{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 214,
  "global_step": 214,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004672897196261682,
      "grad_norm": 1.5234375,
      "learning_rate": 1e-05,
      "loss": 2.9069,
      "step": 1
    },
    {
      "epoch": 0.009345794392523364,
      "grad_norm": 1.546875,
      "learning_rate": 9.953271028037384e-06,
      "loss": 2.9462,
      "step": 2
    },
    {
      "epoch": 0.014018691588785047,
      "grad_norm": 1.4375,
      "learning_rate": 9.906542056074768e-06,
      "loss": 2.8437,
      "step": 3
    },
    {
      "epoch": 0.018691588785046728,
      "grad_norm": 1.4140625,
      "learning_rate": 9.859813084112151e-06,
      "loss": 2.8093,
      "step": 4
    },
    {
      "epoch": 0.02336448598130841,
      "grad_norm": 1.140625,
      "learning_rate": 9.813084112149533e-06,
      "loss": 2.7097,
      "step": 5
    },
    {
      "epoch": 0.028037383177570093,
      "grad_norm": 1.0703125,
      "learning_rate": 9.766355140186918e-06,
      "loss": 2.7076,
      "step": 6
    },
    {
      "epoch": 0.03271028037383177,
      "grad_norm": 0.953125,
      "learning_rate": 9.7196261682243e-06,
      "loss": 2.6449,
      "step": 7
    },
    {
      "epoch": 0.037383177570093455,
      "grad_norm": 0.8984375,
      "learning_rate": 9.672897196261683e-06,
      "loss": 2.6143,
      "step": 8
    },
    {
      "epoch": 0.04205607476635514,
      "grad_norm": 0.91015625,
      "learning_rate": 9.626168224299066e-06,
      "loss": 2.5539,
      "step": 9
    },
    {
      "epoch": 0.04672897196261682,
      "grad_norm": 0.85546875,
      "learning_rate": 9.57943925233645e-06,
      "loss": 2.4964,
      "step": 10
    },
    {
      "epoch": 0.0514018691588785,
      "grad_norm": 0.73046875,
      "learning_rate": 9.532710280373833e-06,
      "loss": 2.4061,
      "step": 11
    },
    {
      "epoch": 0.056074766355140186,
      "grad_norm": 0.6875,
      "learning_rate": 9.485981308411217e-06,
      "loss": 2.3791,
      "step": 12
    },
    {
      "epoch": 0.06074766355140187,
      "grad_norm": 0.66796875,
      "learning_rate": 9.439252336448598e-06,
      "loss": 2.3691,
      "step": 13
    },
    {
      "epoch": 0.06542056074766354,
      "grad_norm": 0.6484375,
      "learning_rate": 9.392523364485983e-06,
      "loss": 2.3177,
      "step": 14
    },
    {
      "epoch": 0.07009345794392523,
      "grad_norm": 0.59375,
      "learning_rate": 9.345794392523365e-06,
      "loss": 2.2444,
      "step": 15
    },
    {
      "epoch": 0.07476635514018691,
      "grad_norm": 0.59375,
      "learning_rate": 9.299065420560748e-06,
      "loss": 2.261,
      "step": 16
    },
    {
      "epoch": 0.0794392523364486,
      "grad_norm": 0.609375,
      "learning_rate": 9.252336448598132e-06,
      "loss": 2.2963,
      "step": 17
    },
    {
      "epoch": 0.08411214953271028,
      "grad_norm": 0.5859375,
      "learning_rate": 9.205607476635515e-06,
      "loss": 2.2225,
      "step": 18
    },
    {
      "epoch": 0.08878504672897196,
      "grad_norm": 0.54296875,
      "learning_rate": 9.158878504672899e-06,
      "loss": 2.1646,
      "step": 19
    },
    {
      "epoch": 0.09345794392523364,
      "grad_norm": 0.5625,
      "learning_rate": 9.112149532710282e-06,
      "loss": 2.1857,
      "step": 20
    },
    {
      "epoch": 0.09813084112149532,
      "grad_norm": 0.578125,
      "learning_rate": 9.065420560747664e-06,
      "loss": 2.1392,
      "step": 21
    },
    {
      "epoch": 0.102803738317757,
      "grad_norm": 0.57421875,
      "learning_rate": 9.018691588785047e-06,
      "loss": 2.1135,
      "step": 22
    },
    {
      "epoch": 0.10747663551401869,
      "grad_norm": 0.55859375,
      "learning_rate": 8.97196261682243e-06,
      "loss": 2.1053,
      "step": 23
    },
    {
      "epoch": 0.11214953271028037,
      "grad_norm": 0.546875,
      "learning_rate": 8.925233644859814e-06,
      "loss": 2.0868,
      "step": 24
    },
    {
      "epoch": 0.11682242990654206,
      "grad_norm": 0.5234375,
      "learning_rate": 8.878504672897197e-06,
      "loss": 2.0354,
      "step": 25
    },
    {
      "epoch": 0.12149532710280374,
      "grad_norm": 0.5,
      "learning_rate": 8.83177570093458e-06,
      "loss": 2.0114,
      "step": 26
    },
    {
      "epoch": 0.1261682242990654,
      "grad_norm": 0.498046875,
      "learning_rate": 8.785046728971963e-06,
      "loss": 2.0158,
      "step": 27
    },
    {
      "epoch": 0.1308411214953271,
      "grad_norm": 0.498046875,
      "learning_rate": 8.738317757009348e-06,
      "loss": 2.0091,
      "step": 28
    },
    {
      "epoch": 0.13551401869158877,
      "grad_norm": 0.47265625,
      "learning_rate": 8.69158878504673e-06,
      "loss": 1.9591,
      "step": 29
    },
    {
      "epoch": 0.14018691588785046,
      "grad_norm": 0.45703125,
      "learning_rate": 8.644859813084113e-06,
      "loss": 1.9576,
      "step": 30
    },
    {
      "epoch": 0.14485981308411214,
      "grad_norm": 0.4609375,
      "learning_rate": 8.598130841121496e-06,
      "loss": 1.9389,
      "step": 31
    },
    {
      "epoch": 0.14953271028037382,
      "grad_norm": 0.44921875,
      "learning_rate": 8.55140186915888e-06,
      "loss": 1.8883,
      "step": 32
    },
    {
      "epoch": 0.1542056074766355,
      "grad_norm": 0.44921875,
      "learning_rate": 8.504672897196263e-06,
      "loss": 1.8913,
      "step": 33
    },
    {
      "epoch": 0.1588785046728972,
      "grad_norm": 0.431640625,
      "learning_rate": 8.457943925233646e-06,
      "loss": 1.8813,
      "step": 34
    },
    {
      "epoch": 0.16355140186915887,
      "grad_norm": 0.44140625,
      "learning_rate": 8.411214953271028e-06,
      "loss": 1.8649,
      "step": 35
    },
    {
      "epoch": 0.16822429906542055,
      "grad_norm": 0.427734375,
      "learning_rate": 8.364485981308411e-06,
      "loss": 1.8555,
      "step": 36
    },
    {
      "epoch": 0.17289719626168223,
      "grad_norm": 0.4140625,
      "learning_rate": 8.317757009345795e-06,
      "loss": 1.8295,
      "step": 37
    },
    {
      "epoch": 0.17757009345794392,
      "grad_norm": 0.40234375,
      "learning_rate": 8.271028037383178e-06,
      "loss": 1.8116,
      "step": 38
    },
    {
      "epoch": 0.1822429906542056,
      "grad_norm": 0.40234375,
      "learning_rate": 8.224299065420562e-06,
      "loss": 1.7998,
      "step": 39
    },
    {
      "epoch": 0.18691588785046728,
      "grad_norm": 0.38671875,
      "learning_rate": 8.177570093457945e-06,
      "loss": 1.7818,
      "step": 40
    },
    {
      "epoch": 0.19158878504672897,
      "grad_norm": 0.38671875,
      "learning_rate": 8.130841121495327e-06,
      "loss": 1.7595,
      "step": 41
    },
    {
      "epoch": 0.19626168224299065,
      "grad_norm": 0.38671875,
      "learning_rate": 8.084112149532712e-06,
      "loss": 1.7451,
      "step": 42
    },
    {
      "epoch": 0.20093457943925233,
      "grad_norm": 0.384765625,
      "learning_rate": 8.037383177570094e-06,
      "loss": 1.7456,
      "step": 43
    },
    {
      "epoch": 0.205607476635514,
      "grad_norm": 0.37109375,
      "learning_rate": 7.990654205607477e-06,
      "loss": 1.7068,
      "step": 44
    },
    {
      "epoch": 0.2102803738317757,
      "grad_norm": 0.361328125,
      "learning_rate": 7.94392523364486e-06,
      "loss": 1.7142,
      "step": 45
    },
    {
      "epoch": 0.21495327102803738,
      "grad_norm": 0.361328125,
      "learning_rate": 7.897196261682244e-06,
      "loss": 1.6928,
      "step": 46
    },
    {
      "epoch": 0.21962616822429906,
      "grad_norm": 0.353515625,
      "learning_rate": 7.850467289719627e-06,
      "loss": 1.665,
      "step": 47
    },
    {
      "epoch": 0.22429906542056074,
      "grad_norm": 0.345703125,
      "learning_rate": 7.80373831775701e-06,
      "loss": 1.6624,
      "step": 48
    },
    {
      "epoch": 0.22897196261682243,
      "grad_norm": 0.3359375,
      "learning_rate": 7.757009345794392e-06,
      "loss": 1.6481,
      "step": 49
    },
    {
      "epoch": 0.2336448598130841,
      "grad_norm": 0.34375,
      "learning_rate": 7.710280373831777e-06,
      "loss": 1.6482,
      "step": 50
    },
    {
      "epoch": 0.2383177570093458,
      "grad_norm": 0.333984375,
      "learning_rate": 7.663551401869159e-06,
      "loss": 1.6233,
      "step": 51
    },
    {
      "epoch": 0.24299065420560748,
      "grad_norm": 0.330078125,
      "learning_rate": 7.616822429906543e-06,
      "loss": 1.6163,
      "step": 52
    },
    {
      "epoch": 0.24766355140186916,
      "grad_norm": 0.32421875,
      "learning_rate": 7.570093457943926e-06,
      "loss": 1.5861,
      "step": 53
    },
    {
      "epoch": 0.2523364485981308,
      "grad_norm": 0.326171875,
      "learning_rate": 7.523364485981309e-06,
      "loss": 1.59,
      "step": 54
    },
    {
      "epoch": 0.2570093457943925,
      "grad_norm": 0.31640625,
      "learning_rate": 7.476635514018692e-06,
      "loss": 1.582,
      "step": 55
    },
    {
      "epoch": 0.2616822429906542,
      "grad_norm": 0.318359375,
      "learning_rate": 7.429906542056075e-06,
      "loss": 1.5681,
      "step": 56
    },
    {
      "epoch": 0.26635514018691586,
      "grad_norm": 0.3125,
      "learning_rate": 7.383177570093458e-06,
      "loss": 1.5647,
      "step": 57
    },
    {
      "epoch": 0.27102803738317754,
      "grad_norm": 0.310546875,
      "learning_rate": 7.336448598130842e-06,
      "loss": 1.539,
      "step": 58
    },
    {
      "epoch": 0.2757009345794392,
      "grad_norm": 0.302734375,
      "learning_rate": 7.289719626168225e-06,
      "loss": 1.5371,
      "step": 59
    },
    {
      "epoch": 0.2803738317757009,
      "grad_norm": 0.29296875,
      "learning_rate": 7.242990654205608e-06,
      "loss": 1.5024,
      "step": 60
    },
    {
      "epoch": 0.2850467289719626,
      "grad_norm": 0.27734375,
      "learning_rate": 7.196261682242991e-06,
      "loss": 1.516,
      "step": 61
    },
    {
      "epoch": 0.2897196261682243,
      "grad_norm": 0.271484375,
      "learning_rate": 7.149532710280375e-06,
      "loss": 1.5012,
      "step": 62
    },
    {
      "epoch": 0.29439252336448596,
      "grad_norm": 0.26171875,
      "learning_rate": 7.1028037383177574e-06,
      "loss": 1.4923,
      "step": 63
    },
    {
      "epoch": 0.29906542056074764,
      "grad_norm": 0.265625,
      "learning_rate": 7.056074766355141e-06,
      "loss": 1.4793,
      "step": 64
    },
    {
      "epoch": 0.3037383177570093,
      "grad_norm": 0.25,
      "learning_rate": 7.009345794392523e-06,
      "loss": 1.486,
      "step": 65
    },
    {
      "epoch": 0.308411214953271,
      "grad_norm": 0.255859375,
      "learning_rate": 6.962616822429908e-06,
      "loss": 1.4806,
      "step": 66
    },
    {
      "epoch": 0.3130841121495327,
      "grad_norm": 0.251953125,
      "learning_rate": 6.91588785046729e-06,
      "loss": 1.4681,
      "step": 67
    },
    {
      "epoch": 0.3177570093457944,
      "grad_norm": 0.248046875,
      "learning_rate": 6.869158878504674e-06,
      "loss": 1.4514,
      "step": 68
    },
    {
      "epoch": 0.32242990654205606,
      "grad_norm": 0.244140625,
      "learning_rate": 6.822429906542056e-06,
      "loss": 1.4378,
      "step": 69
    },
    {
      "epoch": 0.32710280373831774,
      "grad_norm": 0.2451171875,
      "learning_rate": 6.77570093457944e-06,
      "loss": 1.4582,
      "step": 70
    },
    {
      "epoch": 0.3317757009345794,
      "grad_norm": 0.244140625,
      "learning_rate": 6.728971962616823e-06,
      "loss": 1.4322,
      "step": 71
    },
    {
      "epoch": 0.3364485981308411,
      "grad_norm": 0.2333984375,
      "learning_rate": 6.682242990654206e-06,
      "loss": 1.4347,
      "step": 72
    },
    {
      "epoch": 0.3411214953271028,
      "grad_norm": 0.2373046875,
      "learning_rate": 6.635514018691589e-06,
      "loss": 1.4228,
      "step": 73
    },
    {
      "epoch": 0.34579439252336447,
      "grad_norm": 0.232421875,
      "learning_rate": 6.588785046728972e-06,
      "loss": 1.4103,
      "step": 74
    },
    {
      "epoch": 0.35046728971962615,
      "grad_norm": 0.224609375,
      "learning_rate": 6.542056074766355e-06,
      "loss": 1.4139,
      "step": 75
    },
    {
      "epoch": 0.35514018691588783,
      "grad_norm": 0.234375,
      "learning_rate": 6.495327102803739e-06,
      "loss": 1.4196,
      "step": 76
    },
    {
      "epoch": 0.3598130841121495,
      "grad_norm": 0.2265625,
      "learning_rate": 6.448598130841122e-06,
      "loss": 1.3918,
      "step": 77
    },
    {
      "epoch": 0.3644859813084112,
      "grad_norm": 0.224609375,
      "learning_rate": 6.401869158878505e-06,
      "loss": 1.4162,
      "step": 78
    },
    {
      "epoch": 0.3691588785046729,
      "grad_norm": 0.2119140625,
      "learning_rate": 6.355140186915888e-06,
      "loss": 1.3968,
      "step": 79
    },
    {
      "epoch": 0.37383177570093457,
      "grad_norm": 0.22265625,
      "learning_rate": 6.308411214953272e-06,
      "loss": 1.3754,
      "step": 80
    },
    {
      "epoch": 0.37850467289719625,
      "grad_norm": 0.2109375,
      "learning_rate": 6.2616822429906544e-06,
      "loss": 1.4086,
      "step": 81
    },
    {
      "epoch": 0.38317757009345793,
      "grad_norm": 0.2138671875,
      "learning_rate": 6.214953271028038e-06,
      "loss": 1.3931,
      "step": 82
    },
    {
      "epoch": 0.3878504672897196,
      "grad_norm": 0.216796875,
      "learning_rate": 6.16822429906542e-06,
      "loss": 1.3829,
      "step": 83
    },
    {
      "epoch": 0.3925233644859813,
      "grad_norm": 0.20703125,
      "learning_rate": 6.121495327102805e-06,
      "loss": 1.3711,
      "step": 84
    },
    {
      "epoch": 0.397196261682243,
      "grad_norm": 0.2060546875,
      "learning_rate": 6.074766355140187e-06,
      "loss": 1.3579,
      "step": 85
    },
    {
      "epoch": 0.40186915887850466,
      "grad_norm": 0.20703125,
      "learning_rate": 6.028037383177571e-06,
      "loss": 1.3727,
      "step": 86
    },
    {
      "epoch": 0.40654205607476634,
      "grad_norm": 0.2060546875,
      "learning_rate": 5.981308411214953e-06,
      "loss": 1.3692,
      "step": 87
    },
    {
      "epoch": 0.411214953271028,
      "grad_norm": 0.2021484375,
      "learning_rate": 5.9345794392523374e-06,
      "loss": 1.3706,
      "step": 88
    },
    {
      "epoch": 0.4158878504672897,
      "grad_norm": 0.1982421875,
      "learning_rate": 5.88785046728972e-06,
      "loss": 1.3446,
      "step": 89
    },
    {
      "epoch": 0.4205607476635514,
      "grad_norm": 0.216796875,
      "learning_rate": 5.841121495327103e-06,
      "loss": 1.3499,
      "step": 90
    },
    {
      "epoch": 0.4252336448598131,
      "grad_norm": 0.2158203125,
      "learning_rate": 5.794392523364486e-06,
      "loss": 1.3261,
      "step": 91
    },
    {
      "epoch": 0.42990654205607476,
      "grad_norm": 0.2099609375,
      "learning_rate": 5.747663551401869e-06,
      "loss": 1.3459,
      "step": 92
    },
    {
      "epoch": 0.43457943925233644,
      "grad_norm": 0.2060546875,
      "learning_rate": 5.700934579439253e-06,
      "loss": 1.3223,
      "step": 93
    },
    {
      "epoch": 0.4392523364485981,
      "grad_norm": 0.2001953125,
      "learning_rate": 5.654205607476636e-06,
      "loss": 1.3515,
      "step": 94
    },
    {
      "epoch": 0.4439252336448598,
      "grad_norm": 0.2021484375,
      "learning_rate": 5.607476635514019e-06,
      "loss": 1.3395,
      "step": 95
    },
    {
      "epoch": 0.4485981308411215,
      "grad_norm": 0.1982421875,
      "learning_rate": 5.560747663551402e-06,
      "loss": 1.3372,
      "step": 96
    },
    {
      "epoch": 0.4532710280373832,
      "grad_norm": 0.2119140625,
      "learning_rate": 5.514018691588785e-06,
      "loss": 1.3435,
      "step": 97
    },
    {
      "epoch": 0.45794392523364486,
      "grad_norm": 0.1953125,
      "learning_rate": 5.467289719626169e-06,
      "loss": 1.3129,
      "step": 98
    },
    {
      "epoch": 0.46261682242990654,
      "grad_norm": 0.2060546875,
      "learning_rate": 5.4205607476635515e-06,
      "loss": 1.3223,
      "step": 99
    },
    {
      "epoch": 0.4672897196261682,
      "grad_norm": 0.1962890625,
      "learning_rate": 5.373831775700935e-06,
      "loss": 1.3134,
      "step": 100
    },
    {
      "epoch": 0.4719626168224299,
      "grad_norm": 0.1943359375,
      "learning_rate": 5.3271028037383174e-06,
      "loss": 1.3067,
      "step": 101
    },
    {
      "epoch": 0.4766355140186916,
      "grad_norm": 0.2119140625,
      "learning_rate": 5.280373831775702e-06,
      "loss": 1.3027,
      "step": 102
    },
    {
      "epoch": 0.48130841121495327,
      "grad_norm": 0.2060546875,
      "learning_rate": 5.233644859813084e-06,
      "loss": 1.3162,
      "step": 103
    },
    {
      "epoch": 0.48598130841121495,
      "grad_norm": 0.2021484375,
      "learning_rate": 5.186915887850468e-06,
      "loss": 1.3141,
      "step": 104
    },
    {
      "epoch": 0.49065420560747663,
      "grad_norm": 0.201171875,
      "learning_rate": 5.14018691588785e-06,
      "loss": 1.3171,
      "step": 105
    },
    {
      "epoch": 0.4953271028037383,
      "grad_norm": 0.2021484375,
      "learning_rate": 5.0934579439252344e-06,
      "loss": 1.3113,
      "step": 106
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1953125,
      "learning_rate": 5.046728971962617e-06,
      "loss": 1.3141,
      "step": 107
    },
    {
      "epoch": 0.5046728971962616,
      "grad_norm": 0.1953125,
      "learning_rate": 5e-06,
      "loss": 1.2936,
      "step": 108
    },
    {
      "epoch": 0.5093457943925234,
      "grad_norm": 0.1982421875,
      "learning_rate": 4.953271028037384e-06,
      "loss": 1.3028,
      "step": 109
    },
    {
      "epoch": 0.514018691588785,
      "grad_norm": 0.19921875,
      "learning_rate": 4.906542056074766e-06,
      "loss": 1.322,
      "step": 110
    },
    {
      "epoch": 0.5186915887850467,
      "grad_norm": 0.205078125,
      "learning_rate": 4.85981308411215e-06,
      "loss": 1.2738,
      "step": 111
    },
    {
      "epoch": 0.5233644859813084,
      "grad_norm": 0.1943359375,
      "learning_rate": 4.813084112149533e-06,
      "loss": 1.2831,
      "step": 112
    },
    {
      "epoch": 0.5280373831775701,
      "grad_norm": 0.2021484375,
      "learning_rate": 4.766355140186917e-06,
      "loss": 1.2763,
      "step": 113
    },
    {
      "epoch": 0.5327102803738317,
      "grad_norm": 0.1884765625,
      "learning_rate": 4.719626168224299e-06,
      "loss": 1.2935,
      "step": 114
    },
    {
      "epoch": 0.5373831775700935,
      "grad_norm": 0.21875,
      "learning_rate": 4.6728971962616825e-06,
      "loss": 1.2767,
      "step": 115
    },
    {
      "epoch": 0.5420560747663551,
      "grad_norm": 0.19921875,
      "learning_rate": 4.626168224299066e-06,
      "loss": 1.2702,
      "step": 116
    },
    {
      "epoch": 0.5467289719626168,
      "grad_norm": 0.2138671875,
      "learning_rate": 4.579439252336449e-06,
      "loss": 1.3022,
      "step": 117
    },
    {
      "epoch": 0.5514018691588785,
      "grad_norm": 0.2236328125,
      "learning_rate": 4.532710280373832e-06,
      "loss": 1.2769,
      "step": 118
    },
    {
      "epoch": 0.5560747663551402,
      "grad_norm": 0.19921875,
      "learning_rate": 4.485981308411215e-06,
      "loss": 1.2875,
      "step": 119
    },
    {
      "epoch": 0.5607476635514018,
      "grad_norm": 0.1923828125,
      "learning_rate": 4.439252336448599e-06,
      "loss": 1.2648,
      "step": 120
    },
    {
      "epoch": 0.5654205607476636,
      "grad_norm": 0.208984375,
      "learning_rate": 4.392523364485981e-06,
      "loss": 1.2776,
      "step": 121
    },
    {
      "epoch": 0.5700934579439252,
      "grad_norm": 0.2041015625,
      "learning_rate": 4.345794392523365e-06,
      "loss": 1.2699,
      "step": 122
    },
    {
      "epoch": 0.5747663551401869,
      "grad_norm": 0.1904296875,
      "learning_rate": 4.299065420560748e-06,
      "loss": 1.282,
      "step": 123
    },
    {
      "epoch": 0.5794392523364486,
      "grad_norm": 0.1962890625,
      "learning_rate": 4.2523364485981315e-06,
      "loss": 1.275,
      "step": 124
    },
    {
      "epoch": 0.5841121495327103,
      "grad_norm": 0.19921875,
      "learning_rate": 4.205607476635514e-06,
      "loss": 1.2507,
      "step": 125
    },
    {
      "epoch": 0.5887850467289719,
      "grad_norm": 0.2216796875,
      "learning_rate": 4.1588785046728974e-06,
      "loss": 1.2808,
      "step": 126
    },
    {
      "epoch": 0.5934579439252337,
      "grad_norm": 0.2109375,
      "learning_rate": 4.112149532710281e-06,
      "loss": 1.2779,
      "step": 127
    },
    {
      "epoch": 0.5981308411214953,
      "grad_norm": 0.2021484375,
      "learning_rate": 4.065420560747663e-06,
      "loss": 1.2905,
      "step": 128
    },
    {
      "epoch": 0.602803738317757,
      "grad_norm": 0.197265625,
      "learning_rate": 4.018691588785047e-06,
      "loss": 1.2584,
      "step": 129
    },
    {
      "epoch": 0.6074766355140186,
      "grad_norm": 0.19140625,
      "learning_rate": 3.97196261682243e-06,
      "loss": 1.2554,
      "step": 130
    },
    {
      "epoch": 0.6121495327102804,
      "grad_norm": 0.205078125,
      "learning_rate": 3.925233644859814e-06,
      "loss": 1.2852,
      "step": 131
    },
    {
      "epoch": 0.616822429906542,
      "grad_norm": 0.1982421875,
      "learning_rate": 3.878504672897196e-06,
      "loss": 1.263,
      "step": 132
    },
    {
      "epoch": 0.6214953271028038,
      "grad_norm": 0.193359375,
      "learning_rate": 3.8317757009345796e-06,
      "loss": 1.2651,
      "step": 133
    },
    {
      "epoch": 0.6261682242990654,
      "grad_norm": 0.2001953125,
      "learning_rate": 3.785046728971963e-06,
      "loss": 1.2573,
      "step": 134
    },
    {
      "epoch": 0.6308411214953271,
      "grad_norm": 0.1865234375,
      "learning_rate": 3.738317757009346e-06,
      "loss": 1.2498,
      "step": 135
    },
    {
      "epoch": 0.6355140186915887,
      "grad_norm": 0.193359375,
      "learning_rate": 3.691588785046729e-06,
      "loss": 1.2546,
      "step": 136
    },
    {
      "epoch": 0.6401869158878505,
      "grad_norm": 0.1982421875,
      "learning_rate": 3.6448598130841123e-06,
      "loss": 1.2684,
      "step": 137
    },
    {
      "epoch": 0.6448598130841121,
      "grad_norm": 0.2119140625,
      "learning_rate": 3.5981308411214953e-06,
      "loss": 1.2763,
      "step": 138
    },
    {
      "epoch": 0.6495327102803738,
      "grad_norm": 0.1962890625,
      "learning_rate": 3.5514018691588787e-06,
      "loss": 1.2619,
      "step": 139
    },
    {
      "epoch": 0.6542056074766355,
      "grad_norm": 0.2041015625,
      "learning_rate": 3.5046728971962617e-06,
      "loss": 1.2547,
      "step": 140
    },
    {
      "epoch": 0.6588785046728972,
      "grad_norm": 0.2041015625,
      "learning_rate": 3.457943925233645e-06,
      "loss": 1.2526,
      "step": 141
    },
    {
      "epoch": 0.6635514018691588,
      "grad_norm": 0.2060546875,
      "learning_rate": 3.411214953271028e-06,
      "loss": 1.248,
      "step": 142
    },
    {
      "epoch": 0.6682242990654206,
      "grad_norm": 0.2177734375,
      "learning_rate": 3.3644859813084115e-06,
      "loss": 1.2441,
      "step": 143
    },
    {
      "epoch": 0.6728971962616822,
      "grad_norm": 0.1943359375,
      "learning_rate": 3.3177570093457945e-06,
      "loss": 1.2472,
      "step": 144
    },
    {
      "epoch": 0.677570093457944,
      "grad_norm": 0.19921875,
      "learning_rate": 3.2710280373831774e-06,
      "loss": 1.2524,
      "step": 145
    },
    {
      "epoch": 0.6822429906542056,
      "grad_norm": 0.2021484375,
      "learning_rate": 3.224299065420561e-06,
      "loss": 1.2423,
      "step": 146
    },
    {
      "epoch": 0.6869158878504673,
      "grad_norm": 0.2216796875,
      "learning_rate": 3.177570093457944e-06,
      "loss": 1.2397,
      "step": 147
    },
    {
      "epoch": 0.6915887850467289,
      "grad_norm": 0.205078125,
      "learning_rate": 3.1308411214953272e-06,
      "loss": 1.2681,
      "step": 148
    },
    {
      "epoch": 0.6962616822429907,
      "grad_norm": 0.2041015625,
      "learning_rate": 3.08411214953271e-06,
      "loss": 1.2502,
      "step": 149
    },
    {
      "epoch": 0.7009345794392523,
      "grad_norm": 0.1884765625,
      "learning_rate": 3.0373831775700936e-06,
      "loss": 1.2574,
      "step": 150
    },
    {
      "epoch": 0.705607476635514,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.9906542056074766e-06,
      "loss": 1.2453,
      "step": 151
    },
    {
      "epoch": 0.7102803738317757,
      "grad_norm": 0.1953125,
      "learning_rate": 2.94392523364486e-06,
      "loss": 1.2515,
      "step": 152
    },
    {
      "epoch": 0.7149532710280374,
      "grad_norm": 0.19921875,
      "learning_rate": 2.897196261682243e-06,
      "loss": 1.2589,
      "step": 153
    },
    {
      "epoch": 0.719626168224299,
      "grad_norm": 0.2080078125,
      "learning_rate": 2.8504672897196264e-06,
      "loss": 1.2567,
      "step": 154
    },
    {
      "epoch": 0.7242990654205608,
      "grad_norm": 0.2265625,
      "learning_rate": 2.8037383177570094e-06,
      "loss": 1.2359,
      "step": 155
    },
    {
      "epoch": 0.7289719626168224,
      "grad_norm": 0.2158203125,
      "learning_rate": 2.7570093457943923e-06,
      "loss": 1.2416,
      "step": 156
    },
    {
      "epoch": 0.7336448598130841,
      "grad_norm": 0.193359375,
      "learning_rate": 2.7102803738317757e-06,
      "loss": 1.2358,
      "step": 157
    },
    {
      "epoch": 0.7383177570093458,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.6635514018691587e-06,
      "loss": 1.2224,
      "step": 158
    },
    {
      "epoch": 0.7429906542056075,
      "grad_norm": 0.2099609375,
      "learning_rate": 2.616822429906542e-06,
      "loss": 1.2458,
      "step": 159
    },
    {
      "epoch": 0.7476635514018691,
      "grad_norm": 0.193359375,
      "learning_rate": 2.570093457943925e-06,
      "loss": 1.2339,
      "step": 160
    },
    {
      "epoch": 0.7523364485981309,
      "grad_norm": 0.1953125,
      "learning_rate": 2.5233644859813085e-06,
      "loss": 1.2271,
      "step": 161
    },
    {
      "epoch": 0.7570093457943925,
      "grad_norm": 0.2060546875,
      "learning_rate": 2.476635514018692e-06,
      "loss": 1.2401,
      "step": 162
    },
    {
      "epoch": 0.7616822429906542,
      "grad_norm": 0.2060546875,
      "learning_rate": 2.429906542056075e-06,
      "loss": 1.2336,
      "step": 163
    },
    {
      "epoch": 0.7663551401869159,
      "grad_norm": 0.2099609375,
      "learning_rate": 2.3831775700934583e-06,
      "loss": 1.2179,
      "step": 164
    },
    {
      "epoch": 0.7710280373831776,
      "grad_norm": 0.2109375,
      "learning_rate": 2.3364485981308413e-06,
      "loss": 1.2422,
      "step": 165
    },
    {
      "epoch": 0.7757009345794392,
      "grad_norm": 0.21875,
      "learning_rate": 2.2897196261682247e-06,
      "loss": 1.2491,
      "step": 166
    },
    {
      "epoch": 0.780373831775701,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.2429906542056077e-06,
      "loss": 1.2234,
      "step": 167
    },
    {
      "epoch": 0.7850467289719626,
      "grad_norm": 0.1923828125,
      "learning_rate": 2.1962616822429906e-06,
      "loss": 1.225,
      "step": 168
    },
    {
      "epoch": 0.7897196261682243,
      "grad_norm": 0.19140625,
      "learning_rate": 2.149532710280374e-06,
      "loss": 1.2219,
      "step": 169
    },
    {
      "epoch": 0.794392523364486,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.102803738317757e-06,
      "loss": 1.2187,
      "step": 170
    },
    {
      "epoch": 0.7990654205607477,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.0560747663551404e-06,
      "loss": 1.2221,
      "step": 171
    },
    {
      "epoch": 0.8037383177570093,
      "grad_norm": 0.2060546875,
      "learning_rate": 2.0093457943925234e-06,
      "loss": 1.2361,
      "step": 172
    },
    {
      "epoch": 0.8084112149532711,
      "grad_norm": 0.185546875,
      "learning_rate": 1.962616822429907e-06,
      "loss": 1.2293,
      "step": 173
    },
    {
      "epoch": 0.8130841121495327,
      "grad_norm": 0.2138671875,
      "learning_rate": 1.9158878504672898e-06,
      "loss": 1.2535,
      "step": 174
    },
    {
      "epoch": 0.8177570093457944,
      "grad_norm": 0.19921875,
      "learning_rate": 1.869158878504673e-06,
      "loss": 1.2332,
      "step": 175
    },
    {
      "epoch": 0.822429906542056,
      "grad_norm": 0.19921875,
      "learning_rate": 1.8224299065420562e-06,
      "loss": 1.2292,
      "step": 176
    },
    {
      "epoch": 0.8271028037383178,
      "grad_norm": 0.1962890625,
      "learning_rate": 1.7757009345794394e-06,
      "loss": 1.2325,
      "step": 177
    },
    {
      "epoch": 0.8317757009345794,
      "grad_norm": 0.2216796875,
      "learning_rate": 1.7289719626168225e-06,
      "loss": 1.2314,
      "step": 178
    },
    {
      "epoch": 0.8364485981308412,
      "grad_norm": 0.197265625,
      "learning_rate": 1.6822429906542057e-06,
      "loss": 1.2482,
      "step": 179
    },
    {
      "epoch": 0.8411214953271028,
      "grad_norm": 0.2109375,
      "learning_rate": 1.6355140186915887e-06,
      "loss": 1.2413,
      "step": 180
    },
    {
      "epoch": 0.8457943925233645,
      "grad_norm": 0.1923828125,
      "learning_rate": 1.588785046728972e-06,
      "loss": 1.2187,
      "step": 181
    },
    {
      "epoch": 0.8504672897196262,
      "grad_norm": 0.240234375,
      "learning_rate": 1.542056074766355e-06,
      "loss": 1.2416,
      "step": 182
    },
    {
      "epoch": 0.8551401869158879,
      "grad_norm": 0.1962890625,
      "learning_rate": 1.4953271028037383e-06,
      "loss": 1.2207,
      "step": 183
    },
    {
      "epoch": 0.8598130841121495,
      "grad_norm": 0.1962890625,
      "learning_rate": 1.4485981308411215e-06,
      "loss": 1.2447,
      "step": 184
    },
    {
      "epoch": 0.8644859813084113,
      "grad_norm": 0.21484375,
      "learning_rate": 1.4018691588785047e-06,
      "loss": 1.2384,
      "step": 185
    },
    {
      "epoch": 0.8691588785046729,
      "grad_norm": 0.197265625,
      "learning_rate": 1.3551401869158879e-06,
      "loss": 1.2225,
      "step": 186
    },
    {
      "epoch": 0.8738317757009346,
      "grad_norm": 0.1943359375,
      "learning_rate": 1.308411214953271e-06,
      "loss": 1.2498,
      "step": 187
    },
    {
      "epoch": 0.8785046728971962,
      "grad_norm": 0.193359375,
      "learning_rate": 1.2616822429906543e-06,
      "loss": 1.2368,
      "step": 188
    },
    {
      "epoch": 0.883177570093458,
      "grad_norm": 0.189453125,
      "learning_rate": 1.2149532710280374e-06,
      "loss": 1.2277,
      "step": 189
    },
    {
      "epoch": 0.8878504672897196,
      "grad_norm": 0.1875,
      "learning_rate": 1.1682242990654206e-06,
      "loss": 1.2207,
      "step": 190
    },
    {
      "epoch": 0.8925233644859814,
      "grad_norm": 0.2333984375,
      "learning_rate": 1.1214953271028038e-06,
      "loss": 1.229,
      "step": 191
    },
    {
      "epoch": 0.897196261682243,
      "grad_norm": 0.2041015625,
      "learning_rate": 1.074766355140187e-06,
      "loss": 1.2369,
      "step": 192
    },
    {
      "epoch": 0.9018691588785047,
      "grad_norm": 0.208984375,
      "learning_rate": 1.0280373831775702e-06,
      "loss": 1.2409,
      "step": 193
    },
    {
      "epoch": 0.9065420560747663,
      "grad_norm": 0.1982421875,
      "learning_rate": 9.813084112149534e-07,
      "loss": 1.2302,
      "step": 194
    },
    {
      "epoch": 0.9112149532710281,
      "grad_norm": 0.203125,
      "learning_rate": 9.345794392523365e-07,
      "loss": 1.2332,
      "step": 195
    },
    {
      "epoch": 0.9158878504672897,
      "grad_norm": 0.22265625,
      "learning_rate": 8.878504672897197e-07,
      "loss": 1.2479,
      "step": 196
    },
    {
      "epoch": 0.9205607476635514,
      "grad_norm": 0.19921875,
      "learning_rate": 8.411214953271029e-07,
      "loss": 1.2266,
      "step": 197
    },
    {
      "epoch": 0.9252336448598131,
      "grad_norm": 0.1943359375,
      "learning_rate": 7.94392523364486e-07,
      "loss": 1.2167,
      "step": 198
    },
    {
      "epoch": 0.9299065420560748,
      "grad_norm": 0.220703125,
      "learning_rate": 7.476635514018691e-07,
      "loss": 1.216,
      "step": 199
    },
    {
      "epoch": 0.9345794392523364,
      "grad_norm": 0.2109375,
      "learning_rate": 7.009345794392523e-07,
      "loss": 1.2391,
      "step": 200
    },
    {
      "epoch": 0.9392523364485982,
      "grad_norm": 0.20703125,
      "learning_rate": 6.542056074766355e-07,
      "loss": 1.2361,
      "step": 201
    },
    {
      "epoch": 0.9439252336448598,
      "grad_norm": 0.197265625,
      "learning_rate": 6.074766355140187e-07,
      "loss": 1.2316,
      "step": 202
    },
    {
      "epoch": 0.9485981308411215,
      "grad_norm": 0.1982421875,
      "learning_rate": 5.607476635514019e-07,
      "loss": 1.2201,
      "step": 203
    },
    {
      "epoch": 0.9532710280373832,
      "grad_norm": 0.203125,
      "learning_rate": 5.140186915887851e-07,
      "loss": 1.221,
      "step": 204
    },
    {
      "epoch": 0.9579439252336449,
      "grad_norm": 0.2060546875,
      "learning_rate": 4.6728971962616824e-07,
      "loss": 1.2392,
      "step": 205
    },
    {
      "epoch": 0.9626168224299065,
      "grad_norm": 0.18359375,
      "learning_rate": 4.2056074766355143e-07,
      "loss": 1.2295,
      "step": 206
    },
    {
      "epoch": 0.9672897196261683,
      "grad_norm": 0.2021484375,
      "learning_rate": 3.7383177570093457e-07,
      "loss": 1.2338,
      "step": 207
    },
    {
      "epoch": 0.9719626168224299,
      "grad_norm": 0.1953125,
      "learning_rate": 3.2710280373831776e-07,
      "loss": 1.224,
      "step": 208
    },
    {
      "epoch": 0.9766355140186916,
      "grad_norm": 0.2109375,
      "learning_rate": 2.8037383177570096e-07,
      "loss": 1.2317,
      "step": 209
    },
    {
      "epoch": 0.9813084112149533,
      "grad_norm": 0.193359375,
      "learning_rate": 2.3364485981308412e-07,
      "loss": 1.2235,
      "step": 210
    },
    {
      "epoch": 0.985981308411215,
      "grad_norm": 0.20703125,
      "learning_rate": 1.8691588785046729e-07,
      "loss": 1.2238,
      "step": 211
    },
    {
      "epoch": 0.9906542056074766,
      "grad_norm": 0.2080078125,
      "learning_rate": 1.4018691588785048e-07,
      "loss": 1.2146,
      "step": 212
    },
    {
      "epoch": 0.9953271028037384,
      "grad_norm": 0.203125,
      "learning_rate": 9.345794392523364e-08,
      "loss": 1.231,
      "step": 213
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2099609375,
      "learning_rate": 4.672897196261682e-08,
      "loss": 1.2306,
      "step": 214
    },
    {
      "epoch": 1.0,
      "eval_loss": 1.2332667112350464,
      "eval_runtime": 7.4216,
      "eval_samples_per_second": 3.099,
      "eval_steps_per_second": 0.404,
      "step": 214
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 214,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 0,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 7.321152639100518e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}