{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 432,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 7.692307692307694e-07,
      "loss": 2.065,
      "step": 1
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.5384615384615387e-06,
      "loss": 2.0734,
      "step": 2
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.307692307692308e-06,
      "loss": 2.0549,
      "step": 3
    },
    {
      "epoch": 0.03,
      "learning_rate": 3.0769230769230774e-06,
      "loss": 2.0151,
      "step": 4
    },
    {
      "epoch": 0.03,
      "learning_rate": 3.846153846153847e-06,
      "loss": 1.997,
      "step": 5
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.615384615384616e-06,
      "loss": 1.9595,
      "step": 6
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.384615384615385e-06,
      "loss": 1.9403,
      "step": 7
    },
    {
      "epoch": 0.06,
      "learning_rate": 6.153846153846155e-06,
      "loss": 1.9314,
      "step": 8
    },
    {
      "epoch": 0.06,
      "learning_rate": 6.923076923076923e-06,
      "loss": 1.9198,
      "step": 9
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.692307692307694e-06,
      "loss": 1.9514,
      "step": 10
    },
    {
      "epoch": 0.08,
      "learning_rate": 8.461538461538462e-06,
      "loss": 1.9133,
      "step": 11
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.230769230769232e-06,
      "loss": 1.9148,
      "step": 12
    },
    {
      "epoch": 0.09,
      "learning_rate": 1e-05,
      "loss": 1.896,
      "step": 13
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.999859456853116e-06,
      "loss": 1.8821,
      "step": 14
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.99943783531341e-06,
      "loss": 1.8483,
      "step": 15
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.998735159083295e-06,
      "loss": 1.8198,
      "step": 16
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.997751467665295e-06,
      "loss": 1.8479,
      "step": 17
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.996486816359851e-06,
      "loss": 1.8472,
      "step": 18
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.994941276262188e-06,
      "loss": 1.8071,
      "step": 19
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.99311493425834e-06,
      "loss": 1.8517,
      "step": 20
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.991007893020242e-06,
      "loss": 1.816,
      "step": 21
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.98862027099998e-06,
      "loss": 1.8063,
      "step": 22
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.985952202423116e-06,
      "loss": 1.8017,
      "step": 23
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.983003837281152e-06,
      "loss": 1.8223,
      "step": 24
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.979775341323097e-06,
      "loss": 1.8145,
      "step": 25
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.976266896046143e-06,
      "loss": 1.8268,
      "step": 26
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.972478698685463e-06,
      "loss": 1.7692,
      "step": 27
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.968410962203131e-06,
      "loss": 1.773,
      "step": 28
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.964063915276141e-06,
      "loss": 1.7792,
      "step": 29
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.959437802283552e-06,
      "loss": 1.7774,
      "step": 30
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.954532883292761e-06,
      "loss": 1.7651,
      "step": 31
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.949349434044862e-06,
      "loss": 1.8011,
      "step": 32
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.943887745939164e-06,
      "loss": 1.7864,
      "step": 33
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.938148126016805e-06,
      "loss": 1.7789,
      "step": 34
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.932130896943477e-06,
      "loss": 1.7514,
      "step": 35
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.925836396991309e-06,
      "loss": 1.7729,
      "step": 36
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.919264980019829e-06,
      "loss": 1.7676,
      "step": 37
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.912417015456088e-06,
      "loss": 1.7933,
      "step": 38
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.905292888273883e-06,
      "loss": 1.7307,
      "step": 39
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.897892998972113e-06,
      "loss": 1.7404,
      "step": 40
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.89021776355227e-06,
      "loss": 1.7741,
      "step": 41
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.882267613495049e-06,
      "loss": 1.7601,
      "step": 42
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.874042995736095e-06,
      "loss": 1.7748,
      "step": 43
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.865544372640872e-06,
      "loss": 1.7497,
      "step": 44
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.85677222197867e-06,
      "loss": 1.7459,
      "step": 45
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.847727036895759e-06,
      "loss": 1.7574,
      "step": 46
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.838409325887643e-06,
      "loss": 1.7503,
      "step": 47
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.828819612770497e-06,
      "loss": 1.7564,
      "step": 48
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.818958436651704e-06,
      "loss": 1.7693,
      "step": 49
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.808826351899551e-06,
      "loss": 1.7723,
      "step": 50
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.798423928112071e-06,
      "loss": 1.7767,
      "step": 51
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.787751750085014e-06,
      "loss": 1.7676,
      "step": 52
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.77681041777897e-06,
      "loss": 1.7143,
      "step": 53
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.765600546285654e-06,
      "loss": 1.7606,
      "step": 54
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.754122765793306e-06,
      "loss": 1.7771,
      "step": 55
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.742377721551286e-06,
      "loss": 1.7835,
      "step": 56
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.730366073833785e-06,
      "loss": 1.7591,
      "step": 57
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.718088497902709e-06,
      "loss": 1.7789,
      "step": 58
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.705545683969722e-06,
      "loss": 1.7526,
      "step": 59
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.692738337157441e-06,
      "loss": 1.7449,
      "step": 60
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.679667177459794e-06,
      "loss": 1.7617,
      "step": 61
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.66633293970155e-06,
      "loss": 1.7585,
      "step": 62
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.652736373497001e-06,
      "loss": 1.764,
      "step": 63
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.63887824320783e-06,
      "loss": 1.7151,
      "step": 64
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.624759327900131e-06,
      "loss": 1.7299,
      "step": 65
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.610380421300623e-06,
      "loss": 1.7642,
      "step": 66
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.595742331752014e-06,
      "loss": 1.7522,
      "step": 67
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.580845882167574e-06,
      "loss": 1.7303,
      "step": 68
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.565691909984864e-06,
      "loss": 1.7423,
      "step": 69
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.550281267118659e-06,
      "loss": 1.7434,
      "step": 70
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.534614819913056e-06,
      "loss": 1.6931,
      "step": 71
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.518693449092772e-06,
      "loss": 1.7438,
      "step": 72
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.502518049713633e-06,
      "loss": 1.6983,
      "step": 73
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.486089531112247e-06,
      "loss": 1.7422,
      "step": 74
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.469408816854898e-06,
      "loss": 1.7321,
      "step": 75
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.452476844685611e-06,
      "loss": 1.7252,
      "step": 76
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.435294566473453e-06,
      "loss": 1.6962,
      "step": 77
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.417862948158997e-06,
      "loss": 1.7193,
      "step": 78
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.40018296970005e-06,
      "loss": 1.697,
      "step": 79
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.382255625016527e-06,
      "loss": 1.7596,
      "step": 80
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.364081921934607e-06,
      "loss": 1.7381,
      "step": 81
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.345662882130056e-06,
      "loss": 1.7395,
      "step": 82
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.326999541070804e-06,
      "loss": 1.7373,
      "step": 83
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.308092947958725e-06,
      "loss": 1.7659,
      "step": 84
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.288944165670651e-06,
      "loss": 1.7416,
      "step": 85
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.269554270698636e-06,
      "loss": 1.7444,
      "step": 86
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.24992435308942e-06,
      "loss": 1.7693,
      "step": 87
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.23005551638316e-06,
      "loss": 1.7481,
      "step": 88
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.209948877551393e-06,
      "loss": 1.7063,
      "step": 89
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.189605566934235e-06,
      "loss": 1.7348,
      "step": 90
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.169026728176845e-06,
      "loss": 1.756,
      "step": 91
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.148213518165121e-06,
      "loss": 1.7177,
      "step": 92
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.127167106960682e-06,
      "loss": 1.7138,
      "step": 93
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.105888677735069e-06,
      "loss": 1.7358,
      "step": 94
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.084379426703245e-06,
      "loss": 1.7359,
      "step": 95
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.062640563056339e-06,
      "loss": 1.6954,
      "step": 96
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.040673308893677e-06,
      "loss": 1.7139,
      "step": 97
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.018478899154068e-06,
      "loss": 1.7276,
      "step": 98
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.996058581546386e-06,
      "loss": 1.7083,
      "step": 99
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.973413616479429e-06,
      "loss": 1.6832,
      "step": 100
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.95054527699106e-06,
      "loss": 1.7406,
      "step": 101
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.927454848676633e-06,
      "loss": 1.7201,
      "step": 102
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.904143629616735e-06,
      "loss": 1.7369,
      "step": 103
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.880612930304196e-06,
      "loss": 1.6823,
      "step": 104
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.856864073570429e-06,
      "loss": 1.7197,
      "step": 105
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.83289839451106e-06,
      "loss": 1.6927,
      "step": 106
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.80871724041087e-06,
      "loss": 1.7568,
      "step": 107
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.784321970668054e-06,
      "loss": 1.7227,
      "step": 108
    },
    {
      "epoch": 0.76,
      "learning_rate": 8.759713956717804e-06,
      "loss": 1.7387,
      "step": 109
    },
    {
      "epoch": 0.76,
      "learning_rate": 8.734894581955208e-06,
      "loss": 1.7293,
      "step": 110
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.70986524165748e-06,
      "loss": 1.7002,
      "step": 111
    },
    {
      "epoch": 0.78,
      "learning_rate": 8.684627342905519e-06,
      "loss": 1.715,
      "step": 112
    },
    {
      "epoch": 0.78,
      "learning_rate": 8.65918230450481e-06,
      "loss": 1.7531,
      "step": 113
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.63353155690566e-06,
      "loss": 1.6997,
      "step": 114
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.607676542122782e-06,
      "loss": 1.7396,
      "step": 115
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.581618713654239e-06,
      "loss": 1.7285,
      "step": 116
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.55535953639971e-06,
      "loss": 1.741,
      "step": 117
    },
    {
      "epoch": 0.82,
      "learning_rate": 8.528900486578158e-06,
      "loss": 1.7045,
      "step": 118
    },
    {
      "epoch": 0.83,
      "learning_rate": 8.502243051644838e-06,
      "loss": 1.6933,
      "step": 119
    },
    {
      "epoch": 0.83,
      "learning_rate": 8.475388730207662e-06,
      "loss": 1.7161,
      "step": 120
    },
    {
      "epoch": 0.84,
      "learning_rate": 8.44833903194297e-06,
      "loss": 1.7186,
      "step": 121
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.421095477510648e-06,
      "loss": 1.706,
      "step": 122
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.393659598468644e-06,
      "loss": 1.7607,
      "step": 123
    },
    {
      "epoch": 0.86,
      "learning_rate": 8.366032937186869e-06,
      "loss": 1.7455,
      "step": 124
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.33821704676049e-06,
      "loss": 1.7347,
      "step": 125
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.310213490922616e-06,
      "loss": 1.6941,
      "step": 126
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.282023843956392e-06,
      "loss": 1.7156,
      "step": 127
    },
    {
      "epoch": 0.89,
      "learning_rate": 8.253649690606495e-06,
      "loss": 1.7068,
      "step": 128
    },
    {
      "epoch": 0.9,
      "learning_rate": 8.225092625990047e-06,
      "loss": 1.731,
      "step": 129
    },
    {
      "epoch": 0.9,
      "learning_rate": 8.196354255506937e-06,
      "loss": 1.6977,
      "step": 130
    },
    {
      "epoch": 0.91,
      "learning_rate": 8.167436194749576e-06,
      "loss": 1.7639,
      "step": 131
    },
    {
      "epoch": 0.92,
      "learning_rate": 8.138340069412069e-06,
      "loss": 1.7265,
      "step": 132
    },
    {
      "epoch": 0.92,
      "learning_rate": 8.109067515198822e-06,
      "loss": 1.7195,
      "step": 133
    },
    {
      "epoch": 0.93,
      "learning_rate": 8.079620177732587e-06,
      "loss": 1.7243,
      "step": 134
    },
    {
      "epoch": 0.94,
      "learning_rate": 8.049999712461956e-06,
      "loss": 1.7081,
      "step": 135
    },
    {
      "epoch": 0.94,
      "learning_rate": 8.020207784568293e-06,
      "loss": 1.7023,
      "step": 136
    },
    {
      "epoch": 0.95,
      "learning_rate": 7.990246068872111e-06,
      "loss": 1.7047,
      "step": 137
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.960116249738939e-06,
      "loss": 1.688,
      "step": 138
    },
    {
      "epoch": 0.97,
      "learning_rate": 7.92982002098461e-06,
      "loss": 1.7059,
      "step": 139
    },
    {
      "epoch": 0.97,
      "learning_rate": 7.899359085780062e-06,
      "loss": 1.6971,
      "step": 140
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.868735156555567e-06,
      "loss": 1.6947,
      "step": 141
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.83794995490448e-06,
      "loss": 1.7479,
      "step": 142
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.807005211486445e-06,
      "loss": 1.7464,
      "step": 143
    },
    {
      "epoch": 1.0,
      "learning_rate": 7.775902665930114e-06,
      "loss": 1.7386,
      "step": 144
    },
    {
      "epoch": 1.01,
      "learning_rate": 7.744644066735335e-06,
      "loss": 1.6369,
      "step": 145
    },
    {
      "epoch": 1.01,
      "learning_rate": 7.713231171174868e-06,
      "loss": 1.6396,
      "step": 146
    },
    {
      "epoch": 1.02,
      "learning_rate": 7.681665745195593e-06,
      "loss": 1.6174,
      "step": 147
    },
    {
      "epoch": 1.03,
      "learning_rate": 7.649949563319228e-06,
      "loss": 1.5872,
      "step": 148
    },
    {
      "epoch": 1.03,
      "learning_rate": 7.618084408542576e-06,
      "loss": 1.5716,
      "step": 149
    },
    {
      "epoch": 1.04,
      "learning_rate": 7.586072072237291e-06,
      "loss": 1.5423,
      "step": 150
    },
    {
      "epoch": 1.05,
      "learning_rate": 7.5539143540491635e-06,
      "loss": 1.5086,
      "step": 151
    },
    {
      "epoch": 1.06,
      "learning_rate": 7.521613061796957e-06,
      "loss": 1.5367,
      "step": 152
    },
    {
      "epoch": 1.06,
      "learning_rate": 7.48917001137078e-06,
      "loss": 1.5399,
      "step": 153
    },
    {
      "epoch": 1.07,
      "learning_rate": 7.456587026629991e-06,
      "loss": 1.594,
      "step": 154
    },
    {
      "epoch": 1.08,
      "learning_rate": 7.423865939300674e-06,
      "loss": 1.5454,
      "step": 155
    },
    {
      "epoch": 1.08,
      "learning_rate": 7.391008588872661e-06,
      "loss": 1.5158,
      "step": 156
    },
    {
      "epoch": 1.09,
      "learning_rate": 7.358016822496126e-06,
      "loss": 1.4813,
      "step": 157
    },
    {
      "epoch": 1.1,
      "learning_rate": 7.324892494877734e-06,
      "loss": 1.4816,
      "step": 158
    },
    {
      "epoch": 1.1,
      "learning_rate": 7.29163746817638e-06,
      "loss": 1.4641,
      "step": 159
    },
    {
      "epoch": 1.11,
      "learning_rate": 7.258253611898509e-06,
      "loss": 1.374,
      "step": 160
    },
    {
      "epoch": 1.12,
      "learning_rate": 7.224742802793005e-06,
      "loss": 1.4205,
      "step": 161
    },
    {
      "epoch": 1.12,
      "learning_rate": 7.191106924745695e-06,
      "loss": 1.4606,
      "step": 162
    },
    {
      "epoch": 1.13,
      "learning_rate": 7.157347868673441e-06,
      "loss": 1.4101,
      "step": 163
    },
    {
      "epoch": 1.14,
      "learning_rate": 7.1234675324178295e-06,
      "loss": 1.4522,
      "step": 164
    },
    {
      "epoch": 1.15,
      "learning_rate": 7.089467820638491e-06,
      "loss": 1.4079,
      "step": 165
    },
    {
      "epoch": 1.15,
      "learning_rate": 7.055350644706023e-06,
      "loss": 1.4163,
      "step": 166
    },
    {
      "epoch": 1.16,
      "learning_rate": 7.021117922594532e-06,
      "loss": 1.3857,
      "step": 167
    },
    {
      "epoch": 1.17,
      "learning_rate": 6.986771578773812e-06,
      "loss": 1.4118,
      "step": 168
    },
    {
      "epoch": 1.17,
      "learning_rate": 6.952313544101165e-06,
      "loss": 1.4108,
      "step": 169
    },
    {
      "epoch": 1.18,
      "learning_rate": 6.917745755712839e-06,
      "loss": 1.4318,
      "step": 170
    },
    {
      "epoch": 1.19,
      "learning_rate": 6.8830701569151394e-06,
      "loss": 1.3488,
      "step": 171
    },
    {
      "epoch": 1.19,
      "learning_rate": 6.8482886970751785e-06,
      "loss": 1.3547,
      "step": 172
    },
    {
      "epoch": 1.2,
      "learning_rate": 6.81340333151128e-06,
      "loss": 1.3479,
      "step": 173
    },
    {
      "epoch": 1.21,
      "learning_rate": 6.7784160213830696e-06,
      "loss": 1.3658,
      "step": 174
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.743328733581211e-06,
      "loss": 1.3601,
      "step": 175
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.708143440616845e-06,
      "loss": 1.4016,
      "step": 176
    },
    {
      "epoch": 1.23,
      "learning_rate": 6.672862120510688e-06,
      "loss": 1.3709,
      "step": 177
    },
    {
      "epoch": 1.24,
      "learning_rate": 6.637486756681843e-06,
      "loss": 1.3934,
      "step": 178
    },
    {
      "epoch": 1.24,
      "learning_rate": 6.602019337836291e-06,
      "loss": 1.3734,
      "step": 179
    },
    {
      "epoch": 1.25,
      "learning_rate": 6.566461857855096e-06,
      "loss": 1.3656,
      "step": 180
    },
    {
      "epoch": 1.26,
      "learning_rate": 6.5308163156823064e-06,
      "loss": 1.3864,
      "step": 181
    },
    {
      "epoch": 1.26,
      "learning_rate": 6.495084715212597e-06,
      "loss": 1.4066,
      "step": 182
    },
    {
      "epoch": 1.27,
      "learning_rate": 6.459269065178592e-06,
      "loss": 1.3613,
      "step": 183
    },
    {
      "epoch": 1.28,
      "learning_rate": 6.423371379037957e-06,
      "loss": 1.3644,
      "step": 184
    },
    {
      "epoch": 1.28,
      "learning_rate": 6.387393674860205e-06,
      "loss": 1.4032,
      "step": 185
    },
    {
      "epoch": 1.29,
      "learning_rate": 6.351337975213239e-06,
      "loss": 1.3797,
      "step": 186
    },
    {
      "epoch": 1.3,
      "learning_rate": 6.315206307049656e-06,
      "loss": 1.4222,
      "step": 187
    },
    {
      "epoch": 1.31,
      "learning_rate": 6.2790007015927946e-06,
      "loss": 1.3548,
      "step": 188
    },
    {
      "epoch": 1.31,
      "learning_rate": 6.242723194222546e-06,
      "loss": 1.3757,
      "step": 189
    },
    {
      "epoch": 1.32,
      "learning_rate": 6.2063758243609275e-06,
      "loss": 1.3396,
      "step": 190
    },
    {
      "epoch": 1.33,
      "learning_rate": 6.169960635357437e-06,
      "loss": 1.3686,
      "step": 191
    },
    {
      "epoch": 1.33,
      "learning_rate": 6.133479674374176e-06,
      "loss": 1.3789,
      "step": 192
    },
    {
      "epoch": 1.34,
      "learning_rate": 6.0969349922707675e-06,
      "loss": 1.4203,
      "step": 193
    },
    {
      "epoch": 1.35,
      "learning_rate": 6.060328643489064e-06,
      "loss": 1.3963,
      "step": 194
    },
    {
      "epoch": 1.35,
      "learning_rate": 6.023662685937643e-06,
      "loss": 1.4282,
      "step": 195
    },
    {
      "epoch": 1.36,
      "learning_rate": 5.9869391808761315e-06,
      "loss": 1.3772,
      "step": 196
    },
    {
      "epoch": 1.37,
      "learning_rate": 5.9501601927993135e-06,
      "loss": 1.3101,
      "step": 197
    },
    {
      "epoch": 1.38,
      "learning_rate": 5.9133277893210785e-06,
      "loss": 1.3972,
      "step": 198
    },
    {
      "epoch": 1.38,
      "learning_rate": 5.8764440410581846e-06,
      "loss": 1.4262,
      "step": 199
    },
    {
      "epoch": 1.39,
      "learning_rate": 5.839511021513853e-06,
      "loss": 1.3973,
      "step": 200
    },
    {
      "epoch": 1.4,
      "learning_rate": 5.802530806961195e-06,
      "loss": 1.3769,
      "step": 201
    },
    {
      "epoch": 1.4,
      "learning_rate": 5.765505476326505e-06,
      "loss": 1.4184,
      "step": 202
    },
    {
      "epoch": 1.41,
      "learning_rate": 5.728437111072376e-06,
      "loss": 1.3802,
      "step": 203
    },
    {
      "epoch": 1.42,
      "learning_rate": 5.691327795080685e-06,
      "loss": 1.3814,
      "step": 204
    },
    {
      "epoch": 1.42,
      "learning_rate": 5.654179614535457e-06,
      "loss": 1.393,
      "step": 205
    },
    {
      "epoch": 1.43,
      "learning_rate": 5.616994657805566e-06,
      "loss": 1.3777,
      "step": 206
    },
    {
      "epoch": 1.44,
      "learning_rate": 5.579775015327347e-06,
      "loss": 1.4092,
      "step": 207
    },
    {
      "epoch": 1.44,
      "learning_rate": 5.5425227794870715e-06,
      "loss": 1.3382,
      "step": 208
    },
    {
      "epoch": 1.45,
      "learning_rate": 5.505240044503324e-06,
      "loss": 1.3584,
      "step": 209
    },
    {
      "epoch": 1.46,
      "learning_rate": 5.46792890630926e-06,
      "loss": 1.4156,
      "step": 210
    },
    {
      "epoch": 1.47,
      "learning_rate": 5.430591462434792e-06,
      "loss": 1.37,
      "step": 211
    },
    {
      "epoch": 1.47,
      "learning_rate": 5.393229811888663e-06,
      "loss": 1.36,
      "step": 212
    },
    {
      "epoch": 1.48,
      "learning_rate": 5.355846055040449e-06,
      "loss": 1.4008,
      "step": 213
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.318442293502482e-06,
      "loss": 1.3849,
      "step": 214
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.281020630011703e-06,
      "loss": 1.3328,
      "step": 215
    },
    {
      "epoch": 1.5,
      "learning_rate": 5.2435831683114515e-06,
      "loss": 1.3804,
      "step": 216
    },
    {
      "epoch": 1.51,
      "learning_rate": 5.206132013033199e-06,
      "loss": 1.3161,
      "step": 217
    },
    {
      "epoch": 1.51,
      "learning_rate": 5.1686692695782325e-06,
      "loss": 1.3887,
      "step": 218
    },
    {
      "epoch": 1.52,
      "learning_rate": 5.131197043999294e-06,
      "loss": 1.4188,
      "step": 219
    },
    {
      "epoch": 1.53,
      "learning_rate": 5.093717442882185e-06,
      "loss": 1.3621,
      "step": 220
    },
    {
      "epoch": 1.53,
      "learning_rate": 5.0562325732273405e-06,
      "loss": 1.3492,
      "step": 221
    },
    {
      "epoch": 1.54,
      "learning_rate": 5.018744542331376e-06,
      "loss": 1.3909,
      "step": 222
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.981255457668625e-06,
      "loss": 1.3613,
      "step": 223
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.94376742677266e-06,
      "loss": 1.4051,
      "step": 224
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.906282557117817e-06,
      "loss": 1.385,
      "step": 225
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.8688029560007064e-06,
      "loss": 1.3837,
      "step": 226
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.831330730421769e-06,
      "loss": 1.4267,
      "step": 227
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.793867986966802e-06,
      "loss": 1.3869,
      "step": 228
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.756416831688549e-06,
      "loss": 1.3975,
      "step": 229
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.718979369988299e-06,
      "loss": 1.4298,
      "step": 230
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.681557706497519e-06,
      "loss": 1.4423,
      "step": 231
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.644153944959553e-06,
      "loss": 1.3994,
      "step": 232
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.606770188111339e-06,
      "loss": 1.3517,
      "step": 233
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.5694085375652105e-06,
      "loss": 1.3984,
      "step": 234
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.532071093690741e-06,
      "loss": 1.4148,
      "step": 235
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.494759955496678e-06,
      "loss": 1.3898,
      "step": 236
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.457477220512929e-06,
      "loss": 1.3715,
      "step": 237
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.420224984672654e-06,
      "loss": 1.3819,
      "step": 238
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.383005342194436e-06,
      "loss": 1.4257,
      "step": 239
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.345820385464543e-06,
      "loss": 1.3365,
      "step": 240
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.308672204919316e-06,
      "loss": 1.3555,
      "step": 241
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.271562888927626e-06,
      "loss": 1.3831,
      "step": 242
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.234494523673497e-06,
      "loss": 1.3866,
      "step": 243
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.1974691930388055e-06,
      "loss": 1.3544,
      "step": 244
    },
    {
      "epoch": 1.7,
      "learning_rate": 4.16048897848615e-06,
      "loss": 1.3899,
      "step": 245
    },
    {
      "epoch": 1.71,
      "learning_rate": 4.123555958941817e-06,
      "loss": 1.3751,
      "step": 246
    },
    {
      "epoch": 1.72,
      "learning_rate": 4.0866722106789214e-06,
      "loss": 1.38,
      "step": 247
    },
    {
      "epoch": 1.72,
      "learning_rate": 4.049839807200688e-06,
      "loss": 1.3596,
      "step": 248
    },
    {
      "epoch": 1.73,
      "learning_rate": 4.013060819123869e-06,
      "loss": 1.3698,
      "step": 249
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.976337314062358e-06,
      "loss": 1.3362,
      "step": 250
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.9396713565109375e-06,
      "loss": 1.4218,
      "step": 251
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.903065007729234e-06,
      "loss": 1.3871,
      "step": 252
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.866520325625825e-06,
      "loss": 1.399,
      "step": 253
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.830039364642566e-06,
      "loss": 1.4025,
      "step": 254
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.7936241756390746e-06,
      "loss": 1.3634,
      "step": 255
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.7572768057774543e-06,
      "loss": 1.3397,
      "step": 256
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.7209992984072062e-06,
      "loss": 1.4236,
      "step": 257
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.6847936929503446e-06,
      "loss": 1.387,
      "step": 258
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.6486620247867625e-06,
      "loss": 1.4546,
      "step": 259
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.6126063251397968e-06,
      "loss": 1.4263,
      "step": 260
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.576628620962045e-06,
      "loss": 1.4271,
      "step": 261
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.5407309348214094e-06,
      "loss": 1.3993,
      "step": 262
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.5049152847874053e-06,
      "loss": 1.3685,
      "step": 263
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.469183684317694e-06,
      "loss": 1.3821,
      "step": 264
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.4335381421449056e-06,
      "loss": 1.3968,
      "step": 265
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.39798066216371e-06,
      "loss": 1.3889,
      "step": 266
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.3625132433181573e-06,
      "loss": 1.439,
      "step": 267
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.327137879489313e-06,
      "loss": 1.4536,
      "step": 268
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.2918565593831565e-06,
      "loss": 1.3946,
      "step": 269
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.2566712664187907e-06,
      "loss": 1.3681,
      "step": 270
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.221583978616932e-06,
      "loss": 1.395,
      "step": 271
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.1865966684887222e-06,
      "loss": 1.3861,
      "step": 272
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.1517113029248236e-06,
      "loss": 1.4264,
      "step": 273
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.1169298430848605e-06,
      "loss": 1.372,
      "step": 274
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.082254244287163e-06,
      "loss": 1.4616,
      "step": 275
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.0476864558988364e-06,
      "loss": 1.4242,
      "step": 276
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.0132284212261886e-06,
      "loss": 1.4027,
      "step": 277
    },
    {
      "epoch": 1.93,
      "learning_rate": 2.97888207740547e-06,
      "loss": 1.4085,
      "step": 278
    },
    {
      "epoch": 1.94,
      "learning_rate": 2.944649355293979e-06,
      "loss": 1.3835,
      "step": 279
    },
    {
      "epoch": 1.94,
      "learning_rate": 2.9105321793615106e-06,
      "loss": 1.3953,
      "step": 280
    },
    {
      "epoch": 1.95,
      "learning_rate": 2.876532467582174e-06,
      "loss": 1.4009,
      "step": 281
    },
    {
      "epoch": 1.96,
      "learning_rate": 2.842652131326562e-06,
      "loss": 1.3675,
      "step": 282
    },
    {
      "epoch": 1.97,
      "learning_rate": 2.8088930752543063e-06,
      "loss": 1.3713,
      "step": 283
    },
    {
      "epoch": 1.97,
      "learning_rate": 2.775257197206996e-06,
      "loss": 1.4089,
      "step": 284
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.741746388101493e-06,
      "loss": 1.4123,
      "step": 285
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.7083625318236213e-06,
      "loss": 1.4307,
      "step": 286
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.6751075051222684e-06,
      "loss": 1.4278,
      "step": 287
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.6419831775038763e-06,
      "loss": 1.472,
      "step": 288
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.6089914111273398e-06,
      "loss": 1.3458,
      "step": 289
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.576134060699328e-06,
      "loss": 1.3607,
      "step": 290
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.54341297337001e-06,
      "loss": 1.3164,
      "step": 291
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.510829988629222e-06,
      "loss": 1.2873,
      "step": 292
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.478386938203043e-06,
      "loss": 1.2711,
      "step": 293
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.4460856459508374e-06,
      "loss": 1.2507,
      "step": 294
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.4139279277627113e-06,
      "loss": 1.201,
      "step": 295
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.381915591457424e-06,
      "loss": 1.1929,
      "step": 296
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.3500504366807743e-06,
      "loss": 1.2074,
      "step": 297
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.3183342548044067e-06,
      "loss": 1.2939,
      "step": 298
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.286768828825133e-06,
      "loss": 1.2518,
      "step": 299
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.2553559332646675e-06,
      "loss": 1.2253,
      "step": 300
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.2240973340698886e-06,
      "loss": 1.1937,
      "step": 301
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.1929947885135567e-06,
      "loss": 1.1734,
      "step": 302
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.1620500450955224e-06,
      "loss": 1.1867,
      "step": 303
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.1312648434444342e-06,
      "loss": 1.08,
      "step": 304
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.100640914219939e-06,
      "loss": 1.1372,
      "step": 305
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.0701799790153897e-06,
      "loss": 1.178,
      "step": 306
    },
    {
      "epoch": 2.13,
      "learning_rate": 2.039883750261063e-06,
      "loss": 1.0955,
      "step": 307
    },
    {
      "epoch": 2.14,
      "learning_rate": 2.00975393112789e-06,
      "loss": 1.1424,
      "step": 308
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.979792215431709e-06,
      "loss": 1.0761,
      "step": 309
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.9500002875380458e-06,
      "loss": 1.1093,
      "step": 310
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.920379822267414e-06,
      "loss": 1.0658,
      "step": 311
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.8909324848011802e-06,
      "loss": 1.0853,
      "step": 312
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.8616599305879334e-06,
      "loss": 1.0628,
      "step": 313
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.8325638052504235e-06,
      "loss": 1.103,
      "step": 314
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.8036457444930643e-06,
      "loss": 1.0065,
      "step": 315
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.774907374009953e-06,
      "loss": 1.0206,
      "step": 316
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.7463503093935063e-06,
      "loss": 1.0387,
      "step": 317
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.7179761560436097e-06,
      "loss": 1.0256,
      "step": 318
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.6897865090773858e-06,
      "loss": 1.0667,
      "step": 319
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.661782953239512e-06,
      "loss": 1.1087,
      "step": 320
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.6339670628131327e-06,
      "loss": 1.077,
      "step": 321
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.6063404015313583e-06,
      "loss": 1.1042,
      "step": 322
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.578904522489354e-06,
      "loss": 1.0646,
      "step": 323
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.5516609680570316e-06,
      "loss": 1.0694,
      "step": 324
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.5246112697923389e-06,
      "loss": 1.0964,
      "step": 325
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.4977569483551634e-06,
      "loss": 1.0993,
      "step": 326
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.471099513421842e-06,
      "loss": 1.0485,
      "step": 327
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.4446404636002931e-06,
      "loss": 1.0584,
      "step": 328
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.4183812863457624e-06,
      "loss": 1.1169,
      "step": 329
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.3923234578772177e-06,
      "loss": 1.087,
      "step": 330
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.3664684430943431e-06,
      "loss": 1.1103,
      "step": 331
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.3408176954951912e-06,
      "loss": 1.0431,
      "step": 332
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.315372657094483e-06,
      "loss": 1.0704,
      "step": 333
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.29013475834252e-06,
      "loss": 1.0333,
      "step": 334
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.265105418044793e-06,
      "loss": 1.0854,
      "step": 335
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.2402860432821972e-06,
      "loss": 1.0929,
      "step": 336
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.2156780293319476e-06,
      "loss": 1.1359,
      "step": 337
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.1912827595891313e-06,
      "loss": 1.1184,
      "step": 338
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.1671016054889407e-06,
      "loss": 1.1367,
      "step": 339
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.1431359264295717e-06,
      "loss": 1.0994,
      "step": 340
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.1193870696958058e-06,
      "loss": 1.0315,
      "step": 341
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0958563703832675e-06,
      "loss": 1.1019,
      "step": 342
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0725451513233676e-06,
      "loss": 1.142,
      "step": 343
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0494547230089413e-06,
      "loss": 1.1241,
      "step": 344
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.0265863835205709e-06,
      "loss": 1.0894,
      "step": 345
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.0039414184536161e-06,
      "loss": 1.16,
      "step": 346
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.815211008459336e-07,
      "loss": 1.1217,
      "step": 347
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.593266911063253e-07,
      "loss": 1.1184,
      "step": 348
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.373594369436611e-07,
      "loss": 1.0941,
      "step": 349
    },
    {
      "epoch": 2.43,
      "learning_rate": 9.15620573296756e-07,
      "loss": 1.0911,
      "step": 350
    },
    {
      "epoch": 2.44,
      "learning_rate": 8.941113222649328e-07,
      "loss": 1.12,
      "step": 351
    },
    {
      "epoch": 2.44,
      "learning_rate": 8.728328930393188e-07,
      "loss": 1.0704,
      "step": 352
    },
    {
      "epoch": 2.45,
      "learning_rate": 8.517864818348804e-07,
      "loss": 1.0672,
      "step": 353
    },
    {
      "epoch": 2.46,
      "learning_rate": 8.309732718231578e-07,
      "loss": 1.1475,
      "step": 354
    },
    {
      "epoch": 2.47,
      "learning_rate": 8.103944330657665e-07,
      "loss": 1.0936,
      "step": 355
    },
    {
      "epoch": 2.47,
      "learning_rate": 7.900511224486085e-07,
      "loss": 1.0835,
      "step": 356
    },
    {
      "epoch": 2.48,
      "learning_rate": 7.699444836168413e-07,
      "loss": 1.1453,
      "step": 357
    },
    {
      "epoch": 2.49,
      "learning_rate": 7.500756469105819e-07,
      "loss": 1.1123,
      "step": 358
    },
    {
      "epoch": 2.49,
      "learning_rate": 7.304457293013656e-07,
      "loss": 1.0686,
      "step": 359
    },
    {
      "epoch": 2.5,
      "learning_rate": 7.1105583432935e-07,
      "loss": 1.1315,
      "step": 360
    },
    {
      "epoch": 2.51,
      "learning_rate": 6.919070520412768e-07,
      "loss": 1.0504,
      "step": 361
    },
    {
      "epoch": 2.51,
      "learning_rate": 6.730004589291961e-07,
      "loss": 1.1282,
      "step": 362
    },
    {
      "epoch": 2.52,
      "learning_rate": 6.543371178699442e-07,
      "loss": 1.164,
      "step": 363
    },
    {
      "epoch": 2.53,
      "learning_rate": 6.359180780653957e-07,
      "loss": 1.0805,
      "step": 364
    },
    {
      "epoch": 2.53,
      "learning_rate": 6.177443749834743e-07,
      "loss": 1.1122,
      "step": 365
    },
    {
      "epoch": 2.54,
      "learning_rate": 5.998170302999529e-07,
      "loss": 1.1417,
      "step": 366
    },
    {
      "epoch": 2.55,
      "learning_rate": 5.821370518410019e-07,
      "loss": 1.1123,
      "step": 367
    },
    {
      "epoch": 2.56,
      "learning_rate": 5.647054335265489e-07,
      "loss": 1.156,
      "step": 368
    },
    {
      "epoch": 2.56,
      "learning_rate": 5.475231553143906e-07,
      "loss": 1.1302,
      "step": 369
    },
    {
      "epoch": 2.57,
      "learning_rate": 5.305911831451044e-07,
      "loss": 1.1281,
      "step": 370
    },
    {
      "epoch": 2.58,
      "learning_rate": 5.139104688877549e-07,
      "loss": 1.1888,
      "step": 371
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.974819502863687e-07,
      "loss": 1.1384,
      "step": 372
    },
    {
      "epoch": 2.59,
      "learning_rate": 4.813065509072279e-07,
      "loss": 1.1435,
      "step": 373
    },
    {
      "epoch": 2.6,
      "learning_rate": 4.6538518008694465e-07,
      "loss": 1.1718,
      "step": 374
    },
    {
      "epoch": 2.6,
      "learning_rate": 4.4971873288134237e-07,
      "loss": 1.2111,
      "step": 375
    },
    {
      "epoch": 2.61,
      "learning_rate": 4.343080900151375e-07,
      "loss": 1.1525,
      "step": 376
    },
    {
      "epoch": 2.62,
      "learning_rate": 4.1915411783242766e-07,
      "loss": 1.1026,
      "step": 377
    },
    {
      "epoch": 2.62,
      "learning_rate": 4.0425766824798817e-07,
      "loss": 1.1428,
      "step": 378
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.8961957869937893e-07,
      "loss": 1.1742,
      "step": 379
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.752406720998691e-07,
      "loss": 1.1439,
      "step": 380
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.61121756792171e-07,
      "loss": 1.1181,
      "step": 381
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.472636265030005e-07,
      "loss": 1.1227,
      "step": 382
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.3366706029845097e-07,
      "loss": 1.1738,
      "step": 383
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.2033282254020747e-07,
      "loss": 1.0632,
      "step": 384
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.072616628425601e-07,
      "loss": 1.0974,
      "step": 385
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.9445431603027876e-07,
      "loss": 1.1317,
      "step": 386
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.8191150209729233e-07,
      "loss": 1.1371,
      "step": 387
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.696339261662156e-07,
      "loss": 1.1154,
      "step": 388
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.576222784487148e-07,
      "loss": 1.1395,
      "step": 389
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.45877234206694e-07,
      "loss": 1.1419,
      "step": 390
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.3439945371434792e-07,
      "loss": 1.135,
      "step": 391
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.2318958222103004e-07,
      "loss": 1.1152,
      "step": 392
    },
    {
      "epoch": 2.73,
      "learning_rate": 2.1224824991498695e-07,
      "loss": 1.1253,
      "step": 393
    },
    {
      "epoch": 2.74,
      "learning_rate": 2.0157607188792894e-07,
      "loss": 1.1046,
      "step": 394
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.911736481004489e-07,
      "loss": 1.1955,
      "step": 395
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.8104156334829703e-07,
      "loss": 1.1563,
      "step": 396
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.7118038722950313e-07,
      "loss": 1.1846,
      "step": 397
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.615906741123574e-07,
      "loss": 1.1909,
      "step": 398
    },
    {
      "epoch": 2.77,
      "learning_rate": 1.5227296310424244e-07,
      "loss": 1.1598,
      "step": 399
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.432277780213298e-07,
      "loss": 1.118,
      "step": 400
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.3445562735912965e-07,
      "loss": 1.2047,
      "step": 401
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.2595700426390633e-07,
      "loss": 1.1622,
      "step": 402
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.1773238650495122e-07,
      "loss": 1.2584,
      "step": 403
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.0978223644773134e-07,
      "loss": 1.2419,
      "step": 404
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.0210700102788796e-07,
      "loss": 1.2292,
      "step": 405
    },
    {
      "epoch": 2.82,
      "learning_rate": 9.470711172611724e-08,
      "loss": 1.2044,
      "step": 406
    },
    {
      "epoch": 2.83,
      "learning_rate": 8.75829845439119e-08,
      "loss": 1.1776,
      "step": 407
    },
    {
      "epoch": 2.83,
      "learning_rate": 8.073501998017152e-08,
      "loss": 1.1786,
      "step": 408
    },
    {
      "epoch": 2.84,
      "learning_rate": 7.416360300869285e-08,
      "loss": 1.2012,
      "step": 409
    },
    {
      "epoch": 2.85,
      "learning_rate": 6.786910305652373e-08,
      "loss": 1.1945,
      "step": 410
    },
    {
      "epoch": 2.85,
      "learning_rate": 6.185187398319691e-08,
      "loss": 1.2414,
      "step": 411
    },
    {
      "epoch": 2.86,
      "learning_rate": 5.611225406083609e-08,
      "loss": 1.2703,
      "step": 412
    },
    {
      "epoch": 2.87,
      "learning_rate": 5.065056595513984e-08,
      "loss": 1.1913,
      "step": 413
    },
    {
      "epoch": 2.88,
      "learning_rate": 4.546711670724124e-08,
      "loss": 1.1705,
      "step": 414
    },
    {
      "epoch": 2.88,
      "learning_rate": 4.0562197716448316e-08,
      "loss": 1.2055,
      "step": 415
    },
    {
      "epoch": 2.89,
      "learning_rate": 3.593608472386045e-08,
      "loss": 1.1861,
      "step": 416
    },
    {
      "epoch": 2.9,
      "learning_rate": 3.1589037796869725e-08,
      "loss": 1.2427,
      "step": 417
    },
    {
      "epoch": 2.9,
      "learning_rate": 2.7521301314537564e-08,
      "loss": 1.1897,
      "step": 418
    },
    {
      "epoch": 2.91,
      "learning_rate": 2.373310395385797e-08,
      "loss": 1.2836,
      "step": 419
    },
    {
      "epoch": 2.92,
      "learning_rate": 2.022465867690282e-08,
      "loss": 1.2451,
      "step": 420
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.699616271884752e-08,
      "loss": 1.2135,
      "step": 421
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.4047797576885458e-08,
      "loss": 1.2279,
      "step": 422
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.1379729000021711e-08,
      "loss": 1.2035,
      "step": 423
    },
    {
      "epoch": 2.94,
      "learning_rate": 8.99210697975883e-09,
      "loss": 1.2248,
      "step": 424
    },
    {
      "epoch": 2.95,
      "learning_rate": 6.885065741661367e-09,
      "loss": 1.2341,
      "step": 425
    },
    {
      "epoch": 2.96,
      "learning_rate": 5.058723737811355e-09,
      "loss": 1.1999,
      "step": 426
    },
    {
      "epoch": 2.97,
      "learning_rate": 3.5131836401502972e-09,
      "loss": 1.1994,
      "step": 427
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.2485323347054555e-09,
      "loss": 1.2527,
      "step": 428
    },
    {
      "epoch": 2.98,
      "learning_rate": 1.2648409167070886e-09,
      "loss": 1.2552,
      "step": 429
    },
    {
      "epoch": 2.99,
      "learning_rate": 5.621646865899832e-10,
      "loss": 1.2782,
      "step": 430
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.405431468848306e-10,
      "loss": 1.2717,
      "step": 431
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.0,
      "loss": 1.3355,
      "step": 432
    },
    {
      "epoch": 3.0,
      "step": 432,
      "total_flos": 0.0,
      "train_loss": 1.441655464746334,
      "train_runtime": 12001.223,
      "train_samples_per_second": 4.04,
      "train_steps_per_second": 0.036
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 432,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 100,
  "total_flos": 0.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}