "grad_norm": 0.890625,
      "learning_rate": 4.753487461882543e-06,
      "loss": 0.8456,
      "step": 2055
    },
    {
      "epoch": 1.3264516129032258,
      "grad_norm": 0.8515625,
      "learning_rate": 4.749312758542337e-06,
      "loss": 0.7226,
      "step": 2056
    },
    {
      "epoch": 1.3270967741935484,
      "grad_norm": 0.8515625,
      "learning_rate": 4.745138230395797e-06,
      "loss": 0.7027,
      "step": 2057
    },
    {
      "epoch": 1.3277419354838709,
      "grad_norm": 0.7890625,
      "learning_rate": 4.740963880360306e-06,
      "loss": 0.5579,
      "step": 2058
    },
    {
      "epoch": 1.3283870967741935,
      "grad_norm": 1.09375,
      "learning_rate": 4.736789711353125e-06,
      "loss": 0.9029,
      "step": 2059
    },
    {
      "epoch": 1.3290322580645162,
      "grad_norm": 0.78515625,
      "learning_rate": 4.732615726291384e-06,
      "loss": 0.7452,
      "step": 2060
    },
    {
      "epoch": 1.3296774193548386,
      "grad_norm": 0.9921875,
      "learning_rate": 4.728441928092091e-06,
      "loss": 0.836,
      "step": 2061
    },
    {
      "epoch": 1.3303225806451613,
      "grad_norm": 0.83203125,
      "learning_rate": 4.724268319672119e-06,
      "loss": 0.8274,
      "step": 2062
    },
    {
      "epoch": 1.330967741935484,
      "grad_norm": 1.0546875,
      "learning_rate": 4.720094903948202e-06,
      "loss": 0.8634,
      "step": 2063
    },
    {
      "epoch": 1.3316129032258064,
      "grad_norm": 0.89453125,
      "learning_rate": 4.715921683836956e-06,
      "loss": 0.7265,
      "step": 2064
    },
    {
      "epoch": 1.332258064516129,
      "grad_norm": 1.1328125,
      "learning_rate": 4.711748662254845e-06,
      "loss": 0.7558,
      "step": 2065
    },
    {
      "epoch": 1.3329032258064517,
      "grad_norm": 0.640625,
      "learning_rate": 4.7075758421182025e-06,
      "loss": 0.6724,
      "step": 2066
    },
    {
      "epoch": 1.3335483870967741,
      "grad_norm": 0.765625,
      "learning_rate": 4.703403226343215e-06,
      "loss": 0.7398,
      "step": 2067
    },
    {
      "epoch": 1.3341935483870968,
      "grad_norm": 0.65625,
      "learning_rate": 4.699230817845935e-06,
      "loss": 0.5417,
      "step": 2068
    },
    {
      "epoch": 1.3348387096774195,
      "grad_norm": 0.91015625,
      "learning_rate": 4.695058619542261e-06,
      "loss": 0.9464,
      "step": 2069
    },
    {
      "epoch": 1.335483870967742,
      "grad_norm": 1.390625,
      "learning_rate": 4.690886634347951e-06,
      "loss": 0.8734,
      "step": 2070
    },
    {
      "epoch": 1.3361290322580646,
      "grad_norm": 0.96875,
      "learning_rate": 4.686714865178611e-06,
      "loss": 0.7821,
      "step": 2071
    },
    {
      "epoch": 1.3367741935483872,
      "grad_norm": 0.94921875,
      "learning_rate": 4.682543314949693e-06,
      "loss": 0.6854,
      "step": 2072
    },
    {
      "epoch": 1.3374193548387097,
      "grad_norm": 1.28125,
      "learning_rate": 4.6783719865765065e-06,
      "loss": 0.9606,
      "step": 2073
    },
    {
      "epoch": 1.3380645161290323,
      "grad_norm": 0.89453125,
      "learning_rate": 4.674200882974192e-06,
      "loss": 0.5301,
      "step": 2074
    },
    {
      "epoch": 1.3387096774193548,
      "grad_norm": 0.6484375,
      "learning_rate": 4.670030007057744e-06,
      "loss": 0.4926,
      "step": 2075
    },
    {
      "epoch": 1.3393548387096774,
      "grad_norm": 1.0546875,
      "learning_rate": 4.665859361741993e-06,
      "loss": 0.867,
      "step": 2076
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.2109375,
      "learning_rate": 4.661688949941611e-06,
      "loss": 1.0419,
      "step": 2077
    },
    {
      "epoch": 1.3406451612903225,
      "grad_norm": 1.296875,
      "learning_rate": 4.657518774571102e-06,
      "loss": 1.1174,
      "step": 2078
    },
    {
      "epoch": 1.3412903225806452,
      "grad_norm": 0.69140625,
      "learning_rate": 4.653348838544811e-06,
      "loss": 0.6672,
      "step": 2079
    },
    {
      "epoch": 1.3419354838709676,
      "grad_norm": 0.54296875,
      "learning_rate": 4.649179144776912e-06,
      "loss": 0.4197,
      "step": 2080
    },
    {
      "epoch": 1.3425806451612903,
      "grad_norm": 0.8671875,
      "learning_rate": 4.645009696181405e-06,
      "loss": 0.6673,
      "step": 2081
    },
    {
      "epoch": 1.343225806451613,
      "grad_norm": 0.6953125,
      "learning_rate": 4.640840495672131e-06,
      "loss": 0.8646,
      "step": 2082
    },
    {
      "epoch": 1.3438709677419354,
      "grad_norm": 1.46875,
      "learning_rate": 4.6366715461627454e-06,
      "loss": 0.837,
      "step": 2083
    },
    {
      "epoch": 1.344516129032258,
      "grad_norm": 0.7734375,
      "learning_rate": 4.632502850566739e-06,
      "loss": 0.8386,
      "step": 2084
    },
    {
      "epoch": 1.3451612903225807,
      "grad_norm": 0.70703125,
      "learning_rate": 4.628334411797412e-06,
      "loss": 0.565,
      "step": 2085
    },
    {
      "epoch": 1.3458064516129031,
      "grad_norm": 0.8203125,
      "learning_rate": 4.624166232767898e-06,
      "loss": 0.5272,
      "step": 2086
    },
    {
      "epoch": 1.3464516129032258,
      "grad_norm": 0.9609375,
      "learning_rate": 4.619998316391141e-06,
      "loss": 0.9333,
      "step": 2087
    },
    {
      "epoch": 1.3470967741935485,
      "grad_norm": 1.2578125,
      "learning_rate": 4.615830665579908e-06,
      "loss": 0.9657,
      "step": 2088
    },
    {
      "epoch": 1.347741935483871,
      "grad_norm": 1.109375,
      "learning_rate": 4.611663283246773e-06,
      "loss": 0.7297,
      "step": 2089
    },
    {
      "epoch": 1.3483870967741935,
      "grad_norm": 0.71875,
      "learning_rate": 4.607496172304125e-06,
      "loss": 0.6126,
      "step": 2090
    },
    {
      "epoch": 1.3490322580645162,
      "grad_norm": 0.95703125,
      "learning_rate": 4.6033293356641685e-06,
      "loss": 0.7584,
      "step": 2091
    },
    {
      "epoch": 1.3496774193548386,
      "grad_norm": 0.5859375,
      "learning_rate": 4.599162776238906e-06,
      "loss": 0.5535,
      "step": 2092
    },
    {
      "epoch": 1.3503225806451613,
      "grad_norm": 1.09375,
      "learning_rate": 4.594996496940159e-06,
      "loss": 0.8709,
      "step": 2093
    },
    {
      "epoch": 1.350967741935484,
      "grad_norm": 0.66796875,
      "learning_rate": 4.590830500679541e-06,
      "loss": 0.5174,
      "step": 2094
    },
    {
      "epoch": 1.3516129032258064,
      "grad_norm": 0.7890625,
      "learning_rate": 4.586664790368478e-06,
      "loss": 0.6451,
      "step": 2095
    },
    {
      "epoch": 1.352258064516129,
      "grad_norm": 0.8046875,
      "learning_rate": 4.582499368918185e-06,
      "loss": 0.763,
      "step": 2096
    },
    {
      "epoch": 1.3529032258064517,
      "grad_norm": 0.96875,
      "learning_rate": 4.578334239239689e-06,
      "loss": 0.8308,
      "step": 2097
    },
    {
      "epoch": 1.3535483870967742,
      "grad_norm": 0.87890625,
      "learning_rate": 4.574169404243802e-06,
      "loss": 0.8278,
      "step": 2098
    },
    {
      "epoch": 1.3541935483870968,
      "grad_norm": 0.8671875,
      "learning_rate": 4.570004866841132e-06,
      "loss": 0.6904,
      "step": 2099
    },
    {
      "epoch": 1.3548387096774195,
      "grad_norm": 0.8828125,
      "learning_rate": 4.565840629942084e-06,
      "loss": 0.6355,
      "step": 2100
    },
    {
      "epoch": 1.355483870967742,
      "grad_norm": 1.078125,
      "learning_rate": 4.561676696456845e-06,
      "loss": 0.8663,
      "step": 2101
    },
    {
      "epoch": 1.3561290322580646,
      "grad_norm": 1.234375,
      "learning_rate": 4.557513069295402e-06,
      "loss": 1.1701,
      "step": 2102
    },
    {
      "epoch": 1.356774193548387,
      "grad_norm": 1.1484375,
      "learning_rate": 4.553349751367512e-06,
      "loss": 1.048,
      "step": 2103
    },
    {
      "epoch": 1.3574193548387097,
      "grad_norm": 1.125,
      "learning_rate": 4.5491867455827325e-06,
      "loss": 0.8963,
      "step": 2104
    },
    {
      "epoch": 1.3580645161290323,
      "grad_norm": 1.2578125,
      "learning_rate": 4.545024054850388e-06,
      "loss": 0.9666,
      "step": 2105
    },
    {
      "epoch": 1.3587096774193548,
      "grad_norm": 1.5234375,
      "learning_rate": 4.540861682079596e-06,
      "loss": 0.9243,
      "step": 2106
    },
    {
      "epoch": 1.3593548387096774,
      "grad_norm": 1.28125,
      "learning_rate": 4.536699630179241e-06,
      "loss": 0.8713,
      "step": 2107
    },
    {
      "epoch": 1.3599999999999999,
      "grad_norm": 0.703125,
      "learning_rate": 4.532537902057987e-06,
      "loss": 0.4897,
      "step": 2108
    },
    {
      "epoch": 1.3606451612903225,
      "grad_norm": 0.86328125,
      "learning_rate": 4.528376500624277e-06,
      "loss": 0.656,
      "step": 2109
    },
    {
      "epoch": 1.3612903225806452,
      "grad_norm": 0.859375,
      "learning_rate": 4.524215428786315e-06,
      "loss": 0.7018,
      "step": 2110
    },
    {
      "epoch": 1.3619354838709676,
      "grad_norm": 0.671875,
      "learning_rate": 4.520054689452089e-06,
      "loss": 0.672,
      "step": 2111
    },
    {
      "epoch": 1.3625806451612903,
      "grad_norm": 1.1796875,
      "learning_rate": 4.515894285529337e-06,
      "loss": 0.853,
      "step": 2112
    },
    {
      "epoch": 1.363225806451613,
      "grad_norm": 0.82421875,
      "learning_rate": 4.51173421992558e-06,
      "loss": 0.6096,
      "step": 2113
    },
    {
      "epoch": 1.3638709677419354,
      "grad_norm": 0.609375,
      "learning_rate": 4.507574495548089e-06,
      "loss": 0.5162,
      "step": 2114
    },
    {
      "epoch": 1.364516129032258,
      "grad_norm": 0.95703125,
      "learning_rate": 4.503415115303907e-06,
      "loss": 0.8068,
      "step": 2115
    },
    {
      "epoch": 1.3651612903225807,
      "grad_norm": 1.15625,
      "learning_rate": 4.499256082099828e-06,
      "loss": 0.969,
      "step": 2116
    },
    {
      "epoch": 1.3658064516129031,
      "grad_norm": 0.7421875,
      "learning_rate": 4.495097398842406e-06,
      "loss": 0.6725,
      "step": 2117
    },
    {
      "epoch": 1.3664516129032258,
      "grad_norm": 0.83984375,
      "learning_rate": 4.490939068437955e-06,
      "loss": 0.75,
      "step": 2118
    },
    {
      "epoch": 1.3670967741935485,
      "grad_norm": 0.81640625,
      "learning_rate": 4.486781093792536e-06,
      "loss": 0.5288,
      "step": 2119
    },
    {
      "epoch": 1.367741935483871,
      "grad_norm": 0.953125,
      "learning_rate": 4.482623477811968e-06,
      "loss": 0.7377,
      "step": 2120
    },
    {
      "epoch": 1.3683870967741936,
      "grad_norm": 0.99609375,
      "learning_rate": 4.478466223401811e-06,
      "loss": 0.8684,
      "step": 2121
    },
    {
      "epoch": 1.3690322580645162,
      "grad_norm": 1.0625,
      "learning_rate": 4.474309333467381e-06,
      "loss": 1.0697,
      "step": 2122
    },
    {
      "epoch": 1.3696774193548387,
      "grad_norm": 1.0625,
      "learning_rate": 4.470152810913731e-06,
      "loss": 0.8127,
      "step": 2123
    },
    {
      "epoch": 1.3703225806451613,
      "grad_norm": 1.1640625,
      "learning_rate": 4.465996658645668e-06,
      "loss": 0.6652,
      "step": 2124
    },
    {
      "epoch": 1.370967741935484,
      "grad_norm": 1.03125,
      "learning_rate": 4.461840879567729e-06,
      "loss": 0.7456,
      "step": 2125
    },
    {
      "epoch": 1.3716129032258064,
      "grad_norm": 0.78515625,
      "learning_rate": 4.4576854765841935e-06,
      "loss": 0.6915,
      "step": 2126
    },
    {
      "epoch": 1.372258064516129,
      "grad_norm": 0.87890625,
      "learning_rate": 4.453530452599083e-06,
      "loss": 0.8845,
      "step": 2127
    },
    {
      "epoch": 1.3729032258064517,
      "grad_norm": 1.0546875,
      "learning_rate": 4.449375810516149e-06,
      "loss": 0.841,
      "step": 2128
    },
    {
      "epoch": 1.3735483870967742,
      "grad_norm": 0.94140625,
      "learning_rate": 4.4452215532388785e-06,
      "loss": 0.7904,
      "step": 2129
    },
    {
      "epoch": 1.3741935483870968,
      "grad_norm": 0.6484375,
      "learning_rate": 4.441067683670487e-06,
      "loss": 0.5654,
      "step": 2130
    },
    {
      "epoch": 1.3748387096774193,
      "grad_norm": 0.75390625,
      "learning_rate": 4.436914204713926e-06,
      "loss": 0.6441,
      "step": 2131
    },
    {
      "epoch": 1.375483870967742,
      "grad_norm": 0.765625,
      "learning_rate": 4.432761119271863e-06,
      "loss": 0.6555,
      "step": 2132
    },
    {
      "epoch": 1.3761290322580644,
      "grad_norm": 1.0859375,
      "learning_rate": 4.4286084302467014e-06,
      "loss": 0.8395,
      "step": 2133
    },
    {
      "epoch": 1.376774193548387,
      "grad_norm": 0.80859375,
      "learning_rate": 4.424456140540559e-06,
      "loss": 0.7322,
      "step": 2134
    },
    {
      "epoch": 1.3774193548387097,
      "grad_norm": 0.88671875,
      "learning_rate": 4.420304253055279e-06,
      "loss": 0.7608,
      "step": 2135
    },
    {
      "epoch": 1.3780645161290321,
      "grad_norm": 1.25,
      "learning_rate": 4.416152770692424e-06,
      "loss": 0.9955,
      "step": 2136
    },
    {
      "epoch": 1.3787096774193548,
      "grad_norm": 0.9609375,
      "learning_rate": 4.41200169635327e-06,
      "loss": 0.802,
      "step": 2137
    },
    {
      "epoch": 1.3793548387096775,
      "grad_norm": 0.7734375,
      "learning_rate": 4.4078510329388115e-06,
      "loss": 0.6846,
      "step": 2138
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.875,
      "learning_rate": 4.403700783349755e-06,
      "loss": 0.6928,
      "step": 2139
    },
    {
      "epoch": 1.3806451612903226,
      "grad_norm": 0.9375,
      "learning_rate": 4.399550950486515e-06,
      "loss": 1.0683,
      "step": 2140
    },
    {
      "epoch": 1.3812903225806452,
      "grad_norm": 0.8671875,
      "learning_rate": 4.395401537249218e-06,
      "loss": 0.8011,
      "step": 2141
    },
    {
      "epoch": 1.3819354838709677,
      "grad_norm": 0.87109375,
      "learning_rate": 4.391252546537698e-06,
      "loss": 0.7258,
      "step": 2142
    },
    {
      "epoch": 1.3825806451612903,
      "grad_norm": 1.109375,
      "learning_rate": 4.387103981251489e-06,
      "loss": 0.7019,
      "step": 2143
    },
    {
      "epoch": 1.383225806451613,
      "grad_norm": 0.68359375,
      "learning_rate": 4.382955844289831e-06,
      "loss": 0.6447,
      "step": 2144
    },
    {
      "epoch": 1.3838709677419354,
      "grad_norm": 1.28125,
      "learning_rate": 4.378808138551665e-06,
      "loss": 1.0747,
      "step": 2145
    },
    {
      "epoch": 1.384516129032258,
      "grad_norm": 1.046875,
      "learning_rate": 4.374660866935629e-06,
      "loss": 0.8839,
      "step": 2146
    },
    {
      "epoch": 1.3851612903225807,
      "grad_norm": 1.125,
      "learning_rate": 4.37051403234006e-06,
      "loss": 0.9284,
      "step": 2147
    },
    {
      "epoch": 1.3858064516129032,
      "grad_norm": 1.1640625,
      "learning_rate": 4.366367637662986e-06,
      "loss": 0.9029,
      "step": 2148
    },
    {
      "epoch": 1.3864516129032258,
      "grad_norm": 1.0234375,
      "learning_rate": 4.362221685802132e-06,
      "loss": 0.8393,
      "step": 2149
    },
    {
      "epoch": 1.3870967741935485,
      "grad_norm": 0.8515625,
      "learning_rate": 4.358076179654908e-06,
      "loss": 0.7269,
      "step": 2150
    },
    {
      "epoch": 1.387741935483871,
      "grad_norm": 1.5234375,
      "learning_rate": 4.353931122118419e-06,
      "loss": 0.9452,
      "step": 2151
    },
    {
      "epoch": 1.3883870967741936,
      "grad_norm": 0.99609375,
      "learning_rate": 4.349786516089452e-06,
      "loss": 0.8439,
      "step": 2152
    },
    {
      "epoch": 1.3890322580645162,
      "grad_norm": 0.79296875,
      "learning_rate": 4.345642364464477e-06,
      "loss": 0.644,
      "step": 2153
    },
    {
      "epoch": 1.3896774193548387,
      "grad_norm": 1.140625,
      "learning_rate": 4.341498670139652e-06,
      "loss": 0.9274,
      "step": 2154
    },
    {
      "epoch": 1.3903225806451613,
      "grad_norm": 0.7265625,
      "learning_rate": 4.337355436010811e-06,
      "loss": 0.5413,
      "step": 2155
    },
    {
      "epoch": 1.390967741935484,
      "grad_norm": 0.65234375,
      "learning_rate": 4.333212664973471e-06,
      "loss": 0.5312,
      "step": 2156
    },
    {
      "epoch": 1.3916129032258064,
      "grad_norm": 0.8828125,
      "learning_rate": 4.3290703599228165e-06,
      "loss": 0.7735,
      "step": 2157
    },
    {
      "epoch": 1.392258064516129,
      "grad_norm": 1.03125,
      "learning_rate": 4.324928523753717e-06,
      "loss": 0.6618,
      "step": 2158
    },
    {
      "epoch": 1.3929032258064515,
      "grad_norm": 1.3671875,
      "learning_rate": 4.320787159360706e-06,
      "loss": 1.1427,
      "step": 2159
    },
    {
      "epoch": 1.3935483870967742,
      "grad_norm": 0.8828125,
      "learning_rate": 4.316646269637994e-06,
      "loss": 0.6229,
      "step": 2160
    },
    {
      "epoch": 1.3941935483870966,
      "grad_norm": 0.66796875,
      "learning_rate": 4.312505857479454e-06,
      "loss": 0.5618,
      "step": 2161
    },
    {
      "epoch": 1.3948387096774193,
      "grad_norm": 1.0859375,
      "learning_rate": 4.308365925778628e-06,
      "loss": 0.765,
      "step": 2162
    },
    {
      "epoch": 1.395483870967742,
      "grad_norm": 0.91015625,
      "learning_rate": 4.304226477428722e-06,
      "loss": 0.8843,
      "step": 2163
    },
    {
      "epoch": 1.3961290322580644,
      "grad_norm": 1.3203125,
      "learning_rate": 4.300087515322605e-06,
      "loss": 0.9159,
      "step": 2164
    },
    {
      "epoch": 1.396774193548387,
      "grad_norm": 1.0078125,
      "learning_rate": 4.295949042352804e-06,
      "loss": 0.7305,
      "step": 2165
    },
    {
      "epoch": 1.3974193548387097,
      "grad_norm": 1.09375,
      "learning_rate": 4.291811061411506e-06,
      "loss": 0.7471,
      "step": 2166
    },
    {
      "epoch": 1.3980645161290322,
      "grad_norm": 0.91015625,
      "learning_rate": 4.287673575390555e-06,
      "loss": 0.9308,
      "step": 2167
    },
    {
      "epoch": 1.3987096774193548,
      "grad_norm": 0.89453125,
      "learning_rate": 4.283536587181445e-06,
      "loss": 0.6706,
      "step": 2168
    },
    {
      "epoch": 1.3993548387096775,
      "grad_norm": 0.9296875,
      "learning_rate": 4.279400099675328e-06,
      "loss": 0.8239,
      "step": 2169
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.875,
      "learning_rate": 4.2752641157629996e-06,
      "loss": 0.9185,
      "step": 2170
    },
    {
      "epoch": 1.4006451612903226,
      "grad_norm": 0.98046875,
      "learning_rate": 4.27112863833491e-06,
      "loss": 0.7602,
      "step": 2171
    },
    {
      "epoch": 1.4012903225806452,
      "grad_norm": 0.96875,
      "learning_rate": 4.266993670281151e-06,
      "loss": 0.7836,
      "step": 2172
    },
    {
      "epoch": 1.4019354838709677,
      "grad_norm": 1.1875,
      "learning_rate": 4.262859214491457e-06,
      "loss": 0.8016,
      "step": 2173
    },
    {
      "epoch": 1.4025806451612903,
      "grad_norm": 1.2265625,
      "learning_rate": 4.258725273855211e-06,
      "loss": 0.8533,
      "step": 2174
    },
    {
      "epoch": 1.403225806451613,
      "grad_norm": 0.88671875,
      "learning_rate": 4.25459185126143e-06,
      "loss": 0.6244,
      "step": 2175
    },
    {
      "epoch": 1.4038709677419354,
      "grad_norm": 1.359375,
      "learning_rate": 4.250458949598772e-06,
      "loss": 1.4141,
      "step": 2176
    },
    {
      "epoch": 1.404516129032258,
      "grad_norm": 1.296875,
      "learning_rate": 4.246326571755529e-06,
      "loss": 1.0327,
      "step": 2177
    },
    {
      "epoch": 1.4051612903225807,
      "grad_norm": 1.125,
      "learning_rate": 4.242194720619629e-06,
      "loss": 0.9098,
      "step": 2178
    },
    {
      "epoch": 1.4058064516129032,
      "grad_norm": 1.2734375,
      "learning_rate": 4.238063399078629e-06,
      "loss": 0.9127,
      "step": 2179
    },
    {
      "epoch": 1.4064516129032258,
      "grad_norm": 1.1796875,
      "learning_rate": 4.23393261001972e-06,
      "loss": 1.1493,
      "step": 2180
    },
    {
      "epoch": 1.4070967741935485,
      "grad_norm": 0.94921875,
      "learning_rate": 4.229802356329717e-06,
      "loss": 0.7836,
      "step": 2181
    },
    {
      "epoch": 1.407741935483871,
      "grad_norm": 0.60546875,
      "learning_rate": 4.225672640895062e-06,
      "loss": 0.4511,
      "step": 2182
    },
    {
      "epoch": 1.4083870967741936,
      "grad_norm": 1.234375,
      "learning_rate": 4.221543466601824e-06,
      "loss": 0.9838,
      "step": 2183
    },
    {
      "epoch": 1.4090322580645163,
      "grad_norm": 0.59375,
      "learning_rate": 4.2174148363356864e-06,
      "loss": 0.5274,
      "step": 2184
    },
    {
      "epoch": 1.4096774193548387,
      "grad_norm": 1.09375,
      "learning_rate": 4.213286752981961e-06,
      "loss": 1.1571,
      "step": 2185
    },
    {
      "epoch": 1.4103225806451614,
      "grad_norm": 1.3984375,
      "learning_rate": 4.209159219425572e-06,
      "loss": 1.0582,
      "step": 2186
    },
    {
      "epoch": 1.4109677419354838,
      "grad_norm": 0.671875,
      "learning_rate": 4.205032238551062e-06,
      "loss": 0.476,
      "step": 2187
    },
    {
      "epoch": 1.4116129032258065,
      "grad_norm": 0.98828125,
      "learning_rate": 4.200905813242583e-06,
      "loss": 0.8881,
      "step": 2188
    },
    {
      "epoch": 1.412258064516129,
      "grad_norm": 0.7109375,
      "learning_rate": 4.196779946383907e-06,
      "loss": 0.5128,
      "step": 2189
    },
    {
      "epoch": 1.4129032258064516,
      "grad_norm": 0.66796875,
      "learning_rate": 4.1926546408584055e-06,
      "loss": 0.6165,
      "step": 2190
    },
    {
      "epoch": 1.4135483870967742,
      "grad_norm": 1.0546875,
      "learning_rate": 4.188529899549066e-06,
      "loss": 0.774,
      "step": 2191
    },
    {
      "epoch": 1.4141935483870967,
      "grad_norm": 0.9140625,
      "learning_rate": 4.184405725338478e-06,
      "loss": 0.7742,
      "step": 2192
    },
    {
      "epoch": 1.4148387096774193,
      "grad_norm": 0.98046875,
      "learning_rate": 4.180282121108834e-06,
      "loss": 0.8352,
      "step": 2193
    },
    {
      "epoch": 1.415483870967742,
      "grad_norm": 0.8125,
      "learning_rate": 4.1761590897419315e-06,
      "loss": 0.7847,
      "step": 2194
    },
    {
      "epoch": 1.4161290322580644,
      "grad_norm": 0.76171875,
      "learning_rate": 4.172036634119163e-06,
      "loss": 0.5939,
      "step": 2195
    },
    {
      "epoch": 1.416774193548387,
      "grad_norm": 0.90625,
      "learning_rate": 4.167914757121524e-06,
      "loss": 0.8371,
      "step": 2196
    },
    {
      "epoch": 1.4174193548387097,
      "grad_norm": 0.97265625,
      "learning_rate": 4.1637934616296e-06,
      "loss": 0.7478,
      "step": 2197
    },
    {
      "epoch": 1.4180645161290322,
      "grad_norm": 1.1875,
      "learning_rate": 4.159672750523576e-06,
      "loss": 1.042,
      "step": 2198
    },
    {
      "epoch": 1.4187096774193548,
      "grad_norm": 0.86328125,
      "learning_rate": 4.155552626683224e-06,
      "loss": 0.5875,
      "step": 2199
    },
    {
      "epoch": 1.4193548387096775,
      "grad_norm": 1.03125,
      "learning_rate": 4.1514330929879055e-06,
      "loss": 1.1279,
      "step": 2200
    },
    {
      "epoch": 1.4193548387096775,
      "eval_loss": 0.9006433486938477,
      "eval_model_preparation_time": 0.0144,
      "eval_runtime": 82.8354,
      "eval_samples_per_second": 3.597,
      "eval_steps_per_second": 3.597,
      "step": 2200
    }
  ],
  "logging_steps": 1,
  "max_steps": 3875,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 550,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 2.739449206634963e+18,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}