diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7708 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 20.0,
+  "eval_steps": 500,
+  "global_step": 1280,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "learning_rate": 2e-05,
+      "loss": 1.9538,
+      "step": 1
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4e-05,
+      "loss": 2.1597,
+      "step": 2
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6e-05,
+      "loss": 2.1486,
+      "step": 3
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8e-05,
+      "loss": 2.1629,
+      "step": 4
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001,
+      "loss": 2.5289,
+      "step": 5
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.992156862745099e-05,
+      "loss": 1.8736,
+      "step": 6
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.984313725490197e-05,
+      "loss": 1.6761,
+      "step": 7
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.976470588235295e-05,
+      "loss": 1.6344,
+      "step": 8
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.968627450980392e-05,
+      "loss": 1.5802,
+      "step": 9
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.96078431372549e-05,
+      "loss": 1.7118,
+      "step": 10
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.952941176470588e-05,
+      "loss": 1.6162,
+      "step": 11
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.945098039215687e-05,
+      "loss": 1.537,
+      "step": 12
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.937254901960785e-05,
+      "loss": 1.3941,
+      "step": 13
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.929411764705883e-05,
+      "loss": 1.5084,
+      "step": 14
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.921568627450981e-05,
+      "loss": 1.378,
+      "step": 15
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.913725490196078e-05,
+      "loss": 1.4426,
+      "step": 16
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.913725490196078e-05,
+      "loss": 1.3645,
+      "step": 17
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.905882352941176e-05,
+      "loss": 1.3826,
+      "step": 18
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.898039215686275e-05,
+      "loss": 1.266,
+      "step": 19
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.890196078431373e-05,
+      "loss": 1.631,
+      "step": 20
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.882352941176471e-05,
+      "loss": 1.2106,
+      "step": 21
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.874509803921569e-05,
+      "loss": 1.2194,
+      "step": 22
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.866666666666668e-05,
+      "loss": 1.3047,
+      "step": 23
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.858823529411764e-05,
+      "loss": 1.5103,
+      "step": 24
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.850980392156864e-05,
+      "loss": 1.4135,
+      "step": 25
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.843137254901961e-05,
+      "loss": 1.3584,
+      "step": 26
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.835294117647059e-05,
+      "loss": 1.4999,
+      "step": 27
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.827450980392157e-05,
+      "loss": 1.2088,
+      "step": 28
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.819607843137255e-05,
+      "loss": 1.2909,
+      "step": 29
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.811764705882354e-05,
+      "loss": 1.4807,
+      "step": 30
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.80392156862745e-05,
+      "loss": 1.42,
+      "step": 31
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.79607843137255e-05,
+      "loss": 1.3942,
+      "step": 32
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.788235294117648e-05,
+      "loss": 1.223,
+      "step": 33
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.780392156862745e-05,
+      "loss": 1.0867,
+      "step": 34
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.772549019607843e-05,
+      "loss": 1.3569,
+      "step": 35
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.764705882352942e-05,
+      "loss": 1.342,
+      "step": 36
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.75686274509804e-05,
+      "loss": 1.3113,
+      "step": 37
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.749019607843137e-05,
+      "loss": 1.5478,
+      "step": 38
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.741176470588236e-05,
+      "loss": 1.1556,
+      "step": 39
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.733333333333335e-05,
+      "loss": 1.1162,
+      "step": 40
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.725490196078431e-05,
+      "loss": 1.1865,
+      "step": 41
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.71764705882353e-05,
+      "loss": 1.1985,
+      "step": 42
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.709803921568628e-05,
+      "loss": 1.2793,
+      "step": 43
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.701960784313726e-05,
+      "loss": 1.1177,
+      "step": 44
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.694117647058823e-05,
+      "loss": 1.2497,
+      "step": 45
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.686274509803923e-05,
+      "loss": 1.5408,
+      "step": 46
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.678431372549021e-05,
+      "loss": 1.3027,
+      "step": 47
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.670588235294118e-05,
+      "loss": 1.1345,
+      "step": 48
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.662745098039216e-05,
+      "loss": 1.2709,
+      "step": 49
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.654901960784314e-05,
+      "loss": 1.2628,
+      "step": 50
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.647058823529412e-05,
+      "loss": 1.2852,
+      "step": 51
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.639215686274509e-05,
+      "loss": 1.1795,
+      "step": 52
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.631372549019609e-05,
+      "loss": 1.1259,
+      "step": 53
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.623529411764707e-05,
+      "loss": 1.1628,
+      "step": 54
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.615686274509804e-05,
+      "loss": 1.3307,
+      "step": 55
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.607843137254903e-05,
+      "loss": 1.2789,
+      "step": 56
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.6e-05,
+      "loss": 1.3031,
+      "step": 57
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.592156862745098e-05,
+      "loss": 1.0189,
+      "step": 58
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.584313725490197e-05,
+      "loss": 1.2856,
+      "step": 59
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.576470588235295e-05,
+      "loss": 1.1846,
+      "step": 60
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.568627450980393e-05,
+      "loss": 1.4475,
+      "step": 61
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.56078431372549e-05,
+      "loss": 1.2606,
+      "step": 62
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.55294117647059e-05,
+      "loss": 1.2926,
+      "step": 63
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.545098039215686e-05,
+      "loss": 1.3933,
+      "step": 64
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.537254901960785e-05,
+      "loss": 1.2592,
+      "step": 65
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.529411764705883e-05,
+      "loss": 1.044,
+      "step": 66
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.521568627450981e-05,
+      "loss": 1.0397,
+      "step": 67
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.513725490196079e-05,
+      "loss": 1.0966,
+      "step": 68
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.505882352941176e-05,
+      "loss": 0.884,
+      "step": 69
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.498039215686276e-05,
+      "loss": 0.9971,
+      "step": 70
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.490196078431373e-05,
+      "loss": 1.0018,
+      "step": 71
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.482352941176471e-05,
+      "loss": 0.9367,
+      "step": 72
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.474509803921569e-05,
+      "loss": 1.0932,
+      "step": 73
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.466666666666667e-05,
+      "loss": 1.1328,
+      "step": 74
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.458823529411766e-05,
+      "loss": 0.7637,
+      "step": 75
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.450980392156862e-05,
+      "loss": 1.1115,
+      "step": 76
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.443137254901962e-05,
+      "loss": 1.2783,
+      "step": 77
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.435294117647059e-05,
+      "loss": 0.969,
+      "step": 78
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.427450980392157e-05,
+      "loss": 0.7489,
+      "step": 79
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.419607843137255e-05,
+      "loss": 0.9965,
+      "step": 80
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.411764705882353e-05,
+      "loss": 1.0022,
+      "step": 81
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.403921568627452e-05,
+      "loss": 1.1539,
+      "step": 82
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.396078431372549e-05,
+      "loss": 1.0476,
+      "step": 83
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.388235294117648e-05,
+      "loss": 1.025,
+      "step": 84
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.380392156862745e-05,
+      "loss": 1.0464,
+      "step": 85
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.372549019607843e-05,
+      "loss": 0.9128,
+      "step": 86
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.364705882352941e-05,
+      "loss": 1.0464,
+      "step": 87
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.35686274509804e-05,
+      "loss": 1.0524,
+      "step": 88
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.349019607843138e-05,
+      "loss": 0.717,
+      "step": 89
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.341176470588235e-05,
+      "loss": 1.1298,
+      "step": 90
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.333333333333334e-05,
+      "loss": 0.9638,
+      "step": 91
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.325490196078431e-05,
+      "loss": 1.123,
+      "step": 92
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.31764705882353e-05,
+      "loss": 1.0737,
+      "step": 93
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.309803921568629e-05,
+      "loss": 1.1002,
+      "step": 94
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.301960784313726e-05,
+      "loss": 1.0011,
+      "step": 95
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.294117647058824e-05,
+      "loss": 1.0552,
+      "step": 96
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.286274509803922e-05,
+      "loss": 1.0271,
+      "step": 97
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.27843137254902e-05,
+      "loss": 0.9079,
+      "step": 98
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.270588235294117e-05,
+      "loss": 0.73,
+      "step": 99
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.262745098039216e-05,
+      "loss": 1.1459,
+      "step": 100
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.254901960784315e-05,
+      "loss": 1.075,
+      "step": 101
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.247058823529412e-05,
+      "loss": 0.9933,
+      "step": 102
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.23921568627451e-05,
+      "loss": 0.9083,
+      "step": 103
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.231372549019608e-05,
+      "loss": 0.8667,
+      "step": 104
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.223529411764707e-05,
+      "loss": 1.0045,
+      "step": 105
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.215686274509804e-05,
+      "loss": 0.9416,
+      "step": 106
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.207843137254902e-05,
+      "loss": 1.1244,
+      "step": 107
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 0.7778,
+      "step": 108
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.192156862745098e-05,
+      "loss": 0.8804,
+      "step": 109
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.184313725490196e-05,
+      "loss": 1.0487,
+      "step": 110
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.176470588235295e-05,
+      "loss": 1.0215,
+      "step": 111
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.168627450980393e-05,
+      "loss": 0.9556,
+      "step": 112
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.160784313725491e-05,
+      "loss": 1.0114,
+      "step": 113
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.152941176470588e-05,
+      "loss": 1.0112,
+      "step": 114
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.145098039215688e-05,
+      "loss": 0.9658,
+      "step": 115
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.137254901960784e-05,
+      "loss": 1.0448,
+      "step": 116
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.129411764705883e-05,
+      "loss": 0.9674,
+      "step": 117
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.121568627450981e-05,
+      "loss": 0.9712,
+      "step": 118
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.113725490196079e-05,
+      "loss": 1.0485,
+      "step": 119
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.105882352941177e-05,
+      "loss": 1.0545,
+      "step": 120
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.098039215686274e-05,
+      "loss": 0.9935,
+      "step": 121
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.090196078431374e-05,
+      "loss": 0.9879,
+      "step": 122
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.08235294117647e-05,
+      "loss": 0.9226,
+      "step": 123
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.074509803921569e-05,
+      "loss": 1.0172,
+      "step": 124
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.066666666666667e-05,
+      "loss": 1.0499,
+      "step": 125
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.058823529411765e-05,
+      "loss": 1.0008,
+      "step": 126
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.050980392156863e-05,
+      "loss": 1.1101,
+      "step": 127
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.043137254901962e-05,
+      "loss": 1.0055,
+      "step": 128
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.03529411764706e-05,
+      "loss": 0.53,
+      "step": 129
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.027450980392157e-05,
+      "loss": 0.5856,
+      "step": 130
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.019607843137255e-05,
+      "loss": 0.6829,
+      "step": 131
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.011764705882353e-05,
+      "loss": 0.6567,
+      "step": 132
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.003921568627451e-05,
+      "loss": 0.6257,
+      "step": 133
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.99607843137255e-05,
+      "loss": 0.5997,
+      "step": 134
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.988235294117648e-05,
+      "loss": 0.636,
+      "step": 135
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.980392156862746e-05,
+      "loss": 0.5466,
+      "step": 136
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.972549019607843e-05,
+      "loss": 0.5084,
+      "step": 137
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.964705882352941e-05,
+      "loss": 0.67,
+      "step": 138
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.95686274509804e-05,
+      "loss": 0.644,
+      "step": 139
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.949019607843138e-05,
+      "loss": 0.516,
+      "step": 140
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.941176470588236e-05,
+      "loss": 0.5521,
+      "step": 141
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.933333333333334e-05,
+      "loss": 0.4775,
+      "step": 142
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.925490196078432e-05,
+      "loss": 0.5201,
+      "step": 143
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.917647058823529e-05,
+      "loss": 0.5056,
+      "step": 144
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.909803921568627e-05,
+      "loss": 0.5725,
+      "step": 145
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.901960784313726e-05,
+      "loss": 0.4347,
+      "step": 146
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.894117647058824e-05,
+      "loss": 0.6168,
+      "step": 147
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.886274509803922e-05,
+      "loss": 0.5022,
+      "step": 148
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.87843137254902e-05,
+      "loss": 0.7167,
+      "step": 149
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.870588235294118e-05,
+      "loss": 0.4437,
+      "step": 150
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 8.862745098039215e-05,
+      "loss": 0.5978,
+      "step": 151
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 8.854901960784314e-05,
+      "loss": 0.5991,
+      "step": 152
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 8.847058823529413e-05,
+      "loss": 0.533,
+      "step": 153
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 8.83921568627451e-05,
+      "loss": 0.6925,
+      "step": 154
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 8.831372549019608e-05,
+      "loss": 0.5193,
+      "step": 155
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 8.823529411764706e-05,
+      "loss": 0.5601,
+      "step": 156
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 8.815686274509805e-05,
+      "loss": 0.4335,
+      "step": 157
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.807843137254902e-05,
+      "loss": 0.4782,
+      "step": 158
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 0.4825,
+      "step": 159
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.7921568627451e-05,
+      "loss": 0.6073,
+      "step": 160
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.784313725490196e-05,
+      "loss": 0.551,
+      "step": 161
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 8.776470588235294e-05,
+      "loss": 0.5034,
+      "step": 162
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 8.768627450980393e-05,
+      "loss": 0.6109,
+      "step": 163
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 8.760784313725491e-05,
+      "loss": 0.5806,
+      "step": 164
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 8.752941176470588e-05,
+      "loss": 0.5629,
+      "step": 165
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.745098039215687e-05,
+      "loss": 0.6154,
+      "step": 166
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.737254901960786e-05,
+      "loss": 0.6405,
+      "step": 167
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.729411764705882e-05,
+      "loss": 0.5702,
+      "step": 168
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.72156862745098e-05,
+      "loss": 0.5646,
+      "step": 169
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 8.713725490196079e-05,
+      "loss": 0.5897,
+      "step": 170
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 8.705882352941177e-05,
+      "loss": 0.6113,
+      "step": 171
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 8.698039215686274e-05,
+      "loss": 0.5503,
+      "step": 172
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 8.690196078431374e-05,
+      "loss": 0.6661,
+      "step": 173
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 8.682352941176472e-05,
+      "loss": 0.4457,
+      "step": 174
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.674509803921569e-05,
+      "loss": 0.6719,
+      "step": 175
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.666666666666667e-05,
+      "loss": 0.54,
+      "step": 176
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.658823529411765e-05,
+      "loss": 0.4914,
+      "step": 177
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.650980392156863e-05,
+      "loss": 0.5155,
+      "step": 178
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.64313725490196e-05,
+      "loss": 0.615,
+      "step": 179
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.63529411764706e-05,
+      "loss": 0.6426,
+      "step": 180
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.627450980392158e-05,
+      "loss": 0.6963,
+      "step": 181
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.619607843137255e-05,
+      "loss": 0.4658,
+      "step": 182
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.611764705882353e-05,
+      "loss": 0.5623,
+      "step": 183
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.603921568627451e-05,
+      "loss": 0.4873,
+      "step": 184
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.59607843137255e-05,
+      "loss": 0.5005,
+      "step": 185
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.588235294117646e-05,
+      "loss": 0.4677,
+      "step": 186
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.580392156862746e-05,
+      "loss": 0.5745,
+      "step": 187
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.572549019607844e-05,
+      "loss": 0.533,
+      "step": 188
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.564705882352941e-05,
+      "loss": 0.5159,
+      "step": 189
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.55686274509804e-05,
+      "loss": 0.5934,
+      "step": 190
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.549019607843137e-05,
+      "loss": 0.5669,
+      "step": 191
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.541176470588236e-05,
+      "loss": 0.6998,
+      "step": 192
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 8.533333333333334e-05,
+      "loss": 0.3786,
+      "step": 193
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 8.525490196078432e-05,
+      "loss": 0.3003,
+      "step": 194
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 8.51764705882353e-05,
+      "loss": 0.3218,
+      "step": 195
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 8.509803921568627e-05,
+      "loss": 0.3124,
+      "step": 196
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 8.501960784313727e-05,
+      "loss": 0.2532,
+      "step": 197
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 8.494117647058824e-05,
+      "loss": 0.2547,
+      "step": 198
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 8.486274509803922e-05,
+      "loss": 0.226,
+      "step": 199
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 8.47843137254902e-05,
+      "loss": 0.1828,
+      "step": 200
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 8.470588235294118e-05,
+      "loss": 0.2067,
+      "step": 201
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 8.462745098039216e-05,
+      "loss": 0.2229,
+      "step": 202
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.454901960784313e-05,
+      "loss": 0.2521,
+      "step": 203
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.447058823529413e-05,
+      "loss": 0.3148,
+      "step": 204
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.43921568627451e-05,
+      "loss": 0.2308,
+      "step": 205
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.431372549019608e-05,
+      "loss": 0.3097,
+      "step": 206
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.423529411764706e-05,
+      "loss": 0.2641,
+      "step": 207
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.415686274509804e-05,
+      "loss": 0.1943,
+      "step": 208
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.407843137254903e-05,
+      "loss": 0.2683,
+      "step": 209
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.4e-05,
+      "loss": 0.2312,
+      "step": 210
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.392156862745099e-05,
+      "loss": 0.2658,
+      "step": 211
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.384313725490196e-05,
+      "loss": 0.1875,
+      "step": 212
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.376470588235294e-05,
+      "loss": 0.2186,
+      "step": 213
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.368627450980392e-05,
+      "loss": 0.1702,
+      "step": 214
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.36078431372549e-05,
+      "loss": 0.2418,
+      "step": 215
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.352941176470589e-05,
+      "loss": 0.2263,
+      "step": 216
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.345098039215686e-05,
+      "loss": 0.2058,
+      "step": 217
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.337254901960785e-05,
+      "loss": 0.2032,
+      "step": 218
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.329411764705882e-05,
+      "loss": 0.3378,
+      "step": 219
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.32156862745098e-05,
+      "loss": 0.1929,
+      "step": 220
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.313725490196079e-05,
+      "loss": 0.2324,
+      "step": 221
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.305882352941177e-05,
+      "loss": 0.2527,
+      "step": 222
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.298039215686275e-05,
+      "loss": 0.2077,
+      "step": 223
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.290196078431372e-05,
+      "loss": 0.2435,
+      "step": 224
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 8.282352941176471e-05,
+      "loss": 0.2113,
+      "step": 225
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 8.274509803921568e-05,
+      "loss": 0.2206,
+      "step": 226
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 8.266666666666667e-05,
+      "loss": 0.3191,
+      "step": 227
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.258823529411766e-05,
+      "loss": 0.2762,
+      "step": 228
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 8.250980392156863e-05,
+      "loss": 0.2685,
+      "step": 229
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 8.243137254901961e-05,
+      "loss": 0.206,
+      "step": 230
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 8.23529411764706e-05,
+      "loss": 0.2543,
+      "step": 231
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 8.227450980392158e-05,
+      "loss": 0.3673,
+      "step": 232
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 8.219607843137256e-05,
+      "loss": 0.252,
+      "step": 233
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.211764705882353e-05,
+      "loss": 0.2464,
+      "step": 234
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.203921568627452e-05,
+      "loss": 0.226,
+      "step": 235
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.196078431372549e-05,
+      "loss": 0.299,
+      "step": 236
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.188235294117647e-05,
+      "loss": 0.271,
+      "step": 237
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.180392156862746e-05,
+      "loss": 0.2645,
+      "step": 238
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.172549019607844e-05,
+      "loss": 0.2565,
+      "step": 239
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 8.164705882352942e-05,
+      "loss": 0.2216,
+      "step": 240
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 8.156862745098039e-05,
+      "loss": 0.1879,
+      "step": 241
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 8.149019607843139e-05,
+      "loss": 0.2389,
+      "step": 242
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 8.141176470588235e-05,
+      "loss": 0.3006,
+      "step": 243
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 8.133333333333334e-05,
+      "loss": 0.2318,
+      "step": 244
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.125490196078432e-05,
+      "loss": 0.2301,
+      "step": 245
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.11764705882353e-05,
+      "loss": 0.1571,
+      "step": 246
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.109803921568628e-05,
+      "loss": 0.3309,
+      "step": 247
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.101960784313725e-05,
+      "loss": 0.2331,
+      "step": 248
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.094117647058825e-05,
+      "loss": 0.3493,
+      "step": 249
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.086274509803922e-05,
+      "loss": 0.2129,
+      "step": 250
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.07843137254902e-05,
+      "loss": 0.2602,
+      "step": 251
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.070588235294118e-05,
+      "loss": 0.2793,
+      "step": 252
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.062745098039216e-05,
+      "loss": 0.2635,
+      "step": 253
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.054901960784314e-05,
+      "loss": 0.2777,
+      "step": 254
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.047058823529411e-05,
+      "loss": 0.2394,
+      "step": 255
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 8.039215686274511e-05,
+      "loss": 0.1998,
+      "step": 256
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 8.031372549019608e-05,
+      "loss": 0.1736,
+      "step": 257
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 8.023529411764706e-05,
+      "loss": 0.1402,
+      "step": 258
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 8.015686274509804e-05,
+      "loss": 0.1619,
+      "step": 259
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 8.007843137254902e-05,
+      "loss": 0.102,
+      "step": 260
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 8e-05,
+      "loss": 0.1576,
+      "step": 261
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.992156862745099e-05,
+      "loss": 0.1374,
+      "step": 262
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 7.984313725490197e-05,
+      "loss": 0.1287,
+      "step": 263
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.976470588235294e-05,
+      "loss": 0.1248,
+      "step": 264
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.968627450980392e-05,
+      "loss": 0.1144,
+      "step": 265
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.96078431372549e-05,
+      "loss": 0.1907,
+      "step": 266
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.952941176470589e-05,
+      "loss": 0.1158,
+      "step": 267
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 7.945098039215687e-05,
+      "loss": 0.0987,
+      "step": 268
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 7.937254901960785e-05,
+      "loss": 0.1187,
+      "step": 269
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 7.929411764705883e-05,
+      "loss": 0.1501,
+      "step": 270
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 7.92156862745098e-05,
+      "loss": 0.1665,
+      "step": 271
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 7.913725490196078e-05,
+      "loss": 0.1479,
+      "step": 272
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 7.905882352941177e-05,
+      "loss": 0.1736,
+      "step": 273
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 7.898039215686275e-05,
+      "loss": 0.1141,
+      "step": 274
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 7.890196078431373e-05,
+      "loss": 0.142,
+      "step": 275
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 7.882352941176471e-05,
+      "loss": 0.1515,
+      "step": 276
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 7.87450980392157e-05,
+      "loss": 0.138,
+      "step": 277
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 7.866666666666666e-05,
+      "loss": 0.0947,
+      "step": 278
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 7.858823529411765e-05,
+      "loss": 0.1916,
+      "step": 279
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.850980392156864e-05,
+      "loss": 0.1534,
+      "step": 280
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.843137254901961e-05,
+      "loss": 0.1358,
+      "step": 281
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.835294117647059e-05,
+      "loss": 0.1721,
+      "step": 282
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.827450980392157e-05,
+      "loss": 0.1484,
+      "step": 283
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 7.819607843137256e-05,
+      "loss": 0.1692,
+      "step": 284
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 7.811764705882353e-05,
+      "loss": 0.1638,
+      "step": 285
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 7.803921568627451e-05,
+      "loss": 0.1808,
+      "step": 286
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.79607843137255e-05,
+      "loss": 0.1696,
+      "step": 287
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.788235294117647e-05,
+      "loss": 0.1273,
+      "step": 288
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.780392156862745e-05,
+      "loss": 0.1324,
+      "step": 289
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 7.772549019607844e-05,
+      "loss": 0.1576,
+      "step": 290
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 7.764705882352942e-05,
+      "loss": 0.1519,
+      "step": 291
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 7.756862745098039e-05,
+      "loss": 0.1377,
+      "step": 292
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 7.749019607843138e-05,
+      "loss": 0.2005,
+      "step": 293
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 7.741176470588237e-05,
+      "loss": 0.1222,
+      "step": 294
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.733333333333333e-05,
+      "loss": 0.1825,
+      "step": 295
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.725490196078432e-05,
+      "loss": 0.1615,
+      "step": 296
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 7.71764705882353e-05,
+      "loss": 0.1261,
+      "step": 297
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 7.709803921568628e-05,
+      "loss": 0.1617,
+      "step": 298
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 7.701960784313725e-05,
+      "loss": 0.1689,
+      "step": 299
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 7.694117647058824e-05,
+      "loss": 0.154,
+      "step": 300
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 7.686274509803923e-05,
+      "loss": 0.1651,
+      "step": 301
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 7.67843137254902e-05,
+      "loss": 0.1971,
+      "step": 302
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 7.670588235294118e-05,
+      "loss": 0.1112,
+      "step": 303
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 7.662745098039216e-05,
+      "loss": 0.1445,
+      "step": 304
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 7.654901960784314e-05,
+      "loss": 0.1315,
+      "step": 305
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 7.647058823529411e-05,
+      "loss": 0.1233,
+      "step": 306
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.63921568627451e-05,
+      "loss": 0.2014,
+      "step": 307
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.631372549019609e-05,
+      "loss": 0.1653,
+      "step": 308
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 7.623529411764706e-05,
+      "loss": 0.1485,
+      "step": 309
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.615686274509804e-05,
+      "loss": 0.1405,
+      "step": 310
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 7.607843137254902e-05,
+      "loss": 0.1377,
+      "step": 311
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.6e-05,
+      "loss": 0.1303,
+      "step": 312
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.592156862745097e-05,
+      "loss": 0.1357,
+      "step": 313
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.584313725490197e-05,
+      "loss": 0.172,
+      "step": 314
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.576470588235295e-05,
+      "loss": 0.1594,
+      "step": 315
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.568627450980392e-05,
+      "loss": 0.1362,
+      "step": 316
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.56078431372549e-05,
+      "loss": 0.1239,
+      "step": 317
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.552941176470588e-05,
+      "loss": 0.1372,
+      "step": 318
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.545098039215687e-05,
+      "loss": 0.1509,
+      "step": 319
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 7.537254901960785e-05,
+      "loss": 0.1496,
+      "step": 320
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 7.529411764705883e-05,
+      "loss": 0.0656,
+      "step": 321
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 7.521568627450981e-05,
+      "loss": 0.0894,
+      "step": 322
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.513725490196078e-05,
+      "loss": 0.1103,
+      "step": 323
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 7.505882352941176e-05,
+      "loss": 0.089,
+      "step": 324
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.498039215686275e-05,
+      "loss": 0.0811,
+      "step": 325
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 7.490196078431373e-05,
+      "loss": 0.1525,
+      "step": 326
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 7.482352941176471e-05,
+      "loss": 0.0998,
+      "step": 327
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 7.474509803921569e-05,
+      "loss": 0.0963,
+      "step": 328
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 7.466666666666667e-05,
+      "loss": 0.1248,
+      "step": 329
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 7.458823529411764e-05,
+      "loss": 0.131,
+      "step": 330
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 7.450980392156864e-05,
+      "loss": 0.1234,
+      "step": 331
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 7.443137254901961e-05,
+      "loss": 0.0871,
+      "step": 332
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 7.435294117647059e-05,
+      "loss": 0.1357,
+      "step": 333
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 7.427450980392157e-05,
+      "loss": 0.1147,
+      "step": 334
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 7.419607843137255e-05,
+      "loss": 0.1039,
+      "step": 335
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.411764705882354e-05,
+      "loss": 0.1006,
+      "step": 336
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.40392156862745e-05,
+      "loss": 0.1421,
+      "step": 337
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.39607843137255e-05,
+      "loss": 0.0945,
+      "step": 338
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.388235294117647e-05,
+      "loss": 0.0963,
+      "step": 339
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 7.380392156862745e-05,
+      "loss": 0.1049,
+      "step": 340
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 7.372549019607843e-05,
+      "loss": 0.1083,
+      "step": 341
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 7.364705882352942e-05,
+      "loss": 0.1019,
+      "step": 342
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 7.35686274509804e-05,
+      "loss": 0.0891,
+      "step": 343
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 7.349019607843137e-05,
+      "loss": 0.1088,
+      "step": 344
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 7.341176470588236e-05,
+      "loss": 0.1117,
+      "step": 345
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 7.333333333333333e-05,
+      "loss": 0.106,
+      "step": 346
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 7.325490196078431e-05,
+      "loss": 0.1081,
+      "step": 347
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 7.31764705882353e-05,
+      "loss": 0.1084,
+      "step": 348
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 7.309803921568628e-05,
+      "loss": 0.0874,
+      "step": 349
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 7.301960784313726e-05,
+      "loss": 0.0924,
+      "step": 350
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 7.294117647058823e-05,
+      "loss": 0.0997,
+      "step": 351
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 7.286274509803922e-05,
+      "loss": 0.1061,
+      "step": 352
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 7.27843137254902e-05,
+      "loss": 0.0966,
+      "step": 353
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 7.270588235294118e-05,
+      "loss": 0.0861,
+      "step": 354
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 7.262745098039216e-05,
+      "loss": 0.1392,
+      "step": 355
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 7.254901960784314e-05,
+      "loss": 0.1611,
+      "step": 356
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 7.247058823529412e-05,
+      "loss": 0.0756,
+      "step": 357
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 7.239215686274509e-05,
+      "loss": 0.1514,
+      "step": 358
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 7.231372549019609e-05,
+      "loss": 0.1306,
+      "step": 359
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 7.223529411764707e-05,
+      "loss": 0.121,
+      "step": 360
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 7.215686274509804e-05,
+      "loss": 0.1002,
+      "step": 361
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 7.207843137254903e-05,
+      "loss": 0.1309,
+      "step": 362
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 7.2e-05,
+      "loss": 0.1803,
+      "step": 363
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 7.192156862745098e-05,
+      "loss": 0.1262,
+      "step": 364
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 7.184313725490197e-05,
+      "loss": 0.0874,
+      "step": 365
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 7.176470588235295e-05,
+      "loss": 0.1267,
+      "step": 366
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 7.168627450980393e-05,
+      "loss": 0.109,
+      "step": 367
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 7.16078431372549e-05,
+      "loss": 0.0941,
+      "step": 368
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.15294117647059e-05,
+      "loss": 0.105,
+      "step": 369
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.145098039215686e-05,
+      "loss": 0.1218,
+      "step": 370
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.137254901960785e-05,
+      "loss": 0.1091,
+      "step": 371
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 7.129411764705883e-05,
+      "loss": 0.1315,
+      "step": 372
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.121568627450981e-05,
+      "loss": 0.0915,
+      "step": 373
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.113725490196079e-05,
+      "loss": 0.101,
+      "step": 374
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.105882352941176e-05,
+      "loss": 0.1499,
+      "step": 375
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.098039215686276e-05,
+      "loss": 0.1129,
+      "step": 376
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.090196078431373e-05,
+      "loss": 0.1183,
+      "step": 377
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.082352941176471e-05,
+      "loss": 0.1071,
+      "step": 378
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.074509803921569e-05,
+      "loss": 0.1265,
+      "step": 379
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.066666666666667e-05,
+      "loss": 0.0965,
+      "step": 380
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.058823529411765e-05,
+      "loss": 0.0907,
+      "step": 381
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.050980392156862e-05,
+      "loss": 0.1339,
+      "step": 382
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.043137254901962e-05,
+      "loss": 0.1313,
+      "step": 383
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.035294117647059e-05,
+      "loss": 0.1154,
+      "step": 384
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.027450980392157e-05,
+      "loss": 0.0596,
+      "step": 385
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 7.019607843137255e-05,
+      "loss": 0.0634,
+      "step": 386
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 7.011764705882353e-05,
+      "loss": 0.0633,
+      "step": 387
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 7.003921568627452e-05,
+      "loss": 0.0675,
+      "step": 388
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 6.996078431372548e-05,
+      "loss": 0.0923,
+      "step": 389
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 6.988235294117648e-05,
+      "loss": 0.0809,
+      "step": 390
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 6.980392156862745e-05,
+      "loss": 0.1029,
+      "step": 391
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 6.972549019607843e-05,
+      "loss": 0.0676,
+      "step": 392
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 6.964705882352941e-05,
+      "loss": 0.0764,
+      "step": 393
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 6.95686274509804e-05,
+      "loss": 0.0819,
+      "step": 394
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 6.949019607843138e-05,
+      "loss": 0.0628,
+      "step": 395
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 6.941176470588236e-05,
+      "loss": 0.0761,
+      "step": 396
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 6.933333333333334e-05,
+      "loss": 0.0926,
+      "step": 397
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 6.925490196078431e-05,
+      "loss": 0.0602,
+      "step": 398
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 6.91764705882353e-05,
+      "loss": 0.0836,
+      "step": 399
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 6.909803921568629e-05,
+      "loss": 0.0711,
+      "step": 400
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 6.901960784313726e-05,
+      "loss": 0.0756,
+      "step": 401
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 6.894117647058824e-05,
+      "loss": 0.0794,
+      "step": 402
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 6.886274509803922e-05,
+      "loss": 0.1113,
+      "step": 403
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 6.87843137254902e-05,
+      "loss": 0.0611,
+      "step": 404
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 6.870588235294117e-05,
+      "loss": 0.081,
+      "step": 405
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 6.862745098039216e-05,
+      "loss": 0.1448,
+      "step": 406
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 6.854901960784315e-05,
+      "loss": 0.0786,
+      "step": 407
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 6.847058823529412e-05,
+      "loss": 0.0931,
+      "step": 408
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 6.83921568627451e-05,
+      "loss": 0.1108,
+      "step": 409
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 6.831372549019608e-05,
+      "loss": 0.1122,
+      "step": 410
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 6.823529411764707e-05,
+      "loss": 0.0992,
+      "step": 411
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 6.815686274509804e-05,
+      "loss": 0.0676,
+      "step": 412
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 6.807843137254902e-05,
+      "loss": 0.0889,
+      "step": 413
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 0.0813,
+      "step": 414
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 6.792156862745098e-05,
+      "loss": 0.0811,
+      "step": 415
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 6.784313725490196e-05,
+      "loss": 0.0975,
+      "step": 416
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 6.776470588235295e-05,
+      "loss": 0.071,
+      "step": 417
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 6.768627450980393e-05,
+      "loss": 0.0877,
+      "step": 418
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 6.76078431372549e-05,
+      "loss": 0.1611,
+      "step": 419
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 6.752941176470588e-05,
+      "loss": 0.1093,
+      "step": 420
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 6.745098039215687e-05,
+      "loss": 0.0744,
+      "step": 421
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 6.737254901960784e-05,
+      "loss": 0.1172,
+      "step": 422
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 6.729411764705883e-05,
+      "loss": 0.0631,
+      "step": 423
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 6.721568627450981e-05,
+      "loss": 0.0747,
+      "step": 424
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 6.713725490196079e-05,
+      "loss": 0.0969,
+      "step": 425
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 6.705882352941176e-05,
+      "loss": 0.0803,
+      "step": 426
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 6.698039215686275e-05,
+      "loss": 0.0926,
+      "step": 427
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 6.690196078431374e-05,
+      "loss": 0.1095,
+      "step": 428
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 6.68235294117647e-05,
+      "loss": 0.0687,
+      "step": 429
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 6.674509803921569e-05,
+      "loss": 0.1314,
+      "step": 430
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 0.0676,
+      "step": 431
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 6.658823529411765e-05,
+      "loss": 0.0821,
+      "step": 432
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 6.650980392156862e-05,
+      "loss": 0.1156,
+      "step": 433
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 6.643137254901962e-05,
+      "loss": 0.0859,
+      "step": 434
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 6.63529411764706e-05,
+      "loss": 0.0936,
+      "step": 435
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 6.627450980392157e-05,
+      "loss": 0.1102,
+      "step": 436
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 6.619607843137255e-05,
+      "loss": 0.1288,
+      "step": 437
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 6.611764705882353e-05,
+      "loss": 0.0602,
+      "step": 438
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.603921568627451e-05,
+      "loss": 0.1057,
+      "step": 439
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 6.59607843137255e-05,
+      "loss": 0.1349,
+      "step": 440
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 6.588235294117648e-05,
+      "loss": 0.0836,
+      "step": 441
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 6.580392156862746e-05,
+      "loss": 0.0958,
+      "step": 442
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 6.572549019607843e-05,
+      "loss": 0.1139,
+      "step": 443
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 6.564705882352941e-05,
+      "loss": 0.074,
+      "step": 444
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 6.55686274509804e-05,
+      "loss": 0.1084,
+      "step": 445
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 6.549019607843138e-05,
+      "loss": 0.078,
+      "step": 446
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 6.541176470588236e-05,
+      "loss": 0.1003,
+      "step": 447
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 6.533333333333334e-05,
+      "loss": 0.0816,
+      "step": 448
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 6.525490196078432e-05,
+      "loss": 0.087,
+      "step": 449
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 6.517647058823529e-05,
+      "loss": 0.0538,
+      "step": 450
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 6.509803921568627e-05,
+      "loss": 0.0555,
+      "step": 451
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 6.501960784313726e-05,
+      "loss": 0.0659,
+      "step": 452
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 6.494117647058824e-05,
+      "loss": 0.0625,
+      "step": 453
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 6.486274509803922e-05,
+      "loss": 0.0506,
+      "step": 454
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 6.47843137254902e-05,
+      "loss": 0.0656,
+      "step": 455
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 6.470588235294118e-05,
+      "loss": 0.0781,
+      "step": 456
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 6.462745098039215e-05,
+      "loss": 0.1002,
+      "step": 457
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 6.454901960784314e-05,
+      "loss": 0.0709,
+      "step": 458
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 6.447058823529412e-05,
+      "loss": 0.0796,
+      "step": 459
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 6.43921568627451e-05,
+      "loss": 0.0697,
+      "step": 460
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 6.431372549019608e-05,
+      "loss": 0.0808,
+      "step": 461
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 6.423529411764706e-05,
+      "loss": 0.1054,
+      "step": 462
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 6.415686274509805e-05,
+      "loss": 0.0875,
+      "step": 463
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 6.407843137254901e-05,
+      "loss": 0.051,
+      "step": 464
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 0.0669,
+      "step": 465
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 6.392156862745098e-05,
+      "loss": 0.0985,
+      "step": 466
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 6.384313725490196e-05,
+      "loss": 0.0567,
+      "step": 467
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 6.376470588235294e-05,
+      "loss": 0.0576,
+      "step": 468
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 6.368627450980393e-05,
+      "loss": 0.0712,
+      "step": 469
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 6.360784313725491e-05,
+      "loss": 0.0558,
+      "step": 470
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 6.352941176470588e-05,
+      "loss": 0.0457,
+      "step": 471
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 6.345098039215687e-05,
+      "loss": 0.0757,
+      "step": 472
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 6.337254901960784e-05,
+      "loss": 0.0666,
+      "step": 473
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 6.329411764705882e-05,
+      "loss": 0.0644,
+      "step": 474
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 6.32156862745098e-05,
+      "loss": 0.0672,
+      "step": 475
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 6.313725490196079e-05,
+      "loss": 0.0842,
+      "step": 476
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.305882352941177e-05,
+      "loss": 0.0792,
+      "step": 477
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.298039215686274e-05,
+      "loss": 0.077,
+      "step": 478
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.290196078431373e-05,
+      "loss": 0.0881,
+      "step": 479
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.282352941176472e-05,
+      "loss": 0.1171,
+      "step": 480
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.274509803921569e-05,
+      "loss": 0.0952,
+      "step": 481
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 6.266666666666667e-05,
+      "loss": 0.0664,
+      "step": 482
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 6.258823529411765e-05,
+      "loss": 0.0733,
+      "step": 483
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 6.250980392156863e-05,
+      "loss": 0.0799,
+      "step": 484
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 6.24313725490196e-05,
+      "loss": 0.0906,
+      "step": 485
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 6.23529411764706e-05,
+      "loss": 0.0866,
+      "step": 486
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 6.227450980392158e-05,
+      "loss": 0.0813,
+      "step": 487
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 6.219607843137255e-05,
+      "loss": 0.0751,
+      "step": 488
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 6.211764705882353e-05,
+      "loss": 0.0756,
+      "step": 489
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 6.203921568627451e-05,
+      "loss": 0.0871,
+      "step": 490
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 6.19607843137255e-05,
+      "loss": 0.0675,
+      "step": 491
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 6.188235294117646e-05,
+      "loss": 0.0947,
+      "step": 492
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 6.180392156862746e-05,
+      "loss": 0.0586,
+      "step": 493
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 6.172549019607844e-05,
+      "loss": 0.0768,
+      "step": 494
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 6.164705882352941e-05,
+      "loss": 0.0964,
+      "step": 495
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 6.15686274509804e-05,
+      "loss": 0.0734,
+      "step": 496
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 6.149019607843137e-05,
+      "loss": 0.0709,
+      "step": 497
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 6.141176470588236e-05,
+      "loss": 0.0956,
+      "step": 498
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 6.133333333333334e-05,
+      "loss": 0.0555,
+      "step": 499
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 6.125490196078432e-05,
+      "loss": 0.0863,
+      "step": 500
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 6.11764705882353e-05,
+      "loss": 0.0852,
+      "step": 501
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 6.109803921568627e-05,
+      "loss": 0.0923,
+      "step": 502
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.101960784313726e-05,
+      "loss": 0.0642,
+      "step": 503
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.0941176470588235e-05,
+      "loss": 0.0854,
+      "step": 504
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.086274509803922e-05,
+      "loss": 0.1017,
+      "step": 505
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.078431372549019e-05,
+      "loss": 0.0623,
+      "step": 506
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.070588235294118e-05,
+      "loss": 0.0653,
+      "step": 507
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.0627450980392164e-05,
+      "loss": 0.0615,
+      "step": 508
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.054901960784314e-05,
+      "loss": 0.0555,
+      "step": 509
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.047058823529412e-05,
+      "loss": 0.0695,
+      "step": 510
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.03921568627451e-05,
+      "loss": 0.0657,
+      "step": 511
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.031372549019608e-05,
+      "loss": 0.0567,
+      "step": 512
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.0235294117647055e-05,
+      "loss": 0.0469,
+      "step": 513
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.0156862745098044e-05,
+      "loss": 0.0821,
+      "step": 514
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.0078431372549026e-05,
+      "loss": 0.0798,
+      "step": 515
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6e-05,
+      "loss": 0.0451,
+      "step": 516
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 5.9921568627450984e-05,
+      "loss": 0.071,
+      "step": 517
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 5.984313725490196e-05,
+      "loss": 0.0449,
+      "step": 518
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 5.976470588235295e-05,
+      "loss": 0.0646,
+      "step": 519
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 5.968627450980392e-05,
+      "loss": 0.0479,
+      "step": 520
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 5.9607843137254906e-05,
+      "loss": 0.0506,
+      "step": 521
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.952941176470589e-05,
+      "loss": 0.0516,
+      "step": 522
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.9450980392156863e-05,
+      "loss": 0.0585,
+      "step": 523
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.9372549019607846e-05,
+      "loss": 0.0499,
+      "step": 524
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.929411764705882e-05,
+      "loss": 0.0736,
+      "step": 525
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.921568627450981e-05,
+      "loss": 0.0645,
+      "step": 526
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.913725490196079e-05,
+      "loss": 0.0682,
+      "step": 527
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.905882352941177e-05,
+      "loss": 0.051,
+      "step": 528
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.898039215686275e-05,
+      "loss": 0.0592,
+      "step": 529
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.8901960784313725e-05,
+      "loss": 0.0522,
+      "step": 530
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.882352941176471e-05,
+      "loss": 0.0438,
+      "step": 531
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.874509803921568e-05,
+      "loss": 0.0976,
+      "step": 532
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.866666666666667e-05,
+      "loss": 0.0863,
+      "step": 533
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.8588235294117654e-05,
+      "loss": 0.0542,
+      "step": 534
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.850980392156863e-05,
+      "loss": 0.0897,
+      "step": 535
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.843137254901961e-05,
+      "loss": 0.0562,
+      "step": 536
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.835294117647059e-05,
+      "loss": 0.0742,
+      "step": 537
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.8274509803921576e-05,
+      "loss": 0.0822,
+      "step": 538
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.8196078431372545e-05,
+      "loss": 0.0601,
+      "step": 539
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.8117647058823534e-05,
+      "loss": 0.0533,
+      "step": 540
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.8039215686274516e-05,
+      "loss": 0.0837,
+      "step": 541
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 5.796078431372549e-05,
+      "loss": 0.082,
+      "step": 542
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 5.7882352941176474e-05,
+      "loss": 0.0879,
+      "step": 543
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 5.780392156862745e-05,
+      "loss": 0.0617,
+      "step": 544
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 5.772549019607844e-05,
+      "loss": 0.0669,
+      "step": 545
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 5.764705882352941e-05,
+      "loss": 0.0632,
+      "step": 546
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 5.7568627450980396e-05,
+      "loss": 0.0466,
+      "step": 547
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.749019607843138e-05,
+      "loss": 0.0684,
+      "step": 548
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 5.741176470588235e-05,
+      "loss": 0.0884,
+      "step": 549
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 5.7333333333333336e-05,
+      "loss": 0.0586,
+      "step": 550
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 5.725490196078431e-05,
+      "loss": 0.0677,
+      "step": 551
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 5.71764705882353e-05,
+      "loss": 0.0552,
+      "step": 552
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 5.7098039215686275e-05,
+      "loss": 0.0552,
+      "step": 553
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 5.701960784313726e-05,
+      "loss": 0.0685,
+      "step": 554
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 5.694117647058824e-05,
+      "loss": 0.0601,
+      "step": 555
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 5.6862745098039215e-05,
+      "loss": 0.0632,
+      "step": 556
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 5.6784313725490204e-05,
+      "loss": 0.0621,
+      "step": 557
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 5.670588235294117e-05,
+      "loss": 0.1017,
+      "step": 558
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 5.662745098039216e-05,
+      "loss": 0.0499,
+      "step": 559
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 5.654901960784314e-05,
+      "loss": 0.0794,
+      "step": 560
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 5.647058823529412e-05,
+      "loss": 0.0528,
+      "step": 561
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 5.63921568627451e-05,
+      "loss": 0.0799,
+      "step": 562
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.631372549019608e-05,
+      "loss": 0.0474,
+      "step": 563
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 5.6235294117647066e-05,
+      "loss": 0.0947,
+      "step": 564
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.6156862745098035e-05,
+      "loss": 0.1417,
+      "step": 565
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.6078431372549024e-05,
+      "loss": 0.0851,
+      "step": 566
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 0.0555,
+      "step": 567
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.592156862745098e-05,
+      "loss": 0.0836,
+      "step": 568
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.584313725490197e-05,
+      "loss": 0.0557,
+      "step": 569
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.576470588235294e-05,
+      "loss": 0.0672,
+      "step": 570
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.568627450980393e-05,
+      "loss": 0.0702,
+      "step": 571
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.5607843137254903e-05,
+      "loss": 0.0734,
+      "step": 572
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.5529411764705886e-05,
+      "loss": 0.0902,
+      "step": 573
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.545098039215687e-05,
+      "loss": 0.0567,
+      "step": 574
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.537254901960784e-05,
+      "loss": 0.0872,
+      "step": 575
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.529411764705883e-05,
+      "loss": 0.0692,
+      "step": 576
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.52156862745098e-05,
+      "loss": 0.0549,
+      "step": 577
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.513725490196079e-05,
+      "loss": 0.0553,
+      "step": 578
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.5058823529411765e-05,
+      "loss": 0.0362,
+      "step": 579
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 5.498039215686275e-05,
+      "loss": 0.0774,
+      "step": 580
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.490196078431373e-05,
+      "loss": 0.0563,
+      "step": 581
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 5.4823529411764705e-05,
+      "loss": 0.1038,
+      "step": 582
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 5.4745098039215694e-05,
+      "loss": 0.0538,
+      "step": 583
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 5.466666666666666e-05,
+      "loss": 0.0579,
+      "step": 584
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 5.458823529411765e-05,
+      "loss": 0.1042,
+      "step": 585
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 5.450980392156863e-05,
+      "loss": 0.052,
+      "step": 586
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 5.443137254901961e-05,
+      "loss": 0.0477,
+      "step": 587
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 5.43529411764706e-05,
+      "loss": 0.0354,
+      "step": 588
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 5.427450980392157e-05,
+      "loss": 0.069,
+      "step": 589
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 5.4196078431372556e-05,
+      "loss": 0.0626,
+      "step": 590
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 5.411764705882353e-05,
+      "loss": 0.0554,
+      "step": 591
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 5.4039215686274514e-05,
+      "loss": 0.0442,
+      "step": 592
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 5.396078431372549e-05,
+      "loss": 0.0602,
+      "step": 593
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 5.388235294117647e-05,
+      "loss": 0.0663,
+      "step": 594
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 5.380392156862746e-05,
+      "loss": 0.0531,
+      "step": 595
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 5.372549019607843e-05,
+      "loss": 0.0575,
+      "step": 596
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 5.364705882352942e-05,
+      "loss": 0.063,
+      "step": 597
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 5.356862745098039e-05,
+      "loss": 0.074,
+      "step": 598
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 5.3490196078431376e-05,
+      "loss": 0.0354,
+      "step": 599
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 5.341176470588235e-05,
+      "loss": 0.0549,
+      "step": 600
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 5.333333333333333e-05,
+      "loss": 0.0586,
+      "step": 601
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 5.325490196078432e-05,
+      "loss": 0.0609,
+      "step": 602
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 5.31764705882353e-05,
+      "loss": 0.0413,
+      "step": 603
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 5.309803921568628e-05,
+      "loss": 0.0676,
+      "step": 604
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 5.3019607843137255e-05,
+      "loss": 0.065,
+      "step": 605
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 5.294117647058824e-05,
+      "loss": 0.0575,
+      "step": 606
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 5.2862745098039226e-05,
+      "loss": 0.0537,
+      "step": 607
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 5.2784313725490195e-05,
+      "loss": 0.0468,
+      "step": 608
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 5.2705882352941184e-05,
+      "loss": 0.0622,
+      "step": 609
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 5.262745098039216e-05,
+      "loss": 0.0755,
+      "step": 610
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 5.254901960784314e-05,
+      "loss": 0.0568,
+      "step": 611
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 5.247058823529412e-05,
+      "loss": 0.0638,
+      "step": 612
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 5.23921568627451e-05,
+      "loss": 0.0634,
+      "step": 613
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 5.231372549019609e-05,
+      "loss": 0.082,
+      "step": 614
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 5.223529411764706e-05,
+      "loss": 0.0673,
+      "step": 615
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 5.2156862745098046e-05,
+      "loss": 0.1286,
+      "step": 616
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 5.207843137254902e-05,
+      "loss": 0.0532,
+      "step": 617
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 0.0546,
+      "step": 618
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.192156862745098e-05,
+      "loss": 0.0459,
+      "step": 619
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.184313725490196e-05,
+      "loss": 0.0463,
+      "step": 620
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 5.176470588235295e-05,
+      "loss": 0.126,
+      "step": 621
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 5.1686274509803926e-05,
+      "loss": 0.0492,
+      "step": 622
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 5.160784313725491e-05,
+      "loss": 0.0494,
+      "step": 623
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 5.152941176470588e-05,
+      "loss": 0.0819,
+      "step": 624
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 5.1450980392156865e-05,
+      "loss": 0.0786,
+      "step": 625
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 5.137254901960784e-05,
+      "loss": 0.0599,
+      "step": 626
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 5.129411764705882e-05,
+      "loss": 0.0563,
+      "step": 627
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 5.121568627450981e-05,
+      "loss": 0.0619,
+      "step": 628
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.113725490196079e-05,
+      "loss": 0.0744,
+      "step": 629
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.105882352941177e-05,
+      "loss": 0.0701,
+      "step": 630
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 5.0980392156862745e-05,
+      "loss": 0.0609,
+      "step": 631
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.090196078431373e-05,
+      "loss": 0.0693,
+      "step": 632
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.08235294117647e-05,
+      "loss": 0.0562,
+      "step": 633
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 5.0745098039215685e-05,
+      "loss": 0.0756,
+      "step": 634
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 5.0666666666666674e-05,
+      "loss": 0.0489,
+      "step": 635
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 5.058823529411765e-05,
+      "loss": 0.0598,
+      "step": 636
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 5.050980392156863e-05,
+      "loss": 0.0418,
+      "step": 637
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 5.043137254901961e-05,
+      "loss": 0.0446,
+      "step": 638
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 5.035294117647059e-05,
+      "loss": 0.077,
+      "step": 639
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 5.0274509803921565e-05,
+      "loss": 0.1131,
+      "step": 640
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 5.0196078431372554e-05,
+      "loss": 0.0338,
+      "step": 641
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 5.0117647058823536e-05,
+      "loss": 0.0516,
+      "step": 642
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 5.003921568627451e-05,
+      "loss": 0.0434,
+      "step": 643
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.9960784313725493e-05,
+      "loss": 0.0629,
+      "step": 644
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.9882352941176476e-05,
+      "loss": 0.0423,
+      "step": 645
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.980392156862745e-05,
+      "loss": 0.0541,
+      "step": 646
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.972549019607843e-05,
+      "loss": 0.0522,
+      "step": 647
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.9647058823529416e-05,
+      "loss": 0.0839,
+      "step": 648
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.956862745098039e-05,
+      "loss": 0.0392,
+      "step": 649
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.949019607843137e-05,
+      "loss": 0.0433,
+      "step": 650
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.9411764705882355e-05,
+      "loss": 0.056,
+      "step": 651
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.933333333333334e-05,
+      "loss": 0.0427,
+      "step": 652
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.925490196078432e-05,
+      "loss": 0.0471,
+      "step": 653
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.9176470588235295e-05,
+      "loss": 0.0541,
+      "step": 654
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.909803921568628e-05,
+      "loss": 0.0613,
+      "step": 655
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.901960784313725e-05,
+      "loss": 0.0566,
+      "step": 656
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.894117647058824e-05,
+      "loss": 0.0745,
+      "step": 657
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.886274509803922e-05,
+      "loss": 0.0708,
+      "step": 658
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.87843137254902e-05,
+      "loss": 0.07,
+      "step": 659
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.870588235294118e-05,
+      "loss": 0.0681,
+      "step": 660
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.862745098039216e-05,
+      "loss": 0.0704,
+      "step": 661
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.854901960784314e-05,
+      "loss": 0.0619,
+      "step": 662
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.8470588235294115e-05,
+      "loss": 0.0437,
+      "step": 663
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.8392156862745104e-05,
+      "loss": 0.047,
+      "step": 664
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.831372549019608e-05,
+      "loss": 0.1095,
+      "step": 665
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.823529411764706e-05,
+      "loss": 0.0426,
+      "step": 666
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.8156862745098044e-05,
+      "loss": 0.0502,
+      "step": 667
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.807843137254902e-05,
+      "loss": 0.0396,
+      "step": 668
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.8e-05,
+      "loss": 0.0835,
+      "step": 669
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.7921568627450983e-05,
+      "loss": 0.0472,
+      "step": 670
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.7843137254901966e-05,
+      "loss": 0.0834,
+      "step": 671
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.776470588235295e-05,
+      "loss": 0.0429,
+      "step": 672
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.768627450980392e-05,
+      "loss": 0.0429,
+      "step": 673
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.7607843137254905e-05,
+      "loss": 0.0537,
+      "step": 674
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.752941176470588e-05,
+      "loss": 0.0474,
+      "step": 675
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.745098039215686e-05,
+      "loss": 0.0525,
+      "step": 676
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.7372549019607845e-05,
+      "loss": 0.0798,
+      "step": 677
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.729411764705883e-05,
+      "loss": 0.0819,
+      "step": 678
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.721568627450981e-05,
+      "loss": 0.0503,
+      "step": 679
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.7137254901960785e-05,
+      "loss": 0.053,
+      "step": 680
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.705882352941177e-05,
+      "loss": 0.0612,
+      "step": 681
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.698039215686274e-05,
+      "loss": 0.0448,
+      "step": 682
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.6901960784313725e-05,
+      "loss": 0.0487,
+      "step": 683
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.682352941176471e-05,
+      "loss": 0.0414,
+      "step": 684
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.674509803921569e-05,
+      "loss": 0.0611,
+      "step": 685
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.666666666666667e-05,
+      "loss": 0.0607,
+      "step": 686
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.658823529411765e-05,
+      "loss": 0.0594,
+      "step": 687
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.650980392156863e-05,
+      "loss": 0.0504,
+      "step": 688
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.643137254901961e-05,
+      "loss": 0.0699,
+      "step": 689
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.635294117647059e-05,
+      "loss": 0.0461,
+      "step": 690
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.6274509803921576e-05,
+      "loss": 0.0645,
+      "step": 691
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.619607843137255e-05,
+      "loss": 0.0514,
+      "step": 692
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.6117647058823534e-05,
+      "loss": 0.0582,
+      "step": 693
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.603921568627451e-05,
+      "loss": 0.082,
+      "step": 694
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.596078431372549e-05,
+      "loss": 0.0475,
+      "step": 695
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.588235294117647e-05,
+      "loss": 0.0748,
+      "step": 696
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.5803921568627456e-05,
+      "loss": 0.0663,
+      "step": 697
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.572549019607844e-05,
+      "loss": 0.0497,
+      "step": 698
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.564705882352941e-05,
+      "loss": 0.0856,
+      "step": 699
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.5568627450980395e-05,
+      "loss": 0.0435,
+      "step": 700
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.549019607843137e-05,
+      "loss": 0.0732,
+      "step": 701
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.541176470588235e-05,
+      "loss": 0.0553,
+      "step": 702
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.5333333333333335e-05,
+      "loss": 0.1174,
+      "step": 703
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.525490196078432e-05,
+      "loss": 0.0467,
+      "step": 704
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.51764705882353e-05,
+      "loss": 0.0648,
+      "step": 705
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.5098039215686275e-05,
+      "loss": 0.066,
+      "step": 706
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.501960784313726e-05,
+      "loss": 0.052,
+      "step": 707
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.494117647058824e-05,
+      "loss": 0.0776,
+      "step": 708
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.4862745098039215e-05,
+      "loss": 0.0635,
+      "step": 709
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.47843137254902e-05,
+      "loss": 0.0474,
+      "step": 710
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.470588235294118e-05,
+      "loss": 0.0355,
+      "step": 711
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.462745098039216e-05,
+      "loss": 0.051,
+      "step": 712
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.454901960784314e-05,
+      "loss": 0.0611,
+      "step": 713
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.447058823529412e-05,
+      "loss": 0.0612,
+      "step": 714
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.43921568627451e-05,
+      "loss": 0.0428,
+      "step": 715
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.431372549019608e-05,
+      "loss": 0.0454,
+      "step": 716
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.4235294117647066e-05,
+      "loss": 0.0399,
+      "step": 717
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.415686274509804e-05,
+      "loss": 0.04,
+      "step": 718
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.4078431372549023e-05,
+      "loss": 0.0482,
+      "step": 719
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.0399,
+      "step": 720
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.392156862745098e-05,
+      "loss": 0.0835,
+      "step": 721
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.384313725490196e-05,
+      "loss": 0.0605,
+      "step": 722
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.376470588235294e-05,
+      "loss": 0.0479,
+      "step": 723
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.368627450980393e-05,
+      "loss": 0.0504,
+      "step": 724
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.36078431372549e-05,
+      "loss": 0.0661,
+      "step": 725
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.3529411764705885e-05,
+      "loss": 0.0469,
+      "step": 726
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.345098039215687e-05,
+      "loss": 0.0678,
+      "step": 727
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.337254901960784e-05,
+      "loss": 0.053,
+      "step": 728
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.3294117647058825e-05,
+      "loss": 0.043,
+      "step": 729
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.32156862745098e-05,
+      "loss": 0.0586,
+      "step": 730
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.313725490196079e-05,
+      "loss": 0.0497,
+      "step": 731
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.3058823529411765e-05,
+      "loss": 0.0374,
+      "step": 732
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.298039215686275e-05,
+      "loss": 0.0602,
+      "step": 733
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.290196078431373e-05,
+      "loss": 0.0548,
+      "step": 734
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.2823529411764705e-05,
+      "loss": 0.0409,
+      "step": 735
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.274509803921569e-05,
+      "loss": 0.0632,
+      "step": 736
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.266666666666667e-05,
+      "loss": 0.0412,
+      "step": 737
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.258823529411765e-05,
+      "loss": 0.0533,
+      "step": 738
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.2509803921568634e-05,
+      "loss": 0.0905,
+      "step": 739
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.243137254901961e-05,
+      "loss": 0.0654,
+      "step": 740
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.235294117647059e-05,
+      "loss": 0.0802,
+      "step": 741
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.227450980392157e-05,
+      "loss": 0.0519,
+      "step": 742
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.219607843137255e-05,
+      "loss": 0.0435,
+      "step": 743
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.211764705882353e-05,
+      "loss": 0.0563,
+      "step": 744
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.203921568627451e-05,
+      "loss": 0.0491,
+      "step": 745
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.1960784313725496e-05,
+      "loss": 0.0433,
+      "step": 746
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.188235294117647e-05,
+      "loss": 0.0562,
+      "step": 747
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.180392156862745e-05,
+      "loss": 0.0448,
+      "step": 748
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.172549019607843e-05,
+      "loss": 0.0501,
+      "step": 749
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.164705882352941e-05,
+      "loss": 0.0533,
+      "step": 750
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.156862745098039e-05,
+      "loss": 0.0518,
+      "step": 751
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.1490196078431375e-05,
+      "loss": 0.0417,
+      "step": 752
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.141176470588236e-05,
+      "loss": 0.0468,
+      "step": 753
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.133333333333333e-05,
+      "loss": 0.0702,
+      "step": 754
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.1254901960784315e-05,
+      "loss": 0.0519,
+      "step": 755
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.11764705882353e-05,
+      "loss": 0.0476,
+      "step": 756
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.109803921568628e-05,
+      "loss": 0.049,
+      "step": 757
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.101960784313726e-05,
+      "loss": 0.0941,
+      "step": 758
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.094117647058824e-05,
+      "loss": 0.0489,
+      "step": 759
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.086274509803922e-05,
+      "loss": 0.062,
+      "step": 760
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.0784313725490195e-05,
+      "loss": 0.0403,
+      "step": 761
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.070588235294118e-05,
+      "loss": 0.0465,
+      "step": 762
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.062745098039216e-05,
+      "loss": 0.0526,
+      "step": 763
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.054901960784314e-05,
+      "loss": 0.0436,
+      "step": 764
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.0470588235294124e-05,
+      "loss": 0.0471,
+      "step": 765
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.03921568627451e-05,
+      "loss": 0.0514,
+      "step": 766
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.031372549019608e-05,
+      "loss": 0.0479,
+      "step": 767
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.023529411764706e-05,
+      "loss": 0.0607,
+      "step": 768
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.015686274509804e-05,
+      "loss": 0.0591,
+      "step": 769
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.007843137254902e-05,
+      "loss": 0.0424,
+      "step": 770
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4e-05,
+      "loss": 0.0605,
+      "step": 771
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 3.9921568627450985e-05,
+      "loss": 0.0627,
+      "step": 772
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 3.984313725490196e-05,
+      "loss": 0.0512,
+      "step": 773
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 3.976470588235294e-05,
+      "loss": 0.0423,
+      "step": 774
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 3.9686274509803925e-05,
+      "loss": 0.0437,
+      "step": 775
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 3.96078431372549e-05,
+      "loss": 0.0465,
+      "step": 776
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 3.952941176470588e-05,
+      "loss": 0.0399,
+      "step": 777
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 3.9450980392156865e-05,
+      "loss": 0.0484,
+      "step": 778
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 3.937254901960785e-05,
+      "loss": 0.048,
+      "step": 779
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 3.929411764705882e-05,
+      "loss": 0.0532,
+      "step": 780
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 3.9215686274509805e-05,
+      "loss": 0.0392,
+      "step": 781
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 3.913725490196079e-05,
+      "loss": 0.038,
+      "step": 782
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 3.905882352941176e-05,
+      "loss": 0.0395,
+      "step": 783
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 3.898039215686275e-05,
+      "loss": 0.051,
+      "step": 784
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 3.890196078431373e-05,
+      "loss": 0.0596,
+      "step": 785
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 3.882352941176471e-05,
+      "loss": 0.0446,
+      "step": 786
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 3.874509803921569e-05,
+      "loss": 0.0583,
+      "step": 787
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 3.866666666666667e-05,
+      "loss": 0.0448,
+      "step": 788
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 3.858823529411765e-05,
+      "loss": 0.0463,
+      "step": 789
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 3.8509803921568625e-05,
+      "loss": 0.0527,
+      "step": 790
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 3.8431372549019614e-05,
+      "loss": 0.0428,
+      "step": 791
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 3.835294117647059e-05,
+      "loss": 0.0586,
+      "step": 792
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 3.827450980392157e-05,
+      "loss": 0.0553,
+      "step": 793
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 3.819607843137255e-05,
+      "loss": 0.039,
+      "step": 794
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 3.811764705882353e-05,
+      "loss": 0.0443,
+      "step": 795
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 3.803921568627451e-05,
+      "loss": 0.0329,
+      "step": 796
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 3.7960784313725486e-05,
+      "loss": 0.0478,
+      "step": 797
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 3.7882352941176475e-05,
+      "loss": 0.0648,
+      "step": 798
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 3.780392156862745e-05,
+      "loss": 0.0397,
+      "step": 799
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 3.772549019607843e-05,
+      "loss": 0.0355,
+      "step": 800
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 3.7647058823529415e-05,
+      "loss": 0.0588,
+      "step": 801
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 3.756862745098039e-05,
+      "loss": 0.0716,
+      "step": 802
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 3.749019607843137e-05,
+      "loss": 0.0434,
+      "step": 803
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 3.7411764705882355e-05,
+      "loss": 0.0493,
+      "step": 804
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 3.733333333333334e-05,
+      "loss": 0.0704,
+      "step": 805
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 3.725490196078432e-05,
+      "loss": 0.0545,
+      "step": 806
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 3.7176470588235295e-05,
+      "loss": 0.0571,
+      "step": 807
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 3.709803921568628e-05,
+      "loss": 0.0467,
+      "step": 808
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 3.701960784313725e-05,
+      "loss": 0.0463,
+      "step": 809
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 3.6941176470588235e-05,
+      "loss": 0.0711,
+      "step": 810
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 3.686274509803922e-05,
+      "loss": 0.0469,
+      "step": 811
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 3.67843137254902e-05,
+      "loss": 0.0601,
+      "step": 812
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 3.670588235294118e-05,
+      "loss": 0.0468,
+      "step": 813
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 3.662745098039216e-05,
+      "loss": 0.0563,
+      "step": 814
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 3.654901960784314e-05,
+      "loss": 0.0761,
+      "step": 815
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 3.6470588235294114e-05,
+      "loss": 0.0429,
+      "step": 816
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 3.63921568627451e-05,
+      "loss": 0.0437,
+      "step": 817
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 3.631372549019608e-05,
+      "loss": 0.0674,
+      "step": 818
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 3.623529411764706e-05,
+      "loss": 0.049,
+      "step": 819
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 3.615686274509804e-05,
+      "loss": 0.07,
+      "step": 820
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 3.607843137254902e-05,
+      "loss": 0.0645,
+      "step": 821
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 3.6e-05,
+      "loss": 0.0534,
+      "step": 822
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 3.592156862745098e-05,
+      "loss": 0.0496,
+      "step": 823
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 3.5843137254901965e-05,
+      "loss": 0.0679,
+      "step": 824
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 3.576470588235295e-05,
+      "loss": 0.0456,
+      "step": 825
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 3.568627450980392e-05,
+      "loss": 0.0478,
+      "step": 826
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 3.5607843137254905e-05,
+      "loss": 0.0435,
+      "step": 827
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 3.552941176470588e-05,
+      "loss": 0.0389,
+      "step": 828
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 3.545098039215686e-05,
+      "loss": 0.0913,
+      "step": 829
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 3.5372549019607845e-05,
+      "loss": 0.0511,
+      "step": 830
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 3.529411764705883e-05,
+      "loss": 0.081,
+      "step": 831
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 3.521568627450981e-05,
+      "loss": 0.0538,
+      "step": 832
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 3.5137254901960785e-05,
+      "loss": 0.0506,
+      "step": 833
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 3.505882352941177e-05,
+      "loss": 0.0426,
+      "step": 834
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 3.498039215686274e-05,
+      "loss": 0.1086,
+      "step": 835
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 3.4901960784313725e-05,
+      "loss": 0.052,
+      "step": 836
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 3.482352941176471e-05,
+      "loss": 0.0341,
+      "step": 837
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 3.474509803921569e-05,
+      "loss": 0.0433,
+      "step": 838
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 3.466666666666667e-05,
+      "loss": 0.0515,
+      "step": 839
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 3.458823529411765e-05,
+      "loss": 0.0553,
+      "step": 840
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 3.450980392156863e-05,
+      "loss": 0.0415,
+      "step": 841
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 3.443137254901961e-05,
+      "loss": 0.0579,
+      "step": 842
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 3.4352941176470587e-05,
+      "loss": 0.1101,
+      "step": 843
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 3.4274509803921576e-05,
+      "loss": 0.0539,
+      "step": 844
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 3.419607843137255e-05,
+      "loss": 0.0649,
+      "step": 845
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 3.411764705882353e-05,
+      "loss": 0.0646,
+      "step": 846
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 3.403921568627451e-05,
+      "loss": 0.0438,
+      "step": 847
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 3.396078431372549e-05,
+      "loss": 0.0436,
+      "step": 848
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 3.388235294117647e-05,
+      "loss": 0.0374,
+      "step": 849
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 3.380392156862745e-05,
+      "loss": 0.048,
+      "step": 850
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 3.372549019607844e-05,
+      "loss": 0.0857,
+      "step": 851
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 3.364705882352941e-05,
+      "loss": 0.0524,
+      "step": 852
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 3.3568627450980395e-05,
+      "loss": 0.0488,
+      "step": 853
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 3.349019607843138e-05,
+      "loss": 0.0445,
+      "step": 854
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 3.341176470588235e-05,
+      "loss": 0.051,
+      "step": 855
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.0363,
+      "step": 856
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 3.325490196078431e-05,
+      "loss": 0.0411,
+      "step": 857
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 3.31764705882353e-05,
+      "loss": 0.0446,
+      "step": 858
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 3.3098039215686275e-05,
+      "loss": 0.055,
+      "step": 859
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 3.301960784313726e-05,
+      "loss": 0.0405,
+      "step": 860
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 3.294117647058824e-05,
+      "loss": 0.0423,
+      "step": 861
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 3.2862745098039215e-05,
+      "loss": 0.0543,
+      "step": 862
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 3.27843137254902e-05,
+      "loss": 0.0372,
+      "step": 863
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 3.270588235294118e-05,
+      "loss": 0.0514,
+      "step": 864
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 3.262745098039216e-05,
+      "loss": 0.0678,
+      "step": 865
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 3.254901960784314e-05,
+      "loss": 0.0364,
+      "step": 866
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 3.247058823529412e-05,
+      "loss": 0.0438,
+      "step": 867
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 3.23921568627451e-05,
+      "loss": 0.0473,
+      "step": 868
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 3.2313725490196077e-05,
+      "loss": 0.0484,
+      "step": 869
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 3.223529411764706e-05,
+      "loss": 0.0376,
+      "step": 870
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 3.215686274509804e-05,
+      "loss": 0.0427,
+      "step": 871
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 3.207843137254902e-05,
+      "loss": 0.0509,
+      "step": 872
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.0471,
+      "step": 873
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 3.192156862745098e-05,
+      "loss": 0.0552,
+      "step": 874
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 3.184313725490196e-05,
+      "loss": 0.0604,
+      "step": 875
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 3.176470588235294e-05,
+      "loss": 0.0378,
+      "step": 876
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 3.168627450980392e-05,
+      "loss": 0.0722,
+      "step": 877
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 3.16078431372549e-05,
+      "loss": 0.05,
+      "step": 878
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 3.1529411764705885e-05,
+      "loss": 0.0897,
+      "step": 879
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 3.145098039215687e-05,
+      "loss": 0.0401,
+      "step": 880
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 3.137254901960784e-05,
+      "loss": 0.0764,
+      "step": 881
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 3.1294117647058825e-05,
+      "loss": 0.0402,
+      "step": 882
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 3.12156862745098e-05,
+      "loss": 0.0485,
+      "step": 883
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 3.113725490196079e-05,
+      "loss": 0.0345,
+      "step": 884
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 3.1058823529411765e-05,
+      "loss": 0.0638,
+      "step": 885
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 3.098039215686275e-05,
+      "loss": 0.0539,
+      "step": 886
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 3.090196078431373e-05,
+      "loss": 0.0623,
+      "step": 887
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 3.0823529411764705e-05,
+      "loss": 0.0652,
+      "step": 888
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 3.074509803921569e-05,
+      "loss": 0.0485,
+      "step": 889
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 3.066666666666667e-05,
+      "loss": 0.0447,
+      "step": 890
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 3.058823529411765e-05,
+      "loss": 0.0561,
+      "step": 891
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 3.050980392156863e-05,
+      "loss": 0.0482,
+      "step": 892
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 3.043137254901961e-05,
+      "loss": 0.0418,
+      "step": 893
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 3.035294117647059e-05,
+      "loss": 0.0408,
+      "step": 894
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 3.027450980392157e-05,
+      "loss": 0.077,
+      "step": 895
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 3.019607843137255e-05,
+      "loss": 0.0765,
+      "step": 896
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 3.0117647058823527e-05,
+      "loss": 0.0398,
+      "step": 897
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 3.0039215686274513e-05,
+      "loss": 0.0438,
+      "step": 898
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 2.9960784313725492e-05,
+      "loss": 0.0376,
+      "step": 899
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 2.9882352941176474e-05,
+      "loss": 0.0436,
+      "step": 900
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 2.9803921568627453e-05,
+      "loss": 0.0406,
+      "step": 901
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 2.9725490196078432e-05,
+      "loss": 0.0505,
+      "step": 902
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 2.964705882352941e-05,
+      "loss": 0.0395,
+      "step": 903
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 2.9568627450980396e-05,
+      "loss": 0.0501,
+      "step": 904
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 2.9490196078431375e-05,
+      "loss": 0.0416,
+      "step": 905
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 2.9411764705882354e-05,
+      "loss": 0.0434,
+      "step": 906
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 2.9333333333333336e-05,
+      "loss": 0.0306,
+      "step": 907
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 2.9254901960784315e-05,
+      "loss": 0.0532,
+      "step": 908
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 2.9176470588235294e-05,
+      "loss": 0.0591,
+      "step": 909
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 2.9098039215686272e-05,
+      "loss": 0.0385,
+      "step": 910
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 2.9019607843137258e-05,
+      "loss": 0.0428,
+      "step": 911
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 2.8941176470588237e-05,
+      "loss": 0.0423,
+      "step": 912
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 2.886274509803922e-05,
+      "loss": 0.043,
+      "step": 913
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 2.8784313725490198e-05,
+      "loss": 0.0733,
+      "step": 914
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 2.8705882352941177e-05,
+      "loss": 0.0401,
+      "step": 915
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 2.8627450980392155e-05,
+      "loss": 0.0435,
+      "step": 916
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 2.8549019607843138e-05,
+      "loss": 0.0607,
+      "step": 917
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 2.847058823529412e-05,
+      "loss": 0.064,
+      "step": 918
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 2.8392156862745102e-05,
+      "loss": 0.0344,
+      "step": 919
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 2.831372549019608e-05,
+      "loss": 0.0374,
+      "step": 920
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 2.823529411764706e-05,
+      "loss": 0.0406,
+      "step": 921
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 2.815686274509804e-05,
+      "loss": 0.0591,
+      "step": 922
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 2.8078431372549017e-05,
+      "loss": 0.0392,
+      "step": 923
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.0426,
+      "step": 924
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 2.7921568627450985e-05,
+      "loss": 0.0378,
+      "step": 925
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 2.7843137254901964e-05,
+      "loss": 0.0473,
+      "step": 926
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 2.7764705882352943e-05,
+      "loss": 0.0663,
+      "step": 927
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 2.768627450980392e-05,
+      "loss": 0.0465,
+      "step": 928
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 2.76078431372549e-05,
+      "loss": 0.0429,
+      "step": 929
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 2.7529411764705883e-05,
+      "loss": 0.0575,
+      "step": 930
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 2.7450980392156865e-05,
+      "loss": 0.0442,
+      "step": 931
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 2.7372549019607847e-05,
+      "loss": 0.0725,
+      "step": 932
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 2.7294117647058826e-05,
+      "loss": 0.0589,
+      "step": 933
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 2.7215686274509805e-05,
+      "loss": 0.0624,
+      "step": 934
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 2.7137254901960784e-05,
+      "loss": 0.0455,
+      "step": 935
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 2.7058823529411766e-05,
+      "loss": 0.0453,
+      "step": 936
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 2.6980392156862745e-05,
+      "loss": 0.0453,
+      "step": 937
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 2.690196078431373e-05,
+      "loss": 0.0576,
+      "step": 938
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 2.682352941176471e-05,
+      "loss": 0.0462,
+      "step": 939
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 2.6745098039215688e-05,
+      "loss": 0.0483,
+      "step": 940
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.0922,
+      "step": 941
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 2.658823529411765e-05,
+      "loss": 0.0685,
+      "step": 942
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 2.6509803921568628e-05,
+      "loss": 0.0532,
+      "step": 943
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 2.6431372549019613e-05,
+      "loss": 0.0418,
+      "step": 944
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 2.6352941176470592e-05,
+      "loss": 0.0765,
+      "step": 945
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 2.627450980392157e-05,
+      "loss": 0.041,
+      "step": 946
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 2.619607843137255e-05,
+      "loss": 0.0471,
+      "step": 947
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 2.611764705882353e-05,
+      "loss": 0.1111,
+      "step": 948
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 2.603921568627451e-05,
+      "loss": 0.0857,
+      "step": 949
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 2.596078431372549e-05,
+      "loss": 0.0546,
+      "step": 950
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 2.5882352941176475e-05,
+      "loss": 0.051,
+      "step": 951
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 2.5803921568627454e-05,
+      "loss": 0.0491,
+      "step": 952
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 2.5725490196078433e-05,
+      "loss": 0.0617,
+      "step": 953
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 2.564705882352941e-05,
+      "loss": 0.0468,
+      "step": 954
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 2.5568627450980394e-05,
+      "loss": 0.0421,
+      "step": 955
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 2.5490196078431373e-05,
+      "loss": 0.051,
+      "step": 956
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 2.541176470588235e-05,
+      "loss": 0.0453,
+      "step": 957
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 2.5333333333333337e-05,
+      "loss": 0.0553,
+      "step": 958
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 2.5254901960784316e-05,
+      "loss": 0.0531,
+      "step": 959
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 2.5176470588235295e-05,
+      "loss": 0.0643,
+      "step": 960
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 2.5098039215686277e-05,
+      "loss": 0.0464,
+      "step": 961
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 2.5019607843137256e-05,
+      "loss": 0.0465,
+      "step": 962
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 2.4941176470588238e-05,
+      "loss": 0.0463,
+      "step": 963
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 2.4862745098039217e-05,
+      "loss": 0.0597,
+      "step": 964
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 2.4784313725490195e-05,
+      "loss": 0.0521,
+      "step": 965
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 2.4705882352941178e-05,
+      "loss": 0.0415,
+      "step": 966
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 2.462745098039216e-05,
+      "loss": 0.0459,
+      "step": 967
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 2.454901960784314e-05,
+      "loss": 0.0312,
+      "step": 968
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 2.447058823529412e-05,
+      "loss": 0.0395,
+      "step": 969
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 2.43921568627451e-05,
+      "loss": 0.045,
+      "step": 970
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 2.431372549019608e-05,
+      "loss": 0.0593,
+      "step": 971
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 2.4235294117647057e-05,
+      "loss": 0.0417,
+      "step": 972
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 2.415686274509804e-05,
+      "loss": 0.0369,
+      "step": 973
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 2.4078431372549022e-05,
+      "loss": 0.054,
+      "step": 974
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 2.4e-05,
+      "loss": 0.0438,
+      "step": 975
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 2.3921568627450983e-05,
+      "loss": 0.0578,
+      "step": 976
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 2.384313725490196e-05,
+      "loss": 0.094,
+      "step": 977
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 2.376470588235294e-05,
+      "loss": 0.0467,
+      "step": 978
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 2.3686274509803923e-05,
+      "loss": 0.0388,
+      "step": 979
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 2.3607843137254905e-05,
+      "loss": 0.05,
+      "step": 980
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 2.3529411764705884e-05,
+      "loss": 0.0366,
+      "step": 981
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 2.3450980392156862e-05,
+      "loss": 0.0409,
+      "step": 982
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 2.3372549019607845e-05,
+      "loss": 0.0405,
+      "step": 983
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 2.3294117647058824e-05,
+      "loss": 0.0494,
+      "step": 984
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 2.3215686274509806e-05,
+      "loss": 0.0744,
+      "step": 985
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 2.3137254901960788e-05,
+      "loss": 0.0352,
+      "step": 986
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 2.3058823529411767e-05,
+      "loss": 0.0841,
+      "step": 987
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 2.2980392156862746e-05,
+      "loss": 0.0511,
+      "step": 988
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 2.2901960784313728e-05,
+      "loss": 0.0449,
+      "step": 989
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 2.2823529411764707e-05,
+      "loss": 0.0493,
+      "step": 990
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 2.2745098039215685e-05,
+      "loss": 0.0781,
+      "step": 991
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 2.2666666666666668e-05,
+      "loss": 0.0525,
+      "step": 992
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 2.258823529411765e-05,
+      "loss": 0.0431,
+      "step": 993
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 2.250980392156863e-05,
+      "loss": 0.042,
+      "step": 994
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 2.2431372549019607e-05,
+      "loss": 0.0542,
+      "step": 995
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 2.235294117647059e-05,
+      "loss": 0.0461,
+      "step": 996
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 2.227450980392157e-05,
+      "loss": 0.0546,
+      "step": 997
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 2.219607843137255e-05,
+      "loss": 0.1097,
+      "step": 998
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 2.2117647058823533e-05,
+      "loss": 0.0584,
+      "step": 999
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 2.2039215686274512e-05,
+      "loss": 0.0483,
+      "step": 1000
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 2.196078431372549e-05,
+      "loss": 0.048,
+      "step": 1001
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 2.188235294117647e-05,
+      "loss": 0.0493,
+      "step": 1002
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 2.180392156862745e-05,
+      "loss": 0.0444,
+      "step": 1003
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 2.1725490196078434e-05,
+      "loss": 0.0395,
+      "step": 1004
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 2.1647058823529413e-05,
+      "loss": 0.0463,
+      "step": 1005
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 2.1568627450980395e-05,
+      "loss": 0.0523,
+      "step": 1006
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 2.1490196078431374e-05,
+      "loss": 0.0493,
+      "step": 1007
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 2.1411764705882352e-05,
+      "loss": 0.0621,
+      "step": 1008
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 2.1333333333333335e-05,
+      "loss": 0.0446,
+      "step": 1009
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 2.1254901960784317e-05,
+      "loss": 0.0503,
+      "step": 1010
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 2.1176470588235296e-05,
+      "loss": 0.06,
+      "step": 1011
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 2.1098039215686274e-05,
+      "loss": 0.0472,
+      "step": 1012
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 2.1019607843137257e-05,
+      "loss": 0.0421,
+      "step": 1013
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 2.0941176470588235e-05,
+      "loss": 0.0572,
+      "step": 1014
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 2.0862745098039214e-05,
+      "loss": 0.0529,
+      "step": 1015
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 2.0784313725490197e-05,
+      "loss": 0.0591,
+      "step": 1016
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 2.070588235294118e-05,
+      "loss": 0.0732,
+      "step": 1017
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 2.0627450980392158e-05,
+      "loss": 0.04,
+      "step": 1018
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 2.054901960784314e-05,
+      "loss": 0.0587,
+      "step": 1019
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 2.047058823529412e-05,
+      "loss": 0.0392,
+      "step": 1020
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 2.0392156862745097e-05,
+      "loss": 0.0348,
+      "step": 1021
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 2.031372549019608e-05,
+      "loss": 0.0479,
+      "step": 1022
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 2.0235294117647062e-05,
+      "loss": 0.0422,
+      "step": 1023
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 2.015686274509804e-05,
+      "loss": 0.0364,
+      "step": 1024
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 2.007843137254902e-05,
+      "loss": 0.0753,
+      "step": 1025
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 2e-05,
+      "loss": 0.0433,
+      "step": 1026
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 1.992156862745098e-05,
+      "loss": 0.04,
+      "step": 1027
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 1.9843137254901963e-05,
+      "loss": 0.0355,
+      "step": 1028
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 1.976470588235294e-05,
+      "loss": 0.0468,
+      "step": 1029
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 1.9686274509803924e-05,
+      "loss": 0.0476,
+      "step": 1030
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 1.9607843137254903e-05,
+      "loss": 0.0487,
+      "step": 1031
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 1.952941176470588e-05,
+      "loss": 0.0393,
+      "step": 1032
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 1.9450980392156864e-05,
+      "loss": 0.0449,
+      "step": 1033
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 1.9372549019607846e-05,
+      "loss": 0.0458,
+      "step": 1034
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 1.9294117647058825e-05,
+      "loss": 0.0448,
+      "step": 1035
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 1.9215686274509807e-05,
+      "loss": 0.0496,
+      "step": 1036
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 1.9137254901960786e-05,
+      "loss": 0.0518,
+      "step": 1037
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 1.9058823529411764e-05,
+      "loss": 0.0496,
+      "step": 1038
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 1.8980392156862743e-05,
+      "loss": 0.0707,
+      "step": 1039
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 1.8901960784313725e-05,
+      "loss": 0.0492,
+      "step": 1040
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 1.8823529411764708e-05,
+      "loss": 0.0472,
+      "step": 1041
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 1.8745098039215686e-05,
+      "loss": 0.0593,
+      "step": 1042
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 0.0436,
+      "step": 1043
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 1.8588235294117647e-05,
+      "loss": 0.0423,
+      "step": 1044
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 1.8509803921568626e-05,
+      "loss": 0.0379,
+      "step": 1045
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 1.843137254901961e-05,
+      "loss": 0.0414,
+      "step": 1046
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 1.835294117647059e-05,
+      "loss": 0.0547,
+      "step": 1047
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 1.827450980392157e-05,
+      "loss": 0.0752,
+      "step": 1048
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 1.819607843137255e-05,
+      "loss": 0.0397,
+      "step": 1049
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 1.811764705882353e-05,
+      "loss": 0.045,
+      "step": 1050
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 1.803921568627451e-05,
+      "loss": 0.0432,
+      "step": 1051
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 1.796078431372549e-05,
+      "loss": 0.0502,
+      "step": 1052
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 1.7882352941176474e-05,
+      "loss": 0.0646,
+      "step": 1053
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 1.7803921568627453e-05,
+      "loss": 0.0459,
+      "step": 1054
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 1.772549019607843e-05,
+      "loss": 0.0607,
+      "step": 1055
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 1.7647058823529414e-05,
+      "loss": 0.0534,
+      "step": 1056
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 1.7568627450980392e-05,
+      "loss": 0.0393,
+      "step": 1057
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 1.749019607843137e-05,
+      "loss": 0.0543,
+      "step": 1058
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 1.7411764705882353e-05,
+      "loss": 0.0625,
+      "step": 1059
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 0.0352,
+      "step": 1060
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 1.7254901960784314e-05,
+      "loss": 0.0348,
+      "step": 1061
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 1.7176470588235293e-05,
+      "loss": 0.0396,
+      "step": 1062
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 1.7098039215686275e-05,
+      "loss": 0.0489,
+      "step": 1063
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 1.7019607843137254e-05,
+      "loss": 0.0737,
+      "step": 1064
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 1.6941176470588237e-05,
+      "loss": 0.0488,
+      "step": 1065
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 1.686274509803922e-05,
+      "loss": 0.0433,
+      "step": 1066
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 1.6784313725490198e-05,
+      "loss": 0.0658,
+      "step": 1067
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 1.6705882352941176e-05,
+      "loss": 0.055,
+      "step": 1068
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 1.6627450980392155e-05,
+      "loss": 0.0522,
+      "step": 1069
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 1.6549019607843137e-05,
+      "loss": 0.0587,
+      "step": 1070
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 1.647058823529412e-05,
+      "loss": 0.0428,
+      "step": 1071
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 1.63921568627451e-05,
+      "loss": 0.0492,
+      "step": 1072
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 1.631372549019608e-05,
+      "loss": 0.0311,
+      "step": 1073
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 1.623529411764706e-05,
+      "loss": 0.0485,
+      "step": 1074
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 1.6156862745098038e-05,
+      "loss": 0.0783,
+      "step": 1075
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 1.607843137254902e-05,
+      "loss": 0.0419,
+      "step": 1076
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.0474,
+      "step": 1077
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 1.592156862745098e-05,
+      "loss": 0.0405,
+      "step": 1078
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 1.584313725490196e-05,
+      "loss": 0.0443,
+      "step": 1079
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 1.5764705882352943e-05,
+      "loss": 0.0443,
+      "step": 1080
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 1.568627450980392e-05,
+      "loss": 0.0617,
+      "step": 1081
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 1.56078431372549e-05,
+      "loss": 0.0294,
+      "step": 1082
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 1.5529411764705882e-05,
+      "loss": 0.0495,
+      "step": 1083
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 1.5450980392156865e-05,
+      "loss": 0.0417,
+      "step": 1084
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 1.5372549019607843e-05,
+      "loss": 0.0554,
+      "step": 1085
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 1.5294117647058826e-05,
+      "loss": 0.0511,
+      "step": 1086
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 1.5215686274509804e-05,
+      "loss": 0.0406,
+      "step": 1087
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 1.5137254901960785e-05,
+      "loss": 0.0534,
+      "step": 1088
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 1.5058823529411764e-05,
+      "loss": 0.0571,
+      "step": 1089
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 1.4980392156862746e-05,
+      "loss": 0.0757,
+      "step": 1090
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 1.4901960784313726e-05,
+      "loss": 0.0464,
+      "step": 1091
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 1.4823529411764705e-05,
+      "loss": 0.0447,
+      "step": 1092
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 1.4745098039215687e-05,
+      "loss": 0.0598,
+      "step": 1093
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 1.4666666666666668e-05,
+      "loss": 0.0512,
+      "step": 1094
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 1.4588235294117647e-05,
+      "loss": 0.05,
+      "step": 1095
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 1.4509803921568629e-05,
+      "loss": 0.0309,
+      "step": 1096
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 1.443137254901961e-05,
+      "loss": 0.0361,
+      "step": 1097
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 1.4352941176470588e-05,
+      "loss": 0.0806,
+      "step": 1098
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 1.4274509803921569e-05,
+      "loss": 0.0357,
+      "step": 1099
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 1.4196078431372551e-05,
+      "loss": 0.047,
+      "step": 1100
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 1.411764705882353e-05,
+      "loss": 0.0388,
+      "step": 1101
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 1.4039215686274509e-05,
+      "loss": 0.0349,
+      "step": 1102
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 1.3960784313725493e-05,
+      "loss": 0.0473,
+      "step": 1103
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 1.3882352941176471e-05,
+      "loss": 0.0582,
+      "step": 1104
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 1.380392156862745e-05,
+      "loss": 0.0415,
+      "step": 1105
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 1.3725490196078432e-05,
+      "loss": 0.0387,
+      "step": 1106
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 1.3647058823529413e-05,
+      "loss": 0.075,
+      "step": 1107
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 1.3568627450980392e-05,
+      "loss": 0.0537,
+      "step": 1108
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 1.3490196078431372e-05,
+      "loss": 0.0473,
+      "step": 1109
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 1.3411764705882354e-05,
+      "loss": 0.0771,
+      "step": 1110
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.0423,
+      "step": 1111
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 1.3254901960784314e-05,
+      "loss": 0.0397,
+      "step": 1112
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 1.3176470588235296e-05,
+      "loss": 0.0547,
+      "step": 1113
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 1.3098039215686275e-05,
+      "loss": 0.0446,
+      "step": 1114
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 1.3019607843137255e-05,
+      "loss": 0.065,
+      "step": 1115
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 1.2941176470588238e-05,
+      "loss": 0.0513,
+      "step": 1116
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 1.2862745098039216e-05,
+      "loss": 0.0491,
+      "step": 1117
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 1.2784313725490197e-05,
+      "loss": 0.0415,
+      "step": 1118
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 1.2705882352941176e-05,
+      "loss": 0.0426,
+      "step": 1119
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 1.2627450980392158e-05,
+      "loss": 0.0457,
+      "step": 1120
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 1.2549019607843138e-05,
+      "loss": 0.0574,
+      "step": 1121
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 1.2470588235294119e-05,
+      "loss": 0.0468,
+      "step": 1122
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 1.2392156862745098e-05,
+      "loss": 0.0543,
+      "step": 1123
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 1.231372549019608e-05,
+      "loss": 0.0322,
+      "step": 1124
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 1.223529411764706e-05,
+      "loss": 0.0458,
+      "step": 1125
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 1.215686274509804e-05,
+      "loss": 0.0444,
+      "step": 1126
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 1.207843137254902e-05,
+      "loss": 0.0372,
+      "step": 1127
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 1.2e-05,
+      "loss": 0.0549,
+      "step": 1128
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 1.192156862745098e-05,
+      "loss": 0.0468,
+      "step": 1129
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 1.1843137254901961e-05,
+      "loss": 0.0378,
+      "step": 1130
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 1.1764705882352942e-05,
+      "loss": 0.0419,
+      "step": 1131
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 1.1686274509803922e-05,
+      "loss": 0.0463,
+      "step": 1132
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 1.1607843137254903e-05,
+      "loss": 0.0387,
+      "step": 1133
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 1.1529411764705883e-05,
+      "loss": 0.0428,
+      "step": 1134
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 1.1450980392156864e-05,
+      "loss": 0.0388,
+      "step": 1135
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 1.1372549019607843e-05,
+      "loss": 0.0417,
+      "step": 1136
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 1.1294117647058825e-05,
+      "loss": 0.0395,
+      "step": 1137
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 1.1215686274509804e-05,
+      "loss": 0.0551,
+      "step": 1138
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 1.1137254901960784e-05,
+      "loss": 0.0686,
+      "step": 1139
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 1.1058823529411766e-05,
+      "loss": 0.0408,
+      "step": 1140
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 1.0980392156862745e-05,
+      "loss": 0.0841,
+      "step": 1141
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 1.0901960784313726e-05,
+      "loss": 0.0426,
+      "step": 1142
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 1.0823529411764706e-05,
+      "loss": 0.0341,
+      "step": 1143
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 1.0745098039215687e-05,
+      "loss": 0.0492,
+      "step": 1144
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 0.0755,
+      "step": 1145
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 1.0588235294117648e-05,
+      "loss": 0.048,
+      "step": 1146
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 1.0509803921568628e-05,
+      "loss": 0.057,
+      "step": 1147
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 1.0431372549019607e-05,
+      "loss": 0.0482,
+      "step": 1148
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 1.035294117647059e-05,
+      "loss": 0.0317,
+      "step": 1149
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 1.027450980392157e-05,
+      "loss": 0.0459,
+      "step": 1150
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 1.0196078431372549e-05,
+      "loss": 0.0809,
+      "step": 1151
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 1.0117647058823531e-05,
+      "loss": 0.0464,
+      "step": 1152
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 1.003921568627451e-05,
+      "loss": 0.0399,
+      "step": 1153
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 9.96078431372549e-06,
+      "loss": 0.0545,
+      "step": 1154
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 9.88235294117647e-06,
+      "loss": 0.0439,
+      "step": 1155
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 9.803921568627451e-06,
+      "loss": 0.0463,
+      "step": 1156
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 9.725490196078432e-06,
+      "loss": 0.0455,
+      "step": 1157
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 9.647058823529412e-06,
+      "loss": 0.0467,
+      "step": 1158
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 9.568627450980393e-06,
+      "loss": 0.0458,
+      "step": 1159
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 9.490196078431372e-06,
+      "loss": 0.0428,
+      "step": 1160
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 9.411764705882354e-06,
+      "loss": 0.0412,
+      "step": 1161
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 0.0356,
+      "step": 1162
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 9.254901960784313e-06,
+      "loss": 0.0567,
+      "step": 1163
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 9.176470588235295e-06,
+      "loss": 0.0482,
+      "step": 1164
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 9.098039215686274e-06,
+      "loss": 0.0431,
+      "step": 1165
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 9.019607843137255e-06,
+      "loss": 0.0457,
+      "step": 1166
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 8.941176470588237e-06,
+      "loss": 0.0395,
+      "step": 1167
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 8.862745098039216e-06,
+      "loss": 0.0425,
+      "step": 1168
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 8.784313725490196e-06,
+      "loss": 0.064,
+      "step": 1169
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 8.705882352941177e-06,
+      "loss": 0.0519,
+      "step": 1170
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 8.627450980392157e-06,
+      "loss": 0.0419,
+      "step": 1171
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 8.549019607843138e-06,
+      "loss": 0.0565,
+      "step": 1172
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 8.470588235294118e-06,
+      "loss": 0.0686,
+      "step": 1173
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 8.392156862745099e-06,
+      "loss": 0.0441,
+      "step": 1174
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 8.313725490196078e-06,
+      "loss": 0.0449,
+      "step": 1175
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 8.23529411764706e-06,
+      "loss": 0.0431,
+      "step": 1176
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 8.15686274509804e-06,
+      "loss": 0.0499,
+      "step": 1177
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 8.078431372549019e-06,
+      "loss": 0.0435,
+      "step": 1178
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.0517,
+      "step": 1179
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 7.92156862745098e-06,
+      "loss": 0.0345,
+      "step": 1180
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 7.84313725490196e-06,
+      "loss": 0.0488,
+      "step": 1181
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 7.764705882352941e-06,
+      "loss": 0.0618,
+      "step": 1182
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 7.686274509803922e-06,
+      "loss": 0.0631,
+      "step": 1183
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 7.607843137254902e-06,
+      "loss": 0.0756,
+      "step": 1184
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 7.529411764705882e-06,
+      "loss": 0.0431,
+      "step": 1185
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 7.450980392156863e-06,
+      "loss": 0.0414,
+      "step": 1186
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 7.372549019607844e-06,
+      "loss": 0.0384,
+      "step": 1187
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 7.294117647058823e-06,
+      "loss": 0.0364,
+      "step": 1188
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 7.215686274509805e-06,
+      "loss": 0.0352,
+      "step": 1189
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 7.137254901960784e-06,
+      "loss": 0.0721,
+      "step": 1190
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 7.058823529411765e-06,
+      "loss": 0.0381,
+      "step": 1191
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 6.980392156862746e-06,
+      "loss": 0.0358,
+      "step": 1192
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 6.901960784313725e-06,
+      "loss": 0.0419,
+      "step": 1193
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 6.8235294117647065e-06,
+      "loss": 0.034,
+      "step": 1194
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 6.745098039215686e-06,
+      "loss": 0.0499,
+      "step": 1195
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.0451,
+      "step": 1196
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 6.588235294117648e-06,
+      "loss": 0.0395,
+      "step": 1197
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 6.509803921568628e-06,
+      "loss": 0.0634,
+      "step": 1198
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 6.431372549019608e-06,
+      "loss": 0.0458,
+      "step": 1199
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 6.352941176470588e-06,
+      "loss": 0.0513,
+      "step": 1200
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 6.274509803921569e-06,
+      "loss": 0.0671,
+      "step": 1201
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 6.196078431372549e-06,
+      "loss": 0.048,
+      "step": 1202
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 6.11764705882353e-06,
+      "loss": 0.0475,
+      "step": 1203
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 6.03921568627451e-06,
+      "loss": 0.0363,
+      "step": 1204
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 5.96078431372549e-06,
+      "loss": 0.068,
+      "step": 1205
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 5.882352941176471e-06,
+      "loss": 0.0407,
+      "step": 1206
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 5.8039215686274514e-06,
+      "loss": 0.0364,
+      "step": 1207
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 5.725490196078432e-06,
+      "loss": 0.0718,
+      "step": 1208
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 5.6470588235294125e-06,
+      "loss": 0.0361,
+      "step": 1209
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 5.568627450980392e-06,
+      "loss": 0.0733,
+      "step": 1210
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 5.490196078431373e-06,
+      "loss": 0.0381,
+      "step": 1211
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 5.411764705882353e-06,
+      "loss": 0.0405,
+      "step": 1212
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 0.0588,
+      "step": 1213
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 5.254901960784314e-06,
+      "loss": 0.052,
+      "step": 1214
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 5.176470588235295e-06,
+      "loss": 0.0396,
+      "step": 1215
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 5.098039215686274e-06,
+      "loss": 0.0469,
+      "step": 1216
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 5.019607843137255e-06,
+      "loss": 0.0627,
+      "step": 1217
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 4.941176470588235e-06,
+      "loss": 0.0396,
+      "step": 1218
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 4.862745098039216e-06,
+      "loss": 0.0589,
+      "step": 1219
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 4.784313725490196e-06,
+      "loss": 0.0361,
+      "step": 1220
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 4.705882352941177e-06,
+      "loss": 0.0405,
+      "step": 1221
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 4.6274509803921566e-06,
+      "loss": 0.058,
+      "step": 1222
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 4.549019607843137e-06,
+      "loss": 0.0456,
+      "step": 1223
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 4.4705882352941184e-06,
+      "loss": 0.0437,
+      "step": 1224
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 4.392156862745098e-06,
+      "loss": 0.0449,
+      "step": 1225
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 4.313725490196079e-06,
+      "loss": 0.0305,
+      "step": 1226
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 4.235294117647059e-06,
+      "loss": 0.035,
+      "step": 1227
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 4.156862745098039e-06,
+      "loss": 0.0396,
+      "step": 1228
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 4.07843137254902e-06,
+      "loss": 0.0421,
+      "step": 1229
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.0464,
+      "step": 1230
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 3.92156862745098e-06,
+      "loss": 0.0452,
+      "step": 1231
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 3.843137254901961e-06,
+      "loss": 0.06,
+      "step": 1232
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 3.764705882352941e-06,
+      "loss": 0.0716,
+      "step": 1233
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 3.686274509803922e-06,
+      "loss": 0.0511,
+      "step": 1234
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 3.6078431372549024e-06,
+      "loss": 0.0367,
+      "step": 1235
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 3.5294117647058825e-06,
+      "loss": 0.0326,
+      "step": 1236
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 3.4509803921568626e-06,
+      "loss": 0.0358,
+      "step": 1237
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 3.372549019607843e-06,
+      "loss": 0.0468,
+      "step": 1238
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 3.294117647058824e-06,
+      "loss": 0.0463,
+      "step": 1239
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 3.215686274509804e-06,
+      "loss": 0.0434,
+      "step": 1240
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 3.1372549019607846e-06,
+      "loss": 0.0391,
+      "step": 1241
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 3.058823529411765e-06,
+      "loss": 0.0369,
+      "step": 1242
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 2.980392156862745e-06,
+      "loss": 0.051,
+      "step": 1243
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 2.9019607843137257e-06,
+      "loss": 0.044,
+      "step": 1244
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 2.8235294117647062e-06,
+      "loss": 0.0424,
+      "step": 1245
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 2.7450980392156863e-06,
+      "loss": 0.0601,
+      "step": 1246
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 0.0461,
+      "step": 1247
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 2.5882352941176473e-06,
+      "loss": 0.0489,
+      "step": 1248
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 2.5098039215686274e-06,
+      "loss": 0.0303,
+      "step": 1249
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 2.431372549019608e-06,
+      "loss": 0.0445,
+      "step": 1250
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 2.3529411764705885e-06,
+      "loss": 0.0596,
+      "step": 1251
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 2.2745098039215685e-06,
+      "loss": 0.0372,
+      "step": 1252
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 2.196078431372549e-06,
+      "loss": 0.0879,
+      "step": 1253
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 2.1176470588235296e-06,
+      "loss": 0.0626,
+      "step": 1254
+    },
+    {
+      "epoch": 19.61,
+      "learning_rate": 2.03921568627451e-06,
+      "loss": 0.0476,
+      "step": 1255
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 1.96078431372549e-06,
+      "loss": 0.037,
+      "step": 1256
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 1.8823529411764705e-06,
+      "loss": 0.0497,
+      "step": 1257
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 1.8039215686274512e-06,
+      "loss": 0.0581,
+      "step": 1258
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 1.7254901960784313e-06,
+      "loss": 0.0376,
+      "step": 1259
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 1.647058823529412e-06,
+      "loss": 0.0398,
+      "step": 1260
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 1.5686274509803923e-06,
+      "loss": 0.0426,
+      "step": 1261
+    },
+    {
+      "epoch": 19.72,
+      "learning_rate": 1.4901960784313726e-06,
+      "loss": 0.0348,
+      "step": 1262
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 1.4117647058823531e-06,
+      "loss": 0.0367,
+      "step": 1263
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 0.0404,
+      "step": 1264
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 1.2549019607843137e-06,
+      "loss": 0.0454,
+      "step": 1265
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 1.1764705882352942e-06,
+      "loss": 0.0357,
+      "step": 1266
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 1.0980392156862745e-06,
+      "loss": 0.0407,
+      "step": 1267
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 1.019607843137255e-06,
+      "loss": 0.0521,
+      "step": 1268
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 9.411764705882352e-07,
+      "loss": 0.0933,
+      "step": 1269
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 8.627450980392156e-07,
+      "loss": 0.0736,
+      "step": 1270
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 7.843137254901962e-07,
+      "loss": 0.0547,
+      "step": 1271
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 7.058823529411766e-07,
+      "loss": 0.0504,
+      "step": 1272
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 6.274509803921569e-07,
+      "loss": 0.0368,
+      "step": 1273
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 5.490196078431373e-07,
+      "loss": 0.0686,
+      "step": 1274
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 4.705882352941176e-07,
+      "loss": 0.0453,
+      "step": 1275
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 3.921568627450981e-07,
+      "loss": 0.0622,
+      "step": 1276
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 3.1372549019607843e-07,
+      "loss": 0.0415,
+      "step": 1277
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 2.352941176470588e-07,
+      "loss": 0.0483,
+      "step": 1278
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 1.5686274509803921e-07,
+      "loss": 0.046,
+      "step": 1279
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 7.843137254901961e-08,
+      "loss": 0.0732,
+      "step": 1280
+    },
+    {
+      "epoch": 20.0,
+      "step": 1280,
+      "total_flos": 6.53458201903104e+16,
+      "train_loss": 0.21532799505366712,
+      "train_runtime": 2152.3369,
+      "train_samples_per_second": 2.379,
+      "train_steps_per_second": 0.595
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1280,
+  "num_train_epochs": 20,
+  "save_steps": 500,
+  "total_flos": 6.53458201903104e+16,
+  "trial_name": null,
+  "trial_params": null
+}