diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9817 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 6.0,
+  "global_step": 1632,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996936274509804e-05,
+      "loss": 3.5118,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.993872549019608e-05,
+      "loss": 4.5785,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.990808823529412e-05,
+      "loss": 3.2408,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.987745098039216e-05,
+      "loss": 3.0245,
+      "step": 4
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.98468137254902e-05,
+      "loss": 2.9701,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9816176470588236e-05,
+      "loss": 2.4974,
+      "step": 6
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.978553921568628e-05,
+      "loss": 3.0388,
+      "step": 7
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.975490196078432e-05,
+      "loss": 2.7724,
+      "step": 8
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9724264705882356e-05,
+      "loss": 2.7099,
+      "step": 9
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9693627450980394e-05,
+      "loss": 2.8153,
+      "step": 10
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.966299019607843e-05,
+      "loss": 3.0066,
+      "step": 11
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9632352941176476e-05,
+      "loss": 2.7652,
+      "step": 12
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9601715686274514e-05,
+      "loss": 2.7697,
+      "step": 13
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.957107843137255e-05,
+      "loss": 2.5942,
+      "step": 14
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.954044117647059e-05,
+      "loss": 2.7342,
+      "step": 15
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9509803921568634e-05,
+      "loss": 2.8845,
+      "step": 16
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.947916666666667e-05,
+      "loss": 2.7979,
+      "step": 17
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.944852941176471e-05,
+      "loss": 2.836,
+      "step": 18
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.941789215686275e-05,
+      "loss": 2.7855,
+      "step": 19
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9387254901960786e-05,
+      "loss": 3.003,
+      "step": 20
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.935661764705883e-05,
+      "loss": 2.6277,
+      "step": 21
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.932598039215687e-05,
+      "loss": 2.6586,
+      "step": 22
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9295343137254906e-05,
+      "loss": 2.8393,
+      "step": 23
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9264705882352944e-05,
+      "loss": 2.9762,
+      "step": 24
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.923406862745099e-05,
+      "loss": 2.5588,
+      "step": 25
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.920343137254902e-05,
+      "loss": 2.8396,
+      "step": 26
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.917279411764706e-05,
+      "loss": 2.464,
+      "step": 27
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9142156862745095e-05,
+      "loss": 2.7508,
+      "step": 28
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.911151960784314e-05,
+      "loss": 2.814,
+      "step": 29
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.908088235294118e-05,
+      "loss": 2.672,
+      "step": 30
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9050245098039215e-05,
+      "loss": 2.601,
+      "step": 31
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.901960784313725e-05,
+      "loss": 2.9379,
+      "step": 32
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.89889705882353e-05,
+      "loss": 2.7852,
+      "step": 33
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8958333333333335e-05,
+      "loss": 2.8595,
+      "step": 34
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.892769607843137e-05,
+      "loss": 2.7366,
+      "step": 35
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.889705882352941e-05,
+      "loss": 2.7073,
+      "step": 36
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.886642156862745e-05,
+      "loss": 2.65,
+      "step": 37
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.883578431372549e-05,
+      "loss": 2.3047,
+      "step": 38
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.880514705882353e-05,
+      "loss": 2.539,
+      "step": 39
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.877450980392157e-05,
+      "loss": 2.5978,
+      "step": 40
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.874387254901961e-05,
+      "loss": 2.7383,
+      "step": 41
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.871323529411765e-05,
+      "loss": 2.6903,
+      "step": 42
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.868259803921569e-05,
+      "loss": 2.7869,
+      "step": 43
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.865196078431373e-05,
+      "loss": 2.7194,
+      "step": 44
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8621323529411765e-05,
+      "loss": 3.2235,
+      "step": 45
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.85906862745098e-05,
+      "loss": 2.9692,
+      "step": 46
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.856004901960785e-05,
+      "loss": 2.9272,
+      "step": 47
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8529411764705885e-05,
+      "loss": 2.8609,
+      "step": 48
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.849877450980392e-05,
+      "loss": 2.6578,
+      "step": 49
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.846813725490196e-05,
+      "loss": 2.6517,
+      "step": 50
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8437500000000005e-05,
+      "loss": 2.6786,
+      "step": 51
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.840686274509804e-05,
+      "loss": 2.5651,
+      "step": 52
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.837622549019608e-05,
+      "loss": 2.6324,
+      "step": 53
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.834558823529412e-05,
+      "loss": 2.6695,
+      "step": 54
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8314950980392156e-05,
+      "loss": 2.4562,
+      "step": 55
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.82843137254902e-05,
+      "loss": 2.6444,
+      "step": 56
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.825367647058824e-05,
+      "loss": 2.6841,
+      "step": 57
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8223039215686276e-05,
+      "loss": 2.5234,
+      "step": 58
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.8192401960784314e-05,
+      "loss": 2.3478,
+      "step": 59
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.816176470588236e-05,
+      "loss": 2.5619,
+      "step": 60
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.81311274509804e-05,
+      "loss": 2.6184,
+      "step": 61
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8100490196078434e-05,
+      "loss": 2.9987,
+      "step": 62
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.806985294117647e-05,
+      "loss": 2.5148,
+      "step": 63
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.803921568627452e-05,
+      "loss": 2.651,
+      "step": 64
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8008578431372555e-05,
+      "loss": 2.4045,
+      "step": 65
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.797794117647059e-05,
+      "loss": 2.2347,
+      "step": 66
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.794730392156863e-05,
+      "loss": 2.8173,
+      "step": 67
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.791666666666667e-05,
+      "loss": 2.4226,
+      "step": 68
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.788602941176471e-05,
+      "loss": 2.657,
+      "step": 69
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.785539215686275e-05,
+      "loss": 2.4022,
+      "step": 70
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.782475490196079e-05,
+      "loss": 2.7068,
+      "step": 71
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7794117647058826e-05,
+      "loss": 2.665,
+      "step": 72
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.776348039215687e-05,
+      "loss": 2.7504,
+      "step": 73
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.773284313725491e-05,
+      "loss": 2.5097,
+      "step": 74
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7702205882352946e-05,
+      "loss": 2.7736,
+      "step": 75
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7671568627450984e-05,
+      "loss": 2.7904,
+      "step": 76
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.764093137254902e-05,
+      "loss": 2.5558,
+      "step": 77
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.761029411764706e-05,
+      "loss": 2.6493,
+      "step": 78
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.75796568627451e-05,
+      "loss": 2.7398,
+      "step": 79
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7549019607843135e-05,
+      "loss": 2.4867,
+      "step": 80
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.751838235294117e-05,
+      "loss": 2.3607,
+      "step": 81
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.748774509803922e-05,
+      "loss": 2.6477,
+      "step": 82
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7457107843137256e-05,
+      "loss": 2.3804,
+      "step": 83
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.742647058823529e-05,
+      "loss": 2.5733,
+      "step": 84
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.739583333333333e-05,
+      "loss": 2.5984,
+      "step": 85
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.7365196078431376e-05,
+      "loss": 2.6211,
+      "step": 86
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.7334558823529414e-05,
+      "loss": 2.6124,
+      "step": 87
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.730392156862745e-05,
+      "loss": 2.4538,
+      "step": 88
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.727328431372549e-05,
+      "loss": 2.51,
+      "step": 89
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.7242647058823534e-05,
+      "loss": 2.6145,
+      "step": 90
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.721200980392157e-05,
+      "loss": 2.4823,
+      "step": 91
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.718137254901961e-05,
+      "loss": 2.5439,
+      "step": 92
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.715073529411765e-05,
+      "loss": 2.5715,
+      "step": 93
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7120098039215685e-05,
+      "loss": 2.6819,
+      "step": 94
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.708946078431373e-05,
+      "loss": 2.4487,
+      "step": 95
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.705882352941177e-05,
+      "loss": 2.4479,
+      "step": 96
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.7028186274509805e-05,
+      "loss": 2.5691,
+      "step": 97
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.699754901960784e-05,
+      "loss": 2.5309,
+      "step": 98
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.696691176470589e-05,
+      "loss": 2.3055,
+      "step": 99
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6936274509803925e-05,
+      "loss": 2.6269,
+      "step": 100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.690563725490196e-05,
+      "loss": 2.5308,
+      "step": 101
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6875e-05,
+      "loss": 2.2248,
+      "step": 102
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.684436274509804e-05,
+      "loss": 2.4921,
+      "step": 103
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.681372549019608e-05,
+      "loss": 2.5861,
+      "step": 104
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.678308823529412e-05,
+      "loss": 2.5067,
+      "step": 105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.675245098039216e-05,
+      "loss": 2.2451,
+      "step": 106
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.67218137254902e-05,
+      "loss": 2.4245,
+      "step": 107
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.669117647058824e-05,
+      "loss": 2.4757,
+      "step": 108
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.666053921568628e-05,
+      "loss": 2.1759,
+      "step": 109
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.662990196078432e-05,
+      "loss": 2.4127,
+      "step": 110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6599264705882355e-05,
+      "loss": 2.1939,
+      "step": 111
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.656862745098039e-05,
+      "loss": 2.457,
+      "step": 112
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.653799019607844e-05,
+      "loss": 2.3628,
+      "step": 113
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6507352941176475e-05,
+      "loss": 2.5025,
+      "step": 114
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.647671568627451e-05,
+      "loss": 2.6513,
+      "step": 115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.644607843137255e-05,
+      "loss": 2.1793,
+      "step": 116
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.6415441176470595e-05,
+      "loss": 2.5897,
+      "step": 117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.638480392156863e-05,
+      "loss": 2.426,
+      "step": 118
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.635416666666667e-05,
+      "loss": 2.403,
+      "step": 119
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.632352941176471e-05,
+      "loss": 2.4851,
+      "step": 120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.6292892156862746e-05,
+      "loss": 2.4252,
+      "step": 121
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.626225490196079e-05,
+      "loss": 2.2623,
+      "step": 122
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.623161764705883e-05,
+      "loss": 2.4193,
+      "step": 123
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.6200980392156867e-05,
+      "loss": 2.2587,
+      "step": 124
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.6170343137254904e-05,
+      "loss": 2.3598,
+      "step": 125
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.613970588235295e-05,
+      "loss": 2.2216,
+      "step": 126
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.610906862745099e-05,
+      "loss": 2.3176,
+      "step": 127
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.607843137254902e-05,
+      "loss": 2.4784,
+      "step": 128
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.6047794117647056e-05,
+      "loss": 2.249,
+      "step": 129
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.60171568627451e-05,
+      "loss": 2.086,
+      "step": 130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.598651960784314e-05,
+      "loss": 2.3452,
+      "step": 131
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5955882352941176e-05,
+      "loss": 2.3103,
+      "step": 132
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5925245098039214e-05,
+      "loss": 2.3928,
+      "step": 133
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.589460784313726e-05,
+      "loss": 2.2282,
+      "step": 134
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5863970588235296e-05,
+      "loss": 2.2371,
+      "step": 135
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 2.5309,
+      "step": 136
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.580269607843137e-05,
+      "loss": 2.5047,
+      "step": 137
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.577205882352941e-05,
+      "loss": 2.2585,
+      "step": 138
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5741421568627454e-05,
+      "loss": 2.5134,
+      "step": 139
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.571078431372549e-05,
+      "loss": 2.5324,
+      "step": 140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.568014705882353e-05,
+      "loss": 1.9985,
+      "step": 141
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.564950980392157e-05,
+      "loss": 2.5235,
+      "step": 142
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.561887254901961e-05,
+      "loss": 2.4866,
+      "step": 143
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.558823529411765e-05,
+      "loss": 2.3664,
+      "step": 144
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.555759803921569e-05,
+      "loss": 2.1721,
+      "step": 145
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5526960784313725e-05,
+      "loss": 2.2977,
+      "step": 146
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.549632352941176e-05,
+      "loss": 2.3362,
+      "step": 147
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.546568627450981e-05,
+      "loss": 2.1449,
+      "step": 148
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5435049019607846e-05,
+      "loss": 2.3353,
+      "step": 149
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5404411764705883e-05,
+      "loss": 2.3871,
+      "step": 150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.537377450980392e-05,
+      "loss": 2.2133,
+      "step": 151
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5343137254901966e-05,
+      "loss": 2.5061,
+      "step": 152
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5312500000000004e-05,
+      "loss": 2.2117,
+      "step": 153
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.528186274509804e-05,
+      "loss": 2.1194,
+      "step": 154
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.525122549019608e-05,
+      "loss": 2.6798,
+      "step": 155
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.522058823529412e-05,
+      "loss": 2.281,
+      "step": 156
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.518995098039216e-05,
+      "loss": 2.4016,
+      "step": 157
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.51593137254902e-05,
+      "loss": 2.3965,
+      "step": 158
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.512867647058824e-05,
+      "loss": 2.4317,
+      "step": 159
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.5098039215686275e-05,
+      "loss": 2.2113,
+      "step": 160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.506740196078432e-05,
+      "loss": 2.2191,
+      "step": 161
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.503676470588236e-05,
+      "loss": 2.3163,
+      "step": 162
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.5006127450980395e-05,
+      "loss": 2.4287,
+      "step": 163
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.497549019607843e-05,
+      "loss": 2.158,
+      "step": 164
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.494485294117647e-05,
+      "loss": 2.3427,
+      "step": 165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.4914215686274515e-05,
+      "loss": 2.4291,
+      "step": 166
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.488357843137255e-05,
+      "loss": 2.4465,
+      "step": 167
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.485294117647059e-05,
+      "loss": 2.2704,
+      "step": 168
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.482230392156863e-05,
+      "loss": 2.4008,
+      "step": 169
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.4791666666666673e-05,
+      "loss": 2.4054,
+      "step": 170
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.476102941176471e-05,
+      "loss": 2.3167,
+      "step": 171
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.473039215686275e-05,
+      "loss": 1.9715,
+      "step": 172
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.469975490196079e-05,
+      "loss": 2.5775,
+      "step": 173
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.4669117647058825e-05,
+      "loss": 2.4314,
+      "step": 174
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.463848039215687e-05,
+      "loss": 2.4272,
+      "step": 175
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.460784313725491e-05,
+      "loss": 2.0353,
+      "step": 176
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.4577205882352945e-05,
+      "loss": 2.2946,
+      "step": 177
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.454656862745098e-05,
+      "loss": 2.2535,
+      "step": 178
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.451593137254902e-05,
+      "loss": 2.342,
+      "step": 179
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.448529411764706e-05,
+      "loss": 2.1857,
+      "step": 180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.4454656862745096e-05,
+      "loss": 2.1759,
+      "step": 181
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.4424019607843134e-05,
+      "loss": 2.1335,
+      "step": 182
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.439338235294118e-05,
+      "loss": 2.2563,
+      "step": 183
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.4362745098039216e-05,
+      "loss": 2.4938,
+      "step": 184
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.4332107843137254e-05,
+      "loss": 2.2641,
+      "step": 185
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.430147058823529e-05,
+      "loss": 2.4519,
+      "step": 186
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4270833333333337e-05,
+      "loss": 2.4227,
+      "step": 187
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4240196078431374e-05,
+      "loss": 2.2201,
+      "step": 188
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.420955882352941e-05,
+      "loss": 2.2575,
+      "step": 189
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.417892156862745e-05,
+      "loss": 2.2289,
+      "step": 190
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.414828431372549e-05,
+      "loss": 2.1321,
+      "step": 191
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.411764705882353e-05,
+      "loss": 2.1958,
+      "step": 192
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.408700980392157e-05,
+      "loss": 2.3323,
+      "step": 193
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.405637254901961e-05,
+      "loss": 2.0402,
+      "step": 194
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.4025735294117646e-05,
+      "loss": 1.9049,
+      "step": 195
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.399509803921569e-05,
+      "loss": 2.3106,
+      "step": 196
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.396446078431373e-05,
+      "loss": 2.3732,
+      "step": 197
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.3933823529411766e-05,
+      "loss": 1.8018,
+      "step": 198
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.3903186274509804e-05,
+      "loss": 2.0577,
+      "step": 199
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.387254901960784e-05,
+      "loss": 2.2024,
+      "step": 200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.3841911764705886e-05,
+      "loss": 2.3273,
+      "step": 201
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.3811274509803924e-05,
+      "loss": 2.1693,
+      "step": 202
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.378063725490196e-05,
+      "loss": 2.4046,
+      "step": 203
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.375e-05,
+      "loss": 2.1225,
+      "step": 204
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.3719362745098044e-05,
+      "loss": 2.1044,
+      "step": 205
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.368872549019608e-05,
+      "loss": 2.2267,
+      "step": 206
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.365808823529412e-05,
+      "loss": 2.3884,
+      "step": 207
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.362745098039216e-05,
+      "loss": 2.3939,
+      "step": 208
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.3596813725490195e-05,
+      "loss": 2.0336,
+      "step": 209
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.356617647058824e-05,
+      "loss": 2.1361,
+      "step": 210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.353553921568628e-05,
+      "loss": 2.3723,
+      "step": 211
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3504901960784316e-05,
+      "loss": 2.2218,
+      "step": 212
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.347426470588235e-05,
+      "loss": 2.0623,
+      "step": 213
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.34436274509804e-05,
+      "loss": 2.1853,
+      "step": 214
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3412990196078436e-05,
+      "loss": 2.1589,
+      "step": 215
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3382352941176474e-05,
+      "loss": 2.1722,
+      "step": 216
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.335171568627451e-05,
+      "loss": 2.0452,
+      "step": 217
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.3321078431372556e-05,
+      "loss": 2.345,
+      "step": 218
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3290441176470594e-05,
+      "loss": 2.4241,
+      "step": 219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.325980392156863e-05,
+      "loss": 2.3346,
+      "step": 220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.322916666666667e-05,
+      "loss": 2.2368,
+      "step": 221
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.319852941176471e-05,
+      "loss": 2.0703,
+      "step": 222
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.316789215686275e-05,
+      "loss": 2.405,
+      "step": 223
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.313725490196079e-05,
+      "loss": 2.1701,
+      "step": 224
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.310661764705883e-05,
+      "loss": 2.132,
+      "step": 225
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.3075980392156865e-05,
+      "loss": 2.2665,
+      "step": 226
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.304534313725491e-05,
+      "loss": 2.472,
+      "step": 227
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.301470588235295e-05,
+      "loss": 2.2249,
+      "step": 228
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.2984068627450985e-05,
+      "loss": 2.1702,
+      "step": 229
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.2953431372549016e-05,
+      "loss": 2.0627,
+      "step": 230
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.292279411764706e-05,
+      "loss": 2.355,
+      "step": 231
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.28921568627451e-05,
+      "loss": 2.0884,
+      "step": 232
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.2861519607843137e-05,
+      "loss": 2.146,
+      "step": 233
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.2830882352941174e-05,
+      "loss": 2.2043,
+      "step": 234
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.280024509803921e-05,
+      "loss": 2.1336,
+      "step": 235
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.276960784313726e-05,
+      "loss": 2.1111,
+      "step": 236
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2738970588235295e-05,
+      "loss": 2.0117,
+      "step": 237
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.270833333333333e-05,
+      "loss": 2.374,
+      "step": 238
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.267769607843137e-05,
+      "loss": 2.3928,
+      "step": 239
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2647058823529415e-05,
+      "loss": 2.0387,
+      "step": 240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.261642156862745e-05,
+      "loss": 2.1409,
+      "step": 241
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.258578431372549e-05,
+      "loss": 1.919,
+      "step": 242
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.255514705882353e-05,
+      "loss": 2.1419,
+      "step": 243
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.252450980392157e-05,
+      "loss": 2.202,
+      "step": 244
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.249387254901961e-05,
+      "loss": 1.8653,
+      "step": 245
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.246323529411765e-05,
+      "loss": 2.2304,
+      "step": 246
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2432598039215686e-05,
+      "loss": 2.0331,
+      "step": 247
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2401960784313724e-05,
+      "loss": 2.2847,
+      "step": 248
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.237132352941177e-05,
+      "loss": 2.0888,
+      "step": 249
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2340686274509806e-05,
+      "loss": 2.3228,
+      "step": 250
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2310049019607844e-05,
+      "loss": 2.2154,
+      "step": 251
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.227941176470588e-05,
+      "loss": 2.1985,
+      "step": 252
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.224877450980393e-05,
+      "loss": 2.1147,
+      "step": 253
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.2218137254901964e-05,
+      "loss": 2.4517,
+      "step": 254
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.21875e-05,
+      "loss": 2.3431,
+      "step": 255
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.215686274509804e-05,
+      "loss": 2.2463,
+      "step": 256
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.212622549019608e-05,
+      "loss": 2.1827,
+      "step": 257
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.209558823529412e-05,
+      "loss": 2.306,
+      "step": 258
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.206495098039216e-05,
+      "loss": 2.1433,
+      "step": 259
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.20343137254902e-05,
+      "loss": 2.3276,
+      "step": 260
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2003676470588236e-05,
+      "loss": 2.1947,
+      "step": 261
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.197303921568628e-05,
+      "loss": 2.1406,
+      "step": 262
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.194240196078432e-05,
+      "loss": 2.0821,
+      "step": 263
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1911764705882356e-05,
+      "loss": 1.8286,
+      "step": 264
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1881127450980394e-05,
+      "loss": 2.199,
+      "step": 265
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.185049019607843e-05,
+      "loss": 2.1696,
+      "step": 266
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.1819852941176476e-05,
+      "loss": 2.244,
+      "step": 267
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1789215686274514e-05,
+      "loss": 2.0299,
+      "step": 268
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.175857843137255e-05,
+      "loss": 1.855,
+      "step": 269
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.172794117647059e-05,
+      "loss": 2.0488,
+      "step": 270
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1697303921568634e-05,
+      "loss": 2.4275,
+      "step": 271
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 2.3835,
+      "step": 272
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.163602941176471e-05,
+      "loss": 1.9092,
+      "step": 273
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.160539215686275e-05,
+      "loss": 2.025,
+      "step": 274
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1574754901960785e-05,
+      "loss": 1.931,
+      "step": 275
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.154411764705883e-05,
+      "loss": 1.949,
+      "step": 276
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.151348039215687e-05,
+      "loss": 2.0069,
+      "step": 277
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1482843137254906e-05,
+      "loss": 1.7921,
+      "step": 278
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.1452205882352943e-05,
+      "loss": 1.852,
+      "step": 279
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.142156862745099e-05,
+      "loss": 1.8901,
+      "step": 280
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.139093137254902e-05,
+      "loss": 2.0774,
+      "step": 281
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.136029411764706e-05,
+      "loss": 1.9705,
+      "step": 282
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.1329656862745095e-05,
+      "loss": 1.8348,
+      "step": 283
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.129901960784314e-05,
+      "loss": 2.1076,
+      "step": 284
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.126838235294118e-05,
+      "loss": 2.1449,
+      "step": 285
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.1237745098039215e-05,
+      "loss": 2.0556,
+      "step": 286
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.120710784313725e-05,
+      "loss": 1.6693,
+      "step": 287
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.11764705882353e-05,
+      "loss": 1.8824,
+      "step": 288
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.1145833333333335e-05,
+      "loss": 1.8618,
+      "step": 289
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.111519607843137e-05,
+      "loss": 2.0572,
+      "step": 290
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.108455882352941e-05,
+      "loss": 1.8349,
+      "step": 291
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.105392156862745e-05,
+      "loss": 2.043,
+      "step": 292
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.102328431372549e-05,
+      "loss": 1.6203,
+      "step": 293
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.099264705882353e-05,
+      "loss": 1.7903,
+      "step": 294
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.096200980392157e-05,
+      "loss": 2.0009,
+      "step": 295
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.0931372549019607e-05,
+      "loss": 1.9993,
+      "step": 296
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.090073529411765e-05,
+      "loss": 1.7185,
+      "step": 297
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.087009803921569e-05,
+      "loss": 2.1832,
+      "step": 298
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.083946078431373e-05,
+      "loss": 1.7913,
+      "step": 299
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.0808823529411765e-05,
+      "loss": 1.9589,
+      "step": 300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.07781862745098e-05,
+      "loss": 1.8285,
+      "step": 301
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.074754901960785e-05,
+      "loss": 1.9652,
+      "step": 302
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.0716911764705885e-05,
+      "loss": 1.9667,
+      "step": 303
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.068627450980392e-05,
+      "loss": 2.158,
+      "step": 304
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.065563725490196e-05,
+      "loss": 2.0314,
+      "step": 305
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.0625000000000005e-05,
+      "loss": 1.9269,
+      "step": 306
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.059436274509804e-05,
+      "loss": 2.1711,
+      "step": 307
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.056372549019608e-05,
+      "loss": 1.9729,
+      "step": 308
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.053308823529412e-05,
+      "loss": 1.9459,
+      "step": 309
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.0502450980392156e-05,
+      "loss": 1.972,
+      "step": 310
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.04718137254902e-05,
+      "loss": 1.8987,
+      "step": 311
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.044117647058824e-05,
+      "loss": 1.7057,
+      "step": 312
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.0410539215686276e-05,
+      "loss": 1.8183,
+      "step": 313
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.0379901960784314e-05,
+      "loss": 2.0447,
+      "step": 314
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.034926470588236e-05,
+      "loss": 2.0349,
+      "step": 315
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.0318627450980397e-05,
+      "loss": 1.9461,
+      "step": 316
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.0287990196078434e-05,
+      "loss": 2.0576,
+      "step": 317
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.025735294117647e-05,
+      "loss": 2.0213,
+      "step": 318
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.022671568627451e-05,
+      "loss": 2.0006,
+      "step": 319
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.0196078431372555e-05,
+      "loss": 1.8733,
+      "step": 320
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.016544117647059e-05,
+      "loss": 1.8257,
+      "step": 321
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.013480392156863e-05,
+      "loss": 2.0813,
+      "step": 322
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.010416666666667e-05,
+      "loss": 1.9079,
+      "step": 323
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.007352941176471e-05,
+      "loss": 2.0039,
+      "step": 324
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.004289215686275e-05,
+      "loss": 1.878,
+      "step": 325
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.001225490196079e-05,
+      "loss": 1.9918,
+      "step": 326
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.9981617647058826e-05,
+      "loss": 1.9761,
+      "step": 327
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.9950980392156864e-05,
+      "loss": 1.9731,
+      "step": 328
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.992034313725491e-05,
+      "loss": 1.8294,
+      "step": 329
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.9889705882352946e-05,
+      "loss": 1.8657,
+      "step": 330
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.9859068627450984e-05,
+      "loss": 2.0021,
+      "step": 331
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.982843137254902e-05,
+      "loss": 1.967,
+      "step": 332
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.979779411764706e-05,
+      "loss": 1.9019,
+      "step": 333
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.97671568627451e-05,
+      "loss": 2.0132,
+      "step": 334
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.9736519607843135e-05,
+      "loss": 1.9158,
+      "step": 335
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.970588235294117e-05,
+      "loss": 1.7672,
+      "step": 336
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.967524509803922e-05,
+      "loss": 2.1716,
+      "step": 337
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.9644607843137255e-05,
+      "loss": 2.0246,
+      "step": 338
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.961397058823529e-05,
+      "loss": 1.8116,
+      "step": 339
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.958333333333333e-05,
+      "loss": 1.7693,
+      "step": 340
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.9552696078431376e-05,
+      "loss": 1.8338,
+      "step": 341
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.952205882352941e-05,
+      "loss": 1.983,
+      "step": 342
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.949142156862745e-05,
+      "loss": 1.8287,
+      "step": 343
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.946078431372549e-05,
+      "loss": 1.7685,
+      "step": 344
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.943014705882353e-05,
+      "loss": 1.9178,
+      "step": 345
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.939950980392157e-05,
+      "loss": 2.1342,
+      "step": 346
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.936887254901961e-05,
+      "loss": 1.8681,
+      "step": 347
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.933823529411765e-05,
+      "loss": 2.1368,
+      "step": 348
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.9307598039215685e-05,
+      "loss": 1.882,
+      "step": 349
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.927696078431373e-05,
+      "loss": 1.8108,
+      "step": 350
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.924632352941177e-05,
+      "loss": 1.8854,
+      "step": 351
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.9215686274509805e-05,
+      "loss": 2.064,
+      "step": 352
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.918504901960784e-05,
+      "loss": 2.0354,
+      "step": 353
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.915441176470588e-05,
+      "loss": 1.7351,
+      "step": 354
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.9123774509803925e-05,
+      "loss": 1.8782,
+      "step": 355
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.909313725490196e-05,
+      "loss": 1.9751,
+      "step": 356
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.90625e-05,
+      "loss": 2.0059,
+      "step": 357
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.903186274509804e-05,
+      "loss": 1.9633,
+      "step": 358
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.900122549019608e-05,
+      "loss": 2.0914,
+      "step": 359
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.897058823529412e-05,
+      "loss": 2.0495,
+      "step": 360
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.893995098039216e-05,
+      "loss": 1.8114,
+      "step": 361
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.89093137254902e-05,
+      "loss": 1.959,
+      "step": 362
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8878676470588234e-05,
+      "loss": 2.0646,
+      "step": 363
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.884803921568628e-05,
+      "loss": 1.9153,
+      "step": 364
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.881740196078432e-05,
+      "loss": 2.0165,
+      "step": 365
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.8786764705882355e-05,
+      "loss": 2.1385,
+      "step": 366
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.875612745098039e-05,
+      "loss": 2.0246,
+      "step": 367
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.872549019607844e-05,
+      "loss": 1.8779,
+      "step": 368
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.8694852941176475e-05,
+      "loss": 1.9307,
+      "step": 369
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.866421568627451e-05,
+      "loss": 1.8555,
+      "step": 370
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.863357843137255e-05,
+      "loss": 1.8603,
+      "step": 371
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.8602941176470595e-05,
+      "loss": 2.0863,
+      "step": 372
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.857230392156863e-05,
+      "loss": 2.0271,
+      "step": 373
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.854166666666667e-05,
+      "loss": 2.0512,
+      "step": 374
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.851102941176471e-05,
+      "loss": 1.9899,
+      "step": 375
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.8480392156862746e-05,
+      "loss": 1.868,
+      "step": 376
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.844975490196079e-05,
+      "loss": 2.1204,
+      "step": 377
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.841911764705883e-05,
+      "loss": 2.0317,
+      "step": 378
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.8388480392156866e-05,
+      "loss": 1.8943,
+      "step": 379
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.8357843137254904e-05,
+      "loss": 1.6817,
+      "step": 380
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.832720588235295e-05,
+      "loss": 1.8007,
+      "step": 381
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.829656862745099e-05,
+      "loss": 1.9926,
+      "step": 382
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.826593137254902e-05,
+      "loss": 1.5958,
+      "step": 383
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.8235294117647055e-05,
+      "loss": 1.9617,
+      "step": 384
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.82046568627451e-05,
+      "loss": 2.0121,
+      "step": 385
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.817401960784314e-05,
+      "loss": 2.0449,
+      "step": 386
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.8143382352941176e-05,
+      "loss": 1.9553,
+      "step": 387
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.8112745098039213e-05,
+      "loss": 2.118,
+      "step": 388
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.808210784313725e-05,
+      "loss": 1.6562,
+      "step": 389
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.8051470588235296e-05,
+      "loss": 1.8063,
+      "step": 390
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8020833333333334e-05,
+      "loss": 1.8776,
+      "step": 391
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.799019607843137e-05,
+      "loss": 2.0092,
+      "step": 392
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.795955882352941e-05,
+      "loss": 2.0422,
+      "step": 393
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7928921568627454e-05,
+      "loss": 1.8063,
+      "step": 394
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.789828431372549e-05,
+      "loss": 1.9717,
+      "step": 395
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.786764705882353e-05,
+      "loss": 1.9537,
+      "step": 396
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.783700980392157e-05,
+      "loss": 1.9653,
+      "step": 397
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.780637254901961e-05,
+      "loss": 1.9078,
+      "step": 398
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.777573529411765e-05,
+      "loss": 1.733,
+      "step": 399
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.774509803921569e-05,
+      "loss": 1.9314,
+      "step": 400
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.7714460784313725e-05,
+      "loss": 1.8944,
+      "step": 401
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.768382352941176e-05,
+      "loss": 1.825,
+      "step": 402
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.765318627450981e-05,
+      "loss": 1.9577,
+      "step": 403
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.7622549019607845e-05,
+      "loss": 1.657,
+      "step": 404
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.759191176470588e-05,
+      "loss": 2.0773,
+      "step": 405
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.756127450980392e-05,
+      "loss": 1.9045,
+      "step": 406
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.7530637254901966e-05,
+      "loss": 1.8432,
+      "step": 407
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 1.8743,
+      "step": 408
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.746936274509804e-05,
+      "loss": 1.7366,
+      "step": 409
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.743872549019608e-05,
+      "loss": 2.0883,
+      "step": 410
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.740808823529412e-05,
+      "loss": 1.9495,
+      "step": 411
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.737745098039216e-05,
+      "loss": 1.8685,
+      "step": 412
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.73468137254902e-05,
+      "loss": 1.9078,
+      "step": 413
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.731617647058824e-05,
+      "loss": 1.7417,
+      "step": 414
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.7285539215686275e-05,
+      "loss": 2.1359,
+      "step": 415
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.725490196078432e-05,
+      "loss": 2.2246,
+      "step": 416
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.722426470588236e-05,
+      "loss": 1.9227,
+      "step": 417
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.7193627450980395e-05,
+      "loss": 1.929,
+      "step": 418
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.716299019607843e-05,
+      "loss": 1.96,
+      "step": 419
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.713235294117647e-05,
+      "loss": 2.073,
+      "step": 420
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.7101715686274515e-05,
+      "loss": 1.7272,
+      "step": 421
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.707107843137255e-05,
+      "loss": 1.9502,
+      "step": 422
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.704044117647059e-05,
+      "loss": 1.7618,
+      "step": 423
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.700980392156863e-05,
+      "loss": 1.9592,
+      "step": 424
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.697916666666667e-05,
+      "loss": 1.945,
+      "step": 425
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.694852941176471e-05,
+      "loss": 2.045,
+      "step": 426
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.691789215686275e-05,
+      "loss": 1.8333,
+      "step": 427
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.688725490196079e-05,
+      "loss": 1.9429,
+      "step": 428
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.6856617647058825e-05,
+      "loss": 1.9512,
+      "step": 429
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.682598039215687e-05,
+      "loss": 1.9848,
+      "step": 430
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.679534313725491e-05,
+      "loss": 1.6781,
+      "step": 431
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.6764705882352945e-05,
+      "loss": 1.8621,
+      "step": 432
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.673406862745098e-05,
+      "loss": 2.0256,
+      "step": 433
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.670343137254902e-05,
+      "loss": 1.8649,
+      "step": 434
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.667279411764706e-05,
+      "loss": 2.0788,
+      "step": 435
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.6642156862745096e-05,
+      "loss": 1.9107,
+      "step": 436
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.6611519607843134e-05,
+      "loss": 2.0379,
+      "step": 437
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.658088235294118e-05,
+      "loss": 1.9677,
+      "step": 438
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.6550245098039216e-05,
+      "loss": 1.8014,
+      "step": 439
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.6519607843137254e-05,
+      "loss": 1.8772,
+      "step": 440
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.648897058823529e-05,
+      "loss": 1.7015,
+      "step": 441
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.6458333333333336e-05,
+      "loss": 1.7116,
+      "step": 442
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.6427696078431374e-05,
+      "loss": 1.8549,
+      "step": 443
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.639705882352941e-05,
+      "loss": 1.8121,
+      "step": 444
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.636642156862745e-05,
+      "loss": 1.8071,
+      "step": 445
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.633578431372549e-05,
+      "loss": 1.9324,
+      "step": 446
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.630514705882353e-05,
+      "loss": 2.0323,
+      "step": 447
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.627450980392157e-05,
+      "loss": 1.7462,
+      "step": 448
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.624387254901961e-05,
+      "loss": 1.7846,
+      "step": 449
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.6213235294117646e-05,
+      "loss": 1.9438,
+      "step": 450
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.618259803921569e-05,
+      "loss": 1.8369,
+      "step": 451
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.615196078431373e-05,
+      "loss": 1.7862,
+      "step": 452
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.6121323529411766e-05,
+      "loss": 1.9134,
+      "step": 453
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.6090686274509804e-05,
+      "loss": 2.0405,
+      "step": 454
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.606004901960784e-05,
+      "loss": 2.0288,
+      "step": 455
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.6029411764705886e-05,
+      "loss": 1.8688,
+      "step": 456
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.5998774509803924e-05,
+      "loss": 2.1993,
+      "step": 457
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.596813725490196e-05,
+      "loss": 1.7871,
+      "step": 458
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.59375e-05,
+      "loss": 1.9402,
+      "step": 459
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.5906862745098044e-05,
+      "loss": 1.9078,
+      "step": 460
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.587622549019608e-05,
+      "loss": 1.8933,
+      "step": 461
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.584558823529412e-05,
+      "loss": 2.0978,
+      "step": 462
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.581495098039216e-05,
+      "loss": 1.7515,
+      "step": 463
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.5784313725490195e-05,
+      "loss": 2.1172,
+      "step": 464
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.575367647058824e-05,
+      "loss": 1.99,
+      "step": 465
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.572303921568628e-05,
+      "loss": 2.0716,
+      "step": 466
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.5692401960784315e-05,
+      "loss": 2.1353,
+      "step": 467
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.566176470588235e-05,
+      "loss": 1.8881,
+      "step": 468
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.56311274509804e-05,
+      "loss": 1.7231,
+      "step": 469
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.5600490196078436e-05,
+      "loss": 1.7643,
+      "step": 470
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.5569852941176473e-05,
+      "loss": 1.7493,
+      "step": 471
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.553921568627451e-05,
+      "loss": 2.0446,
+      "step": 472
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.550857843137255e-05,
+      "loss": 1.8594,
+      "step": 473
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.5477941176470594e-05,
+      "loss": 1.7812,
+      "step": 474
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.544730392156863e-05,
+      "loss": 1.6997,
+      "step": 475
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.541666666666667e-05,
+      "loss": 1.8309,
+      "step": 476
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.538602941176471e-05,
+      "loss": 1.928,
+      "step": 477
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.535539215686275e-05,
+      "loss": 1.9739,
+      "step": 478
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.532475490196079e-05,
+      "loss": 1.8249,
+      "step": 479
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.529411764705883e-05,
+      "loss": 1.8884,
+      "step": 480
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.5263480392156865e-05,
+      "loss": 1.7544,
+      "step": 481
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.52328431372549e-05,
+      "loss": 2.0147,
+      "step": 482
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.520220588235295e-05,
+      "loss": 1.9665,
+      "step": 483
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.5171568627450985e-05,
+      "loss": 2.0832,
+      "step": 484
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.5140931372549016e-05,
+      "loss": 1.6139,
+      "step": 485
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.511029411764706e-05,
+      "loss": 1.9746,
+      "step": 486
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.50796568627451e-05,
+      "loss": 2.0346,
+      "step": 487
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.5049019607843136e-05,
+      "loss": 2.0741,
+      "step": 488
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.5018382352941174e-05,
+      "loss": 1.7232,
+      "step": 489
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.498774509803921e-05,
+      "loss": 1.8538,
+      "step": 490
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.495710784313726e-05,
+      "loss": 1.8615,
+      "step": 491
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.4926470588235294e-05,
+      "loss": 1.959,
+      "step": 492
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.489583333333333e-05,
+      "loss": 1.802,
+      "step": 493
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.486519607843137e-05,
+      "loss": 1.8703,
+      "step": 494
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.4834558823529415e-05,
+      "loss": 1.9265,
+      "step": 495
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.480392156862745e-05,
+      "loss": 2.0489,
+      "step": 496
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.477328431372549e-05,
+      "loss": 1.7964,
+      "step": 497
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.474264705882353e-05,
+      "loss": 1.9704,
+      "step": 498
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.4712009803921566e-05,
+      "loss": 2.116,
+      "step": 499
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.468137254901961e-05,
+      "loss": 1.9838,
+      "step": 500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.465073529411765e-05,
+      "loss": 1.9908,
+      "step": 501
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.4620098039215686e-05,
+      "loss": 2.1425,
+      "step": 502
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.4589460784313724e-05,
+      "loss": 2.0332,
+      "step": 503
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.455882352941177e-05,
+      "loss": 1.6816,
+      "step": 504
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.4528186274509806e-05,
+      "loss": 1.96,
+      "step": 505
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.4497549019607844e-05,
+      "loss": 1.7254,
+      "step": 506
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.446691176470588e-05,
+      "loss": 2.0473,
+      "step": 507
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.443627450980392e-05,
+      "loss": 2.074,
+      "step": 508
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.4405637254901964e-05,
+      "loss": 1.8824,
+      "step": 509
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.4375e-05,
+      "loss": 2.0204,
+      "step": 510
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.434436274509804e-05,
+      "loss": 1.7945,
+      "step": 511
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.431372549019608e-05,
+      "loss": 2.1147,
+      "step": 512
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.428308823529412e-05,
+      "loss": 1.9908,
+      "step": 513
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.425245098039216e-05,
+      "loss": 1.7443,
+      "step": 514
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.42218137254902e-05,
+      "loss": 1.6125,
+      "step": 515
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.4191176470588236e-05,
+      "loss": 2.1982,
+      "step": 516
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.416053921568628e-05,
+      "loss": 1.8349,
+      "step": 517
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.412990196078432e-05,
+      "loss": 1.945,
+      "step": 518
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.4099264705882356e-05,
+      "loss": 2.0423,
+      "step": 519
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.4068627450980394e-05,
+      "loss": 1.8873,
+      "step": 520
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.403799019607843e-05,
+      "loss": 1.7465,
+      "step": 521
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.4007352941176476e-05,
+      "loss": 1.8102,
+      "step": 522
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.3976715686274514e-05,
+      "loss": 1.7168,
+      "step": 523
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.394607843137255e-05,
+      "loss": 1.8876,
+      "step": 524
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.391544117647059e-05,
+      "loss": 1.8993,
+      "step": 525
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.3884803921568634e-05,
+      "loss": 1.9045,
+      "step": 526
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.385416666666667e-05,
+      "loss": 2.1143,
+      "step": 527
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.382352941176471e-05,
+      "loss": 1.839,
+      "step": 528
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.379289215686275e-05,
+      "loss": 1.734,
+      "step": 529
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.3762254901960785e-05,
+      "loss": 1.8295,
+      "step": 530
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.373161764705883e-05,
+      "loss": 1.9893,
+      "step": 531
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.370098039215687e-05,
+      "loss": 1.914,
+      "step": 532
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.3670343137254906e-05,
+      "loss": 1.624,
+      "step": 533
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.363970588235294e-05,
+      "loss": 1.9573,
+      "step": 534
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.360906862745099e-05,
+      "loss": 1.8269,
+      "step": 535
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.357843137254902e-05,
+      "loss": 1.9862,
+      "step": 536
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.354779411764706e-05,
+      "loss": 2.0143,
+      "step": 537
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.3517156862745095e-05,
+      "loss": 1.8909,
+      "step": 538
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.348651960784314e-05,
+      "loss": 1.8889,
+      "step": 539
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.345588235294118e-05,
+      "loss": 1.7834,
+      "step": 540
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.3425245098039215e-05,
+      "loss": 1.8342,
+      "step": 541
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.339460784313725e-05,
+      "loss": 1.8376,
+      "step": 542
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.33639705882353e-05,
+      "loss": 1.9145,
+      "step": 543
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 2.3114,
+      "step": 544
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.330269607843137e-05,
+      "loss": 1.7952,
+      "step": 545
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.327205882352941e-05,
+      "loss": 1.7951,
+      "step": 546
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.324142156862745e-05,
+      "loss": 1.7298,
+      "step": 547
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.321078431372549e-05,
+      "loss": 1.6547,
+      "step": 548
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.318014705882353e-05,
+      "loss": 1.5407,
+      "step": 549
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.314950980392157e-05,
+      "loss": 1.5092,
+      "step": 550
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3118872549019606e-05,
+      "loss": 1.5908,
+      "step": 551
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.308823529411765e-05,
+      "loss": 1.7911,
+      "step": 552
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.305759803921569e-05,
+      "loss": 1.7413,
+      "step": 553
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3026960784313727e-05,
+      "loss": 1.7531,
+      "step": 554
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.2996323529411764e-05,
+      "loss": 1.8203,
+      "step": 555
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.29656862745098e-05,
+      "loss": 1.6736,
+      "step": 556
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.293504901960785e-05,
+      "loss": 1.5981,
+      "step": 557
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.2904411764705885e-05,
+      "loss": 1.8698,
+      "step": 558
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.287377450980392e-05,
+      "loss": 1.7287,
+      "step": 559
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.284313725490196e-05,
+      "loss": 1.565,
+      "step": 560
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.2812500000000005e-05,
+      "loss": 1.623,
+      "step": 561
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.278186274509804e-05,
+      "loss": 1.9199,
+      "step": 562
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.275122549019608e-05,
+      "loss": 1.851,
+      "step": 563
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.272058823529412e-05,
+      "loss": 1.9228,
+      "step": 564
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2689950980392156e-05,
+      "loss": 1.7315,
+      "step": 565
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.26593137254902e-05,
+      "loss": 2.0345,
+      "step": 566
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.262867647058824e-05,
+      "loss": 1.7714,
+      "step": 567
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2598039215686276e-05,
+      "loss": 1.8917,
+      "step": 568
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2567401960784314e-05,
+      "loss": 1.6894,
+      "step": 569
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.253676470588236e-05,
+      "loss": 1.544,
+      "step": 570
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2506127450980396e-05,
+      "loss": 1.4216,
+      "step": 571
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2475490196078434e-05,
+      "loss": 1.7259,
+      "step": 572
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.244485294117647e-05,
+      "loss": 1.7639,
+      "step": 573
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.241421568627451e-05,
+      "loss": 1.8631,
+      "step": 574
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2383578431372554e-05,
+      "loss": 1.6301,
+      "step": 575
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.235294117647059e-05,
+      "loss": 1.4856,
+      "step": 576
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.232230392156863e-05,
+      "loss": 1.5458,
+      "step": 577
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.229166666666667e-05,
+      "loss": 1.9028,
+      "step": 578
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.226102941176471e-05,
+      "loss": 1.5576,
+      "step": 579
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.223039215686275e-05,
+      "loss": 1.831,
+      "step": 580
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.219975490196079e-05,
+      "loss": 1.6631,
+      "step": 581
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.2169117647058826e-05,
+      "loss": 1.6729,
+      "step": 582
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.2138480392156864e-05,
+      "loss": 1.8232,
+      "step": 583
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.210784313725491e-05,
+      "loss": 1.6511,
+      "step": 584
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.2077205882352946e-05,
+      "loss": 1.6036,
+      "step": 585
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.2046568627450984e-05,
+      "loss": 1.6299,
+      "step": 586
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.201593137254902e-05,
+      "loss": 1.4939,
+      "step": 587
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.198529411764706e-05,
+      "loss": 1.7581,
+      "step": 588
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.19546568627451e-05,
+      "loss": 1.5677,
+      "step": 589
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1924019607843135e-05,
+      "loss": 1.7612,
+      "step": 590
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.189338235294117e-05,
+      "loss": 1.7087,
+      "step": 591
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.186274509803922e-05,
+      "loss": 1.7167,
+      "step": 592
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1832107843137255e-05,
+      "loss": 1.8358,
+      "step": 593
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.180147058823529e-05,
+      "loss": 1.7835,
+      "step": 594
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.177083333333333e-05,
+      "loss": 1.8237,
+      "step": 595
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1740196078431375e-05,
+      "loss": 1.7136,
+      "step": 596
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.170955882352941e-05,
+      "loss": 1.887,
+      "step": 597
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.167892156862745e-05,
+      "loss": 1.6598,
+      "step": 598
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.164828431372549e-05,
+      "loss": 1.5705,
+      "step": 599
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.161764705882353e-05,
+      "loss": 1.8555,
+      "step": 600
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.158700980392157e-05,
+      "loss": 1.6725,
+      "step": 601
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.155637254901961e-05,
+      "loss": 1.5554,
+      "step": 602
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.152573529411765e-05,
+      "loss": 1.4466,
+      "step": 603
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.1495098039215685e-05,
+      "loss": 1.6107,
+      "step": 604
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.146446078431373e-05,
+      "loss": 1.6739,
+      "step": 605
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.143382352941177e-05,
+      "loss": 1.7555,
+      "step": 606
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.1403186274509805e-05,
+      "loss": 1.7826,
+      "step": 607
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.137254901960784e-05,
+      "loss": 1.4934,
+      "step": 608
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.134191176470588e-05,
+      "loss": 1.6698,
+      "step": 609
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1311274509803925e-05,
+      "loss": 1.6754,
+      "step": 610
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.128063725490196e-05,
+      "loss": 1.477,
+      "step": 611
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.125e-05,
+      "loss": 1.615,
+      "step": 612
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.121936274509804e-05,
+      "loss": 1.5828,
+      "step": 613
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.118872549019608e-05,
+      "loss": 1.5955,
+      "step": 614
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.115808823529412e-05,
+      "loss": 1.6883,
+      "step": 615
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.112745098039216e-05,
+      "loss": 1.8553,
+      "step": 616
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.1096813725490196e-05,
+      "loss": 1.6608,
+      "step": 617
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.1066176470588234e-05,
+      "loss": 1.65,
+      "step": 618
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.103553921568628e-05,
+      "loss": 1.5788,
+      "step": 619
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.100490196078432e-05,
+      "loss": 1.7806,
+      "step": 620
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.0974264705882354e-05,
+      "loss": 1.6769,
+      "step": 621
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.094362745098039e-05,
+      "loss": 1.3841,
+      "step": 622
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.091299019607844e-05,
+      "loss": 1.5978,
+      "step": 623
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.0882352941176475e-05,
+      "loss": 1.5807,
+      "step": 624
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.085171568627451e-05,
+      "loss": 1.6677,
+      "step": 625
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.082107843137255e-05,
+      "loss": 1.5943,
+      "step": 626
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.079044117647059e-05,
+      "loss": 1.7355,
+      "step": 627
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.075980392156863e-05,
+      "loss": 1.6901,
+      "step": 628
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.072916666666667e-05,
+      "loss": 1.6211,
+      "step": 629
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.069852941176471e-05,
+      "loss": 1.9127,
+      "step": 630
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.0667892156862746e-05,
+      "loss": 1.9395,
+      "step": 631
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.063725490196079e-05,
+      "loss": 1.8865,
+      "step": 632
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.060661764705883e-05,
+      "loss": 1.7826,
+      "step": 633
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.0575980392156866e-05,
+      "loss": 1.738,
+      "step": 634
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.0545343137254904e-05,
+      "loss": 1.6746,
+      "step": 635
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.0514705882352945e-05,
+      "loss": 1.6644,
+      "step": 636
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.0484068627450983e-05,
+      "loss": 1.4546,
+      "step": 637
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.0453431372549018e-05,
+      "loss": 1.7369,
+      "step": 638
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.042279411764706e-05,
+      "loss": 1.6875,
+      "step": 639
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.0392156862745097e-05,
+      "loss": 1.6156,
+      "step": 640
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.0361519607843138e-05,
+      "loss": 1.6868,
+      "step": 641
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.0330882352941176e-05,
+      "loss": 1.7613,
+      "step": 642
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.0300245098039213e-05,
+      "loss": 1.4136,
+      "step": 643
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.0269607843137255e-05,
+      "loss": 1.7128,
+      "step": 644
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.0238970588235292e-05,
+      "loss": 1.7185,
+      "step": 645
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.0208333333333334e-05,
+      "loss": 1.8337,
+      "step": 646
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.017769607843137e-05,
+      "loss": 1.6078,
+      "step": 647
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.0147058823529413e-05,
+      "loss": 1.6927,
+      "step": 648
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.011642156862745e-05,
+      "loss": 1.6439,
+      "step": 649
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.008578431372549e-05,
+      "loss": 1.6697,
+      "step": 650
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.005514705882353e-05,
+      "loss": 1.6961,
+      "step": 651
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.002450980392157e-05,
+      "loss": 1.5553,
+      "step": 652
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.999387254901961e-05,
+      "loss": 1.4761,
+      "step": 653
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.9963235294117646e-05,
+      "loss": 1.636,
+      "step": 654
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.9932598039215687e-05,
+      "loss": 1.7085,
+      "step": 655
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.9901960784313725e-05,
+      "loss": 1.6929,
+      "step": 656
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.9871323529411766e-05,
+      "loss": 1.7305,
+      "step": 657
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.9840686274509804e-05,
+      "loss": 1.7444,
+      "step": 658
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.9810049019607845e-05,
+      "loss": 1.7815,
+      "step": 659
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.9779411764705883e-05,
+      "loss": 1.5991,
+      "step": 660
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.9748774509803924e-05,
+      "loss": 1.7003,
+      "step": 661
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.9718137254901962e-05,
+      "loss": 1.6164,
+      "step": 662
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.96875e-05,
+      "loss": 1.721,
+      "step": 663
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.965686274509804e-05,
+      "loss": 1.5968,
+      "step": 664
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.962622549019608e-05,
+      "loss": 1.6929,
+      "step": 665
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.959558823529412e-05,
+      "loss": 1.6225,
+      "step": 666
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.9564950980392158e-05,
+      "loss": 1.7525,
+      "step": 667
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.95343137254902e-05,
+      "loss": 1.7172,
+      "step": 668
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.9503676470588237e-05,
+      "loss": 1.7414,
+      "step": 669
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.9473039215686278e-05,
+      "loss": 1.8308,
+      "step": 670
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.9442401960784316e-05,
+      "loss": 1.5539,
+      "step": 671
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.9411764705882354e-05,
+      "loss": 1.6541,
+      "step": 672
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.9381127450980395e-05,
+      "loss": 1.9077,
+      "step": 673
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.9350490196078433e-05,
+      "loss": 1.7314,
+      "step": 674
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.9319852941176474e-05,
+      "loss": 1.6238,
+      "step": 675
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.9289215686274512e-05,
+      "loss": 1.7858,
+      "step": 676
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.9258578431372553e-05,
+      "loss": 1.6681,
+      "step": 677
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.922794117647059e-05,
+      "loss": 1.7997,
+      "step": 678
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.9197303921568632e-05,
+      "loss": 1.681,
+      "step": 679
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.916666666666667e-05,
+      "loss": 1.99,
+      "step": 680
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.9136029411764708e-05,
+      "loss": 1.6752,
+      "step": 681
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.910539215686275e-05,
+      "loss": 1.7652,
+      "step": 682
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.9074754901960787e-05,
+      "loss": 1.7809,
+      "step": 683
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.9044117647058828e-05,
+      "loss": 1.782,
+      "step": 684
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.9013480392156866e-05,
+      "loss": 1.5684,
+      "step": 685
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.8982843137254907e-05,
+      "loss": 1.5092,
+      "step": 686
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.8952205882352945e-05,
+      "loss": 1.6763,
+      "step": 687
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.8921568627450986e-05,
+      "loss": 1.723,
+      "step": 688
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.8890931372549017e-05,
+      "loss": 1.7951,
+      "step": 689
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.8860294117647058e-05,
+      "loss": 1.8777,
+      "step": 690
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.8829656862745096e-05,
+      "loss": 1.6097,
+      "step": 691
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.8799019607843137e-05,
+      "loss": 1.6305,
+      "step": 692
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.8768382352941175e-05,
+      "loss": 1.6756,
+      "step": 693
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.8737745098039216e-05,
+      "loss": 1.7516,
+      "step": 694
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.8707107843137254e-05,
+      "loss": 1.6561,
+      "step": 695
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.8676470588235295e-05,
+      "loss": 1.6908,
+      "step": 696
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.8645833333333333e-05,
+      "loss": 1.7896,
+      "step": 697
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.861519607843137e-05,
+      "loss": 1.726,
+      "step": 698
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.8584558823529412e-05,
+      "loss": 1.7645,
+      "step": 699
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.855392156862745e-05,
+      "loss": 1.8206,
+      "step": 700
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.852328431372549e-05,
+      "loss": 1.8183,
+      "step": 701
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.849264705882353e-05,
+      "loss": 1.6613,
+      "step": 702
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.846200980392157e-05,
+      "loss": 1.7351,
+      "step": 703
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8431372549019608e-05,
+      "loss": 1.7487,
+      "step": 704
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.840073529411765e-05,
+      "loss": 1.8985,
+      "step": 705
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.8370098039215687e-05,
+      "loss": 1.8533,
+      "step": 706
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.8339460784313724e-05,
+      "loss": 1.6274,
+      "step": 707
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.8308823529411766e-05,
+      "loss": 1.9091,
+      "step": 708
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.8278186274509803e-05,
+      "loss": 1.6962,
+      "step": 709
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.8247549019607845e-05,
+      "loss": 1.757,
+      "step": 710
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.8216911764705882e-05,
+      "loss": 1.5606,
+      "step": 711
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.8186274509803924e-05,
+      "loss": 1.6163,
+      "step": 712
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.815563725490196e-05,
+      "loss": 1.8571,
+      "step": 713
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.8125000000000003e-05,
+      "loss": 1.5966,
+      "step": 714
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.809436274509804e-05,
+      "loss": 1.6934,
+      "step": 715
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.806372549019608e-05,
+      "loss": 1.6796,
+      "step": 716
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.803308823529412e-05,
+      "loss": 1.646,
+      "step": 717
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.8002450980392157e-05,
+      "loss": 1.6574,
+      "step": 718
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.79718137254902e-05,
+      "loss": 1.5397,
+      "step": 719
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.7941176470588236e-05,
+      "loss": 1.4647,
+      "step": 720
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.7910539215686277e-05,
+      "loss": 1.7253,
+      "step": 721
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.7879901960784315e-05,
+      "loss": 1.6082,
+      "step": 722
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.7849264705882356e-05,
+      "loss": 1.8735,
+      "step": 723
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.7818627450980394e-05,
+      "loss": 1.6772,
+      "step": 724
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.7787990196078435e-05,
+      "loss": 1.9125,
+      "step": 725
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.7757352941176473e-05,
+      "loss": 1.8906,
+      "step": 726
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.772671568627451e-05,
+      "loss": 1.5515,
+      "step": 727
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.7696078431372552e-05,
+      "loss": 1.6465,
+      "step": 728
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.766544117647059e-05,
+      "loss": 1.7526,
+      "step": 729
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.763480392156863e-05,
+      "loss": 1.7528,
+      "step": 730
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.760416666666667e-05,
+      "loss": 1.7291,
+      "step": 731
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.757352941176471e-05,
+      "loss": 1.5997,
+      "step": 732
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.7542892156862748e-05,
+      "loss": 1.6914,
+      "step": 733
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.751225490196079e-05,
+      "loss": 1.8801,
+      "step": 734
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.7481617647058827e-05,
+      "loss": 1.825,
+      "step": 735
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.7450980392156865e-05,
+      "loss": 1.6848,
+      "step": 736
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.7420343137254906e-05,
+      "loss": 1.5738,
+      "step": 737
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.7389705882352944e-05,
+      "loss": 1.7428,
+      "step": 738
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.7359068627450985e-05,
+      "loss": 1.7726,
+      "step": 739
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.732843137254902e-05,
+      "loss": 1.4557,
+      "step": 740
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.7297794117647057e-05,
+      "loss": 1.45,
+      "step": 741
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.72671568627451e-05,
+      "loss": 1.6858,
+      "step": 742
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7236519607843136e-05,
+      "loss": 1.7778,
+      "step": 743
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7205882352941174e-05,
+      "loss": 1.6315,
+      "step": 744
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7175245098039215e-05,
+      "loss": 1.8072,
+      "step": 745
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.7144607843137253e-05,
+      "loss": 1.6312,
+      "step": 746
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.7113970588235294e-05,
+      "loss": 1.511,
+      "step": 747
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.7083333333333332e-05,
+      "loss": 1.7458,
+      "step": 748
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.7052696078431373e-05,
+      "loss": 1.5593,
+      "step": 749
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.702205882352941e-05,
+      "loss": 1.8739,
+      "step": 750
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.6991421568627452e-05,
+      "loss": 1.7651,
+      "step": 751
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.696078431372549e-05,
+      "loss": 1.7267,
+      "step": 752
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.6930147058823528e-05,
+      "loss": 1.5968,
+      "step": 753
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.689950980392157e-05,
+      "loss": 1.7457,
+      "step": 754
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6868872549019607e-05,
+      "loss": 1.6769,
+      "step": 755
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6838235294117648e-05,
+      "loss": 1.6302,
+      "step": 756
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6807598039215686e-05,
+      "loss": 1.6564,
+      "step": 757
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6776960784313727e-05,
+      "loss": 1.918,
+      "step": 758
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6746323529411765e-05,
+      "loss": 1.4914,
+      "step": 759
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6715686274509806e-05,
+      "loss": 1.6885,
+      "step": 760
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.6685049019607844e-05,
+      "loss": 1.9419,
+      "step": 761
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.6654411764705882e-05,
+      "loss": 1.6646,
+      "step": 762
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.6623774509803923e-05,
+      "loss": 1.8451,
+      "step": 763
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.659313725490196e-05,
+      "loss": 1.7758,
+      "step": 764
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.6562500000000002e-05,
+      "loss": 1.6429,
+      "step": 765
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.653186274509804e-05,
+      "loss": 1.5146,
+      "step": 766
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.650122549019608e-05,
+      "loss": 1.7488,
+      "step": 767
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.647058823529412e-05,
+      "loss": 1.6575,
+      "step": 768
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.643995098039216e-05,
+      "loss": 1.835,
+      "step": 769
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.6409313725490198e-05,
+      "loss": 1.6868,
+      "step": 770
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.6378676470588236e-05,
+      "loss": 1.6606,
+      "step": 771
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6348039215686277e-05,
+      "loss": 1.573,
+      "step": 772
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6317401960784315e-05,
+      "loss": 1.5283,
+      "step": 773
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.6286764705882356e-05,
+      "loss": 1.707,
+      "step": 774
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.6256127450980394e-05,
+      "loss": 1.6334,
+      "step": 775
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.6225490196078435e-05,
+      "loss": 1.7618,
+      "step": 776
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.6194852941176473e-05,
+      "loss": 1.7658,
+      "step": 777
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.6164215686274514e-05,
+      "loss": 1.7922,
+      "step": 778
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.613357843137255e-05,
+      "loss": 1.7159,
+      "step": 779
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.6102941176470593e-05,
+      "loss": 1.7026,
+      "step": 780
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.607230392156863e-05,
+      "loss": 1.7477,
+      "step": 781
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.604166666666667e-05,
+      "loss": 1.7535,
+      "step": 782
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.601102941176471e-05,
+      "loss": 1.6735,
+      "step": 783
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.5980392156862747e-05,
+      "loss": 1.7532,
+      "step": 784
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.594975490196079e-05,
+      "loss": 1.6197,
+      "step": 785
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.5919117647058826e-05,
+      "loss": 1.7445,
+      "step": 786
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.5888480392156868e-05,
+      "loss": 1.6087,
+      "step": 787
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.5857843137254905e-05,
+      "loss": 1.7531,
+      "step": 788
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.5827205882352947e-05,
+      "loss": 1.7144,
+      "step": 789
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.5796568627450984e-05,
+      "loss": 1.8403,
+      "step": 790
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.576593137254902e-05,
+      "loss": 1.5874,
+      "step": 791
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.5735294117647057e-05,
+      "loss": 1.7882,
+      "step": 792
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.5704656862745098e-05,
+      "loss": 1.8619,
+      "step": 793
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.5674019607843136e-05,
+      "loss": 1.6761,
+      "step": 794
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.5643382352941177e-05,
+      "loss": 1.62,
+      "step": 795
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5612745098039215e-05,
+      "loss": 1.6485,
+      "step": 796
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5582107843137252e-05,
+      "loss": 1.4833,
+      "step": 797
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5551470588235294e-05,
+      "loss": 1.6203,
+      "step": 798
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.552083333333333e-05,
+      "loss": 1.6494,
+      "step": 799
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.5490196078431373e-05,
+      "loss": 1.825,
+      "step": 800
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.545955882352941e-05,
+      "loss": 1.7217,
+      "step": 801
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.542892156862745e-05,
+      "loss": 1.6608,
+      "step": 802
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.539828431372549e-05,
+      "loss": 1.5793,
+      "step": 803
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.536764705882353e-05,
+      "loss": 1.5895,
+      "step": 804
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.533700980392157e-05,
+      "loss": 1.8166,
+      "step": 805
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.530637254901961e-05,
+      "loss": 1.7811,
+      "step": 806
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.5275735294117647e-05,
+      "loss": 1.679,
+      "step": 807
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.5245098039215685e-05,
+      "loss": 1.9753,
+      "step": 808
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.5214460784313726e-05,
+      "loss": 1.4458,
+      "step": 809
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5183823529411764e-05,
+      "loss": 1.8178,
+      "step": 810
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5153186274509805e-05,
+      "loss": 1.5732,
+      "step": 811
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.5122549019607843e-05,
+      "loss": 1.6766,
+      "step": 812
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.5091911764705884e-05,
+      "loss": 1.7434,
+      "step": 813
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.5061274509803922e-05,
+      "loss": 1.6887,
+      "step": 814
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.5030637254901963e-05,
+      "loss": 1.5294,
+      "step": 815
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.5e-05,
+      "loss": 1.9789,
+      "step": 816
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.496936274509804e-05,
+      "loss": 1.508,
+      "step": 817
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.493872549019608e-05,
+      "loss": 1.6912,
+      "step": 818
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.4908088235294118e-05,
+      "loss": 1.6087,
+      "step": 819
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.487745098039216e-05,
+      "loss": 1.5319,
+      "step": 820
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.4846813725490197e-05,
+      "loss": 1.6139,
+      "step": 821
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.4816176470588238e-05,
+      "loss": 1.5932,
+      "step": 822
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.4785539215686276e-05,
+      "loss": 1.6491,
+      "step": 823
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.4754901960784317e-05,
+      "loss": 1.6825,
+      "step": 824
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.4724264705882355e-05,
+      "loss": 1.5624,
+      "step": 825
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.4693627450980393e-05,
+      "loss": 1.6406,
+      "step": 826
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.4662990196078434e-05,
+      "loss": 1.5463,
+      "step": 827
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.4632352941176472e-05,
+      "loss": 1.4845,
+      "step": 828
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.460171568627451e-05,
+      "loss": 1.4856,
+      "step": 829
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.4571078431372547e-05,
+      "loss": 1.6792,
+      "step": 830
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.454044117647059e-05,
+      "loss": 1.4757,
+      "step": 831
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.4509803921568626e-05,
+      "loss": 1.3586,
+      "step": 832
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.4479166666666668e-05,
+      "loss": 1.4772,
+      "step": 833
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.4448529411764705e-05,
+      "loss": 1.5748,
+      "step": 834
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.4417892156862747e-05,
+      "loss": 1.5252,
+      "step": 835
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.4387254901960784e-05,
+      "loss": 1.3208,
+      "step": 836
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.4356617647058826e-05,
+      "loss": 1.4126,
+      "step": 837
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.4325980392156863e-05,
+      "loss": 1.2898,
+      "step": 838
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.42953431372549e-05,
+      "loss": 1.4466,
+      "step": 839
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.4264705882352942e-05,
+      "loss": 1.5796,
+      "step": 840
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.423406862745098e-05,
+      "loss": 1.5257,
+      "step": 841
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.420343137254902e-05,
+      "loss": 1.618,
+      "step": 842
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.417279411764706e-05,
+      "loss": 1.6186,
+      "step": 843
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.41421568627451e-05,
+      "loss": 1.4637,
+      "step": 844
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.4111519607843138e-05,
+      "loss": 1.4887,
+      "step": 845
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.408088235294118e-05,
+      "loss": 1.4642,
+      "step": 846
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.4050245098039217e-05,
+      "loss": 1.6438,
+      "step": 847
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.401960784313726e-05,
+      "loss": 1.6198,
+      "step": 848
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.3988970588235296e-05,
+      "loss": 1.6453,
+      "step": 849
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.3958333333333334e-05,
+      "loss": 1.3465,
+      "step": 850
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.3927696078431375e-05,
+      "loss": 1.3834,
+      "step": 851
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.3897058823529413e-05,
+      "loss": 1.5953,
+      "step": 852
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.3866421568627454e-05,
+      "loss": 1.6504,
+      "step": 853
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.3835784313725492e-05,
+      "loss": 1.6159,
+      "step": 854
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.380514705882353e-05,
+      "loss": 1.5286,
+      "step": 855
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.3774509803921568e-05,
+      "loss": 1.3925,
+      "step": 856
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.374387254901961e-05,
+      "loss": 1.6244,
+      "step": 857
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.3713235294117647e-05,
+      "loss": 1.5326,
+      "step": 858
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.3682598039215688e-05,
+      "loss": 1.6073,
+      "step": 859
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.3651960784313726e-05,
+      "loss": 1.555,
+      "step": 860
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.3621323529411767e-05,
+      "loss": 1.5101,
+      "step": 861
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.3590686274509805e-05,
+      "loss": 1.6417,
+      "step": 862
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.3560049019607842e-05,
+      "loss": 1.4235,
+      "step": 863
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.3529411764705884e-05,
+      "loss": 1.6864,
+      "step": 864
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.349877450980392e-05,
+      "loss": 1.4341,
+      "step": 865
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.3468137254901963e-05,
+      "loss": 1.6632,
+      "step": 866
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.34375e-05,
+      "loss": 1.5138,
+      "step": 867
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.340686274509804e-05,
+      "loss": 1.4685,
+      "step": 868
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.337622549019608e-05,
+      "loss": 1.4749,
+      "step": 869
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.334558823529412e-05,
+      "loss": 1.4851,
+      "step": 870
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.331495098039216e-05,
+      "loss": 1.7464,
+      "step": 871
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.3284313725490196e-05,
+      "loss": 1.6437,
+      "step": 872
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.3253676470588237e-05,
+      "loss": 1.4297,
+      "step": 873
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.3223039215686275e-05,
+      "loss": 1.5368,
+      "step": 874
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.3192401960784316e-05,
+      "loss": 1.5292,
+      "step": 875
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.3161764705882354e-05,
+      "loss": 1.5702,
+      "step": 876
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.3131127450980395e-05,
+      "loss": 1.396,
+      "step": 877
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.3100490196078433e-05,
+      "loss": 1.5315,
+      "step": 878
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.3069852941176474e-05,
+      "loss": 1.629,
+      "step": 879
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.303921568627451e-05,
+      "loss": 1.5076,
+      "step": 880
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.300857843137255e-05,
+      "loss": 1.6033,
+      "step": 881
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.2977941176470588e-05,
+      "loss": 1.7086,
+      "step": 882
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.294730392156863e-05,
+      "loss": 1.6474,
+      "step": 883
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.2916666666666667e-05,
+      "loss": 1.6159,
+      "step": 884
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.2886029411764705e-05,
+      "loss": 1.8269,
+      "step": 885
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.2855392156862746e-05,
+      "loss": 1.6248,
+      "step": 886
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.2824754901960784e-05,
+      "loss": 1.5715,
+      "step": 887
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.2794117647058825e-05,
+      "loss": 1.5921,
+      "step": 888
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.2763480392156863e-05,
+      "loss": 1.4613,
+      "step": 889
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.2732843137254904e-05,
+      "loss": 1.6952,
+      "step": 890
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.2702205882352942e-05,
+      "loss": 1.6108,
+      "step": 891
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.2671568627450983e-05,
+      "loss": 1.4386,
+      "step": 892
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.264093137254902e-05,
+      "loss": 1.4097,
+      "step": 893
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.261029411764706e-05,
+      "loss": 1.5939,
+      "step": 894
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.25796568627451e-05,
+      "loss": 1.7092,
+      "step": 895
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.2549019607843138e-05,
+      "loss": 1.5576,
+      "step": 896
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.251838235294118e-05,
+      "loss": 1.393,
+      "step": 897
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.2487745098039217e-05,
+      "loss": 1.4851,
+      "step": 898
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.2457107843137258e-05,
+      "loss": 1.6144,
+      "step": 899
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.2426470588235296e-05,
+      "loss": 1.4565,
+      "step": 900
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.2395833333333337e-05,
+      "loss": 1.5017,
+      "step": 901
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.2365196078431375e-05,
+      "loss": 1.5192,
+      "step": 902
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.2334558823529412e-05,
+      "loss": 1.5872,
+      "step": 903
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.2303921568627454e-05,
+      "loss": 1.352,
+      "step": 904
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.227328431372549e-05,
+      "loss": 1.6783,
+      "step": 905
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.224264705882353e-05,
+      "loss": 1.5861,
+      "step": 906
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.2212009803921567e-05,
+      "loss": 1.5041,
+      "step": 907
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.2181372549019608e-05,
+      "loss": 1.6788,
+      "step": 908
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.2150735294117646e-05,
+      "loss": 1.5606,
+      "step": 909
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2.2120098039215687e-05,
+      "loss": 1.4601,
+      "step": 910
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2.2089460784313725e-05,
+      "loss": 1.5232,
+      "step": 911
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2.2058823529411766e-05,
+      "loss": 1.4606,
+      "step": 912
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2.2028186274509804e-05,
+      "loss": 1.3622,
+      "step": 913
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2.1997549019607845e-05,
+      "loss": 1.6201,
+      "step": 914
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2.1966911764705883e-05,
+      "loss": 1.432,
+      "step": 915
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2.193627450980392e-05,
+      "loss": 1.4297,
+      "step": 916
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2.1905637254901962e-05,
+      "loss": 1.5281,
+      "step": 917
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2.1875e-05,
+      "loss": 1.5213,
+      "step": 918
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2.184436274509804e-05,
+      "loss": 1.5982,
+      "step": 919
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2.181372549019608e-05,
+      "loss": 1.3857,
+      "step": 920
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2.178308823529412e-05,
+      "loss": 1.363,
+      "step": 921
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2.1752450980392158e-05,
+      "loss": 1.5072,
+      "step": 922
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2.17218137254902e-05,
+      "loss": 1.4379,
+      "step": 923
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.1691176470588237e-05,
+      "loss": 1.5319,
+      "step": 924
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.1660539215686278e-05,
+      "loss": 1.5453,
+      "step": 925
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.1629901960784316e-05,
+      "loss": 1.639,
+      "step": 926
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2.1599264705882354e-05,
+      "loss": 1.5594,
+      "step": 927
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2.1568627450980395e-05,
+      "loss": 1.4636,
+      "step": 928
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.1537990196078433e-05,
+      "loss": 1.456,
+      "step": 929
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.1507352941176474e-05,
+      "loss": 1.5959,
+      "step": 930
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.1476715686274508e-05,
+      "loss": 1.6251,
+      "step": 931
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.144607843137255e-05,
+      "loss": 1.5132,
+      "step": 932
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.1415441176470587e-05,
+      "loss": 1.6091,
+      "step": 933
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.138480392156863e-05,
+      "loss": 1.6581,
+      "step": 934
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.1354166666666666e-05,
+      "loss": 1.3676,
+      "step": 935
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.1323529411764707e-05,
+      "loss": 1.3934,
+      "step": 936
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.1292892156862745e-05,
+      "loss": 1.6582,
+      "step": 937
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.1262254901960786e-05,
+      "loss": 1.574,
+      "step": 938
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.1231617647058824e-05,
+      "loss": 1.5802,
+      "step": 939
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.1200980392156862e-05,
+      "loss": 1.6085,
+      "step": 940
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.1170343137254903e-05,
+      "loss": 1.6661,
+      "step": 941
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.113970588235294e-05,
+      "loss": 1.3744,
+      "step": 942
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.1109068627450982e-05,
+      "loss": 1.5277,
+      "step": 943
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.107843137254902e-05,
+      "loss": 1.4371,
+      "step": 944
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.104779411764706e-05,
+      "loss": 1.4866,
+      "step": 945
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.10171568627451e-05,
+      "loss": 1.3851,
+      "step": 946
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.098651960784314e-05,
+      "loss": 1.7386,
+      "step": 947
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.0955882352941178e-05,
+      "loss": 1.6862,
+      "step": 948
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.0925245098039216e-05,
+      "loss": 1.585,
+      "step": 949
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.0894607843137257e-05,
+      "loss": 1.5413,
+      "step": 950
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.0863970588235295e-05,
+      "loss": 1.8423,
+      "step": 951
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.0833333333333336e-05,
+      "loss": 1.6848,
+      "step": 952
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.0802696078431374e-05,
+      "loss": 1.5684,
+      "step": 953
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.0772058823529415e-05,
+      "loss": 1.4456,
+      "step": 954
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.0741421568627453e-05,
+      "loss": 1.5642,
+      "step": 955
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.0710784313725494e-05,
+      "loss": 1.5921,
+      "step": 956
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.068014705882353e-05,
+      "loss": 1.4654,
+      "step": 957
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.064950980392157e-05,
+      "loss": 1.4524,
+      "step": 958
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.0618872549019607e-05,
+      "loss": 1.61,
+      "step": 959
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.058823529411765e-05,
+      "loss": 1.4104,
+      "step": 960
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.0557598039215686e-05,
+      "loss": 1.4319,
+      "step": 961
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.0526960784313724e-05,
+      "loss": 1.3024,
+      "step": 962
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.0496323529411765e-05,
+      "loss": 1.5844,
+      "step": 963
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.0465686274509803e-05,
+      "loss": 1.3626,
+      "step": 964
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.0435049019607844e-05,
+      "loss": 1.3839,
+      "step": 965
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.0404411764705882e-05,
+      "loss": 1.431,
+      "step": 966
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.0373774509803923e-05,
+      "loss": 1.504,
+      "step": 967
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.034313725490196e-05,
+      "loss": 1.6262,
+      "step": 968
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.0312500000000002e-05,
+      "loss": 1.613,
+      "step": 969
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.028186274509804e-05,
+      "loss": 1.6319,
+      "step": 970
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.0251225490196078e-05,
+      "loss": 1.5713,
+      "step": 971
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.022058823529412e-05,
+      "loss": 1.552,
+      "step": 972
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.0189950980392157e-05,
+      "loss": 1.5736,
+      "step": 973
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.0159313725490198e-05,
+      "loss": 1.4066,
+      "step": 974
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.0128676470588236e-05,
+      "loss": 1.7287,
+      "step": 975
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.0098039215686277e-05,
+      "loss": 1.3724,
+      "step": 976
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.0067401960784315e-05,
+      "loss": 1.5404,
+      "step": 977
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.0036764705882356e-05,
+      "loss": 1.5954,
+      "step": 978
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.0006127450980394e-05,
+      "loss": 1.6406,
+      "step": 979
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.9975490196078432e-05,
+      "loss": 1.5424,
+      "step": 980
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.9944852941176473e-05,
+      "loss": 1.6974,
+      "step": 981
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.991421568627451e-05,
+      "loss": 1.4865,
+      "step": 982
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.988357843137255e-05,
+      "loss": 1.6517,
+      "step": 983
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.9852941176470586e-05,
+      "loss": 1.5225,
+      "step": 984
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.9822303921568628e-05,
+      "loss": 1.4705,
+      "step": 985
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.9791666666666665e-05,
+      "loss": 1.4987,
+      "step": 986
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.9761029411764707e-05,
+      "loss": 1.6709,
+      "step": 987
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.9730392156862744e-05,
+      "loss": 1.4193,
+      "step": 988
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.9699754901960786e-05,
+      "loss": 1.6389,
+      "step": 989
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.9669117647058824e-05,
+      "loss": 1.399,
+      "step": 990
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.9638480392156865e-05,
+      "loss": 1.5216,
+      "step": 991
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.9607843137254903e-05,
+      "loss": 1.5682,
+      "step": 992
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.957720588235294e-05,
+      "loss": 1.7917,
+      "step": 993
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.954656862745098e-05,
+      "loss": 1.6278,
+      "step": 994
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.951593137254902e-05,
+      "loss": 1.5897,
+      "step": 995
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.948529411764706e-05,
+      "loss": 1.4257,
+      "step": 996
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.94546568627451e-05,
+      "loss": 1.7129,
+      "step": 997
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.942401960784314e-05,
+      "loss": 1.5397,
+      "step": 998
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.9393382352941177e-05,
+      "loss": 1.5283,
+      "step": 999
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.936274509803922e-05,
+      "loss": 1.5265,
+      "step": 1000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.9332107843137256e-05,
+      "loss": 1.4652,
+      "step": 1001
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.9301470588235298e-05,
+      "loss": 1.4885,
+      "step": 1002
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.9270833333333335e-05,
+      "loss": 1.7961,
+      "step": 1003
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.9240196078431373e-05,
+      "loss": 1.5605,
+      "step": 1004
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.9209558823529414e-05,
+      "loss": 1.4952,
+      "step": 1005
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.9178921568627452e-05,
+      "loss": 1.4335,
+      "step": 1006
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.9148284313725493e-05,
+      "loss": 1.5382,
+      "step": 1007
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.9117647058823528e-05,
+      "loss": 1.4546,
+      "step": 1008
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.908700980392157e-05,
+      "loss": 1.3929,
+      "step": 1009
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.9056372549019607e-05,
+      "loss": 1.628,
+      "step": 1010
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.9025735294117648e-05,
+      "loss": 1.555,
+      "step": 1011
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.8995098039215686e-05,
+      "loss": 1.3743,
+      "step": 1012
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.8964460784313727e-05,
+      "loss": 1.5469,
+      "step": 1013
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.8933823529411765e-05,
+      "loss": 1.482,
+      "step": 1014
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.8903186274509806e-05,
+      "loss": 1.5306,
+      "step": 1015
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.8872549019607844e-05,
+      "loss": 1.701,
+      "step": 1016
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.884191176470588e-05,
+      "loss": 1.4249,
+      "step": 1017
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.8811274509803923e-05,
+      "loss": 1.6122,
+      "step": 1018
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.878063725490196e-05,
+      "loss": 1.6724,
+      "step": 1019
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 1.4886,
+      "step": 1020
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.871936274509804e-05,
+      "loss": 1.3535,
+      "step": 1021
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.868872549019608e-05,
+      "loss": 1.6116,
+      "step": 1022
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.865808823529412e-05,
+      "loss": 1.5913,
+      "step": 1023
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.862745098039216e-05,
+      "loss": 1.704,
+      "step": 1024
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.8596813725490198e-05,
+      "loss": 1.7096,
+      "step": 1025
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.8566176470588235e-05,
+      "loss": 1.396,
+      "step": 1026
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.8535539215686277e-05,
+      "loss": 1.6076,
+      "step": 1027
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.8504901960784314e-05,
+      "loss": 1.4886,
+      "step": 1028
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.8474264705882356e-05,
+      "loss": 1.7009,
+      "step": 1029
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.8443627450980393e-05,
+      "loss": 1.5554,
+      "step": 1030
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.8412990196078435e-05,
+      "loss": 1.7629,
+      "step": 1031
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.8382352941176472e-05,
+      "loss": 1.4118,
+      "step": 1032
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.835171568627451e-05,
+      "loss": 1.6175,
+      "step": 1033
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.8321078431372548e-05,
+      "loss": 1.7014,
+      "step": 1034
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.829044117647059e-05,
+      "loss": 1.3622,
+      "step": 1035
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.8259803921568627e-05,
+      "loss": 1.6063,
+      "step": 1036
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.8229166666666668e-05,
+      "loss": 1.4624,
+      "step": 1037
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.8198529411764706e-05,
+      "loss": 1.5711,
+      "step": 1038
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.8167892156862744e-05,
+      "loss": 1.496,
+      "step": 1039
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.8137254901960785e-05,
+      "loss": 1.6719,
+      "step": 1040
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.8106617647058823e-05,
+      "loss": 1.5207,
+      "step": 1041
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.8075980392156864e-05,
+      "loss": 1.6701,
+      "step": 1042
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.8045343137254902e-05,
+      "loss": 1.4148,
+      "step": 1043
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.8014705882352943e-05,
+      "loss": 1.572,
+      "step": 1044
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.798406862745098e-05,
+      "loss": 1.4964,
+      "step": 1045
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.7953431372549022e-05,
+      "loss": 1.6377,
+      "step": 1046
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.792279411764706e-05,
+      "loss": 1.4529,
+      "step": 1047
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.7892156862745098e-05,
+      "loss": 1.5799,
+      "step": 1048
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.786151960784314e-05,
+      "loss": 1.59,
+      "step": 1049
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.7830882352941177e-05,
+      "loss": 1.5418,
+      "step": 1050
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.7800245098039218e-05,
+      "loss": 1.4932,
+      "step": 1051
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7769607843137256e-05,
+      "loss": 1.4915,
+      "step": 1052
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7738970588235297e-05,
+      "loss": 1.409,
+      "step": 1053
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7708333333333335e-05,
+      "loss": 1.6133,
+      "step": 1054
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7677696078431376e-05,
+      "loss": 1.6306,
+      "step": 1055
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7647058823529414e-05,
+      "loss": 1.5727,
+      "step": 1056
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.761642156862745e-05,
+      "loss": 1.7368,
+      "step": 1057
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.7585784313725493e-05,
+      "loss": 1.5548,
+      "step": 1058
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.755514705882353e-05,
+      "loss": 1.7127,
+      "step": 1059
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.7524509803921568e-05,
+      "loss": 1.5802,
+      "step": 1060
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.7493872549019606e-05,
+      "loss": 1.5354,
+      "step": 1061
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.7463235294117647e-05,
+      "loss": 1.522,
+      "step": 1062
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.7432598039215685e-05,
+      "loss": 1.6673,
+      "step": 1063
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.7401960784313726e-05,
+      "loss": 1.5218,
+      "step": 1064
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.7371323529411764e-05,
+      "loss": 1.6205,
+      "step": 1065
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.7340686274509805e-05,
+      "loss": 1.5405,
+      "step": 1066
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.7310049019607843e-05,
+      "loss": 1.346,
+      "step": 1067
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.7279411764705884e-05,
+      "loss": 1.6793,
+      "step": 1068
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.7248774509803922e-05,
+      "loss": 1.2233,
+      "step": 1069
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.721813725490196e-05,
+      "loss": 1.6132,
+      "step": 1070
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.71875e-05,
+      "loss": 1.5816,
+      "step": 1071
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.715686274509804e-05,
+      "loss": 1.54,
+      "step": 1072
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.712622549019608e-05,
+      "loss": 1.6634,
+      "step": 1073
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.7095588235294118e-05,
+      "loss": 1.5432,
+      "step": 1074
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.706495098039216e-05,
+      "loss": 1.5715,
+      "step": 1075
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.7034313725490197e-05,
+      "loss": 1.4424,
+      "step": 1076
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.7003676470588238e-05,
+      "loss": 1.5179,
+      "step": 1077
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.6973039215686276e-05,
+      "loss": 1.744,
+      "step": 1078
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.6942401960784317e-05,
+      "loss": 1.6881,
+      "step": 1079
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.6911764705882355e-05,
+      "loss": 1.5537,
+      "step": 1080
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.6881127450980393e-05,
+      "loss": 1.6063,
+      "step": 1081
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.6850490196078434e-05,
+      "loss": 1.6029,
+      "step": 1082
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.681985294117647e-05,
+      "loss": 1.475,
+      "step": 1083
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.678921568627451e-05,
+      "loss": 1.4981,
+      "step": 1084
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.6758578431372547e-05,
+      "loss": 1.3653,
+      "step": 1085
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.672794117647059e-05,
+      "loss": 1.4921,
+      "step": 1086
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.6697303921568626e-05,
+      "loss": 1.4773,
+      "step": 1087
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 1.504,
+      "step": 1088
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.6636029411764705e-05,
+      "loss": 1.5356,
+      "step": 1089
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.6605392156862746e-05,
+      "loss": 1.3539,
+      "step": 1090
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.6574754901960784e-05,
+      "loss": 1.3739,
+      "step": 1091
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.6544117647058825e-05,
+      "loss": 1.4307,
+      "step": 1092
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.6513480392156863e-05,
+      "loss": 1.3339,
+      "step": 1093
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.64828431372549e-05,
+      "loss": 1.3471,
+      "step": 1094
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.6452205882352942e-05,
+      "loss": 1.4513,
+      "step": 1095
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.642156862745098e-05,
+      "loss": 1.4716,
+      "step": 1096
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.639093137254902e-05,
+      "loss": 1.5006,
+      "step": 1097
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.636029411764706e-05,
+      "loss": 1.467,
+      "step": 1098
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.63296568627451e-05,
+      "loss": 1.443,
+      "step": 1099
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.6299019607843138e-05,
+      "loss": 1.3968,
+      "step": 1100
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.626838235294118e-05,
+      "loss": 1.3921,
+      "step": 1101
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.6237745098039217e-05,
+      "loss": 1.4596,
+      "step": 1102
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.6207107843137255e-05,
+      "loss": 1.3681,
+      "step": 1103
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.6176470588235296e-05,
+      "loss": 1.4834,
+      "step": 1104
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.6145833333333334e-05,
+      "loss": 1.6039,
+      "step": 1105
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.6115196078431375e-05,
+      "loss": 1.6279,
+      "step": 1106
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.6084558823529413e-05,
+      "loss": 1.4857,
+      "step": 1107
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.6053921568627454e-05,
+      "loss": 1.4268,
+      "step": 1108
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.6023284313725492e-05,
+      "loss": 1.4589,
+      "step": 1109
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.599264705882353e-05,
+      "loss": 1.5033,
+      "step": 1110
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.5962009803921568e-05,
+      "loss": 1.3222,
+      "step": 1111
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.593137254901961e-05,
+      "loss": 1.3264,
+      "step": 1112
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.5900735294117647e-05,
+      "loss": 1.3563,
+      "step": 1113
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.5870098039215688e-05,
+      "loss": 1.394,
+      "step": 1114
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.5839460784313726e-05,
+      "loss": 1.5443,
+      "step": 1115
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.5808823529411763e-05,
+      "loss": 1.4671,
+      "step": 1116
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.5778186274509805e-05,
+      "loss": 1.3731,
+      "step": 1117
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.5747549019607842e-05,
+      "loss": 1.5471,
+      "step": 1118
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.5716911764705884e-05,
+      "loss": 1.407,
+      "step": 1119
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.568627450980392e-05,
+      "loss": 1.3053,
+      "step": 1120
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5655637254901963e-05,
+      "loss": 1.4517,
+      "step": 1121
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5625e-05,
+      "loss": 1.4396,
+      "step": 1122
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.559436274509804e-05,
+      "loss": 1.3251,
+      "step": 1123
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.556372549019608e-05,
+      "loss": 1.358,
+      "step": 1124
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5533088235294117e-05,
+      "loss": 1.2287,
+      "step": 1125
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.550245098039216e-05,
+      "loss": 1.5335,
+      "step": 1126
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5471813725490196e-05,
+      "loss": 1.5099,
+      "step": 1127
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.5441176470588237e-05,
+      "loss": 1.4488,
+      "step": 1128
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.5410539215686275e-05,
+      "loss": 1.4609,
+      "step": 1129
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.5379901960784316e-05,
+      "loss": 1.5629,
+      "step": 1130
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.5349264705882354e-05,
+      "loss": 1.4651,
+      "step": 1131
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.5318627450980395e-05,
+      "loss": 1.4663,
+      "step": 1132
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.5287990196078433e-05,
+      "loss": 1.3572,
+      "step": 1133
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.5257352941176473e-05,
+      "loss": 1.3044,
+      "step": 1134
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.5226715686274509e-05,
+      "loss": 1.346,
+      "step": 1135
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.5196078431372548e-05,
+      "loss": 1.576,
+      "step": 1136
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.5165441176470588e-05,
+      "loss": 1.4592,
+      "step": 1137
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.5134803921568627e-05,
+      "loss": 1.4666,
+      "step": 1138
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.5104166666666667e-05,
+      "loss": 1.4533,
+      "step": 1139
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.5073529411764706e-05,
+      "loss": 1.2686,
+      "step": 1140
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.5042892156862746e-05,
+      "loss": 1.3723,
+      "step": 1141
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.5012254901960785e-05,
+      "loss": 1.5566,
+      "step": 1142
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.4981617647058823e-05,
+      "loss": 1.3036,
+      "step": 1143
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.4950980392156863e-05,
+      "loss": 1.489,
+      "step": 1144
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.4920343137254902e-05,
+      "loss": 1.3525,
+      "step": 1145
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.4889705882352942e-05,
+      "loss": 1.4423,
+      "step": 1146
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.4859068627450981e-05,
+      "loss": 1.5834,
+      "step": 1147
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.482843137254902e-05,
+      "loss": 1.4495,
+      "step": 1148
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.479779411764706e-05,
+      "loss": 1.3656,
+      "step": 1149
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.47671568627451e-05,
+      "loss": 1.547,
+      "step": 1150
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.4736519607843139e-05,
+      "loss": 1.3298,
+      "step": 1151
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.4705882352941177e-05,
+      "loss": 1.4765,
+      "step": 1152
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.4675245098039216e-05,
+      "loss": 1.6715,
+      "step": 1153
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.4644607843137256e-05,
+      "loss": 1.5626,
+      "step": 1154
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.4613970588235295e-05,
+      "loss": 1.3637,
+      "step": 1155
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.4583333333333335e-05,
+      "loss": 1.3981,
+      "step": 1156
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.4552696078431374e-05,
+      "loss": 1.5173,
+      "step": 1157
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.4522058823529414e-05,
+      "loss": 1.4549,
+      "step": 1158
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.4491421568627453e-05,
+      "loss": 1.414,
+      "step": 1159
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.4460784313725493e-05,
+      "loss": 1.3995,
+      "step": 1160
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.4430147058823529e-05,
+      "loss": 1.3785,
+      "step": 1161
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.4399509803921568e-05,
+      "loss": 1.4794,
+      "step": 1162
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.4368872549019608e-05,
+      "loss": 1.5118,
+      "step": 1163
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.4338235294117647e-05,
+      "loss": 1.5511,
+      "step": 1164
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.4307598039215685e-05,
+      "loss": 1.2987,
+      "step": 1165
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4276960784313725e-05,
+      "loss": 1.1585,
+      "step": 1166
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4246323529411764e-05,
+      "loss": 1.4099,
+      "step": 1167
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4215686274509804e-05,
+      "loss": 1.3679,
+      "step": 1168
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4185049019607843e-05,
+      "loss": 1.4253,
+      "step": 1169
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4154411764705883e-05,
+      "loss": 1.4279,
+      "step": 1170
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4123774509803922e-05,
+      "loss": 1.4769,
+      "step": 1171
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4093137254901962e-05,
+      "loss": 1.4458,
+      "step": 1172
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4062500000000001e-05,
+      "loss": 1.598,
+      "step": 1173
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.403186274509804e-05,
+      "loss": 1.5248,
+      "step": 1174
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.4001225490196079e-05,
+      "loss": 1.5028,
+      "step": 1175
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3970588235294118e-05,
+      "loss": 1.5491,
+      "step": 1176
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3939950980392158e-05,
+      "loss": 1.3717,
+      "step": 1177
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3909313725490197e-05,
+      "loss": 1.3945,
+      "step": 1178
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3878676470588237e-05,
+      "loss": 1.3055,
+      "step": 1179
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3848039215686276e-05,
+      "loss": 1.457,
+      "step": 1180
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3817401960784316e-05,
+      "loss": 1.3625,
+      "step": 1181
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.3786764705882355e-05,
+      "loss": 1.4535,
+      "step": 1182
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.3756127450980395e-05,
+      "loss": 1.5061,
+      "step": 1183
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.3725490196078432e-05,
+      "loss": 1.3891,
+      "step": 1184
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.3694852941176472e-05,
+      "loss": 1.351,
+      "step": 1185
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.366421568627451e-05,
+      "loss": 1.5714,
+      "step": 1186
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.363357843137255e-05,
+      "loss": 1.5037,
+      "step": 1187
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.3602941176470587e-05,
+      "loss": 1.6476,
+      "step": 1188
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.3572303921568627e-05,
+      "loss": 1.3652,
+      "step": 1189
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.3541666666666666e-05,
+      "loss": 1.3413,
+      "step": 1190
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.3511029411764706e-05,
+      "loss": 1.3947,
+      "step": 1191
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.3480392156862745e-05,
+      "loss": 1.3582,
+      "step": 1192
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.3449754901960785e-05,
+      "loss": 1.3612,
+      "step": 1193
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.3419117647058824e-05,
+      "loss": 1.3941,
+      "step": 1194
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.3388480392156864e-05,
+      "loss": 1.3693,
+      "step": 1195
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.3357843137254903e-05,
+      "loss": 1.4676,
+      "step": 1196
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.3327205882352941e-05,
+      "loss": 1.4294,
+      "step": 1197
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.329656862745098e-05,
+      "loss": 1.3346,
+      "step": 1198
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.326593137254902e-05,
+      "loss": 1.287,
+      "step": 1199
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.323529411764706e-05,
+      "loss": 1.3284,
+      "step": 1200
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.3204656862745099e-05,
+      "loss": 1.3557,
+      "step": 1201
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.3174019607843138e-05,
+      "loss": 1.3663,
+      "step": 1202
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.3143382352941178e-05,
+      "loss": 1.5017,
+      "step": 1203
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.3112745098039217e-05,
+      "loss": 1.2696,
+      "step": 1204
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.3082107843137257e-05,
+      "loss": 1.3,
+      "step": 1205
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.3051470588235296e-05,
+      "loss": 1.544,
+      "step": 1206
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.3020833333333334e-05,
+      "loss": 1.4967,
+      "step": 1207
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.2990196078431374e-05,
+      "loss": 1.4816,
+      "step": 1208
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.2959558823529413e-05,
+      "loss": 1.4502,
+      "step": 1209
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.2928921568627453e-05,
+      "loss": 1.5155,
+      "step": 1210
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.2898284313725492e-05,
+      "loss": 1.5773,
+      "step": 1211
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.2867647058823528e-05,
+      "loss": 1.4237,
+      "step": 1212
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.2837009803921568e-05,
+      "loss": 1.4368,
+      "step": 1213
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.2806372549019607e-05,
+      "loss": 1.3629,
+      "step": 1214
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.2775735294117647e-05,
+      "loss": 1.2714,
+      "step": 1215
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.2745098039215686e-05,
+      "loss": 1.5067,
+      "step": 1216
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.2714460784313726e-05,
+      "loss": 1.2424,
+      "step": 1217
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.2683823529411765e-05,
+      "loss": 1.4043,
+      "step": 1218
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.2653186274509805e-05,
+      "loss": 1.3204,
+      "step": 1219
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.2622549019607843e-05,
+      "loss": 1.6397,
+      "step": 1220
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.2591911764705882e-05,
+      "loss": 1.4565,
+      "step": 1221
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.2561274509803922e-05,
+      "loss": 1.4553,
+      "step": 1222
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.2530637254901961e-05,
+      "loss": 1.4254,
+      "step": 1223
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.25e-05,
+      "loss": 1.4327,
+      "step": 1224
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.246936274509804e-05,
+      "loss": 1.3844,
+      "step": 1225
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.243872549019608e-05,
+      "loss": 1.4984,
+      "step": 1226
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2408088235294119e-05,
+      "loss": 1.5168,
+      "step": 1227
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2377450980392159e-05,
+      "loss": 1.4606,
+      "step": 1228
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2346813725490196e-05,
+      "loss": 1.4188,
+      "step": 1229
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2316176470588236e-05,
+      "loss": 1.4727,
+      "step": 1230
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.2285539215686274e-05,
+      "loss": 1.3193,
+      "step": 1231
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.2254901960784313e-05,
+      "loss": 1.4957,
+      "step": 1232
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.2224264705882353e-05,
+      "loss": 1.371,
+      "step": 1233
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.2193627450980392e-05,
+      "loss": 1.4556,
+      "step": 1234
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.2162990196078432e-05,
+      "loss": 1.3747,
+      "step": 1235
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.2132352941176471e-05,
+      "loss": 1.5551,
+      "step": 1236
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.210171568627451e-05,
+      "loss": 1.3392,
+      "step": 1237
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.207107843137255e-05,
+      "loss": 1.3955,
+      "step": 1238
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.204044117647059e-05,
+      "loss": 1.2239,
+      "step": 1239
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.200980392156863e-05,
+      "loss": 1.3046,
+      "step": 1240
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.1979166666666667e-05,
+      "loss": 1.3456,
+      "step": 1241
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.1948529411764707e-05,
+      "loss": 1.4442,
+      "step": 1242
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.1917892156862746e-05,
+      "loss": 1.4228,
+      "step": 1243
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.1887254901960784e-05,
+      "loss": 1.5372,
+      "step": 1244
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.1856617647058823e-05,
+      "loss": 1.3335,
+      "step": 1245
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.1825980392156863e-05,
+      "loss": 1.4749,
+      "step": 1246
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.1795343137254902e-05,
+      "loss": 1.6256,
+      "step": 1247
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.1764705882352942e-05,
+      "loss": 1.385,
+      "step": 1248
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.1734068627450981e-05,
+      "loss": 1.428,
+      "step": 1249
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.170343137254902e-05,
+      "loss": 1.4012,
+      "step": 1250
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.167279411764706e-05,
+      "loss": 1.5967,
+      "step": 1251
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.1642156862745098e-05,
+      "loss": 1.2575,
+      "step": 1252
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.1611519607843138e-05,
+      "loss": 1.4903,
+      "step": 1253
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.1580882352941177e-05,
+      "loss": 1.4447,
+      "step": 1254
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.1550245098039217e-05,
+      "loss": 1.455,
+      "step": 1255
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.1519607843137254e-05,
+      "loss": 1.5362,
+      "step": 1256
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.1488970588235294e-05,
+      "loss": 1.3095,
+      "step": 1257
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.1458333333333333e-05,
+      "loss": 1.4242,
+      "step": 1258
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.1427696078431373e-05,
+      "loss": 1.525,
+      "step": 1259
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.1397058823529412e-05,
+      "loss": 1.3701,
+      "step": 1260
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.1366421568627452e-05,
+      "loss": 1.355,
+      "step": 1261
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.1335784313725491e-05,
+      "loss": 1.4957,
+      "step": 1262
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.130514705882353e-05,
+      "loss": 1.3589,
+      "step": 1263
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.1274509803921569e-05,
+      "loss": 1.4259,
+      "step": 1264
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.1243872549019608e-05,
+      "loss": 1.4381,
+      "step": 1265
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.1213235294117648e-05,
+      "loss": 1.5621,
+      "step": 1266
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.1182598039215687e-05,
+      "loss": 1.3248,
+      "step": 1267
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.1151960784313727e-05,
+      "loss": 1.4263,
+      "step": 1268
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.1121323529411765e-05,
+      "loss": 1.3911,
+      "step": 1269
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.1090686274509804e-05,
+      "loss": 1.5409,
+      "step": 1270
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.1060049019607844e-05,
+      "loss": 1.4962,
+      "step": 1271
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.1029411764705883e-05,
+      "loss": 1.2959,
+      "step": 1272
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0998774509803923e-05,
+      "loss": 1.3762,
+      "step": 1273
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.096813725490196e-05,
+      "loss": 1.5678,
+      "step": 1274
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.09375e-05,
+      "loss": 1.3904,
+      "step": 1275
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.090686274509804e-05,
+      "loss": 1.3816,
+      "step": 1276
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0876225490196079e-05,
+      "loss": 1.2756,
+      "step": 1277
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0845588235294118e-05,
+      "loss": 1.3998,
+      "step": 1278
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.0814950980392158e-05,
+      "loss": 1.4528,
+      "step": 1279
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0784313725490197e-05,
+      "loss": 1.3792,
+      "step": 1280
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0753676470588237e-05,
+      "loss": 1.2258,
+      "step": 1281
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.0723039215686275e-05,
+      "loss": 1.449,
+      "step": 1282
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0692401960784314e-05,
+      "loss": 1.4884,
+      "step": 1283
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0661764705882354e-05,
+      "loss": 1.3689,
+      "step": 1284
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.0631127450980393e-05,
+      "loss": 1.5394,
+      "step": 1285
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0600490196078431e-05,
+      "loss": 1.347,
+      "step": 1286
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.056985294117647e-05,
+      "loss": 1.309,
+      "step": 1287
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.053921568627451e-05,
+      "loss": 1.3844,
+      "step": 1288
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.050857843137255e-05,
+      "loss": 1.4357,
+      "step": 1289
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0477941176470589e-05,
+      "loss": 1.3567,
+      "step": 1290
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0447303921568629e-05,
+      "loss": 1.6525,
+      "step": 1291
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0416666666666668e-05,
+      "loss": 1.4266,
+      "step": 1292
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0386029411764708e-05,
+      "loss": 1.5506,
+      "step": 1293
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0355392156862747e-05,
+      "loss": 1.4294,
+      "step": 1294
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0324754901960785e-05,
+      "loss": 1.4653,
+      "step": 1295
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.0294117647058824e-05,
+      "loss": 1.4107,
+      "step": 1296
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0263480392156862e-05,
+      "loss": 1.5791,
+      "step": 1297
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0232843137254902e-05,
+      "loss": 1.5723,
+      "step": 1298
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0202205882352941e-05,
+      "loss": 1.4471,
+      "step": 1299
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.017156862745098e-05,
+      "loss": 1.4392,
+      "step": 1300
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.014093137254902e-05,
+      "loss": 1.5031,
+      "step": 1301
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.011029411764706e-05,
+      "loss": 1.5184,
+      "step": 1302
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0079656862745099e-05,
+      "loss": 1.4192,
+      "step": 1303
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.0049019607843139e-05,
+      "loss": 1.5441,
+      "step": 1304
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.0018382352941178e-05,
+      "loss": 1.3691,
+      "step": 1305
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 9.987745098039216e-06,
+      "loss": 1.2795,
+      "step": 1306
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 9.957107843137255e-06,
+      "loss": 1.4674,
+      "step": 1307
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 9.926470588235293e-06,
+      "loss": 1.3477,
+      "step": 1308
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 9.895833333333333e-06,
+      "loss": 1.3581,
+      "step": 1309
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.865196078431372e-06,
+      "loss": 1.4442,
+      "step": 1310
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.834558823529412e-06,
+      "loss": 1.3559,
+      "step": 1311
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.803921568627451e-06,
+      "loss": 1.5118,
+      "step": 1312
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 9.77328431372549e-06,
+      "loss": 1.4683,
+      "step": 1313
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 9.74264705882353e-06,
+      "loss": 1.5387,
+      "step": 1314
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 9.71200980392157e-06,
+      "loss": 1.4573,
+      "step": 1315
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 9.68137254901961e-06,
+      "loss": 1.3425,
+      "step": 1316
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 9.650735294117649e-06,
+      "loss": 1.3819,
+      "step": 1317
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 9.620098039215687e-06,
+      "loss": 1.3142,
+      "step": 1318
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 9.589460784313726e-06,
+      "loss": 1.3131,
+      "step": 1319
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 9.558823529411764e-06,
+      "loss": 1.4972,
+      "step": 1320
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 9.528186274509803e-06,
+      "loss": 1.2616,
+      "step": 1321
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 9.497549019607843e-06,
+      "loss": 1.4856,
+      "step": 1322
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 9.466911764705882e-06,
+      "loss": 1.4967,
+      "step": 1323
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.436274509803922e-06,
+      "loss": 1.5492,
+      "step": 1324
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.405637254901961e-06,
+      "loss": 1.442,
+      "step": 1325
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 9.375000000000001e-06,
+      "loss": 1.3386,
+      "step": 1326
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 9.34436274509804e-06,
+      "loss": 1.4034,
+      "step": 1327
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 9.31372549019608e-06,
+      "loss": 1.4639,
+      "step": 1328
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 9.283088235294118e-06,
+      "loss": 1.2968,
+      "step": 1329
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 9.252450980392157e-06,
+      "loss": 1.4808,
+      "step": 1330
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 9.221813725490197e-06,
+      "loss": 1.3063,
+      "step": 1331
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 9.191176470588236e-06,
+      "loss": 1.386,
+      "step": 1332
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 9.160539215686274e-06,
+      "loss": 1.3305,
+      "step": 1333
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 9.129901960784313e-06,
+      "loss": 1.3694,
+      "step": 1334
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 9.099264705882353e-06,
+      "loss": 1.4481,
+      "step": 1335
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 9.068627450980392e-06,
+      "loss": 1.4869,
+      "step": 1336
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 9.037990196078432e-06,
+      "loss": 1.4148,
+      "step": 1337
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 9.007352941176471e-06,
+      "loss": 1.407,
+      "step": 1338
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 8.976715686274511e-06,
+      "loss": 1.6586,
+      "step": 1339
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 8.946078431372549e-06,
+      "loss": 1.4268,
+      "step": 1340
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 8.915441176470588e-06,
+      "loss": 1.5512,
+      "step": 1341
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 8.884803921568628e-06,
+      "loss": 1.5415,
+      "step": 1342
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.854166666666667e-06,
+      "loss": 1.4795,
+      "step": 1343
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.823529411764707e-06,
+      "loss": 1.3405,
+      "step": 1344
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.792892156862746e-06,
+      "loss": 1.3937,
+      "step": 1345
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.762254901960784e-06,
+      "loss": 1.4259,
+      "step": 1346
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.731617647058824e-06,
+      "loss": 1.3503,
+      "step": 1347
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 8.700980392156863e-06,
+      "loss": 1.531,
+      "step": 1348
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 8.670343137254903e-06,
+      "loss": 1.3949,
+      "step": 1349
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 8.639705882352942e-06,
+      "loss": 1.3392,
+      "step": 1350
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 8.60906862745098e-06,
+      "loss": 1.3883,
+      "step": 1351
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 8.57843137254902e-06,
+      "loss": 1.4137,
+      "step": 1352
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 8.547794117647059e-06,
+      "loss": 1.6342,
+      "step": 1353
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.517156862745098e-06,
+      "loss": 1.6366,
+      "step": 1354
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.486519607843138e-06,
+      "loss": 1.4365,
+      "step": 1355
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 8.455882352941177e-06,
+      "loss": 1.3931,
+      "step": 1356
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 8.425245098039217e-06,
+      "loss": 1.4836,
+      "step": 1357
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 8.394607843137255e-06,
+      "loss": 1.4096,
+      "step": 1358
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 8.363970588235294e-06,
+      "loss": 1.3158,
+      "step": 1359
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 1.389,
+      "step": 1360
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 8.302696078431373e-06,
+      "loss": 1.3151,
+      "step": 1361
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 8.272058823529413e-06,
+      "loss": 1.4303,
+      "step": 1362
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 8.24142156862745e-06,
+      "loss": 1.2624,
+      "step": 1363
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 8.21078431372549e-06,
+      "loss": 1.1676,
+      "step": 1364
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 8.18014705882353e-06,
+      "loss": 1.386,
+      "step": 1365
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 8.149509803921569e-06,
+      "loss": 1.2936,
+      "step": 1366
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 8.118872549019609e-06,
+      "loss": 1.4249,
+      "step": 1367
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 8.088235294117648e-06,
+      "loss": 1.4556,
+      "step": 1368
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 8.057598039215688e-06,
+      "loss": 1.376,
+      "step": 1369
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 8.026960784313727e-06,
+      "loss": 1.3012,
+      "step": 1370
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 7.996323529411765e-06,
+      "loss": 1.3203,
+      "step": 1371
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 7.965686274509804e-06,
+      "loss": 1.3053,
+      "step": 1372
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.935049019607844e-06,
+      "loss": 1.3044,
+      "step": 1373
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 7.904411764705882e-06,
+      "loss": 1.2894,
+      "step": 1374
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 7.873774509803921e-06,
+      "loss": 1.2726,
+      "step": 1375
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 7.84313725490196e-06,
+      "loss": 1.2429,
+      "step": 1376
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 7.8125e-06,
+      "loss": 1.3949,
+      "step": 1377
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 7.78186274509804e-06,
+      "loss": 1.3328,
+      "step": 1378
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 7.75122549019608e-06,
+      "loss": 1.297,
+      "step": 1379
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 7.720588235294119e-06,
+      "loss": 1.2876,
+      "step": 1380
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.689950980392158e-06,
+      "loss": 1.2755,
+      "step": 1381
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.659313725490198e-06,
+      "loss": 1.4501,
+      "step": 1382
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 7.628676470588236e-06,
+      "loss": 1.3352,
+      "step": 1383
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 7.598039215686274e-06,
+      "loss": 1.5587,
+      "step": 1384
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 7.567401960784314e-06,
+      "loss": 1.3972,
+      "step": 1385
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.536764705882353e-06,
+      "loss": 1.3933,
+      "step": 1386
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.506127450980393e-06,
+      "loss": 1.3135,
+      "step": 1387
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 7.475490196078431e-06,
+      "loss": 1.4927,
+      "step": 1388
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 7.444852941176471e-06,
+      "loss": 1.3657,
+      "step": 1389
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 7.41421568627451e-06,
+      "loss": 1.3142,
+      "step": 1390
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 7.38357843137255e-06,
+      "loss": 1.3301,
+      "step": 1391
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 7.3529411764705884e-06,
+      "loss": 1.3348,
+      "step": 1392
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 7.322303921568628e-06,
+      "loss": 1.3596,
+      "step": 1393
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 7.2916666666666674e-06,
+      "loss": 1.3528,
+      "step": 1394
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 7.261029411764707e-06,
+      "loss": 1.3327,
+      "step": 1395
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 7.2303921568627464e-06,
+      "loss": 1.3356,
+      "step": 1396
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 7.199754901960784e-06,
+      "loss": 1.2757,
+      "step": 1397
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 7.169117647058824e-06,
+      "loss": 1.314,
+      "step": 1398
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 7.138480392156862e-06,
+      "loss": 1.3611,
+      "step": 1399
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.107843137254902e-06,
+      "loss": 1.3965,
+      "step": 1400
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.077205882352941e-06,
+      "loss": 1.3052,
+      "step": 1401
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 7.046568627450981e-06,
+      "loss": 1.6027,
+      "step": 1402
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 7.01593137254902e-06,
+      "loss": 1.3454,
+      "step": 1403
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 6.985294117647059e-06,
+      "loss": 1.3814,
+      "step": 1404
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.9546568627450986e-06,
+      "loss": 1.3829,
+      "step": 1405
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.924019607843138e-06,
+      "loss": 1.2014,
+      "step": 1406
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 6.8933823529411776e-06,
+      "loss": 1.3531,
+      "step": 1407
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.862745098039216e-06,
+      "loss": 1.3345,
+      "step": 1408
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.832107843137255e-06,
+      "loss": 1.3291,
+      "step": 1409
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 6.8014705882352935e-06,
+      "loss": 1.3112,
+      "step": 1410
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.770833333333333e-06,
+      "loss": 1.3974,
+      "step": 1411
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.7401960784313725e-06,
+      "loss": 1.3526,
+      "step": 1412
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 6.709558823529412e-06,
+      "loss": 1.3284,
+      "step": 1413
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.6789215686274515e-06,
+      "loss": 1.3124,
+      "step": 1414
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 6.64828431372549e-06,
+      "loss": 1.3025,
+      "step": 1415
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.61764705882353e-06,
+      "loss": 1.3556,
+      "step": 1416
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.587009803921569e-06,
+      "loss": 1.2002,
+      "step": 1417
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 6.556372549019609e-06,
+      "loss": 1.2804,
+      "step": 1418
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.525735294117648e-06,
+      "loss": 1.5184,
+      "step": 1419
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.495098039215687e-06,
+      "loss": 1.4451,
+      "step": 1420
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 6.464460784313726e-06,
+      "loss": 1.3879,
+      "step": 1421
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.433823529411764e-06,
+      "loss": 1.3042,
+      "step": 1422
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 6.403186274509804e-06,
+      "loss": 1.3802,
+      "step": 1423
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.372549019607843e-06,
+      "loss": 1.3642,
+      "step": 1424
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.341911764705883e-06,
+      "loss": 1.235,
+      "step": 1425
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 6.311274509803921e-06,
+      "loss": 1.3146,
+      "step": 1426
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.280637254901961e-06,
+      "loss": 1.3801,
+      "step": 1427
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.25e-06,
+      "loss": 1.378,
+      "step": 1428
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 6.21936274509804e-06,
+      "loss": 1.2427,
+      "step": 1429
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.188725490196079e-06,
+      "loss": 1.3531,
+      "step": 1430
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.158088235294118e-06,
+      "loss": 1.4706,
+      "step": 1431
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 6.127450980392157e-06,
+      "loss": 1.3509,
+      "step": 1432
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.096813725490196e-06,
+      "loss": 1.3532,
+      "step": 1433
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 6.066176470588236e-06,
+      "loss": 1.4325,
+      "step": 1434
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.035539215686275e-06,
+      "loss": 1.1805,
+      "step": 1435
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 6.004901960784315e-06,
+      "loss": 1.4125,
+      "step": 1436
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 5.974264705882353e-06,
+      "loss": 1.3231,
+      "step": 1437
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 5.943627450980392e-06,
+      "loss": 1.5148,
+      "step": 1438
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 5.912990196078431e-06,
+      "loss": 1.2919,
+      "step": 1439
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 5.882352941176471e-06,
+      "loss": 1.3798,
+      "step": 1440
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 5.85171568627451e-06,
+      "loss": 1.2696,
+      "step": 1441
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 5.821078431372549e-06,
+      "loss": 1.3772,
+      "step": 1442
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 5.7904411764705886e-06,
+      "loss": 1.2347,
+      "step": 1443
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 5.759803921568627e-06,
+      "loss": 1.3938,
+      "step": 1444
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 5.729166666666667e-06,
+      "loss": 1.1959,
+      "step": 1445
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 5.698529411764706e-06,
+      "loss": 1.4088,
+      "step": 1446
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 5.667892156862746e-06,
+      "loss": 1.3357,
+      "step": 1447
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 5.637254901960784e-06,
+      "loss": 1.2897,
+      "step": 1448
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 5.606617647058824e-06,
+      "loss": 1.4405,
+      "step": 1449
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 5.575980392156863e-06,
+      "loss": 1.4486,
+      "step": 1450
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 5.545343137254902e-06,
+      "loss": 1.3635,
+      "step": 1451
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 5.5147058823529415e-06,
+      "loss": 1.4221,
+      "step": 1452
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 5.48406862745098e-06,
+      "loss": 1.2789,
+      "step": 1453
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 5.45343137254902e-06,
+      "loss": 1.4844,
+      "step": 1454
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 5.422794117647059e-06,
+      "loss": 1.4011,
+      "step": 1455
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 5.392156862745099e-06,
+      "loss": 1.3918,
+      "step": 1456
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.361519607843137e-06,
+      "loss": 1.4862,
+      "step": 1457
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.330882352941177e-06,
+      "loss": 1.4181,
+      "step": 1458
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.3002450980392155e-06,
+      "loss": 1.3656,
+      "step": 1459
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.269607843137255e-06,
+      "loss": 1.3658,
+      "step": 1460
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.2389705882352945e-06,
+      "loss": 1.4512,
+      "step": 1461
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.208333333333334e-06,
+      "loss": 1.4083,
+      "step": 1462
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.1776960784313735e-06,
+      "loss": 1.3507,
+      "step": 1463
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.147058823529412e-06,
+      "loss": 1.4279,
+      "step": 1464
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.116421568627451e-06,
+      "loss": 1.3177,
+      "step": 1465
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.08578431372549e-06,
+      "loss": 1.3388,
+      "step": 1466
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.05514705882353e-06,
+      "loss": 1.3707,
+      "step": 1467
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.024509803921569e-06,
+      "loss": 1.3089,
+      "step": 1468
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.993872549019608e-06,
+      "loss": 1.3342,
+      "step": 1469
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 4.963235294117647e-06,
+      "loss": 1.2706,
+      "step": 1470
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.932598039215686e-06,
+      "loss": 1.4257,
+      "step": 1471
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.901960784313726e-06,
+      "loss": 1.4162,
+      "step": 1472
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.871323529411765e-06,
+      "loss": 1.3846,
+      "step": 1473
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.840686274509805e-06,
+      "loss": 1.5194,
+      "step": 1474
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.810049019607843e-06,
+      "loss": 1.3468,
+      "step": 1475
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.779411764705882e-06,
+      "loss": 1.3427,
+      "step": 1476
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.7487745098039214e-06,
+      "loss": 1.403,
+      "step": 1477
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.718137254901961e-06,
+      "loss": 1.5622,
+      "step": 1478
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.6875000000000004e-06,
+      "loss": 1.35,
+      "step": 1479
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.65686274509804e-06,
+      "loss": 1.2039,
+      "step": 1480
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.626225490196079e-06,
+      "loss": 1.2873,
+      "step": 1481
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.595588235294118e-06,
+      "loss": 1.3726,
+      "step": 1482
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.564950980392157e-06,
+      "loss": 1.3988,
+      "step": 1483
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.534313725490196e-06,
+      "loss": 1.4315,
+      "step": 1484
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.503676470588236e-06,
+      "loss": 1.2885,
+      "step": 1485
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.473039215686274e-06,
+      "loss": 1.228,
+      "step": 1486
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.442401960784314e-06,
+      "loss": 1.404,
+      "step": 1487
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.411764705882353e-06,
+      "loss": 1.3827,
+      "step": 1488
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.381127450980392e-06,
+      "loss": 1.3999,
+      "step": 1489
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.3504901960784316e-06,
+      "loss": 1.4251,
+      "step": 1490
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.319852941176471e-06,
+      "loss": 1.2958,
+      "step": 1491
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.28921568627451e-06,
+      "loss": 1.3771,
+      "step": 1492
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.258578431372549e-06,
+      "loss": 1.3828,
+      "step": 1493
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 4.227941176470589e-06,
+      "loss": 1.4641,
+      "step": 1494
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.197303921568627e-06,
+      "loss": 1.3418,
+      "step": 1495
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 1.3596,
+      "step": 1496
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.136029411764706e-06,
+      "loss": 1.4111,
+      "step": 1497
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.105392156862745e-06,
+      "loss": 1.3755,
+      "step": 1498
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.0747549019607845e-06,
+      "loss": 1.2282,
+      "step": 1499
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 4.044117647058824e-06,
+      "loss": 1.3756,
+      "step": 1500
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.0134803921568635e-06,
+      "loss": 1.3529,
+      "step": 1501
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.982843137254902e-06,
+      "loss": 1.3308,
+      "step": 1502
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.952205882352941e-06,
+      "loss": 1.3572,
+      "step": 1503
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.92156862745098e-06,
+      "loss": 1.4223,
+      "step": 1504
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.89093137254902e-06,
+      "loss": 1.1864,
+      "step": 1505
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.860294117647059e-06,
+      "loss": 1.2461,
+      "step": 1506
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.829656862745099e-06,
+      "loss": 1.3607,
+      "step": 1507
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.799019607843137e-06,
+      "loss": 1.3134,
+      "step": 1508
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.7683823529411766e-06,
+      "loss": 1.3372,
+      "step": 1509
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.7377450980392156e-06,
+      "loss": 1.3405,
+      "step": 1510
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.707107843137255e-06,
+      "loss": 1.3257,
+      "step": 1511
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.6764705882352942e-06,
+      "loss": 1.4348,
+      "step": 1512
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.6458333333333337e-06,
+      "loss": 1.3397,
+      "step": 1513
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.6151960784313732e-06,
+      "loss": 1.3834,
+      "step": 1514
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.584558823529412e-06,
+      "loss": 1.3306,
+      "step": 1515
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.553921568627451e-06,
+      "loss": 1.2967,
+      "step": 1516
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.5232843137254905e-06,
+      "loss": 1.4311,
+      "step": 1517
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.4926470588235295e-06,
+      "loss": 1.1779,
+      "step": 1518
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.462009803921569e-06,
+      "loss": 1.2954,
+      "step": 1519
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.431372549019608e-06,
+      "loss": 1.4043,
+      "step": 1520
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.4007352941176468e-06,
+      "loss": 1.3324,
+      "step": 1521
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.3700980392156863e-06,
+      "loss": 1.4851,
+      "step": 1522
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.3394607843137258e-06,
+      "loss": 1.2778,
+      "step": 1523
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.308823529411765e-06,
+      "loss": 1.3242,
+      "step": 1524
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.2781862745098043e-06,
+      "loss": 1.3616,
+      "step": 1525
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.2475490196078434e-06,
+      "loss": 1.2087,
+      "step": 1526
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.216911764705882e-06,
+      "loss": 1.2976,
+      "step": 1527
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.1862745098039216e-06,
+      "loss": 1.4759,
+      "step": 1528
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.1556372549019606e-06,
+      "loss": 1.2936,
+      "step": 1529
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.125e-06,
+      "loss": 1.4411,
+      "step": 1530
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.0943627450980397e-06,
+      "loss": 1.4209,
+      "step": 1531
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.0637254901960783e-06,
+      "loss": 1.4859,
+      "step": 1532
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.033088235294118e-06,
+      "loss": 1.209,
+      "step": 1533
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.0024509803921573e-06,
+      "loss": 1.2898,
+      "step": 1534
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.971813725490196e-06,
+      "loss": 1.2862,
+      "step": 1535
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.9411764705882355e-06,
+      "loss": 1.3604,
+      "step": 1536
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.9105392156862745e-06,
+      "loss": 1.4272,
+      "step": 1537
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.8799019607843136e-06,
+      "loss": 1.2584,
+      "step": 1538
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.849264705882353e-06,
+      "loss": 1.4146,
+      "step": 1539
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.818627450980392e-06,
+      "loss": 1.3974,
+      "step": 1540
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.7879901960784317e-06,
+      "loss": 1.4027,
+      "step": 1541
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.7573529411764708e-06,
+      "loss": 1.3628,
+      "step": 1542
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.72671568627451e-06,
+      "loss": 1.3352,
+      "step": 1543
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.6960784313725493e-06,
+      "loss": 1.2352,
+      "step": 1544
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.6654411764705884e-06,
+      "loss": 1.3221,
+      "step": 1545
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.6348039215686275e-06,
+      "loss": 1.3001,
+      "step": 1546
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.604166666666667e-06,
+      "loss": 1.2622,
+      "step": 1547
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.573529411764706e-06,
+      "loss": 1.1456,
+      "step": 1548
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.542892156862745e-06,
+      "loss": 1.2954,
+      "step": 1549
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.5122549019607847e-06,
+      "loss": 1.3937,
+      "step": 1550
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.4816176470588233e-06,
+      "loss": 1.4733,
+      "step": 1551
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.450980392156863e-06,
+      "loss": 1.3507,
+      "step": 1552
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.4203431372549023e-06,
+      "loss": 1.127,
+      "step": 1553
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.389705882352941e-06,
+      "loss": 1.3298,
+      "step": 1554
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.3590686274509805e-06,
+      "loss": 1.3087,
+      "step": 1555
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.32843137254902e-06,
+      "loss": 1.3157,
+      "step": 1556
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.297794117647059e-06,
+      "loss": 1.2992,
+      "step": 1557
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.267156862745098e-06,
+      "loss": 1.41,
+      "step": 1558
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.236519607843137e-06,
+      "loss": 1.3936,
+      "step": 1559
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.2058823529411767e-06,
+      "loss": 1.3702,
+      "step": 1560
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.1752450980392158e-06,
+      "loss": 1.2999,
+      "step": 1561
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.144607843137255e-06,
+      "loss": 1.3051,
+      "step": 1562
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.1139705882352944e-06,
+      "loss": 1.2498,
+      "step": 1563
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.0833333333333334e-06,
+      "loss": 1.2821,
+      "step": 1564
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.0526960784313725e-06,
+      "loss": 1.3378,
+      "step": 1565
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.022058823529412e-06,
+      "loss": 1.2227,
+      "step": 1566
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.991421568627451e-06,
+      "loss": 1.4524,
+      "step": 1567
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.96078431372549e-06,
+      "loss": 1.3363,
+      "step": 1568
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.9301470588235297e-06,
+      "loss": 1.2427,
+      "step": 1569
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.8995098039215685e-06,
+      "loss": 1.3986,
+      "step": 1570
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.8688725490196078e-06,
+      "loss": 1.4094,
+      "step": 1571
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.8382352941176471e-06,
+      "loss": 1.3821,
+      "step": 1572
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.8075980392156866e-06,
+      "loss": 1.4598,
+      "step": 1573
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.7769607843137255e-06,
+      "loss": 1.2013,
+      "step": 1574
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.7463235294117648e-06,
+      "loss": 1.2467,
+      "step": 1575
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.715686274509804e-06,
+      "loss": 1.2598,
+      "step": 1576
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.6850490196078431e-06,
+      "loss": 1.4044,
+      "step": 1577
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.6544117647058824e-06,
+      "loss": 1.4162,
+      "step": 1578
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.6237745098039217e-06,
+      "loss": 1.3605,
+      "step": 1579
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.5931372549019608e-06,
+      "loss": 1.4473,
+      "step": 1580
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.5625e-06,
+      "loss": 1.1182,
+      "step": 1581
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.5318627450980392e-06,
+      "loss": 1.3216,
+      "step": 1582
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.5012254901960787e-06,
+      "loss": 1.3245,
+      "step": 1583
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.4705882352941177e-06,
+      "loss": 1.4301,
+      "step": 1584
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.4399509803921568e-06,
+      "loss": 1.3103,
+      "step": 1585
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.409313725490196e-06,
+      "loss": 1.2685,
+      "step": 1586
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.3786764705882354e-06,
+      "loss": 1.3375,
+      "step": 1587
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.3480392156862747e-06,
+      "loss": 1.4271,
+      "step": 1588
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.3174019607843138e-06,
+      "loss": 1.3489,
+      "step": 1589
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.286764705882353e-06,
+      "loss": 1.3495,
+      "step": 1590
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.2561274509803923e-06,
+      "loss": 1.3817,
+      "step": 1591
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.2254901960784314e-06,
+      "loss": 1.3489,
+      "step": 1592
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.1948529411764705e-06,
+      "loss": 1.4411,
+      "step": 1593
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.16421568627451e-06,
+      "loss": 1.3191,
+      "step": 1594
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.133578431372549e-06,
+      "loss": 1.3374,
+      "step": 1595
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.1029411764705884e-06,
+      "loss": 1.3493,
+      "step": 1596
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.0723039215686274e-06,
+      "loss": 1.2171,
+      "step": 1597
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.0416666666666667e-06,
+      "loss": 1.2148,
+      "step": 1598
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.011029411764706e-06,
+      "loss": 1.457,
+      "step": 1599
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 9.80392156862745e-07,
+      "loss": 1.3792,
+      "step": 1600
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 9.497549019607843e-07,
+      "loss": 1.4415,
+      "step": 1601
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 9.191176470588236e-07,
+      "loss": 1.5127,
+      "step": 1602
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 8.884803921568627e-07,
+      "loss": 1.186,
+      "step": 1603
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 8.57843137254902e-07,
+      "loss": 1.2269,
+      "step": 1604
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 8.272058823529412e-07,
+      "loss": 1.2867,
+      "step": 1605
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.965686274509804e-07,
+      "loss": 1.352,
+      "step": 1606
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.659313725490196e-07,
+      "loss": 1.3858,
+      "step": 1607
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.352941176470589e-07,
+      "loss": 1.3844,
+      "step": 1608
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.04656862745098e-07,
+      "loss": 1.4637,
+      "step": 1609
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.740196078431373e-07,
+      "loss": 1.4068,
+      "step": 1610
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 6.433823529411765e-07,
+      "loss": 1.4805,
+      "step": 1611
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.127450980392157e-07,
+      "loss": 1.2372,
+      "step": 1612
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 5.82107843137255e-07,
+      "loss": 1.4152,
+      "step": 1613
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 5.514705882352942e-07,
+      "loss": 1.4519,
+      "step": 1614
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 5.208333333333334e-07,
+      "loss": 1.1806,
+      "step": 1615
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.901960784313725e-07,
+      "loss": 1.3203,
+      "step": 1616
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.595588235294118e-07,
+      "loss": 1.4597,
+      "step": 1617
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.28921568627451e-07,
+      "loss": 1.276,
+      "step": 1618
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.982843137254902e-07,
+      "loss": 1.2572,
+      "step": 1619
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.6764705882352943e-07,
+      "loss": 1.3598,
+      "step": 1620
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.3700980392156867e-07,
+      "loss": 1.3351,
+      "step": 1621
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.0637254901960785e-07,
+      "loss": 1.4287,
+      "step": 1622
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.757352941176471e-07,
+      "loss": 1.4305,
+      "step": 1623
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.4509803921568627e-07,
+      "loss": 1.3029,
+      "step": 1624
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.144607843137255e-07,
+      "loss": 1.4285,
+      "step": 1625
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.8382352941176472e-07,
+      "loss": 1.2801,
+      "step": 1626
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.5318627450980393e-07,
+      "loss": 1.3033,
+      "step": 1627
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.2254901960784314e-07,
+      "loss": 1.3917,
+      "step": 1628
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 9.191176470588236e-08,
+      "loss": 1.2055,
+      "step": 1629
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.127450980392157e-08,
+      "loss": 1.3366,
+      "step": 1630
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.0637254901960784e-08,
+      "loss": 1.2871,
+      "step": 1631
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.0,
+      "loss": 1.2639,
+      "step": 1632
+    },
+    {
+      "epoch": 6.0,
+      "step": 1632,
+      "total_flos": 6.250550402639462e+16,
+      "train_loss": 1.7286054671687239,
+      "train_runtime": 2952.8969,
+      "train_samples_per_second": 8.823,
+      "train_steps_per_second": 0.553
+    }
+  ],
+  "max_steps": 1632,
+  "num_train_epochs": 6,
+  "total_flos": 6.250550402639462e+16,
+  "trial_name": null,
+  "trial_params": null
+}