diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,40720 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 6782,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.901960784313726e-08,
+      "loss": 2.2344,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.803921568627452e-08,
+      "loss": 1.9673,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4705882352941178e-07,
+      "loss": 2.4105,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9607843137254904e-07,
+      "loss": 2.1124,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4509803921568627e-07,
+      "loss": 1.9979,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9411764705882356e-07,
+      "loss": 2.4906,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4313725490196084e-07,
+      "loss": 2.1457,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.921568627450981e-07,
+      "loss": 2.1406,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4117647058823536e-07,
+      "loss": 1.9778,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.901960784313725e-07,
+      "loss": 2.3251,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.392156862745098e-07,
+      "loss": 2.4013,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.882352941176471e-07,
+      "loss": 2.5159,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.372549019607843e-07,
+      "loss": 2.037,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.862745098039217e-07,
+      "loss": 2.0223,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.352941176470589e-07,
+      "loss": 2.5548,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.843137254901962e-07,
+      "loss": 2.0072,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 1.9432,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.823529411764707e-07,
+      "loss": 2.3919,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.313725490196079e-07,
+      "loss": 2.2813,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.80392156862745e-07,
+      "loss": 2.4744,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0294117647058825e-06,
+      "loss": 2.3581,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0784313725490197e-06,
+      "loss": 2.3634,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.127450980392157e-06,
+      "loss": 2.2455,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1764705882352942e-06,
+      "loss": 1.9889,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2254901960784314e-06,
+      "loss": 1.9922,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2745098039215686e-06,
+      "loss": 1.967,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.323529411764706e-06,
+      "loss": 1.9223,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3725490196078434e-06,
+      "loss": 1.9402,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4215686274509805e-06,
+      "loss": 2.1122,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4705882352941177e-06,
+      "loss": 2.2423,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.519607843137255e-06,
+      "loss": 2.285,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5686274509803923e-06,
+      "loss": 2.3161,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6176470588235297e-06,
+      "loss": 2.3289,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 2.2441,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.715686274509804e-06,
+      "loss": 1.7668,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7647058823529414e-06,
+      "loss": 2.133,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8137254901960786e-06,
+      "loss": 2.1466,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8627450980392158e-06,
+      "loss": 2.1498,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9117647058823528e-06,
+      "loss": 2.1133,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.96078431372549e-06,
+      "loss": 1.612,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0098039215686276e-06,
+      "loss": 2.1942,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.058823529411765e-06,
+      "loss": 2.2504,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1078431372549023e-06,
+      "loss": 1.8754,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1568627450980393e-06,
+      "loss": 1.6117,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2058823529411767e-06,
+      "loss": 2.1003,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.254901960784314e-06,
+      "loss": 1.8921,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.303921568627451e-06,
+      "loss": 2.3267,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3529411764705885e-06,
+      "loss": 1.8387,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.401960784313726e-06,
+      "loss": 1.2302,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.450980392156863e-06,
+      "loss": 2.1555,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-06,
+      "loss": 2.1925,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.549019607843137e-06,
+      "loss": 2.0862,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.598039215686275e-06,
+      "loss": 2.2132,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.647058823529412e-06,
+      "loss": 2.0449,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.696078431372549e-06,
+      "loss": 2.1558,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7450980392156867e-06,
+      "loss": 1.721,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7941176470588237e-06,
+      "loss": 1.9284,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.843137254901961e-06,
+      "loss": 1.9184,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8921568627450985e-06,
+      "loss": 2.0569,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9411764705882355e-06,
+      "loss": 1.6391,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.990196078431373e-06,
+      "loss": 1.9417,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.03921568627451e-06,
+      "loss": 2.2305,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0882352941176476e-06,
+      "loss": 1.7883,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1372549019607846e-06,
+      "loss": 1.5322,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1862745098039216e-06,
+      "loss": 1.6597,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2352941176470594e-06,
+      "loss": 1.9777,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2843137254901964e-06,
+      "loss": 1.9077,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.6632,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.382352941176471e-06,
+      "loss": 1.6095,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.431372549019608e-06,
+      "loss": 1.9071,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.480392156862745e-06,
+      "loss": 1.7772,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.529411764705883e-06,
+      "loss": 1.3924,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.57843137254902e-06,
+      "loss": 1.7126,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6274509803921573e-06,
+      "loss": 1.8624,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6764705882352946e-06,
+      "loss": 1.7238,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7254901960784316e-06,
+      "loss": 1.9014,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.774509803921569e-06,
+      "loss": 1.8978,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8235294117647055e-06,
+      "loss": 1.6738,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.872549019607843e-06,
+      "loss": 1.7192,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.92156862745098e-06,
+      "loss": 2.1092,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.970588235294118e-06,
+      "loss": 1.6503,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.019607843137255e-06,
+      "loss": 1.6697,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.068627450980392e-06,
+      "loss": 1.848,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.11764705882353e-06,
+      "loss": 1.6919,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 1.7317,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.215686274509805e-06,
+      "loss": 1.9306,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.264705882352942e-06,
+      "loss": 1.7061,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.313725490196079e-06,
+      "loss": 1.7795,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3627450980392164e-06,
+      "loss": 1.7494,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.411764705882353e-06,
+      "loss": 1.4801,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.460784313725491e-06,
+      "loss": 1.5161,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.509803921568628e-06,
+      "loss": 1.9264,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.558823529411765e-06,
+      "loss": 1.7132,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.607843137254902e-06,
+      "loss": 1.6329,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.65686274509804e-06,
+      "loss": 1.8112,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.705882352941177e-06,
+      "loss": 1.8105,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.754901960784314e-06,
+      "loss": 1.5189,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.803921568627452e-06,
+      "loss": 1.5423,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.852941176470589e-06,
+      "loss": 1.5294,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.901960784313726e-06,
+      "loss": 1.8319,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9509803921568634e-06,
+      "loss": 1.7446,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 1.7355,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.049019607843137e-06,
+      "loss": 1.6667,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.098039215686274e-06,
+      "loss": 1.3655,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.147058823529411e-06,
+      "loss": 1.6856,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.19607843137255e-06,
+      "loss": 1.7373,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.245098039215687e-06,
+      "loss": 1.9254,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.294117647058824e-06,
+      "loss": 1.6764,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.343137254901961e-06,
+      "loss": 1.4615,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.392156862745098e-06,
+      "loss": 1.6227,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.441176470588236e-06,
+      "loss": 1.8696,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4901960784313735e-06,
+      "loss": 1.772,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5392156862745104e-06,
+      "loss": 1.4496,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.588235294117647e-06,
+      "loss": 1.6038,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.637254901960784e-06,
+      "loss": 1.5164,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.686274509803922e-06,
+      "loss": 1.6828,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.735294117647059e-06,
+      "loss": 1.5905,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.784313725490197e-06,
+      "loss": 1.8818,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.833333333333334e-06,
+      "loss": 1.6723,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.882352941176471e-06,
+      "loss": 1.7534,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.931372549019609e-06,
+      "loss": 1.5901,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.980392156862746e-06,
+      "loss": 1.2185,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.029411764705883e-06,
+      "loss": 1.5631,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.07843137254902e-06,
+      "loss": 1.3533,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.1274509803921575e-06,
+      "loss": 1.7354,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.176470588235295e-06,
+      "loss": 1.8175,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.225490196078432e-06,
+      "loss": 1.4738,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.274509803921569e-06,
+      "loss": 1.7899,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.323529411764706e-06,
+      "loss": 1.4438,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.372549019607843e-06,
+      "loss": 1.5377,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.421568627450982e-06,
+      "loss": 1.4813,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.470588235294119e-06,
+      "loss": 1.4316,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.519607843137256e-06,
+      "loss": 1.8588,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.568627450980393e-06,
+      "loss": 1.4009,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.61764705882353e-06,
+      "loss": 1.6417,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.3006,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.715686274509804e-06,
+      "loss": 1.6206,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.764705882352942e-06,
+      "loss": 1.4093,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.813725490196079e-06,
+      "loss": 1.2932,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.862745098039216e-06,
+      "loss": 1.3548,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.911764705882353e-06,
+      "loss": 1.5825,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.96078431372549e-06,
+      "loss": 1.7146,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.009803921568628e-06,
+      "loss": 1.5856,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.058823529411766e-06,
+      "loss": 1.5372,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.107843137254903e-06,
+      "loss": 1.6908,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.15686274509804e-06,
+      "loss": 1.7683,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.205882352941177e-06,
+      "loss": 1.6344,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.2549019607843145e-06,
+      "loss": 1.5937,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.3039215686274515e-06,
+      "loss": 1.448,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.352941176470589e-06,
+      "loss": 1.8815,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.401960784313726e-06,
+      "loss": 1.4341,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.450980392156863e-06,
+      "loss": 2.0356,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 1.3832,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.549019607843138e-06,
+      "loss": 1.4053,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.598039215686275e-06,
+      "loss": 1.8025,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.647058823529411e-06,
+      "loss": 1.5143,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.69607843137255e-06,
+      "loss": 1.7861,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.745098039215687e-06,
+      "loss": 1.6,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.794117647058825e-06,
+      "loss": 1.5798,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.84313725490196e-06,
+      "loss": 1.5191,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.892156862745098e-06,
+      "loss": 1.3068,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.941176470588236e-06,
+      "loss": 1.7093,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.990196078431374e-06,
+      "loss": 1.5956,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.03921568627451e-06,
+      "loss": 1.7645,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.088235294117648e-06,
+      "loss": 1.4287,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.137254901960784e-06,
+      "loss": 1.6076,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.186274509803922e-06,
+      "loss": 1.8178,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.23529411764706e-06,
+      "loss": 1.6651,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.284313725490198e-06,
+      "loss": 1.7245,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 1.5255,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.382352941176472e-06,
+      "loss": 1.8861,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.43137254901961e-06,
+      "loss": 1.5971,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.480392156862745e-06,
+      "loss": 1.5408,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.529411764705883e-06,
+      "loss": 1.6193,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.57843137254902e-06,
+      "loss": 1.4013,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.627450980392157e-06,
+      "loss": 1.8985,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.676470588235295e-06,
+      "loss": 1.6237,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.725490196078433e-06,
+      "loss": 1.5535,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.774509803921569e-06,
+      "loss": 1.3085,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.823529411764707e-06,
+      "loss": 1.5867,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.872549019607843e-06,
+      "loss": 1.682,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.921568627450982e-06,
+      "loss": 1.522,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.970588235294119e-06,
+      "loss": 1.6854,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.019607843137256e-06,
+      "loss": 1.418,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.068627450980392e-06,
+      "loss": 1.5572,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.11764705882353e-06,
+      "loss": 1.5689,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 1.4212,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.215686274509804e-06,
+      "loss": 1.7375,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.264705882352942e-06,
+      "loss": 1.3738,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.31372549019608e-06,
+      "loss": 1.491,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.362745098039216e-06,
+      "loss": 1.5094,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.411764705882354e-06,
+      "loss": 1.5126,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.46078431372549e-06,
+      "loss": 1.9123,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.509803921568628e-06,
+      "loss": 1.8545,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.558823529411766e-06,
+      "loss": 2.1227,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.607843137254903e-06,
+      "loss": 1.5373,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.65686274509804e-06,
+      "loss": 1.9784,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.705882352941177e-06,
+      "loss": 1.9127,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.754901960784315e-06,
+      "loss": 1.7023,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.803921568627451e-06,
+      "loss": 1.4075,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.852941176470589e-06,
+      "loss": 1.6949,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.901960784313727e-06,
+      "loss": 1.5231,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.950980392156863e-06,
+      "loss": 1.6294,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 1.3522,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999999429767446e-06,
+      "loss": 1.5935,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999997719069912e-06,
+      "loss": 1.5636,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99999486790779e-06,
+      "loss": 1.4435,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99999087628173e-06,
+      "loss": 1.686,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999985744192642e-06,
+      "loss": 1.5424,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999979471641694e-06,
+      "loss": 1.8544,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999972058630322e-06,
+      "loss": 1.3468,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999963505160212e-06,
+      "loss": 1.299,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99995381123332e-06,
+      "loss": 1.4661,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999942976851853e-06,
+      "loss": 1.6928,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999931002018283e-06,
+      "loss": 1.4696,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999917886735342e-06,
+      "loss": 1.7348,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999903631006022e-06,
+      "loss": 1.9065,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999888234833575e-06,
+      "loss": 1.6642,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999871698221511e-06,
+      "loss": 1.8948,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9998540211736e-06,
+      "loss": 1.4923,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99983520369388e-06,
+      "loss": 1.4832,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99981524578664e-06,
+      "loss": 1.639,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99979414745643e-06,
+      "loss": 1.6015,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999771908708065e-06,
+      "loss": 1.9197,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999748529546618e-06,
+      "loss": 1.4565,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999724009977419e-06,
+      "loss": 1.5791,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999698350006063e-06,
+      "loss": 1.6575,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999671549638403e-06,
+      "loss": 1.4708,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999643608880551e-06,
+      "loss": 1.6849,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999614527738882e-06,
+      "loss": 1.6312,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999584306220023e-06,
+      "loss": 1.7248,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999552944330875e-06,
+      "loss": 1.5611,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999520442078587e-06,
+      "loss": 1.7523,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999486799470575e-06,
+      "loss": 1.2805,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99945201651451e-06,
+      "loss": 1.7862,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999416093218329e-06,
+      "loss": 1.7149,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999379029590224e-06,
+      "loss": 1.5474,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999340825638647e-06,
+      "loss": 1.9409,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999301481372314e-06,
+      "loss": 1.6789,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9992609968002e-06,
+      "loss": 1.5797,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99921937193154e-06,
+      "loss": 1.4411,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999176606775825e-06,
+      "loss": 1.511,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99913270134281e-06,
+      "loss": 1.3841,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999087655642514e-06,
+      "loss": 1.541,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999041469685206e-06,
+      "loss": 1.5094,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998994143481424e-06,
+      "loss": 1.6445,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99894567704196e-06,
+      "loss": 1.5805,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998896070377873e-06,
+      "loss": 1.8436,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998845323500475e-06,
+      "loss": 1.4944,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998793436421342e-06,
+      "loss": 1.522,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998740409152308e-06,
+      "loss": 1.5747,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99868624170547e-06,
+      "loss": 1.6863,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998630934093179e-06,
+      "loss": 1.7522,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998574486328056e-06,
+      "loss": 1.6787,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998516898422971e-06,
+      "loss": 1.7169,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998458170391065e-06,
+      "loss": 1.4978,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99839830224573e-06,
+      "loss": 1.8507,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998337294000621e-06,
+      "loss": 1.6143,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998275145669655e-06,
+      "loss": 1.5608,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998211857267008e-06,
+      "loss": 1.5493,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998147428807113e-06,
+      "loss": 1.7783,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99808186030467e-06,
+      "loss": 1.693,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99801515177463e-06,
+      "loss": 1.6735,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997947303232211e-06,
+      "loss": 1.703,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99787831469289e-06,
+      "loss": 1.8444,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997808186172402e-06,
+      "loss": 1.4977,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997736917686743e-06,
+      "loss": 1.6046,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997664509252165e-06,
+      "loss": 1.5306,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99759096088519e-06,
+      "loss": 1.6429,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99751627260259e-06,
+      "loss": 1.6025,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9974404444214e-06,
+      "loss": 1.8291,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997363476358921e-06,
+      "loss": 1.3221,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997285368432704e-06,
+      "loss": 1.6162,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997206120660565e-06,
+      "loss": 1.693,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997125733060584e-06,
+      "loss": 1.7701,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997044205651093e-06,
+      "loss": 1.365,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996961538450689e-06,
+      "loss": 1.5603,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996877731478227e-06,
+      "loss": 1.7438,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996792784752826e-06,
+      "loss": 1.5682,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996706698293858e-06,
+      "loss": 1.4417,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99661947212096e-06,
+      "loss": 1.3503,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996531106254027e-06,
+      "loss": 1.7228,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996441600713215e-06,
+      "loss": 1.9496,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996350955518943e-06,
+      "loss": 1.5788,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996259170691881e-06,
+      "loss": 1.5086,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996166246252967e-06,
+      "loss": 1.4017,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996072182223398e-06,
+      "loss": 1.3598,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995976978624628e-06,
+      "loss": 1.7454,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99588063547837e-06,
+      "loss": 1.5049,
+      "step": 289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995783152806602e-06,
+      "loss": 1.7542,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99568453063156e-06,
+      "loss": 1.6602,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995584768975735e-06,
+      "loss": 1.5919,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995483867861885e-06,
+      "loss": 1.4832,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995381827313025e-06,
+      "loss": 1.3368,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995278647352428e-06,
+      "loss": 1.4827,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99517432800363e-06,
+      "loss": 1.4452,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995068869290425e-06,
+      "loss": 1.6528,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994962271236868e-06,
+      "loss": 1.7891,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994854533867273e-06,
+      "loss": 1.4847,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994745657206214e-06,
+      "loss": 1.7543,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994635641278522e-06,
+      "loss": 1.3948,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994524486109296e-06,
+      "loss": 1.4799,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994412191723888e-06,
+      "loss": 1.725,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99429875814791e-06,
+      "loss": 1.4462,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994184185407236e-06,
+      "loss": 1.6811,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994068473528001e-06,
+      "loss": 1.8197,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993951622536596e-06,
+      "loss": 1.6624,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993833632459675e-06,
+      "loss": 1.8406,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99371450332415e-06,
+      "loss": 1.6469,
+      "step": 309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993594235157194e-06,
+      "loss": 1.3456,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99347282798624e-06,
+      "loss": 1.2507,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993350281838978e-06,
+      "loss": 1.6479,
+      "step": 312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993226596743362e-06,
+      "loss": 1.6064,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993101772727602e-06,
+      "loss": 1.7355,
+      "step": 314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992975809820172e-06,
+      "loss": 1.506,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9928487080498e-06,
+      "loss": 1.7132,
+      "step": 316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99272046744548e-06,
+      "loss": 1.48,
+      "step": 317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992591088036463e-06,
+      "loss": 1.6251,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992460569852256e-06,
+      "loss": 1.6123,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992328912922631e-06,
+      "loss": 1.4671,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99219611727762e-06,
+      "loss": 1.5457,
+      "step": 321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992062182947509e-06,
+      "loss": 1.7262,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991927109962851e-06,
+      "loss": 1.5484,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991790898354454e-06,
+      "loss": 1.4308,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991653548153387e-06,
+      "loss": 1.294,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991515059390977e-06,
+      "loss": 1.6132,
+      "step": 326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991375432098815e-06,
+      "loss": 1.6437,
+      "step": 327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991234666308747e-06,
+      "loss": 1.8556,
+      "step": 328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991092762052883e-06,
+      "loss": 1.359,
+      "step": 329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990949719363586e-06,
+      "loss": 1.4923,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990805538273486e-06,
+      "loss": 1.5964,
+      "step": 331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990660218815473e-06,
+      "loss": 1.6949,
+      "step": 332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990513761022687e-06,
+      "loss": 1.8398,
+      "step": 333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990366164928538e-06,
+      "loss": 1.6964,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99021743056669e-06,
+      "loss": 1.3916,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990067557971068e-06,
+      "loss": 1.6391,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98991654717586e-06,
+      "loss": 1.4661,
+      "step": 337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989764398215505e-06,
+      "loss": 1.656,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989611111124712e-06,
+      "loss": 1.5427,
+      "step": 339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989456685938444e-06,
+      "loss": 1.4238,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989301122691922e-06,
+      "loss": 1.7418,
+      "step": 341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98914442142063e-06,
+      "loss": 1.5051,
+      "step": 342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988986582160312e-06,
+      "loss": 1.5275,
+      "step": 343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988827604946967e-06,
+      "loss": 1.5216,
+      "step": 344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988667489816859e-06,
+      "loss": 1.4892,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98850623680651e-06,
+      "loss": 1.7219,
+      "step": 346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988343845952697e-06,
+      "loss": 1.7229,
+      "step": 347
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988180317292463e-06,
+      "loss": 1.1875,
+      "step": 348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988015650863108e-06,
+      "loss": 1.4902,
+      "step": 349
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98784984670219e-06,
+      "loss": 1.5979,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987682904847527e-06,
+      "loss": 1.6984,
+      "step": 351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987514825337198e-06,
+      "loss": 1.8098,
+      "step": 352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987345608209543e-06,
+      "loss": 1.3701,
+      "step": 353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987175253503156e-06,
+      "loss": 1.6954,
+      "step": 354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987003761256896e-06,
+      "loss": 1.4692,
+      "step": 355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986831131509878e-06,
+      "loss": 1.743,
+      "step": 356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986657364301479e-06,
+      "loss": 1.6297,
+      "step": 357
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986482459671332e-06,
+      "loss": 1.6225,
+      "step": 358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986306417659333e-06,
+      "loss": 1.7433,
+      "step": 359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986129238305635e-06,
+      "loss": 1.6151,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985950921650652e-06,
+      "loss": 1.7503,
+      "step": 361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985771467735057e-06,
+      "loss": 1.5558,
+      "step": 362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985590876599782e-06,
+      "loss": 1.4771,
+      "step": 363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985409148286018e-06,
+      "loss": 1.9488,
+      "step": 364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985226282835216e-06,
+      "loss": 2.027,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985042280289087e-06,
+      "loss": 1.5517,
+      "step": 366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984857140689602e-06,
+      "loss": 1.7426,
+      "step": 367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984670864078986e-06,
+      "loss": 1.3778,
+      "step": 368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984483450499732e-06,
+      "loss": 1.6089,
+      "step": 369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984294899994584e-06,
+      "loss": 1.5996,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984105212606552e-06,
+      "loss": 1.4915,
+      "step": 371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983914388378902e-06,
+      "loss": 1.6402,
+      "step": 372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983722427355157e-06,
+      "loss": 1.7076,
+      "step": 373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983529329579103e-06,
+      "loss": 1.5976,
+      "step": 374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983335095094785e-06,
+      "loss": 1.5969,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983139723946508e-06,
+      "loss": 1.744,
+      "step": 376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982943216178834e-06,
+      "loss": 1.7092,
+      "step": 377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982745571836581e-06,
+      "loss": 1.5824,
+      "step": 378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982546790964837e-06,
+      "loss": 1.4575,
+      "step": 379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982346873608936e-06,
+      "loss": 1.4568,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982145819814484e-06,
+      "loss": 1.6098,
+      "step": 381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981943629627334e-06,
+      "loss": 1.6244,
+      "step": 382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98174030309361e-06,
+      "loss": 1.4183,
+      "step": 383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981535840259684e-06,
+      "loss": 1.4131,
+      "step": 384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981330241172195e-06,
+      "loss": 1.6977,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98112350587804e-06,
+      "loss": 1.4158,
+      "step": 386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980915634424373e-06,
+      "loss": 1.4829,
+      "step": 387
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980706626858607e-06,
+      "loss": 1.5395,
+      "step": 388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980496483228415e-06,
+      "loss": 1.5764,
+      "step": 389
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98028520358173e-06,
+      "loss": 1.3263,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980072787966744e-06,
+      "loss": 1.2227,
+      "step": 391
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979859236431907e-06,
+      "loss": 1.479,
+      "step": 392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979644549025928e-06,
+      "loss": 1.4815,
+      "step": 393
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979428725797777e-06,
+      "loss": 1.7932,
+      "step": 394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97921176679668e-06,
+      "loss": 1.6599,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978993672072125e-06,
+      "loss": 1.2562,
+      "step": 396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97877444167386e-06,
+      "loss": 1.573,
+      "step": 397
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978554075651886e-06,
+      "loss": 1.5502,
+      "step": 398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978332574056468e-06,
+      "loss": 1.4017,
+      "step": 399
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97810993693813e-06,
+      "loss": 1.7498,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977886164347655e-06,
+      "loss": 1.3943,
+      "step": 401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977661256336081e-06,
+      "loss": 1.4099,
+      "step": 402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97743521295471e-06,
+      "loss": 1.7617,
+      "step": 403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977208034255098e-06,
+      "loss": 1.6135,
+      "step": 404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97697972028907e-06,
+      "loss": 1.4172,
+      "step": 405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976750271108694e-06,
+      "loss": 1.5756,
+      "step": 406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976519686766313e-06,
+      "loss": 1.6976,
+      "step": 407
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976287967314517e-06,
+      "loss": 1.4763,
+      "step": 408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976055112806162e-06,
+      "loss": 1.6287,
+      "step": 409
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975821123294357e-06,
+      "loss": 1.2835,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97558599883248e-06,
+      "loss": 1.66,
+      "step": 411
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975349739474156e-06,
+      "loss": 1.7295,
+      "step": 412
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975112345273274e-06,
+      "loss": 1.5575,
+      "step": 413
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974873816283984e-06,
+      "loss": 1.4137,
+      "step": 414
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974634152560692e-06,
+      "loss": 1.6903,
+      "step": 415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974393354158063e-06,
+      "loss": 1.4685,
+      "step": 416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974151421131023e-06,
+      "loss": 1.6246,
+      "step": 417
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973908353534755e-06,
+      "loss": 1.5024,
+      "step": 418
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973664151424699e-06,
+      "loss": 2.0028,
+      "step": 419
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973418814856559e-06,
+      "loss": 1.5241,
+      "step": 420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973172343886291e-06,
+      "loss": 1.5507,
+      "step": 421
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972924738570115e-06,
+      "loss": 1.5927,
+      "step": 422
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97267599896451e-06,
+      "loss": 1.6636,
+      "step": 423
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972426125126208e-06,
+      "loss": 1.5213,
+      "step": 424
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972175117112208e-06,
+      "loss": 1.9006,
+      "step": 425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971922974979758e-06,
+      "loss": 1.3288,
+      "step": 426
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971669698786374e-06,
+      "loss": 1.4563,
+      "step": 427
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971415288589826e-06,
+      "loss": 1.7362,
+      "step": 428
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97115974444814e-06,
+      "loss": 1.43,
+      "step": 429
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970903066419605e-06,
+      "loss": 1.5692,
+      "step": 430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970645254562771e-06,
+      "loss": 1.5982,
+      "step": 431
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97038630893644e-06,
+      "loss": 1.397,
+      "step": 432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970126229599675e-06,
+      "loss": 1.5441,
+      "step": 433
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9698650166118e-06,
+      "loss": 1.6536,
+      "step": 434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969602670032396e-06,
+      "loss": 1.4951,
+      "step": 435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969339189921302e-06,
+      "loss": 1.8622,
+      "step": 436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969074576338615e-06,
+      "loss": 1.4475,
+      "step": 437
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968808829344692e-06,
+      "loss": 1.741,
+      "step": 438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96854194900015e-06,
+      "loss": 1.4943,
+      "step": 439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96827393536586e-06,
+      "loss": 1.4112,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.968004788502955e-06,
+      "loss": 1.4214,
+      "step": 441
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967734508472824e-06,
+      "loss": 1.5675,
+      "step": 442
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967463095337119e-06,
+      "loss": 1.7205,
+      "step": 443
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967190549157746e-06,
+      "loss": 1.5216,
+      "step": 444
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96691686999687e-06,
+      "loss": 1.359,
+      "step": 445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966642057916915e-06,
+      "loss": 1.58,
+      "step": 446
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966366112980565e-06,
+      "loss": 1.4501,
+      "step": 447
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966089035250763e-06,
+      "loss": 1.1024,
+      "step": 448
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965810824790705e-06,
+      "loss": 1.5817,
+      "step": 449
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96553148166385e-06,
+      "loss": 1.5612,
+      "step": 450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965251005933915e-06,
+      "loss": 1.3791,
+      "step": 451
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964969397664874e-06,
+      "loss": 1.7318,
+      "step": 452
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96468665692096e-06,
+      "loss": 1.5082,
+      "step": 453
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964402783766665e-06,
+      "loss": 1.6826,
+      "step": 454
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964117778266735e-06,
+      "loss": 1.5496,
+      "step": 455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963831640486183e-06,
+      "loss": 1.3804,
+      "step": 456
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96354437049027e-06,
+      "loss": 1.7518,
+      "step": 457
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963255968344525e-06,
+      "loss": 1.478,
+      "step": 458
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962966434114727e-06,
+      "loss": 1.721,
+      "step": 459
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962675767866917e-06,
+      "loss": 1.6925,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962383969667397e-06,
+      "loss": 1.4853,
+      "step": 461
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96209103958272e-06,
+      "loss": 1.4629,
+      "step": 462
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961796977679705e-06,
+      "loss": 1.5578,
+      "step": 463
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961501784025423e-06,
+      "loss": 1.7494,
+      "step": 464
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961205458687203e-06,
+      "loss": 1.121,
+      "step": 465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960908001732643e-06,
+      "loss": 1.4179,
+      "step": 466
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960609413229584e-06,
+      "loss": 1.6575,
+      "step": 467
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960309693246135e-06,
+      "loss": 1.532,
+      "step": 468
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960008841850659e-06,
+      "loss": 1.3026,
+      "step": 469
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959706859111775e-06,
+      "loss": 2.0288,
+      "step": 470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95940374509837e-06,
+      "loss": 1.8289,
+      "step": 471
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959099499879577e-06,
+      "loss": 1.6623,
+      "step": 472
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958794123524794e-06,
+      "loss": 1.8074,
+      "step": 473
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958487616103674e-06,
+      "loss": 1.3478,
+      "step": 474
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95817997768613e-06,
+      "loss": 1.3631,
+      "step": 475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957871208342331e-06,
+      "loss": 1.9722,
+      "step": 476
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95756130814271e-06,
+      "loss": 1.5096,
+      "step": 477
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957250277157946e-06,
+      "loss": 1.5719,
+      "step": 478
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956938115458987e-06,
+      "loss": 1.487,
+      "step": 479
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956624823117036e-06,
+      "loss": 1.5689,
+      "step": 480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95631040020355e-06,
+      "loss": 1.9901,
+      "step": 481
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955994846790248e-06,
+      "loss": 1.3598,
+      "step": 482
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955678162949105e-06,
+      "loss": 1.4813,
+      "step": 483
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955360348752353e-06,
+      "loss": 1.8786,
+      "step": 484
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955041404272489e-06,
+      "loss": 1.4658,
+      "step": 485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954721329582255e-06,
+      "loss": 1.5585,
+      "step": 486
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954400124754662e-06,
+      "loss": 1.574,
+      "step": 487
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954077789862972e-06,
+      "loss": 1.4747,
+      "step": 488
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953754324980711e-06,
+      "loss": 1.66,
+      "step": 489
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953429730181653e-06,
+      "loss": 1.7497,
+      "step": 490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953104005539843e-06,
+      "loss": 1.5897,
+      "step": 491
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952777151129571e-06,
+      "loss": 1.4744,
+      "step": 492
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952449167025394e-06,
+      "loss": 1.5082,
+      "step": 493
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95212005330212e-06,
+      "loss": 1.7607,
+      "step": 494
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95178981003482e-06,
+      "loss": 1.5176,
+      "step": 495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951458437298817e-06,
+      "loss": 1.5696,
+      "step": 496
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951125935169697e-06,
+      "loss": 1.584,
+      "step": 497
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950792303723302e-06,
+      "loss": 1.4016,
+      "step": 498
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95045754303573e-06,
+      "loss": 1.3496,
+      "step": 499
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950121653183336e-06,
+      "loss": 1.9397,
+      "step": 500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949784634242737e-06,
+      "loss": 1.189,
+      "step": 501
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949446486290803e-06,
+      "loss": 1.5316,
+      "step": 502
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949107209404664e-06,
+      "loss": 1.3954,
+      "step": 503
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948766803661707e-06,
+      "loss": 1.4541,
+      "step": 504
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948425269139574e-06,
+      "loss": 1.3311,
+      "step": 505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.94808260591617e-06,
+      "loss": 1.5581,
+      "step": 506
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947738814069652e-06,
+      "loss": 1.7266,
+      "step": 507
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947393893678437e-06,
+      "loss": 1.4577,
+      "step": 508
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.947047844821195e-06,
+      "loss": 1.5369,
+      "step": 509
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946700667576865e-06,
+      "loss": 1.6643,
+      "step": 510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946352362024632e-06,
+      "loss": 1.8495,
+      "step": 511
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94600292824394e-06,
+      "loss": 1.128,
+      "step": 512
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945652366314495e-06,
+      "loss": 1.5744,
+      "step": 513
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945300676316256e-06,
+      "loss": 1.5897,
+      "step": 514
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944947858329443e-06,
+      "loss": 1.8057,
+      "step": 515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94459391243453e-06,
+      "loss": 1.4097,
+      "step": 516
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94423883871225e-06,
+      "loss": 1.7258,
+      "step": 517
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943882637243592e-06,
+      "loss": 1.5302,
+      "step": 518
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943525308109806e-06,
+      "loss": 1.3099,
+      "step": 519
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943166851392392e-06,
+      "loss": 1.4125,
+      "step": 520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942807267173114e-06,
+      "loss": 1.4499,
+      "step": 521
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94244655553399e-06,
+      "loss": 1.4371,
+      "step": 522
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942084716557297e-06,
+      "loss": 1.3972,
+      "step": 523
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941721750325565e-06,
+      "loss": 1.6085,
+      "step": 524
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941357656921586e-06,
+      "loss": 1.9309,
+      "step": 525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94099243642841e-06,
+      "loss": 1.3,
+      "step": 526
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940626088929337e-06,
+      "loss": 1.6394,
+      "step": 527
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94025861450793e-06,
+      "loss": 1.5656,
+      "step": 528
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939890013248006e-06,
+      "loss": 1.512,
+      "step": 529
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939520285233642e-06,
+      "loss": 1.9122,
+      "step": 530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939149430549171e-06,
+      "loss": 1.5352,
+      "step": 531
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93877744927918e-06,
+      "loss": 1.3341,
+      "step": 532
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938404341508516e-06,
+      "loss": 1.5371,
+      "step": 533
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938030107322284e-06,
+      "loss": 1.546,
+      "step": 534
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937654746805843e-06,
+      "loss": 1.4844,
+      "step": 535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937278260044809e-06,
+      "loss": 1.6383,
+      "step": 536
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936900647125058e-06,
+      "loss": 1.4757,
+      "step": 537
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936521908132722e-06,
+      "loss": 1.3869,
+      "step": 538
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936142043154183e-06,
+      "loss": 1.7344,
+      "step": 539
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935761052276092e-06,
+      "loss": 1.4607,
+      "step": 540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935378935585347e-06,
+      "loss": 1.3326,
+      "step": 541
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934995693169104e-06,
+      "loss": 1.826,
+      "step": 542
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934611325114784e-06,
+      "loss": 1.4168,
+      "step": 543
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934225831510056e-06,
+      "loss": 1.3375,
+      "step": 544
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933839212442846e-06,
+      "loss": 1.4687,
+      "step": 545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933451468001341e-06,
+      "loss": 1.2541,
+      "step": 546
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933062598273983e-06,
+      "loss": 1.3308,
+      "step": 547
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932672603349472e-06,
+      "loss": 1.9399,
+      "step": 548
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932281483316759e-06,
+      "loss": 1.57,
+      "step": 549
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931889238265058e-06,
+      "loss": 1.6337,
+      "step": 550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931495868283838e-06,
+      "loss": 1.5195,
+      "step": 551
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931101373462823e-06,
+      "loss": 1.5528,
+      "step": 552
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930705753891996e-06,
+      "loss": 1.5108,
+      "step": 553
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930309009661593e-06,
+      "loss": 1.6981,
+      "step": 554
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929911140862109e-06,
+      "loss": 1.4705,
+      "step": 555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929512147584297e-06,
+      "loss": 1.5119,
+      "step": 556
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929112029919163e-06,
+      "loss": 1.4626,
+      "step": 557
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928710787957972e-06,
+      "loss": 1.5363,
+      "step": 558
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928308421792243e-06,
+      "loss": 1.5455,
+      "step": 559
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927904931513751e-06,
+      "loss": 1.738,
+      "step": 560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927500317214536e-06,
+      "loss": 1.6213,
+      "step": 561
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927094578986883e-06,
+      "loss": 1.6135,
+      "step": 562
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926687716923335e-06,
+      "loss": 1.6753,
+      "step": 563
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926279731116701e-06,
+      "loss": 1.7287,
+      "step": 564
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925870621660039e-06,
+      "loss": 1.6285,
+      "step": 565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925460388646659e-06,
+      "loss": 1.5298,
+      "step": 566
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925049032170135e-06,
+      "loss": 1.6238,
+      "step": 567
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924636552324296e-06,
+      "loss": 1.4037,
+      "step": 568
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924222949203224e-06,
+      "loss": 1.5605,
+      "step": 569
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923808222901258e-06,
+      "loss": 1.5812,
+      "step": 570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923392373512997e-06,
+      "loss": 1.5303,
+      "step": 571
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922975401133292e-06,
+      "loss": 1.3953,
+      "step": 572
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922557305857253e-06,
+      "loss": 1.1794,
+      "step": 573
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92213808778024e-06,
+      "loss": 1.3158,
+      "step": 574
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921717746997879e-06,
+      "loss": 1.3675,
+      "step": 575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921296283606043e-06,
+      "loss": 1.5253,
+      "step": 576
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920873697700867e-06,
+      "loss": 1.4483,
+      "step": 577
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920449989378741e-06,
+      "loss": 1.1039,
+      "step": 578
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920025158736307e-06,
+      "loss": 1.5368,
+      "step": 579
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919599205870466e-06,
+      "loss": 1.524,
+      "step": 580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919172130878378e-06,
+      "loss": 1.7529,
+      "step": 581
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918743933857453e-06,
+      "loss": 1.5887,
+      "step": 582
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918314614905362e-06,
+      "loss": 1.5652,
+      "step": 583
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917884174120026e-06,
+      "loss": 1.4737,
+      "step": 584
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91745261159963e-06,
+      "loss": 1.5799,
+      "step": 585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917019927442605e-06,
+      "loss": 1.4693,
+      "step": 586
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91658612174765e-06,
+      "loss": 1.7466,
+      "step": 587
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916151194613708e-06,
+      "loss": 1.534,
+      "step": 588
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915715146139984e-06,
+      "loss": 1.683,
+      "step": 589
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915277976425938e-06,
+      "loss": 1.6011,
+      "step": 590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914839685571285e-06,
+      "loss": 1.4653,
+      "step": 591
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914400273675997e-06,
+      "loss": 1.5284,
+      "step": 592
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.9139597408403e-06,
+      "loss": 1.756,
+      "step": 593
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913518087164678e-06,
+      "loss": 1.5993,
+      "step": 594
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913075312749867e-06,
+      "loss": 1.6556,
+      "step": 595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912631417696861e-06,
+      "loss": 1.5137,
+      "step": 596
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91218640210691e-06,
+      "loss": 1.6689,
+      "step": 597
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911740266081517e-06,
+      "loss": 1.7226,
+      "step": 598
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911293009722448e-06,
+      "loss": 1.5683,
+      "step": 599
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910844633131712e-06,
+      "loss": 1.3827,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910395136411589e-06,
+      "loss": 1.5889,
+      "step": 601
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909944519664596e-06,
+      "loss": 1.4418,
+      "step": 602
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909492782993521e-06,
+      "loss": 1.481,
+      "step": 603
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909039926501405e-06,
+      "loss": 1.1881,
+      "step": 604
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908585950291536e-06,
+      "loss": 1.56,
+      "step": 605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908130854467463e-06,
+      "loss": 1.4529,
+      "step": 606
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907674639132995e-06,
+      "loss": 1.4402,
+      "step": 607
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907217304392186e-06,
+      "loss": 1.7192,
+      "step": 608
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906758850349356e-06,
+      "loss": 0.9867,
+      "step": 609
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906299277109071e-06,
+      "loss": 1.8127,
+      "step": 610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905838584776158e-06,
+      "loss": 1.714,
+      "step": 611
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905376773455697e-06,
+      "loss": 1.5541,
+      "step": 612
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904913843253025e-06,
+      "loss": 1.5749,
+      "step": 613
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904449794273734e-06,
+      "loss": 1.4623,
+      "step": 614
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903984626623668e-06,
+      "loss": 1.4456,
+      "step": 615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903518340408932e-06,
+      "loss": 1.2588,
+      "step": 616
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903050935735876e-06,
+      "loss": 1.3206,
+      "step": 617
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90258241271112e-06,
+      "loss": 1.8127,
+      "step": 618
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902112771441524e-06,
+      "loss": 1.6776,
+      "step": 619
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901642012034214e-06,
+      "loss": 1.7384,
+      "step": 620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901170134596565e-06,
+      "loss": 1.5313,
+      "step": 621
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90069713923621e-06,
+      "loss": 1.4615,
+      "step": 622
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900223026061035e-06,
+      "loss": 1.5528,
+      "step": 623
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899747795179184e-06,
+      "loss": 1.3043,
+      "step": 624
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89927144669905e-06,
+      "loss": 1.5103,
+      "step": 625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898793980729288e-06,
+      "loss": 1.4674,
+      "step": 626
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898315397378805e-06,
+      "loss": 1.3751,
+      "step": 627
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897835696756759e-06,
+      "loss": 1.5105,
+      "step": 628
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897354878972571e-06,
+      "loss": 1.6079,
+      "step": 629
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896872944135908e-06,
+      "loss": 1.5866,
+      "step": 630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.8963898923567e-06,
+      "loss": 1.5994,
+      "step": 631
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895905723745123e-06,
+      "loss": 1.5792,
+      "step": 632
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895420438411616e-06,
+      "loss": 1.4559,
+      "step": 633
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894934036466868e-06,
+      "loss": 1.7699,
+      "step": 634
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894446518021822e-06,
+      "loss": 1.3935,
+      "step": 635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893957883187684e-06,
+      "loss": 1.4906,
+      "step": 636
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893468132075901e-06,
+      "loss": 1.3967,
+      "step": 637
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892977264798185e-06,
+      "loss": 1.1142,
+      "step": 638
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.8924852814665e-06,
+      "loss": 1.7828,
+      "step": 639
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891992182193061e-06,
+      "loss": 1.5886,
+      "step": 640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891497967090344e-06,
+      "loss": 1.629,
+      "step": 641
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891002636271075e-06,
+      "loss": 1.7151,
+      "step": 642
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890506189848233e-06,
+      "loss": 1.5922,
+      "step": 643
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890008627935057e-06,
+      "loss": 1.4044,
+      "step": 644
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889509950645038e-06,
+      "loss": 1.4456,
+      "step": 645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889010158091917e-06,
+      "loss": 1.8154,
+      "step": 646
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888509250389695e-06,
+      "loss": 1.4982,
+      "step": 647
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888007227652627e-06,
+      "loss": 1.6124,
+      "step": 648
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887504089995222e-06,
+      "loss": 1.5228,
+      "step": 649
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886999837532237e-06,
+      "loss": 1.4694,
+      "step": 650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886494470378691e-06,
+      "loss": 1.9768,
+      "step": 651
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885987988649857e-06,
+      "loss": 1.651,
+      "step": 652
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885480392461258e-06,
+      "loss": 1.9581,
+      "step": 653
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884971681928674e-06,
+      "loss": 1.1611,
+      "step": 654
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884461857168137e-06,
+      "loss": 1.4817,
+      "step": 655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883950918295934e-06,
+      "loss": 1.7017,
+      "step": 656
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883438865428609e-06,
+      "loss": 1.5292,
+      "step": 657
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882925698682955e-06,
+      "loss": 1.6886,
+      "step": 658
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882411418176023e-06,
+      "loss": 1.5946,
+      "step": 659
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881896024025119e-06,
+      "loss": 1.3972,
+      "step": 660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881379516347797e-06,
+      "loss": 1.6662,
+      "step": 661
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88086189526187e-06,
+      "loss": 1.6954,
+      "step": 662
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880343160885406e-06,
+      "loss": 1.6994,
+      "step": 663
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879823313336723e-06,
+      "loss": 1.5972,
+      "step": 664
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879302352734395e-06,
+      "loss": 1.6346,
+      "step": 665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878780279197246e-06,
+      "loss": 1.5839,
+      "step": 666
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878257092844363e-06,
+      "loss": 1.3247,
+      "step": 667
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877732793795078e-06,
+      "loss": 1.4522,
+      "step": 668
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87720738216898e-06,
+      "loss": 1.4161,
+      "step": 669
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876680858085912e-06,
+      "loss": 1.4794,
+      "step": 670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876153221665971e-06,
+      "loss": 1.2934,
+      "step": 671
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875624473029508e-06,
+      "loss": 1.5938,
+      "step": 672
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875094612297124e-06,
+      "loss": 1.4479,
+      "step": 673
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87456363958968e-06,
+      "loss": 1.6953,
+      "step": 674
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874031555028285e-06,
+      "loss": 1.3522,
+      "step": 675
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873498358734304e-06,
+      "loss": 1.1769,
+      "step": 676
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872964050829355e-06,
+      "loss": 1.4984,
+      "step": 677
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87242863143531e-06,
+      "loss": 1.7154,
+      "step": 678
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871892100674297e-06,
+      "loss": 1.4628,
+      "step": 679
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87135445866869e-06,
+      "loss": 1.5878,
+      "step": 680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870815705541126e-06,
+      "loss": 1.5562,
+      "step": 681
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870275841414489e-06,
+      "loss": 1.8696,
+      "step": 682
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869734866411917e-06,
+      "loss": 1.4447,
+      "step": 683
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869192780656806e-06,
+      "loss": 1.5809,
+      "step": 684
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8686495842728e-06,
+      "loss": 1.7078,
+      "step": 685
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868105277383797e-06,
+      "loss": 1.6436,
+      "step": 686
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86755986011395e-06,
+      "loss": 1.5466,
+      "step": 687
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867013332587667e-06,
+      "loss": 1.5929,
+      "step": 688
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866465694929605e-06,
+      "loss": 1.4959,
+      "step": 689
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865916947264677e-06,
+      "loss": 1.6399,
+      "step": 690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86536708971805e-06,
+      "loss": 1.6295,
+      "step": 691
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864816122415139e-06,
+      "loss": 1.3543,
+      "step": 692
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86426404548162e-06,
+      "loss": 1.3559,
+      "step": 693
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863710859043416e-06,
+      "loss": 2.0963,
+      "step": 694
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863156563226703e-06,
+      "loss": 1.0959,
+      "step": 695
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862601158157916e-06,
+      "loss": 1.627,
+      "step": 696
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862044643963737e-06,
+      "loss": 1.2871,
+      "step": 697
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861487020771103e-06,
+      "loss": 1.2443,
+      "step": 698
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860928288707203e-06,
+      "loss": 1.5211,
+      "step": 699
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86036844789948e-06,
+      "loss": 1.6305,
+      "step": 700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859807498475633e-06,
+      "loss": 1.4059,
+      "step": 701
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859245440563606e-06,
+      "loss": 1.1757,
+      "step": 702
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858682274291605e-06,
+      "loss": 1.4282,
+      "step": 703
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85811799978808e-06,
+      "loss": 1.6226,
+      "step": 704
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85755261718174e-06,
+      "loss": 1.2743,
+      "step": 705
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856986126601546e-06,
+      "loss": 1.2687,
+      "step": 706
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85641852817671e-06,
+      "loss": 1.3168,
+      "step": 707
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855849822036695e-06,
+      "loss": 1.4914,
+      "step": 708
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855280008311223e-06,
+      "loss": 1.8088,
+      "step": 709
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854709087130261e-06,
+      "loss": 1.5148,
+      "step": 710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854137058624034e-06,
+      "loss": 1.7648,
+      "step": 711
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.853563922923015e-06,
+      "loss": 1.6352,
+      "step": 712
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852989680157936e-06,
+      "loss": 1.5515,
+      "step": 713
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852414330459776e-06,
+      "loss": 1.3262,
+      "step": 714
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851837873959768e-06,
+      "loss": 1.7384,
+      "step": 715
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851260310789396e-06,
+      "loss": 1.2284,
+      "step": 716
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850681641080405e-06,
+      "loss": 1.6252,
+      "step": 717
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850101864964776e-06,
+      "loss": 1.3803,
+      "step": 718
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849520982574759e-06,
+      "loss": 1.6032,
+      "step": 719
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848938994042844e-06,
+      "loss": 1.5065,
+      "step": 720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848355899501786e-06,
+      "loss": 1.2507,
+      "step": 721
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847771699084577e-06,
+      "loss": 1.6248,
+      "step": 722
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84718639292447e-06,
+      "loss": 1.3287,
+      "step": 723
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846599981154975e-06,
+      "loss": 1.7183,
+      "step": 724
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846012463909845e-06,
+      "loss": 1.4378,
+      "step": 725
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845423841323085e-06,
+      "loss": 1.2852,
+      "step": 726
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844834113528962e-06,
+      "loss": 1.4184,
+      "step": 727
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844243280661984e-06,
+      "loss": 1.3176,
+      "step": 728
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843651342856919e-06,
+      "loss": 1.5111,
+      "step": 729
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843058300248783e-06,
+      "loss": 1.3917,
+      "step": 730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842464152972843e-06,
+      "loss": 1.2483,
+      "step": 731
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841868901164621e-06,
+      "loss": 1.516,
+      "step": 732
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841272544959892e-06,
+      "loss": 1.6016,
+      "step": 733
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84067508449468e-06,
+      "loss": 1.4881,
+      "step": 734
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840076519905257e-06,
+      "loss": 1.4611,
+      "step": 735
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839476851328157e-06,
+      "loss": 1.3731,
+      "step": 736
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838876078900158e-06,
+      "loss": 1.1527,
+      "step": 737
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83827420275829e-06,
+      "loss": 1.2926,
+      "step": 738
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83767122303984e-06,
+      "loss": 1.3744,
+      "step": 739
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837067139882342e-06,
+      "loss": 1.9387,
+      "step": 740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836461953423584e-06,
+      "loss": 1.3649,
+      "step": 741
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835855663801603e-06,
+      "loss": 1.6315,
+      "step": 742
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835248271154693e-06,
+      "loss": 1.3138,
+      "step": 743
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83463977562139e-06,
+      "loss": 1.4969,
+      "step": 744
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834030177340494e-06,
+      "loss": 1.5427,
+      "step": 745
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833419476451047e-06,
+      "loss": 1.4993,
+      "step": 746
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832807673092346e-06,
+      "loss": 1.4817,
+      "step": 747
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832194767403938e-06,
+      "loss": 1.7172,
+      "step": 748
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831580759525626e-06,
+      "loss": 1.4221,
+      "step": 749
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830965649597455e-06,
+      "loss": 1.7103,
+      "step": 750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830349437759731e-06,
+      "loss": 1.8245,
+      "step": 751
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829732124153012e-06,
+      "loss": 1.3566,
+      "step": 752
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829113708918092e-06,
+      "loss": 1.4303,
+      "step": 753
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828494192196037e-06,
+      "loss": 1.3129,
+      "step": 754
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82787357412815e-06,
+      "loss": 1.4439,
+      "step": 755
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827251854855992e-06,
+      "loss": 1.3404,
+      "step": 756
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82662903452137e-06,
+      "loss": 1.5391,
+      "step": 757
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826005113266346e-06,
+      "loss": 1.6735,
+      "step": 758
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825380091233234e-06,
+      "loss": 1.5463,
+      "step": 759
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824753968564593e-06,
+      "loss": 1.6669,
+      "step": 760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824126745403241e-06,
+      "loss": 1.1105,
+      "step": 761
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823498421892243e-06,
+      "loss": 1.4457,
+      "step": 762
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822868998174914e-06,
+      "loss": 1.484,
+      "step": 763
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822238474394823e-06,
+      "loss": 1.7306,
+      "step": 764
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821606850695784e-06,
+      "loss": 1.3981,
+      "step": 765
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820974127221869e-06,
+      "loss": 1.5863,
+      "step": 766
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820340304117397e-06,
+      "loss": 1.3422,
+      "step": 767
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81970538152694e-06,
+      "loss": 1.4114,
+      "step": 768
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819069359595316e-06,
+      "loss": 1.4764,
+      "step": 769
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818432238467603e-06,
+      "loss": 1.4183,
+      "step": 770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817794018289118e-06,
+      "loss": 1.7543,
+      "step": 771
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817154699205438e-06,
+      "loss": 1.7805,
+      "step": 772
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816514281362385e-06,
+      "loss": 1.7625,
+      "step": 773
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815872764906036e-06,
+      "loss": 1.5714,
+      "step": 774
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815230149982714e-06,
+      "loss": 1.2829,
+      "step": 775
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814586436738998e-06,
+      "loss": 1.3956,
+      "step": 776
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81394162532171e-06,
+      "loss": 1.2609,
+      "step": 777
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813295715877933e-06,
+      "loss": 1.6083,
+      "step": 778
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.812648708554991e-06,
+      "loss": 1.752,
+      "step": 779
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.812000603500462e-06,
+      "loss": 1.3568,
+      "step": 780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.811351400862172e-06,
+      "loss": 1.5168,
+      "step": 781
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810701100788205e-06,
+      "loss": 1.7195,
+      "step": 782
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810049703426885e-06,
+      "loss": 1.6539,
+      "step": 783
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809397208926793e-06,
+      "loss": 1.2104,
+      "step": 784
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80874361743676e-06,
+      "loss": 1.4576,
+      "step": 785
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808088929105865e-06,
+      "loss": 1.3451,
+      "step": 786
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807433144083436e-06,
+      "loss": 1.5562,
+      "step": 787
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806776262519053e-06,
+      "loss": 1.4091,
+      "step": 788
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806118284562547e-06,
+      "loss": 1.3687,
+      "step": 789
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805459210364e-06,
+      "loss": 1.4947,
+      "step": 790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804799040073739e-06,
+      "loss": 1.3341,
+      "step": 791
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804137773842346e-06,
+      "loss": 1.5473,
+      "step": 792
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803475411820651e-06,
+      "loss": 1.5183,
+      "step": 793
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802811954159734e-06,
+      "loss": 1.4207,
+      "step": 794
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802147401010926e-06,
+      "loss": 1.6394,
+      "step": 795
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801481752525806e-06,
+      "loss": 1.352,
+      "step": 796
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800815008856204e-06,
+      "loss": 1.4081,
+      "step": 797
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.8001471701542e-06,
+      "loss": 1.5054,
+      "step": 798
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.799478236572123e-06,
+      "loss": 1.3142,
+      "step": 799
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798808208262549e-06,
+      "loss": 1.2322,
+      "step": 800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798137085378312e-06,
+      "loss": 1.2586,
+      "step": 801
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797464868072489e-06,
+      "loss": 1.5035,
+      "step": 802
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796791556498405e-06,
+      "loss": 1.7881,
+      "step": 803
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796117150809641e-06,
+      "loss": 1.3252,
+      "step": 804
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.795441651160022e-06,
+      "loss": 1.4734,
+      "step": 805
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794765057703627e-06,
+      "loss": 1.2776,
+      "step": 806
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794087370594781e-06,
+      "loss": 1.4009,
+      "step": 807
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79340858998806e-06,
+      "loss": 1.4405,
+      "step": 808
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792728716038289e-06,
+      "loss": 1.3544,
+      "step": 809
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792047748900542e-06,
+      "loss": 1.4684,
+      "step": 810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.791365688730143e-06,
+      "loss": 1.8824,
+      "step": 811
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790682535682667e-06,
+      "loss": 1.4916,
+      "step": 812
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789998289913934e-06,
+      "loss": 1.7739,
+      "step": 813
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789312951580017e-06,
+      "loss": 1.5521,
+      "step": 814
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788626520837235e-06,
+      "loss": 1.5345,
+      "step": 815
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787938997842161e-06,
+      "loss": 1.8411,
+      "step": 816
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787250382751612e-06,
+      "loss": 1.3518,
+      "step": 817
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786560675722657e-06,
+      "loss": 1.9003,
+      "step": 818
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785869876912613e-06,
+      "loss": 1.5555,
+      "step": 819
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785177986479049e-06,
+      "loss": 1.5302,
+      "step": 820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784485004579775e-06,
+      "loss": 1.561,
+      "step": 821
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78379093137286e-06,
+      "loss": 1.5246,
+      "step": 822
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783095767016617e-06,
+      "loss": 1.2851,
+      "step": 823
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782399511669605e-06,
+      "loss": 1.8957,
+      "step": 824
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781702165490638e-06,
+      "loss": 1.8275,
+      "step": 825
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781003728638775e-06,
+      "loss": 1.3237,
+      "step": 826
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780304201273324e-06,
+      "loss": 1.3863,
+      "step": 827
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.779603583553842e-06,
+      "loss": 1.4259,
+      "step": 828
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778901875640135e-06,
+      "loss": 1.5343,
+      "step": 829
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77819907769226e-06,
+      "loss": 1.4264,
+      "step": 830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777495189870518e-06,
+      "loss": 1.4979,
+      "step": 831
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776790212335461e-06,
+      "loss": 1.6392,
+      "step": 832
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776084145247889e-06,
+      "loss": 1.4452,
+      "step": 833
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775376988768853e-06,
+      "loss": 1.4323,
+      "step": 834
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77466874305965e-06,
+      "loss": 1.3904,
+      "step": 835
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773959408281823e-06,
+      "loss": 1.5723,
+      "step": 836
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77324898459717e-06,
+      "loss": 1.4484,
+      "step": 837
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772537472167733e-06,
+      "loss": 1.5031,
+      "step": 838
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771824871155801e-06,
+      "loss": 1.5384,
+      "step": 839
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771111181723916e-06,
+      "loss": 1.3519,
+      "step": 840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.770396404034863e-06,
+      "loss": 1.3909,
+      "step": 841
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76968053825168e-06,
+      "loss": 1.6751,
+      "step": 842
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76896358453765e-06,
+      "loss": 1.4605,
+      "step": 843
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768245543056303e-06,
+      "loss": 1.4643,
+      "step": 844
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767526413971424e-06,
+      "loss": 1.701,
+      "step": 845
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766806197447037e-06,
+      "loss": 1.5347,
+      "step": 846
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766084893647421e-06,
+      "loss": 1.7214,
+      "step": 847
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.765362502737098e-06,
+      "loss": 1.3511,
+      "step": 848
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.764639024880842e-06,
+      "loss": 1.6685,
+      "step": 849
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763914460243675e-06,
+      "loss": 1.3169,
+      "step": 850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763188808990862e-06,
+      "loss": 1.6245,
+      "step": 851
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.762462071287919e-06,
+      "loss": 1.5327,
+      "step": 852
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761734247300611e-06,
+      "loss": 1.6129,
+      "step": 853
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76100533719495e-06,
+      "loss": 1.7445,
+      "step": 854
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.760275341137195e-06,
+      "loss": 1.718,
+      "step": 855
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759544259293854e-06,
+      "loss": 1.7868,
+      "step": 856
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75881209183168e-06,
+      "loss": 1.464,
+      "step": 857
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758078838917676e-06,
+      "loss": 1.5235,
+      "step": 858
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75734450071909e-06,
+      "loss": 1.6145,
+      "step": 859
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.756609077403421e-06,
+      "loss": 1.5013,
+      "step": 860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755872569138414e-06,
+      "loss": 1.2072,
+      "step": 861
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755134976092063e-06,
+      "loss": 1.4464,
+      "step": 862
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.754396298432604e-06,
+      "loss": 1.551,
+      "step": 863
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753656536328529e-06,
+      "loss": 1.6506,
+      "step": 864
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752915689948568e-06,
+      "loss": 1.513,
+      "step": 865
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752173759461705e-06,
+      "loss": 1.6574,
+      "step": 866
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75143074503717e-06,
+      "loss": 1.4834,
+      "step": 867
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.750686646844438e-06,
+      "loss": 1.3465,
+      "step": 868
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749941465053235e-06,
+      "loss": 1.9299,
+      "step": 869
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749195199833527e-06,
+      "loss": 1.5411,
+      "step": 870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748447851355534e-06,
+      "loss": 1.3023,
+      "step": 871
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.747699419789724e-06,
+      "loss": 1.643,
+      "step": 872
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.746949905306805e-06,
+      "loss": 1.4894,
+      "step": 873
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74619930807774e-06,
+      "loss": 1.0792,
+      "step": 874
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745447628273733e-06,
+      "loss": 1.4694,
+      "step": 875
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744694866066234e-06,
+      "loss": 1.7273,
+      "step": 876
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.743941021626945e-06,
+      "loss": 1.3253,
+      "step": 877
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.743186095127816e-06,
+      "loss": 1.6865,
+      "step": 878
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742430086741036e-06,
+      "loss": 1.1774,
+      "step": 879
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.741672996639046e-06,
+      "loss": 1.4252,
+      "step": 880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740914824994536e-06,
+      "loss": 1.6003,
+      "step": 881
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740155571980435e-06,
+      "loss": 1.656,
+      "step": 882
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739395237769927e-06,
+      "loss": 1.6109,
+      "step": 883
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.738633822536438e-06,
+      "loss": 1.4093,
+      "step": 884
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737871326453641e-06,
+      "loss": 1.726,
+      "step": 885
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737107749695456e-06,
+      "loss": 1.4827,
+      "step": 886
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73634309243605e-06,
+      "loss": 1.3406,
+      "step": 887
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.735577354849834e-06,
+      "loss": 1.5148,
+      "step": 888
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.734810537111472e-06,
+      "loss": 1.2333,
+      "step": 889
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.734042639395865e-06,
+      "loss": 1.553,
+      "step": 890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.733273661878167e-06,
+      "loss": 1.3598,
+      "step": 891
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.732503604733776e-06,
+      "loss": 1.8178,
+      "step": 892
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731732468138338e-06,
+      "loss": 1.4297,
+      "step": 893
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730960252267744e-06,
+      "loss": 1.7174,
+      "step": 894
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730186957298128e-06,
+      "loss": 1.2063,
+      "step": 895
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.729412583405877e-06,
+      "loss": 1.9062,
+      "step": 896
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728637130767618e-06,
+      "loss": 1.3966,
+      "step": 897
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727860599560226e-06,
+      "loss": 1.3111,
+      "step": 898
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727082989960824e-06,
+      "loss": 1.5672,
+      "step": 899
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726304302146776e-06,
+      "loss": 1.638,
+      "step": 900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.725524536295701e-06,
+      "loss": 1.4362,
+      "step": 901
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724743692585454e-06,
+      "loss": 1.5911,
+      "step": 902
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.723961771194139e-06,
+      "loss": 1.4613,
+      "step": 903
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72317877230011e-06,
+      "loss": 1.5291,
+      "step": 904
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722394696081963e-06,
+      "loss": 1.4189,
+      "step": 905
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72160954271854e-06,
+      "loss": 1.1073,
+      "step": 906
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720823312388927e-06,
+      "loss": 1.6013,
+      "step": 907
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72003600527246e-06,
+      "loss": 1.5227,
+      "step": 908
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.719247621548718e-06,
+      "loss": 1.6432,
+      "step": 909
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.718458161397524e-06,
+      "loss": 1.5091,
+      "step": 910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71766762499895e-06,
+      "loss": 1.7519,
+      "step": 911
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716876012533312e-06,
+      "loss": 1.7433,
+      "step": 912
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716083324181171e-06,
+      "loss": 1.335,
+      "step": 913
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.715289560123333e-06,
+      "loss": 1.8473,
+      "step": 914
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71449472054085e-06,
+      "loss": 1.638,
+      "step": 915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.713698805615021e-06,
+      "loss": 1.7109,
+      "step": 916
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.712901815527387e-06,
+      "loss": 1.5051,
+      "step": 917
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.712103750459737e-06,
+      "loss": 1.8568,
+      "step": 918
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711304610594104e-06,
+      "loss": 1.6829,
+      "step": 919
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.710504396112765e-06,
+      "loss": 1.278,
+      "step": 920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.709703107198243e-06,
+      "loss": 1.6786,
+      "step": 921
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708900744033308e-06,
+      "loss": 1.3932,
+      "step": 922
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708097306800973e-06,
+      "loss": 1.6589,
+      "step": 923
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.707292795684498e-06,
+      "loss": 1.8568,
+      "step": 924
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70648721086738e-06,
+      "loss": 1.3298,
+      "step": 925
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.705680552533376e-06,
+      "loss": 1.7356,
+      "step": 926
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.704872820866476e-06,
+      "loss": 1.5478,
+      "step": 927
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.704064016050914e-06,
+      "loss": 1.3208,
+      "step": 928
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703254138271178e-06,
+      "loss": 1.6202,
+      "step": 929
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702443187711991e-06,
+      "loss": 1.4207,
+      "step": 930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.701631164558329e-06,
+      "loss": 1.5663,
+      "step": 931
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.700818068995407e-06,
+      "loss": 1.1636,
+      "step": 932
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.700003901208689e-06,
+      "loss": 1.6348,
+      "step": 933
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699188661383875e-06,
+      "loss": 1.5834,
+      "step": 934
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69837234970692e-06,
+      "loss": 1.6704,
+      "step": 935
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69755496636402e-06,
+      "loss": 1.356,
+      "step": 936
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.696736511541611e-06,
+      "loss": 1.5867,
+      "step": 937
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69591698542638e-06,
+      "loss": 1.2581,
+      "step": 938
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.695096388205251e-06,
+      "loss": 1.3931,
+      "step": 939
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.6942747200654e-06,
+      "loss": 1.4864,
+      "step": 940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.693451981194244e-06,
+      "loss": 1.7282,
+      "step": 941
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.692628171779441e-06,
+      "loss": 1.3014,
+      "step": 942
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.6918032920089e-06,
+      "loss": 1.1249,
+      "step": 943
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690977342070769e-06,
+      "loss": 1.4069,
+      "step": 944
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69015032215344e-06,
+      "loss": 1.4904,
+      "step": 945
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68932223244555e-06,
+      "loss": 1.5643,
+      "step": 946
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.688493073135982e-06,
+      "loss": 1.5039,
+      "step": 947
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687662844413862e-06,
+      "loss": 1.5132,
+      "step": 948
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.686831546468557e-06,
+      "loss": 1.4376,
+      "step": 949
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.685999179489684e-06,
+      "loss": 1.4905,
+      "step": 950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.685165743667097e-06,
+      "loss": 1.5694,
+      "step": 951
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.6843312391909e-06,
+      "loss": 1.3044,
+      "step": 952
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683495666251433e-06,
+      "loss": 1.2102,
+      "step": 953
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.682659025039287e-06,
+      "loss": 1.5407,
+      "step": 954
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.681821315745294e-06,
+      "loss": 1.7901,
+      "step": 955
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.680982538560531e-06,
+      "loss": 1.3517,
+      "step": 956
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.680142693676315e-06,
+      "loss": 1.4589,
+      "step": 957
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.679301781284209e-06,
+      "loss": 1.5831,
+      "step": 958
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67845980157602e-06,
+      "loss": 1.7163,
+      "step": 959
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.677616754743797e-06,
+      "loss": 1.6055,
+      "step": 960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.676772640979833e-06,
+      "loss": 1.3944,
+      "step": 961
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675927460476667e-06,
+      "loss": 1.1248,
+      "step": 962
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675081213427076e-06,
+      "loss": 1.6006,
+      "step": 963
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.674233900024082e-06,
+      "loss": 1.2961,
+      "step": 964
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.673385520460955e-06,
+      "loss": 1.508,
+      "step": 965
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.672536074931202e-06,
+      "loss": 1.6451,
+      "step": 966
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.671685563628576e-06,
+      "loss": 1.628,
+      "step": 967
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.670833986747074e-06,
+      "loss": 1.55,
+      "step": 968
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669981344480932e-06,
+      "loss": 1.259,
+      "step": 969
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669127637024634e-06,
+      "loss": 1.4742,
+      "step": 970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.668272864572904e-06,
+      "loss": 1.4383,
+      "step": 971
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66741702732071e-06,
+      "loss": 1.249,
+      "step": 972
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.666560125463263e-06,
+      "loss": 1.5714,
+      "step": 973
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665702159196014e-06,
+      "loss": 1.6358,
+      "step": 974
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.664843128714661e-06,
+      "loss": 1.4242,
+      "step": 975
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663983034215143e-06,
+      "loss": 1.2497,
+      "step": 976
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663121875893641e-06,
+      "loss": 1.7149,
+      "step": 977
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.662259653946578e-06,
+      "loss": 1.7886,
+      "step": 978
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661396368570623e-06,
+      "loss": 1.2305,
+      "step": 979
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660532019962685e-06,
+      "loss": 1.2388,
+      "step": 980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.659666608319915e-06,
+      "loss": 1.5912,
+      "step": 981
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.658800133839708e-06,
+      "loss": 1.5118,
+      "step": 982
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.657932596719699e-06,
+      "loss": 1.4288,
+      "step": 983
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65706399715777e-06,
+      "loss": 1.5012,
+      "step": 984
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65619433535204e-06,
+      "loss": 1.2149,
+      "step": 985
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.655323611500876e-06,
+      "loss": 1.4613,
+      "step": 986
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.654451825802881e-06,
+      "loss": 1.5415,
+      "step": 987
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.653578978456904e-06,
+      "loss": 1.6894,
+      "step": 988
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.652705069662035e-06,
+      "loss": 1.5697,
+      "step": 989
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.651830099617607e-06,
+      "loss": 1.4373,
+      "step": 990
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650954068523196e-06,
+      "loss": 1.503,
+      "step": 991
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650076976578615e-06,
+      "loss": 1.8807,
+      "step": 992
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.649198823983927e-06,
+      "loss": 1.1919,
+      "step": 993
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64831961093943e-06,
+      "loss": 1.4173,
+      "step": 994
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.647439337645666e-06,
+      "loss": 1.6179,
+      "step": 995
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.646558004303419e-06,
+      "loss": 1.3524,
+      "step": 996
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.645675611113715e-06,
+      "loss": 1.3988,
+      "step": 997
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.644792158277826e-06,
+      "loss": 1.4626,
+      "step": 998
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.643907645997257e-06,
+      "loss": 1.8531,
+      "step": 999
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.643022074473758e-06,
+      "loss": 1.4534,
+      "step": 1000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.642135443909323e-06,
+      "loss": 1.5121,
+      "step": 1001
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64124775450619e-06,
+      "loss": 1.6012,
+      "step": 1002
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64035900646683e-06,
+      "loss": 1.2617,
+      "step": 1003
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.639469199993962e-06,
+      "loss": 1.8295,
+      "step": 1004
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.638578335290546e-06,
+      "loss": 1.4884,
+      "step": 1005
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63768641255978e-06,
+      "loss": 1.476,
+      "step": 1006
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.636793432005106e-06,
+      "loss": 1.3209,
+      "step": 1007
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.635899393830206e-06,
+      "loss": 1.6976,
+      "step": 1008
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.635004298239004e-06,
+      "loss": 1.5399,
+      "step": 1009
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.634108145435665e-06,
+      "loss": 1.203,
+      "step": 1010
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.633210935624598e-06,
+      "loss": 1.5377,
+      "step": 1011
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.632312669010446e-06,
+      "loss": 1.7412,
+      "step": 1012
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.6314133457981e-06,
+      "loss": 1.2214,
+      "step": 1013
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.630512966192688e-06,
+      "loss": 1.5492,
+      "step": 1014
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.629611530399584e-06,
+      "loss": 1.7114,
+      "step": 1015
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.628709038624394e-06,
+      "loss": 1.509,
+      "step": 1016
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.627805491072972e-06,
+      "loss": 1.6256,
+      "step": 1017
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62690088795141e-06,
+      "loss": 1.5467,
+      "step": 1018
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.625995229466045e-06,
+      "loss": 1.4868,
+      "step": 1019
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62508851582345e-06,
+      "loss": 1.7778,
+      "step": 1020
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.624180747230439e-06,
+      "loss": 1.5362,
+      "step": 1021
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.623271923894066e-06,
+      "loss": 1.3214,
+      "step": 1022
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62236204602163e-06,
+      "loss": 1.4458,
+      "step": 1023
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.621451113820667e-06,
+      "loss": 1.7235,
+      "step": 1024
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.620539127498954e-06,
+      "loss": 1.2655,
+      "step": 1025
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61962608726451e-06,
+      "loss": 1.7154,
+      "step": 1026
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.618711993325592e-06,
+      "loss": 1.0951,
+      "step": 1027
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617796845890698e-06,
+      "loss": 1.5891,
+      "step": 1028
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.616880645168567e-06,
+      "loss": 1.6723,
+      "step": 1029
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61596339136818e-06,
+      "loss": 1.5891,
+      "step": 1030
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615045084698753e-06,
+      "loss": 1.5589,
+      "step": 1031
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.614125725369748e-06,
+      "loss": 1.3864,
+      "step": 1032
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61320531359086e-06,
+      "loss": 1.22,
+      "step": 1033
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.612283849572037e-06,
+      "loss": 1.5922,
+      "step": 1034
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61136133352345e-06,
+      "loss": 1.6394,
+      "step": 1035
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.610437765655522e-06,
+      "loss": 1.5633,
+      "step": 1036
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.609513146178912e-06,
+      "loss": 1.4472,
+      "step": 1037
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.608587475304521e-06,
+      "loss": 1.837,
+      "step": 1038
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607660753243484e-06,
+      "loss": 1.6412,
+      "step": 1039
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.606732980207186e-06,
+      "loss": 1.5025,
+      "step": 1040
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.605804156407239e-06,
+      "loss": 1.4295,
+      "step": 1041
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.604874282055506e-06,
+      "loss": 1.3261,
+      "step": 1042
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60394335736408e-06,
+      "loss": 1.5341,
+      "step": 1043
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.603011382545305e-06,
+      "loss": 1.4314,
+      "step": 1044
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60207835781175e-06,
+      "loss": 1.4073,
+      "step": 1045
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.601144283376239e-06,
+      "loss": 1.7661,
+      "step": 1046
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.600209159451824e-06,
+      "loss": 1.5322,
+      "step": 1047
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.5992729862518e-06,
+      "loss": 1.5697,
+      "step": 1048
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.598335763989703e-06,
+      "loss": 1.3711,
+      "step": 1049
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.597397492879308e-06,
+      "loss": 1.2951,
+      "step": 1050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.596458173134624e-06,
+      "loss": 1.4056,
+      "step": 1051
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.595517804969907e-06,
+      "loss": 1.5444,
+      "step": 1052
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.594576388599647e-06,
+      "loss": 1.2825,
+      "step": 1053
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.593633924238575e-06,
+      "loss": 1.6441,
+      "step": 1054
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.59269041210166e-06,
+      "loss": 1.3448,
+      "step": 1055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.591745852404108e-06,
+      "loss": 1.4507,
+      "step": 1056
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.590800245361372e-06,
+      "loss": 1.4261,
+      "step": 1057
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.589853591189135e-06,
+      "loss": 1.5989,
+      "step": 1058
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.588905890103322e-06,
+      "loss": 1.8019,
+      "step": 1059
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.587957142320098e-06,
+      "loss": 1.8088,
+      "step": 1060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.587007348055866e-06,
+      "loss": 1.4616,
+      "step": 1061
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.586056507527266e-06,
+      "loss": 1.6042,
+      "step": 1062
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58510462095118e-06,
+      "loss": 1.4491,
+      "step": 1063
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.584151688544726e-06,
+      "loss": 1.6207,
+      "step": 1064
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.583197710525258e-06,
+      "loss": 1.3892,
+      "step": 1065
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.582242687110377e-06,
+      "loss": 1.5834,
+      "step": 1066
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.581286618517916e-06,
+      "loss": 1.4372,
+      "step": 1067
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.580329504965944e-06,
+      "loss": 1.6289,
+      "step": 1068
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.579371346672776e-06,
+      "loss": 1.4161,
+      "step": 1069
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57841214385696e-06,
+      "loss": 1.5621,
+      "step": 1070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.577451896737281e-06,
+      "loss": 1.5114,
+      "step": 1071
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.576490605532766e-06,
+      "loss": 1.474,
+      "step": 1072
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.575528270462684e-06,
+      "loss": 1.5645,
+      "step": 1073
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.574564891746529e-06,
+      "loss": 1.6603,
+      "step": 1074
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.573600469604044e-06,
+      "loss": 1.5604,
+      "step": 1075
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.572635004255208e-06,
+      "loss": 1.797,
+      "step": 1076
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.571668495920236e-06,
+      "loss": 1.2889,
+      "step": 1077
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.570700944819584e-06,
+      "loss": 1.4402,
+      "step": 1078
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.569732351173941e-06,
+      "loss": 1.6627,
+      "step": 1079
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.568762715204236e-06,
+      "loss": 1.5633,
+      "step": 1080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.567792037131637e-06,
+      "loss": 1.3848,
+      "step": 1081
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56682031717755e-06,
+      "loss": 1.4322,
+      "step": 1082
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.565847555563617e-06,
+      "loss": 1.7149,
+      "step": 1083
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.564873752511719e-06,
+      "loss": 1.3944,
+      "step": 1084
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56389890824397e-06,
+      "loss": 1.4152,
+      "step": 1085
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56292302298273e-06,
+      "loss": 1.5967,
+      "step": 1086
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.561946096950588e-06,
+      "loss": 1.6062,
+      "step": 1087
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.560968130370376e-06,
+      "loss": 1.326,
+      "step": 1088
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55998912346516e-06,
+      "loss": 1.218,
+      "step": 1089
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.559009076458248e-06,
+      "loss": 1.2768,
+      "step": 1090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.558027989573177e-06,
+      "loss": 1.3822,
+      "step": 1091
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55704586303373e-06,
+      "loss": 1.5131,
+      "step": 1092
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55606269706392e-06,
+      "loss": 1.4665,
+      "step": 1093
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.555078491888002e-06,
+      "loss": 1.3133,
+      "step": 1094
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.554093247730467e-06,
+      "loss": 1.3072,
+      "step": 1095
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.553106964816041e-06,
+      "loss": 1.4869,
+      "step": 1096
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.552119643369689e-06,
+      "loss": 1.6559,
+      "step": 1097
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.551131283616611e-06,
+      "loss": 1.7645,
+      "step": 1098
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.550141885782247e-06,
+      "loss": 1.6771,
+      "step": 1099
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54915145009227e-06,
+      "loss": 1.4029,
+      "step": 1100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.548159976772593e-06,
+      "loss": 1.7136,
+      "step": 1101
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.547167466049363e-06,
+      "loss": 1.5877,
+      "step": 1102
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.546173918148964e-06,
+      "loss": 1.3064,
+      "step": 1103
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.545179333298018e-06,
+      "loss": 1.0022,
+      "step": 1104
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.544183711723386e-06,
+      "loss": 1.595,
+      "step": 1105
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.543187053652156e-06,
+      "loss": 1.496,
+      "step": 1106
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.542189359311664e-06,
+      "loss": 1.6041,
+      "step": 1107
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.541190628929475e-06,
+      "loss": 1.6018,
+      "step": 1108
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.540190862733394e-06,
+      "loss": 1.57,
+      "step": 1109
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.539190060951459e-06,
+      "loss": 1.6958,
+      "step": 1110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.538188223811946e-06,
+      "loss": 1.4735,
+      "step": 1111
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.537185351543367e-06,
+      "loss": 1.3427,
+      "step": 1112
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.536181444374472e-06,
+      "loss": 1.6854,
+      "step": 1113
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.535176502534242e-06,
+      "loss": 1.3773,
+      "step": 1114
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.534170526251901e-06,
+      "loss": 1.5125,
+      "step": 1115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.533163515756902e-06,
+      "loss": 1.7061,
+      "step": 1116
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.53215547127894e-06,
+      "loss": 1.5464,
+      "step": 1117
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.53114639304794e-06,
+      "loss": 1.1787,
+      "step": 1118
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.530136281294067e-06,
+      "loss": 1.5355,
+      "step": 1119
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.52912513624772e-06,
+      "loss": 1.1403,
+      "step": 1120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.528112958139535e-06,
+      "loss": 1.4243,
+      "step": 1121
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.527099747200382e-06,
+      "loss": 1.5949,
+      "step": 1122
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.526085503661368e-06,
+      "loss": 1.4822,
+      "step": 1123
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.525070227753835e-06,
+      "loss": 1.6049,
+      "step": 1124
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.524053919709359e-06,
+      "loss": 1.6012,
+      "step": 1125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.523036579759755e-06,
+      "loss": 1.633,
+      "step": 1126
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.522018208137066e-06,
+      "loss": 1.512,
+      "step": 1127
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.520998805073583e-06,
+      "loss": 1.701,
+      "step": 1128
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.51997837080182e-06,
+      "loss": 1.5978,
+      "step": 1129
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.518956905554533e-06,
+      "loss": 1.6715,
+      "step": 1130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.517934409564708e-06,
+      "loss": 1.5826,
+      "step": 1131
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.51691088306557e-06,
+      "loss": 1.713,
+      "step": 1132
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.515886326290583e-06,
+      "loss": 1.3618,
+      "step": 1133
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.514860739473436e-06,
+      "loss": 1.3135,
+      "step": 1134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.51383412284806e-06,
+      "loss": 1.2038,
+      "step": 1135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.512806476648619e-06,
+      "loss": 1.3978,
+      "step": 1136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.511777801109512e-06,
+      "loss": 1.6956,
+      "step": 1137
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.510748096465372e-06,
+      "loss": 1.506,
+      "step": 1138
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.509717362951068e-06,
+      "loss": 1.4899,
+      "step": 1139
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.508685600801704e-06,
+      "loss": 1.5985,
+      "step": 1140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.507652810252616e-06,
+      "loss": 1.429,
+      "step": 1141
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.506618991539378e-06,
+      "loss": 1.5698,
+      "step": 1142
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.505584144897797e-06,
+      "loss": 1.3061,
+      "step": 1143
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.50454827056391e-06,
+      "loss": 1.736,
+      "step": 1144
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.503511368774001e-06,
+      "loss": 1.6135,
+      "step": 1145
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.502473439764572e-06,
+      "loss": 1.4474,
+      "step": 1146
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.501434483772371e-06,
+      "loss": 1.436,
+      "step": 1147
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.500394501034376e-06,
+      "loss": 1.4017,
+      "step": 1148
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.499353491787799e-06,
+      "loss": 1.5926,
+      "step": 1149
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.498311456270091e-06,
+      "loss": 1.3285,
+      "step": 1150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.497268394718926e-06,
+      "loss": 1.6043,
+      "step": 1151
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.496224307372225e-06,
+      "loss": 1.6706,
+      "step": 1152
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.495179194468135e-06,
+      "loss": 1.6031,
+      "step": 1153
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.494133056245039e-06,
+      "loss": 1.6968,
+      "step": 1154
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.493085892941553e-06,
+      "loss": 1.4931,
+      "step": 1155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49203770479653e-06,
+      "loss": 1.4301,
+      "step": 1156
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.490988492049052e-06,
+      "loss": 1.3359,
+      "step": 1157
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.489938254938438e-06,
+      "loss": 1.3412,
+      "step": 1158
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.488886993704239e-06,
+      "loss": 1.6609,
+      "step": 1159
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.487834708586241e-06,
+      "loss": 1.2589,
+      "step": 1160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.486781399824464e-06,
+      "loss": 1.7841,
+      "step": 1161
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.485727067659159e-06,
+      "loss": 1.5423,
+      "step": 1162
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48467171233081e-06,
+      "loss": 1.4445,
+      "step": 1163
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.483615334080141e-06,
+      "loss": 1.3888,
+      "step": 1164
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.482557933148102e-06,
+      "loss": 1.8437,
+      "step": 1165
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.481499509775878e-06,
+      "loss": 1.6067,
+      "step": 1166
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.480440064204888e-06,
+      "loss": 1.4854,
+      "step": 1167
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.479379596676785e-06,
+      "loss": 1.65,
+      "step": 1168
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.478318107433452e-06,
+      "loss": 1.3232,
+      "step": 1169
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.477255596717012e-06,
+      "loss": 1.3872,
+      "step": 1170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.476192064769813e-06,
+      "loss": 1.3666,
+      "step": 1171
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.475127511834438e-06,
+      "loss": 1.4814,
+      "step": 1172
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.474061938153709e-06,
+      "loss": 1.5768,
+      "step": 1173
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.47299534397067e-06,
+      "loss": 1.2363,
+      "step": 1174
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.471927729528609e-06,
+      "loss": 1.4002,
+      "step": 1175
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.470859095071036e-06,
+      "loss": 1.6171,
+      "step": 1176
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.469789440841703e-06,
+      "loss": 1.2375,
+      "step": 1177
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.468718767084591e-06,
+      "loss": 1.615,
+      "step": 1178
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.467647074043911e-06,
+      "loss": 1.6113,
+      "step": 1179
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.466574361964108e-06,
+      "loss": 1.4737,
+      "step": 1180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.465500631089864e-06,
+      "loss": 1.5018,
+      "step": 1181
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.464425881666085e-06,
+      "loss": 1.1992,
+      "step": 1182
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.463350113937917e-06,
+      "loss": 1.4281,
+      "step": 1183
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.462273328150734e-06,
+      "loss": 1.4336,
+      "step": 1184
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.461195524550145e-06,
+      "loss": 1.4896,
+      "step": 1185
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.460116703381985e-06,
+      "loss": 1.4439,
+      "step": 1186
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.459036864892331e-06,
+      "loss": 1.7641,
+      "step": 1187
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.457956009327482e-06,
+      "loss": 1.6706,
+      "step": 1188
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.456874136933977e-06,
+      "loss": 1.9218,
+      "step": 1189
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.45579124795858e-06,
+      "loss": 1.5794,
+      "step": 1190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.454707342648297e-06,
+      "loss": 1.3974,
+      "step": 1191
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.453622421250353e-06,
+      "loss": 1.5574,
+      "step": 1192
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.452536484012212e-06,
+      "loss": 1.5217,
+      "step": 1193
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.451449531181571e-06,
+      "loss": 1.4315,
+      "step": 1194
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.450361563006354e-06,
+      "loss": 1.3994,
+      "step": 1195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.44927257973472e-06,
+      "loss": 1.1895,
+      "step": 1196
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.44818258161506e-06,
+      "loss": 1.3709,
+      "step": 1197
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.447091568895991e-06,
+      "loss": 1.6319,
+      "step": 1198
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.44599954182637e-06,
+      "loss": 1.5211,
+      "step": 1199
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.444906500655276e-06,
+      "loss": 1.7548,
+      "step": 1200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.443812445632028e-06,
+      "loss": 1.25,
+      "step": 1201
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.442717377006172e-06,
+      "loss": 1.4348,
+      "step": 1202
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.441621295027483e-06,
+      "loss": 1.2944,
+      "step": 1203
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.440524199945972e-06,
+      "loss": 1.8535,
+      "step": 1204
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.439426092011877e-06,
+      "loss": 1.3875,
+      "step": 1205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.43832697147567e-06,
+      "loss": 1.6578,
+      "step": 1206
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.437226838588052e-06,
+      "loss": 1.2502,
+      "step": 1207
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.436125693599955e-06,
+      "loss": 1.4248,
+      "step": 1208
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.435023536762546e-06,
+      "loss": 1.5847,
+      "step": 1209
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.433920368327215e-06,
+      "loss": 1.5397,
+      "step": 1210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.43281618854559e-06,
+      "loss": 1.574,
+      "step": 1211
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.431710997669523e-06,
+      "loss": 1.528,
+      "step": 1212
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.430604795951104e-06,
+      "loss": 1.1437,
+      "step": 1213
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42949758364265e-06,
+      "loss": 1.6166,
+      "step": 1214
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.428389360996705e-06,
+      "loss": 1.5999,
+      "step": 1215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.427280128266049e-06,
+      "loss": 1.4728,
+      "step": 1216
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.426169885703693e-06,
+      "loss": 1.3925,
+      "step": 1217
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42505863356287e-06,
+      "loss": 1.3518,
+      "step": 1218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.423946372097053e-06,
+      "loss": 1.5656,
+      "step": 1219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.422833101559939e-06,
+      "loss": 1.7177,
+      "step": 1220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.421718822205457e-06,
+      "loss": 1.2889,
+      "step": 1221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.420603534287767e-06,
+      "loss": 1.7856,
+      "step": 1222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.41948723806126e-06,
+      "loss": 1.3694,
+      "step": 1223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.418369933780554e-06,
+      "loss": 1.37,
+      "step": 1224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.417251621700499e-06,
+      "loss": 1.3928,
+      "step": 1225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.416132302076173e-06,
+      "loss": 1.905,
+      "step": 1226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.415011975162885e-06,
+      "loss": 1.3965,
+      "step": 1227
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.413890641216173e-06,
+      "loss": 1.0713,
+      "step": 1228
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.412768300491807e-06,
+      "loss": 1.9821,
+      "step": 1229
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.411644953245788e-06,
+      "loss": 1.553,
+      "step": 1230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.410520599734338e-06,
+      "loss": 1.6485,
+      "step": 1231
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.409395240213917e-06,
+      "loss": 1.253,
+      "step": 1232
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.408268874941211e-06,
+      "loss": 1.5252,
+      "step": 1233
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.407141504173137e-06,
+      "loss": 1.3976,
+      "step": 1234
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40601312816684e-06,
+      "loss": 1.4254,
+      "step": 1235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.404883747179695e-06,
+      "loss": 1.7027,
+      "step": 1236
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.403753361469304e-06,
+      "loss": 1.3688,
+      "step": 1237
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.4026219712935e-06,
+      "loss": 1.707,
+      "step": 1238
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40148957691035e-06,
+      "loss": 1.5411,
+      "step": 1239
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.400356178578139e-06,
+      "loss": 1.9324,
+      "step": 1240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.399221776555392e-06,
+      "loss": 1.3915,
+      "step": 1241
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.398086371100856e-06,
+      "loss": 1.7221,
+      "step": 1242
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.39694996247351e-06,
+      "loss": 1.0538,
+      "step": 1243
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.395812550932559e-06,
+      "loss": 1.601,
+      "step": 1244
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.39467413673744e-06,
+      "loss": 1.6618,
+      "step": 1245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.393534720147819e-06,
+      "loss": 1.6142,
+      "step": 1246
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.392394301423586e-06,
+      "loss": 1.4409,
+      "step": 1247
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.391252880824864e-06,
+      "loss": 1.7287,
+      "step": 1248
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.390110458612002e-06,
+      "loss": 1.7455,
+      "step": 1249
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.388967035045582e-06,
+      "loss": 1.5984,
+      "step": 1250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.387822610386406e-06,
+      "loss": 1.6616,
+      "step": 1251
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.386677184895513e-06,
+      "loss": 1.2086,
+      "step": 1252
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.385530758834165e-06,
+      "loss": 1.7706,
+      "step": 1253
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.384383332463853e-06,
+      "loss": 1.3401,
+      "step": 1254
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.383234906046298e-06,
+      "loss": 1.3705,
+      "step": 1255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.38208547984345e-06,
+      "loss": 1.4969,
+      "step": 1256
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.38093505411748e-06,
+      "loss": 1.5295,
+      "step": 1257
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.379783629130796e-06,
+      "loss": 1.4626,
+      "step": 1258
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.378631205146032e-06,
+      "loss": 1.5125,
+      "step": 1259
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.377477782426041e-06,
+      "loss": 1.3496,
+      "step": 1260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.376323361233917e-06,
+      "loss": 1.5996,
+      "step": 1261
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.375167941832974e-06,
+      "loss": 1.6351,
+      "step": 1262
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.374011524486752e-06,
+      "loss": 1.5473,
+      "step": 1263
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.372854109459024e-06,
+      "loss": 1.4076,
+      "step": 1264
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.37169569701379e-06,
+      "loss": 1.5645,
+      "step": 1265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.370536287415273e-06,
+      "loss": 1.3335,
+      "step": 1266
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.369375880927926e-06,
+      "loss": 1.5327,
+      "step": 1267
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.368214477816433e-06,
+      "loss": 1.7011,
+      "step": 1268
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.3670520783457e-06,
+      "loss": 1.4544,
+      "step": 1269
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.365888682780862e-06,
+      "loss": 1.4302,
+      "step": 1270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.36472429138728e-06,
+      "loss": 1.5059,
+      "step": 1271
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.363558904430546e-06,
+      "loss": 1.488,
+      "step": 1272
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.362392522176476e-06,
+      "loss": 1.2605,
+      "step": 1273
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.361225144891113e-06,
+      "loss": 1.4088,
+      "step": 1274
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.360056772840727e-06,
+      "loss": 1.4026,
+      "step": 1275
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.358887406291817e-06,
+      "loss": 1.5257,
+      "step": 1276
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.357717045511106e-06,
+      "loss": 1.4148,
+      "step": 1277
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.356545690765545e-06,
+      "loss": 1.3501,
+      "step": 1278
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.355373342322316e-06,
+      "loss": 1.6404,
+      "step": 1279
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.354200000448818e-06,
+      "loss": 1.4076,
+      "step": 1280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.353025665412685e-06,
+      "loss": 1.5028,
+      "step": 1281
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.351850337481774e-06,
+      "loss": 1.7257,
+      "step": 1282
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.35067401692417e-06,
+      "loss": 1.2044,
+      "step": 1283
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.349496704008181e-06,
+      "loss": 1.2978,
+      "step": 1284
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.348318399002347e-06,
+      "loss": 1.4601,
+      "step": 1285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.347139102175427e-06,
+      "loss": 1.2976,
+      "step": 1286
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.345958813796415e-06,
+      "loss": 1.4812,
+      "step": 1287
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.344777534134524e-06,
+      "loss": 1.5418,
+      "step": 1288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.343595263459196e-06,
+      "loss": 1.3357,
+      "step": 1289
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.3424120020401e-06,
+      "loss": 1.4532,
+      "step": 1290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.341227750147127e-06,
+      "loss": 1.3907,
+      "step": 1291
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.3400425080504e-06,
+      "loss": 1.5607,
+      "step": 1292
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.338856276020261e-06,
+      "loss": 1.2369,
+      "step": 1293
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.337669054327283e-06,
+      "loss": 1.2037,
+      "step": 1294
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.336480843242261e-06,
+      "loss": 1.6168,
+      "step": 1295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.335291643036221e-06,
+      "loss": 1.68,
+      "step": 1296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.33410145398041e-06,
+      "loss": 1.6809,
+      "step": 1297
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.332910276346302e-06,
+      "loss": 1.4213,
+      "step": 1298
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.331718110405596e-06,
+      "loss": 1.5308,
+      "step": 1299
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.330524956430216e-06,
+      "loss": 1.6593,
+      "step": 1300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.329330814692312e-06,
+      "loss": 1.6268,
+      "step": 1301
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.32813568546426e-06,
+      "loss": 1.7163,
+      "step": 1302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.326939569018662e-06,
+      "loss": 1.2245,
+      "step": 1303
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.325742465628342e-06,
+      "loss": 1.0942,
+      "step": 1304
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.324544375566352e-06,
+      "loss": 1.3443,
+      "step": 1305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.323345299105967e-06,
+      "loss": 1.1903,
+      "step": 1306
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.322145236520688e-06,
+      "loss": 1.491,
+      "step": 1307
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.320944188084241e-06,
+      "loss": 1.2553,
+      "step": 1308
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.319742154070578e-06,
+      "loss": 1.5637,
+      "step": 1309
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.318539134753873e-06,
+      "loss": 1.3587,
+      "step": 1310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.317335130408527e-06,
+      "loss": 1.3446,
+      "step": 1311
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.316130141309166e-06,
+      "loss": 1.8556,
+      "step": 1312
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.314924167730637e-06,
+      "loss": 1.2735,
+      "step": 1313
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.313717209948017e-06,
+      "loss": 1.4641,
+      "step": 1314
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.312509268236602e-06,
+      "loss": 1.3353,
+      "step": 1315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.311300342871919e-06,
+      "loss": 1.4336,
+      "step": 1316
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31009043412971e-06,
+      "loss": 1.3964,
+      "step": 1317
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.30887954228595e-06,
+      "loss": 1.5835,
+      "step": 1318
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.307667667616836e-06,
+      "loss": 1.1466,
+      "step": 1319
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.306454810398786e-06,
+      "loss": 1.6919,
+      "step": 1320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.305240970908444e-06,
+      "loss": 1.1541,
+      "step": 1321
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.30402614942268e-06,
+      "loss": 1.5263,
+      "step": 1322
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.302810346218587e-06,
+      "loss": 1.6452,
+      "step": 1323
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.30159356157348e-06,
+      "loss": 1.4165,
+      "step": 1324
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.300375795764897e-06,
+      "loss": 1.492,
+      "step": 1325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.299157049070604e-06,
+      "loss": 1.5503,
+      "step": 1326
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.297937321768588e-06,
+      "loss": 1.4768,
+      "step": 1327
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.296716614137062e-06,
+      "loss": 1.6878,
+      "step": 1328
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.295494926454461e-06,
+      "loss": 1.4842,
+      "step": 1329
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.29427225899944e-06,
+      "loss": 1.3918,
+      "step": 1330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.293048612050883e-06,
+      "loss": 1.1728,
+      "step": 1331
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.291823985887896e-06,
+      "loss": 1.4353,
+      "step": 1332
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.290598380789806e-06,
+      "loss": 1.3135,
+      "step": 1333
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.289371797036167e-06,
+      "loss": 1.4445,
+      "step": 1334
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.288144234906753e-06,
+      "loss": 1.7299,
+      "step": 1335
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.286915694681561e-06,
+      "loss": 1.8139,
+      "step": 1336
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.285686176640817e-06,
+      "loss": 1.2489,
+      "step": 1337
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.28445568106496e-06,
+      "loss": 1.5158,
+      "step": 1338
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.28322420823466e-06,
+      "loss": 1.7456,
+      "step": 1339
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.281991758430808e-06,
+      "loss": 1.1974,
+      "step": 1340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.280758331934517e-06,
+      "loss": 1.2752,
+      "step": 1341
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.279523929027122e-06,
+      "loss": 1.6412,
+      "step": 1342
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.278288549990183e-06,
+      "loss": 1.7151,
+      "step": 1343
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.277052195105478e-06,
+      "loss": 1.4685,
+      "step": 1344
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.275814864655016e-06,
+      "loss": 1.8505,
+      "step": 1345
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.274576558921018e-06,
+      "loss": 1.5816,
+      "step": 1346
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.273337278185936e-06,
+      "loss": 1.3594,
+      "step": 1347
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.272097022732444e-06,
+      "loss": 1.4592,
+      "step": 1348
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.27085579284343e-06,
+      "loss": 1.3919,
+      "step": 1349
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.269613588802013e-06,
+      "loss": 1.5463,
+      "step": 1350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.268370410891531e-06,
+      "loss": 1.4778,
+      "step": 1351
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.267126259395545e-06,
+      "loss": 1.4847,
+      "step": 1352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.265881134597833e-06,
+      "loss": 1.4961,
+      "step": 1353
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.264635036782406e-06,
+      "loss": 1.3888,
+      "step": 1354
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.263387966233483e-06,
+      "loss": 1.5007,
+      "step": 1355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.262139923235518e-06,
+      "loss": 1.5145,
+      "step": 1356
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.26089090807318e-06,
+      "loss": 1.4508,
+      "step": 1357
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.259640921031355e-06,
+      "loss": 1.5187,
+      "step": 1358
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.258389962395163e-06,
+      "loss": 1.3165,
+      "step": 1359
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.257138032449936e-06,
+      "loss": 1.3881,
+      "step": 1360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.255885131481231e-06,
+      "loss": 1.584,
+      "step": 1361
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.254631259774826e-06,
+      "loss": 1.2969,
+      "step": 1362
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25337641761672e-06,
+      "loss": 1.6151,
+      "step": 1363
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.252120605293133e-06,
+      "loss": 1.5037,
+      "step": 1364
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.250863823090511e-06,
+      "loss": 1.4866,
+      "step": 1365
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.249606071295512e-06,
+      "loss": 1.5953,
+      "step": 1366
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.248347350195021e-06,
+      "loss": 1.4344,
+      "step": 1367
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.247087660076148e-06,
+      "loss": 1.1392,
+      "step": 1368
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.245827001226217e-06,
+      "loss": 1.2017,
+      "step": 1369
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.244565373932775e-06,
+      "loss": 1.3253,
+      "step": 1370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.24330277848359e-06,
+      "loss": 1.5794,
+      "step": 1371
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.242039215166652e-06,
+      "loss": 1.3382,
+      "step": 1372
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.240774684270168e-06,
+      "loss": 1.1093,
+      "step": 1373
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.239509186082574e-06,
+      "loss": 1.5305,
+      "step": 1374
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23824272089252e-06,
+      "loss": 1.4782,
+      "step": 1375
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.236975288988875e-06,
+      "loss": 1.5431,
+      "step": 1376
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.235706890660735e-06,
+      "loss": 1.9019,
+      "step": 1377
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.234437526197408e-06,
+      "loss": 1.2614,
+      "step": 1378
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23316719588843e-06,
+      "loss": 1.2298,
+      "step": 1379
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.231895900023556e-06,
+      "loss": 1.5327,
+      "step": 1380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.230623638892758e-06,
+      "loss": 1.3288,
+      "step": 1381
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.22935041278623e-06,
+      "loss": 1.7685,
+      "step": 1382
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.228076221994386e-06,
+      "loss": 1.7747,
+      "step": 1383
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.22680106680786e-06,
+      "loss": 1.8166,
+      "step": 1384
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.225524947517505e-06,
+      "loss": 1.1567,
+      "step": 1385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.224247864414398e-06,
+      "loss": 1.5078,
+      "step": 1386
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.222969817789829e-06,
+      "loss": 1.413,
+      "step": 1387
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.221690807935314e-06,
+      "loss": 1.488,
+      "step": 1388
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.220410835142586e-06,
+      "loss": 1.6404,
+      "step": 1389
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.219129899703596e-06,
+      "loss": 1.3657,
+      "step": 1390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.217848001910518e-06,
+      "loss": 1.4852,
+      "step": 1391
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.216565142055745e-06,
+      "loss": 1.8905,
+      "step": 1392
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.215281320431887e-06,
+      "loss": 1.5178,
+      "step": 1393
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.213996537331773e-06,
+      "loss": 1.4481,
+      "step": 1394
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.212710793048457e-06,
+      "loss": 1.5754,
+      "step": 1395
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.211424087875205e-06,
+      "loss": 1.5531,
+      "step": 1396
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.210136422105508e-06,
+      "loss": 1.5437,
+      "step": 1397
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.20884779603307e-06,
+      "loss": 1.462,
+      "step": 1398
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.207558209951822e-06,
+      "loss": 1.4514,
+      "step": 1399
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.206267664155906e-06,
+      "loss": 1.5986,
+      "step": 1400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.204976158939688e-06,
+      "loss": 1.5324,
+      "step": 1401
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.203683694597754e-06,
+      "loss": 1.5529,
+      "step": 1402
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.2023902714249e-06,
+      "loss": 1.3742,
+      "step": 1403
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.201095889716152e-06,
+      "loss": 1.7949,
+      "step": 1404
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.199800549766746e-06,
+      "loss": 1.5593,
+      "step": 1405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.198504251872143e-06,
+      "loss": 1.2071,
+      "step": 1406
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.197206996328019e-06,
+      "loss": 1.4689,
+      "step": 1407
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.195908783430264e-06,
+      "loss": 1.4248,
+      "step": 1408
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.194609613475001e-06,
+      "loss": 1.6955,
+      "step": 1409
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.193309486758552e-06,
+      "loss": 1.3082,
+      "step": 1410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.192008403577472e-06,
+      "loss": 1.7874,
+      "step": 1411
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.190706364228528e-06,
+      "loss": 1.6267,
+      "step": 1412
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.189403369008704e-06,
+      "loss": 1.3339,
+      "step": 1413
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.188099418215208e-06,
+      "loss": 1.3433,
+      "step": 1414
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.18679451214546e-06,
+      "loss": 1.7237,
+      "step": 1415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.185488651097099e-06,
+      "loss": 1.3304,
+      "step": 1416
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.184181835367985e-06,
+      "loss": 1.6131,
+      "step": 1417
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.182874065256191e-06,
+      "loss": 1.6742,
+      "step": 1418
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.181565341060013e-06,
+      "loss": 1.4017,
+      "step": 1419
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.18025566307796e-06,
+      "loss": 1.128,
+      "step": 1420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17894503160876e-06,
+      "loss": 1.8292,
+      "step": 1421
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17763344695136e-06,
+      "loss": 1.4424,
+      "step": 1422
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.176320909404925e-06,
+      "loss": 1.5819,
+      "step": 1423
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17500741926883e-06,
+      "loss": 1.373,
+      "step": 1424
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17369297684268e-06,
+      "loss": 1.3017,
+      "step": 1425
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.172377582426286e-06,
+      "loss": 1.5888,
+      "step": 1426
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.171061236319681e-06,
+      "loss": 1.7281,
+      "step": 1427
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.169743938823115e-06,
+      "loss": 1.4841,
+      "step": 1428
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.168425690237052e-06,
+      "loss": 1.2195,
+      "step": 1429
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.167106490862178e-06,
+      "loss": 1.6903,
+      "step": 1430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.165786340999393e-06,
+      "loss": 1.5631,
+      "step": 1431
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.164465240949814e-06,
+      "loss": 1.683,
+      "step": 1432
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.163143191014772e-06,
+      "loss": 1.4679,
+      "step": 1433
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.16182019149582e-06,
+      "loss": 1.4639,
+      "step": 1434
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.160496242694725e-06,
+      "loss": 1.5863,
+      "step": 1435
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.159171344913469e-06,
+      "loss": 1.504,
+      "step": 1436
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.157845498454253e-06,
+      "loss": 1.4004,
+      "step": 1437
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.156518703619492e-06,
+      "loss": 1.5891,
+      "step": 1438
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.155190960711822e-06,
+      "loss": 1.4472,
+      "step": 1439
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.153862270034088e-06,
+      "loss": 1.4899,
+      "step": 1440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.152532631889357e-06,
+      "loss": 1.2824,
+      "step": 1441
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.151202046580908e-06,
+      "loss": 1.7709,
+      "step": 1442
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.149870514412243e-06,
+      "loss": 1.6287,
+      "step": 1443
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.148538035687071e-06,
+      "loss": 1.3426,
+      "step": 1444
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.147204610709322e-06,
+      "loss": 1.3674,
+      "step": 1445
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.145870239783143e-06,
+      "loss": 1.8416,
+      "step": 1446
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.144534923212892e-06,
+      "loss": 1.7522,
+      "step": 1447
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.143198661303146e-06,
+      "loss": 1.6784,
+      "step": 1448
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.141861454358696e-06,
+      "loss": 1.3589,
+      "step": 1449
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.140523302684553e-06,
+      "loss": 1.431,
+      "step": 1450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.139184206585937e-06,
+      "loss": 1.5746,
+      "step": 1451
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.137844166368289e-06,
+      "loss": 1.16,
+      "step": 1452
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.13650318233726e-06,
+      "loss": 1.4406,
+      "step": 1453
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.135161254798719e-06,
+      "loss": 1.737,
+      "step": 1454
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.133818384058755e-06,
+      "loss": 1.6398,
+      "step": 1455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.132474570423662e-06,
+      "loss": 1.6738,
+      "step": 1456
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.131129814199957e-06,
+      "loss": 1.2528,
+      "step": 1457
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.129784115694368e-06,
+      "loss": 1.6538,
+      "step": 1458
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.128437475213842e-06,
+      "loss": 1.2118,
+      "step": 1459
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.127089893065538e-06,
+      "loss": 1.6572,
+      "step": 1460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.12574136955683e-06,
+      "loss": 1.3085,
+      "step": 1461
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.124391904995302e-06,
+      "loss": 1.2694,
+      "step": 1462
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.123041499688764e-06,
+      "loss": 1.6776,
+      "step": 1463
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.121690153945228e-06,
+      "loss": 1.4796,
+      "step": 1464
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.120337868072933e-06,
+      "loss": 1.4544,
+      "step": 1465
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.118984642380322e-06,
+      "loss": 1.6323,
+      "step": 1466
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.117630477176058e-06,
+      "loss": 1.3922,
+      "step": 1467
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.116275372769015e-06,
+      "loss": 1.4372,
+      "step": 1468
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.114919329468283e-06,
+      "loss": 1.4961,
+      "step": 1469
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.113562347583166e-06,
+      "loss": 1.2419,
+      "step": 1470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.112204427423184e-06,
+      "loss": 1.5755,
+      "step": 1471
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.110845569298068e-06,
+      "loss": 1.543,
+      "step": 1472
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.109485773517764e-06,
+      "loss": 1.2684,
+      "step": 1473
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.108125040392432e-06,
+      "loss": 1.4663,
+      "step": 1474
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.106763370232444e-06,
+      "loss": 1.4359,
+      "step": 1475
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.10540076334839e-06,
+      "loss": 1.4899,
+      "step": 1476
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.104037220051071e-06,
+      "loss": 1.385,
+      "step": 1477
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.1026727406515e-06,
+      "loss": 1.5997,
+      "step": 1478
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.101307325460907e-06,
+      "loss": 1.3658,
+      "step": 1479
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09994097479073e-06,
+      "loss": 1.4826,
+      "step": 1480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09857368895263e-06,
+      "loss": 1.485,
+      "step": 1481
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09720546825847e-06,
+      "loss": 1.755,
+      "step": 1482
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.095836313020335e-06,
+      "loss": 1.4471,
+      "step": 1483
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.094466223550519e-06,
+      "loss": 1.5819,
+      "step": 1484
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.093095200161528e-06,
+      "loss": 1.4008,
+      "step": 1485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.091723243166086e-06,
+      "loss": 1.5025,
+      "step": 1486
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.090350352877123e-06,
+      "loss": 1.6323,
+      "step": 1487
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.088976529607789e-06,
+      "loss": 1.4559,
+      "step": 1488
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.087601773671441e-06,
+      "loss": 1.4069,
+      "step": 1489
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.086226085381654e-06,
+      "loss": 1.515,
+      "step": 1490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.08484946505221e-06,
+      "loss": 1.6039,
+      "step": 1491
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.08347191299711e-06,
+      "loss": 1.4526,
+      "step": 1492
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.082093429530562e-06,
+      "loss": 1.5809,
+      "step": 1493
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.080714014966985e-06,
+      "loss": 1.8075,
+      "step": 1494
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.07933366962102e-06,
+      "loss": 1.3046,
+      "step": 1495
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.07795239380751e-06,
+      "loss": 1.3658,
+      "step": 1496
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.076570187841516e-06,
+      "loss": 1.3633,
+      "step": 1497
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.07518705203831e-06,
+      "loss": 1.4413,
+      "step": 1498
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.073802986713374e-06,
+      "loss": 1.4,
+      "step": 1499
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.072417992182403e-06,
+      "loss": 1.6928,
+      "step": 1500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.071032068761309e-06,
+      "loss": 1.5858,
+      "step": 1501
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.069645216766207e-06,
+      "loss": 1.3928,
+      "step": 1502
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06825743651343e-06,
+      "loss": 1.3684,
+      "step": 1503
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.066868728319522e-06,
+      "loss": 1.3609,
+      "step": 1504
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.065479092501235e-06,
+      "loss": 1.3491,
+      "step": 1505
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.064088529375538e-06,
+      "loss": 1.7695,
+      "step": 1506
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.062697039259605e-06,
+      "loss": 1.8069,
+      "step": 1507
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06130462247083e-06,
+      "loss": 1.3424,
+      "step": 1508
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.05991127932681e-06,
+      "loss": 1.817,
+      "step": 1509
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.058517010145359e-06,
+      "loss": 1.4647,
+      "step": 1510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.057121815244498e-06,
+      "loss": 1.3225,
+      "step": 1511
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.055725694942463e-06,
+      "loss": 1.4305,
+      "step": 1512
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.054328649557699e-06,
+      "loss": 1.4682,
+      "step": 1513
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.05293067940886e-06,
+      "loss": 1.5825,
+      "step": 1514
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.051531784814817e-06,
+      "loss": 1.529,
+      "step": 1515
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.050131966094646e-06,
+      "loss": 1.7362,
+      "step": 1516
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.048731223567636e-06,
+      "loss": 1.4242,
+      "step": 1517
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.047329557553286e-06,
+      "loss": 1.5627,
+      "step": 1518
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.045926968371308e-06,
+      "loss": 1.5342,
+      "step": 1519
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.044523456341624e-06,
+      "loss": 1.5431,
+      "step": 1520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.04311902178436e-06,
+      "loss": 1.45,
+      "step": 1521
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.04171366501986e-06,
+      "loss": 1.6326,
+      "step": 1522
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.040307386368678e-06,
+      "loss": 1.2927,
+      "step": 1523
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.038900186151574e-06,
+      "loss": 1.2922,
+      "step": 1524
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.037492064689523e-06,
+      "loss": 1.4462,
+      "step": 1525
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.036083022303707e-06,
+      "loss": 1.259,
+      "step": 1526
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.034673059315518e-06,
+      "loss": 1.4311,
+      "step": 1527
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.033262176046558e-06,
+      "loss": 1.4765,
+      "step": 1528
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.031850372818642e-06,
+      "loss": 1.6471,
+      "step": 1529
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.03043764995379e-06,
+      "loss": 1.5118,
+      "step": 1530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.029024007774235e-06,
+      "loss": 1.6054,
+      "step": 1531
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.027609446602419e-06,
+      "loss": 1.5182,
+      "step": 1532
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.026193966760994e-06,
+      "loss": 1.611,
+      "step": 1533
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.02477756857282e-06,
+      "loss": 1.5496,
+      "step": 1534
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.023360252360969e-06,
+      "loss": 1.6269,
+      "step": 1535
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.021942018448722e-06,
+      "loss": 1.5459,
+      "step": 1536
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.020522867159564e-06,
+      "loss": 1.6276,
+      "step": 1537
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.019102798817196e-06,
+      "loss": 1.4123,
+      "step": 1538
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.017681813745526e-06,
+      "loss": 1.3537,
+      "step": 1539
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.016259912268672e-06,
+      "loss": 1.1274,
+      "step": 1540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.014837094710957e-06,
+      "loss": 1.567,
+      "step": 1541
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.013413361396917e-06,
+      "loss": 1.5604,
+      "step": 1542
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.011988712651295e-06,
+      "loss": 1.4973,
+      "step": 1543
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.010563148799045e-06,
+      "loss": 1.6945,
+      "step": 1544
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.009136670165326e-06,
+      "loss": 1.8897,
+      "step": 1545
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.007709277075512e-06,
+      "loss": 1.0486,
+      "step": 1546
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.006280969855175e-06,
+      "loss": 1.8008,
+      "step": 1547
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.004851748830107e-06,
+      "loss": 1.3861,
+      "step": 1548
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.003421614326302e-06,
+      "loss": 1.3104,
+      "step": 1549
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.001990566669962e-06,
+      "loss": 1.6654,
+      "step": 1550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.000558606187502e-06,
+      "loss": 1.3507,
+      "step": 1551
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.999125733205541e-06,
+      "loss": 1.3964,
+      "step": 1552
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.997691948050904e-06,
+      "loss": 1.4805,
+      "step": 1553
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.996257251050633e-06,
+      "loss": 1.4782,
+      "step": 1554
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.994821642531968e-06,
+      "loss": 1.5154,
+      "step": 1555
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.993385122822364e-06,
+      "loss": 1.2983,
+      "step": 1556
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.99194769224948e-06,
+      "loss": 1.5056,
+      "step": 1557
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.990509351141184e-06,
+      "loss": 1.9269,
+      "step": 1558
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.989070099825551e-06,
+      "loss": 1.8205,
+      "step": 1559
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.987629938630866e-06,
+      "loss": 1.4983,
+      "step": 1560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.986188867885617e-06,
+      "loss": 1.5927,
+      "step": 1561
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.984746887918504e-06,
+      "loss": 1.3168,
+      "step": 1562
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.983303999058433e-06,
+      "loss": 1.3042,
+      "step": 1563
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.981860201634516e-06,
+      "loss": 1.2642,
+      "step": 1564
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.980415495976072e-06,
+      "loss": 1.3914,
+      "step": 1565
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97896988241263e-06,
+      "loss": 1.57,
+      "step": 1566
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.977523361273924e-06,
+      "loss": 1.2912,
+      "step": 1567
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.976075932889896e-06,
+      "loss": 1.4623,
+      "step": 1568
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.974627597590693e-06,
+      "loss": 1.4348,
+      "step": 1569
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.973178355706669e-06,
+      "loss": 1.5121,
+      "step": 1570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.971728207568389e-06,
+      "loss": 1.4713,
+      "step": 1571
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97027715350662e-06,
+      "loss": 1.5432,
+      "step": 1572
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.968825193852338e-06,
+      "loss": 1.2768,
+      "step": 1573
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.967372328936726e-06,
+      "loss": 1.3109,
+      "step": 1574
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.965918559091169e-06,
+      "loss": 1.7635,
+      "step": 1575
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.964463884647264e-06,
+      "loss": 1.5415,
+      "step": 1576
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.96300830593681e-06,
+      "loss": 1.5525,
+      "step": 1577
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.961551823291818e-06,
+      "loss": 1.672,
+      "step": 1578
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.9600944370445e-06,
+      "loss": 1.4372,
+      "step": 1579
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.958636147527275e-06,
+      "loss": 1.4864,
+      "step": 1580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.957176955072768e-06,
+      "loss": 1.3791,
+      "step": 1581
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.955716860013812e-06,
+      "loss": 1.3198,
+      "step": 1582
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.954255862683443e-06,
+      "loss": 1.4154,
+      "step": 1583
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.952793963414908e-06,
+      "loss": 1.2819,
+      "step": 1584
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.95133116254165e-06,
+      "loss": 1.4566,
+      "step": 1585
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.949867460397329e-06,
+      "loss": 1.3141,
+      "step": 1586
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.948402857315803e-06,
+      "loss": 1.3055,
+      "step": 1587
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.946937353631136e-06,
+      "loss": 1.5483,
+      "step": 1588
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.945470949677603e-06,
+      "loss": 1.7631,
+      "step": 1589
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.944003645789678e-06,
+      "loss": 1.4244,
+      "step": 1590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.942535442302045e-06,
+      "loss": 1.4748,
+      "step": 1591
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.941066339549586e-06,
+      "loss": 1.4668,
+      "step": 1592
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.9395963378674e-06,
+      "loss": 1.7404,
+      "step": 1593
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.938125437590778e-06,
+      "loss": 1.2417,
+      "step": 1594
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.936653639055225e-06,
+      "loss": 1.5241,
+      "step": 1595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.93518094259645e-06,
+      "loss": 1.6511,
+      "step": 1596
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.933707348550361e-06,
+      "loss": 1.4595,
+      "step": 1597
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.932232857253076e-06,
+      "loss": 1.4919,
+      "step": 1598
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.930757469040917e-06,
+      "loss": 1.381,
+      "step": 1599
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.929281184250407e-06,
+      "loss": 1.2841,
+      "step": 1600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.927804003218278e-06,
+      "loss": 1.3126,
+      "step": 1601
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.926325926281468e-06,
+      "loss": 1.2286,
+      "step": 1602
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.924846953777109e-06,
+      "loss": 1.6192,
+      "step": 1603
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.923367086042548e-06,
+      "loss": 1.5731,
+      "step": 1604
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.921886323415336e-06,
+      "loss": 1.5297,
+      "step": 1605
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.92040466623322e-06,
+      "loss": 1.5181,
+      "step": 1606
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.918922114834156e-06,
+      "loss": 1.6803,
+      "step": 1607
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.917438669556307e-06,
+      "loss": 1.1008,
+      "step": 1608
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.915954330738032e-06,
+      "loss": 1.6616,
+      "step": 1609
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.914469098717901e-06,
+      "loss": 1.566,
+      "step": 1610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.912982973834686e-06,
+      "loss": 1.7469,
+      "step": 1611
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.911495956427358e-06,
+      "loss": 1.7276,
+      "step": 1612
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.910008046835097e-06,
+      "loss": 1.4768,
+      "step": 1613
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.908519245397286e-06,
+      "loss": 1.3456,
+      "step": 1614
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.90702955245351e-06,
+      "loss": 1.5423,
+      "step": 1615
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.905538968343558e-06,
+      "loss": 1.5472,
+      "step": 1616
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.904047493407419e-06,
+      "loss": 1.8241,
+      "step": 1617
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.902555127985291e-06,
+      "loss": 1.3433,
+      "step": 1618
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.901061872417571e-06,
+      "loss": 1.5088,
+      "step": 1619
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89956772704486e-06,
+      "loss": 1.3253,
+      "step": 1620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.898072692207964e-06,
+      "loss": 1.5914,
+      "step": 1621
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.896576768247887e-06,
+      "loss": 1.4285,
+      "step": 1622
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89507995550584e-06,
+      "loss": 1.477,
+      "step": 1623
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.893582254323236e-06,
+      "loss": 1.3056,
+      "step": 1624
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89208366504169e-06,
+      "loss": 1.2437,
+      "step": 1625
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89058418800302e-06,
+      "loss": 1.8733,
+      "step": 1626
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.889083823549245e-06,
+      "loss": 1.3955,
+      "step": 1627
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.88758257202259e-06,
+      "loss": 1.5286,
+      "step": 1628
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.886080433765479e-06,
+      "loss": 1.7463,
+      "step": 1629
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.884577409120535e-06,
+      "loss": 1.3249,
+      "step": 1630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.883073498430596e-06,
+      "loss": 1.553,
+      "step": 1631
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.881568702038687e-06,
+      "loss": 1.5372,
+      "step": 1632
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.880063020288044e-06,
+      "loss": 1.5851,
+      "step": 1633
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.8785564535221e-06,
+      "loss": 1.6982,
+      "step": 1634
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.877049002084496e-06,
+      "loss": 1.3246,
+      "step": 1635
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.875540666319069e-06,
+      "loss": 1.3439,
+      "step": 1636
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.87403144656986e-06,
+      "loss": 1.7974,
+      "step": 1637
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.87252134318111e-06,
+      "loss": 1.1225,
+      "step": 1638
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.871010356497268e-06,
+      "loss": 1.5626,
+      "step": 1639
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.869498486862976e-06,
+      "loss": 1.8638,
+      "step": 1640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.867985734623079e-06,
+      "loss": 1.6113,
+      "step": 1641
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.866472100122629e-06,
+      "loss": 1.3439,
+      "step": 1642
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.864957583706873e-06,
+      "loss": 1.3527,
+      "step": 1643
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.863442185721264e-06,
+      "loss": 1.3408,
+      "step": 1644
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.86192590651145e-06,
+      "loss": 1.46,
+      "step": 1645
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.86040874642329e-06,
+      "loss": 1.4517,
+      "step": 1646
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.85889070580283e-06,
+      "loss": 1.2746,
+      "step": 1647
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.857371784996329e-06,
+      "loss": 1.3005,
+      "step": 1648
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.855851984350243e-06,
+      "loss": 1.2431,
+      "step": 1649
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.854331304211223e-06,
+      "loss": 1.478,
+      "step": 1650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.852809744926131e-06,
+      "loss": 1.3027,
+      "step": 1651
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.85128730684202e-06,
+      "loss": 1.2867,
+      "step": 1652
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.849763990306153e-06,
+      "loss": 1.5917,
+      "step": 1653
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.84823979566598e-06,
+      "loss": 1.3064,
+      "step": 1654
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.846714723269166e-06,
+      "loss": 1.3623,
+      "step": 1655
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.845188773463567e-06,
+      "loss": 1.4732,
+      "step": 1656
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.84366194659724e-06,
+      "loss": 1.2984,
+      "step": 1657
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.842134243018446e-06,
+      "loss": 1.3247,
+      "step": 1658
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.84060566307564e-06,
+      "loss": 1.8694,
+      "step": 1659
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.839076207117485e-06,
+      "loss": 1.4126,
+      "step": 1660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.837545875492835e-06,
+      "loss": 1.3752,
+      "step": 1661
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.836014668550752e-06,
+      "loss": 1.2948,
+      "step": 1662
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.83448258664049e-06,
+      "loss": 1.3243,
+      "step": 1663
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.832949630111508e-06,
+      "loss": 1.3907,
+      "step": 1664
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.831415799313463e-06,
+      "loss": 1.4576,
+      "step": 1665
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.82988109459621e-06,
+      "loss": 1.4646,
+      "step": 1666
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.828345516309805e-06,
+      "loss": 1.7235,
+      "step": 1667
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.826809064804503e-06,
+      "loss": 1.3404,
+      "step": 1668
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.825271740430756e-06,
+      "loss": 1.48,
+      "step": 1669
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.823733543539221e-06,
+      "loss": 1.3135,
+      "step": 1670
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.822194474480747e-06,
+      "loss": 1.807,
+      "step": 1671
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.820654533606383e-06,
+      "loss": 1.344,
+      "step": 1672
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.819113721267385e-06,
+      "loss": 1.4346,
+      "step": 1673
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.817572037815195e-06,
+      "loss": 1.3953,
+      "step": 1674
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.816029483601463e-06,
+      "loss": 1.7888,
+      "step": 1675
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.814486058978035e-06,
+      "loss": 1.4452,
+      "step": 1676
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.812941764296955e-06,
+      "loss": 1.6224,
+      "step": 1677
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.811396599910467e-06,
+      "loss": 1.4979,
+      "step": 1678
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.80985056617101e-06,
+      "loss": 1.5203,
+      "step": 1679
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.808303663431226e-06,
+      "loss": 1.8462,
+      "step": 1680
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.80675589204395e-06,
+      "loss": 1.6629,
+      "step": 1681
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.80520725236222e-06,
+      "loss": 1.5618,
+      "step": 1682
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.80365774473927e-06,
+      "loss": 1.466,
+      "step": 1683
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.802107369528529e-06,
+      "loss": 1.6859,
+      "step": 1684
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.80055612708363e-06,
+      "loss": 1.0945,
+      "step": 1685
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.7990040177584e-06,
+      "loss": 1.1346,
+      "step": 1686
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.797451041906863e-06,
+      "loss": 1.6313,
+      "step": 1687
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.795897199883243e-06,
+      "loss": 1.235,
+      "step": 1688
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79434249204196e-06,
+      "loss": 1.2041,
+      "step": 1689
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.792786918737632e-06,
+      "loss": 1.6253,
+      "step": 1690
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.791230480325076e-06,
+      "loss": 1.5902,
+      "step": 1691
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.789673177159302e-06,
+      "loss": 1.3528,
+      "step": 1692
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.788115009595522e-06,
+      "loss": 1.3255,
+      "step": 1693
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.786555977989142e-06,
+      "loss": 1.2763,
+      "step": 1694
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.784996082695766e-06,
+      "loss": 1.3898,
+      "step": 1695
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.7834353240712e-06,
+      "loss": 1.2596,
+      "step": 1696
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.781873702471433e-06,
+      "loss": 1.5341,
+      "step": 1697
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.780311218252667e-06,
+      "loss": 1.64,
+      "step": 1698
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.778747871771293e-06,
+      "loss": 1.6225,
+      "step": 1699
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.777183663383897e-06,
+      "loss": 1.581,
+      "step": 1700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.775618593447265e-06,
+      "loss": 1.2394,
+      "step": 1701
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.77405266231838e-06,
+      "loss": 1.282,
+      "step": 1702
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.772485870354417e-06,
+      "loss": 1.6343,
+      "step": 1703
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.770918217912753e-06,
+      "loss": 1.3583,
+      "step": 1704
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.769349705350959e-06,
+      "loss": 1.2282,
+      "step": 1705
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.767780333026798e-06,
+      "loss": 1.7633,
+      "step": 1706
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.766210101298239e-06,
+      "loss": 1.3949,
+      "step": 1707
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.764639010523433e-06,
+      "loss": 1.2894,
+      "step": 1708
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.763067061060742e-06,
+      "loss": 1.6193,
+      "step": 1709
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.761494253268712e-06,
+      "loss": 1.6411,
+      "step": 1710
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.759920587506093e-06,
+      "loss": 1.4676,
+      "step": 1711
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.758346064131824e-06,
+      "loss": 1.3114,
+      "step": 1712
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.756770683505044e-06,
+      "loss": 1.7217,
+      "step": 1713
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.755194445985087e-06,
+      "loss": 1.4969,
+      "step": 1714
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.753617351931483e-06,
+      "loss": 1.7592,
+      "step": 1715
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.752039401703954e-06,
+      "loss": 1.4756,
+      "step": 1716
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.750460595662418e-06,
+      "loss": 1.4304,
+      "step": 1717
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.748880934166992e-06,
+      "loss": 1.6942,
+      "step": 1718
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.747300417577987e-06,
+      "loss": 1.6808,
+      "step": 1719
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.745719046255904e-06,
+      "loss": 1.5322,
+      "step": 1720
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.744136820561446e-06,
+      "loss": 1.627,
+      "step": 1721
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.742553740855507e-06,
+      "loss": 1.5255,
+      "step": 1722
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.740969807499175e-06,
+      "loss": 1.3152,
+      "step": 1723
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.739385020853735e-06,
+      "loss": 1.3271,
+      "step": 1724
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.737799381280667e-06,
+      "loss": 1.7658,
+      "step": 1725
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.736212889141642e-06,
+      "loss": 1.4844,
+      "step": 1726
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.73462554479853e-06,
+      "loss": 1.6017,
+      "step": 1727
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.733037348613391e-06,
+      "loss": 1.5008,
+      "step": 1728
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.731448300948484e-06,
+      "loss": 1.3027,
+      "step": 1729
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.729858402166259e-06,
+      "loss": 1.4338,
+      "step": 1730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.728267652629357e-06,
+      "loss": 1.5026,
+      "step": 1731
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.726676052700623e-06,
+      "loss": 1.3177,
+      "step": 1732
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.725083602743084e-06,
+      "loss": 1.3581,
+      "step": 1733
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.723490303119973e-06,
+      "loss": 1.5453,
+      "step": 1734
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.721896154194704e-06,
+      "loss": 1.3147,
+      "step": 1735
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.720301156330896e-06,
+      "loss": 1.2917,
+      "step": 1736
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.718705309892354e-06,
+      "loss": 1.3908,
+      "step": 1737
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.717108615243081e-06,
+      "loss": 1.5283,
+      "step": 1738
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.715511072747274e-06,
+      "loss": 1.7886,
+      "step": 1739
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.713912682769316e-06,
+      "loss": 1.4944,
+      "step": 1740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.712313445673792e-06,
+      "loss": 1.2272,
+      "step": 1741
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.710713361825478e-06,
+      "loss": 1.526,
+      "step": 1742
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.709112431589338e-06,
+      "loss": 1.6585,
+      "step": 1743
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.707510655330536e-06,
+      "loss": 1.617,
+      "step": 1744
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.705908033414426e-06,
+      "loss": 1.4361,
+      "step": 1745
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.704304566206553e-06,
+      "loss": 1.6276,
+      "step": 1746
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.702700254072658e-06,
+      "loss": 1.3722,
+      "step": 1747
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.701095097378674e-06,
+      "loss": 1.4782,
+      "step": 1748
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.699489096490726e-06,
+      "loss": 1.2547,
+      "step": 1749
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.697882251775129e-06,
+      "loss": 1.2514,
+      "step": 1750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.696274563598395e-06,
+      "loss": 1.5904,
+      "step": 1751
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.694666032327226e-06,
+      "loss": 1.6928,
+      "step": 1752
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.693056658328517e-06,
+      "loss": 1.5265,
+      "step": 1753
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.691446441969356e-06,
+      "loss": 1.4722,
+      "step": 1754
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.689835383617022e-06,
+      "loss": 1.5212,
+      "step": 1755
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.688223483638985e-06,
+      "loss": 1.6317,
+      "step": 1756
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.686610742402908e-06,
+      "loss": 1.5301,
+      "step": 1757
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.684997160276647e-06,
+      "loss": 1.6666,
+      "step": 1758
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.683382737628247e-06,
+      "loss": 1.4591,
+      "step": 1759
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.681767474825949e-06,
+      "loss": 1.5032,
+      "step": 1760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.680151372238181e-06,
+      "loss": 1.3883,
+      "step": 1761
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.678534430233567e-06,
+      "loss": 1.8188,
+      "step": 1762
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.67691664918092e-06,
+      "loss": 1.4032,
+      "step": 1763
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.675298029449241e-06,
+      "loss": 1.5268,
+      "step": 1764
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.673678571407731e-06,
+      "loss": 1.4005,
+      "step": 1765
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.672058275425773e-06,
+      "loss": 1.3187,
+      "step": 1766
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.670437141872945e-06,
+      "loss": 1.7154,
+      "step": 1767
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.66881517111902e-06,
+      "loss": 1.4475,
+      "step": 1768
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.667192363533958e-06,
+      "loss": 1.3923,
+      "step": 1769
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.665568719487905e-06,
+      "loss": 1.2417,
+      "step": 1770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.663944239351208e-06,
+      "loss": 1.4526,
+      "step": 1771
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.662318923494397e-06,
+      "loss": 1.8005,
+      "step": 1772
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.660692772288195e-06,
+      "loss": 1.5868,
+      "step": 1773
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.659065786103518e-06,
+      "loss": 1.5722,
+      "step": 1774
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.657437965311468e-06,
+      "loss": 1.2783,
+      "step": 1775
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.655809310283341e-06,
+      "loss": 1.0151,
+      "step": 1776
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.65417982139062e-06,
+      "loss": 1.7598,
+      "step": 1777
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.652549499004983e-06,
+      "loss": 2.0153,
+      "step": 1778
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.650918343498294e-06,
+      "loss": 1.3773,
+      "step": 1779
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.649286355242606e-06,
+      "loss": 1.1881,
+      "step": 1780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.647653534610166e-06,
+      "loss": 1.8038,
+      "step": 1781
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.646019881973409e-06,
+      "loss": 1.6256,
+      "step": 1782
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.64438539770496e-06,
+      "loss": 1.6664,
+      "step": 1783
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.642750082177633e-06,
+      "loss": 1.852,
+      "step": 1784
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.64111393576443e-06,
+      "loss": 1.4515,
+      "step": 1785
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.63947695883855e-06,
+      "loss": 1.5275,
+      "step": 1786
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.63783915177337e-06,
+      "loss": 1.988,
+      "step": 1787
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.636200514942466e-06,
+      "loss": 1.4181,
+      "step": 1788
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.634561048719599e-06,
+      "loss": 1.2964,
+      "step": 1789
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.63292075347872e-06,
+      "loss": 1.2898,
+      "step": 1790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.631279629593968e-06,
+      "loss": 1.2998,
+      "step": 1791
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.62963767743967e-06,
+      "loss": 1.4063,
+      "step": 1792
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.627994897390348e-06,
+      "loss": 1.5821,
+      "step": 1793
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.626351289820707e-06,
+      "loss": 1.3783,
+      "step": 1794
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.624706855105643e-06,
+      "loss": 1.6404,
+      "step": 1795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.623061593620236e-06,
+      "loss": 1.6062,
+      "step": 1796
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.621415505739766e-06,
+      "loss": 1.1878,
+      "step": 1797
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.619768591839685e-06,
+      "loss": 1.4147,
+      "step": 1798
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.618120852295652e-06,
+      "loss": 1.1164,
+      "step": 1799
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.616472287483499e-06,
+      "loss": 1.4567,
+      "step": 1800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.614822897779253e-06,
+      "loss": 1.4715,
+      "step": 1801
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.61317268355913e-06,
+      "loss": 1.481,
+      "step": 1802
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.611521645199532e-06,
+      "loss": 1.3918,
+      "step": 1803
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.609869783077048e-06,
+      "loss": 1.2389,
+      "step": 1804
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.608217097568456e-06,
+      "loss": 1.7604,
+      "step": 1805
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.606563589050722e-06,
+      "loss": 1.4018,
+      "step": 1806
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.604909257901003e-06,
+      "loss": 1.5706,
+      "step": 1807
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.603254104496637e-06,
+      "loss": 1.7512,
+      "step": 1808
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.601598129215154e-06,
+      "loss": 1.5633,
+      "step": 1809
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.59994133243427e-06,
+      "loss": 1.5179,
+      "step": 1810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.59828371453189e-06,
+      "loss": 1.5028,
+      "step": 1811
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.596625275886103e-06,
+      "loss": 1.702,
+      "step": 1812
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.594966016875189e-06,
+      "loss": 1.3899,
+      "step": 1813
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.593305937877614e-06,
+      "loss": 1.3858,
+      "step": 1814
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.591645039272028e-06,
+      "loss": 1.4955,
+      "step": 1815
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.589983321437271e-06,
+      "loss": 1.8671,
+      "step": 1816
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.588320784752372e-06,
+      "loss": 1.4258,
+      "step": 1817
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.586657429596542e-06,
+      "loss": 1.4522,
+      "step": 1818
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.58499325634918e-06,
+      "loss": 1.8705,
+      "step": 1819
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.583328265389872e-06,
+      "loss": 1.5051,
+      "step": 1820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.581662457098391e-06,
+      "loss": 1.3598,
+      "step": 1821
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.5799958318547e-06,
+      "loss": 1.5079,
+      "step": 1822
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.578328390038942e-06,
+      "loss": 1.6353,
+      "step": 1823
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.576660132031445e-06,
+      "loss": 1.5287,
+      "step": 1824
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.574991058212731e-06,
+      "loss": 1.4207,
+      "step": 1825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.573321168963507e-06,
+      "loss": 1.7104,
+      "step": 1826
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.571650464664658e-06,
+      "loss": 1.4697,
+      "step": 1827
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.56997894569726e-06,
+      "loss": 1.5491,
+      "step": 1828
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.568306612442579e-06,
+      "loss": 1.4234,
+      "step": 1829
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.566633465282058e-06,
+      "loss": 1.425,
+      "step": 1830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.564959504597332e-06,
+      "loss": 1.4726,
+      "step": 1831
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.563284730770222e-06,
+      "loss": 1.3477,
+      "step": 1832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.561609144182728e-06,
+      "loss": 1.6387,
+      "step": 1833
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.559932745217043e-06,
+      "loss": 1.7656,
+      "step": 1834
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.558255534255541e-06,
+      "loss": 1.3511,
+      "step": 1835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.556577511680781e-06,
+      "loss": 1.6777,
+      "step": 1836
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.554898677875509e-06,
+      "loss": 1.3261,
+      "step": 1837
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.553219033222657e-06,
+      "loss": 1.419,
+      "step": 1838
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.551538578105338e-06,
+      "loss": 1.6423,
+      "step": 1839
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.549857312906852e-06,
+      "loss": 1.4158,
+      "step": 1840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.548175238010684e-06,
+      "loss": 1.5682,
+      "step": 1841
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.546492353800504e-06,
+      "loss": 1.4607,
+      "step": 1842
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.544808660660168e-06,
+      "loss": 1.6186,
+      "step": 1843
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.543124158973712e-06,
+      "loss": 1.6235,
+      "step": 1844
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.541438849125359e-06,
+      "loss": 1.3403,
+      "step": 1845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.539752731499518e-06,
+      "loss": 1.44,
+      "step": 1846
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.53806580648078e-06,
+      "loss": 1.7357,
+      "step": 1847
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.53637807445392e-06,
+      "loss": 1.6924,
+      "step": 1848
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.534689535803899e-06,
+      "loss": 1.9203,
+      "step": 1849
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.533000190915862e-06,
+      "loss": 1.453,
+      "step": 1850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.531310040175135e-06,
+      "loss": 1.522,
+      "step": 1851
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.529619083967229e-06,
+      "loss": 1.6825,
+      "step": 1852
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.52792732267784e-06,
+      "loss": 1.7457,
+      "step": 1853
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.52623475669285e-06,
+      "loss": 1.2543,
+      "step": 1854
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.524541386398318e-06,
+      "loss": 1.4773,
+      "step": 1855
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.522847212180489e-06,
+      "loss": 1.4177,
+      "step": 1856
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.521152234425793e-06,
+      "loss": 1.6876,
+      "step": 1857
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.519456453520847e-06,
+      "loss": 1.4398,
+      "step": 1858
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.51775986985244e-06,
+      "loss": 1.7195,
+      "step": 1859
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.516062483807556e-06,
+      "loss": 1.5538,
+      "step": 1860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.514364295773354e-06,
+      "loss": 1.5168,
+      "step": 1861
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.512665306137182e-06,
+      "loss": 1.571,
+      "step": 1862
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.510965515286564e-06,
+      "loss": 1.2588,
+      "step": 1863
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.509264923609215e-06,
+      "loss": 1.3198,
+      "step": 1864
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.507563531493022e-06,
+      "loss": 1.5707,
+      "step": 1865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.505861339326065e-06,
+      "loss": 1.3525,
+      "step": 1866
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.504158347496602e-06,
+      "loss": 1.1443,
+      "step": 1867
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.502454556393071e-06,
+      "loss": 1.6538,
+      "step": 1868
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.500749966404098e-06,
+      "loss": 1.5353,
+      "step": 1869
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.499044577918484e-06,
+      "loss": 1.2932,
+      "step": 1870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.497338391325223e-06,
+      "loss": 1.6323,
+      "step": 1871
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.495631407013475e-06,
+      "loss": 1.7081,
+      "step": 1872
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.4939236253726e-06,
+      "loss": 1.6397,
+      "step": 1873
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.492215046792127e-06,
+      "loss": 1.3822,
+      "step": 1874
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.49050567166177e-06,
+      "loss": 1.3378,
+      "step": 1875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.488795500371427e-06,
+      "loss": 1.5667,
+      "step": 1876
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.487084533311176e-06,
+      "loss": 1.4551,
+      "step": 1877
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.485372770871277e-06,
+      "loss": 1.038,
+      "step": 1878
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.48366021344217e-06,
+      "loss": 1.5566,
+      "step": 1879
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.48194686141448e-06,
+      "loss": 1.3404,
+      "step": 1880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.480232715179004e-06,
+      "loss": 1.7366,
+      "step": 1881
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.478517775126733e-06,
+      "loss": 1.6839,
+      "step": 1882
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.476802041648832e-06,
+      "loss": 1.5205,
+      "step": 1883
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.47508551513665e-06,
+      "loss": 1.1136,
+      "step": 1884
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.473368195981708e-06,
+      "loss": 1.584,
+      "step": 1885
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.47165008457572e-06,
+      "loss": 1.418,
+      "step": 1886
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.469931181310573e-06,
+      "loss": 1.5149,
+      "step": 1887
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.468211486578337e-06,
+      "loss": 2.0012,
+      "step": 1888
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.466491000771264e-06,
+      "loss": 1.5816,
+      "step": 1889
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.464769724281782e-06,
+      "loss": 1.6476,
+      "step": 1890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.463047657502504e-06,
+      "loss": 1.3683,
+      "step": 1891
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.461324800826221e-06,
+      "loss": 1.7172,
+      "step": 1892
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.459601154645906e-06,
+      "loss": 1.2548,
+      "step": 1893
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.457876719354708e-06,
+      "loss": 1.3603,
+      "step": 1894
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.456151495345962e-06,
+      "loss": 1.514,
+      "step": 1895
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.454425483013176e-06,
+      "loss": 1.6652,
+      "step": 1896
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.452698682750042e-06,
+      "loss": 1.2444,
+      "step": 1897
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.450971094950433e-06,
+      "loss": 1.2813,
+      "step": 1898
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.449242720008399e-06,
+      "loss": 1.4036,
+      "step": 1899
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.447513558318171e-06,
+      "loss": 1.5482,
+      "step": 1900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.445783610274157e-06,
+      "loss": 1.352,
+      "step": 1901
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.444052876270947e-06,
+      "loss": 1.5287,
+      "step": 1902
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.442321356703309e-06,
+      "loss": 1.4583,
+      "step": 1903
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.44058905196619e-06,
+      "loss": 1.2446,
+      "step": 1904
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.438855962454717e-06,
+      "loss": 1.4476,
+      "step": 1905
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.437122088564197e-06,
+      "loss": 1.491,
+      "step": 1906
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.435387430690114e-06,
+      "loss": 1.6469,
+      "step": 1907
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43365198922813e-06,
+      "loss": 1.7366,
+      "step": 1908
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.431915764574087e-06,
+      "loss": 1.4312,
+      "step": 1909
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43017875712401e-06,
+      "loss": 1.5457,
+      "step": 1910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.428440967274092e-06,
+      "loss": 1.3536,
+      "step": 1911
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.426702395420715e-06,
+      "loss": 1.3098,
+      "step": 1912
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.424963041960434e-06,
+      "loss": 1.2387,
+      "step": 1913
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.423222907289981e-06,
+      "loss": 1.5588,
+      "step": 1914
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.421481991806272e-06,
+      "loss": 1.5495,
+      "step": 1915
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.419740295906397e-06,
+      "loss": 1.5148,
+      "step": 1916
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.417997819987624e-06,
+      "loss": 1.3865,
+      "step": 1917
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.4162545644474e-06,
+      "loss": 1.3497,
+      "step": 1918
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.414510529683347e-06,
+      "loss": 1.4439,
+      "step": 1919
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.412765716093273e-06,
+      "loss": 1.318,
+      "step": 1920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.41102012407515e-06,
+      "loss": 1.2018,
+      "step": 1921
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.409273754027141e-06,
+      "loss": 1.563,
+      "step": 1922
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.40752660634758e-06,
+      "loss": 1.4329,
+      "step": 1923
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.405778681434977e-06,
+      "loss": 1.4248,
+      "step": 1924
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.404029979688024e-06,
+      "loss": 1.3132,
+      "step": 1925
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.402280501505587e-06,
+      "loss": 1.2561,
+      "step": 1926
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.40053024728671e-06,
+      "loss": 1.5641,
+      "step": 1927
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.398779217430612e-06,
+      "loss": 1.6223,
+      "step": 1928
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.39702741233669e-06,
+      "loss": 1.132,
+      "step": 1929
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.395274832404524e-06,
+      "loss": 1.7801,
+      "step": 1930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.39352147803386e-06,
+      "loss": 1.5791,
+      "step": 1931
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.391767349624627e-06,
+      "loss": 1.7737,
+      "step": 1932
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.390012447576931e-06,
+      "loss": 1.6129,
+      "step": 1933
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.388256772291052e-06,
+      "loss": 1.6782,
+      "step": 1934
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.386500324167446e-06,
+      "loss": 1.5297,
+      "step": 1935
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.384743103606748e-06,
+      "loss": 1.1324,
+      "step": 1936
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.382985111009768e-06,
+      "loss": 1.3744,
+      "step": 1937
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.38122634677749e-06,
+      "loss": 1.3844,
+      "step": 1938
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.379466811311078e-06,
+      "loss": 1.2488,
+      "step": 1939
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.377706505011869e-06,
+      "loss": 1.493,
+      "step": 1940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.375945428281376e-06,
+      "loss": 1.4857,
+      "step": 1941
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.374183581521288e-06,
+      "loss": 1.4166,
+      "step": 1942
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.372420965133473e-06,
+      "loss": 1.4883,
+      "step": 1943
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.370657579519968e-06,
+      "loss": 1.4904,
+      "step": 1944
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.368893425082991e-06,
+      "loss": 1.2572,
+      "step": 1945
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.367128502224931e-06,
+      "loss": 1.6607,
+      "step": 1946
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.365362811348358e-06,
+      "loss": 1.4838,
+      "step": 1947
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.36359635285601e-06,
+      "loss": 1.4923,
+      "step": 1948
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.361829127150807e-06,
+      "loss": 1.3001,
+      "step": 1949
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.36006113463584e-06,
+      "loss": 1.6666,
+      "step": 1950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.358292375714375e-06,
+      "loss": 1.5616,
+      "step": 1951
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.356522850789852e-06,
+      "loss": 1.8263,
+      "step": 1952
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.354752560265891e-06,
+      "loss": 1.6579,
+      "step": 1953
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.35298150454628e-06,
+      "loss": 1.3941,
+      "step": 1954
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.351209684034987e-06,
+      "loss": 1.3929,
+      "step": 1955
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.34943709913615e-06,
+      "loss": 1.4489,
+      "step": 1956
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.347663750254082e-06,
+      "loss": 1.4467,
+      "step": 1957
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.345889637793276e-06,
+      "loss": 1.4903,
+      "step": 1958
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.344114762158391e-06,
+      "loss": 1.4443,
+      "step": 1959
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.342339123754265e-06,
+      "loss": 1.4746,
+      "step": 1960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.340562722985906e-06,
+      "loss": 1.6133,
+      "step": 1961
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.338785560258502e-06,
+      "loss": 1.5781,
+      "step": 1962
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.337007635977411e-06,
+      "loss": 1.4841,
+      "step": 1963
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.335228950548164e-06,
+      "loss": 1.8979,
+      "step": 1964
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.333449504376468e-06,
+      "loss": 1.4075,
+      "step": 1965
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.331669297868201e-06,
+      "loss": 1.6647,
+      "step": 1966
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.329888331429417e-06,
+      "loss": 1.4056,
+      "step": 1967
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.328106605466339e-06,
+      "loss": 1.7261,
+      "step": 1968
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.32632412038537e-06,
+      "loss": 1.6041,
+      "step": 1969
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.324540876593082e-06,
+      "loss": 1.4539,
+      "step": 1970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.32275687449622e-06,
+      "loss": 1.5469,
+      "step": 1971
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.320972114501698e-06,
+      "loss": 1.5539,
+      "step": 1972
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.319186597016612e-06,
+      "loss": 1.5452,
+      "step": 1973
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.317400322448227e-06,
+      "loss": 1.3252,
+      "step": 1974
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.315613291203977e-06,
+      "loss": 1.7249,
+      "step": 1975
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.313825503691473e-06,
+      "loss": 1.7425,
+      "step": 1976
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.312036960318494e-06,
+      "loss": 1.5491,
+      "step": 1977
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.310247661492998e-06,
+      "loss": 1.4008,
+      "step": 1978
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.308457607623109e-06,
+      "loss": 1.1367,
+      "step": 1979
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.306666799117129e-06,
+      "loss": 1.5072,
+      "step": 1980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.304875236383525e-06,
+      "loss": 1.4168,
+      "step": 1981
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.30308291983094e-06,
+      "loss": 1.2749,
+      "step": 1982
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.301289849868193e-06,
+      "loss": 1.4456,
+      "step": 1983
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.299496026904268e-06,
+      "loss": 1.668,
+      "step": 1984
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.297701451348324e-06,
+      "loss": 1.5817,
+      "step": 1985
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.29590612360969e-06,
+      "loss": 1.448,
+      "step": 1986
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.29411004409787e-06,
+      "loss": 1.3491,
+      "step": 1987
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.292313213222534e-06,
+      "loss": 1.4595,
+      "step": 1988
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.290515631393528e-06,
+      "loss": 0.9418,
+      "step": 1989
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.28871729902087e-06,
+      "loss": 1.204,
+      "step": 1990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.286918216514745e-06,
+      "loss": 1.4062,
+      "step": 1991
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.285118384285511e-06,
+      "loss": 1.3102,
+      "step": 1992
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.283317802743699e-06,
+      "loss": 1.7167,
+      "step": 1993
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.281516472300007e-06,
+      "loss": 1.8095,
+      "step": 1994
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.279714393365307e-06,
+      "loss": 1.431,
+      "step": 1995
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.277911566350641e-06,
+      "loss": 1.344,
+      "step": 1996
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.276107991667219e-06,
+      "loss": 1.4505,
+      "step": 1997
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.274303669726427e-06,
+      "loss": 1.6298,
+      "step": 1998
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.272498600939815e-06,
+      "loss": 1.4743,
+      "step": 1999
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.270692785719109e-06,
+      "loss": 1.855,
+      "step": 2000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.268886224476202e-06,
+      "loss": 1.5937,
+      "step": 2001
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.26707891762316e-06,
+      "loss": 1.2077,
+      "step": 2002
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.265270865572213e-06,
+      "loss": 1.5404,
+      "step": 2003
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.263462068735766e-06,
+      "loss": 1.3129,
+      "step": 2004
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.261652527526397e-06,
+      "loss": 1.3993,
+      "step": 2005
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.259842242356846e-06,
+      "loss": 1.3011,
+      "step": 2006
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.258031213640027e-06,
+      "loss": 1.7253,
+      "step": 2007
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.256219441789023e-06,
+      "loss": 1.3861,
+      "step": 2008
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.254406927217087e-06,
+      "loss": 1.3472,
+      "step": 2009
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.252593670337641e-06,
+      "loss": 1.4588,
+      "step": 2010
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.250779671564277e-06,
+      "loss": 1.5505,
+      "step": 2011
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.248964931310752e-06,
+      "loss": 1.2845,
+      "step": 2012
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.247149449991002e-06,
+      "loss": 1.5293,
+      "step": 2013
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.245333228019117e-06,
+      "loss": 1.2944,
+      "step": 2014
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.243516265809373e-06,
+      "loss": 1.4068,
+      "step": 2015
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.241698563776201e-06,
+      "loss": 1.4784,
+      "step": 2016
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.239880122334208e-06,
+      "loss": 1.5638,
+      "step": 2017
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.238060941898168e-06,
+      "loss": 1.5122,
+      "step": 2018
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.236241022883023e-06,
+      "loss": 1.3516,
+      "step": 2019
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.234420365703883e-06,
+      "loss": 1.3169,
+      "step": 2020
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.232598970776027e-06,
+      "loss": 1.3298,
+      "step": 2021
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.230776838514905e-06,
+      "loss": 1.6717,
+      "step": 2022
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.228953969336131e-06,
+      "loss": 1.3142,
+      "step": 2023
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.22713036365549e-06,
+      "loss": 1.3271,
+      "step": 2024
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.225306021888931e-06,
+      "loss": 1.3597,
+      "step": 2025
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.223480944452575e-06,
+      "loss": 1.5151,
+      "step": 2026
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.22165513176271e-06,
+      "loss": 1.2166,
+      "step": 2027
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.219828584235792e-06,
+      "loss": 1.1427,
+      "step": 2028
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.218001302288442e-06,
+      "loss": 1.5457,
+      "step": 2029
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.216173286337449e-06,
+      "loss": 1.422,
+      "step": 2030
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.214344536799774e-06,
+      "loss": 1.7567,
+      "step": 2031
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.21251505409254e-06,
+      "loss": 1.3202,
+      "step": 2032
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.210684838633039e-06,
+      "loss": 1.32,
+      "step": 2033
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.208853890838733e-06,
+      "loss": 1.2185,
+      "step": 2034
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.207022211127244e-06,
+      "loss": 1.7586,
+      "step": 2035
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.205189799916369e-06,
+      "loss": 1.6928,
+      "step": 2036
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.20335665762407e-06,
+      "loss": 1.6029,
+      "step": 2037
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.201522784668467e-06,
+      "loss": 1.6174,
+      "step": 2038
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.19968818146786e-06,
+      "loss": 1.4383,
+      "step": 2039
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.197852848440704e-06,
+      "loss": 1.4028,
+      "step": 2040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.196016786005632e-06,
+      "loss": 1.4226,
+      "step": 2041
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.19417999458143e-06,
+      "loss": 1.2442,
+      "step": 2042
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.192342474587062e-06,
+      "loss": 1.6868,
+      "step": 2043
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.190504226441654e-06,
+      "loss": 1.5422,
+      "step": 2044
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.188665250564497e-06,
+      "loss": 1.3963,
+      "step": 2045
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.186825547375043e-06,
+      "loss": 1.2307,
+      "step": 2046
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.184985117292922e-06,
+      "loss": 1.2161,
+      "step": 2047
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.183143960737922e-06,
+      "loss": 1.7009,
+      "step": 2048
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.181302078129998e-06,
+      "loss": 1.429,
+      "step": 2049
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.179459469889269e-06,
+      "loss": 1.4156,
+      "step": 2050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.17761613643602e-06,
+      "loss": 1.3827,
+      "step": 2051
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.175772078190706e-06,
+      "loss": 1.3769,
+      "step": 2052
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.173927295573943e-06,
+      "loss": 1.5206,
+      "step": 2053
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.172081789006509e-06,
+      "loss": 1.3731,
+      "step": 2054
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.170235558909357e-06,
+      "loss": 1.0634,
+      "step": 2055
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.168388605703594e-06,
+      "loss": 1.2265,
+      "step": 2056
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.166540929810502e-06,
+      "loss": 1.3529,
+      "step": 2057
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.164692531651518e-06,
+      "loss": 1.4941,
+      "step": 2058
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.162843411648256e-06,
+      "loss": 1.2688,
+      "step": 2059
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.16099357022248e-06,
+      "loss": 1.41,
+      "step": 2060
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.159143007796127e-06,
+      "loss": 1.4805,
+      "step": 2061
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.157291724791301e-06,
+      "loss": 1.3321,
+      "step": 2062
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.155439721630265e-06,
+      "loss": 1.2443,
+      "step": 2063
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.153586998735448e-06,
+      "loss": 1.7333,
+      "step": 2064
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.151733556529441e-06,
+      "loss": 1.4128,
+      "step": 2065
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.149879395435004e-06,
+      "loss": 1.4525,
+      "step": 2066
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.148024515875057e-06,
+      "loss": 1.4786,
+      "step": 2067
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.146168918272688e-06,
+      "loss": 1.5576,
+      "step": 2068
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.144312603051141e-06,
+      "loss": 1.4314,
+      "step": 2069
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.14245557063383e-06,
+      "loss": 1.3723,
+      "step": 2070
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.140597821444334e-06,
+      "loss": 1.3658,
+      "step": 2071
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.138739355906389e-06,
+      "loss": 1.6187,
+      "step": 2072
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.136880174443901e-06,
+      "loss": 1.3846,
+      "step": 2073
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.135020277480933e-06,
+      "loss": 1.3826,
+      "step": 2074
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.13315966544172e-06,
+      "loss": 1.2642,
+      "step": 2075
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.131298338750648e-06,
+      "loss": 1.5242,
+      "step": 2076
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.129436297832275e-06,
+      "loss": 1.3699,
+      "step": 2077
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.127573543111325e-06,
+      "loss": 1.6286,
+      "step": 2078
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.12571007501267e-06,
+      "loss": 1.6405,
+      "step": 2079
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.12384589396136e-06,
+      "loss": 1.4915,
+      "step": 2080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.121981000382602e-06,
+      "loss": 1.5032,
+      "step": 2081
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.120115394701763e-06,
+      "loss": 1.5535,
+      "step": 2082
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.118249077344373e-06,
+      "loss": 1.3262,
+      "step": 2083
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.11638204873613e-06,
+      "loss": 1.3622,
+      "step": 2084
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.114514309302888e-06,
+      "loss": 1.5905,
+      "step": 2085
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.112645859470665e-06,
+      "loss": 1.6535,
+      "step": 2086
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.110776699665643e-06,
+      "loss": 1.3792,
+      "step": 2087
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.108906830314165e-06,
+      "loss": 1.5072,
+      "step": 2088
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.10703625184273e-06,
+      "loss": 1.4369,
+      "step": 2089
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.105164964678009e-06,
+      "loss": 1.4133,
+      "step": 2090
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.103292969246829e-06,
+      "loss": 1.6834,
+      "step": 2091
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.101420265976179e-06,
+      "loss": 1.2122,
+      "step": 2092
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.099546855293206e-06,
+      "loss": 1.5178,
+      "step": 2093
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.097672737625226e-06,
+      "loss": 1.2272,
+      "step": 2094
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.09579791339971e-06,
+      "loss": 1.656,
+      "step": 2095
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.093922383044293e-06,
+      "loss": 1.2731,
+      "step": 2096
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.09204614698677e-06,
+      "loss": 1.5215,
+      "step": 2097
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.090169205655098e-06,
+      "loss": 1.6693,
+      "step": 2098
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.088291559477395e-06,
+      "loss": 1.6161,
+      "step": 2099
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.086413208881938e-06,
+      "loss": 1.4219,
+      "step": 2100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.084534154297165e-06,
+      "loss": 1.2408,
+      "step": 2101
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.082654396151676e-06,
+      "loss": 1.3775,
+      "step": 2102
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.080773934874231e-06,
+      "loss": 1.648,
+      "step": 2103
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.07889277089375e-06,
+      "loss": 1.5253,
+      "step": 2104
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.07701090463931e-06,
+      "loss": 1.893,
+      "step": 2105
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.075128336540157e-06,
+      "loss": 1.2422,
+      "step": 2106
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.07324506702569e-06,
+      "loss": 1.2709,
+      "step": 2107
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.071361096525468e-06,
+      "loss": 1.2784,
+      "step": 2108
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.069476425469212e-06,
+      "loss": 1.5689,
+      "step": 2109
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.067591054286803e-06,
+      "loss": 1.3526,
+      "step": 2110
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.065704983408281e-06,
+      "loss": 1.314,
+      "step": 2111
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.063818213263846e-06,
+      "loss": 1.6118,
+      "step": 2112
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.061930744283855e-06,
+      "loss": 1.6178,
+      "step": 2113
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.060042576898828e-06,
+      "loss": 1.3434,
+      "step": 2114
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.058153711539444e-06,
+      "loss": 1.3563,
+      "step": 2115
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.05626414863654e-06,
+      "loss": 1.7057,
+      "step": 2116
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.054373888621108e-06,
+      "loss": 1.4848,
+      "step": 2117
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.052482931924307e-06,
+      "loss": 1.3054,
+      "step": 2118
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.05059127897745e-06,
+      "loss": 1.4728,
+      "step": 2119
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.04869893021201e-06,
+      "loss": 1.2228,
+      "step": 2120
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.046805886059619e-06,
+      "loss": 1.0723,
+      "step": 2121
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.044912146952066e-06,
+      "loss": 1.5824,
+      "step": 2122
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.043017713321301e-06,
+      "loss": 1.6749,
+      "step": 2123
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.041122585599428e-06,
+      "loss": 1.2126,
+      "step": 2124
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.039226764218718e-06,
+      "loss": 1.4616,
+      "step": 2125
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.037330249611591e-06,
+      "loss": 1.4265,
+      "step": 2126
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.035433042210628e-06,
+      "loss": 1.6929,
+      "step": 2127
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.03353514244857e-06,
+      "loss": 1.4663,
+      "step": 2128
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.031636550758313e-06,
+      "loss": 1.6489,
+      "step": 2129
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.029737267572915e-06,
+      "loss": 1.6515,
+      "step": 2130
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.02783729332559e-06,
+      "loss": 1.3377,
+      "step": 2131
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.025936628449706e-06,
+      "loss": 1.3859,
+      "step": 2132
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.024035273378792e-06,
+      "loss": 1.4345,
+      "step": 2133
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.022133228546536e-06,
+      "loss": 1.5678,
+      "step": 2134
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.020230494386779e-06,
+      "loss": 1.2314,
+      "step": 2135
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.018327071333521e-06,
+      "loss": 1.9491,
+      "step": 2136
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.016422959820922e-06,
+      "loss": 1.3499,
+      "step": 2137
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.014518160283293e-06,
+      "loss": 1.3622,
+      "step": 2138
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.01261267315511e-06,
+      "loss": 1.3711,
+      "step": 2139
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.010706498870997e-06,
+      "loss": 1.6346,
+      "step": 2140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.008799637865741e-06,
+      "loss": 1.36,
+      "step": 2141
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.006892090574286e-06,
+      "loss": 1.2494,
+      "step": 2142
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.004983857431726e-06,
+      "loss": 1.5078,
+      "step": 2143
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.003074938873316e-06,
+      "loss": 1.3562,
+      "step": 2144
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.00116533533447e-06,
+      "loss": 1.4471,
+      "step": 2145
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.999255047250754e-06,
+      "loss": 1.5028,
+      "step": 2146
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.997344075057891e-06,
+      "loss": 1.412,
+      "step": 2147
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.99543241919176e-06,
+      "loss": 1.4015,
+      "step": 2148
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.993520080088397e-06,
+      "loss": 1.6482,
+      "step": 2149
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.991607058183995e-06,
+      "loss": 1.1622,
+      "step": 2150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.989693353914895e-06,
+      "loss": 1.3937,
+      "step": 2151
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.987778967717607e-06,
+      "loss": 1.4391,
+      "step": 2152
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.985863900028785e-06,
+      "loss": 1.3877,
+      "step": 2153
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.983948151285242e-06,
+      "loss": 1.1203,
+      "step": 2154
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.982031721923949e-06,
+      "loss": 1.3373,
+      "step": 2155
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.98011461238203e-06,
+      "loss": 1.2512,
+      "step": 2156
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.978196823096765e-06,
+      "loss": 1.3963,
+      "step": 2157
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.976278354505584e-06,
+      "loss": 1.817,
+      "step": 2158
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.974359207046081e-06,
+      "loss": 1.5421,
+      "step": 2159
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.972439381155999e-06,
+      "loss": 1.6605,
+      "step": 2160
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.970518877273235e-06,
+      "loss": 1.5532,
+      "step": 2161
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.968597695835845e-06,
+      "loss": 1.2801,
+      "step": 2162
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.966675837282034e-06,
+      "loss": 1.0331,
+      "step": 2163
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.96475330205017e-06,
+      "loss": 1.1722,
+      "step": 2164
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.962830090578764e-06,
+      "loss": 1.674,
+      "step": 2165
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.96090620330649e-06,
+      "loss": 1.5165,
+      "step": 2166
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.958981640672173e-06,
+      "loss": 1.3178,
+      "step": 2167
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.957056403114792e-06,
+      "loss": 1.4454,
+      "step": 2168
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.955130491073482e-06,
+      "loss": 1.4411,
+      "step": 2169
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.953203904987526e-06,
+      "loss": 1.1842,
+      "step": 2170
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.951276645296368e-06,
+      "loss": 1.71,
+      "step": 2171
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.949348712439602e-06,
+      "loss": 1.2492,
+      "step": 2172
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.947420106856975e-06,
+      "loss": 1.3703,
+      "step": 2173
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.94549082898839e-06,
+      "loss": 1.5577,
+      "step": 2174
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.943560879273901e-06,
+      "loss": 1.1948,
+      "step": 2175
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.941630258153717e-06,
+      "loss": 1.7403,
+      "step": 2176
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.939698966068197e-06,
+      "loss": 1.6845,
+      "step": 2177
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.937767003457858e-06,
+      "loss": 1.6243,
+      "step": 2178
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.935834370763364e-06,
+      "loss": 1.4157,
+      "step": 2179
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.933901068425539e-06,
+      "loss": 1.3052,
+      "step": 2180
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.931967096885354e-06,
+      "loss": 1.1614,
+      "step": 2181
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.930032456583931e-06,
+      "loss": 1.319,
+      "step": 2182
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.928097147962552e-06,
+      "loss": 1.8114,
+      "step": 2183
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.926161171462647e-06,
+      "loss": 1.3211,
+      "step": 2184
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.924224527525799e-06,
+      "loss": 1.0607,
+      "step": 2185
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.922287216593741e-06,
+      "loss": 1.3307,
+      "step": 2186
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.920349239108362e-06,
+      "loss": 1.5598,
+      "step": 2187
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.918410595511698e-06,
+      "loss": 1.4682,
+      "step": 2188
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.916471286245945e-06,
+      "loss": 1.4785,
+      "step": 2189
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.914531311753443e-06,
+      "loss": 1.4412,
+      "step": 2190
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.912590672476686e-06,
+      "loss": 1.7624,
+      "step": 2191
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.910649368858321e-06,
+      "loss": 1.369,
+      "step": 2192
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.908707401341146e-06,
+      "loss": 1.5443,
+      "step": 2193
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.906764770368112e-06,
+      "loss": 1.3584,
+      "step": 2194
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.904821476382316e-06,
+      "loss": 1.48,
+      "step": 2195
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.902877519827012e-06,
+      "loss": 1.45,
+      "step": 2196
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.900932901145603e-06,
+      "loss": 1.5265,
+      "step": 2197
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.898987620781643e-06,
+      "loss": 1.2888,
+      "step": 2198
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.897041679178837e-06,
+      "loss": 1.3046,
+      "step": 2199
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.895095076781038e-06,
+      "loss": 1.317,
+      "step": 2200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.893147814032253e-06,
+      "loss": 1.6527,
+      "step": 2201
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.891199891376644e-06,
+      "loss": 1.7214,
+      "step": 2202
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.889251309258514e-06,
+      "loss": 1.4107,
+      "step": 2203
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.887302068122322e-06,
+      "loss": 1.4849,
+      "step": 2204
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.885352168412677e-06,
+      "loss": 1.417,
+      "step": 2205
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.883401610574338e-06,
+      "loss": 1.2582,
+      "step": 2206
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.88145039505221e-06,
+      "loss": 1.6319,
+      "step": 2207
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.879498522291355e-06,
+      "loss": 1.3536,
+      "step": 2208
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.877545992736982e-06,
+      "loss": 1.4654,
+      "step": 2209
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.875592806834445e-06,
+      "loss": 1.2648,
+      "step": 2210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.873638965029257e-06,
+      "loss": 1.3375,
+      "step": 2211
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.871684467767076e-06,
+      "loss": 1.4119,
+      "step": 2212
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.869729315493704e-06,
+      "loss": 1.63,
+      "step": 2213
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.867773508655103e-06,
+      "loss": 1.7846,
+      "step": 2214
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.865817047697376e-06,
+      "loss": 1.1061,
+      "step": 2215
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.863859933066777e-06,
+      "loss": 1.295,
+      "step": 2216
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.861902165209711e-06,
+      "loss": 1.4106,
+      "step": 2217
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.859943744572735e-06,
+      "loss": 1.3968,
+      "step": 2218
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.857984671602547e-06,
+      "loss": 1.353,
+      "step": 2219
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.856024946746e-06,
+      "loss": 1.5004,
+      "step": 2220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.854064570450091e-06,
+      "loss": 1.2592,
+      "step": 2221
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.85210354316197e-06,
+      "loss": 1.327,
+      "step": 2222
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.850141865328934e-06,
+      "loss": 1.2615,
+      "step": 2223
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.848179537398427e-06,
+      "loss": 1.1761,
+      "step": 2224
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.846216559818042e-06,
+      "loss": 1.5522,
+      "step": 2225
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.844252933035522e-06,
+      "loss": 1.2744,
+      "step": 2226
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.842288657498756e-06,
+      "loss": 1.5747,
+      "step": 2227
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.84032373365578e-06,
+      "loss": 1.4278,
+      "step": 2228
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.838358161954781e-06,
+      "loss": 1.6148,
+      "step": 2229
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.836391942844092e-06,
+      "loss": 1.3721,
+      "step": 2230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.834425076772194e-06,
+      "loss": 1.1615,
+      "step": 2231
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.832457564187715e-06,
+      "loss": 1.2152,
+      "step": 2232
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.830489405539431e-06,
+      "loss": 1.3727,
+      "step": 2233
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.828520601276266e-06,
+      "loss": 1.719,
+      "step": 2234
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.82655115184729e-06,
+      "loss": 1.3587,
+      "step": 2235
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.824581057701718e-06,
+      "loss": 1.2763,
+      "step": 2236
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.82261031928892e-06,
+      "loss": 1.295,
+      "step": 2237
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.820638937058402e-06,
+      "loss": 1.3484,
+      "step": 2238
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.818666911459827e-06,
+      "loss": 1.5366,
+      "step": 2239
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.816694242943e-06,
+      "loss": 1.3477,
+      "step": 2240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.814720931957871e-06,
+      "loss": 1.4835,
+      "step": 2241
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.812746978954539e-06,
+      "loss": 1.9177,
+      "step": 2242
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.810772384383246e-06,
+      "loss": 1.492,
+      "step": 2243
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.808797148694392e-06,
+      "loss": 1.3895,
+      "step": 2244
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.806821272338504e-06,
+      "loss": 1.4081,
+      "step": 2245
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.804844755766273e-06,
+      "loss": 1.3301,
+      "step": 2246
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.802867599428524e-06,
+      "loss": 1.408,
+      "step": 2247
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.800889803776235e-06,
+      "loss": 1.3163,
+      "step": 2248
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.798911369260526e-06,
+      "loss": 1.4045,
+      "step": 2249
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.796932296332666e-06,
+      "loss": 1.7123,
+      "step": 2250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.794952585444068e-06,
+      "loss": 1.5142,
+      "step": 2251
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.792972237046288e-06,
+      "loss": 1.6105,
+      "step": 2252
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.790991251591031e-06,
+      "loss": 1.8666,
+      "step": 2253
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.789009629530144e-06,
+      "loss": 1.6527,
+      "step": 2254
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.787027371315626e-06,
+      "loss": 1.3247,
+      "step": 2255
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.78504447739961e-06,
+      "loss": 1.5226,
+      "step": 2256
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.783060948234384e-06,
+      "loss": 1.648,
+      "step": 2257
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.781076784272377e-06,
+      "loss": 1.6934,
+      "step": 2258
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.779091985966161e-06,
+      "loss": 1.1172,
+      "step": 2259
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.777106553768457e-06,
+      "loss": 1.5516,
+      "step": 2260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.775120488132127e-06,
+      "loss": 1.7037,
+      "step": 2261
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.77313378951018e-06,
+      "loss": 0.9397,
+      "step": 2262
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.771146458355765e-06,
+      "loss": 1.5377,
+      "step": 2263
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.769158495122183e-06,
+      "loss": 1.6276,
+      "step": 2264
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.76716990026287e-06,
+      "loss": 1.39,
+      "step": 2265
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.765180674231414e-06,
+      "loss": 1.576,
+      "step": 2266
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.763190817481541e-06,
+      "loss": 1.5656,
+      "step": 2267
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.761200330467124e-06,
+      "loss": 1.4676,
+      "step": 2268
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.759209213642182e-06,
+      "loss": 1.7167,
+      "step": 2269
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.75721746746087e-06,
+      "loss": 1.605,
+      "step": 2270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.755225092377498e-06,
+      "loss": 1.1227,
+      "step": 2271
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.753232088846505e-06,
+      "loss": 1.3815,
+      "step": 2272
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.751238457322487e-06,
+      "loss": 1.6382,
+      "step": 2273
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.749244198260175e-06,
+      "loss": 1.5501,
+      "step": 2274
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.747249312114446e-06,
+      "loss": 1.5689,
+      "step": 2275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.745253799340322e-06,
+      "loss": 1.8602,
+      "step": 2276
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.743257660392959e-06,
+      "loss": 1.3876,
+      "step": 2277
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.74126089572767e-06,
+      "loss": 1.5166,
+      "step": 2278
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.7392635057999e-06,
+      "loss": 1.9915,
+      "step": 2279
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.737265491065238e-06,
+      "loss": 1.2114,
+      "step": 2280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.735266851979417e-06,
+      "loss": 1.5825,
+      "step": 2281
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.733267588998316e-06,
+      "loss": 1.5765,
+      "step": 2282
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.73126770257795e-06,
+      "loss": 1.4974,
+      "step": 2283
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.729267193174483e-06,
+      "loss": 1.4945,
+      "step": 2284
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.727266061244214e-06,
+      "loss": 1.6329,
+      "step": 2285
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.725264307243587e-06,
+      "loss": 1.2855,
+      "step": 2286
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.723261931629188e-06,
+      "loss": 1.242,
+      "step": 2287
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.721258934857748e-06,
+      "loss": 1.6751,
+      "step": 2288
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.719255317386135e-06,
+      "loss": 1.3632,
+      "step": 2289
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.717251079671356e-06,
+      "loss": 1.6582,
+      "step": 2290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.715246222170568e-06,
+      "loss": 1.3684,
+      "step": 2291
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.713240745341065e-06,
+      "loss": 1.5031,
+      "step": 2292
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.711234649640282e-06,
+      "loss": 1.0721,
+      "step": 2293
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.709227935525796e-06,
+      "loss": 1.6731,
+      "step": 2294
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.70722060345532e-06,
+      "loss": 1.0329,
+      "step": 2295
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.70521265388672e-06,
+      "loss": 1.3102,
+      "step": 2296
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.703204087277989e-06,
+      "loss": 1.2631,
+      "step": 2297
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.701194904087271e-06,
+      "loss": 1.6383,
+      "step": 2298
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.699185104772842e-06,
+      "loss": 1.4791,
+      "step": 2299
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.697174689793126e-06,
+      "loss": 1.4916,
+      "step": 2300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.695163659606687e-06,
+      "loss": 1.7572,
+      "step": 2301
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.693152014672223e-06,
+      "loss": 1.448,
+      "step": 2302
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.691139755448577e-06,
+      "loss": 1.2833,
+      "step": 2303
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.689126882394735e-06,
+      "loss": 1.1845,
+      "step": 2304
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.687113395969814e-06,
+      "loss": 1.448,
+      "step": 2305
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.685099296633079e-06,
+      "loss": 1.3569,
+      "step": 2306
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.683084584843931e-06,
+      "loss": 1.727,
+      "step": 2307
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.681069261061912e-06,
+      "loss": 1.454,
+      "step": 2308
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.679053325746705e-06,
+      "loss": 1.4078,
+      "step": 2309
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.67703677935813e-06,
+      "loss": 1.7273,
+      "step": 2310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.675019622356146e-06,
+      "loss": 1.1818,
+      "step": 2311
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.673001855200852e-06,
+      "loss": 1.2966,
+      "step": 2312
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.670983478352487e-06,
+      "loss": 1.4905,
+      "step": 2313
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.668964492271431e-06,
+      "loss": 1.577,
+      "step": 2314
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.666944897418199e-06,
+      "loss": 1.4705,
+      "step": 2315
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.664924694253444e-06,
+      "loss": 1.6358,
+      "step": 2316
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.662903883237963e-06,
+      "loss": 1.8409,
+      "step": 2317
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.660882464832692e-06,
+      "loss": 1.3606,
+      "step": 2318
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.658860439498694e-06,
+      "loss": 1.293,
+      "step": 2319
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.656837807697187e-06,
+      "loss": 1.4768,
+      "step": 2320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.654814569889515e-06,
+      "loss": 1.4928,
+      "step": 2321
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.652790726537166e-06,
+      "loss": 1.5129,
+      "step": 2322
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.650766278101762e-06,
+      "loss": 1.5359,
+      "step": 2323
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.64874122504507e-06,
+      "loss": 1.4455,
+      "step": 2324
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.646715567828985e-06,
+      "loss": 1.0273,
+      "step": 2325
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.644689306915551e-06,
+      "loss": 1.5935,
+      "step": 2326
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.642662442766942e-06,
+      "loss": 1.4269,
+      "step": 2327
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.640634975845468e-06,
+      "loss": 1.4215,
+      "step": 2328
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.638606906613587e-06,
+      "loss": 1.1833,
+      "step": 2329
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.63657823553388e-06,
+      "loss": 1.2967,
+      "step": 2330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.634548963069076e-06,
+      "loss": 1.6991,
+      "step": 2331
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.632519089682038e-06,
+      "loss": 1.5147,
+      "step": 2332
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.630488615835766e-06,
+      "loss": 1.3473,
+      "step": 2333
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.628457541993397e-06,
+      "loss": 1.4701,
+      "step": 2334
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.626425868618203e-06,
+      "loss": 1.5541,
+      "step": 2335
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.624393596173598e-06,
+      "loss": 1.3911,
+      "step": 2336
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.622360725123128e-06,
+      "loss": 1.296,
+      "step": 2337
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.620327255930475e-06,
+      "loss": 1.5739,
+      "step": 2338
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.61829318905946e-06,
+      "loss": 1.4682,
+      "step": 2339
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.61625852497404e-06,
+      "loss": 1.3903,
+      "step": 2340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.6142232641383065e-06,
+      "loss": 1.5012,
+      "step": 2341
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.6121874070164895e-06,
+      "loss": 1.544,
+      "step": 2342
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.610150954072953e-06,
+      "loss": 1.3394,
+      "step": 2343
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.6081139057721976e-06,
+      "loss": 1.5114,
+      "step": 2344
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.6060762625788585e-06,
+      "loss": 1.6916,
+      "step": 2345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.6040380249577124e-06,
+      "loss": 1.2754,
+      "step": 2346
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.601999193373664e-06,
+      "loss": 1.4718,
+      "step": 2347
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.599959768291756e-06,
+      "loss": 1.5568,
+      "step": 2348
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.597919750177168e-06,
+      "loss": 1.785,
+      "step": 2349
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.595879139495215e-06,
+      "loss": 1.3125,
+      "step": 2350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.593837936711344e-06,
+      "loss": 1.6435,
+      "step": 2351
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.591796142291139e-06,
+      "loss": 1.5142,
+      "step": 2352
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.589753756700322e-06,
+      "loss": 1.4225,
+      "step": 2353
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.587710780404742e-06,
+      "loss": 1.1397,
+      "step": 2354
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.585667213870394e-06,
+      "loss": 1.4415,
+      "step": 2355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.583623057563394e-06,
+      "loss": 1.5842,
+      "step": 2356
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.581578311950006e-06,
+      "loss": 1.4586,
+      "step": 2357
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.579532977496619e-06,
+      "loss": 1.728,
+      "step": 2358
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.577487054669761e-06,
+      "loss": 1.2773,
+      "step": 2359
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.575440543936092e-06,
+      "loss": 1.7341,
+      "step": 2360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.573393445762407e-06,
+      "loss": 1.2514,
+      "step": 2361
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5713457606156335e-06,
+      "loss": 1.6384,
+      "step": 2362
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.569297488962837e-06,
+      "loss": 1.4911,
+      "step": 2363
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5672486312712115e-06,
+      "loss": 1.3877,
+      "step": 2364
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5651991880080875e-06,
+      "loss": 1.5987,
+      "step": 2365
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.563149159640929e-06,
+      "loss": 1.5619,
+      "step": 2366
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.561098546637333e-06,
+      "loss": 1.7687,
+      "step": 2367
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.559047349465032e-06,
+      "loss": 1.1947,
+      "step": 2368
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.556995568591887e-06,
+      "loss": 1.6163,
+      "step": 2369
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.554943204485896e-06,
+      "loss": 1.4419,
+      "step": 2370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.552890257615189e-06,
+      "loss": 1.4267,
+      "step": 2371
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.55083672844803e-06,
+      "loss": 1.254,
+      "step": 2372
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.548782617452812e-06,
+      "loss": 1.3657,
+      "step": 2373
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.546727925098065e-06,
+      "loss": 1.463,
+      "step": 2374
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5446726518524505e-06,
+      "loss": 1.2482,
+      "step": 2375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5426167981847605e-06,
+      "loss": 1.4697,
+      "step": 2376
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.540560364563922e-06,
+      "loss": 1.4349,
+      "step": 2377
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.538503351458993e-06,
+      "loss": 1.5123,
+      "step": 2378
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.536445759339164e-06,
+      "loss": 1.3827,
+      "step": 2379
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.534387588673757e-06,
+      "loss": 1.1206,
+      "step": 2380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.532328839932226e-06,
+      "loss": 1.6704,
+      "step": 2381
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.530269513584158e-06,
+      "loss": 1.4311,
+      "step": 2382
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.528209610099271e-06,
+      "loss": 1.6845,
+      "step": 2383
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.526149129947414e-06,
+      "loss": 1.5288,
+      "step": 2384
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.524088073598568e-06,
+      "loss": 1.3409,
+      "step": 2385
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.522026441522847e-06,
+      "loss": 1.9549,
+      "step": 2386
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.519964234190493e-06,
+      "loss": 1.5863,
+      "step": 2387
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.51790145207188e-06,
+      "loss": 1.492,
+      "step": 2388
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.51583809563752e-06,
+      "loss": 1.6629,
+      "step": 2389
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.513774165358044e-06,
+      "loss": 1.7346,
+      "step": 2390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.511709661704223e-06,
+      "loss": 1.5577,
+      "step": 2391
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5096445851469555e-06,
+      "loss": 1.5469,
+      "step": 2392
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.507578936157272e-06,
+      "loss": 1.3929,
+      "step": 2393
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5055127152063305e-06,
+      "loss": 1.511,
+      "step": 2394
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5034459227654225e-06,
+      "loss": 1.6024,
+      "step": 2395
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.501378559305969e-06,
+      "loss": 1.5998,
+      "step": 2396
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.49931062529952e-06,
+      "loss": 1.8443,
+      "step": 2397
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.49724212121776e-06,
+      "loss": 1.1354,
+      "step": 2398
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.495173047532498e-06,
+      "loss": 1.5645,
+      "step": 2399
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.493103404715675e-06,
+      "loss": 1.807,
+      "step": 2400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.4910331932393634e-06,
+      "loss": 1.2648,
+      "step": 2401
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.488962413575764e-06,
+      "loss": 1.7207,
+      "step": 2402
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.486891066197205e-06,
+      "loss": 1.6053,
+      "step": 2403
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.484819151576148e-06,
+      "loss": 1.5406,
+      "step": 2404
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.482746670185182e-06,
+      "loss": 1.4583,
+      "step": 2405
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.480673622497027e-06,
+      "loss": 1.7469,
+      "step": 2406
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.4786000089845275e-06,
+      "loss": 1.4823,
+      "step": 2407
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4765258301206635e-06,
+      "loss": 1.4212,
+      "step": 2408
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.474451086378539e-06,
+      "loss": 1.5407,
+      "step": 2409
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.472375778231389e-06,
+      "loss": 1.5406,
+      "step": 2410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.470299906152576e-06,
+      "loss": 1.3984,
+      "step": 2411
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.468223470615593e-06,
+      "loss": 1.1498,
+      "step": 2412
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.46614647209406e-06,
+      "loss": 1.4553,
+      "step": 2413
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.464068911061726e-06,
+      "loss": 1.441,
+      "step": 2414
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.461990787992468e-06,
+      "loss": 1.6324,
+      "step": 2415
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.459912103360291e-06,
+      "loss": 1.6804,
+      "step": 2416
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.457832857639331e-06,
+      "loss": 1.4866,
+      "step": 2417
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.455753051303846e-06,
+      "loss": 1.2243,
+      "step": 2418
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.453672684828226e-06,
+      "loss": 1.2849,
+      "step": 2419
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.45159175868699e-06,
+      "loss": 1.3238,
+      "step": 2420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4495102733547805e-06,
+      "loss": 1.5928,
+      "step": 2421
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4474282293063725e-06,
+      "loss": 0.7882,
+      "step": 2422
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.445345627016662e-06,
+      "loss": 1.6756,
+      "step": 2423
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.443262466960679e-06,
+      "loss": 1.6865,
+      "step": 2424
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.441178749613578e-06,
+      "loss": 1.5516,
+      "step": 2425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.439094475450638e-06,
+      "loss": 1.6626,
+      "step": 2426
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.437009644947268e-06,
+      "loss": 1.3445,
+      "step": 2427
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.434924258579005e-06,
+      "loss": 1.4657,
+      "step": 2428
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.432838316821509e-06,
+      "loss": 1.1185,
+      "step": 2429
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4307518201505705e-06,
+      "loss": 1.2158,
+      "step": 2430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.428664769042104e-06,
+      "loss": 1.3959,
+      "step": 2431
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.42657716397215e-06,
+      "loss": 1.534,
+      "step": 2432
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.42448900541688e-06,
+      "loss": 1.5131,
+      "step": 2433
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.422400293852585e-06,
+      "loss": 1.6259,
+      "step": 2434
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.420311029755688e-06,
+      "loss": 1.1464,
+      "step": 2435
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.418221213602734e-06,
+      "loss": 1.5047,
+      "step": 2436
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.416130845870396e-06,
+      "loss": 1.4249,
+      "step": 2437
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.414039927035472e-06,
+      "loss": 1.1348,
+      "step": 2438
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.411948457574888e-06,
+      "loss": 1.7375,
+      "step": 2439
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.40985643796569e-06,
+      "loss": 1.5813,
+      "step": 2440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.407763868685056e-06,
+      "loss": 1.2725,
+      "step": 2441
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.405670750210283e-06,
+      "loss": 1.4813,
+      "step": 2442
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.403577083018801e-06,
+      "loss": 1.2235,
+      "step": 2443
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.401482867588159e-06,
+      "loss": 1.6068,
+      "step": 2444
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3993881043960335e-06,
+      "loss": 1.7302,
+      "step": 2445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.397292793920224e-06,
+      "loss": 1.6094,
+      "step": 2446
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.395196936638658e-06,
+      "loss": 1.5044,
+      "step": 2447
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.393100533029383e-06,
+      "loss": 1.6053,
+      "step": 2448
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3910035835705775e-06,
+      "loss": 1.4376,
+      "step": 2449
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.388906088740539e-06,
+      "loss": 1.77,
+      "step": 2450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.386808049017692e-06,
+      "loss": 1.5853,
+      "step": 2451
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.384709464880584e-06,
+      "loss": 1.4169,
+      "step": 2452
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.382610336807887e-06,
+      "loss": 1.5606,
+      "step": 2453
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3805106652784e-06,
+      "loss": 1.478,
+      "step": 2454
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.378410450771042e-06,
+      "loss": 1.6478,
+      "step": 2455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.376309693764857e-06,
+      "loss": 1.3005,
+      "step": 2456
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.374208394739013e-06,
+      "loss": 1.5075,
+      "step": 2457
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.372106554172802e-06,
+      "loss": 1.3188,
+      "step": 2458
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.370004172545639e-06,
+      "loss": 1.6473,
+      "step": 2459
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3679012503370635e-06,
+      "loss": 1.4692,
+      "step": 2460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3657977880267365e-06,
+      "loss": 1.7528,
+      "step": 2461
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.363693786094441e-06,
+      "loss": 1.6714,
+      "step": 2462
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.361589245020089e-06,
+      "loss": 1.4025,
+      "step": 2463
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.359484165283708e-06,
+      "loss": 1.7579,
+      "step": 2464
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.357378547365454e-06,
+      "loss": 1.2955,
+      "step": 2465
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.355272391745605e-06,
+      "loss": 1.3076,
+      "step": 2466
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.353165698904558e-06,
+      "loss": 1.2745,
+      "step": 2467
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.351058469322838e-06,
+      "loss": 1.6225,
+      "step": 2468
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.348950703481086e-06,
+      "loss": 1.6904,
+      "step": 2469
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.346842401860069e-06,
+      "loss": 1.5875,
+      "step": 2470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.344733564940678e-06,
+      "loss": 1.4414,
+      "step": 2471
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.342624193203922e-06,
+      "loss": 1.3013,
+      "step": 2472
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.340514287130935e-06,
+      "loss": 1.3703,
+      "step": 2473
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.338403847202972e-06,
+      "loss": 1.1936,
+      "step": 2474
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3362928739014085e-06,
+      "loss": 1.2592,
+      "step": 2475
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.334181367707744e-06,
+      "loss": 1.7277,
+      "step": 2476
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.3320693291035986e-06,
+      "loss": 1.3538,
+      "step": 2477
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.329956758570712e-06,
+      "loss": 1.4655,
+      "step": 2478
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.327843656590948e-06,
+      "loss": 1.3341,
+      "step": 2479
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.3257300236462906e-06,
+      "loss": 1.4519,
+      "step": 2480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.323615860218844e-06,
+      "loss": 1.4984,
+      "step": 2481
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.321501166790834e-06,
+      "loss": 1.6391,
+      "step": 2482
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.319385943844608e-06,
+      "loss": 1.9646,
+      "step": 2483
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.317270191862633e-06,
+      "loss": 1.0461,
+      "step": 2484
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.315153911327496e-06,
+      "loss": 1.7167,
+      "step": 2485
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.3130371027219095e-06,
+      "loss": 1.5215,
+      "step": 2486
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.3109197665287e-06,
+      "loss": 1.2983,
+      "step": 2487
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.308801903230817e-06,
+      "loss": 1.3532,
+      "step": 2488
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.306683513311333e-06,
+      "loss": 1.1146,
+      "step": 2489
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.304564597253436e-06,
+      "loss": 1.5682,
+      "step": 2490
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.302445155540435e-06,
+      "loss": 1.521,
+      "step": 2491
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.300325188655762e-06,
+      "loss": 1.4894,
+      "step": 2492
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.298204697082965e-06,
+      "loss": 1.1965,
+      "step": 2493
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2960836813057136e-06,
+      "loss": 1.3854,
+      "step": 2494
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2939621418077964e-06,
+      "loss": 1.9598,
+      "step": 2495
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.291840079073122e-06,
+      "loss": 1.7009,
+      "step": 2496
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2897174935857196e-06,
+      "loss": 1.305,
+      "step": 2497
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.287594385829735e-06,
+      "loss": 1.2507,
+      "step": 2498
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.285470756289435e-06,
+      "loss": 1.863,
+      "step": 2499
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.283346605449203e-06,
+      "loss": 1.7386,
+      "step": 2500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.281221933793544e-06,
+      "loss": 1.3252,
+      "step": 2501
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.279096741807081e-06,
+      "loss": 1.3278,
+      "step": 2502
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.276971029974555e-06,
+      "loss": 1.3037,
+      "step": 2503
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.274844798780826e-06,
+      "loss": 1.6069,
+      "step": 2504
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2727180487108725e-06,
+      "loss": 1.3802,
+      "step": 2505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.270590780249792e-06,
+      "loss": 1.3489,
+      "step": 2506
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.268462993882799e-06,
+      "loss": 1.4374,
+      "step": 2507
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.266334690095227e-06,
+      "loss": 1.6181,
+      "step": 2508
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.264205869372527e-06,
+      "loss": 1.4922,
+      "step": 2509
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2620765322002685e-06,
+      "loss": 1.5885,
+      "step": 2510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.259946679064137e-06,
+      "loss": 1.2847,
+      "step": 2511
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.25781631044994e-06,
+      "loss": 1.3746,
+      "step": 2512
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2556854268435975e-06,
+      "loss": 1.4434,
+      "step": 2513
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.253554028731149e-06,
+      "loss": 1.5756,
+      "step": 2514
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.251422116598753e-06,
+      "loss": 0.8568,
+      "step": 2515
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.249289690932684e-06,
+      "loss": 1.5977,
+      "step": 2516
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.247156752219331e-06,
+      "loss": 1.2021,
+      "step": 2517
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.245023300945203e-06,
+      "loss": 1.5144,
+      "step": 2518
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.242889337596928e-06,
+      "loss": 1.4925,
+      "step": 2519
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.240754862661246e-06,
+      "loss": 1.4013,
+      "step": 2520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.238619876625016e-06,
+      "loss": 1.6039,
+      "step": 2521
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.236484379975214e-06,
+      "loss": 1.2045,
+      "step": 2522
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.234348373198932e-06,
+      "loss": 1.3493,
+      "step": 2523
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2322118567833765e-06,
+      "loss": 1.0894,
+      "step": 2524
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.230074831215874e-06,
+      "loss": 1.7547,
+      "step": 2525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.227937296983865e-06,
+      "loss": 1.4509,
+      "step": 2526
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2257992545749045e-06,
+      "loss": 1.5061,
+      "step": 2527
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.223660704476666e-06,
+      "loss": 1.0158,
+      "step": 2528
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.221521647176939e-06,
+      "loss": 1.6875,
+      "step": 2529
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.219382083163625e-06,
+      "loss": 1.749,
+      "step": 2530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.217242012924747e-06,
+      "loss": 1.8077,
+      "step": 2531
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2151014369484375e-06,
+      "loss": 1.1935,
+      "step": 2532
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2129603557229465e-06,
+      "loss": 1.5263,
+      "step": 2533
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.210818769736642e-06,
+      "loss": 1.7151,
+      "step": 2534
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.208676679478003e-06,
+      "loss": 1.364,
+      "step": 2535
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.206534085435626e-06,
+      "loss": 1.4412,
+      "step": 2536
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.204390988098222e-06,
+      "loss": 1.4705,
+      "step": 2537
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2022473879546175e-06,
+      "loss": 1.5015,
+      "step": 2538
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.200103285493749e-06,
+      "loss": 1.3508,
+      "step": 2539
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.197958681204675e-06,
+      "loss": 1.3102,
+      "step": 2540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.195813575576565e-06,
+      "loss": 1.4834,
+      "step": 2541
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.1936679690987e-06,
+      "loss": 1.2568,
+      "step": 2542
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.191521862260479e-06,
+      "loss": 1.4553,
+      "step": 2543
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.189375255551413e-06,
+      "loss": 1.7201,
+      "step": 2544
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.18722814946113e-06,
+      "loss": 1.4446,
+      "step": 2545
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.18508054447937e-06,
+      "loss": 1.8239,
+      "step": 2546
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.182932441095985e-06,
+      "loss": 1.4907,
+      "step": 2547
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.180783839800942e-06,
+      "loss": 1.6736,
+      "step": 2548
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1786347410843246e-06,
+      "loss": 1.2863,
+      "step": 2549
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.176485145436325e-06,
+      "loss": 1.895,
+      "step": 2550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.174335053347252e-06,
+      "loss": 1.4215,
+      "step": 2551
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.172184465307526e-06,
+      "loss": 1.5375,
+      "step": 2552
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.170033381807681e-06,
+      "loss": 1.3876,
+      "step": 2553
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.167881803338366e-06,
+      "loss": 1.5195,
+      "step": 2554
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.165729730390338e-06,
+      "loss": 1.4133,
+      "step": 2555
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.163577163454472e-06,
+      "loss": 1.1283,
+      "step": 2556
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.161424103021752e-06,
+      "loss": 1.428,
+      "step": 2557
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.159270549583278e-06,
+      "loss": 1.6124,
+      "step": 2558
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.157116503630259e-06,
+      "loss": 1.4674,
+      "step": 2559
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.15496196565402e-06,
+      "loss": 1.3874,
+      "step": 2560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1528069361459905e-06,
+      "loss": 1.4027,
+      "step": 2561
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1506514155977245e-06,
+      "loss": 1.2356,
+      "step": 2562
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.148495404500877e-06,
+      "loss": 1.6757,
+      "step": 2563
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.146338903347221e-06,
+      "loss": 1.3016,
+      "step": 2564
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.14418191262864e-06,
+      "loss": 1.1588,
+      "step": 2565
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1420244328371244e-06,
+      "loss": 1.4663,
+      "step": 2566
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.139866464464785e-06,
+      "loss": 1.6005,
+      "step": 2567
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.137708008003838e-06,
+      "loss": 1.1221,
+      "step": 2568
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.13554906394661e-06,
+      "loss": 1.7264,
+      "step": 2569
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.133389632785543e-06,
+      "loss": 1.8595,
+      "step": 2570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.131229715013189e-06,
+      "loss": 1.7338,
+      "step": 2571
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.129069311122208e-06,
+      "loss": 1.5028,
+      "step": 2572
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1269084216053756e-06,
+      "loss": 1.3672,
+      "step": 2573
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.124747046955574e-06,
+      "loss": 1.4908,
+      "step": 2574
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.122585187665796e-06,
+      "loss": 1.382,
+      "step": 2575
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.12042284422915e-06,
+      "loss": 1.6298,
+      "step": 2576
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.11826001713885e-06,
+      "loss": 1.2041,
+      "step": 2577
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.116096706888222e-06,
+      "loss": 1.3433,
+      "step": 2578
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.113932913970701e-06,
+      "loss": 1.2181,
+      "step": 2579
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.111768638879834e-06,
+      "loss": 1.2894,
+      "step": 2580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.109603882109277e-06,
+      "loss": 1.2217,
+      "step": 2581
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1074386441527945e-06,
+      "loss": 1.3664,
+      "step": 2582
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1052729255042645e-06,
+      "loss": 1.443,
+      "step": 2583
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.10310672665767e-06,
+      "loss": 1.6072,
+      "step": 2584
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.100940048107109e-06,
+      "loss": 1.3007,
+      "step": 2585
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.098772890346782e-06,
+      "loss": 1.4367,
+      "step": 2586
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.096605253871005e-06,
+      "loss": 1.2892,
+      "step": 2587
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.094437139174202e-06,
+      "loss": 1.9482,
+      "step": 2588
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0922685467509e-06,
+      "loss": 1.3992,
+      "step": 2589
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.090099477095744e-06,
+      "loss": 1.3022,
+      "step": 2590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.087929930703483e-06,
+      "loss": 1.451,
+      "step": 2591
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0857599080689746e-06,
+      "loss": 1.2292,
+      "step": 2592
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.083589409687184e-06,
+      "loss": 1.3219,
+      "step": 2593
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.08141843605319e-06,
+      "loss": 1.4142,
+      "step": 2594
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.079246987662176e-06,
+      "loss": 1.6139,
+      "step": 2595
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0770750650094335e-06,
+      "loss": 1.653,
+      "step": 2596
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.074902668590362e-06,
+      "loss": 1.3134,
+      "step": 2597
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0727297989004705e-06,
+      "loss": 1.5032,
+      "step": 2598
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.070556456435378e-06,
+      "loss": 1.4369,
+      "step": 2599
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0683826416908045e-06,
+      "loss": 1.1907,
+      "step": 2600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.066208355162585e-06,
+      "loss": 1.5582,
+      "step": 2601
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.064033597346658e-06,
+      "loss": 1.6333,
+      "step": 2602
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.06185836873907e-06,
+      "loss": 1.3463,
+      "step": 2603
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.059682669835976e-06,
+      "loss": 1.4367,
+      "step": 2604
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.057506501133638e-06,
+      "loss": 1.2363,
+      "step": 2605
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.055329863128424e-06,
+      "loss": 1.4477,
+      "step": 2606
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.053152756316812e-06,
+      "loss": 1.6136,
+      "step": 2607
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.050975181195384e-06,
+      "loss": 1.3042,
+      "step": 2608
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.048797138260829e-06,
+      "loss": 1.4841,
+      "step": 2609
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.046618628009942e-06,
+      "loss": 1.4723,
+      "step": 2610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.044439650939629e-06,
+      "loss": 1.3979,
+      "step": 2611
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.042260207546899e-06,
+      "loss": 1.2902,
+      "step": 2612
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0400802983288665e-06,
+      "loss": 1.3764,
+      "step": 2613
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.037899923782754e-06,
+      "loss": 1.7038,
+      "step": 2614
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.035719084405888e-06,
+      "loss": 1.5518,
+      "step": 2615
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.033537780695705e-06,
+      "loss": 1.5274,
+      "step": 2616
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.031356013149746e-06,
+      "loss": 1.6638,
+      "step": 2617
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0291737822656545e-06,
+      "loss": 1.5111,
+      "step": 2618
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.026991088541184e-06,
+      "loss": 1.3504,
+      "step": 2619
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0248079324741915e-06,
+      "loss": 1.7212,
+      "step": 2620
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.022624314562637e-06,
+      "loss": 1.476,
+      "step": 2621
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.020440235304593e-06,
+      "loss": 1.3927,
+      "step": 2622
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.018255695198229e-06,
+      "loss": 1.3412,
+      "step": 2623
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.016070694741824e-06,
+      "loss": 1.3023,
+      "step": 2624
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0138852344337625e-06,
+      "loss": 1.19,
+      "step": 2625
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.011699314772533e-06,
+      "loss": 1.6953,
+      "step": 2626
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.009512936256727e-06,
+      "loss": 1.1714,
+      "step": 2627
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.007326099385045e-06,
+      "loss": 1.4607,
+      "step": 2628
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.005138804656285e-06,
+      "loss": 1.6279,
+      "step": 2629
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0029510525693565e-06,
+      "loss": 1.41,
+      "step": 2630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.00076284362327e-06,
+      "loss": 1.334,
+      "step": 2631
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.998574178317141e-06,
+      "loss": 1.6892,
+      "step": 2632
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.996385057150188e-06,
+      "loss": 1.4393,
+      "step": 2633
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.994195480621735e-06,
+      "loss": 1.4792,
+      "step": 2634
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9920054492312086e-06,
+      "loss": 1.6568,
+      "step": 2635
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.989814963478139e-06,
+      "loss": 1.755,
+      "step": 2636
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.987624023862162e-06,
+      "loss": 0.9644,
+      "step": 2637
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.985432630883016e-06,
+      "loss": 1.5019,
+      "step": 2638
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.983240785040539e-06,
+      "loss": 1.6947,
+      "step": 2639
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.981048486834682e-06,
+      "loss": 1.6296,
+      "step": 2640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.978855736765486e-06,
+      "loss": 1.5842,
+      "step": 2641
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.976662535333107e-06,
+      "loss": 1.6953,
+      "step": 2642
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.974468883037799e-06,
+      "loss": 1.561,
+      "step": 2643
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.972274780379914e-06,
+      "loss": 1.1989,
+      "step": 2644
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.970080227859915e-06,
+      "loss": 1.7931,
+      "step": 2645
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.967885225978366e-06,
+      "loss": 1.6994,
+      "step": 2646
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.965689775235926e-06,
+      "loss": 1.4,
+      "step": 2647
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.963493876133367e-06,
+      "loss": 1.2369,
+      "step": 2648
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.961297529171556e-06,
+      "loss": 1.4378,
+      "step": 2649
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.959100734851465e-06,
+      "loss": 1.41,
+      "step": 2650
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.956903493674167e-06,
+      "loss": 1.3259,
+      "step": 2651
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.954705806140838e-06,
+      "loss": 1.3765,
+      "step": 2652
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9525076727527554e-06,
+      "loss": 1.3443,
+      "step": 2653
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.950309094011298e-06,
+      "loss": 1.5911,
+      "step": 2654
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.948110070417945e-06,
+      "loss": 1.6898,
+      "step": 2655
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.945910602474279e-06,
+      "loss": 1.3759,
+      "step": 2656
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.943710690681984e-06,
+      "loss": 1.5808,
+      "step": 2657
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.941510335542843e-06,
+      "loss": 1.3202,
+      "step": 2658
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9393095375587436e-06,
+      "loss": 1.653,
+      "step": 2659
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.937108297231672e-06,
+      "loss": 1.473,
+      "step": 2660
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.934906615063716e-06,
+      "loss": 1.2738,
+      "step": 2661
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.932704491557062e-06,
+      "loss": 1.3012,
+      "step": 2662
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.930501927213999e-06,
+      "loss": 1.3578,
+      "step": 2663
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.92829892253692e-06,
+      "loss": 1.5491,
+      "step": 2664
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.926095478028312e-06,
+      "loss": 1.2895,
+      "step": 2665
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.923891594190767e-06,
+      "loss": 1.5027,
+      "step": 2666
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.921687271526973e-06,
+      "loss": 1.5565,
+      "step": 2667
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.919482510539723e-06,
+      "loss": 1.7671,
+      "step": 2668
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.917277311731908e-06,
+      "loss": 1.5733,
+      "step": 2669
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.915071675606516e-06,
+      "loss": 1.4791,
+      "step": 2670
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9128656026666385e-06,
+      "loss": 1.5193,
+      "step": 2671
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.910659093415466e-06,
+      "loss": 1.185,
+      "step": 2672
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.908452148356287e-06,
+      "loss": 1.9837,
+      "step": 2673
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.90624476799249e-06,
+      "loss": 1.3858,
+      "step": 2674
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.904036952827565e-06,
+      "loss": 1.3612,
+      "step": 2675
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.901828703365095e-06,
+      "loss": 1.6319,
+      "step": 2676
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.899620020108772e-06,
+      "loss": 1.6062,
+      "step": 2677
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.897410903562377e-06,
+      "loss": 1.2775,
+      "step": 2678
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.895201354229796e-06,
+      "loss": 1.6089,
+      "step": 2679
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.89299137261501e-06,
+      "loss": 1.5333,
+      "step": 2680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.890780959222103e-06,
+      "loss": 1.4768,
+      "step": 2681
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.888570114555253e-06,
+      "loss": 1.2097,
+      "step": 2682
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.886358839118739e-06,
+      "loss": 1.3738,
+      "step": 2683
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.884147133416936e-06,
+      "loss": 1.3692,
+      "step": 2684
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.881934997954321e-06,
+      "loss": 1.4466,
+      "step": 2685
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.879722433235464e-06,
+      "loss": 1.5707,
+      "step": 2686
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8775094397650375e-06,
+      "loss": 1.2069,
+      "step": 2687
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.87529601804781e-06,
+      "loss": 1.5616,
+      "step": 2688
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.873082168588647e-06,
+      "loss": 1.574,
+      "step": 2689
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.870867891892511e-06,
+      "loss": 1.6702,
+      "step": 2690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.868653188464465e-06,
+      "loss": 1.6239,
+      "step": 2691
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.866438058809667e-06,
+      "loss": 1.7751,
+      "step": 2692
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.864222503433372e-06,
+      "loss": 1.5494,
+      "step": 2693
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.862006522840931e-06,
+      "loss": 1.3907,
+      "step": 2694
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.859790117537799e-06,
+      "loss": 1.6248,
+      "step": 2695
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8575732880295164e-06,
+      "loss": 1.6076,
+      "step": 2696
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.85535603482173e-06,
+      "loss": 1.3221,
+      "step": 2697
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.853138358420179e-06,
+      "loss": 1.2603,
+      "step": 2698
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.850920259330701e-06,
+      "loss": 1.299,
+      "step": 2699
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8487017380592266e-06,
+      "loss": 1.4094,
+      "step": 2700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.846482795111786e-06,
+      "loss": 1.1581,
+      "step": 2701
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.844263430994507e-06,
+      "loss": 1.3999,
+      "step": 2702
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.842043646213607e-06,
+      "loss": 1.4653,
+      "step": 2703
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.839823441275408e-06,
+      "loss": 1.7248,
+      "step": 2704
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.837602816686318e-06,
+      "loss": 1.5488,
+      "step": 2705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8353817729528496e-06,
+      "loss": 1.5562,
+      "step": 2706
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.833160310581607e-06,
+      "loss": 1.4732,
+      "step": 2707
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.830938430079289e-06,
+      "loss": 1.2191,
+      "step": 2708
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8287161319526915e-06,
+      "loss": 1.3282,
+      "step": 2709
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.826493416708706e-06,
+      "loss": 1.3811,
+      "step": 2710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8242702848543185e-06,
+      "loss": 1.4938,
+      "step": 2711
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.822046736896607e-06,
+      "loss": 1.6267,
+      "step": 2712
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.81982277334275e-06,
+      "loss": 1.5045,
+      "step": 2713
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.817598394700019e-06,
+      "loss": 1.481,
+      "step": 2714
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.815373601475776e-06,
+      "loss": 1.5129,
+      "step": 2715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.813148394177485e-06,
+      "loss": 1.5858,
+      "step": 2716
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8109227733126956e-06,
+      "loss": 1.313,
+      "step": 2717
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.808696739389059e-06,
+      "loss": 1.6051,
+      "step": 2718
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.806470292914318e-06,
+      "loss": 1.4722,
+      "step": 2719
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.804243434396308e-06,
+      "loss": 1.8904,
+      "step": 2720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.802016164342963e-06,
+      "loss": 1.6773,
+      "step": 2721
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.799788483262303e-06,
+      "loss": 1.6895,
+      "step": 2722
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.797560391662451e-06,
+      "loss": 1.6582,
+      "step": 2723
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.795331890051618e-06,
+      "loss": 1.3417,
+      "step": 2724
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.793102978938107e-06,
+      "loss": 1.2799,
+      "step": 2725
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.790873658830321e-06,
+      "loss": 1.6183,
+      "step": 2726
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.788643930236749e-06,
+      "loss": 1.4276,
+      "step": 2727
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.786413793665979e-06,
+      "loss": 1.349,
+      "step": 2728
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.784183249626688e-06,
+      "loss": 1.1722,
+      "step": 2729
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.781952298627649e-06,
+      "loss": 1.6777,
+      "step": 2730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.779720941177725e-06,
+      "loss": 1.6113,
+      "step": 2731
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7774891777858746e-06,
+      "loss": 1.4593,
+      "step": 2732
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.775257008961144e-06,
+      "loss": 1.3938,
+      "step": 2733
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.773024435212678e-06,
+      "loss": 1.6209,
+      "step": 2734
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.770791457049713e-06,
+      "loss": 1.579,
+      "step": 2735
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.768558074981572e-06,
+      "loss": 1.0871,
+      "step": 2736
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.766324289517676e-06,
+      "loss": 1.5071,
+      "step": 2737
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.764090101167535e-06,
+      "loss": 1.4121,
+      "step": 2738
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.761855510440752e-06,
+      "loss": 1.3229,
+      "step": 2739
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.759620517847023e-06,
+      "loss": 1.7283,
+      "step": 2740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.757385123896132e-06,
+      "loss": 1.4297,
+      "step": 2741
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7551493290979564e-06,
+      "loss": 1.4161,
+      "step": 2742
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7529131339624686e-06,
+      "loss": 1.3107,
+      "step": 2743
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.750676538999726e-06,
+      "loss": 1.6281,
+      "step": 2744
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7484395447198805e-06,
+      "loss": 1.6374,
+      "step": 2745
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.746202151633176e-06,
+      "loss": 1.5598,
+      "step": 2746
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.743964360249947e-06,
+      "loss": 1.4329,
+      "step": 2747
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.741726171080618e-06,
+      "loss": 1.5973,
+      "step": 2748
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.739487584635702e-06,
+      "loss": 1.3819,
+      "step": 2749
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.737248601425808e-06,
+      "loss": 1.6956,
+      "step": 2750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.735009221961628e-06,
+      "loss": 1.3844,
+      "step": 2751
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.732769446753954e-06,
+      "loss": 1.4188,
+      "step": 2752
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.730529276313661e-06,
+      "loss": 1.2443,
+      "step": 2753
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7282887111517144e-06,
+      "loss": 1.6944,
+      "step": 2754
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7260477517791735e-06,
+      "loss": 1.2997,
+      "step": 2755
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.723806398707186e-06,
+      "loss": 0.8917,
+      "step": 2756
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.721564652446987e-06,
+      "loss": 1.5737,
+      "step": 2757
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7193225135099045e-06,
+      "loss": 1.4283,
+      "step": 2758
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.717079982407355e-06,
+      "loss": 1.3645,
+      "step": 2759
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7148370596508435e-06,
+      "loss": 1.4148,
+      "step": 2760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.712593745751965e-06,
+      "loss": 1.0906,
+      "step": 2761
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.710350041222404e-06,
+      "loss": 1.8166,
+      "step": 2762
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.708105946573934e-06,
+      "loss": 1.484,
+      "step": 2763
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.705861462318417e-06,
+      "loss": 1.4523,
+      "step": 2764
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.703616588967804e-06,
+      "loss": 1.2142,
+      "step": 2765
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.701371327034134e-06,
+      "loss": 1.6921,
+      "step": 2766
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.699125677029538e-06,
+      "loss": 1.6546,
+      "step": 2767
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.696879639466232e-06,
+      "loss": 1.2729,
+      "step": 2768
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.69463321485652e-06,
+      "loss": 1.4709,
+      "step": 2769
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.692386403712801e-06,
+      "loss": 1.8872,
+      "step": 2770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.69013920654755e-06,
+      "loss": 1.6213,
+      "step": 2771
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.687891623873341e-06,
+      "loss": 1.1036,
+      "step": 2772
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.68564365620283e-06,
+      "loss": 1.5728,
+      "step": 2773
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.683395304048764e-06,
+      "loss": 1.5043,
+      "step": 2774
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.681146567923977e-06,
+      "loss": 1.5269,
+      "step": 2775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.678897448341388e-06,
+      "loss": 1.5043,
+      "step": 2776
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6766479458140075e-06,
+      "loss": 1.2764,
+      "step": 2777
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.674398060854931e-06,
+      "loss": 1.5978,
+      "step": 2778
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.672147793977341e-06,
+      "loss": 1.427,
+      "step": 2779
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.669897145694507e-06,
+      "loss": 1.333,
+      "step": 2780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.667646116519788e-06,
+      "loss": 1.283,
+      "step": 2781
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.665394706966626e-06,
+      "loss": 1.5947,
+      "step": 2782
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.663142917548553e-06,
+      "loss": 1.492,
+      "step": 2783
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.660890748779186e-06,
+      "loss": 1.738,
+      "step": 2784
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.658638201172231e-06,
+      "loss": 1.3717,
+      "step": 2785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.656385275241475e-06,
+      "loss": 1.4618,
+      "step": 2786
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.654131971500796e-06,
+      "loss": 1.5203,
+      "step": 2787
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6518782904641576e-06,
+      "loss": 1.6212,
+      "step": 2788
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.649624232645607e-06,
+      "loss": 1.5598,
+      "step": 2789
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.647369798559282e-06,
+      "loss": 1.0921,
+      "step": 2790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.645114988719401e-06,
+      "loss": 1.7225,
+      "step": 2791
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.642859803640272e-06,
+      "loss": 1.5697,
+      "step": 2792
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.640604243836285e-06,
+      "loss": 1.2716,
+      "step": 2793
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.638348309821918e-06,
+      "loss": 1.307,
+      "step": 2794
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.636092002111736e-06,
+      "loss": 1.2652,
+      "step": 2795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.633835321220383e-06,
+      "loss": 1.3036,
+      "step": 2796
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.631578267662596e-06,
+      "loss": 1.5067,
+      "step": 2797
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6293208419531915e-06,
+      "loss": 1.377,
+      "step": 2798
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6270630446070726e-06,
+      "loss": 1.6148,
+      "step": 2799
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.624804876139227e-06,
+      "loss": 1.4048,
+      "step": 2800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.622546337064728e-06,
+      "loss": 1.3294,
+      "step": 2801
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.620287427898731e-06,
+      "loss": 1.5912,
+      "step": 2802
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.618028149156479e-06,
+      "loss": 1.343,
+      "step": 2803
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.615768501353297e-06,
+      "loss": 1.3164,
+      "step": 2804
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.613508485004594e-06,
+      "loss": 1.4644,
+      "step": 2805
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.611248100625866e-06,
+      "loss": 1.6937,
+      "step": 2806
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.608987348732689e-06,
+      "loss": 1.3533,
+      "step": 2807
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.606726229840725e-06,
+      "loss": 1.37,
+      "step": 2808
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.604464744465722e-06,
+      "loss": 1.3809,
+      "step": 2809
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.602202893123506e-06,
+      "loss": 1.7417,
+      "step": 2810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.59994067632999e-06,
+      "loss": 1.1199,
+      "step": 2811
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.597678094601172e-06,
+      "loss": 1.4161,
+      "step": 2812
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.595415148453128e-06,
+      "loss": 1.4848,
+      "step": 2813
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.593151838402023e-06,
+      "loss": 1.6883,
+      "step": 2814
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5908881649641e-06,
+      "loss": 1.3089,
+      "step": 2815
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.588624128655688e-06,
+      "loss": 1.5461,
+      "step": 2816
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5863597299932e-06,
+      "loss": 1.3783,
+      "step": 2817
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.584094969493126e-06,
+      "loss": 1.5036,
+      "step": 2818
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.581829847672043e-06,
+      "loss": 1.2921,
+      "step": 2819
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.579564365046609e-06,
+      "loss": 1.422,
+      "step": 2820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.577298522133569e-06,
+      "loss": 1.6201,
+      "step": 2821
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.57503231944974e-06,
+      "loss": 1.587,
+      "step": 2822
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.572765757512031e-06,
+      "loss": 1.6419,
+      "step": 2823
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.570498836837428e-06,
+      "loss": 1.151,
+      "step": 2824
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.568231557942999e-06,
+      "loss": 1.1916,
+      "step": 2825
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.565963921345896e-06,
+      "loss": 1.6638,
+      "step": 2826
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5636959275633494e-06,
+      "loss": 1.3797,
+      "step": 2827
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.561427577112673e-06,
+      "loss": 1.2504,
+      "step": 2828
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.559158870511264e-06,
+      "loss": 1.6638,
+      "step": 2829
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5568898082765945e-06,
+      "loss": 1.3876,
+      "step": 2830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.554620390926225e-06,
+      "loss": 1.2366,
+      "step": 2831
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.552350618977793e-06,
+      "loss": 1.635,
+      "step": 2832
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.550080492949017e-06,
+      "loss": 1.6921,
+      "step": 2833
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.547810013357697e-06,
+      "loss": 1.5891,
+      "step": 2834
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.545539180721715e-06,
+      "loss": 1.6281,
+      "step": 2835
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5432679955590305e-06,
+      "loss": 1.5712,
+      "step": 2836
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.540996458387685e-06,
+      "loss": 1.4174,
+      "step": 2837
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.538724569725802e-06,
+      "loss": 1.2342,
+      "step": 2838
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.536452330091581e-06,
+      "loss": 1.3323,
+      "step": 2839
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.534179740003305e-06,
+      "loss": 1.4745,
+      "step": 2840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.531906799979335e-06,
+      "loss": 1.3823,
+      "step": 2841
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.529633510538116e-06,
+      "loss": 1.308,
+      "step": 2842
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.527359872198166e-06,
+      "loss": 1.4647,
+      "step": 2843
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.525085885478088e-06,
+      "loss": 1.4213,
+      "step": 2844
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5228115508965626e-06,
+      "loss": 1.6275,
+      "step": 2845
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.520536868972348e-06,
+      "loss": 1.2728,
+      "step": 2846
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5182618402242846e-06,
+      "loss": 1.3372,
+      "step": 2847
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.51598646517129e-06,
+      "loss": 1.4011,
+      "step": 2848
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.513710744332361e-06,
+      "loss": 1.2881,
+      "step": 2849
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.511434678226574e-06,
+      "loss": 1.5693,
+      "step": 2850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.509158267373086e-06,
+      "loss": 1.4503,
+      "step": 2851
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.506881512291125e-06,
+      "loss": 1.0858,
+      "step": 2852
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.504604413500009e-06,
+      "loss": 1.5154,
+      "step": 2853
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.502326971519125e-06,
+      "loss": 1.7414,
+      "step": 2854
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.500049186867943e-06,
+      "loss": 1.4312,
+      "step": 2855
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.497771060066008e-06,
+      "loss": 1.7118,
+      "step": 2856
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.495492591632947e-06,
+      "loss": 1.6783,
+      "step": 2857
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.493213782088461e-06,
+      "loss": 1.1296,
+      "step": 2858
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.490934631952332e-06,
+      "loss": 1.3651,
+      "step": 2859
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.488655141744417e-06,
+      "loss": 1.4427,
+      "step": 2860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.486375311984651e-06,
+      "loss": 1.6046,
+      "step": 2861
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.48409514319305e-06,
+      "loss": 1.1895,
+      "step": 2862
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4818146358897016e-06,
+      "loss": 1.6908,
+      "step": 2863
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.479533790594777e-06,
+      "loss": 1.14,
+      "step": 2864
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.477252607828518e-06,
+      "loss": 1.5901,
+      "step": 2865
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4749710881112485e-06,
+      "loss": 1.4432,
+      "step": 2866
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4726892319633665e-06,
+      "loss": 1.4927,
+      "step": 2867
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.470407039905347e-06,
+      "loss": 1.4207,
+      "step": 2868
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.468124512457743e-06,
+      "loss": 1.3868,
+      "step": 2869
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.465841650141183e-06,
+      "loss": 1.3276,
+      "step": 2870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.46355845347637e-06,
+      "loss": 1.8023,
+      "step": 2871
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.461274922984087e-06,
+      "loss": 1.3621,
+      "step": 2872
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.458991059185191e-06,
+      "loss": 1.6423,
+      "step": 2873
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.456706862600616e-06,
+      "loss": 1.2798,
+      "step": 2874
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.454422333751372e-06,
+      "loss": 1.4038,
+      "step": 2875
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.45213747315854e-06,
+      "loss": 1.5549,
+      "step": 2876
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.449852281343286e-06,
+      "loss": 1.4237,
+      "step": 2877
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.447566758826843e-06,
+      "loss": 1.4554,
+      "step": 2878
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.445280906130524e-06,
+      "loss": 1.3373,
+      "step": 2879
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4429947237757174e-06,
+      "loss": 1.682,
+      "step": 2880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.440708212283881e-06,
+      "loss": 1.5042,
+      "step": 2881
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4384213721765565e-06,
+      "loss": 1.522,
+      "step": 2882
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.436134203975355e-06,
+      "loss": 1.8298,
+      "step": 2883
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.43384670820196e-06,
+      "loss": 1.6289,
+      "step": 2884
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.4315588853781394e-06,
+      "loss": 1.3403,
+      "step": 2885
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.429270736025728e-06,
+      "loss": 1.2954,
+      "step": 2886
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.426982260666634e-06,
+      "loss": 1.7457,
+      "step": 2887
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.424693459822843e-06,
+      "loss": 1.8318,
+      "step": 2888
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.422404334016417e-06,
+      "loss": 1.39,
+      "step": 2889
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.420114883769487e-06,
+      "loss": 1.454,
+      "step": 2890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.417825109604263e-06,
+      "loss": 1.3809,
+      "step": 2891
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.415535012043025e-06,
+      "loss": 1.6123,
+      "step": 2892
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.4132445916081285e-06,
+      "loss": 1.4585,
+      "step": 2893
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.410953848822003e-06,
+      "loss": 1.4158,
+      "step": 2894
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.408662784207149e-06,
+      "loss": 1.3498,
+      "step": 2895
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.406371398286144e-06,
+      "loss": 1.59,
+      "step": 2896
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.404079691581638e-06,
+      "loss": 1.3345,
+      "step": 2897
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.401787664616352e-06,
+      "loss": 1.3563,
+      "step": 2898
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.399495317913082e-06,
+      "loss": 1.4844,
+      "step": 2899
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.397202651994696e-06,
+      "loss": 1.3968,
+      "step": 2900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3949096673841345e-06,
+      "loss": 1.2173,
+      "step": 2901
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.392616364604412e-06,
+      "loss": 1.8007,
+      "step": 2902
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3903227441786155e-06,
+      "loss": 1.274,
+      "step": 2903
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.388028806629902e-06,
+      "loss": 1.2907,
+      "step": 2904
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.385734552481504e-06,
+      "loss": 1.3818,
+      "step": 2905
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.383439982256723e-06,
+      "loss": 1.287,
+      "step": 2906
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.381145096478938e-06,
+      "loss": 1.1616,
+      "step": 2907
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.378849895671594e-06,
+      "loss": 1.553,
+      "step": 2908
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3765543803582096e-06,
+      "loss": 1.0658,
+      "step": 2909
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.374258551062377e-06,
+      "loss": 1.6316,
+      "step": 2910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.371962408307761e-06,
+      "loss": 1.5068,
+      "step": 2911
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.369665952618091e-06,
+      "loss": 1.6007,
+      "step": 2912
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.367369184517177e-06,
+      "loss": 1.7172,
+      "step": 2913
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.365072104528892e-06,
+      "loss": 1.829,
+      "step": 2914
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.362774713177187e-06,
+      "loss": 1.2219,
+      "step": 2915
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.360477010986079e-06,
+      "loss": 1.5475,
+      "step": 2916
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3581789984796585e-06,
+      "loss": 1.2087,
+      "step": 2917
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.355880676182086e-06,
+      "loss": 1.2978,
+      "step": 2918
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.353582044617593e-06,
+      "loss": 1.5154,
+      "step": 2919
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.351283104310479e-06,
+      "loss": 1.3713,
+      "step": 2920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.348983855785122e-06,
+      "loss": 1.1862,
+      "step": 2921
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.346684299565959e-06,
+      "loss": 1.3913,
+      "step": 2922
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3443844361775056e-06,
+      "loss": 1.2195,
+      "step": 2923
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.342084266144343e-06,
+      "loss": 1.3825,
+      "step": 2924
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.339783789991125e-06,
+      "loss": 1.8184,
+      "step": 2925
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.337483008242573e-06,
+      "loss": 1.5272,
+      "step": 2926
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.335181921423481e-06,
+      "loss": 1.6837,
+      "step": 2927
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.332880530058709e-06,
+      "loss": 1.7515,
+      "step": 2928
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.330578834673191e-06,
+      "loss": 1.0284,
+      "step": 2929
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.328276835791924e-06,
+      "loss": 1.2862,
+      "step": 2930
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.32597453393998e-06,
+      "loss": 1.402,
+      "step": 2931
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3236719296424985e-06,
+      "loss": 1.6662,
+      "step": 2932
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.321369023424686e-06,
+      "loss": 1.4686,
+      "step": 2933
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3190658158118205e-06,
+      "loss": 1.3817,
+      "step": 2934
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.316762307329248e-06,
+      "loss": 1.4794,
+      "step": 2935
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.31445849850238e-06,
+      "loss": 1.6116,
+      "step": 2936
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.312154389856702e-06,
+      "loss": 1.3964,
+      "step": 2937
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.309849981917762e-06,
+      "loss": 1.3513,
+      "step": 2938
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.307545275211183e-06,
+      "loss": 1.6027,
+      "step": 2939
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.305240270262651e-06,
+      "loss": 1.3888,
+      "step": 2940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.302934967597922e-06,
+      "loss": 1.5491,
+      "step": 2941
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.30062936774282e-06,
+      "loss": 1.4496,
+      "step": 2942
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.2983234712232334e-06,
+      "loss": 1.527,
+      "step": 2943
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.296017278565124e-06,
+      "loss": 1.5579,
+      "step": 2944
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.2937107902945175e-06,
+      "loss": 1.5892,
+      "step": 2945
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.291404006937507e-06,
+      "loss": 1.3858,
+      "step": 2946
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.289096929020254e-06,
+      "loss": 1.5011,
+      "step": 2947
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.286789557068987e-06,
+      "loss": 1.5022,
+      "step": 2948
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.284481891610001e-06,
+      "loss": 1.5166,
+      "step": 2949
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.28217393316966e-06,
+      "loss": 1.7139,
+      "step": 2950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.279865682274391e-06,
+      "loss": 1.6788,
+      "step": 2951
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.277557139450692e-06,
+      "loss": 1.3251,
+      "step": 2952
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.275248305225124e-06,
+      "loss": 1.5196,
+      "step": 2953
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.272939180124316e-06,
+      "loss": 1.0894,
+      "step": 2954
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2706297646749655e-06,
+      "loss": 1.3848,
+      "step": 2955
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.268320059403832e-06,
+      "loss": 1.2938,
+      "step": 2956
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.266010064837742e-06,
+      "loss": 1.6508,
+      "step": 2957
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.263699781503591e-06,
+      "loss": 1.81,
+      "step": 2958
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2613892099283405e-06,
+      "loss": 1.5966,
+      "step": 2959
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.25907835063901e-06,
+      "loss": 1.3547,
+      "step": 2960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2567672041626956e-06,
+      "loss": 1.4766,
+      "step": 2961
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.254455771026552e-06,
+      "loss": 1.0473,
+      "step": 2962
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2521440517578e-06,
+      "loss": 1.3095,
+      "step": 2963
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.249832046883729e-06,
+      "loss": 1.4704,
+      "step": 2964
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.247519756931689e-06,
+      "loss": 1.1917,
+      "step": 2965
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.245207182429098e-06,
+      "loss": 1.1875,
+      "step": 2966
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2428943239034375e-06,
+      "loss": 1.3604,
+      "step": 2967
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2405811818822545e-06,
+      "loss": 1.5032,
+      "step": 2968
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.238267756893163e-06,
+      "loss": 1.7033,
+      "step": 2969
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.235954049463834e-06,
+      "loss": 1.4075,
+      "step": 2970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2336400601220125e-06,
+      "loss": 1.3447,
+      "step": 2971
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.231325789395501e-06,
+      "loss": 1.5169,
+      "step": 2972
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.229011237812172e-06,
+      "loss": 1.3282,
+      "step": 2973
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2266964058999535e-06,
+      "loss": 1.7626,
+      "step": 2974
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.2243812941868455e-06,
+      "loss": 1.4597,
+      "step": 2975
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.222065903200909e-06,
+      "loss": 1.6415,
+      "step": 2976
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.219750233470267e-06,
+      "loss": 1.456,
+      "step": 2977
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.217434285523109e-06,
+      "loss": 1.4735,
+      "step": 2978
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.215118059887685e-06,
+      "loss": 1.4964,
+      "step": 2979
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.212801557092312e-06,
+      "loss": 1.4025,
+      "step": 2980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.210484777665366e-06,
+      "loss": 1.2776,
+      "step": 2981
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.208167722135289e-06,
+      "loss": 1.1936,
+      "step": 2982
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.205850391030586e-06,
+      "loss": 1.4356,
+      "step": 2983
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.203532784879824e-06,
+      "loss": 1.8519,
+      "step": 2984
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.201214904211631e-06,
+      "loss": 1.4603,
+      "step": 2985
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1988967495547016e-06,
+      "loss": 1.2933,
+      "step": 2986
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1965783214377895e-06,
+      "loss": 1.2832,
+      "step": 2987
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.194259620389712e-06,
+      "loss": 1.179,
+      "step": 2988
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.19194064693935e-06,
+      "loss": 1.3887,
+      "step": 2989
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.189621401615643e-06,
+      "loss": 1.3917,
+      "step": 2990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.187301884947597e-06,
+      "loss": 1.709,
+      "step": 2991
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.184982097464273e-06,
+      "loss": 1.4499,
+      "step": 2992
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1826620396948035e-06,
+      "loss": 1.1682,
+      "step": 2993
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.180341712168376e-06,
+      "loss": 1.3007,
+      "step": 2994
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.178021115414241e-06,
+      "loss": 1.3436,
+      "step": 2995
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.175700249961709e-06,
+      "loss": 1.3865,
+      "step": 2996
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.173379116340155e-06,
+      "loss": 1.096,
+      "step": 2997
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.171057715079012e-06,
+      "loss": 1.5959,
+      "step": 2998
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.168736046707777e-06,
+      "loss": 0.9753,
+      "step": 2999
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.166414111756005e-06,
+      "loss": 1.4845,
+      "step": 3000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.164091910753313e-06,
+      "loss": 1.1543,
+      "step": 3001
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.16176944422938e-06,
+      "loss": 1.4737,
+      "step": 3002
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.159446712713942e-06,
+      "loss": 1.3221,
+      "step": 3003
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.157123716736802e-06,
+      "loss": 1.3043,
+      "step": 3004
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.154800456827816e-06,
+      "loss": 1.6035,
+      "step": 3005
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1524769335169044e-06,
+      "loss": 1.4645,
+      "step": 3006
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.150153147334046e-06,
+      "loss": 1.5725,
+      "step": 3007
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.147829098809282e-06,
+      "loss": 1.0392,
+      "step": 3008
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1455047884727095e-06,
+      "loss": 1.6076,
+      "step": 3009
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.143180216854488e-06,
+      "loss": 1.4029,
+      "step": 3010
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1408553844848365e-06,
+      "loss": 1.3455,
+      "step": 3011
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.138530291894033e-06,
+      "loss": 1.677,
+      "step": 3012
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.136204939612414e-06,
+      "loss": 1.2477,
+      "step": 3013
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1338793281703775e-06,
+      "loss": 1.3083,
+      "step": 3014
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.131553458098378e-06,
+      "loss": 1.557,
+      "step": 3015
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.129227329926932e-06,
+      "loss": 1.3499,
+      "step": 3016
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.126900944186612e-06,
+      "loss": 1.4516,
+      "step": 3017
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.12457430140805e-06,
+      "loss": 1.4301,
+      "step": 3018
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.122247402121938e-06,
+      "loss": 1.5272,
+      "step": 3019
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.119920246859025e-06,
+      "loss": 1.3335,
+      "step": 3020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.117592836150118e-06,
+      "loss": 1.3773,
+      "step": 3021
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.115265170526084e-06,
+      "loss": 1.3049,
+      "step": 3022
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.112937250517846e-06,
+      "loss": 1.3802,
+      "step": 3023
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.110609076656389e-06,
+      "loss": 1.3403,
+      "step": 3024
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.108280649472751e-06,
+      "loss": 1.4133,
+      "step": 3025
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.105951969498029e-06,
+      "loss": 1.4847,
+      "step": 3026
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.103623037263383e-06,
+      "loss": 1.2845,
+      "step": 3027
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.101293853300023e-06,
+      "loss": 1.2683,
+      "step": 3028
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.098964418139219e-06,
+      "loss": 1.5733,
+      "step": 3029
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.096634732312301e-06,
+      "loss": 1.2978,
+      "step": 3030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.094304796350653e-06,
+      "loss": 1.7353,
+      "step": 3031
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0919746107857165e-06,
+      "loss": 1.2431,
+      "step": 3032
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0896441761489925e-06,
+      "loss": 1.292,
+      "step": 3033
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.087313492972034e-06,
+      "loss": 1.496,
+      "step": 3034
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0849825617864565e-06,
+      "loss": 1.5101,
+      "step": 3035
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0826513831239274e-06,
+      "loss": 1.7316,
+      "step": 3036
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.080319957516173e-06,
+      "loss": 1.2798,
+      "step": 3037
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0779882854949745e-06,
+      "loss": 1.3198,
+      "step": 3038
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.075656367592171e-06,
+      "loss": 1.3669,
+      "step": 3039
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.073324204339657e-06,
+      "loss": 1.687,
+      "step": 3040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.07099179626938e-06,
+      "loss": 1.3775,
+      "step": 3041
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.068659143913349e-06,
+      "loss": 1.3637,
+      "step": 3042
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.066326247803624e-06,
+      "loss": 1.5664,
+      "step": 3043
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.063993108472323e-06,
+      "loss": 1.6393,
+      "step": 3044
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.06165972645162e-06,
+      "loss": 1.5111,
+      "step": 3045
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0593261022737405e-06,
+      "loss": 1.2701,
+      "step": 3046
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.05699223647097e-06,
+      "loss": 1.2323,
+      "step": 3047
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.054658129575646e-06,
+      "loss": 1.5794,
+      "step": 3048
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0523237821201615e-06,
+      "loss": 1.3397,
+      "step": 3049
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.049989194636966e-06,
+      "loss": 1.6444,
+      "step": 3050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.047654367658563e-06,
+      "loss": 1.5891,
+      "step": 3051
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.045319301717507e-06,
+      "loss": 1.2924,
+      "step": 3052
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.042983997346415e-06,
+      "loss": 1.5676,
+      "step": 3053
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.040648455077951e-06,
+      "loss": 1.2563,
+      "step": 3054
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.038312675444836e-06,
+      "loss": 1.3752,
+      "step": 3055
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.035976658979846e-06,
+      "loss": 1.35,
+      "step": 3056
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.033640406215808e-06,
+      "loss": 1.1178,
+      "step": 3057
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.031303917685608e-06,
+      "loss": 1.965,
+      "step": 3058
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.028967193922179e-06,
+      "loss": 1.4344,
+      "step": 3059
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.026630235458516e-06,
+      "loss": 1.3883,
+      "step": 3060
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.024293042827658e-06,
+      "loss": 1.4535,
+      "step": 3061
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.021955616562706e-06,
+      "loss": 1.2695,
+      "step": 3062
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.019617957196809e-06,
+      "loss": 1.4569,
+      "step": 3063
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0172800652631706e-06,
+      "loss": 1.2709,
+      "step": 3064
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.014941941295048e-06,
+      "loss": 1.5199,
+      "step": 3065
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.012603585825752e-06,
+      "loss": 1.3893,
+      "step": 3066
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.010264999388643e-06,
+      "loss": 1.4041,
+      "step": 3067
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0079261825171365e-06,
+      "loss": 1.4766,
+      "step": 3068
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.005587135744702e-06,
+      "loss": 1.5089,
+      "step": 3069
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.003247859604861e-06,
+      "loss": 1.4965,
+      "step": 3070
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.000908354631183e-06,
+      "loss": 1.4608,
+      "step": 3071
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.998568621357294e-06,
+      "loss": 1.6715,
+      "step": 3072
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.99622866031687e-06,
+      "loss": 1.5536,
+      "step": 3073
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.993888472043641e-06,
+      "loss": 1.2458,
+      "step": 3074
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9915480570713865e-06,
+      "loss": 1.7613,
+      "step": 3075
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.98920741593394e-06,
+      "loss": 1.7946,
+      "step": 3076
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.986866549165185e-06,
+      "loss": 1.7425,
+      "step": 3077
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9845254572990555e-06,
+      "loss": 1.5643,
+      "step": 3078
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.982184140869539e-06,
+      "loss": 1.3196,
+      "step": 3079
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.979842600410676e-06,
+      "loss": 1.5526,
+      "step": 3080
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.977500836456552e-06,
+      "loss": 1.517,
+      "step": 3081
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.975158849541308e-06,
+      "loss": 1.2634,
+      "step": 3082
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9728166401991375e-06,
+      "loss": 1.0584,
+      "step": 3083
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.970474208964278e-06,
+      "loss": 1.1909,
+      "step": 3084
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.968131556371024e-06,
+      "loss": 1.2815,
+      "step": 3085
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.965788682953717e-06,
+      "loss": 1.3547,
+      "step": 3086
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.963445589246752e-06,
+      "loss": 1.3645,
+      "step": 3087
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9611022757845705e-06,
+      "loss": 1.5036,
+      "step": 3088
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.958758743101667e-06,
+      "loss": 1.7928,
+      "step": 3089
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9564149917325845e-06,
+      "loss": 1.7029,
+      "step": 3090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.954071022211916e-06,
+      "loss": 1.4049,
+      "step": 3091
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.951726835074305e-06,
+      "loss": 1.5694,
+      "step": 3092
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9493824308544455e-06,
+      "loss": 1.5703,
+      "step": 3093
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.947037810087077e-06,
+      "loss": 1.0344,
+      "step": 3094
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.944692973306995e-06,
+      "loss": 1.3666,
+      "step": 3095
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.942347921049035e-06,
+      "loss": 1.3299,
+      "step": 3096
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.940002653848093e-06,
+      "loss": 1.3364,
+      "step": 3097
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.937657172239103e-06,
+      "loss": 1.2833,
+      "step": 3098
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.935311476757056e-06,
+      "loss": 1.5687,
+      "step": 3099
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.932965567936987e-06,
+      "loss": 1.5839,
+      "step": 3100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.930619446313983e-06,
+      "loss": 1.4087,
+      "step": 3101
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.928273112423177e-06,
+      "loss": 1.8308,
+      "step": 3102
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.925926566799754e-06,
+      "loss": 1.6064,
+      "step": 3103
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.92357980997894e-06,
+      "loss": 1.3766,
+      "step": 3104
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.921232842496016e-06,
+      "loss": 1.1489,
+      "step": 3105
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.918885664886311e-06,
+      "loss": 1.715,
+      "step": 3106
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.916538277685196e-06,
+      "loss": 1.389,
+      "step": 3107
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.914190681428098e-06,
+      "loss": 1.1458,
+      "step": 3108
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.911842876650483e-06,
+      "loss": 1.2184,
+      "step": 3109
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.909494863887872e-06,
+      "loss": 1.1938,
+      "step": 3110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.907146643675828e-06,
+      "loss": 1.739,
+      "step": 3111
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.904798216549965e-06,
+      "loss": 1.4365,
+      "step": 3112
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9024495830459435e-06,
+      "loss": 1.3783,
+      "step": 3113
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.900100743699469e-06,
+      "loss": 1.1836,
+      "step": 3114
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.897751699046297e-06,
+      "loss": 1.4621,
+      "step": 3115
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.895402449622226e-06,
+      "loss": 1.3076,
+      "step": 3116
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.893052995963105e-06,
+      "loss": 1.3691,
+      "step": 3117
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8907033386048275e-06,
+      "loss": 1.2737,
+      "step": 3118
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.888353478083333e-06,
+      "loss": 1.3731,
+      "step": 3119
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.88600341493461e-06,
+      "loss": 1.6563,
+      "step": 3120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.88365314969469e-06,
+      "loss": 1.169,
+      "step": 3121
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.881302682899654e-06,
+      "loss": 1.4187,
+      "step": 3122
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.878952015085626e-06,
+      "loss": 1.2944,
+      "step": 3123
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.876601146788776e-06,
+      "loss": 1.4102,
+      "step": 3124
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8742500785453226e-06,
+      "loss": 1.5804,
+      "step": 3125
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.871898810891526e-06,
+      "loss": 1.0668,
+      "step": 3126
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.869547344363695e-06,
+      "loss": 1.452,
+      "step": 3127
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.867195679498183e-06,
+      "loss": 1.303,
+      "step": 3128
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.864843816831388e-06,
+      "loss": 1.1936,
+      "step": 3129
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.862491756899753e-06,
+      "loss": 1.5298,
+      "step": 3130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.860139500239768e-06,
+      "loss": 1.7281,
+      "step": 3131
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8577870473879625e-06,
+      "loss": 1.6698,
+      "step": 3132
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.855434398880919e-06,
+      "loss": 1.4384,
+      "step": 3133
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8530815552552585e-06,
+      "loss": 1.5147,
+      "step": 3134
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.850728517047649e-06,
+      "loss": 1.3018,
+      "step": 3135
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8483752847948e-06,
+      "loss": 1.3578,
+      "step": 3136
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.846021859033471e-06,
+      "loss": 1.5474,
+      "step": 3137
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8436682403004575e-06,
+      "loss": 1.7062,
+      "step": 3138
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.841314429132606e-06,
+      "loss": 1.5637,
+      "step": 3139
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.838960426066804e-06,
+      "loss": 1.3247,
+      "step": 3140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.836606231639984e-06,
+      "loss": 1.4776,
+      "step": 3141
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8342518463891195e-06,
+      "loss": 1.398,
+      "step": 3142
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.83189727085123e-06,
+      "loss": 1.3101,
+      "step": 3143
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.82954250556338e-06,
+      "loss": 1.626,
+      "step": 3144
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.827187551062672e-06,
+      "loss": 1.5707,
+      "step": 3145
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8248324078862565e-06,
+      "loss": 1.2316,
+      "step": 3146
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.822477076571324e-06,
+      "loss": 1.7007,
+      "step": 3147
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.820121557655109e-06,
+      "loss": 1.5235,
+      "step": 3148
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.817765851674892e-06,
+      "loss": 1.3134,
+      "step": 3149
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.815409959167989e-06,
+      "loss": 1.393,
+      "step": 3150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.813053880671765e-06,
+      "loss": 1.4829,
+      "step": 3151
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8106976167236236e-06,
+      "loss": 1.1006,
+      "step": 3152
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.808341167861015e-06,
+      "loss": 1.6086,
+      "step": 3153
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.805984534621424e-06,
+      "loss": 1.1479,
+      "step": 3154
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.803627717542386e-06,
+      "loss": 1.4768,
+      "step": 3155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.801270717161474e-06,
+      "loss": 1.7105,
+      "step": 3156
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.798913534016303e-06,
+      "loss": 1.0771,
+      "step": 3157
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.796556168644529e-06,
+      "loss": 1.5987,
+      "step": 3158
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.794198621583853e-06,
+      "loss": 1.3809,
+      "step": 3159
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.791840893372013e-06,
+      "loss": 1.4118,
+      "step": 3160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.789482984546791e-06,
+      "loss": 1.2242,
+      "step": 3161
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.787124895646008e-06,
+      "loss": 1.6086,
+      "step": 3162
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.78476662720753e-06,
+      "loss": 1.6235,
+      "step": 3163
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.78240817976926e-06,
+      "loss": 1.0856,
+      "step": 3164
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.780049553869143e-06,
+      "loss": 1.3177,
+      "step": 3165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.777690750045168e-06,
+      "loss": 1.4507,
+      "step": 3166
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.77533176883536e-06,
+      "loss": 1.4823,
+      "step": 3167
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7729726107777855e-06,
+      "loss": 1.0658,
+      "step": 3168
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7706132764105515e-06,
+      "loss": 1.405,
+      "step": 3169
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7682537662718075e-06,
+      "loss": 1.484,
+      "step": 3170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.76589408089974e-06,
+      "loss": 1.8174,
+      "step": 3171
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7635342208325775e-06,
+      "loss": 1.0757,
+      "step": 3172
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7611741866085866e-06,
+      "loss": 1.6152,
+      "step": 3173
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.758813978766077e-06,
+      "loss": 1.2083,
+      "step": 3174
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.756453597843392e-06,
+      "loss": 0.8938,
+      "step": 3175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.754093044378921e-06,
+      "loss": 1.5486,
+      "step": 3176
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.75173231891109e-06,
+      "loss": 1.3071,
+      "step": 3177
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.749371421978361e-06,
+      "loss": 1.3283,
+      "step": 3178
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7470103541192405e-06,
+      "loss": 1.0415,
+      "step": 3179
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7446491158722716e-06,
+      "loss": 1.2371,
+      "step": 3180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.742287707776034e-06,
+      "loss": 1.4508,
+      "step": 3181
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.739926130369153e-06,
+      "loss": 1.266,
+      "step": 3182
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.73756438419028e-06,
+      "loss": 1.3858,
+      "step": 3183
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.735202469778122e-06,
+      "loss": 1.7736,
+      "step": 3184
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7328403876714105e-06,
+      "loss": 1.5177,
+      "step": 3185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.730478138408919e-06,
+      "loss": 1.1475,
+      "step": 3186
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.728115722529463e-06,
+      "loss": 1.5145,
+      "step": 3187
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.725753140571891e-06,
+      "loss": 1.5238,
+      "step": 3188
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.723390393075092e-06,
+      "loss": 1.0826,
+      "step": 3189
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.721027480577993e-06,
+      "loss": 1.5876,
+      "step": 3190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.718664403619557e-06,
+      "loss": 1.4465,
+      "step": 3191
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.716301162738784e-06,
+      "loss": 1.5276,
+      "step": 3192
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.713937758474717e-06,
+      "loss": 1.6703,
+      "step": 3193
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.711574191366427e-06,
+      "loss": 1.2979,
+      "step": 3194
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.70921046195303e-06,
+      "loss": 1.6218,
+      "step": 3195
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.706846570773677e-06,
+      "loss": 1.2932,
+      "step": 3196
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.704482518367551e-06,
+      "loss": 1.1283,
+      "step": 3197
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.70211830527388e-06,
+      "loss": 1.4724,
+      "step": 3198
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.699753932031923e-06,
+      "loss": 1.5797,
+      "step": 3199
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.697389399180977e-06,
+      "loss": 1.4939,
+      "step": 3200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.695024707260376e-06,
+      "loss": 1.6181,
+      "step": 3201
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.69265985680949e-06,
+      "loss": 1.3426,
+      "step": 3202
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.690294848367724e-06,
+      "loss": 1.6475,
+      "step": 3203
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.687929682474518e-06,
+      "loss": 1.646,
+      "step": 3204
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.685564359669354e-06,
+      "loss": 1.6248,
+      "step": 3205
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.683198880491743e-06,
+      "loss": 1.5506,
+      "step": 3206
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.680833245481234e-06,
+      "loss": 1.1389,
+      "step": 3207
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.678467455177413e-06,
+      "loss": 1.5688,
+      "step": 3208
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6761015101199e-06,
+      "loss": 1.7663,
+      "step": 3209
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.673735410848351e-06,
+      "loss": 1.5926,
+      "step": 3210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.671369157902456e-06,
+      "loss": 1.4291,
+      "step": 3211
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.669002751821942e-06,
+      "loss": 1.3136,
+      "step": 3212
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.666636193146568e-06,
+      "loss": 1.5092,
+      "step": 3213
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6642694824161305e-06,
+      "loss": 1.3706,
+      "step": 3214
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.661902620170458e-06,
+      "loss": 1.3166,
+      "step": 3215
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.659535606949419e-06,
+      "loss": 1.3678,
+      "step": 3216
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.657168443292909e-06,
+      "loss": 1.4375,
+      "step": 3217
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.654801129740863e-06,
+      "loss": 1.3424,
+      "step": 3218
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.652433666833248e-06,
+      "loss": 1.6024,
+      "step": 3219
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.650066055110067e-06,
+      "loss": 1.7761,
+      "step": 3220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.647698295111356e-06,
+      "loss": 1.487,
+      "step": 3221
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.645330387377182e-06,
+      "loss": 1.3707,
+      "step": 3222
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.642962332447651e-06,
+      "loss": 1.4409,
+      "step": 3223
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6405941308629e-06,
+      "loss": 1.6145,
+      "step": 3224
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.638225783163096e-06,
+      "loss": 1.1548,
+      "step": 3225
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6358572898884445e-06,
+      "loss": 1.3608,
+      "step": 3226
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.633488651579183e-06,
+      "loss": 1.3299,
+      "step": 3227
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6311198687755805e-06,
+      "loss": 1.6094,
+      "step": 3228
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.628750942017939e-06,
+      "loss": 1.5518,
+      "step": 3229
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.626381871846597e-06,
+      "loss": 1.5894,
+      "step": 3230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.62401265880192e-06,
+      "loss": 1.5287,
+      "step": 3231
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.62164330342431e-06,
+      "loss": 1.5507,
+      "step": 3232
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6192738062542e-06,
+      "loss": 1.2748,
+      "step": 3233
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.616904167832058e-06,
+      "loss": 1.6978,
+      "step": 3234
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.614534388698379e-06,
+      "loss": 1.0321,
+      "step": 3235
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.612164469393694e-06,
+      "loss": 1.801,
+      "step": 3236
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.609794410458566e-06,
+      "loss": 1.5797,
+      "step": 3237
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.607424212433588e-06,
+      "loss": 1.505,
+      "step": 3238
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.605053875859386e-06,
+      "loss": 1.7692,
+      "step": 3239
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6026834012766155e-06,
+      "loss": 1.5381,
+      "step": 3240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.600312789225968e-06,
+      "loss": 1.424,
+      "step": 3241
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.597942040248163e-06,
+      "loss": 1.1784,
+      "step": 3242
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5955711548839505e-06,
+      "loss": 1.536,
+      "step": 3243
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.593200133674113e-06,
+      "loss": 1.473,
+      "step": 3244
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.590828977159466e-06,
+      "loss": 1.1069,
+      "step": 3245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.588457685880851e-06,
+      "loss": 1.671,
+      "step": 3246
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.586086260379146e-06,
+      "loss": 1.7006,
+      "step": 3247
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.583714701195253e-06,
+      "loss": 1.1276,
+      "step": 3248
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.581343008870111e-06,
+      "loss": 1.3561,
+      "step": 3249
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5789711839446845e-06,
+      "loss": 1.4644,
+      "step": 3250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.576599226959971e-06,
+      "loss": 1.3978,
+      "step": 3251
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.574227138456998e-06,
+      "loss": 1.5426,
+      "step": 3252
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.571854918976821e-06,
+      "loss": 1.2277,
+      "step": 3253
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.569482569060529e-06,
+      "loss": 1.2767,
+      "step": 3254
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.567110089249236e-06,
+      "loss": 1.4803,
+      "step": 3255
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.564737480084089e-06,
+      "loss": 1.2846,
+      "step": 3256
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.562364742106264e-06,
+      "loss": 1.588,
+      "step": 3257
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.559991875856966e-06,
+      "loss": 1.5295,
+      "step": 3258
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.557618881877428e-06,
+      "loss": 0.947,
+      "step": 3259
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.555245760708915e-06,
+      "loss": 1.2466,
+      "step": 3260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.552872512892718e-06,
+      "loss": 1.6138,
+      "step": 3261
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.550499138970158e-06,
+      "loss": 1.6612,
+      "step": 3262
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.548125639482587e-06,
+      "loss": 1.5075,
+      "step": 3263
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.545752014971383e-06,
+      "loss": 1.367,
+      "step": 3264
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5433782659779525e-06,
+      "loss": 1.729,
+      "step": 3265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.541004393043731e-06,
+      "loss": 1.7469,
+      "step": 3266
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.538630396710182e-06,
+      "loss": 1.3629,
+      "step": 3267
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.536256277518799e-06,
+      "loss": 1.4567,
+      "step": 3268
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.533882036011101e-06,
+      "loss": 1.3372,
+      "step": 3269
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5315076727286366e-06,
+      "loss": 1.4222,
+      "step": 3270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.52913318821298e-06,
+      "loss": 1.7158,
+      "step": 3271
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.526758583005736e-06,
+      "loss": 1.4674,
+      "step": 3272
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.524383857648534e-06,
+      "loss": 1.3551,
+      "step": 3273
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.522009012683035e-06,
+      "loss": 1.6629,
+      "step": 3274
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.519634048650922e-06,
+      "loss": 1.6115,
+      "step": 3275
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.51725896609391e-06,
+      "loss": 1.4605,
+      "step": 3276
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5148837655537366e-06,
+      "loss": 1.1564,
+      "step": 3277
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.512508447572169e-06,
+      "loss": 1.6059,
+      "step": 3278
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.510133012691002e-06,
+      "loss": 1.4068,
+      "step": 3279
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.507757461452053e-06,
+      "loss": 1.5686,
+      "step": 3280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.50538179439717e-06,
+      "loss": 1.2508,
+      "step": 3281
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5030060120682275e-06,
+      "loss": 1.3054,
+      "step": 3282
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.500630115007121e-06,
+      "loss": 1.5964,
+      "step": 3283
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4982541037557825e-06,
+      "loss": 1.2043,
+      "step": 3284
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.495877978856159e-06,
+      "loss": 1.4235,
+      "step": 3285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.493501740850228e-06,
+      "loss": 1.5933,
+      "step": 3286
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.491125390279994e-06,
+      "loss": 1.3943,
+      "step": 3287
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.488748927687485e-06,
+      "loss": 1.4183,
+      "step": 3288
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4863723536147575e-06,
+      "loss": 1.0894,
+      "step": 3289
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.48399566860389e-06,
+      "loss": 1.2081,
+      "step": 3290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.481618873196986e-06,
+      "loss": 1.3753,
+      "step": 3291
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.47924196793618e-06,
+      "loss": 1.1865,
+      "step": 3292
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.476864953363624e-06,
+      "loss": 1.6908,
+      "step": 3293
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.474487830021501e-06,
+      "loss": 1.5217,
+      "step": 3294
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.472110598452014e-06,
+      "loss": 1.6444,
+      "step": 3295
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.469733259197395e-06,
+      "loss": 1.3326,
+      "step": 3296
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.467355812799896e-06,
+      "loss": 1.6179,
+      "step": 3297
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.464978259801797e-06,
+      "loss": 1.4822,
+      "step": 3298
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.462600600745404e-06,
+      "loss": 1.5726,
+      "step": 3299
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.460222836173041e-06,
+      "loss": 1.5537,
+      "step": 3300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4578449666270605e-06,
+      "loss": 1.4191,
+      "step": 3301
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.455466992649838e-06,
+      "loss": 1.7444,
+      "step": 3302
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.453088914783773e-06,
+      "loss": 1.4314,
+      "step": 3303
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.450710733571288e-06,
+      "loss": 1.6631,
+      "step": 3304
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.448332449554831e-06,
+      "loss": 1.529,
+      "step": 3305
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.44595406327687e-06,
+      "loss": 1.1427,
+      "step": 3306
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.443575575279898e-06,
+      "loss": 1.3664,
+      "step": 3307
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.441196986106434e-06,
+      "loss": 1.6954,
+      "step": 3308
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.438818296299015e-06,
+      "loss": 1.6162,
+      "step": 3309
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.436439506400206e-06,
+      "loss": 1.773,
+      "step": 3310
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4340606169525915e-06,
+      "loss": 1.8016,
+      "step": 3311
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.431681628498779e-06,
+      "loss": 1.7688,
+      "step": 3312
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.429302541581398e-06,
+      "loss": 1.5014,
+      "step": 3313
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.426923356743102e-06,
+      "loss": 1.247,
+      "step": 3314
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.42454407452657e-06,
+      "loss": 1.6934,
+      "step": 3315
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.422164695474497e-06,
+      "loss": 1.4254,
+      "step": 3316
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.419785220129602e-06,
+      "loss": 1.5497,
+      "step": 3317
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.417405649034628e-06,
+      "loss": 1.3604,
+      "step": 3318
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.415025982732338e-06,
+      "loss": 1.3716,
+      "step": 3319
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.412646221765518e-06,
+      "loss": 1.3157,
+      "step": 3320
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.410266366676974e-06,
+      "loss": 1.1806,
+      "step": 3321
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.407886418009535e-06,
+      "loss": 1.5151,
+      "step": 3322
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.40550637630605e-06,
+      "loss": 1.2036,
+      "step": 3323
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.40312624210939e-06,
+      "loss": 1.5818,
+      "step": 3324
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.400746015962448e-06,
+      "loss": 1.4888,
+      "step": 3325
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.398365698408135e-06,
+      "loss": 1.4248,
+      "step": 3326
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.395985289989387e-06,
+      "loss": 1.2834,
+      "step": 3327
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.393604791249158e-06,
+      "loss": 1.6553,
+      "step": 3328
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3912242027304216e-06,
+      "loss": 1.4882,
+      "step": 3329
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.388843524976175e-06,
+      "loss": 1.5309,
+      "step": 3330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.386462758529435e-06,
+      "loss": 1.0252,
+      "step": 3331
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.384081903933235e-06,
+      "loss": 1.3556,
+      "step": 3332
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.381700961730634e-06,
+      "loss": 1.5545,
+      "step": 3333
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.379319932464706e-06,
+      "loss": 1.5807,
+      "step": 3334
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.37693881667855e-06,
+      "loss": 1.4755,
+      "step": 3335
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.374557614915279e-06,
+      "loss": 1.1932,
+      "step": 3336
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.372176327718029e-06,
+      "loss": 1.3851,
+      "step": 3337
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.369794955629958e-06,
+      "loss": 1.6692,
+      "step": 3338
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.367413499194236e-06,
+      "loss": 1.6851,
+      "step": 3339
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.36503195895406e-06,
+      "loss": 1.6795,
+      "step": 3340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.362650335452642e-06,
+      "loss": 1.7278,
+      "step": 3341
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.360268629233212e-06,
+      "loss": 1.657,
+      "step": 3342
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.357886840839021e-06,
+      "loss": 1.134,
+      "step": 3343
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.355504970813341e-06,
+      "loss": 1.3233,
+      "step": 3344
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.353123019699456e-06,
+      "loss": 1.3064,
+      "step": 3345
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3507409880406765e-06,
+      "loss": 1.282,
+      "step": 3346
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.348358876380324e-06,
+      "loss": 1.5413,
+      "step": 3347
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.345976685261741e-06,
+      "loss": 1.423,
+      "step": 3348
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.343594415228293e-06,
+      "loss": 1.3772,
+      "step": 3349
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.341212066823356e-06,
+      "loss": 1.3234,
+      "step": 3350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.338829640590327e-06,
+      "loss": 1.588,
+      "step": 3351
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.336447137072621e-06,
+      "loss": 1.4609,
+      "step": 3352
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.334064556813672e-06,
+      "loss": 1.4707,
+      "step": 3353
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.331681900356929e-06,
+      "loss": 1.3116,
+      "step": 3354
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3292991682458576e-06,
+      "loss": 1.492,
+      "step": 3355
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.326916361023945e-06,
+      "loss": 1.3692,
+      "step": 3356
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.324533479234691e-06,
+      "loss": 1.5416,
+      "step": 3357
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3221505234216156e-06,
+      "loss": 1.6428,
+      "step": 3358
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.319767494128252e-06,
+      "loss": 1.28,
+      "step": 3359
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.317384391898156e-06,
+      "loss": 1.4255,
+      "step": 3360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.315001217274896e-06,
+      "loss": 1.5429,
+      "step": 3361
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.312617970802054e-06,
+      "loss": 1.6834,
+      "step": 3362
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3102346530232365e-06,
+      "loss": 1.5838,
+      "step": 3363
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.307851264482059e-06,
+      "loss": 1.6356,
+      "step": 3364
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.305467805722155e-06,
+      "loss": 1.3176,
+      "step": 3365
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.303084277287179e-06,
+      "loss": 1.8121,
+      "step": 3366
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3007006797207915e-06,
+      "loss": 1.2626,
+      "step": 3367
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.298317013566678e-06,
+      "loss": 1.3152,
+      "step": 3368
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.295933279368536e-06,
+      "loss": 1.6627,
+      "step": 3369
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.293549477670077e-06,
+      "loss": 1.8853,
+      "step": 3370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.291165609015032e-06,
+      "loss": 1.3206,
+      "step": 3371
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.288781673947143e-06,
+      "loss": 1.3954,
+      "step": 3372
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.286397673010169e-06,
+      "loss": 1.4482,
+      "step": 3373
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2840136067478845e-06,
+      "loss": 1.324,
+      "step": 3374
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.281629475704077e-06,
+      "loss": 1.5854,
+      "step": 3375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2792452804225535e-06,
+      "loss": 1.4318,
+      "step": 3376
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.276861021447129e-06,
+      "loss": 1.3733,
+      "step": 3377
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.274476699321638e-06,
+      "loss": 1.3886,
+      "step": 3378
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.272092314589927e-06,
+      "loss": 1.8044,
+      "step": 3379
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2697078677958565e-06,
+      "loss": 1.382,
+      "step": 3380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.267323359483305e-06,
+      "loss": 1.0615,
+      "step": 3381
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.26493879019616e-06,
+      "loss": 1.4138,
+      "step": 3382
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.262554160478326e-06,
+      "loss": 1.1862,
+      "step": 3383
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.260169470873721e-06,
+      "loss": 1.57,
+      "step": 3384
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2577847219262735e-06,
+      "loss": 1.4027,
+      "step": 3385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.255399914179932e-06,
+      "loss": 1.2498,
+      "step": 3386
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.25301504817865e-06,
+      "loss": 1.7015,
+      "step": 3387
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.250630124466402e-06,
+      "loss": 1.5025,
+      "step": 3388
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.248245143587172e-06,
+      "loss": 1.8526,
+      "step": 3389
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2458601060849565e-06,
+      "loss": 1.4657,
+      "step": 3390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2434750125037655e-06,
+      "loss": 1.3099,
+      "step": 3391
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.241089863387625e-06,
+      "loss": 1.5228,
+      "step": 3392
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.238704659280568e-06,
+      "loss": 1.7917,
+      "step": 3393
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2363194007266435e-06,
+      "loss": 1.3518,
+      "step": 3394
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.233934088269914e-06,
+      "loss": 1.5872,
+      "step": 3395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.231548722454451e-06,
+      "loss": 1.5064,
+      "step": 3396
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.22916330382434e-06,
+      "loss": 1.4283,
+      "step": 3397
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.22677783292368e-06,
+      "loss": 1.6228,
+      "step": 3398
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2243923102965765e-06,
+      "loss": 1.6005,
+      "step": 3399
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.222006736487154e-06,
+      "loss": 1.6593,
+      "step": 3400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.219621112039545e-06,
+      "loss": 1.0941,
+      "step": 3401
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2172354374978905e-06,
+      "loss": 1.4649,
+      "step": 3402
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2148497134063495e-06,
+      "loss": 1.2061,
+      "step": 3403
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.21246394030909e-06,
+      "loss": 1.1047,
+      "step": 3404
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.210078118750286e-06,
+      "loss": 1.4722,
+      "step": 3405
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.20769224927413e-06,
+      "loss": 1.4492,
+      "step": 3406
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.205306332424821e-06,
+      "loss": 1.4612,
+      "step": 3407
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.202920368746571e-06,
+      "loss": 1.3129,
+      "step": 3408
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2005343587836e-06,
+      "loss": 1.3623,
+      "step": 3409
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.198148303080142e-06,
+      "loss": 1.5593,
+      "step": 3410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1957622021804385e-06,
+      "loss": 1.4663,
+      "step": 3411
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.193376056628743e-06,
+      "loss": 1.4334,
+      "step": 3412
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.190989866969317e-06,
+      "loss": 1.3311,
+      "step": 3413
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1886036337464365e-06,
+      "loss": 1.3778,
+      "step": 3414
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.186217357504382e-06,
+      "loss": 1.2662,
+      "step": 3415
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.183831038787449e-06,
+      "loss": 1.3168,
+      "step": 3416
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.181444678139939e-06,
+      "loss": 1.5973,
+      "step": 3417
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.179058276106163e-06,
+      "loss": 1.2861,
+      "step": 3418
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.176671833230444e-06,
+      "loss": 1.6184,
+      "step": 3419
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.174285350057113e-06,
+      "loss": 1.475,
+      "step": 3420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1718988271305105e-06,
+      "loss": 1.7972,
+      "step": 3421
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.169512264994984e-06,
+      "loss": 1.4378,
+      "step": 3422
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.167125664194893e-06,
+      "loss": 1.3141,
+      "step": 3423
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.164739025274604e-06,
+      "loss": 1.2934,
+      "step": 3424
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.162352348778493e-06,
+      "loss": 1.6038,
+      "step": 3425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.159965635250945e-06,
+      "loss": 1.3315,
+      "step": 3426
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.157578885236352e-06,
+      "loss": 1.504,
+      "step": 3427
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.155192099279113e-06,
+      "loss": 1.3467,
+      "step": 3428
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.152805277923641e-06,
+      "loss": 1.5949,
+      "step": 3429
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.150418421714349e-06,
+      "loss": 1.8096,
+      "step": 3430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.148031531195668e-06,
+      "loss": 1.1447,
+      "step": 3431
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.145644606912025e-06,
+      "loss": 1.0269,
+      "step": 3432
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1432576494078645e-06,
+      "loss": 1.329,
+      "step": 3433
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.140870659227635e-06,
+      "loss": 1.5485,
+      "step": 3434
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1384836369157895e-06,
+      "loss": 1.4974,
+      "step": 3435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.136096583016793e-06,
+      "loss": 1.7049,
+      "step": 3436
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.133709498075117e-06,
+      "loss": 1.2904,
+      "step": 3437
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.131322382635236e-06,
+      "loss": 1.0451,
+      "step": 3438
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1289352372416364e-06,
+      "loss": 1.2006,
+      "step": 3439
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.126548062438809e-06,
+      "loss": 1.3847,
+      "step": 3440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.124160858771252e-06,
+      "loss": 1.485,
+      "step": 3441
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1217736267834696e-06,
+      "loss": 1.7034,
+      "step": 3442
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.119386367019972e-06,
+      "loss": 1.6606,
+      "step": 3443
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.116999080025277e-06,
+      "loss": 1.4272,
+      "step": 3444
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.114611766343908e-06,
+      "loss": 1.424,
+      "step": 3445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.112224426520396e-06,
+      "loss": 1.2673,
+      "step": 3446
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.109837061099274e-06,
+      "loss": 1.596,
+      "step": 3447
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.107449670625086e-06,
+      "loss": 1.232,
+      "step": 3448
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1050622556423775e-06,
+      "loss": 1.8298,
+      "step": 3449
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.102674816695703e-06,
+      "loss": 1.5818,
+      "step": 3450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.100287354329618e-06,
+      "loss": 1.3568,
+      "step": 3451
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.097899869088687e-06,
+      "loss": 1.5709,
+      "step": 3452
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0955123615174795e-06,
+      "loss": 1.5385,
+      "step": 3453
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.093124832160569e-06,
+      "loss": 1.3322,
+      "step": 3454
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.090737281562534e-06,
+      "loss": 1.3843,
+      "step": 3455
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.088349710267958e-06,
+      "loss": 1.5111,
+      "step": 3456
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.085962118821429e-06,
+      "loss": 1.246,
+      "step": 3457
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.083574507767542e-06,
+      "loss": 1.7053,
+      "step": 3458
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.081186877650892e-06,
+      "loss": 1.2732,
+      "step": 3459
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.078799229016083e-06,
+      "loss": 1.5318,
+      "step": 3460
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.076411562407718e-06,
+      "loss": 1.4935,
+      "step": 3461
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.07402387837041e-06,
+      "loss": 1.4883,
+      "step": 3462
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.071636177448772e-06,
+      "loss": 1.4897,
+      "step": 3463
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.069248460187422e-06,
+      "loss": 1.6626,
+      "step": 3464
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.06686072713098e-06,
+      "loss": 1.2123,
+      "step": 3465
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0644729788240745e-06,
+      "loss": 1.2802,
+      "step": 3466
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.06208521581133e-06,
+      "loss": 1.5844,
+      "step": 3467
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.059697438637382e-06,
+      "loss": 1.2985,
+      "step": 3468
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.057309647846867e-06,
+      "loss": 1.2668,
+      "step": 3469
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0549218439844185e-06,
+      "loss": 1.1247,
+      "step": 3470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0525340275946826e-06,
+      "loss": 1.4872,
+      "step": 3471
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0501461992223e-06,
+      "loss": 1.5351,
+      "step": 3472
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.04775835941192e-06,
+      "loss": 1.4764,
+      "step": 3473
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.045370508708192e-06,
+      "loss": 1.7314,
+      "step": 3474
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.042982647655766e-06,
+      "loss": 1.5404,
+      "step": 3475
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0405947767993e-06,
+      "loss": 1.2158,
+      "step": 3476
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.038206896683446e-06,
+      "loss": 1.2259,
+      "step": 3477
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.035819007852868e-06,
+      "loss": 1.5259,
+      "step": 3478
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.033431110852224e-06,
+      "loss": 1.5309,
+      "step": 3479
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0310432062261764e-06,
+      "loss": 1.2898,
+      "step": 3480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.02865529451939e-06,
+      "loss": 1.3941,
+      "step": 3481
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0262673762765316e-06,
+      "loss": 1.6332,
+      "step": 3482
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.023879452042266e-06,
+      "loss": 1.3176,
+      "step": 3483
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.021491522361266e-06,
+      "loss": 1.4288,
+      "step": 3484
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.019103587778199e-06,
+      "loss": 1.4625,
+      "step": 3485
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.016715648837737e-06,
+      "loss": 1.4318,
+      "step": 3486
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.014327706084553e-06,
+      "loss": 1.6479,
+      "step": 3487
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.011939760063317e-06,
+      "loss": 1.5672,
+      "step": 3488
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.009551811318707e-06,
+      "loss": 1.4533,
+      "step": 3489
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.007163860395394e-06,
+      "loss": 1.2752,
+      "step": 3490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.004775907838057e-06,
+      "loss": 1.459,
+      "step": 3491
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.002387954191367e-06,
+      "loss": 1.2682,
+      "step": 3492
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5e-06,
+      "loss": 1.4283,
+      "step": 3493
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.997612045808634e-06,
+      "loss": 1.2578,
+      "step": 3494
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.995224092161946e-06,
+      "loss": 1.6361,
+      "step": 3495
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.992836139604606e-06,
+      "loss": 1.3536,
+      "step": 3496
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.990448188681295e-06,
+      "loss": 1.3158,
+      "step": 3497
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.988060239936684e-06,
+      "loss": 1.386,
+      "step": 3498
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.98567229391545e-06,
+      "loss": 1.4979,
+      "step": 3499
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.983284351162264e-06,
+      "loss": 1.5124,
+      "step": 3500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.980896412221803e-06,
+      "loss": 1.3982,
+      "step": 3501
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.978508477638736e-06,
+      "loss": 1.3376,
+      "step": 3502
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.976120547957736e-06,
+      "loss": 1.4913,
+      "step": 3503
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.973732623723471e-06,
+      "loss": 1.7194,
+      "step": 3504
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.97134470548061e-06,
+      "loss": 1.4979,
+      "step": 3505
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.968956793773825e-06,
+      "loss": 1.6163,
+      "step": 3506
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.966568889147776e-06,
+      "loss": 1.3649,
+      "step": 3507
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.964180992147133e-06,
+      "loss": 1.4486,
+      "step": 3508
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.961793103316554e-06,
+      "loss": 1.6313,
+      "step": 3509
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.959405223200703e-06,
+      "loss": 1.5193,
+      "step": 3510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.957017352344235e-06,
+      "loss": 1.1311,
+      "step": 3511
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.954629491291811e-06,
+      "loss": 1.2287,
+      "step": 3512
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9522416405880815e-06,
+      "loss": 1.3554,
+      "step": 3513
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.949853800777702e-06,
+      "loss": 1.5741,
+      "step": 3514
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.94746597240532e-06,
+      "loss": 1.4949,
+      "step": 3515
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.945078156015582e-06,
+      "loss": 1.4979,
+      "step": 3516
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.942690352153136e-06,
+      "loss": 1.5733,
+      "step": 3517
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.940302561362618e-06,
+      "loss": 1.3485,
+      "step": 3518
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9379147841886715e-06,
+      "loss": 1.6603,
+      "step": 3519
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.935527021175928e-06,
+      "loss": 1.6437,
+      "step": 3520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.933139272869022e-06,
+      "loss": 1.3588,
+      "step": 3521
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.93075153981258e-06,
+      "loss": 1.1962,
+      "step": 3522
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928363822551231e-06,
+      "loss": 1.5892,
+      "step": 3523
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.925976121629591e-06,
+      "loss": 1.0417,
+      "step": 3524
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.923588437592284e-06,
+      "loss": 1.638,
+      "step": 3525
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.921200770983919e-06,
+      "loss": 1.062,
+      "step": 3526
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.918813122349108e-06,
+      "loss": 1.6922,
+      "step": 3527
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9164254922324595e-06,
+      "loss": 1.6396,
+      "step": 3528
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.914037881178571e-06,
+      "loss": 1.1144,
+      "step": 3529
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.911650289732044e-06,
+      "loss": 1.2628,
+      "step": 3530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.909262718437467e-06,
+      "loss": 1.8791,
+      "step": 3531
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.906875167839433e-06,
+      "loss": 1.2699,
+      "step": 3532
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.904487638482521e-06,
+      "loss": 1.2987,
+      "step": 3533
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9021001309113154e-06,
+      "loss": 1.4142,
+      "step": 3534
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.899712645670383e-06,
+      "loss": 1.4655,
+      "step": 3535
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8973251833043e-06,
+      "loss": 1.2437,
+      "step": 3536
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.894937744357623e-06,
+      "loss": 1.5892,
+      "step": 3537
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.892550329374915e-06,
+      "loss": 1.5565,
+      "step": 3538
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8901629389007274e-06,
+      "loss": 1.4568,
+      "step": 3539
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.887775573479606e-06,
+      "loss": 1.9801,
+      "step": 3540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.885388233656094e-06,
+      "loss": 1.5132,
+      "step": 3541
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.883000919974724e-06,
+      "loss": 1.3115,
+      "step": 3542
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8806136329800305e-06,
+      "loss": 1.364,
+      "step": 3543
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.878226373216532e-06,
+      "loss": 1.1084,
+      "step": 3544
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.875839141228751e-06,
+      "loss": 1.6258,
+      "step": 3545
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.873451937561192e-06,
+      "loss": 1.7221,
+      "step": 3546
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.871064762758365e-06,
+      "loss": 0.9525,
+      "step": 3547
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8686776173647655e-06,
+      "loss": 1.2393,
+      "step": 3548
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.866290501924885e-06,
+      "loss": 1.3743,
+      "step": 3549
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.863903416983208e-06,
+      "loss": 1.7779,
+      "step": 3550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.861516363084212e-06,
+      "loss": 1.7415,
+      "step": 3551
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.859129340772367e-06,
+      "loss": 1.1701,
+      "step": 3552
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.856742350592136e-06,
+      "loss": 1.5256,
+      "step": 3553
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.854355393087976e-06,
+      "loss": 1.5189,
+      "step": 3554
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.851968468804335e-06,
+      "loss": 1.3396,
+      "step": 3555
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.849581578285652e-06,
+      "loss": 1.0012,
+      "step": 3556
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.847194722076362e-06,
+      "loss": 1.7024,
+      "step": 3557
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.844807900720888e-06,
+      "loss": 1.4853,
+      "step": 3558
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.84242111476365e-06,
+      "loss": 1.2841,
+      "step": 3559
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.840034364749056e-06,
+      "loss": 1.4873,
+      "step": 3560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.837647651221508e-06,
+      "loss": 1.507,
+      "step": 3561
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.835260974725397e-06,
+      "loss": 1.2257,
+      "step": 3562
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.832874335805109e-06,
+      "loss": 1.4729,
+      "step": 3563
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.830487735005018e-06,
+      "loss": 1.6597,
+      "step": 3564
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.828101172869491e-06,
+      "loss": 1.3637,
+      "step": 3565
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.825714649942888e-06,
+      "loss": 1.1612,
+      "step": 3566
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.823328166769557e-06,
+      "loss": 1.3869,
+      "step": 3567
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.820941723893838e-06,
+      "loss": 1.4492,
+      "step": 3568
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.818555321860062e-06,
+      "loss": 1.228,
+      "step": 3569
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.816168961212553e-06,
+      "loss": 1.5189,
+      "step": 3570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.813782642495618e-06,
+      "loss": 1.362,
+      "step": 3571
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.811396366253566e-06,
+      "loss": 1.3954,
+      "step": 3572
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.809010133030684e-06,
+      "loss": 1.722,
+      "step": 3573
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.80662394337126e-06,
+      "loss": 1.3494,
+      "step": 3574
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.804237797819563e-06,
+      "loss": 1.4653,
+      "step": 3575
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.801851696919861e-06,
+      "loss": 1.6277,
+      "step": 3576
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7994656412164006e-06,
+      "loss": 1.3824,
+      "step": 3577
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.797079631253431e-06,
+      "loss": 1.4455,
+      "step": 3578
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7946936675751794e-06,
+      "loss": 1.2473,
+      "step": 3579
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.79230775072587e-06,
+      "loss": 1.5429,
+      "step": 3580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.789921881249715e-06,
+      "loss": 1.6224,
+      "step": 3581
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.787536059690911e-06,
+      "loss": 1.1906,
+      "step": 3582
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.785150286593651e-06,
+      "loss": 1.548,
+      "step": 3583
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.78276456250211e-06,
+      "loss": 1.4979,
+      "step": 3584
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7803788879604585e-06,
+      "loss": 1.4836,
+      "step": 3585
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.777993263512848e-06,
+      "loss": 1.4797,
+      "step": 3586
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.775607689703426e-06,
+      "loss": 1.4584,
+      "step": 3587
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.773222167076323e-06,
+      "loss": 1.4274,
+      "step": 3588
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.770836696175662e-06,
+      "loss": 1.3996,
+      "step": 3589
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.76845127754555e-06,
+      "loss": 1.5337,
+      "step": 3590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.766065911730086e-06,
+      "loss": 1.5777,
+      "step": 3591
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.763680599273357e-06,
+      "loss": 1.3615,
+      "step": 3592
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.761295340719433e-06,
+      "loss": 1.5371,
+      "step": 3593
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.758910136612377e-06,
+      "loss": 1.5195,
+      "step": 3594
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7565249874962345e-06,
+      "loss": 1.4119,
+      "step": 3595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.754139893915046e-06,
+      "loss": 1.4654,
+      "step": 3596
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.75175485641283e-06,
+      "loss": 1.4442,
+      "step": 3597
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7493698755336e-06,
+      "loss": 1.5713,
+      "step": 3598
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7469849518213515e-06,
+      "loss": 1.1715,
+      "step": 3599
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7446000858200715e-06,
+      "loss": 0.9956,
+      "step": 3600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.742215278073727e-06,
+      "loss": 1.2682,
+      "step": 3601
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.73983052912628e-06,
+      "loss": 1.2716,
+      "step": 3602
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7374458395216746e-06,
+      "loss": 1.4202,
+      "step": 3603
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.73506120980384e-06,
+      "loss": 1.2499,
+      "step": 3604
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7326766405166964e-06,
+      "loss": 1.5015,
+      "step": 3605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7302921322041435e-06,
+      "loss": 1.6246,
+      "step": 3606
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.727907685410076e-06,
+      "loss": 1.5776,
+      "step": 3607
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7255233006783626e-06,
+      "loss": 1.6865,
+      "step": 3608
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.723138978552873e-06,
+      "loss": 1.5076,
+      "step": 3609
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.720754719577448e-06,
+      "loss": 1.8383,
+      "step": 3610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.718370524295924e-06,
+      "loss": 1.3175,
+      "step": 3611
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.715986393252117e-06,
+      "loss": 1.4505,
+      "step": 3612
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.713602326989831e-06,
+      "loss": 1.3668,
+      "step": 3613
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.711218326052859e-06,
+      "loss": 0.9939,
+      "step": 3614
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.708834390984969e-06,
+      "loss": 1.575,
+      "step": 3615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.706450522329924e-06,
+      "loss": 1.5825,
+      "step": 3616
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.704066720631465e-06,
+      "loss": 1.5198,
+      "step": 3617
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.701682986433324e-06,
+      "loss": 1.6455,
+      "step": 3618
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.69929932027921e-06,
+      "loss": 1.3558,
+      "step": 3619
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.696915722712824e-06,
+      "loss": 1.7823,
+      "step": 3620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.694532194277846e-06,
+      "loss": 1.4733,
+      "step": 3621
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.692148735517944e-06,
+      "loss": 1.4093,
+      "step": 3622
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.689765346976765e-06,
+      "loss": 1.4331,
+      "step": 3623
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.687382029197946e-06,
+      "loss": 1.5886,
+      "step": 3624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.684998782725107e-06,
+      "loss": 1.3201,
+      "step": 3625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6826156081018444e-06,
+      "loss": 1.5422,
+      "step": 3626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.680232505871749e-06,
+      "loss": 1.3252,
+      "step": 3627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.677849476578386e-06,
+      "loss": 1.9479,
+      "step": 3628
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.675466520765311e-06,
+      "loss": 1.7833,
+      "step": 3629
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.673083638976057e-06,
+      "loss": 1.5471,
+      "step": 3630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.670700831754145e-06,
+      "loss": 1.4997,
+      "step": 3631
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.668318099643073e-06,
+      "loss": 1.3401,
+      "step": 3632
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.665935443186331e-06,
+      "loss": 1.5682,
+      "step": 3633
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.66355286292738e-06,
+      "loss": 1.4675,
+      "step": 3634
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6611703594096735e-06,
+      "loss": 1.5573,
+      "step": 3635
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6587879331766465e-06,
+      "loss": 1.2533,
+      "step": 3636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.656405584771708e-06,
+      "loss": 1.3571,
+      "step": 3637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.65402331473826e-06,
+      "loss": 1.6752,
+      "step": 3638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.651641123619677e-06,
+      "loss": 1.4806,
+      "step": 3639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.649259011959326e-06,
+      "loss": 1.4314,
+      "step": 3640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6468769803005445e-06,
+      "loss": 1.1052,
+      "step": 3641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.644495029186661e-06,
+      "loss": 1.5411,
+      "step": 3642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6421131591609794e-06,
+      "loss": 1.5244,
+      "step": 3643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.639731370766788e-06,
+      "loss": 1.38,
+      "step": 3644
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.63734966454736e-06,
+      "loss": 1.7126,
+      "step": 3645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.634968041045941e-06,
+      "loss": 1.5401,
+      "step": 3646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6325865008057656e-06,
+      "loss": 1.4371,
+      "step": 3647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.630205044370043e-06,
+      "loss": 1.4845,
+      "step": 3648
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.627823672281972e-06,
+      "loss": 1.5646,
+      "step": 3649
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6254423850847225e-06,
+      "loss": 1.7846,
+      "step": 3650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6230611833214526e-06,
+      "loss": 1.6442,
+      "step": 3651
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.620680067535295e-06,
+      "loss": 1.5127,
+      "step": 3652
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.618299038269369e-06,
+      "loss": 1.415,
+      "step": 3653
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.615918096066766e-06,
+      "loss": 1.5535,
+      "step": 3654
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.613537241470566e-06,
+      "loss": 1.3177,
+      "step": 3655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.611156475023826e-06,
+      "loss": 1.372,
+      "step": 3656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.608775797269578e-06,
+      "loss": 1.3508,
+      "step": 3657
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.606395208750844e-06,
+      "loss": 1.3594,
+      "step": 3658
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.604014710010614e-06,
+      "loss": 1.7094,
+      "step": 3659
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.601634301591867e-06,
+      "loss": 1.3014,
+      "step": 3660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.599253984037553e-06,
+      "loss": 1.4108,
+      "step": 3661
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.596873757890612e-06,
+      "loss": 1.3539,
+      "step": 3662
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.594493623693951e-06,
+      "loss": 1.3528,
+      "step": 3663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5921135819904675e-06,
+      "loss": 1.556,
+      "step": 3664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.589733633323028e-06,
+      "loss": 1.352,
+      "step": 3665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5873537782344834e-06,
+      "loss": 1.4011,
+      "step": 3666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5849740172676635e-06,
+      "loss": 1.5576,
+      "step": 3667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.582594350965374e-06,
+      "loss": 1.3514,
+      "step": 3668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.580214779870399e-06,
+      "loss": 1.0286,
+      "step": 3669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.577835304525505e-06,
+      "loss": 1.7018,
+      "step": 3670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.575455925473432e-06,
+      "loss": 1.539,
+      "step": 3671
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.573076643256899e-06,
+      "loss": 1.6254,
+      "step": 3672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5706974584186045e-06,
+      "loss": 1.0338,
+      "step": 3673
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.568318371501224e-06,
+      "loss": 1.5204,
+      "step": 3674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.565939383047411e-06,
+      "loss": 1.6947,
+      "step": 3675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5635604935997954e-06,
+      "loss": 1.5894,
+      "step": 3676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.561181703700986e-06,
+      "loss": 1.61,
+      "step": 3677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.558803013893568e-06,
+      "loss": 1.2514,
+      "step": 3678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.556424424720103e-06,
+      "loss": 1.5026,
+      "step": 3679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.554045936723132e-06,
+      "loss": 1.6391,
+      "step": 3680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.551667550445171e-06,
+      "loss": 1.4178,
+      "step": 3681
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.549289266428713e-06,
+      "loss": 1.5317,
+      "step": 3682
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.546911085216229e-06,
+      "loss": 1.3723,
+      "step": 3683
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.544533007350163e-06,
+      "loss": 1.2715,
+      "step": 3684
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.542155033372941e-06,
+      "loss": 1.2533,
+      "step": 3685
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.53977716382696e-06,
+      "loss": 1.3869,
+      "step": 3686
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.537399399254598e-06,
+      "loss": 1.466,
+      "step": 3687
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.535021740198202e-06,
+      "loss": 1.4891,
+      "step": 3688
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5326441872001055e-06,
+      "loss": 1.4472,
+      "step": 3689
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.530266740802607e-06,
+      "loss": 1.3401,
+      "step": 3690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5278894015479866e-06,
+      "loss": 1.3943,
+      "step": 3691
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.525512169978501e-06,
+      "loss": 1.4993,
+      "step": 3692
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.523135046636377e-06,
+      "loss": 1.2614,
+      "step": 3693
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.520758032063822e-06,
+      "loss": 1.4042,
+      "step": 3694
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5183811268030145e-06,
+      "loss": 1.1454,
+      "step": 3695
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.516004331396112e-06,
+      "loss": 1.59,
+      "step": 3696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.513627646385244e-06,
+      "loss": 1.5989,
+      "step": 3697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.511251072312516e-06,
+      "loss": 1.6168,
+      "step": 3698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.508874609720007e-06,
+      "loss": 1.5282,
+      "step": 3699
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.506498259149774e-06,
+      "loss": 1.3565,
+      "step": 3700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.504122021143842e-06,
+      "loss": 1.3826,
+      "step": 3701
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.501745896244219e-06,
+      "loss": 1.5201,
+      "step": 3702
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.499369884992878e-06,
+      "loss": 1.2535,
+      "step": 3703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.496993987931775e-06,
+      "loss": 1.3908,
+      "step": 3704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.494618205602832e-06,
+      "loss": 1.5512,
+      "step": 3705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.49224253854795e-06,
+      "loss": 1.2871,
+      "step": 3706
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.489866987309001e-06,
+      "loss": 1.5797,
+      "step": 3707
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.487491552427833e-06,
+      "loss": 1.7426,
+      "step": 3708
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.485116234446265e-06,
+      "loss": 1.6139,
+      "step": 3709
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.482741033906091e-06,
+      "loss": 1.3577,
+      "step": 3710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4803659513490785e-06,
+      "loss": 1.1639,
+      "step": 3711
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4779909873169646e-06,
+      "loss": 1.5094,
+      "step": 3712
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4756161423514665e-06,
+      "loss": 1.293,
+      "step": 3713
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.473241416994265e-06,
+      "loss": 1.821,
+      "step": 3714
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.470866811787022e-06,
+      "loss": 1.5792,
+      "step": 3715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.468492327271365e-06,
+      "loss": 1.2727,
+      "step": 3716
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.466117963988901e-06,
+      "loss": 1.9628,
+      "step": 3717
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.463743722481202e-06,
+      "loss": 1.5278,
+      "step": 3718
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.461369603289821e-06,
+      "loss": 1.6258,
+      "step": 3719
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.458995606956271e-06,
+      "loss": 1.6017,
+      "step": 3720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.456621734022048e-06,
+      "loss": 1.5979,
+      "step": 3721
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.454247985028619e-06,
+      "loss": 1.2979,
+      "step": 3722
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.451874360517413e-06,
+      "loss": 1.3539,
+      "step": 3723
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4495008610298435e-06,
+      "loss": 1.3417,
+      "step": 3724
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.447127487107283e-06,
+      "loss": 1.1016,
+      "step": 3725
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.444754239291088e-06,
+      "loss": 1.5868,
+      "step": 3726
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.442381118122573e-06,
+      "loss": 1.6307,
+      "step": 3727
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.440008124143037e-06,
+      "loss": 1.5686,
+      "step": 3728
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4376352578937375e-06,
+      "loss": 1.5323,
+      "step": 3729
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4352625199159135e-06,
+      "loss": 1.3584,
+      "step": 3730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4328899107507655e-06,
+      "loss": 1.6556,
+      "step": 3731
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4305174309394714e-06,
+      "loss": 1.1276,
+      "step": 3732
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4281450810231794e-06,
+      "loss": 1.7147,
+      "step": 3733
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.425772861543002e-06,
+      "loss": 1.6687,
+      "step": 3734
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.423400773040031e-06,
+      "loss": 1.0528,
+      "step": 3735
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.421028816055316e-06,
+      "loss": 1.2037,
+      "step": 3736
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.418656991129892e-06,
+      "loss": 1.385,
+      "step": 3737
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4162852988047484e-06,
+      "loss": 1.8003,
+      "step": 3738
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.413913739620857e-06,
+      "loss": 1.3797,
+      "step": 3739
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.41154231411915e-06,
+      "loss": 1.5111,
+      "step": 3740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.409171022840536e-06,
+      "loss": 1.1435,
+      "step": 3741
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4067998663258875e-06,
+      "loss": 1.5467,
+      "step": 3742
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.40442884511605e-06,
+      "loss": 1.3427,
+      "step": 3743
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4020579597518385e-06,
+      "loss": 1.2212,
+      "step": 3744
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.399687210774032e-06,
+      "loss": 1.3315,
+      "step": 3745
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.397316598723385e-06,
+      "loss": 1.3443,
+      "step": 3746
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.394946124140615e-06,
+      "loss": 1.707,
+      "step": 3747
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.392575787566415e-06,
+      "loss": 1.3371,
+      "step": 3748
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.390205589541435e-06,
+      "loss": 1.4584,
+      "step": 3749
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.387835530606308e-06,
+      "loss": 1.4186,
+      "step": 3750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.385465611301622e-06,
+      "loss": 1.5028,
+      "step": 3751
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.383095832167943e-06,
+      "loss": 1.3327,
+      "step": 3752
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3807261937458005e-06,
+      "loss": 1.5033,
+      "step": 3753
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.37835669657569e-06,
+      "loss": 1.5596,
+      "step": 3754
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.375987341198082e-06,
+      "loss": 1.6184,
+      "step": 3755
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.373618128153404e-06,
+      "loss": 1.4808,
+      "step": 3756
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.371249057982062e-06,
+      "loss": 1.5184,
+      "step": 3757
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.36888013122442e-06,
+      "loss": 1.4025,
+      "step": 3758
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.366511348420819e-06,
+      "loss": 1.2176,
+      "step": 3759
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.364142710111556e-06,
+      "loss": 1.0974,
+      "step": 3760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.361774216836907e-06,
+      "loss": 1.3736,
+      "step": 3761
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.359405869137103e-06,
+      "loss": 1.4191,
+      "step": 3762
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.357037667552349e-06,
+      "loss": 1.5127,
+      "step": 3763
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3546696126228185e-06,
+      "loss": 1.5724,
+      "step": 3764
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.352301704888645e-06,
+      "loss": 1.4487,
+      "step": 3765
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.349933944889934e-06,
+      "loss": 1.7224,
+      "step": 3766
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.347566333166753e-06,
+      "loss": 1.4467,
+      "step": 3767
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.345198870259139e-06,
+      "loss": 1.6309,
+      "step": 3768
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.342831556707093e-06,
+      "loss": 1.3012,
+      "step": 3769
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.340464393050584e-06,
+      "loss": 1.772,
+      "step": 3770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3380973798295425e-06,
+      "loss": 1.2822,
+      "step": 3771
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.335730517583873e-06,
+      "loss": 1.2873,
+      "step": 3772
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.333363806853433e-06,
+      "loss": 1.6235,
+      "step": 3773
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.330997248178059e-06,
+      "loss": 1.4521,
+      "step": 3774
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3286308420975445e-06,
+      "loss": 1.4938,
+      "step": 3775
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.326264589151649e-06,
+      "loss": 1.4529,
+      "step": 3776
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.323898489880101e-06,
+      "loss": 1.6648,
+      "step": 3777
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.321532544822588e-06,
+      "loss": 1.5276,
+      "step": 3778
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.319166754518768e-06,
+      "loss": 1.1441,
+      "step": 3779
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.316801119508259e-06,
+      "loss": 1.3035,
+      "step": 3780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.314435640330649e-06,
+      "loss": 1.4625,
+      "step": 3781
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.312070317525483e-06,
+      "loss": 1.5348,
+      "step": 3782
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.30970515163228e-06,
+      "loss": 1.8174,
+      "step": 3783
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.307340143190512e-06,
+      "loss": 1.5351,
+      "step": 3784
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.304975292739625e-06,
+      "loss": 1.438,
+      "step": 3785
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.302610600819024e-06,
+      "loss": 1.4161,
+      "step": 3786
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.300246067968078e-06,
+      "loss": 1.3035,
+      "step": 3787
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.297881694726121e-06,
+      "loss": 1.7474,
+      "step": 3788
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.29551748163245e-06,
+      "loss": 1.3347,
+      "step": 3789
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2931534292263265e-06,
+      "loss": 1.3871,
+      "step": 3790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.290789538046971e-06,
+      "loss": 1.2451,
+      "step": 3791
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2884258086335755e-06,
+      "loss": 1.1034,
+      "step": 3792
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.286062241525285e-06,
+      "loss": 1.3176,
+      "step": 3793
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.283698837261217e-06,
+      "loss": 1.0726,
+      "step": 3794
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.281335596380445e-06,
+      "loss": 1.0737,
+      "step": 3795
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.278972519422009e-06,
+      "loss": 1.5271,
+      "step": 3796
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.276609606924909e-06,
+      "loss": 1.4769,
+      "step": 3797
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.27424685942811e-06,
+      "loss": 1.3828,
+      "step": 3798
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.271884277470538e-06,
+      "loss": 1.3561,
+      "step": 3799
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.269521861591082e-06,
+      "loss": 1.4134,
+      "step": 3800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.267159612328591e-06,
+      "loss": 1.5042,
+      "step": 3801
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.26479753022188e-06,
+      "loss": 1.7763,
+      "step": 3802
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2624356158097205e-06,
+      "loss": 1.5477,
+      "step": 3803
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.26007386963085e-06,
+      "loss": 1.454,
+      "step": 3804
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.257712292223967e-06,
+      "loss": 1.4573,
+      "step": 3805
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.25535088412773e-06,
+      "loss": 1.1821,
+      "step": 3806
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.25298964588076e-06,
+      "loss": 1.2885,
+      "step": 3807
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2506285780216405e-06,
+      "loss": 1.6287,
+      "step": 3808
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.248267681088913e-06,
+      "loss": 1.3719,
+      "step": 3809
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2459069556210795e-06,
+      "loss": 1.3036,
+      "step": 3810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2435464021566084e-06,
+      "loss": 1.6362,
+      "step": 3811
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.241186021233925e-06,
+      "loss": 1.3128,
+      "step": 3812
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.238825813391414e-06,
+      "loss": 1.3191,
+      "step": 3813
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.236465779167424e-06,
+      "loss": 1.4155,
+      "step": 3814
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.234105919100261e-06,
+      "loss": 1.4949,
+      "step": 3815
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.231746233728194e-06,
+      "loss": 1.5426,
+      "step": 3816
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.22938672358945e-06,
+      "loss": 1.4996,
+      "step": 3817
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.227027389222215e-06,
+      "loss": 1.3832,
+      "step": 3818
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.224668231164642e-06,
+      "loss": 1.3997,
+      "step": 3819
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2223092499548325e-06,
+      "loss": 1.4965,
+      "step": 3820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.219950446130858e-06,
+      "loss": 1.2299,
+      "step": 3821
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2175918202307416e-06,
+      "loss": 1.7811,
+      "step": 3822
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2152333727924716e-06,
+      "loss": 1.5094,
+      "step": 3823
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.212875104353993e-06,
+      "loss": 1.1089,
+      "step": 3824
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.210517015453212e-06,
+      "loss": 1.3633,
+      "step": 3825
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.208159106627989e-06,
+      "loss": 1.1633,
+      "step": 3826
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.205801378416148e-06,
+      "loss": 1.1605,
+      "step": 3827
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2034438313554715e-06,
+      "loss": 1.1502,
+      "step": 3828
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.201086465983697e-06,
+      "loss": 1.2891,
+      "step": 3829
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.198729282838527e-06,
+      "loss": 1.3771,
+      "step": 3830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.196372282457614e-06,
+      "loss": 1.0815,
+      "step": 3831
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.194015465378578e-06,
+      "loss": 1.4779,
+      "step": 3832
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.191658832138987e-06,
+      "loss": 1.2674,
+      "step": 3833
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.189302383276378e-06,
+      "loss": 1.3466,
+      "step": 3834
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.186946119328237e-06,
+      "loss": 1.3104,
+      "step": 3835
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.184590040832013e-06,
+      "loss": 1.0535,
+      "step": 3836
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.18223414832511e-06,
+      "loss": 1.3221,
+      "step": 3837
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.179878442344892e-06,
+      "loss": 1.0748,
+      "step": 3838
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.177522923428677e-06,
+      "loss": 1.3616,
+      "step": 3839
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.175167592113744e-06,
+      "loss": 1.3072,
+      "step": 3840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1728124489373296e-06,
+      "loss": 1.0491,
+      "step": 3841
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.17045749443662e-06,
+      "loss": 1.737,
+      "step": 3842
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.168102729148771e-06,
+      "loss": 1.4568,
+      "step": 3843
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.165748153610881e-06,
+      "loss": 1.4485,
+      "step": 3844
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.163393768360019e-06,
+      "loss": 1.7602,
+      "step": 3845
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.161039573933197e-06,
+      "loss": 1.4345,
+      "step": 3846
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.158685570867396e-06,
+      "loss": 1.5724,
+      "step": 3847
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.156331759699544e-06,
+      "loss": 1.1862,
+      "step": 3848
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.15397814096653e-06,
+      "loss": 1.5784,
+      "step": 3849
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.151624715205201e-06,
+      "loss": 2.1576,
+      "step": 3850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.149271482952351e-06,
+      "loss": 1.7763,
+      "step": 3851
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.146918444744742e-06,
+      "loss": 1.3461,
+      "step": 3852
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.14456560111908e-06,
+      "loss": 1.2965,
+      "step": 3853
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.142212952612038e-06,
+      "loss": 1.4826,
+      "step": 3854
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.139860499760234e-06,
+      "loss": 1.2946,
+      "step": 3855
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.137508243100249e-06,
+      "loss": 1.1388,
+      "step": 3856
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1351561831686136e-06,
+      "loss": 1.4894,
+      "step": 3857
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.13280432050182e-06,
+      "loss": 1.5261,
+      "step": 3858
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.130452655636306e-06,
+      "loss": 1.695,
+      "step": 3859
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.128101189108475e-06,
+      "loss": 1.3385,
+      "step": 3860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.125749921454679e-06,
+      "loss": 1.3333,
+      "step": 3861
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.123398853211224e-06,
+      "loss": 1.62,
+      "step": 3862
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.121047984914376e-06,
+      "loss": 1.3112,
+      "step": 3863
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.118697317100347e-06,
+      "loss": 1.3282,
+      "step": 3864
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1163468503053115e-06,
+      "loss": 1.4009,
+      "step": 3865
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.113996585065391e-06,
+      "loss": 1.3494,
+      "step": 3866
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.111646521916669e-06,
+      "loss": 1.624,
+      "step": 3867
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.109296661395174e-06,
+      "loss": 1.5226,
+      "step": 3868
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.106947004036898e-06,
+      "loss": 1.5042,
+      "step": 3869
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.104597550377776e-06,
+      "loss": 1.1202,
+      "step": 3870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.102248300953704e-06,
+      "loss": 1.5011,
+      "step": 3871
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0998992563005326e-06,
+      "loss": 1.298,
+      "step": 3872
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.097550416954057e-06,
+      "loss": 1.5143,
+      "step": 3873
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.095201783450037e-06,
+      "loss": 1.2428,
+      "step": 3874
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.092853356324173e-06,
+      "loss": 1.7502,
+      "step": 3875
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.09050513611213e-06,
+      "loss": 1.2401,
+      "step": 3876
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0881571233495185e-06,
+      "loss": 1.2361,
+      "step": 3877
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.085809318571905e-06,
+      "loss": 1.4432,
+      "step": 3878
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.083461722314805e-06,
+      "loss": 1.3246,
+      "step": 3879
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.081114335113692e-06,
+      "loss": 1.5477,
+      "step": 3880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.078767157503985e-06,
+      "loss": 1.398,
+      "step": 3881
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.076420190021061e-06,
+      "loss": 1.1113,
+      "step": 3882
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.074073433200249e-06,
+      "loss": 1.4654,
+      "step": 3883
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.071726887576823e-06,
+      "loss": 1.2856,
+      "step": 3884
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.069380553686018e-06,
+      "loss": 1.7213,
+      "step": 3885
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.067034432063013e-06,
+      "loss": 1.2944,
+      "step": 3886
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.064688523242947e-06,
+      "loss": 1.568,
+      "step": 3887
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.062342827760899e-06,
+      "loss": 0.9504,
+      "step": 3888
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.059997346151911e-06,
+      "loss": 1.6049,
+      "step": 3889
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.057652078950966e-06,
+      "loss": 1.5632,
+      "step": 3890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.055307026693008e-06,
+      "loss": 1.7914,
+      "step": 3891
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.052962189912924e-06,
+      "loss": 1.9414,
+      "step": 3892
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.050617569145555e-06,
+      "loss": 1.4578,
+      "step": 3893
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0482731649256955e-06,
+      "loss": 1.5895,
+      "step": 3894
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.045928977788085e-06,
+      "loss": 1.3622,
+      "step": 3895
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.043585008267418e-06,
+      "loss": 1.5674,
+      "step": 3896
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.041241256898334e-06,
+      "loss": 1.4532,
+      "step": 3897
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.038897724215431e-06,
+      "loss": 1.443,
+      "step": 3898
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0365544107532495e-06,
+      "loss": 1.435,
+      "step": 3899
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.034211317046285e-06,
+      "loss": 1.4366,
+      "step": 3900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.031868443628978e-06,
+      "loss": 1.7893,
+      "step": 3901
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.029525791035725e-06,
+      "loss": 1.2575,
+      "step": 3902
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.027183359800865e-06,
+      "loss": 1.5379,
+      "step": 3903
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.024841150458692e-06,
+      "loss": 1.5084,
+      "step": 3904
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.022499163543449e-06,
+      "loss": 1.5171,
+      "step": 3905
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.020157399589325e-06,
+      "loss": 1.1333,
+      "step": 3906
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.017815859130462e-06,
+      "loss": 1.4249,
+      "step": 3907
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.015474542700945e-06,
+      "loss": 1.6118,
+      "step": 3908
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.013133450834818e-06,
+      "loss": 1.5489,
+      "step": 3909
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.010792584066062e-06,
+      "loss": 1.6065,
+      "step": 3910
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.008451942928616e-06,
+      "loss": 1.0199,
+      "step": 3911
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.006111527956361e-06,
+      "loss": 1.5244,
+      "step": 3912
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.003771339683132e-06,
+      "loss": 1.3508,
+      "step": 3913
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.001431378642708e-06,
+      "loss": 1.4626,
+      "step": 3914
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.999091645368819e-06,
+      "loss": 1.6449,
+      "step": 3915
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.99675214039514e-06,
+      "loss": 1.2459,
+      "step": 3916
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.994412864255297e-06,
+      "loss": 1.1703,
+      "step": 3917
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.992073817482864e-06,
+      "loss": 1.1928,
+      "step": 3918
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.989735000611358e-06,
+      "loss": 1.158,
+      "step": 3919
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.987396414174251e-06,
+      "loss": 1.6715,
+      "step": 3920
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.985058058704954e-06,
+      "loss": 1.4654,
+      "step": 3921
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.982719934736832e-06,
+      "loss": 1.3773,
+      "step": 3922
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.980382042803192e-06,
+      "loss": 1.35,
+      "step": 3923
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9780443834372965e-06,
+      "loss": 1.4234,
+      "step": 3924
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.975706957172343e-06,
+      "loss": 1.6481,
+      "step": 3925
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.973369764541486e-06,
+      "loss": 1.4862,
+      "step": 3926
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9710328060778215e-06,
+      "loss": 1.5442,
+      "step": 3927
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.968696082314394e-06,
+      "loss": 1.4547,
+      "step": 3928
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9663595937841925e-06,
+      "loss": 1.2463,
+      "step": 3929
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9640233410201555e-06,
+      "loss": 1.4007,
+      "step": 3930
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.961687324555165e-06,
+      "loss": 1.4372,
+      "step": 3931
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.95935154492205e-06,
+      "loss": 1.687,
+      "step": 3932
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9570160026535865e-06,
+      "loss": 1.2194,
+      "step": 3933
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.954680698282494e-06,
+      "loss": 1.3714,
+      "step": 3934
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.95234563234144e-06,
+      "loss": 1.3818,
+      "step": 3935
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9500108053630345e-06,
+      "loss": 1.2667,
+      "step": 3936
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.94767621787984e-06,
+      "loss": 1.4616,
+      "step": 3937
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9453418704243565e-06,
+      "loss": 1.0911,
+      "step": 3938
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.943007763529032e-06,
+      "loss": 1.0235,
+      "step": 3939
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.940673897726261e-06,
+      "loss": 1.3757,
+      "step": 3940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.938340273548382e-06,
+      "loss": 1.4984,
+      "step": 3941
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.936006891527678e-06,
+      "loss": 1.04,
+      "step": 3942
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.933673752196377e-06,
+      "loss": 1.3465,
+      "step": 3943
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.931340856086652e-06,
+      "loss": 1.616,
+      "step": 3944
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.929008203730621e-06,
+      "loss": 1.4274,
+      "step": 3945
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.926675795660344e-06,
+      "loss": 1.3793,
+      "step": 3946
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.924343632407831e-06,
+      "loss": 1.8807,
+      "step": 3947
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9220117145050254e-06,
+      "loss": 1.4548,
+      "step": 3948
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.919680042483829e-06,
+      "loss": 1.5131,
+      "step": 3949
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.917348616876073e-06,
+      "loss": 1.5402,
+      "step": 3950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.915017438213545e-06,
+      "loss": 1.0379,
+      "step": 3951
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.912686507027967e-06,
+      "loss": 1.7114,
+      "step": 3952
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.910355823851008e-06,
+      "loss": 1.4134,
+      "step": 3953
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.908025389214284e-06,
+      "loss": 1.3449,
+      "step": 3954
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.905695203649348e-06,
+      "loss": 1.232,
+      "step": 3955
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9033652676877e-06,
+      "loss": 1.1359,
+      "step": 3956
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.901035581860781e-06,
+      "loss": 1.4638,
+      "step": 3957
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.898706146699978e-06,
+      "loss": 1.4026,
+      "step": 3958
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.896376962736617e-06,
+      "loss": 1.6474,
+      "step": 3959
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8940480305019715e-06,
+      "loss": 1.4096,
+      "step": 3960
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.89171935052725e-06,
+      "loss": 1.5934,
+      "step": 3961
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.889390923343613e-06,
+      "loss": 1.4451,
+      "step": 3962
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.887062749482155e-06,
+      "loss": 1.4978,
+      "step": 3963
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.884734829473919e-06,
+      "loss": 1.4996,
+      "step": 3964
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.882407163849884e-06,
+      "loss": 1.4404,
+      "step": 3965
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.880079753140978e-06,
+      "loss": 1.4439,
+      "step": 3966
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.877752597878063e-06,
+      "loss": 1.6239,
+      "step": 3967
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.87542569859195e-06,
+      "loss": 1.5927,
+      "step": 3968
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.873099055813389e-06,
+      "loss": 1.3344,
+      "step": 3969
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.870772670073067e-06,
+      "loss": 1.538,
+      "step": 3970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.868446541901623e-06,
+      "loss": 1.3702,
+      "step": 3971
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.866120671829623e-06,
+      "loss": 1.4034,
+      "step": 3972
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8637950603875875e-06,
+      "loss": 1.1236,
+      "step": 3973
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.861469708105969e-06,
+      "loss": 1.735,
+      "step": 3974
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.859144615515166e-06,
+      "loss": 1.3338,
+      "step": 3975
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.856819783145514e-06,
+      "loss": 1.2739,
+      "step": 3976
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.854495211527294e-06,
+      "loss": 1.3929,
+      "step": 3977
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.85217090119072e-06,
+      "loss": 1.2904,
+      "step": 3978
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8498468526659535e-06,
+      "loss": 1.3787,
+      "step": 3979
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.847523066483097e-06,
+      "loss": 1.4119,
+      "step": 3980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.845199543172185e-06,
+      "loss": 1.4369,
+      "step": 3981
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.842876283263199e-06,
+      "loss": 1.535,
+      "step": 3982
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.840553287286058e-06,
+      "loss": 1.4336,
+      "step": 3983
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.838230555770623e-06,
+      "loss": 1.105,
+      "step": 3984
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.835908089246688e-06,
+      "loss": 1.4088,
+      "step": 3985
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.833585888243998e-06,
+      "loss": 1.3943,
+      "step": 3986
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.831263953292225e-06,
+      "loss": 1.4693,
+      "step": 3987
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.82894228492099e-06,
+      "loss": 1.3779,
+      "step": 3988
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.826620883659846e-06,
+      "loss": 1.4532,
+      "step": 3989
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.824299750038291e-06,
+      "loss": 1.5941,
+      "step": 3990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.821978884585761e-06,
+      "loss": 1.5705,
+      "step": 3991
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.819658287831624e-06,
+      "loss": 1.3789,
+      "step": 3992
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.817337960305197e-06,
+      "loss": 1.2408,
+      "step": 3993
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.815017902535727e-06,
+      "loss": 1.1121,
+      "step": 3994
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8126981150524068e-06,
+      "loss": 1.1901,
+      "step": 3995
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.810378598384358e-06,
+      "loss": 1.3604,
+      "step": 3996
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8080593530606525e-06,
+      "loss": 1.2975,
+      "step": 3997
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.805740379610289e-06,
+      "loss": 1.4444,
+      "step": 3998
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.803421678562213e-06,
+      "loss": 1.1329,
+      "step": 3999
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8011032504453e-06,
+      "loss": 1.3493,
+      "step": 4000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7987850957883693e-06,
+      "loss": 1.4337,
+      "step": 4001
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.796467215120178e-06,
+      "loss": 1.1506,
+      "step": 4002
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7941496089694146e-06,
+      "loss": 1.5528,
+      "step": 4003
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7918322778647126e-06,
+      "loss": 1.1484,
+      "step": 4004
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7895152223346354e-06,
+      "loss": 1.7323,
+      "step": 4005
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7871984429076907e-06,
+      "loss": 1.4054,
+      "step": 4006
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.784881940112316e-06,
+      "loss": 1.6563,
+      "step": 4007
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.782565714476894e-06,
+      "loss": 1.2942,
+      "step": 4008
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7802497665297345e-06,
+      "loss": 1.345,
+      "step": 4009
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.777934096799094e-06,
+      "loss": 1.5719,
+      "step": 4010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7756187058131553e-06,
+      "loss": 1.3964,
+      "step": 4011
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.773303594100047e-06,
+      "loss": 1.5436,
+      "step": 4012
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7709887621878305e-06,
+      "loss": 1.664,
+      "step": 4013
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.768674210604498e-06,
+      "loss": 1.6736,
+      "step": 4014
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.766359939877989e-06,
+      "loss": 1.0938,
+      "step": 4015
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7640459505361667e-06,
+      "loss": 1.1003,
+      "step": 4016
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.76173224310684e-06,
+      "loss": 1.652,
+      "step": 4017
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.759418818117746e-06,
+      "loss": 1.5022,
+      "step": 4018
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.757105676096565e-06,
+      "loss": 1.2954,
+      "step": 4019
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7547928175709037e-06,
+      "loss": 1.0705,
+      "step": 4020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7524802430683137e-06,
+      "loss": 1.4855,
+      "step": 4021
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.750167953116272e-06,
+      "loss": 1.5098,
+      "step": 4022
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7478559482421993e-06,
+      "loss": 1.2976,
+      "step": 4023
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.745544228973449e-06,
+      "loss": 1.3572,
+      "step": 4024
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7432327958373044e-06,
+      "loss": 1.4906,
+      "step": 4025
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.740921649360991e-06,
+      "loss": 1.6097,
+      "step": 4026
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7386107900716616e-06,
+      "loss": 1.083,
+      "step": 4027
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7363002184964103e-06,
+      "loss": 1.3716,
+      "step": 4028
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7339899351622587e-06,
+      "loss": 1.1524,
+      "step": 4029
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7316799405961713e-06,
+      "loss": 1.5872,
+      "step": 4030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.729370235325036e-06,
+      "loss": 1.6619,
+      "step": 4031
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7270608198756852e-06,
+      "loss": 1.3885,
+      "step": 4032
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7247516947748774e-06,
+      "loss": 1.835,
+      "step": 4033
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7224428605493087e-06,
+      "loss": 1.8699,
+      "step": 4034
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.72013431772561e-06,
+      "loss": 1.52,
+      "step": 4035
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.717826066830341e-06,
+      "loss": 1.5914,
+      "step": 4036
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7155181083900005e-06,
+      "loss": 1.1823,
+      "step": 4037
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7132104429310144e-06,
+      "loss": 1.4597,
+      "step": 4038
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.710903070979749e-06,
+      "loss": 1.1808,
+      "step": 4039
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7085959930624946e-06,
+      "loss": 1.4942,
+      "step": 4040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.706289209705485e-06,
+      "loss": 1.6274,
+      "step": 4041
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.703982721434877e-06,
+      "loss": 1.6586,
+      "step": 4042
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7016765287767674e-06,
+      "loss": 1.2852,
+      "step": 4043
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.699370632257182e-06,
+      "loss": 1.5054,
+      "step": 4044
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6970650324020784e-06,
+      "loss": 1.644,
+      "step": 4045
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6947597297373495e-06,
+      "loss": 1.1516,
+      "step": 4046
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6924547247888175e-06,
+      "loss": 1.7017,
+      "step": 4047
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6901500180822393e-06,
+      "loss": 1.3183,
+      "step": 4048
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6878456101433003e-06,
+      "loss": 1.2415,
+      "step": 4049
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6855415014976227e-06,
+      "loss": 1.4695,
+      "step": 4050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.683237692670755e-06,
+      "loss": 1.554,
+      "step": 4051
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.680934184188182e-06,
+      "loss": 1.3831,
+      "step": 4052
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.678630976575315e-06,
+      "loss": 1.7146,
+      "step": 4053
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.676328070357503e-06,
+      "loss": 1.5638,
+      "step": 4054
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6740254660600205e-06,
+      "loss": 1.7452,
+      "step": 4055
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6717231642080776e-06,
+      "loss": 1.5114,
+      "step": 4056
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6694211653268107e-06,
+      "loss": 1.7578,
+      "step": 4057
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6671194699412914e-06,
+      "loss": 1.1,
+      "step": 4058
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6648180785765205e-06,
+      "loss": 1.2451,
+      "step": 4059
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.662516991757428e-06,
+      "loss": 1.2607,
+      "step": 4060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6602162100088768e-06,
+      "loss": 1.6235,
+      "step": 4061
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6579157338556582e-06,
+      "loss": 1.2468,
+      "step": 4062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.655615563822496e-06,
+      "loss": 1.4836,
+      "step": 4063
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6533157004340426e-06,
+      "loss": 1.4257,
+      "step": 4064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6510161442148783e-06,
+      "loss": 1.456,
+      "step": 4065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6487168956895215e-06,
+      "loss": 1.2941,
+      "step": 4066
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6464179553824086e-06,
+      "loss": 1.5545,
+      "step": 4067
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6441193238179152e-06,
+      "loss": 1.4649,
+      "step": 4068
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.641821001520343e-06,
+      "loss": 1.3134,
+      "step": 4069
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6395229890139226e-06,
+      "loss": 1.6714,
+      "step": 4070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6372252868228147e-06,
+      "loss": 1.5478,
+      "step": 4071
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6349278954711085e-06,
+      "loss": 1.7943,
+      "step": 4072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.632630815482825e-06,
+      "loss": 1.787,
+      "step": 4073
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.63033404738191e-06,
+      "loss": 1.2731,
+      "step": 4074
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6280375916922406e-06,
+      "loss": 1.2995,
+      "step": 4075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6257414489376217e-06,
+      "loss": 1.4538,
+      "step": 4076
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6234456196417913e-06,
+      "loss": 1.4548,
+      "step": 4077
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.621150104328407e-06,
+      "loss": 1.3037,
+      "step": 4078
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6188549035210635e-06,
+      "loss": 1.1926,
+      "step": 4079
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.616560017743277e-06,
+      "loss": 1.3685,
+      "step": 4080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.614265447518498e-06,
+      "loss": 1.622,
+      "step": 4081
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6119711933700995e-06,
+      "loss": 1.7472,
+      "step": 4082
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.609677255821387e-06,
+      "loss": 1.4478,
+      "step": 4083
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.607383635395589e-06,
+      "loss": 1.5818,
+      "step": 4084
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6050903326158663e-06,
+      "loss": 1.3939,
+      "step": 4085
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.602797348005305e-06,
+      "loss": 1.5091,
+      "step": 4086
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.600504682086918e-06,
+      "loss": 1.4065,
+      "step": 4087
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5982123353836484e-06,
+      "loss": 1.3197,
+      "step": 4088
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.595920308418362e-06,
+      "loss": 1.6673,
+      "step": 4089
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5936286017138566e-06,
+      "loss": 1.2938,
+      "step": 4090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5913372157928515e-06,
+      "loss": 1.2389,
+      "step": 4091
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.589046151178e-06,
+      "loss": 1.2592,
+      "step": 4092
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.586755408391873e-06,
+      "loss": 1.6743,
+      "step": 4093
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5844649879569777e-06,
+      "loss": 1.5382,
+      "step": 4094
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.582174890395739e-06,
+      "loss": 1.4673,
+      "step": 4095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.579885116230515e-06,
+      "loss": 1.8996,
+      "step": 4096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5775956659835854e-06,
+      "loss": 1.5094,
+      "step": 4097
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5753065401771577e-06,
+      "loss": 1.2477,
+      "step": 4098
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.573017739333369e-06,
+      "loss": 1.5488,
+      "step": 4099
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.570729263974273e-06,
+      "loss": 1.0139,
+      "step": 4100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.568441114621861e-06,
+      "loss": 1.3917,
+      "step": 4101
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.566153291798039e-06,
+      "loss": 1.3339,
+      "step": 4102
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.563865796024648e-06,
+      "loss": 1.9091,
+      "step": 4103
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5615786278234443e-06,
+      "loss": 1.5077,
+      "step": 4104
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.559291787716121e-06,
+      "loss": 1.3885,
+      "step": 4105
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5570052762242847e-06,
+      "loss": 1.32,
+      "step": 4106
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5547190938694774e-06,
+      "loss": 1.3149,
+      "step": 4107
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5524332411731575e-06,
+      "loss": 1.2728,
+      "step": 4108
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5501477186567146e-06,
+      "loss": 1.3715,
+      "step": 4109
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5478625268414613e-06,
+      "loss": 1.6159,
+      "step": 4110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5455776662486297e-06,
+      "loss": 1.6856,
+      "step": 4111
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5432931373993858e-06,
+      "loss": 1.0747,
+      "step": 4112
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5410089408148096e-06,
+      "loss": 1.3623,
+      "step": 4113
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5387250770159152e-06,
+      "loss": 1.1495,
+      "step": 4114
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.536441546523632e-06,
+      "loss": 1.4401,
+      "step": 4115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5341583498588207e-06,
+      "loss": 1.4708,
+      "step": 4116
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5318754875422588e-06,
+      "loss": 1.2429,
+      "step": 4117
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5295929600946554e-06,
+      "loss": 1.5204,
+      "step": 4118
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.527310768036635e-06,
+      "loss": 1.2141,
+      "step": 4119
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5250289118887515e-06,
+      "loss": 1.5256,
+      "step": 4120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.522747392171483e-06,
+      "loss": 1.6175,
+      "step": 4121
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5204662094052232e-06,
+      "loss": 1.4016,
+      "step": 4122
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5181853641102993e-06,
+      "loss": 1.4153,
+      "step": 4123
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.515904856806951e-06,
+      "loss": 1.6378,
+      "step": 4124
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.513624688015351e-06,
+      "loss": 1.4849,
+      "step": 4125
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5113448582555855e-06,
+      "loss": 1.3559,
+      "step": 4126
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.509065368047671e-06,
+      "loss": 1.7007,
+      "step": 4127
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5067862179115408e-06,
+      "loss": 1.5294,
+      "step": 4128
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.504507408367056e-06,
+      "loss": 1.2052,
+      "step": 4129
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5022289399339933e-06,
+      "loss": 1.1607,
+      "step": 4130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4999508131320577e-06,
+      "loss": 1.7434,
+      "step": 4131
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.497673028480876e-06,
+      "loss": 1.5504,
+      "step": 4132
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.495395586499991e-06,
+      "loss": 1.3415,
+      "step": 4133
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4931184877088754e-06,
+      "loss": 1.6039,
+      "step": 4134
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4908417326269156e-06,
+      "loss": 1.1938,
+      "step": 4135
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4885653217734273e-06,
+      "loss": 1.3452,
+      "step": 4136
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.48628925566764e-06,
+      "loss": 1.4371,
+      "step": 4137
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.484013534828713e-06,
+      "loss": 1.0786,
+      "step": 4138
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.481738159775717e-06,
+      "loss": 1.7181,
+      "step": 4139
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4794631310276544e-06,
+      "loss": 1.4378,
+      "step": 4140
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.477188449103439e-06,
+      "loss": 1.1666,
+      "step": 4141
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4749141145219118e-06,
+      "loss": 1.3032,
+      "step": 4142
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4726401278018353e-06,
+      "loss": 1.0271,
+      "step": 4143
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4703664894618842e-06,
+      "loss": 1.3948,
+      "step": 4144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4680932000206657e-06,
+      "loss": 0.845,
+      "step": 4145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4658202599966963e-06,
+      "loss": 1.2335,
+      "step": 4146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.463547669908422e-06,
+      "loss": 1.4126,
+      "step": 4147
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4612754302742e-06,
+      "loss": 1.597,
+      "step": 4148
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.459003541612317e-06,
+      "loss": 1.4498,
+      "step": 4149
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.456732004440971e-06,
+      "loss": 1.5062,
+      "step": 4150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4544608192782855e-06,
+      "loss": 1.7598,
+      "step": 4151
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4521899866423034e-06,
+      "loss": 1.1599,
+      "step": 4152
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4499195070509833e-06,
+      "loss": 1.3047,
+      "step": 4153
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.447649381022209e-06,
+      "loss": 1.6847,
+      "step": 4154
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.445379609073776e-06,
+      "loss": 1.5972,
+      "step": 4155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.443110191723407e-06,
+      "loss": 1.5093,
+      "step": 4156
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4408411294887378e-06,
+      "loss": 1.415,
+      "step": 4157
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4385724228873285e-06,
+      "loss": 1.6128,
+      "step": 4158
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.436304072436652e-06,
+      "loss": 1.6401,
+      "step": 4159
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4340360786541067e-06,
+      "loss": 1.7321,
+      "step": 4160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.431768442057002e-06,
+      "loss": 1.2837,
+      "step": 4161
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4295011631625723e-06,
+      "loss": 1.4132,
+      "step": 4162
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.42723424248797e-06,
+      "loss": 1.6278,
+      "step": 4163
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.424967680550261e-06,
+      "loss": 1.241,
+      "step": 4164
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4227014778664334e-06,
+      "loss": 1.2723,
+      "step": 4165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.420435634953391e-06,
+      "loss": 1.2704,
+      "step": 4166
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4181701523279596e-06,
+      "loss": 1.2013,
+      "step": 4167
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4159050305068764e-06,
+      "loss": 1.4548,
+      "step": 4168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4136402700068034e-06,
+      "loss": 1.6538,
+      "step": 4169
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4113758713443124e-06,
+      "loss": 1.3688,
+      "step": 4170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.409111835035902e-06,
+      "loss": 1.419,
+      "step": 4171
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.406848161597979e-06,
+      "loss": 1.4904,
+      "step": 4172
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.404584851546873e-06,
+      "loss": 1.6916,
+      "step": 4173
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.40232190539883e-06,
+      "loss": 1.4928,
+      "step": 4174
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.400059323670011e-06,
+      "loss": 1.2566,
+      "step": 4175
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.397797106876496e-06,
+      "loss": 1.6556,
+      "step": 4176
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.39553525553428e-06,
+      "loss": 1.494,
+      "step": 4177
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.393273770159276e-06,
+      "loss": 1.3491,
+      "step": 4178
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.391012651267313e-06,
+      "loss": 1.5389,
+      "step": 4179
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3887518993741363e-06,
+      "loss": 1.5836,
+      "step": 4180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3864915149954073e-06,
+      "loss": 1.5291,
+      "step": 4181
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.384231498646706e-06,
+      "loss": 1.3427,
+      "step": 4182
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.381971850843523e-06,
+      "loss": 1.6421,
+      "step": 4183
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3797125721012703e-06,
+      "loss": 1.3451,
+      "step": 4184
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.377453662935274e-06,
+      "loss": 1.6933,
+      "step": 4185
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3751951238607745e-06,
+      "loss": 1.4201,
+      "step": 4186
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3729369553929287e-06,
+      "loss": 1.7232,
+      "step": 4187
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3706791580468098e-06,
+      "loss": 1.3526,
+      "step": 4188
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.368421732337405e-06,
+      "loss": 1.4192,
+      "step": 4189
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3661646787796176e-06,
+      "loss": 1.2056,
+      "step": 4190
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.363907997888266e-06,
+      "loss": 1.2746,
+      "step": 4191
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3616516901780826e-06,
+      "loss": 1.3892,
+      "step": 4192
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3593957561637165e-06,
+      "loss": 1.1834,
+      "step": 4193
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3571401963597295e-06,
+      "loss": 1.2882,
+      "step": 4194
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3548850112805985e-06,
+      "loss": 1.509,
+      "step": 4195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3526302014407198e-06,
+      "loss": 1.343,
+      "step": 4196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.350375767354393e-06,
+      "loss": 1.6513,
+      "step": 4197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3481217095358437e-06,
+      "loss": 1.5603,
+      "step": 4198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3458680284992053e-06,
+      "loss": 1.2928,
+      "step": 4199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.343614724758527e-06,
+      "loss": 1.386,
+      "step": 4200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3413617988277714e-06,
+      "loss": 1.5561,
+      "step": 4201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3391092512208144e-06,
+      "loss": 1.4289,
+      "step": 4202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.336857082451448e-06,
+      "loss": 1.5027,
+      "step": 4203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3346052930333755e-06,
+      "loss": 1.5972,
+      "step": 4204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3323538834802136e-06,
+      "loss": 1.3805,
+      "step": 4205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3301028543054935e-06,
+      "loss": 1.3094,
+      "step": 4206
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.327852206022661e-06,
+      "loss": 1.4106,
+      "step": 4207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3256019391450696e-06,
+      "loss": 1.5332,
+      "step": 4208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3233520541859938e-06,
+      "loss": 1.745,
+      "step": 4209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3211025516586125e-06,
+      "loss": 1.6298,
+      "step": 4210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.318853432076025e-06,
+      "loss": 1.4441,
+      "step": 4211
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3166046959512366e-06,
+      "loss": 1.3067,
+      "step": 4212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3143563437971725e-06,
+      "loss": 1.1835,
+      "step": 4213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3121083761266614e-06,
+      "loss": 1.4003,
+      "step": 4214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.309860793452453e-06,
+      "loss": 1.4149,
+      "step": 4215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3076135962872014e-06,
+      "loss": 1.2563,
+      "step": 4216
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.305366785143479e-06,
+      "loss": 1.5763,
+      "step": 4217
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.303120360533769e-06,
+      "loss": 1.3848,
+      "step": 4218
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.300874322970462e-06,
+      "loss": 1.2164,
+      "step": 4219
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2986286729658673e-06,
+      "loss": 1.4833,
+      "step": 4220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.296383411032198e-06,
+      "loss": 1.5079,
+      "step": 4221
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.294138537681586e-06,
+      "loss": 1.2234,
+      "step": 4222
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2918940534260674e-06,
+      "loss": 1.6498,
+      "step": 4223
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2896499587775982e-06,
+      "loss": 1.5028,
+      "step": 4224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2874062542480363e-06,
+      "loss": 1.2491,
+      "step": 4225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2851629403491586e-06,
+      "loss": 1.2708,
+      "step": 4226
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2829200175926457e-06,
+      "loss": 1.5404,
+      "step": 4227
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2806774864900954e-06,
+      "loss": 1.2586,
+      "step": 4228
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.278435347553014e-06,
+      "loss": 1.7365,
+      "step": 4229
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2761936012928147e-06,
+      "loss": 1.4254,
+      "step": 4230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.273952248220828e-06,
+      "loss": 1.304,
+      "step": 4231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.271711288848287e-06,
+      "loss": 1.0609,
+      "step": 4232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2694707236863423e-06,
+      "loss": 1.5431,
+      "step": 4233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.267230553246047e-06,
+      "loss": 1.6959,
+      "step": 4234
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.264990778038374e-06,
+      "loss": 1.2945,
+      "step": 4235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2627513985741943e-06,
+      "loss": 1.2463,
+      "step": 4236
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.260512415364301e-06,
+      "loss": 1.0949,
+      "step": 4237
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2582738289193837e-06,
+      "loss": 1.6863,
+      "step": 4238
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2560356397500524e-06,
+      "loss": 1.4756,
+      "step": 4239
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.253797848366825e-06,
+      "loss": 1.3979,
+      "step": 4240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2515604552801207e-06,
+      "loss": 1.2214,
+      "step": 4241
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.249323461000277e-06,
+      "loss": 1.6964,
+      "step": 4242
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.247086866037533e-06,
+      "loss": 1.6178,
+      "step": 4243
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2448506709020457e-06,
+      "loss": 1.3133,
+      "step": 4244
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2426148761038702e-06,
+      "loss": 1.5848,
+      "step": 4245
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.24037948215298e-06,
+      "loss": 1.4424,
+      "step": 4246
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2381444895592483e-06,
+      "loss": 1.3331,
+      "step": 4247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.235909898832465e-06,
+      "loss": 1.0147,
+      "step": 4248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.233675710482326e-06,
+      "loss": 1.5364,
+      "step": 4249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.231441925018428e-06,
+      "loss": 1.0921,
+      "step": 4250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.229208542950289e-06,
+      "loss": 1.0799,
+      "step": 4251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.226975564787322e-06,
+      "loss": 1.3132,
+      "step": 4252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2247429910388585e-06,
+      "loss": 1.2744,
+      "step": 4253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2225108222141275e-06,
+      "loss": 1.0649,
+      "step": 4254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2202790588222775e-06,
+      "loss": 1.1551,
+      "step": 4255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.218047701372352e-06,
+      "loss": 1.366,
+      "step": 4256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.215816750373314e-06,
+      "loss": 1.2342,
+      "step": 4257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.213586206334022e-06,
+      "loss": 1.2266,
+      "step": 4258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2113560697632513e-06,
+      "loss": 1.7879,
+      "step": 4259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.209126341169681e-06,
+      "loss": 1.4075,
+      "step": 4260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2068970210618934e-06,
+      "loss": 1.2909,
+      "step": 4261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2046681099483845e-06,
+      "loss": 1.4605,
+      "step": 4262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2024396083375495e-06,
+      "loss": 1.2912,
+      "step": 4263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2002115167376985e-06,
+      "loss": 1.4904,
+      "step": 4264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1979838356570393e-06,
+      "loss": 1.498,
+      "step": 4265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1957565656036938e-06,
+      "loss": 1.7195,
+      "step": 4266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1935297070856836e-06,
+      "loss": 1.1081,
+      "step": 4267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1913032606109435e-06,
+      "loss": 1.6602,
+      "step": 4268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.189077226687306e-06,
+      "loss": 1.3433,
+      "step": 4269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1868516058225165e-06,
+      "loss": 1.5147,
+      "step": 4270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1846263985242243e-06,
+      "loss": 1.6561,
+      "step": 4271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.182401605299982e-06,
+      "loss": 1.294,
+      "step": 4272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.180177226657251e-06,
+      "loss": 1.2184,
+      "step": 4273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.177953263103394e-06,
+      "loss": 1.103,
+      "step": 4274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1757297151456844e-06,
+      "loss": 1.2401,
+      "step": 4275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1735065832912947e-06,
+      "loss": 1.3186,
+      "step": 4276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.17128386804731e-06,
+      "loss": 1.3518,
+      "step": 4277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1690615699207122e-06,
+      "loss": 1.5994,
+      "step": 4278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1668396894183955e-06,
+      "loss": 1.4097,
+      "step": 4279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1646182270471517e-06,
+      "loss": 1.3432,
+      "step": 4280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1623971833136826e-06,
+      "loss": 1.1818,
+      "step": 4281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.160176558724595e-06,
+      "loss": 1.1849,
+      "step": 4282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1579563537863926e-06,
+      "loss": 1.3898,
+      "step": 4283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.155736569005495e-06,
+      "loss": 1.4952,
+      "step": 4284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1535172048882133e-06,
+      "loss": 1.1222,
+      "step": 4285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.151298261940775e-06,
+      "loss": 1.346,
+      "step": 4286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1490797406693003e-06,
+      "loss": 1.2162,
+      "step": 4287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1468616415798224e-06,
+      "loss": 1.5112,
+      "step": 4288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.144643965178271e-06,
+      "loss": 1.4312,
+      "step": 4289
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1424267119704852e-06,
+      "loss": 1.6796,
+      "step": 4290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.140209882462203e-06,
+      "loss": 1.3618,
+      "step": 4291
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.137993477159069e-06,
+      "loss": 1.3398,
+      "step": 4292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1357774965666297e-06,
+      "loss": 1.642,
+      "step": 4293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1335619411903332e-06,
+      "loss": 1.5487,
+      "step": 4294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1313468115355362e-06,
+      "loss": 1.1924,
+      "step": 4295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1291321081074887e-06,
+      "loss": 1.2101,
+      "step": 4296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1269178314113544e-06,
+      "loss": 1.3339,
+      "step": 4297
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1247039819521907e-06,
+      "loss": 1.2213,
+      "step": 4298
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.122490560234964e-06,
+      "loss": 1.4618,
+      "step": 4299
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1202775667645368e-06,
+      "loss": 1.6035,
+      "step": 4300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.118065002045682e-06,
+      "loss": 1.1778,
+      "step": 4301
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1158528665830657e-06,
+      "loss": 1.3751,
+      "step": 4302
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.113641160881263e-06,
+      "loss": 1.5092,
+      "step": 4303
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.111429885444749e-06,
+      "loss": 1.3333,
+      "step": 4304
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.109219040777899e-06,
+      "loss": 1.4948,
+      "step": 4305
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.107008627384991e-06,
+      "loss": 1.5765,
+      "step": 4306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.104798645770206e-06,
+      "loss": 1.2662,
+      "step": 4307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1025890964376246e-06,
+      "loss": 1.2252,
+      "step": 4308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1003799798912298e-06,
+      "loss": 1.3794,
+      "step": 4309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0981712966349053e-06,
+      "loss": 1.6987,
+      "step": 4310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.095963047172437e-06,
+      "loss": 1.3981,
+      "step": 4311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0937552320075116e-06,
+      "loss": 1.5765,
+      "step": 4312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0915478516437137e-06,
+      "loss": 1.4305,
+      "step": 4313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0893409065845346e-06,
+      "loss": 1.4337,
+      "step": 4314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0871343973333623e-06,
+      "loss": 1.3999,
+      "step": 4315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.084928324393485e-06,
+      "loss": 1.561,
+      "step": 4316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.082722688268094e-06,
+      "loss": 1.6122,
+      "step": 4317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0805174894602775e-06,
+      "loss": 1.3796,
+      "step": 4318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0783127284730275e-06,
+      "loss": 1.1219,
+      "step": 4319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.076108405809235e-06,
+      "loss": 1.1854,
+      "step": 4320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0739045219716888e-06,
+      "loss": 1.7433,
+      "step": 4321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0717010774630807e-06,
+      "loss": 1.3517,
+      "step": 4322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0694980727860015e-06,
+      "loss": 1.5475,
+      "step": 4323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0672955084429403e-06,
+      "loss": 1.5108,
+      "step": 4324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.065093384936285e-06,
+      "loss": 1.4152,
+      "step": 4325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0628917027683293e-06,
+      "loss": 1.6862,
+      "step": 4326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.060690462441256e-06,
+      "loss": 1.4938,
+      "step": 4327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0584896644571583e-06,
+      "loss": 1.4544,
+      "step": 4328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0562893093180168e-06,
+      "loss": 1.569,
+      "step": 4329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0540893975257217e-06,
+      "loss": 1.3045,
+      "step": 4330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0518899295820566e-06,
+      "loss": 1.3801,
+      "step": 4331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0496909059887036e-06,
+      "loss": 1.6019,
+      "step": 4332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.047492327247246e-06,
+      "loss": 1.5457,
+      "step": 4333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0452941938591628e-06,
+      "loss": 1.3756,
+      "step": 4334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0430965063258345e-06,
+      "loss": 1.4547,
+      "step": 4335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.040899265148536e-06,
+      "loss": 1.3101,
+      "step": 4336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0387024708284457e-06,
+      "loss": 1.4987,
+      "step": 4337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0365061238666336e-06,
+      "loss": 1.4738,
+      "step": 4338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0343102247640755e-06,
+      "loss": 1.6397,
+      "step": 4339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.032114774021636e-06,
+      "loss": 1.1091,
+      "step": 4340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0299197721400863e-06,
+      "loss": 1.4269,
+      "step": 4341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.027725219620087e-06,
+      "loss": 1.2502,
+      "step": 4342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0255311169622043e-06,
+      "loss": 1.1146,
+      "step": 4343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0233374646668935e-06,
+      "loss": 1.4713,
+      "step": 4344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0211442632345133e-06,
+      "loss": 1.6629,
+      "step": 4345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.01895151316532e-06,
+      "loss": 1.4545,
+      "step": 4346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0167592149594606e-06,
+      "loss": 1.7011,
+      "step": 4347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0145673691169863e-06,
+      "loss": 1.4203,
+      "step": 4348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.012375976137839e-06,
+      "loss": 1.756,
+      "step": 4349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0101850365218633e-06,
+      "loss": 1.6259,
+      "step": 4350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.007994550768793e-06,
+      "loss": 1.3401,
+      "step": 4351
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0058045193782676e-06,
+      "loss": 1.379,
+      "step": 4352
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.003614942849813e-06,
+      "loss": 1.6508,
+      "step": 4353
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.001425821682861e-06,
+      "loss": 1.2512,
+      "step": 4354
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9992371563767307e-06,
+      "loss": 1.7301,
+      "step": 4355
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9970489474306435e-06,
+      "loss": 1.3178,
+      "step": 4356
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.994861195343717e-06,
+      "loss": 1.0486,
+      "step": 4357
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9926739006149564e-06,
+      "loss": 1.6215,
+      "step": 4358
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9904870637432737e-06,
+      "loss": 1.5715,
+      "step": 4359
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.988300685227468e-06,
+      "loss": 1.1634,
+      "step": 4360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9861147655662387e-06,
+      "loss": 1.4017,
+      "step": 4361
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9839293052581767e-06,
+      "loss": 1.3603,
+      "step": 4362
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9817443048017734e-06,
+      "loss": 1.0242,
+      "step": 4363
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.979559764695409e-06,
+      "loss": 1.3447,
+      "step": 4364
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.977375685437365e-06,
+      "loss": 1.2826,
+      "step": 4365
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9751920675258106e-06,
+      "loss": 1.3459,
+      "step": 4366
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.973008911458816e-06,
+      "loss": 1.8176,
+      "step": 4367
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9708262177343468e-06,
+      "loss": 1.2232,
+      "step": 4368
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.968643986850255e-06,
+      "loss": 1.7143,
+      "step": 4369
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.966462219304296e-06,
+      "loss": 1.5544,
+      "step": 4370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.964280915594113e-06,
+      "loss": 1.4226,
+      "step": 4371
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.962100076217249e-06,
+      "loss": 1.514,
+      "step": 4372
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9599197016711356e-06,
+      "loss": 1.5708,
+      "step": 4373
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9577397924531038e-06,
+      "loss": 1.4109,
+      "step": 4374
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.955560349060371e-06,
+      "loss": 1.4483,
+      "step": 4375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9533813719900594e-06,
+      "loss": 1.6469,
+      "step": 4376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.951202861739173e-06,
+      "loss": 1.6123,
+      "step": 4377
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9490248188046166e-06,
+      "loss": 1.3079,
+      "step": 4378
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9468472436831887e-06,
+      "loss": 1.4487,
+      "step": 4379
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9446701368715758e-06,
+      "loss": 1.5576,
+      "step": 4380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.942493498866364e-06,
+      "loss": 1.5004,
+      "step": 4381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9403173301640254e-06,
+      "loss": 1.662,
+      "step": 4382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.938141631260933e-06,
+      "loss": 1.3566,
+      "step": 4383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9359664026533443e-06,
+      "loss": 1.5911,
+      "step": 4384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9337916448374182e-06,
+      "loss": 1.3902,
+      "step": 4385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9316173583091968e-06,
+      "loss": 1.2365,
+      "step": 4386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9294435435646253e-06,
+      "loss": 1.6624,
+      "step": 4387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9272702010995303e-06,
+      "loss": 1.0236,
+      "step": 4388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9250973314096386e-06,
+      "loss": 1.2744,
+      "step": 4389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9229249349905686e-06,
+      "loss": 1.4593,
+      "step": 4390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9207530123378246e-06,
+      "loss": 1.4348,
+      "step": 4391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9185815639468118e-06,
+      "loss": 1.4285,
+      "step": 4392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9164105903128175e-06,
+      "loss": 1.4332,
+      "step": 4393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.91424009193103e-06,
+      "loss": 1.3845,
+      "step": 4394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9120700692965198e-06,
+      "loss": 1.3534,
+      "step": 4395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.909900522904257e-06,
+      "loss": 1.7641,
+      "step": 4396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.907731453249102e-06,
+      "loss": 1.3054,
+      "step": 4397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9055628608258e-06,
+      "loss": 1.3741,
+      "step": 4398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.903394746128997e-06,
+      "loss": 1.3645,
+      "step": 4399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9012271096532185e-06,
+      "loss": 1.7378,
+      "step": 4400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8990599518928943e-06,
+      "loss": 1.4984,
+      "step": 4401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8968932733423307e-06,
+      "loss": 1.4252,
+      "step": 4402
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8947270744957385e-06,
+      "loss": 1.5454,
+      "step": 4403
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.892561355847207e-06,
+      "loss": 1.1933,
+      "step": 4404
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.890396117890727e-06,
+      "loss": 1.3123,
+      "step": 4405
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8882313611201684e-06,
+      "loss": 1.434,
+      "step": 4406
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8860670860293027e-06,
+      "loss": 1.3709,
+      "step": 4407
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8839032931117804e-06,
+      "loss": 1.878,
+      "step": 4408
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8817399828611537e-06,
+      "loss": 1.7786,
+      "step": 4409
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.879577155770852e-06,
+      "loss": 1.4628,
+      "step": 4410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.877414812334205e-06,
+      "loss": 1.5451,
+      "step": 4411
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8752529530444296e-06,
+      "loss": 1.4492,
+      "step": 4412
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.873091578394626e-06,
+      "loss": 1.8287,
+      "step": 4413
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8709306888777945e-06,
+      "loss": 1.4162,
+      "step": 4414
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8687702849868126e-06,
+      "loss": 1.4082,
+      "step": 4415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8666103672144597e-06,
+      "loss": 1.7706,
+      "step": 4416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8644509360533925e-06,
+      "loss": 1.4148,
+      "step": 4417
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8622919919961658e-06,
+      "loss": 1.5745,
+      "step": 4418
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8601335355352165e-06,
+      "loss": 1.1615,
+      "step": 4419
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8579755671628785e-06,
+      "loss": 1.2812,
+      "step": 4420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8558180873713637e-06,
+      "loss": 1.2769,
+      "step": 4421
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.85366109665278e-06,
+      "loss": 1.5147,
+      "step": 4422
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.851504595499125e-06,
+      "loss": 1.6237,
+      "step": 4423
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.849348584402277e-06,
+      "loss": 1.4814,
+      "step": 4424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.847193063854011e-06,
+      "loss": 1.3647,
+      "step": 4425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8450380343459837e-06,
+      "loss": 1.3761,
+      "step": 4426
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8428834963697436e-06,
+      "loss": 1.3008,
+      "step": 4427
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8407294504167238e-06,
+      "loss": 1.5254,
+      "step": 4428
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8385758969782507e-06,
+      "loss": 1.3623,
+      "step": 4429
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8364228365455306e-06,
+      "loss": 1.4946,
+      "step": 4430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8342702696096657e-06,
+      "loss": 1.5613,
+      "step": 4431
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8321181966616372e-06,
+      "loss": 1.4763,
+      "step": 4432
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8299666181923204e-06,
+      "loss": 1.1356,
+      "step": 4433
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.827815534692476e-06,
+      "loss": 1.3641,
+      "step": 4434
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.82566494665275e-06,
+      "loss": 1.7679,
+      "step": 4435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8235148545636776e-06,
+      "loss": 1.3346,
+      "step": 4436
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.821365258915677e-06,
+      "loss": 1.6137,
+      "step": 4437
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.81921616019906e-06,
+      "loss": 1.4955,
+      "step": 4438
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8170675589040175e-06,
+      "loss": 1.57,
+      "step": 4439
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8149194555206333e-06,
+      "loss": 1.4604,
+      "step": 4440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.812771850538871e-06,
+      "loss": 1.2185,
+      "step": 4441
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.810624744448588e-06,
+      "loss": 1.1818,
+      "step": 4442
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.808478137739524e-06,
+      "loss": 1.508,
+      "step": 4443
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.806332030901302e-06,
+      "loss": 1.1931,
+      "step": 4444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8041864244234373e-06,
+      "loss": 1.4654,
+      "step": 4445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8020413187953256e-06,
+      "loss": 1.6712,
+      "step": 4446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7998967145062528e-06,
+      "loss": 1.1976,
+      "step": 4447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7977526120453846e-06,
+      "loss": 1.5147,
+      "step": 4448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.79560901190178e-06,
+      "loss": 1.6767,
+      "step": 4449
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.793465914564375e-06,
+      "loss": 1.3383,
+      "step": 4450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.791323320521999e-06,
+      "loss": 1.2511,
+      "step": 4451
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7891812302633593e-06,
+      "loss": 1.3246,
+      "step": 4452
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7870396442770543e-06,
+      "loss": 1.5158,
+      "step": 4453
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7848985630515646e-06,
+      "loss": 1.0654,
+      "step": 4454
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7827579870752542e-06,
+      "loss": 1.6995,
+      "step": 4455
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7806179168363755e-06,
+      "loss": 1.4703,
+      "step": 4456
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.778478352823062e-06,
+      "loss": 1.4796,
+      "step": 4457
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7763392955233354e-06,
+      "loss": 1.4535,
+      "step": 4458
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7742007454250963e-06,
+      "loss": 1.4254,
+      "step": 4459
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7720627030161373e-06,
+      "loss": 1.7443,
+      "step": 4460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7699251687841265e-06,
+      "loss": 1.5912,
+      "step": 4461
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7677881432166256e-06,
+      "loss": 0.9634,
+      "step": 4462
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7656516268010697e-06,
+      "loss": 1.0696,
+      "step": 4463
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.763515620024786e-06,
+      "loss": 1.4575,
+      "step": 4464
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.761380123374985e-06,
+      "loss": 1.0428,
+      "step": 4465
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7592451373387542e-06,
+      "loss": 1.117,
+      "step": 4466
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7571106624030735e-06,
+      "loss": 1.4309,
+      "step": 4467
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7549766990547973e-06,
+      "loss": 1.2868,
+      "step": 4468
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7528432477806717e-06,
+      "loss": 1.5595,
+      "step": 4469
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.750710309067318e-06,
+      "loss": 1.4395,
+      "step": 4470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.748577883401249e-06,
+      "loss": 2.0306,
+      "step": 4471
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7464459712688517e-06,
+      "loss": 1.4337,
+      "step": 4472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7443145731564046e-06,
+      "loss": 1.2926,
+      "step": 4473
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7421836895500613e-06,
+      "loss": 1.3837,
+      "step": 4474
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7400533209358625e-06,
+      "loss": 1.3899,
+      "step": 4475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7379234677997335e-06,
+      "loss": 1.4464,
+      "step": 4476
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7357941306274738e-06,
+      "loss": 1.4444,
+      "step": 4477
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7336653099047748e-06,
+      "loss": 1.3974,
+      "step": 4478
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.731537006117202e-06,
+      "loss": 1.5286,
+      "step": 4479
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7294092197502096e-06,
+      "loss": 1.4331,
+      "step": 4480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.727281951289128e-06,
+      "loss": 1.4811,
+      "step": 4481
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7251552012191763e-06,
+      "loss": 1.4972,
+      "step": 4482
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7230289700254464e-06,
+      "loss": 1.3605,
+      "step": 4483
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7209032581929217e-06,
+      "loss": 1.5809,
+      "step": 4484
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.718778066206457e-06,
+      "loss": 1.3241,
+      "step": 4485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.716653394550798e-06,
+      "loss": 1.8028,
+      "step": 4486
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7145292437105665e-06,
+      "loss": 1.6152,
+      "step": 4487
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.712405614170265e-06,
+      "loss": 1.3042,
+      "step": 4488
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7102825064142813e-06,
+      "loss": 1.9381,
+      "step": 4489
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.708159920926878e-06,
+      "loss": 1.5947,
+      "step": 4490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7060378581922052e-06,
+      "loss": 1.3565,
+      "step": 4491
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7039163186942885e-06,
+      "loss": 1.0805,
+      "step": 4492
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.701795302917038e-06,
+      "loss": 1.5131,
+      "step": 4493
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6996748113442397e-06,
+      "loss": 1.7763,
+      "step": 4494
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6975548444595665e-06,
+      "loss": 1.3971,
+      "step": 4495
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6954354027465655e-06,
+      "loss": 1.5491,
+      "step": 4496
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.693316486688667e-06,
+      "loss": 1.3881,
+      "step": 4497
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.691198096769183e-06,
+      "loss": 1.6045,
+      "step": 4498
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6890802334713e-06,
+      "loss": 1.5212,
+      "step": 4499
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.686962897278092e-06,
+      "loss": 1.0566,
+      "step": 4500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6848460886725044e-06,
+      "loss": 1.4428,
+      "step": 4501
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.68272980813737e-06,
+      "loss": 1.2797,
+      "step": 4502
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.680614056155394e-06,
+      "loss": 1.5293,
+      "step": 4503
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6784988332091686e-06,
+      "loss": 1.3628,
+      "step": 4504
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6763841397811576e-06,
+      "loss": 1.2356,
+      "step": 4505
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6742699763537115e-06,
+      "loss": 1.5853,
+      "step": 4506
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.672156343409053e-06,
+      "loss": 1.0407,
+      "step": 4507
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.670043241429288e-06,
+      "loss": 1.4145,
+      "step": 4508
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6679306708964027e-06,
+      "loss": 1.3698,
+      "step": 4509
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6658186322922562e-06,
+      "loss": 1.3701,
+      "step": 4510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6637071260985923e-06,
+      "loss": 1.5776,
+      "step": 4511
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6615961527970293e-06,
+      "loss": 1.5412,
+      "step": 4512
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6594857128690663e-06,
+      "loss": 1.2562,
+      "step": 4513
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.657375806796079e-06,
+      "loss": 1.4784,
+      "step": 4514
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.655266435059325e-06,
+      "loss": 1.331,
+      "step": 4515
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.653157598139932e-06,
+      "loss": 1.3962,
+      "step": 4516
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.651049296518917e-06,
+      "loss": 1.4804,
+      "step": 4517
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6489415306771637e-06,
+      "loss": 1.3416,
+      "step": 4518
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.646834301095441e-06,
+      "loss": 1.6385,
+      "step": 4519
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.644727608254396e-06,
+      "loss": 1.588,
+      "step": 4520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6426214526345456e-06,
+      "loss": 1.3963,
+      "step": 4521
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6405158347162934e-06,
+      "loss": 1.2811,
+      "step": 4522
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.638410754979913e-06,
+      "loss": 1.5581,
+      "step": 4523
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6363062139055613e-06,
+      "loss": 1.2767,
+      "step": 4524
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6342022119732656e-06,
+      "loss": 1.6383,
+      "step": 4525
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.632098749662939e-06,
+      "loss": 1.6823,
+      "step": 4526
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.629995827454361e-06,
+      "loss": 1.5803,
+      "step": 4527
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6278934458271998e-06,
+      "loss": 1.2152,
+      "step": 4528
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6257916052609876e-06,
+      "loss": 1.3765,
+      "step": 4529
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6236903062351424e-06,
+      "loss": 1.9219,
+      "step": 4530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.621589549228959e-06,
+      "loss": 1.356,
+      "step": 4531
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6194893347215998e-06,
+      "loss": 1.61,
+      "step": 4532
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6173896631921134e-06,
+      "loss": 1.3487,
+      "step": 4533
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6152905351194175e-06,
+      "loss": 1.4357,
+      "step": 4534
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6131919509823114e-06,
+      "loss": 1.413,
+      "step": 4535
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6110939112594626e-06,
+      "loss": 1.4161,
+      "step": 4536
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.608996416429425e-06,
+      "loss": 1.5108,
+      "step": 4537
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6068994669706184e-06,
+      "loss": 1.3128,
+      "step": 4538
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6048030633613453e-06,
+      "loss": 1.4179,
+      "step": 4539
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6027072060797776e-06,
+      "loss": 1.6128,
+      "step": 4540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6006118956039673e-06,
+      "loss": 1.3824,
+      "step": 4541
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.598517132411842e-06,
+      "loss": 1.6327,
+      "step": 4542
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5964229169811987e-06,
+      "loss": 1.1962,
+      "step": 4543
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5943292497897176e-06,
+      "loss": 1.6922,
+      "step": 4544
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5922361313149457e-06,
+      "loss": 1.2749,
+      "step": 4545
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.590143562034312e-06,
+      "loss": 1.4314,
+      "step": 4546
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.588051542425114e-06,
+      "loss": 1.4027,
+      "step": 4547
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5859600729645294e-06,
+      "loss": 1.4807,
+      "step": 4548
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5838691541296046e-06,
+      "loss": 1.2412,
+      "step": 4549
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.581778786397266e-06,
+      "loss": 1.0084,
+      "step": 4550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5796889702443127e-06,
+      "loss": 1.5525,
+      "step": 4551
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5775997061474143e-06,
+      "loss": 1.3828,
+      "step": 4552
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5755109945831212e-06,
+      "loss": 1.8776,
+      "step": 4553
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.57342283602785e-06,
+      "loss": 1.2744,
+      "step": 4554
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.571335230957898e-06,
+      "loss": 1.3935,
+      "step": 4555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5692481798494308e-06,
+      "loss": 1.5009,
+      "step": 4556
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.567161683178493e-06,
+      "loss": 1.569,
+      "step": 4557
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5650757414209963e-06,
+      "loss": 1.534,
+      "step": 4558
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5629903550527343e-06,
+      "loss": 1.8612,
+      "step": 4559
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.560905524549364e-06,
+      "loss": 1.887,
+      "step": 4560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.558821250386423e-06,
+      "loss": 1.4638,
+      "step": 4561
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.556737533039322e-06,
+      "loss": 1.7149,
+      "step": 4562
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.554654372983338e-06,
+      "loss": 1.4357,
+      "step": 4563
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.552571770693629e-06,
+      "loss": 1.5351,
+      "step": 4564
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.550489726645219e-06,
+      "loss": 1.2805,
+      "step": 4565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.548408241313011e-06,
+      "loss": 1.4439,
+      "step": 4566
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.546327315171774e-06,
+      "loss": 1.2459,
+      "step": 4567
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.544246948696156e-06,
+      "loss": 1.6095,
+      "step": 4568
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5421671423606696e-06,
+      "loss": 1.3573,
+      "step": 4569
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5400878966397095e-06,
+      "loss": 1.5191,
+      "step": 4570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5380092120075323e-06,
+      "loss": 1.3499,
+      "step": 4571
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.535931088938274e-06,
+      "loss": 1.2189,
+      "step": 4572
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5338535279059408e-06,
+      "loss": 1.358,
+      "step": 4573
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.531776529384407e-06,
+      "loss": 1.6435,
+      "step": 4574
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5297000938474246e-06,
+      "loss": 1.3783,
+      "step": 4575
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5276242217686115e-06,
+      "loss": 1.4949,
+      "step": 4576
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5255489136214622e-06,
+      "loss": 1.5629,
+      "step": 4577
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5234741698793365e-06,
+      "loss": 1.9348,
+      "step": 4578
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.521399991015473e-06,
+      "loss": 1.4632,
+      "step": 4579
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5193263775029737e-06,
+      "loss": 1.5259,
+      "step": 4580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.517253329814819e-06,
+      "loss": 1.2442,
+      "step": 4581
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.515180848423853e-06,
+      "loss": 1.3956,
+      "step": 4582
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5131089338027948e-06,
+      "loss": 1.5586,
+      "step": 4583
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5110375864242377e-06,
+      "loss": 1.5282,
+      "step": 4584
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5089668067606365e-06,
+      "loss": 1.4289,
+      "step": 4585
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5068965952843254e-06,
+      "loss": 1.3484,
+      "step": 4586
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5048269524675017e-06,
+      "loss": 1.689,
+      "step": 4587
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5027578787822403e-06,
+      "loss": 1.4675,
+      "step": 4588
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.500689374700479e-06,
+      "loss": 1.2289,
+      "step": 4589
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4986214406940325e-06,
+      "loss": 1.458,
+      "step": 4590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4965540772345783e-06,
+      "loss": 1.9311,
+      "step": 4591
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.494487284793671e-06,
+      "loss": 1.4319,
+      "step": 4592
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4924210638427287e-06,
+      "loss": 1.3284,
+      "step": 4593
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4903554148530432e-06,
+      "loss": 1.5063,
+      "step": 4594
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.488290338295777e-06,
+      "loss": 1.0305,
+      "step": 4595
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4862258346419556e-06,
+      "loss": 1.8101,
+      "step": 4596
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.484161904362481e-06,
+      "loss": 1.3318,
+      "step": 4597
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4820985479281184e-06,
+      "loss": 1.4187,
+      "step": 4598
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4800357658095085e-06,
+      "loss": 1.0889,
+      "step": 4599
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4779735584771542e-06,
+      "loss": 1.2719,
+      "step": 4600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.475911926401433e-06,
+      "loss": 1.3025,
+      "step": 4601
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4738508700525864e-06,
+      "loss": 1.6042,
+      "step": 4602
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4717903899007306e-06,
+      "loss": 1.5177,
+      "step": 4603
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.469730486415842e-06,
+      "loss": 1.5197,
+      "step": 4604
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4676711600677733e-06,
+      "loss": 1.5205,
+      "step": 4605
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4656124113262436e-06,
+      "loss": 1.4303,
+      "step": 4606
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4635542406608352e-06,
+      "loss": 1.0672,
+      "step": 4607
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4614966485410073e-06,
+      "loss": 1.6155,
+      "step": 4608
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.459439635436078e-06,
+      "loss": 1.5036,
+      "step": 4609
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4573832018152403e-06,
+      "loss": 1.4803,
+      "step": 4610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.45532734814755e-06,
+      "loss": 1.2891,
+      "step": 4611
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4532720749019364e-06,
+      "loss": 1.2349,
+      "step": 4612
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4512173825471885e-06,
+      "loss": 1.5119,
+      "step": 4613
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.449163271551972e-06,
+      "loss": 1.4016,
+      "step": 4614
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.447109742384811e-06,
+      "loss": 1.6287,
+      "step": 4615
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.445056795514105e-06,
+      "loss": 1.4274,
+      "step": 4616
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4430044314081134e-06,
+      "loss": 1.2204,
+      "step": 4617
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.44095265053497e-06,
+      "loss": 1.3791,
+      "step": 4618
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.438901453362667e-06,
+      "loss": 1.2706,
+      "step": 4619
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.436850840359073e-06,
+      "loss": 1.3288,
+      "step": 4620
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.434800811991914e-06,
+      "loss": 1.6702,
+      "step": 4621
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4327513687287897e-06,
+      "loss": 0.85,
+      "step": 4622
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.430702511037165e-06,
+      "loss": 1.3201,
+      "step": 4623
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4286542393843665e-06,
+      "loss": 1.595,
+      "step": 4624
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.426606554237595e-06,
+      "loss": 1.4061,
+      "step": 4625
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4245594560639086e-06,
+      "loss": 1.0585,
+      "step": 4626
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4225129453302403e-06,
+      "loss": 1.3465,
+      "step": 4627
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.420467022503381e-06,
+      "loss": 1.4363,
+      "step": 4628
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4184216880499956e-06,
+      "loss": 1.3286,
+      "step": 4629
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4163769424366064e-06,
+      "loss": 1.4035,
+      "step": 4630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4143327861296088e-06,
+      "loss": 1.8411,
+      "step": 4631
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4122892195952585e-06,
+      "loss": 1.5251,
+      "step": 4632
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.410246243299681e-06,
+      "loss": 1.4292,
+      "step": 4633
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.408203857708862e-06,
+      "loss": 1.2243,
+      "step": 4634
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4061620632886583e-06,
+      "loss": 1.2029,
+      "step": 4635
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4041208605047866e-06,
+      "loss": 1.6455,
+      "step": 4636
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4020802498228333e-06,
+      "loss": 1.7671,
+      "step": 4637
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.400040231708245e-06,
+      "loss": 1.5352,
+      "step": 4638
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3980008066263384e-06,
+      "loss": 1.2933,
+      "step": 4639
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3959619750422884e-06,
+      "loss": 1.2507,
+      "step": 4640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.393923737421141e-06,
+      "loss": 1.0728,
+      "step": 4641
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.391886094227805e-06,
+      "loss": 0.988,
+      "step": 4642
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.389849045927049e-06,
+      "loss": 1.3918,
+      "step": 4643
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3878125929835134e-06,
+      "loss": 1.2484,
+      "step": 4644
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.385776735861695e-06,
+      "loss": 1.7239,
+      "step": 4645
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.383741475025963e-06,
+      "loss": 1.3994,
+      "step": 4646
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3817068109405415e-06,
+      "loss": 1.5953,
+      "step": 4647
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.379672744069527e-06,
+      "loss": 1.4772,
+      "step": 4648
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3776392748768733e-06,
+      "loss": 1.5908,
+      "step": 4649
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3756064038264033e-06,
+      "loss": 1.4514,
+      "step": 4650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3735741313817974e-06,
+      "loss": 1.6008,
+      "step": 4651
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3715424580066053e-06,
+      "loss": 1.72,
+      "step": 4652
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.369511384164236e-06,
+      "loss": 1.5328,
+      "step": 4653
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3674809103179654e-06,
+      "loss": 1.5436,
+      "step": 4654
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.365451036930927e-06,
+      "loss": 1.3302,
+      "step": 4655
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3634217644661246e-06,
+      "loss": 1.2972,
+      "step": 4656
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.361393093386417e-06,
+      "loss": 1.4044,
+      "step": 4657
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3593650241545323e-06,
+      "loss": 1.6332,
+      "step": 4658
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3573375572330603e-06,
+      "loss": 1.4641,
+      "step": 4659
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3553106930844495e-06,
+      "loss": 1.8195,
+      "step": 4660
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3532844321710154e-06,
+      "loss": 1.4313,
+      "step": 4661
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3512587749549314e-06,
+      "loss": 1.7348,
+      "step": 4662
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3492337218982396e-06,
+      "loss": 1.7195,
+      "step": 4663
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.347209273462836e-06,
+      "loss": 1.5264,
+      "step": 4664
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3451854301104877e-06,
+      "loss": 1.1279,
+      "step": 4665
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3431621923028146e-06,
+      "loss": 1.4836,
+      "step": 4666
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3411395605013082e-06,
+      "loss": 1.3559,
+      "step": 4667
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.339117535167312e-06,
+      "loss": 1.486,
+      "step": 4668
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.337096116762037e-06,
+      "loss": 1.5433,
+      "step": 4669
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.335075305746558e-06,
+      "loss": 1.4353,
+      "step": 4670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3330551025818038e-06,
+      "loss": 1.3112,
+      "step": 4671
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.331035507728572e-06,
+      "loss": 1.353,
+      "step": 4672
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.329016521647514e-06,
+      "loss": 1.3089,
+      "step": 4673
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3269981447991506e-06,
+      "loss": 1.5289,
+      "step": 4674
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.324980377643856e-06,
+      "loss": 1.4652,
+      "step": 4675
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3229632206418727e-06,
+      "loss": 1.5174,
+      "step": 4676
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.320946674253296e-06,
+      "loss": 1.5476,
+      "step": 4677
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3189307389380893e-06,
+      "loss": 1.2167,
+      "step": 4678
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.316915415156071e-06,
+      "loss": 1.2471,
+      "step": 4679
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3149007033669226e-06,
+      "loss": 1.4181,
+      "step": 4680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.312886604030189e-06,
+      "loss": 1.5349,
+      "step": 4681
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3108731176052673e-06,
+      "loss": 1.1282,
+      "step": 4682
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3088602445514246e-06,
+      "loss": 1.4726,
+      "step": 4683
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3068479853277793e-06,
+      "loss": 1.3678,
+      "step": 4684
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3048363403933164e-06,
+      "loss": 1.5676,
+      "step": 4685
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3028253102068754e-06,
+      "loss": 2.0614,
+      "step": 4686
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3008148952271615e-06,
+      "loss": 1.5701,
+      "step": 4687
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.298805095912733e-06,
+      "loss": 1.4532,
+      "step": 4688
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.296795912722014e-06,
+      "loss": 1.2006,
+      "step": 4689
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2947873461132823e-06,
+      "loss": 1.5334,
+      "step": 4690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.292779396544679e-06,
+      "loss": 1.6828,
+      "step": 4691
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2907720644742064e-06,
+      "loss": 1.6226,
+      "step": 4692
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2887653503597183e-06,
+      "loss": 1.3585,
+      "step": 4693
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2867592546589362e-06,
+      "loss": 1.5037,
+      "step": 4694
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2847537778294325e-06,
+      "loss": 1.1847,
+      "step": 4695
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2827489203286464e-06,
+      "loss": 1.5809,
+      "step": 4696
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2807446826138686e-06,
+      "loss": 1.5339,
+      "step": 4697
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2787410651422546e-06,
+      "loss": 1.3979,
+      "step": 4698
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.276738068370812e-06,
+      "loss": 1.5782,
+      "step": 4699
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2747356927564156e-06,
+      "loss": 1.5854,
+      "step": 4700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.272733938755788e-06,
+      "loss": 1.2867,
+      "step": 4701
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.270732806825517e-06,
+      "loss": 1.4723,
+      "step": 4702
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.26873229742205e-06,
+      "loss": 1.2388,
+      "step": 4703
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.266732411001684e-06,
+      "loss": 1.5536,
+      "step": 4704
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2647331480205842e-06,
+      "loss": 1.294,
+      "step": 4705
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.262734508934764e-06,
+      "loss": 1.4071,
+      "step": 4706
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.260736494200103e-06,
+      "loss": 1.3794,
+      "step": 4707
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2587391042723313e-06,
+      "loss": 1.1822,
+      "step": 4708
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.256742339607042e-06,
+      "loss": 1.3081,
+      "step": 4709
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2547462006596803e-06,
+      "loss": 1.4637,
+      "step": 4710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2527506878855554e-06,
+      "loss": 1.2362,
+      "step": 4711
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2507558017398263e-06,
+      "loss": 1.2533,
+      "step": 4712
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2487615426775133e-06,
+      "loss": 1.4262,
+      "step": 4713
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2467679111534963e-06,
+      "loss": 1.3107,
+      "step": 4714
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.244774907622504e-06,
+      "loss": 1.3191,
+      "step": 4715
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2427825325391306e-06,
+      "loss": 1.2707,
+      "step": 4716
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.240790786357819e-06,
+      "loss": 1.148,
+      "step": 4717
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.238799669532877e-06,
+      "loss": 1.4003,
+      "step": 4718
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.23680918251846e-06,
+      "loss": 1.1187,
+      "step": 4719
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2348193257685887e-06,
+      "loss": 1.4072,
+      "step": 4720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.232830099737131e-06,
+      "loss": 1.4389,
+      "step": 4721
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2308415048778197e-06,
+      "loss": 1.3042,
+      "step": 4722
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2288535416442358e-06,
+      "loss": 1.4774,
+      "step": 4723
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2268662104898216e-06,
+      "loss": 1.5963,
+      "step": 4724
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2248795118678746e-06,
+      "loss": 0.9891,
+      "step": 4725
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2228934462315437e-06,
+      "loss": 1.2473,
+      "step": 4726
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.220908014033841e-06,
+      "loss": 1.2612,
+      "step": 4727
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2189232157276247e-06,
+      "loss": 1.5365,
+      "step": 4728
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2169390517656185e-06,
+      "loss": 1.5856,
+      "step": 4729
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2149555226003915e-06,
+      "loss": 1.4967,
+      "step": 4730
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2129726286843777e-06,
+      "loss": 1.6154,
+      "step": 4731
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2109903704698565e-06,
+      "loss": 1.4265,
+      "step": 4732
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2090087484089716e-06,
+      "loss": 1.593,
+      "step": 4733
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2070277629537133e-06,
+      "loss": 1.3738,
+      "step": 4734
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2050474145559326e-06,
+      "loss": 1.3088,
+      "step": 4735
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2030677036673345e-06,
+      "loss": 1.3484,
+      "step": 4736
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2010886307394733e-06,
+      "loss": 1.2354,
+      "step": 4737
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.199110196223767e-06,
+      "loss": 1.4856,
+      "step": 4738
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1971324005714774e-06,
+      "loss": 1.5974,
+      "step": 4739
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.19515524423373e-06,
+      "loss": 1.8089,
+      "step": 4740
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1931787276614968e-06,
+      "loss": 1.5998,
+      "step": 4741
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1912028513056118e-06,
+      "loss": 1.642,
+      "step": 4742
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1892276156167535e-06,
+      "loss": 1.5439,
+      "step": 4743
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1872530210454623e-06,
+      "loss": 1.4197,
+      "step": 4744
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1852790680421305e-06,
+      "loss": 1.5402,
+      "step": 4745
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1833057570570006e-06,
+      "loss": 1.617,
+      "step": 4746
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1813330885401736e-06,
+      "loss": 1.6574,
+      "step": 4747
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1793610629415972e-06,
+      "loss": 1.4384,
+      "step": 4748
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1773896807110824e-06,
+      "loss": 1.5314,
+      "step": 4749
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1754189422982825e-06,
+      "loss": 1.2646,
+      "step": 4750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.173448848152713e-06,
+      "loss": 1.0666,
+      "step": 4751
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1714793987237353e-06,
+      "loss": 1.532,
+      "step": 4752
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.169510594460571e-06,
+      "loss": 1.2884,
+      "step": 4753
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.167542435812286e-06,
+      "loss": 1.3604,
+      "step": 4754
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.165574923227806e-06,
+      "loss": 1.5239,
+      "step": 4755
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1636080571559094e-06,
+      "loss": 1.2807,
+      "step": 4756
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1616418380452194e-06,
+      "loss": 1.5357,
+      "step": 4757
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.159676266344222e-06,
+      "loss": 1.5661,
+      "step": 4758
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1577113425012457e-06,
+      "loss": 1.1007,
+      "step": 4759
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.15574706696448e-06,
+      "loss": 1.4869,
+      "step": 4760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.153783440181959e-06,
+      "loss": 1.6796,
+      "step": 4761
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1518204626015753e-06,
+      "loss": 1.5954,
+      "step": 4762
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1498581346710673e-06,
+      "loss": 1.3768,
+      "step": 4763
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1478964568380316e-06,
+      "loss": 1.3862,
+      "step": 4764
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.14593542954991e-06,
+      "loss": 1.369,
+      "step": 4765
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1439750532540012e-06,
+      "loss": 1.4227,
+      "step": 4766
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.142015328397454e-06,
+      "loss": 1.6312,
+      "step": 4767
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1400562554272653e-06,
+      "loss": 1.2581,
+      "step": 4768
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1380978347902892e-06,
+      "loss": 1.0108,
+      "step": 4769
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1361400669332237e-06,
+      "loss": 1.2448,
+      "step": 4770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.134182952302627e-06,
+      "loss": 1.6479,
+      "step": 4771
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.132226491344899e-06,
+      "loss": 1.5424,
+      "step": 4772
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1302706845062976e-06,
+      "loss": 1.746,
+      "step": 4773
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1283155322329256e-06,
+      "loss": 1.5042,
+      "step": 4774
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.126361034970744e-06,
+      "loss": 1.6477,
+      "step": 4775
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1244071931655552e-06,
+      "loss": 1.4078,
+      "step": 4776
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1224540072630195e-06,
+      "loss": 1.2673,
+      "step": 4777
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.120501477708647e-06,
+      "loss": 1.419,
+      "step": 4778
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.118549604947791e-06,
+      "loss": 1.4756,
+      "step": 4779
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1165983894256647e-06,
+      "loss": 1.5349,
+      "step": 4780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1146478315873238e-06,
+      "loss": 1.6901,
+      "step": 4781
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.112697931877679e-06,
+      "loss": 1.5753,
+      "step": 4782
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1107486907414865e-06,
+      "loss": 1.1693,
+      "step": 4783
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1088001086233574e-06,
+      "loss": 1.2943,
+      "step": 4784
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1068521859677463e-06,
+      "loss": 1.6199,
+      "step": 4785
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.104904923218965e-06,
+      "loss": 1.3664,
+      "step": 4786
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.102958320821166e-06,
+      "loss": 1.595,
+      "step": 4787
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.101012379218358e-06,
+      "loss": 1.3031,
+      "step": 4788
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.099067098854398e-06,
+      "loss": 1.5223,
+      "step": 4789
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.097122480172988e-06,
+      "loss": 1.6604,
+      "step": 4790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.095178523617685e-06,
+      "loss": 1.439,
+      "step": 4791
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.093235229631889e-06,
+      "loss": 1.4313,
+      "step": 4792
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0912925986588547e-06,
+      "loss": 1.631,
+      "step": 4793
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0893506311416794e-06,
+      "loss": 1.529,
+      "step": 4794
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0874093275233158e-06,
+      "loss": 1.325,
+      "step": 4795
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0854686882465587e-06,
+      "loss": 1.7251,
+      "step": 4796
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0835287137540566e-06,
+      "loss": 1.2563,
+      "step": 4797
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0815894044883016e-06,
+      "loss": 1.302,
+      "step": 4798
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.079650760891639e-06,
+      "loss": 1.1861,
+      "step": 4799
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.07771278340626e-06,
+      "loss": 1.6248,
+      "step": 4800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0757754724742013e-06,
+      "loss": 1.5275,
+      "step": 4801
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0738388285373532e-06,
+      "loss": 1.541,
+      "step": 4802
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0719028520374475e-06,
+      "loss": 1.4348,
+      "step": 4803
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.06996754341607e-06,
+      "loss": 1.4093,
+      "step": 4804
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.068032903114648e-06,
+      "loss": 1.3908,
+      "step": 4805
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0660989315744624e-06,
+      "loss": 1.4312,
+      "step": 4806
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.064165629236636e-06,
+      "loss": 1.7611,
+      "step": 4807
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.062232996542144e-06,
+      "loss": 1.2492,
+      "step": 4808
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.060301033931803e-06,
+      "loss": 1.5308,
+      "step": 4809
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0583697418462828e-06,
+      "loss": 1.6738,
+      "step": 4810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.056439120726099e-06,
+      "loss": 1.5962,
+      "step": 4811
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0545091710116098e-06,
+      "loss": 1.0096,
+      "step": 4812
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0525798931430256e-06,
+      "loss": 1.5652,
+      "step": 4813
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.050651287560399e-06,
+      "loss": 1.6789,
+      "step": 4814
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.048723354703634e-06,
+      "loss": 1.6762,
+      "step": 4815
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.046796095012475e-06,
+      "loss": 1.5242,
+      "step": 4816
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0448695089265207e-06,
+      "loss": 1.7678,
+      "step": 4817
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0429435968852084e-06,
+      "loss": 1.4135,
+      "step": 4818
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0410183593278287e-06,
+      "loss": 1.133,
+      "step": 4819
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.039093796693511e-06,
+      "loss": 1.3601,
+      "step": 4820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.037169909421236e-06,
+      "loss": 1.3719,
+      "step": 4821
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.035246697949832e-06,
+      "loss": 1.5573,
+      "step": 4822
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.033324162717965e-06,
+      "loss": 1.4273,
+      "step": 4823
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0314023041641567e-06,
+      "loss": 1.6633,
+      "step": 4824
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.029481122726766e-06,
+      "loss": 1.5675,
+      "step": 4825
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0275606188440035e-06,
+      "loss": 1.1555,
+      "step": 4826
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0256407929539203e-06,
+      "loss": 1.7725,
+      "step": 4827
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0237216454944176e-06,
+      "loss": 1.2079,
+      "step": 4828
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.021803176903237e-06,
+      "loss": 1.7536,
+      "step": 4829
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0198853876179715e-06,
+      "loss": 1.2594,
+      "step": 4830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.017968278076051e-06,
+      "loss": 1.4532,
+      "step": 4831
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.016051848714758e-06,
+      "loss": 1.2758,
+      "step": 4832
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0141360999712165e-06,
+      "loss": 1.6315,
+      "step": 4833
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0122210322823936e-06,
+      "loss": 1.7886,
+      "step": 4834
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0103066460851054e-06,
+      "loss": 1.4263,
+      "step": 4835
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.008392941816007e-06,
+      "loss": 1.2883,
+      "step": 4836
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0064799199116043e-06,
+      "loss": 1.4614,
+      "step": 4837
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0045675808082405e-06,
+      "loss": 0.9657,
+      "step": 4838
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.002655924942111e-06,
+      "loss": 1.4435,
+      "step": 4839
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.000744952749246e-06,
+      "loss": 1.3667,
+      "step": 4840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9988346646655292e-06,
+      "loss": 1.6701,
+      "step": 4841
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.996925061126684e-06,
+      "loss": 1.1157,
+      "step": 4842
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9950161425682747e-06,
+      "loss": 1.5793,
+      "step": 4843
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9931079094257156e-06,
+      "loss": 1.2504,
+      "step": 4844
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.991200362134258e-06,
+      "loss": 1.4248,
+      "step": 4845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9892935011290037e-06,
+      "loss": 1.2929,
+      "step": 4846
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9873873268448906e-06,
+      "loss": 1.5225,
+      "step": 4847
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9854818397167074e-06,
+      "loss": 1.5092,
+      "step": 4848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.983577040179079e-06,
+      "loss": 1.3217,
+      "step": 4849
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.98167292866648e-06,
+      "loss": 1.5114,
+      "step": 4850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.979769505613222e-06,
+      "loss": 1.0328,
+      "step": 4851
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9778667714534637e-06,
+      "loss": 1.5715,
+      "step": 4852
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9759647266212084e-06,
+      "loss": 1.6736,
+      "step": 4853
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9740633715502945e-06,
+      "loss": 1.3578,
+      "step": 4854
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9721627066744114e-06,
+      "loss": 1.1874,
+      "step": 4855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9702627324270845e-06,
+      "loss": 1.2294,
+      "step": 4856
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.968363449241688e-06,
+      "loss": 1.5339,
+      "step": 4857
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9664648575514316e-06,
+      "loss": 1.4108,
+      "step": 4858
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9645669577893742e-06,
+      "loss": 1.3832,
+      "step": 4859
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.962669750388411e-06,
+      "loss": 1.4035,
+      "step": 4860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9607732357812836e-06,
+      "loss": 1.4753,
+      "step": 4861
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9588774144005712e-06,
+      "loss": 1.6734,
+      "step": 4862
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.956982286678701e-06,
+      "loss": 1.4016,
+      "step": 4863
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9550878530479345e-06,
+      "loss": 1.3079,
+      "step": 4864
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9531941139403827e-06,
+      "loss": 1.5324,
+      "step": 4865
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9513010697879903e-06,
+      "loss": 1.2971,
+      "step": 4866
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9494087210225494e-06,
+      "loss": 1.7887,
+      "step": 4867
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.947517068075694e-06,
+      "loss": 1.7825,
+      "step": 4868
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9456261113788923e-06,
+      "loss": 1.6163,
+      "step": 4869
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9437358513634625e-06,
+      "loss": 1.2771,
+      "step": 4870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9418462884605555e-06,
+      "loss": 1.4484,
+      "step": 4871
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.939957423101172e-06,
+      "loss": 1.5258,
+      "step": 4872
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9380692557161453e-06,
+      "loss": 1.3073,
+      "step": 4873
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.936181786736156e-06,
+      "loss": 1.3951,
+      "step": 4874
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9342950165917194e-06,
+      "loss": 1.5393,
+      "step": 4875
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9324089457131984e-06,
+      "loss": 1.2951,
+      "step": 4876
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9305235745307887e-06,
+      "loss": 1.6127,
+      "step": 4877
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.928638903474534e-06,
+      "loss": 1.606,
+      "step": 4878
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9267549329743114e-06,
+      "loss": 1.2185,
+      "step": 4879
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9248716634598442e-06,
+      "loss": 1.338,
+      "step": 4880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9229890953606908e-06,
+      "loss": 1.3384,
+      "step": 4881
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.921107229106254e-06,
+      "loss": 1.444,
+      "step": 4882
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.919226065125771e-06,
+      "loss": 1.4727,
+      "step": 4883
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9173456038483244e-06,
+      "loss": 1.4391,
+      "step": 4884
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9154658457028364e-06,
+      "loss": 1.7134,
+      "step": 4885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9135867911180627e-06,
+      "loss": 1.2009,
+      "step": 4886
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.911708440522606e-06,
+      "loss": 1.4563,
+      "step": 4887
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9098307943449017e-06,
+      "loss": 1.498,
+      "step": 4888
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9079538530132315e-06,
+      "loss": 1.3203,
+      "step": 4889
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9060776169557083e-06,
+      "loss": 1.5192,
+      "step": 4890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9042020866002919e-06,
+      "loss": 1.2059,
+      "step": 4891
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9023272623747756e-06,
+      "loss": 1.2632,
+      "step": 4892
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9004531447067964e-06,
+      "loss": 1.4005,
+      "step": 4893
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8985797340238233e-06,
+      "loss": 1.3929,
+      "step": 4894
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8967070307531727e-06,
+      "loss": 1.2552,
+      "step": 4895
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8948350353219913e-06,
+      "loss": 1.5145,
+      "step": 4896
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8929637481572715e-06,
+      "loss": 1.5991,
+      "step": 4897
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8910931696858376e-06,
+      "loss": 1.372,
+      "step": 4898
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8892233003343585e-06,
+      "loss": 1.5153,
+      "step": 4899
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.887354140529336e-06,
+      "loss": 1.7427,
+      "step": 4900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8854856906971146e-06,
+      "loss": 1.426,
+      "step": 4901
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.883617951263872e-06,
+      "loss": 1.1603,
+      "step": 4902
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.881750922655628e-06,
+      "loss": 1.4723,
+      "step": 4903
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.87988460529824e-06,
+      "loss": 1.5559,
+      "step": 4904
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8780189996173998e-06,
+      "loss": 1.282,
+      "step": 4905
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8761541060386413e-06,
+      "loss": 1.2448,
+      "step": 4906
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.874289924987331e-06,
+      "loss": 1.6595,
+      "step": 4907
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8724264568886785e-06,
+      "loss": 1.6252,
+      "step": 4908
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8705637021677248e-06,
+      "loss": 1.1877,
+      "step": 4909
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8687016612493542e-06,
+      "loss": 1.695,
+      "step": 4910
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8668403345582824e-06,
+      "loss": 1.3373,
+      "step": 4911
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.864979722519068e-06,
+      "loss": 1.7259,
+      "step": 4912
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8631198255561007e-06,
+      "loss": 1.5574,
+      "step": 4913
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8612606440936133e-06,
+      "loss": 1.236,
+      "step": 4914
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8594021785556682e-06,
+      "loss": 1.3058,
+      "step": 4915
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8575444293661727e-06,
+      "loss": 1.4084,
+      "step": 4916
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.855687396948862e-06,
+      "loss": 1.2886,
+      "step": 4917
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8538310817273143e-06,
+      "loss": 1.1963,
+      "step": 4918
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8519754841249444e-06,
+      "loss": 1.4259,
+      "step": 4919
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8501206045649973e-06,
+      "loss": 1.6779,
+      "step": 4920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.848266443470561e-06,
+      "loss": 1.4172,
+      "step": 4921
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8464130012645547e-06,
+      "loss": 1.5418,
+      "step": 4922
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8445602783697375e-06,
+      "loss": 1.6015,
+      "step": 4923
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8427082752087005e-06,
+      "loss": 1.3518,
+      "step": 4924
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.840856992203875e-06,
+      "loss": 1.3942,
+      "step": 4925
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.839006429777523e-06,
+      "loss": 1.4358,
+      "step": 4926
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8371565883517478e-06,
+      "loss": 1.7947,
+      "step": 4927
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8353074683484817e-06,
+      "loss": 1.3713,
+      "step": 4928
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.833459070189499e-06,
+      "loss": 1.3537,
+      "step": 4929
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.831611394296407e-06,
+      "loss": 1.4971,
+      "step": 4930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8297644410906451e-06,
+      "loss": 1.3262,
+      "step": 4931
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8279182109934929e-06,
+      "loss": 1.3567,
+      "step": 4932
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8260727044260602e-06,
+      "loss": 1.4806,
+      "step": 4933
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8242279218092968e-06,
+      "loss": 1.0932,
+      "step": 4934
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8223838635639811e-06,
+      "loss": 1.2184,
+      "step": 4935
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8205405301107343e-06,
+      "loss": 1.6547,
+      "step": 4936
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.818697921870004e-06,
+      "loss": 1.7606,
+      "step": 4937
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8168560392620794e-06,
+      "loss": 1.5133,
+      "step": 4938
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8150148827070785e-06,
+      "loss": 1.6711,
+      "step": 4939
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8131744526249567e-06,
+      "loss": 1.1313,
+      "step": 4940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.811334749435506e-06,
+      "loss": 1.3428,
+      "step": 4941
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8094957735583463e-06,
+      "loss": 1.4264,
+      "step": 4942
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8076575254129386e-06,
+      "loss": 1.657,
+      "step": 4943
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8058200054185704e-06,
+      "loss": 1.4343,
+      "step": 4944
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8039832139943714e-06,
+      "loss": 1.3312,
+      "step": 4945
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.802147151559297e-06,
+      "loss": 1.7164,
+      "step": 4946
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8003118185321438e-06,
+      "loss": 1.6576,
+      "step": 4947
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7984772153315349e-06,
+      "loss": 1.2747,
+      "step": 4948
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7966433423759327e-06,
+      "loss": 1.6427,
+      "step": 4949
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7948102000836321e-06,
+      "loss": 1.6001,
+      "step": 4950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7929777888727562e-06,
+      "loss": 1.2893,
+      "step": 4951
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7911461091612697e-06,
+      "loss": 1.349,
+      "step": 4952
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7893151613669618e-06,
+      "loss": 1.7114,
+      "step": 4953
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7874849459074628e-06,
+      "loss": 1.2125,
+      "step": 4954
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7856554632002282e-06,
+      "loss": 1.6898,
+      "step": 4955
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7838267136625536e-06,
+      "loss": 1.0768,
+      "step": 4956
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7819986977115611e-06,
+      "loss": 1.6979,
+      "step": 4957
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7801714157642114e-06,
+      "loss": 1.2821,
+      "step": 4958
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7783448682372917e-06,
+      "loss": 1.1887,
+      "step": 4959
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.776519055547426e-06,
+      "loss": 1.2701,
+      "step": 4960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7746939781110717e-06,
+      "loss": 1.6983,
+      "step": 4961
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772869636344512e-06,
+      "loss": 1.4421,
+      "step": 4962
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7710460306638711e-06,
+      "loss": 1.3706,
+      "step": 4963
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7692231614850958e-06,
+      "loss": 1.4279,
+      "step": 4964
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7674010292239746e-06,
+      "loss": 1.2407,
+      "step": 4965
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.765579634296119e-06,
+      "loss": 1.5633,
+      "step": 4966
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7637589771169804e-06,
+      "loss": 1.5078,
+      "step": 4967
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7619390581018341e-06,
+      "loss": 1.495,
+      "step": 4968
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7601198776657947e-06,
+      "loss": 1.7261,
+      "step": 4969
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.758301436223801e-06,
+      "loss": 1.4013,
+      "step": 4970
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7564837341906283e-06,
+      "loss": 1.2894,
+      "step": 4971
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.754666771980884e-06,
+      "loss": 1.5405,
+      "step": 4972
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7528505500090003e-06,
+      "loss": 1.3504,
+      "step": 4973
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7510350686892491e-06,
+      "loss": 1.5639,
+      "step": 4974
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7492203284357245e-06,
+      "loss": 1.4509,
+      "step": 4975
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7474063296623605e-06,
+      "loss": 1.6205,
+      "step": 4976
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.745593072782914e-06,
+      "loss": 1.3062,
+      "step": 4977
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.743780558210979e-06,
+      "loss": 1.5094,
+      "step": 4978
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7419687863599749e-06,
+      "loss": 1.4196,
+      "step": 4979
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7401577576431573e-06,
+      "loss": 1.4148,
+      "step": 4980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.738347472473605e-06,
+      "loss": 1.6545,
+      "step": 4981
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.736537931264234e-06,
+      "loss": 1.2539,
+      "step": 4982
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7347291344277894e-06,
+      "loss": 1.4822,
+      "step": 4983
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7329210823768422e-06,
+      "loss": 1.2686,
+      "step": 4984
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7311137755237994e-06,
+      "loss": 1.417,
+      "step": 4985
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.729307214280892e-06,
+      "loss": 1.0711,
+      "step": 4986
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7275013990601868e-06,
+      "loss": 1.3928,
+      "step": 4987
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7256963302735752e-06,
+      "loss": 1.4092,
+      "step": 4988
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.723892008332783e-06,
+      "loss": 1.2523,
+      "step": 4989
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.722088433649361e-06,
+      "loss": 1.4325,
+      "step": 4990
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.720285606634695e-06,
+      "loss": 1.2321,
+      "step": 4991
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.718483527699994e-06,
+      "loss": 1.8233,
+      "step": 4992
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7166821972563013e-06,
+      "loss": 1.3713,
+      "step": 4993
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7148816157144899e-06,
+      "loss": 1.4967,
+      "step": 4994
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7130817834852553e-06,
+      "loss": 1.7838,
+      "step": 4995
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.711282700979131e-06,
+      "loss": 1.5011,
+      "step": 4996
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7094843686064721e-06,
+      "loss": 1.7686,
+      "step": 4997
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7076867867774681e-06,
+      "loss": 1.5791,
+      "step": 4998
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7058899559021324e-06,
+      "loss": 1.3577,
+      "step": 4999
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.704093876390312e-06,
+      "loss": 1.6437,
+      "step": 5000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.702298548651678e-06,
+      "loss": 1.4059,
+      "step": 5001
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.700503973095734e-06,
+      "loss": 1.4099,
+      "step": 5002
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6987101501318075e-06,
+      "loss": 1.355,
+      "step": 5003
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6969170801690592e-06,
+      "loss": 1.4189,
+      "step": 5004
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.695124763616477e-06,
+      "loss": 1.5442,
+      "step": 5005
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6933332008828723e-06,
+      "loss": 1.3256,
+      "step": 5006
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.691542392376892e-06,
+      "loss": 1.5192,
+      "step": 5007
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6897523385070026e-06,
+      "loss": 1.3245,
+      "step": 5008
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6879630396815073e-06,
+      "loss": 1.2924,
+      "step": 5009
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.686174496308529e-06,
+      "loss": 1.5808,
+      "step": 5010
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6843867087960252e-06,
+      "loss": 1.4485,
+      "step": 5011
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6825996775517744e-06,
+      "loss": 1.3946,
+      "step": 5012
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6808134029833896e-06,
+      "loss": 1.3864,
+      "step": 5013
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6790278854983033e-06,
+      "loss": 1.5546,
+      "step": 5014
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6772431255037824e-06,
+      "loss": 1.3527,
+      "step": 5015
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6754591234069194e-06,
+      "loss": 1.414,
+      "step": 5016
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6736758796146296e-06,
+      "loss": 1.2044,
+      "step": 5017
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6718933945336619e-06,
+      "loss": 1.2329,
+      "step": 5018
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6701116685705843e-06,
+      "loss": 1.3426,
+      "step": 5019
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6683307021318006e-06,
+      "loss": 1.5992,
+      "step": 5020
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6665504956235329e-06,
+      "loss": 1.2616,
+      "step": 5021
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.664771049451837e-06,
+      "loss": 1.3115,
+      "step": 5022
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6629923640225898e-06,
+      "loss": 1.3779,
+      "step": 5023
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6612144397414997e-06,
+      "loss": 1.0804,
+      "step": 5024
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6594372770140954e-06,
+      "loss": 1.6048,
+      "step": 5025
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.657660876245737e-06,
+      "loss": 1.2333,
+      "step": 5026
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6558852378416113e-06,
+      "loss": 1.7161,
+      "step": 5027
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.654110362206725e-06,
+      "loss": 1.4446,
+      "step": 5028
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6523362497459183e-06,
+      "loss": 1.6701,
+      "step": 5029
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6505629008638512e-06,
+      "loss": 0.8656,
+      "step": 5030
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6487903159650148e-06,
+      "loss": 1.4859,
+      "step": 5031
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6470184954537205e-06,
+      "loss": 1.6764,
+      "step": 5032
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6452474397341111e-06,
+      "loss": 1.2452,
+      "step": 5033
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6434771492101487e-06,
+      "loss": 1.7424,
+      "step": 5034
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6417076242856284e-06,
+      "loss": 1.2892,
+      "step": 5035
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6399388653641617e-06,
+      "loss": 1.447,
+      "step": 5036
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6381708728491936e-06,
+      "loss": 1.3158,
+      "step": 5037
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6364036471439908e-06,
+      "loss": 1.135,
+      "step": 5038
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6346371886516432e-06,
+      "loss": 1.3835,
+      "step": 5039
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6328714977750698e-06,
+      "loss": 1.2071,
+      "step": 5040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6311065749170102e-06,
+      "loss": 1.4973,
+      "step": 5041
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6293424204800334e-06,
+      "loss": 1.4391,
+      "step": 5042
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6275790348665277e-06,
+      "loss": 1.292,
+      "step": 5043
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6258164184787123e-06,
+      "loss": 1.5404,
+      "step": 5044
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6240545717186252e-06,
+      "loss": 1.6453,
+      "step": 5045
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6222934949881313e-06,
+      "loss": 1.3779,
+      "step": 5046
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6205331886889231e-06,
+      "loss": 1.2944,
+      "step": 5047
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6187736532225106e-06,
+      "loss": 1.6183,
+      "step": 5048
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6170148889902342e-06,
+      "loss": 0.9436,
+      "step": 5049
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.615256896393253e-06,
+      "loss": 1.3904,
+      "step": 5050
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.613499675832556e-06,
+      "loss": 1.5955,
+      "step": 5051
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6117432277089496e-06,
+      "loss": 1.46,
+      "step": 5052
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6099875524230707e-06,
+      "loss": 1.2488,
+      "step": 5053
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6082326503753732e-06,
+      "loss": 1.8498,
+      "step": 5054
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6064785219661416e-06,
+      "loss": 1.1831,
+      "step": 5055
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6047251675954766e-06,
+      "loss": 1.3881,
+      "step": 5056
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6029725876633085e-06,
+      "loss": 1.5532,
+      "step": 5057
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6012207825693893e-06,
+      "loss": 1.3452,
+      "step": 5058
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5994697527132907e-06,
+      "loss": 1.3719,
+      "step": 5059
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5977194984944133e-06,
+      "loss": 1.2531,
+      "step": 5060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5959700203119755e-06,
+      "loss": 1.5487,
+      "step": 5061
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5942213185650234e-06,
+      "loss": 1.5882,
+      "step": 5062
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5924733936524205e-06,
+      "loss": 1.3604,
+      "step": 5063
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.59072624597286e-06,
+      "loss": 1.4201,
+      "step": 5064
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5889798759248504e-06,
+      "loss": 1.4971,
+      "step": 5065
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5872342839067305e-06,
+      "loss": 1.4931,
+      "step": 5066
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5854894703166536e-06,
+      "loss": 1.6124,
+      "step": 5067
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.583745435552601e-06,
+      "loss": 1.3752,
+      "step": 5068
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5820021800123774e-06,
+      "loss": 1.2072,
+      "step": 5069
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5802597040936036e-06,
+      "loss": 1.4015,
+      "step": 5070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.578518008193729e-06,
+      "loss": 1.8392,
+      "step": 5071
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5767770927100197e-06,
+      "loss": 1.3651,
+      "step": 5072
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5750369580395685e-06,
+      "loss": 1.5115,
+      "step": 5073
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5732976045792859e-06,
+      "loss": 1.2082,
+      "step": 5074
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5715590327259095e-06,
+      "loss": 1.3757,
+      "step": 5075
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5698212428759913e-06,
+      "loss": 1.5957,
+      "step": 5076
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.568084235425913e-06,
+      "loss": 1.3705,
+      "step": 5077
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5663480107718704e-06,
+      "loss": 1.4375,
+      "step": 5078
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5646125693098863e-06,
+      "loss": 1.2673,
+      "step": 5079
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5628779114358034e-06,
+      "loss": 1.6252,
+      "step": 5080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5611440375452825e-06,
+      "loss": 1.5001,
+      "step": 5081
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.559410948033811e-06,
+      "loss": 1.3753,
+      "step": 5082
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5576786432966922e-06,
+      "loss": 1.3192,
+      "step": 5083
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5559471237290548e-06,
+      "loss": 1.2692,
+      "step": 5084
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5542163897258438e-06,
+      "loss": 1.623,
+      "step": 5085
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5524864416818302e-06,
+      "loss": 1.3314,
+      "step": 5086
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.550757279991601e-06,
+      "loss": 1.5241,
+      "step": 5087
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5490289050495678e-06,
+      "loss": 1.4885,
+      "step": 5088
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5473013172499584e-06,
+      "loss": 1.4405,
+      "step": 5089
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5455745169868247e-06,
+      "loss": 1.6341,
+      "step": 5090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5438485046540396e-06,
+      "loss": 1.1724,
+      "step": 5091
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.542123280645292e-06,
+      "loss": 1.454,
+      "step": 5092
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5403988453540952e-06,
+      "loss": 1.7502,
+      "step": 5093
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.538675199173779e-06,
+      "loss": 1.2725,
+      "step": 5094
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5369523424974974e-06,
+      "loss": 1.6936,
+      "step": 5095
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.535230275718219e-06,
+      "loss": 1.173,
+      "step": 5096
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5335089992287382e-06,
+      "loss": 1.3272,
+      "step": 5097
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5317885134216636e-06,
+      "loss": 1.4479,
+      "step": 5098
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.530068818689429e-06,
+      "loss": 1.2056,
+      "step": 5099
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5283499154242815e-06,
+      "loss": 1.1487,
+      "step": 5100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.526631804018292e-06,
+      "loss": 1.5494,
+      "step": 5101
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.524914484863352e-06,
+      "loss": 1.4807,
+      "step": 5102
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5231979583511668e-06,
+      "loss": 1.1576,
+      "step": 5103
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5214822248732669e-06,
+      "loss": 1.5255,
+      "step": 5104
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.519767284820996e-06,
+      "loss": 1.0809,
+      "step": 5105
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5180531385855236e-06,
+      "loss": 1.4747,
+      "step": 5106
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5163397865578306e-06,
+      "loss": 1.7157,
+      "step": 5107
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5146272291287245e-06,
+      "loss": 1.7586,
+      "step": 5108
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.512915466688824e-06,
+      "loss": 1.4622,
+      "step": 5109
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.511204499628574e-06,
+      "loss": 1.1272,
+      "step": 5110
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.50949432833823e-06,
+      "loss": 1.3235,
+      "step": 5111
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.507784953207873e-06,
+      "loss": 1.2691,
+      "step": 5112
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5060763746274004e-06,
+      "loss": 1.2547,
+      "step": 5113
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5043685929865231e-06,
+      "loss": 1.2214,
+      "step": 5114
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5026616086747785e-06,
+      "loss": 1.5112,
+      "step": 5115
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5009554220815148e-06,
+      "loss": 1.4541,
+      "step": 5116
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4992500335959037e-06,
+      "loss": 1.2323,
+      "step": 5117
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4975454436069292e-06,
+      "loss": 1.3026,
+      "step": 5118
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4958416525033998e-06,
+      "loss": 1.6974,
+      "step": 5119
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.494138660673935e-06,
+      "loss": 1.1157,
+      "step": 5120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4924364685069791e-06,
+      "loss": 1.3321,
+      "step": 5121
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4907350763907862e-06,
+      "loss": 1.4716,
+      "step": 5122
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4890344847134364e-06,
+      "loss": 1.2283,
+      "step": 5123
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4873346938628186e-06,
+      "loss": 1.4319,
+      "step": 5124
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4856357042266463e-06,
+      "loss": 1.5066,
+      "step": 5125
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4839375161924446e-06,
+      "loss": 1.4618,
+      "step": 5126
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4822401301475615e-06,
+      "loss": 1.3245,
+      "step": 5127
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4805435464791551e-06,
+      "loss": 1.4615,
+      "step": 5128
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4788477655742067e-06,
+      "loss": 1.1384,
+      "step": 5129
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.477152787819513e-06,
+      "loss": 1.3361,
+      "step": 5130
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4754586136016841e-06,
+      "loss": 1.3998,
+      "step": 5131
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4737652433071515e-06,
+      "loss": 1.6198,
+      "step": 5132
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4720726773221588e-06,
+      "loss": 1.4491,
+      "step": 5133
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.470380916032772e-06,
+      "loss": 1.4474,
+      "step": 5134
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4686899598248656e-06,
+      "loss": 1.6226,
+      "step": 5135
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4669998090841391e-06,
+      "loss": 1.2595,
+      "step": 5136
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4653104641961009e-06,
+      "loss": 1.3396,
+      "step": 5137
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.463621925546081e-06,
+      "loss": 1.6346,
+      "step": 5138
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.461934193519221e-06,
+      "loss": 1.5888,
+      "step": 5139
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4602472685004838e-06,
+      "loss": 1.4602,
+      "step": 5140
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4585611508746422e-06,
+      "loss": 1.7389,
+      "step": 5141
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4568758410262907e-06,
+      "loss": 1.4414,
+      "step": 5142
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4551913393398337e-06,
+      "loss": 1.1836,
+      "step": 5143
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4535076461994974e-06,
+      "loss": 1.1182,
+      "step": 5144
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4518247619893173e-06,
+      "loss": 1.417,
+      "step": 5145
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4501426870931506e-06,
+      "loss": 1.5241,
+      "step": 5146
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.448461421894664e-06,
+      "loss": 1.3239,
+      "step": 5147
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4467809667773436e-06,
+      "loss": 1.3917,
+      "step": 5148
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4451013221244915e-06,
+      "loss": 1.5766,
+      "step": 5149
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4434224883192195e-06,
+      "loss": 1.3424,
+      "step": 5150
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4417444657444608e-06,
+      "loss": 1.2993,
+      "step": 5151
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4400672547829574e-06,
+      "loss": 1.3697,
+      "step": 5152
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4383908558172733e-06,
+      "loss": 1.3819,
+      "step": 5153
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4367152692297799e-06,
+      "loss": 1.2892,
+      "step": 5154
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4350404954026692e-06,
+      "loss": 1.2616,
+      "step": 5155
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4333665347179437e-06,
+      "loss": 1.519,
+      "step": 5156
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.431693387557424e-06,
+      "loss": 1.271,
+      "step": 5157
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4300210543027414e-06,
+      "loss": 1.4111,
+      "step": 5158
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4283495353353454e-06,
+      "loss": 2.0571,
+      "step": 5159
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4266788310364955e-06,
+      "loss": 1.375,
+      "step": 5160
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4250089417872697e-06,
+      "loss": 1.4626,
+      "step": 5161
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4233398679685566e-06,
+      "loss": 1.4633,
+      "step": 5162
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4216716099610623e-06,
+      "loss": 1.5461,
+      "step": 5163
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.420004168145302e-06,
+      "loss": 1.1207,
+      "step": 5164
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4183375429016088e-06,
+      "loss": 1.7187,
+      "step": 5165
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.41667173461013e-06,
+      "loss": 1.3676,
+      "step": 5166
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4150067436508224e-06,
+      "loss": 1.5964,
+      "step": 5167
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.413342570403461e-06,
+      "loss": 1.5212,
+      "step": 5168
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4116792152476295e-06,
+      "loss": 1.7329,
+      "step": 5169
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4100166785627301e-06,
+      "loss": 1.6228,
+      "step": 5170
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.408354960727974e-06,
+      "loss": 1.3413,
+      "step": 5171
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.406694062122389e-06,
+      "loss": 1.1944,
+      "step": 5172
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4050339831248122e-06,
+      "loss": 1.4578,
+      "step": 5173
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4033747241138994e-06,
+      "loss": 1.1296,
+      "step": 5174
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.401716285468112e-06,
+      "loss": 1.5157,
+      "step": 5175
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4000586675657312e-06,
+      "loss": 1.9373,
+      "step": 5176
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3984018707848478e-06,
+      "loss": 1.3618,
+      "step": 5177
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3967458955033642e-06,
+      "loss": 1.4533,
+      "step": 5178
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.395090742098999e-06,
+      "loss": 1.498,
+      "step": 5179
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.393436410949278e-06,
+      "loss": 1.5251,
+      "step": 5180
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3917829024315465e-06,
+      "loss": 1.4463,
+      "step": 5181
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3901302169229542e-06,
+      "loss": 1.591,
+      "step": 5182
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3884783548004704e-06,
+      "loss": 1.2627,
+      "step": 5183
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3868273164408707e-06,
+      "loss": 1.3735,
+      "step": 5184
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3851771022207483e-06,
+      "loss": 1.3475,
+      "step": 5185
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3835277125165026e-06,
+      "loss": 1.2119,
+      "step": 5186
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.381879147704349e-06,
+      "loss": 1.3656,
+      "step": 5187
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3802314081603151e-06,
+      "loss": 1.2283,
+      "step": 5188
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3785844942602366e-06,
+      "loss": 1.6551,
+      "step": 5189
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3769384063797652e-06,
+      "loss": 1.5391,
+      "step": 5190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3752931448943595e-06,
+      "loss": 1.6516,
+      "step": 5191
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.373648710179295e-06,
+      "loss": 1.4826,
+      "step": 5192
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3720051026096526e-06,
+      "loss": 1.5005,
+      "step": 5193
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3703623225603313e-06,
+      "loss": 1.5002,
+      "step": 5194
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3687203704060343e-06,
+      "loss": 1.8545,
+      "step": 5195
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3670792465212828e-06,
+      "loss": 1.4478,
+      "step": 5196
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3654389512804022e-06,
+      "loss": 1.5009,
+      "step": 5197
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3637994850575342e-06,
+      "loss": 1.3515,
+      "step": 5198
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3621608482266308e-06,
+      "loss": 1.3849,
+      "step": 5199
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3605230411614512e-06,
+      "loss": 1.5077,
+      "step": 5200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3588860642355705e-06,
+      "loss": 1.7115,
+      "step": 5201
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3572499178223686e-06,
+      "loss": 1.2695,
+      "step": 5202
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3556146022950417e-06,
+      "loss": 1.3487,
+      "step": 5203
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3539801180265915e-06,
+      "loss": 1.3274,
+      "step": 5204
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3523464653898355e-06,
+      "loss": 1.2754,
+      "step": 5205
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3507136447573955e-06,
+      "loss": 1.3777,
+      "step": 5206
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3490816565017085e-06,
+      "loss": 1.6706,
+      "step": 5207
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3474505009950179e-06,
+      "loss": 1.5646,
+      "step": 5208
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3458201786093795e-06,
+      "loss": 1.3613,
+      "step": 5209
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3441906897166606e-06,
+      "loss": 1.3672,
+      "step": 5210
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3425620346885325e-06,
+      "loss": 1.6244,
+      "step": 5211
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3409342138964836e-06,
+      "loss": 1.5504,
+      "step": 5212
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3393072277118057e-06,
+      "loss": 1.4218,
+      "step": 5213
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3376810765056053e-06,
+      "loss": 1.6611,
+      "step": 5214
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.336055760648794e-06,
+      "loss": 1.4829,
+      "step": 5215
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.334431280512097e-06,
+      "loss": 1.3426,
+      "step": 5216
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3328076364660447e-06,
+      "loss": 1.327,
+      "step": 5217
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3311848288809815e-06,
+      "loss": 1.2327,
+      "step": 5218
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3295628581270553e-06,
+      "loss": 1.245,
+      "step": 5219
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3279417245742288e-06,
+      "loss": 1.7295,
+      "step": 5220
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3263214285922716e-06,
+      "loss": 1.4049,
+      "step": 5221
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3247019705507596e-06,
+      "loss": 1.2635,
+      "step": 5222
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3230833508190826e-06,
+      "loss": 1.4022,
+      "step": 5223
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3214655697664336e-06,
+      "loss": 1.1846,
+      "step": 5224
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.31984862776182e-06,
+      "loss": 1.6386,
+      "step": 5225
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3182325251740524e-06,
+      "loss": 1.277,
+      "step": 5226
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3166172623717549e-06,
+      "loss": 1.2563,
+      "step": 5227
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.315002839723355e-06,
+      "loss": 1.5555,
+      "step": 5228
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3133892575970946e-06,
+      "loss": 1.1994,
+      "step": 5229
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.311776516361017e-06,
+      "loss": 1.5296,
+      "step": 5230
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3101646163829785e-06,
+      "loss": 1.0202,
+      "step": 5231
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3085535580306447e-06,
+      "loss": 1.3103,
+      "step": 5232
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3069433416714827e-06,
+      "loss": 1.178,
+      "step": 5233
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3053339676727755e-06,
+      "loss": 1.1698,
+      "step": 5234
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3037254364016068e-06,
+      "loss": 1.5091,
+      "step": 5235
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3021177482248737e-06,
+      "loss": 1.7584,
+      "step": 5236
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3005109035092766e-06,
+      "loss": 1.4196,
+      "step": 5237
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2989049026213285e-06,
+      "loss": 1.8037,
+      "step": 5238
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2972997459273428e-06,
+      "loss": 1.5752,
+      "step": 5239
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2956954337934475e-06,
+      "loss": 1.3146,
+      "step": 5240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2940919665855762e-06,
+      "loss": 1.8579,
+      "step": 5241
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2924893446694648e-06,
+      "loss": 1.1925,
+      "step": 5242
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2908875684106636e-06,
+      "loss": 1.409,
+      "step": 5243
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2892866381745234e-06,
+      "loss": 1.4447,
+      "step": 5244
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.287686554326209e-06,
+      "loss": 1.7971,
+      "step": 5245
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2860873172306849e-06,
+      "loss": 1.4596,
+      "step": 5246
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2844889272527283e-06,
+      "loss": 1.4683,
+      "step": 5247
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2828913847569185e-06,
+      "loss": 1.3455,
+      "step": 5248
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2812946901076472e-06,
+      "loss": 1.4479,
+      "step": 5249
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2796988436691056e-06,
+      "loss": 1.6015,
+      "step": 5250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2781038458052964e-06,
+      "loss": 1.5342,
+      "step": 5251
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2765096968800295e-06,
+      "loss": 1.1782,
+      "step": 5252
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2749163972569156e-06,
+      "loss": 1.2231,
+      "step": 5253
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2733239472993792e-06,
+      "loss": 1.2507,
+      "step": 5254
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2717323473706434e-06,
+      "loss": 1.5107,
+      "step": 5255
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.270141597833744e-06,
+      "loss": 1.2612,
+      "step": 5256
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2685516990515172e-06,
+      "loss": 1.6497,
+      "step": 5257
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.26696265138661e-06,
+      "loss": 1.6718,
+      "step": 5258
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2653744552014712e-06,
+      "loss": 1.2272,
+      "step": 5259
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.26378711085836e-06,
+      "loss": 1.2776,
+      "step": 5260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2622006187193348e-06,
+      "loss": 1.4751,
+      "step": 5261
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.260614979146265e-06,
+      "loss": 1.3085,
+      "step": 5262
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2590301925008268e-06,
+      "loss": 1.131,
+      "step": 5263
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.257446259144494e-06,
+      "loss": 1.5512,
+      "step": 5264
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2558631794385556e-06,
+      "loss": 1.2296,
+      "step": 5265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2542809537440969e-06,
+      "loss": 1.3278,
+      "step": 5266
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2526995824220156e-06,
+      "loss": 1.5676,
+      "step": 5267
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2511190658330086e-06,
+      "loss": 1.748,
+      "step": 5268
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2495394043375842e-06,
+      "loss": 1.2286,
+      "step": 5269
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2479605982960486e-06,
+      "loss": 1.4692,
+      "step": 5270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2463826480685192e-06,
+      "loss": 1.1742,
+      "step": 5271
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2448055540149135e-06,
+      "loss": 1.4169,
+      "step": 5272
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2432293164949561e-06,
+      "loss": 1.2563,
+      "step": 5273
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2416539358681772e-06,
+      "loss": 1.5027,
+      "step": 5274
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2400794124939081e-06,
+      "loss": 1.3791,
+      "step": 5275
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2385057467312888e-06,
+      "loss": 1.5643,
+      "step": 5276
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2369329389392592e-06,
+      "loss": 1.6359,
+      "step": 5277
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2353609894765677e-06,
+      "loss": 1.7173,
+      "step": 5278
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2337898987017633e-06,
+      "loss": 1.4595,
+      "step": 5279
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2322196669732028e-06,
+      "loss": 1.5807,
+      "step": 5280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2306502946490422e-06,
+      "loss": 1.2767,
+      "step": 5281
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.229081782087248e-06,
+      "loss": 1.6276,
+      "step": 5282
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2275141296455833e-06,
+      "loss": 1.5354,
+      "step": 5283
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.225947337681621e-06,
+      "loss": 1.3533,
+      "step": 5284
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.224381406552736e-06,
+      "loss": 1.3362,
+      "step": 5285
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.222816336616104e-06,
+      "loss": 1.2381,
+      "step": 5286
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2212521282287093e-06,
+      "loss": 1.0485,
+      "step": 5287
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2196887817473336e-06,
+      "loss": 1.2321,
+      "step": 5288
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2181262975285685e-06,
+      "loss": 1.6925,
+      "step": 5289
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2165646759288025e-06,
+      "loss": 1.0533,
+      "step": 5290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2150039173042345e-06,
+      "loss": 1.1269,
+      "step": 5291
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2134440220108585e-06,
+      "loss": 1.2619,
+      "step": 5292
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.21188499040448e-06,
+      "loss": 1.6797,
+      "step": 5293
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2103268228406988e-06,
+      "loss": 1.1977,
+      "step": 5294
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2087695196749244e-06,
+      "loss": 1.3676,
+      "step": 5295
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2072130812623684e-06,
+      "loss": 1.7278,
+      "step": 5296
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.20565750795804e-06,
+      "loss": 1.1309,
+      "step": 5297
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2041028001167576e-06,
+      "loss": 1.2862,
+      "step": 5298
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.202548958093137e-06,
+      "loss": 1.0581,
+      "step": 5299
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2009959822416012e-06,
+      "loss": 1.1632,
+      "step": 5300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1994438729163698e-06,
+      "loss": 1.5091,
+      "step": 5301
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1978926304714722e-06,
+      "loss": 1.4967,
+      "step": 5302
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1963422552607317e-06,
+      "loss": 1.4181,
+      "step": 5303
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.194792747637782e-06,
+      "loss": 1.3199,
+      "step": 5304
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1932441079560508e-06,
+      "loss": 1.5154,
+      "step": 5305
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.191696336568775e-06,
+      "loss": 1.1911,
+      "step": 5306
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.190149433828991e-06,
+      "loss": 1.5244,
+      "step": 5307
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1886034000895341e-06,
+      "loss": 1.0753,
+      "step": 5308
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1870582357030463e-06,
+      "loss": 1.5286,
+      "step": 5309
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1855139410219657e-06,
+      "loss": 1.5797,
+      "step": 5310
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1839705163985387e-06,
+      "loss": 1.4863,
+      "step": 5311
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1824279621848068e-06,
+      "loss": 1.3892,
+      "step": 5312
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1808862787326176e-06,
+      "loss": 1.4033,
+      "step": 5313
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1793454663936171e-06,
+      "loss": 1.1835,
+      "step": 5314
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1778055255192556e-06,
+      "loss": 1.5014,
+      "step": 5315
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1762664564607801e-06,
+      "loss": 1.4058,
+      "step": 5316
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1747282595692433e-06,
+      "loss": 1.5757,
+      "step": 5317
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1731909351954985e-06,
+      "loss": 1.4168,
+      "step": 5318
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1716544836901956e-06,
+      "loss": 1.1534,
+      "step": 5319
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1701189054037915e-06,
+      "loss": 1.6089,
+      "step": 5320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1685842006865378e-06,
+      "loss": 1.6736,
+      "step": 5321
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1670503698884934e-06,
+      "loss": 1.3488,
+      "step": 5322
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.165517413359511e-06,
+      "loss": 1.5327,
+      "step": 5323
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.16398533144925e-06,
+      "loss": 1.2976,
+      "step": 5324
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1624541245071658e-06,
+      "loss": 1.4883,
+      "step": 5325
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1609237928825174e-06,
+      "loss": 1.436,
+      "step": 5326
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.159394336924361e-06,
+      "loss": 1.5829,
+      "step": 5327
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1578657569815554e-06,
+      "loss": 1.417,
+      "step": 5328
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1563380534027613e-06,
+      "loss": 1.2494,
+      "step": 5329
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1548112265364336e-06,
+      "loss": 1.2495,
+      "step": 5330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1532852767308344e-06,
+      "loss": 1.5922,
+      "step": 5331
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1517602043340192e-06,
+      "loss": 1.4046,
+      "step": 5332
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1502360096938486e-06,
+      "loss": 1.3004,
+      "step": 5333
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.148712693157979e-06,
+      "loss": 1.4923,
+      "step": 5334
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.14719025507387e-06,
+      "loss": 1.2352,
+      "step": 5335
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.145668695788777e-06,
+      "loss": 1.3407,
+      "step": 5336
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1441480156497581e-06,
+      "loss": 1.3102,
+      "step": 5337
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1426282150036715e-06,
+      "loss": 1.5447,
+      "step": 5338
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1411092941971702e-06,
+      "loss": 1.608,
+      "step": 5339
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1395912535767117e-06,
+      "loss": 1.6138,
+      "step": 5340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1380740934885486e-06,
+      "loss": 1.2024,
+      "step": 5341
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.136557814278737e-06,
+      "loss": 1.4644,
+      "step": 5342
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1350424162931268e-06,
+      "loss": 1.5036,
+      "step": 5343
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1335278998773725e-06,
+      "loss": 1.5617,
+      "step": 5344
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.132014265376921e-06,
+      "loss": 1.5479,
+      "step": 5345
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1305015131370261e-06,
+      "loss": 1.6323,
+      "step": 5346
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1289896435027326e-06,
+      "loss": 1.3746,
+      "step": 5347
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1274786568188889e-06,
+      "loss": 0.9877,
+      "step": 5348
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1259685534301418e-06,
+      "loss": 1.375,
+      "step": 5349
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1244593336809323e-06,
+      "loss": 1.5573,
+      "step": 5350
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1229509979155057e-06,
+      "loss": 1.2876,
+      "step": 5351
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1214435464779006e-06,
+      "loss": 1.6272,
+      "step": 5352
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1199369797119586e-06,
+      "loss": 1.1174,
+      "step": 5353
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1184312979613139e-06,
+      "loss": 1.391,
+      "step": 5354
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1169265015694058e-06,
+      "loss": 1.4534,
+      "step": 5355
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.115422590879464e-06,
+      "loss": 1.1705,
+      "step": 5356
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1139195662345236e-06,
+      "loss": 1.0837,
+      "step": 5357
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1124174279774107e-06,
+      "loss": 1.2295,
+      "step": 5358
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1109161764507541e-06,
+      "loss": 1.3979,
+      "step": 5359
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.109415811996981e-06,
+      "loss": 1.437,
+      "step": 5360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.10791633495831e-06,
+      "loss": 1.5946,
+      "step": 5361
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.106417745676765e-06,
+      "loss": 1.3681,
+      "step": 5362
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.10492004449416e-06,
+      "loss": 1.5506,
+      "step": 5363
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1034232317521142e-06,
+      "loss": 1.5033,
+      "step": 5364
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1019273077920366e-06,
+      "loss": 1.1294,
+      "step": 5365
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1004322729551398e-06,
+      "loss": 1.5896,
+      "step": 5366
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0989381275824284e-06,
+      "loss": 1.1323,
+      "step": 5367
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0974448720147096e-06,
+      "loss": 1.5607,
+      "step": 5368
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0959525065925807e-06,
+      "loss": 1.3899,
+      "step": 5369
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0944610316564436e-06,
+      "loss": 1.6973,
+      "step": 5370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0929704475464898e-06,
+      "loss": 1.6184,
+      "step": 5371
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0914807546027145e-06,
+      "loss": 1.6349,
+      "step": 5372
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0899919531649034e-06,
+      "loss": 1.4012,
+      "step": 5373
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.088504043572643e-06,
+      "loss": 1.6477,
+      "step": 5374
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0870170261653163e-06,
+      "loss": 1.5832,
+      "step": 5375
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0855309012820992e-06,
+      "loss": 1.5167,
+      "step": 5376
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.084045669261969e-06,
+      "loss": 1.4599,
+      "step": 5377
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0825613304436938e-06,
+      "loss": 1.3417,
+      "step": 5378
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0810778851658438e-06,
+      "loss": 1.4377,
+      "step": 5379
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0795953337667803e-06,
+      "loss": 1.3031,
+      "step": 5380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0781136765846644e-06,
+      "loss": 1.5185,
+      "step": 5381
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0766329139574505e-06,
+      "loss": 1.4834,
+      "step": 5382
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.075153046222892e-06,
+      "loss": 1.4313,
+      "step": 5383
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.073674073718534e-06,
+      "loss": 1.1234,
+      "step": 5384
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.072195996781723e-06,
+      "loss": 1.5385,
+      "step": 5385
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0707188157495947e-06,
+      "loss": 1.3183,
+      "step": 5386
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0692425309590859e-06,
+      "loss": 1.3303,
+      "step": 5387
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0677671427469255e-06,
+      "loss": 1.4836,
+      "step": 5388
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0662926514496409e-06,
+      "loss": 1.4408,
+      "step": 5389
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.064819057403551e-06,
+      "loss": 1.6051,
+      "step": 5390
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0633463609447753e-06,
+      "loss": 1.0984,
+      "step": 5391
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0618745624092226e-06,
+      "loss": 1.32,
+      "step": 5392
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0604036621326009e-06,
+      "loss": 1.3373,
+      "step": 5393
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.058933660450414e-06,
+      "loss": 0.9669,
+      "step": 5394
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0574645576979563e-06,
+      "loss": 1.0531,
+      "step": 5395
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.055996354210323e-06,
+      "loss": 1.5084,
+      "step": 5396
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0545290503223976e-06,
+      "loss": 1.6251,
+      "step": 5397
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0530626463688648e-06,
+      "loss": 1.5064,
+      "step": 5398
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0515971426841992e-06,
+      "loss": 1.2861,
+      "step": 5399
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0501325396026736e-06,
+      "loss": 1.895,
+      "step": 5400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0486688374583509e-06,
+      "loss": 1.2789,
+      "step": 5401
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.047206036585095e-06,
+      "loss": 1.373,
+      "step": 5402
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0457441373165577e-06,
+      "loss": 1.7865,
+      "step": 5403
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0442831399861903e-06,
+      "loss": 1.4518,
+      "step": 5404
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0428230449272336e-06,
+      "loss": 1.462,
+      "step": 5405
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0413638524727277e-06,
+      "loss": 1.5822,
+      "step": 5406
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0399055629555017e-06,
+      "loss": 1.3454,
+      "step": 5407
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0384481767081832e-06,
+      "loss": 0.8914,
+      "step": 5408
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0369916940631903e-06,
+      "loss": 1.4883,
+      "step": 5409
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0355361153527377e-06,
+      "loss": 1.633,
+      "step": 5410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0340814409088335e-06,
+      "loss": 1.327,
+      "step": 5411
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.032627671063276e-06,
+      "loss": 1.8249,
+      "step": 5412
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0311748061476633e-06,
+      "loss": 1.2253,
+      "step": 5413
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0297228464933805e-06,
+      "loss": 1.3702,
+      "step": 5414
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0282717924316128e-06,
+      "loss": 1.4341,
+      "step": 5415
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0268216442933322e-06,
+      "loss": 1.2238,
+      "step": 5416
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0253724024093103e-06,
+      "loss": 1.5493,
+      "step": 5417
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0239240671101065e-06,
+      "loss": 1.209,
+      "step": 5418
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0224766387260782e-06,
+      "loss": 1.3431,
+      "step": 5419
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0210301175873715e-06,
+      "loss": 1.5962,
+      "step": 5420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0195845040239304e-06,
+      "loss": 1.2311,
+      "step": 5421
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0181397983654862e-06,
+      "loss": 1.3693,
+      "step": 5422
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0166960009415695e-06,
+      "loss": 1.7884,
+      "step": 5423
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0152531120814973e-06,
+      "loss": 1.4958,
+      "step": 5424
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.013811132114384e-06,
+      "loss": 1.6547,
+      "step": 5425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0123700613691363e-06,
+      "loss": 1.4578,
+      "step": 5426
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.01092990017445e-06,
+      "loss": 1.4605,
+      "step": 5427
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.009490648858818e-06,
+      "loss": 1.6886,
+      "step": 5428
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0080523077505212e-06,
+      "loss": 1.374,
+      "step": 5429
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.006614877177638e-06,
+      "loss": 1.0687,
+      "step": 5430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.005178357468033e-06,
+      "loss": 1.472,
+      "step": 5431
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0037427489493694e-06,
+      "loss": 1.416,
+      "step": 5432
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0023080519490968e-06,
+      "loss": 1.5028,
+      "step": 5433
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0008742667944625e-06,
+      "loss": 1.5925,
+      "step": 5434
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.994413938124997e-07,
+      "loss": 1.5114,
+      "step": 5435
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.980094333300388e-07,
+      "loss": 1.5504,
+      "step": 5436
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.965783856737005e-07,
+      "loss": 1.4911,
+      "step": 5437
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.95148251169894e-07,
+      "loss": 1.3481,
+      "step": 5438
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.937190301448269e-07,
+      "loss": 1.3954,
+      "step": 5439
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.922907229244905e-07,
+      "loss": 1.3403,
+      "step": 5440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.908633298346753e-07,
+      "loss": 1.5735,
+      "step": 5441
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.894368512009566e-07,
+      "loss": 1.5396,
+      "step": 5442
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.880112873487068e-07,
+      "loss": 1.1402,
+      "step": 5443
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.865866386030847e-07,
+      "loss": 1.5783,
+      "step": 5444
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.851629052890444e-07,
+      "loss": 1.7161,
+      "step": 5445
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.837400877313302e-07,
+      "loss": 1.3256,
+      "step": 5446
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.823181862544745e-07,
+      "loss": 1.6262,
+      "step": 5447
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.808972011828055e-07,
+      "loss": 1.3615,
+      "step": 5448
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.794771328404373e-07,
+      "loss": 1.2731,
+      "step": 5449
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.780579815512807e-07,
+      "loss": 1.3191,
+      "step": 5450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.766397476390315e-07,
+      "loss": 1.5319,
+      "step": 5451
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.752224314271813e-07,
+      "loss": 1.5366,
+      "step": 5452
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.73806033239007e-07,
+      "loss": 1.5821,
+      "step": 5453
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.72390553397583e-07,
+      "loss": 1.3744,
+      "step": 5454
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.709759922257667e-07,
+      "loss": 1.6836,
+      "step": 5455
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.695623500462114e-07,
+      "loss": 1.1719,
+      "step": 5456
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.681496271813602e-07,
+      "loss": 1.4553,
+      "step": 5457
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.667378239534426e-07,
+      "loss": 1.076,
+      "step": 5458
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.65326940684484e-07,
+      "loss": 1.2596,
+      "step": 5459
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.639169776962938e-07,
+      "loss": 1.47,
+      "step": 5460
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.625079353104778e-07,
+      "loss": 1.5134,
+      "step": 5461
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.610998138484262e-07,
+      "loss": 1.4788,
+      "step": 5462
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.596926136313245e-07,
+      "loss": 1.3961,
+      "step": 5463
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.582863349801418e-07,
+      "loss": 1.4688,
+      "step": 5464
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.568809782156435e-07,
+      "loss": 1.6142,
+      "step": 5465
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.554765436583796e-07,
+      "loss": 1.5053,
+      "step": 5466
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.540730316286922e-07,
+      "loss": 1.3492,
+      "step": 5467
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.526704424467142e-07,
+      "loss": 1.6232,
+      "step": 5468
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.512687764323647e-07,
+      "loss": 1.3402,
+      "step": 5469
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.498680339053551e-07,
+      "loss": 1.1636,
+      "step": 5470
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.484682151851832e-07,
+      "loss": 1.4921,
+      "step": 5471
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.470693205911402e-07,
+      "loss": 1.5795,
+      "step": 5472
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.45671350442302e-07,
+      "loss": 1.4358,
+      "step": 5473
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.442743050575382e-07,
+      "loss": 1.6906,
+      "step": 5474
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.428781847555024e-07,
+      "loss": 1.3176,
+      "step": 5475
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.414829898546429e-07,
+      "loss": 1.793,
+      "step": 5476
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.400887206731907e-07,
+      "loss": 1.1577,
+      "step": 5477
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.386953775291708e-07,
+      "loss": 1.3356,
+      "step": 5478
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.373029607403955e-07,
+      "loss": 1.7807,
+      "step": 5479
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.359114706244637e-07,
+      "loss": 0.9803,
+      "step": 5480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.345209074987666e-07,
+      "loss": 1.2981,
+      "step": 5481
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.331312716804791e-07,
+      "loss": 1.494,
+      "step": 5482
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.317425634865712e-07,
+      "loss": 1.2984,
+      "step": 5483
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.303547832337934e-07,
+      "loss": 1.1024,
+      "step": 5484
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.289679312386929e-07,
+      "loss": 1.6164,
+      "step": 5485
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.275820078175967e-07,
+      "loss": 1.626,
+      "step": 5486
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.261970132866282e-07,
+      "loss": 1.1861,
+      "step": 5487
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.248129479616919e-07,
+      "loss": 1.3786,
+      "step": 5488
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.23429812158485e-07,
+      "loss": 1.3496,
+      "step": 5489
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.220476061924915e-07,
+      "loss": 1.238,
+      "step": 5490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.206663303789815e-07,
+      "loss": 1.6179,
+      "step": 5491
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.192859850330166e-07,
+      "loss": 1.4481,
+      "step": 5492
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.179065704694406e-07,
+      "loss": 1.32,
+      "step": 5493
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.165280870028919e-07,
+      "loss": 1.3335,
+      "step": 5494
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.151505349477901e-07,
+      "loss": 1.5185,
+      "step": 5495
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.137739146183478e-07,
+      "loss": 1.4408,
+      "step": 5496
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.123982263285597e-07,
+      "loss": 1.3953,
+      "step": 5497
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.110234703922133e-07,
+      "loss": 1.2963,
+      "step": 5498
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.096496471228783e-07,
+      "loss": 1.3885,
+      "step": 5499
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.082767568339157e-07,
+      "loss": 1.5919,
+      "step": 5500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.069047998384728e-07,
+      "loss": 1.6205,
+      "step": 5501
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.05533776449482e-07,
+      "loss": 1.38,
+      "step": 5502
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.041636869796661e-07,
+      "loss": 1.6434,
+      "step": 5503
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.027945317415305e-07,
+      "loss": 1.4543,
+      "step": 5504
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.01426311047372e-07,
+      "loss": 1.4856,
+      "step": 5505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.000590252092701e-07,
+      "loss": 1.1142,
+      "step": 5506
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.986926745390956e-07,
+      "loss": 1.2665,
+      "step": 5507
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.973272593485011e-07,
+      "loss": 1.7718,
+      "step": 5508
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.959627799489306e-07,
+      "loss": 1.4773,
+      "step": 5509
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.945992366516104e-07,
+      "loss": 1.5081,
+      "step": 5510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.93236629767556e-07,
+      "loss": 1.5408,
+      "step": 5511
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.918749596075698e-07,
+      "loss": 1.4644,
+      "step": 5512
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.905142264822364e-07,
+      "loss": 1.5643,
+      "step": 5513
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.891544307019329e-07,
+      "loss": 1.3395,
+      "step": 5514
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.877955725768161e-07,
+      "loss": 1.3742,
+      "step": 5515
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.864376524168344e-07,
+      "loss": 1.5712,
+      "step": 5516
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.850806705317183e-07,
+      "loss": 1.6356,
+      "step": 5517
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.837246272309873e-07,
+      "loss": 1.1488,
+      "step": 5518
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.823695228239438e-07,
+      "loss": 1.6831,
+      "step": 5519
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.810153576196795e-07,
+      "loss": 1.2945,
+      "step": 5520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.796621319270676e-07,
+      "loss": 1.3584,
+      "step": 5521
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.783098460547712e-07,
+      "loss": 1.6935,
+      "step": 5522
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.769585003112385e-07,
+      "loss": 1.7646,
+      "step": 5523
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.756080950046991e-07,
+      "loss": 1.42,
+      "step": 5524
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.742586304431733e-07,
+      "loss": 1.5388,
+      "step": 5525
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.729101069344626e-07,
+      "loss": 1.3971,
+      "step": 5526
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.715625247861581e-07,
+      "loss": 1.3386,
+      "step": 5527
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.702158843056319e-07,
+      "loss": 1.2883,
+      "step": 5528
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.688701858000448e-07,
+      "loss": 1.4583,
+      "step": 5529
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.675254295763397e-07,
+      "loss": 1.5005,
+      "step": 5530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.661816159412479e-07,
+      "loss": 1.5412,
+      "step": 5531
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.648387452012813e-07,
+      "loss": 1.3285,
+      "step": 5532
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.634968176627412e-07,
+      "loss": 1.3596,
+      "step": 5533
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.621558336317132e-07,
+      "loss": 1.6688,
+      "step": 5534
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.608157934140638e-07,
+      "loss": 1.5091,
+      "step": 5535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.594766973154489e-07,
+      "loss": 1.6483,
+      "step": 5536
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.581385456413044e-07,
+      "loss": 1.4066,
+      "step": 5537
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.568013386968565e-07,
+      "loss": 1.7224,
+      "step": 5538
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.554650767871098e-07,
+      "loss": 1.3693,
+      "step": 5539
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.541297602168591e-07,
+      "loss": 1.7628,
+      "step": 5540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.527953892906782e-07,
+      "loss": 1.3793,
+      "step": 5541
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.51461964312929e-07,
+      "loss": 1.4551,
+      "step": 5542
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.501294855877579e-07,
+      "loss": 1.5734,
+      "step": 5543
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.48797953419091e-07,
+      "loss": 1.3569,
+      "step": 5544
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.474673681106444e-07,
+      "loss": 1.3876,
+      "step": 5545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.46137729965913e-07,
+      "loss": 1.4416,
+      "step": 5546
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.448090392881797e-07,
+      "loss": 1.3372,
+      "step": 5547
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.434812963805078e-07,
+      "loss": 1.0637,
+      "step": 5548
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.421545015457483e-07,
+      "loss": 1.6601,
+      "step": 5549
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.408286550865319e-07,
+      "loss": 1.3758,
+      "step": 5550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.395037573052766e-07,
+      "loss": 1.5118,
+      "step": 5551
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.381798085041804e-07,
+      "loss": 1.2964,
+      "step": 5552
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.368568089852281e-07,
+      "loss": 1.4451,
+      "step": 5553
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.355347590501877e-07,
+      "loss": 1.2629,
+      "step": 5554
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.342136590006067e-07,
+      "loss": 1.39,
+      "step": 5555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.328935091378221e-07,
+      "loss": 1.3662,
+      "step": 5556
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.315743097629481e-07,
+      "loss": 1.339,
+      "step": 5557
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.302560611768867e-07,
+      "loss": 1.3224,
+      "step": 5558
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.289387636803192e-07,
+      "loss": 1.3809,
+      "step": 5559
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.276224175737152e-07,
+      "loss": 1.4734,
+      "step": 5560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.263070231573201e-07,
+      "loss": 1.7486,
+      "step": 5561
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.249925807311699e-07,
+      "loss": 1.3672,
+      "step": 5562
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.236790905950764e-07,
+      "loss": 1.3863,
+      "step": 5563
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.223665530486397e-07,
+      "loss": 1.4522,
+      "step": 5564
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.210549683912406e-07,
+      "loss": 1.1642,
+      "step": 5565
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.19744336922041e-07,
+      "loss": 1.689,
+      "step": 5566
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.184346589399883e-07,
+      "loss": 1.0754,
+      "step": 5567
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.171259347438093e-07,
+      "loss": 1.2322,
+      "step": 5568
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.158181646320162e-07,
+      "loss": 1.6179,
+      "step": 5569
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.145113489029011e-07,
+      "loss": 1.4826,
+      "step": 5570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.132054878545415e-07,
+      "loss": 1.3939,
+      "step": 5571
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.119005817847924e-07,
+      "loss": 1.4176,
+      "step": 5572
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.105966309912966e-07,
+      "loss": 1.3185,
+      "step": 5573
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.092936357714737e-07,
+      "loss": 1.1795,
+      "step": 5574
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.079915964225288e-07,
+      "loss": 1.3444,
+      "step": 5575
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.066905132414493e-07,
+      "loss": 1.3031,
+      "step": 5576
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.053903865250007e-07,
+      "loss": 1.0004,
+      "step": 5577
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.040912165697357e-07,
+      "loss": 1.5194,
+      "step": 5578
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.027930036719828e-07,
+      "loss": 1.4606,
+      "step": 5579
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.014957481278579e-07,
+      "loss": 1.7204,
+      "step": 5580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.001994502332538e-07,
+      "loss": 1.4062,
+      "step": 5581
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.989041102838496e-07,
+      "loss": 1.375,
+      "step": 5582
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.976097285751006e-07,
+      "loss": 1.1902,
+      "step": 5583
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.963163054022483e-07,
+      "loss": 1.4405,
+      "step": 5584
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.950238410603117e-07,
+      "loss": 1.2795,
+      "step": 5585
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.937323358440935e-07,
+      "loss": 1.1707,
+      "step": 5586
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.92441790048179e-07,
+      "loss": 1.2602,
+      "step": 5587
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.911522039669295e-07,
+      "loss": 1.5049,
+      "step": 5588
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.898635778944935e-07,
+      "loss": 1.5813,
+      "step": 5589
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.885759121247949e-07,
+      "loss": 1.0719,
+      "step": 5590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.872892069515442e-07,
+      "loss": 1.5502,
+      "step": 5591
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.860034626682267e-07,
+      "loss": 1.6279,
+      "step": 5592
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.847186795681144e-07,
+      "loss": 1.3543,
+      "step": 5593
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.834348579442552e-07,
+      "loss": 1.6185,
+      "step": 5594
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.821519980894821e-07,
+      "loss": 1.591,
+      "step": 5595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.808701002964042e-07,
+      "loss": 1.2693,
+      "step": 5596
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.795891648574144e-07,
+      "loss": 1.4127,
+      "step": 5597
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.783091920646868e-07,
+      "loss": 1.3604,
+      "step": 5598
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.770301822101712e-07,
+      "loss": 1.3132,
+      "step": 5599
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.757521355856035e-07,
+      "loss": 1.4361,
+      "step": 5600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.744750524824951e-07,
+      "loss": 1.6546,
+      "step": 5601
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.731989331921419e-07,
+      "loss": 1.3507,
+      "step": 5602
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.719237780056149e-07,
+      "loss": 1.5428,
+      "step": 5603
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.706495872137715e-07,
+      "loss": 1.4363,
+      "step": 5604
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.693763611072431e-07,
+      "loss": 1.3926,
+      "step": 5605
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.681040999764455e-07,
+      "loss": 1.3506,
+      "step": 5606
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.668328041115702e-07,
+      "loss": 1.4077,
+      "step": 5607
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.65562473802593e-07,
+      "loss": 1.7551,
+      "step": 5608
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.642931093392675e-07,
+      "loss": 1.1732,
+      "step": 5609
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.630247110111256e-07,
+      "loss": 1.6333,
+      "step": 5610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.617572791074812e-07,
+      "loss": 1.6954,
+      "step": 5611
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.604908139174255e-07,
+      "loss": 1.393,
+      "step": 5612
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.592253157298319e-07,
+      "loss": 1.4932,
+      "step": 5613
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.579607848333498e-07,
+      "loss": 1.3393,
+      "step": 5614
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.56697221516412e-07,
+      "loss": 1.616,
+      "step": 5615
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.554346260672263e-07,
+      "loss": 1.2955,
+      "step": 5616
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.541729987737845e-07,
+      "loss": 1.3322,
+      "step": 5617
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.52912339923852e-07,
+      "loss": 1.5767,
+      "step": 5618
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.516526498049776e-07,
+      "loss": 1.2354,
+      "step": 5619
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.503939287044892e-07,
+      "loss": 1.3282,
+      "step": 5620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.491361769094902e-07,
+      "loss": 1.3476,
+      "step": 5621
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.478793947068669e-07,
+      "loss": 1.3835,
+      "step": 5622
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.466235823832807e-07,
+      "loss": 1.3851,
+      "step": 5623
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.453687402251752e-07,
+      "loss": 1.6301,
+      "step": 5624
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.441148685187694e-07,
+      "loss": 1.1943,
+      "step": 5625
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.428619675500648e-07,
+      "loss": 1.3003,
+      "step": 5626
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.416100376048374e-07,
+      "loss": 1.7234,
+      "step": 5627
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.403590789686454e-07,
+      "loss": 1.4229,
+      "step": 5628
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.391090919268224e-07,
+      "loss": 1.2396,
+      "step": 5629
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.378600767644828e-07,
+      "loss": 1.3588,
+      "step": 5630
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.366120337665166e-07,
+      "loss": 1.519,
+      "step": 5631
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.353649632175957e-07,
+      "loss": 1.3855,
+      "step": 5632
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.341188654021664e-07,
+      "loss": 1.5697,
+      "step": 5633
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.328737406044567e-07,
+      "loss": 1.6618,
+      "step": 5634
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.31629589108469e-07,
+      "loss": 1.2609,
+      "step": 5635
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.303864111979864e-07,
+      "loss": 1.4623,
+      "step": 5636
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.291442071565707e-07,
+      "loss": 1.0895,
+      "step": 5637
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.279029772675572e-07,
+      "loss": 1.9887,
+      "step": 5638
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.266627218140637e-07,
+      "loss": 1.273,
+      "step": 5639
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.254234410789828e-07,
+      "loss": 1.6632,
+      "step": 5640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.241851353449869e-07,
+      "loss": 1.562,
+      "step": 5641
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.229478048945232e-07,
+      "loss": 1.2822,
+      "step": 5642
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.217114500098199e-07,
+      "loss": 1.4528,
+      "step": 5643
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.204760709728792e-07,
+      "loss": 1.6406,
+      "step": 5644
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.192416680654845e-07,
+      "loss": 1.5858,
+      "step": 5645
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.180082415691925e-07,
+      "loss": 1.2504,
+      "step": 5646
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.167757917653412e-07,
+      "loss": 1.3026,
+      "step": 5647
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.155443189350408e-07,
+      "loss": 1.2126,
+      "step": 5648
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.143138233591856e-07,
+      "loss": 1.1686,
+      "step": 5649
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.130843053184394e-07,
+      "loss": 1.4647,
+      "step": 5650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.11855765093249e-07,
+      "loss": 1.7772,
+      "step": 5651
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.106282029638345e-07,
+      "loss": 1.6765,
+      "step": 5652
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.094016192101954e-07,
+      "loss": 1.1414,
+      "step": 5653
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.081760141121058e-07,
+      "loss": 1.5858,
+      "step": 5654
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.06951387949118e-07,
+      "loss": 1.1007,
+      "step": 5655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.057277410005619e-07,
+      "loss": 1.2179,
+      "step": 5656
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.045050735455411e-07,
+      "loss": 1.3119,
+      "step": 5657
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.032833858629389e-07,
+      "loss": 1.3796,
+      "step": 5658
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.020626782314122e-07,
+      "loss": 1.2371,
+      "step": 5659
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.008429509293979e-07,
+      "loss": 1.4804,
+      "step": 5660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.996242042351048e-07,
+      "loss": 1.4491,
+      "step": 5661
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.984064384265232e-07,
+      "loss": 1.3632,
+      "step": 5662
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.971896537814144e-07,
+      "loss": 1.5023,
+      "step": 5663
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.959738505773211e-07,
+      "loss": 1.2044,
+      "step": 5664
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.947590290915568e-07,
+      "loss": 1.4846,
+      "step": 5665
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.935451896012163e-07,
+      "loss": 1.6886,
+      "step": 5666
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.92332332383166e-07,
+      "loss": 1.103,
+      "step": 5667
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.911204577140518e-07,
+      "loss": 1.5271,
+      "step": 5668
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.899095658702914e-07,
+      "loss": 1.198,
+      "step": 5669
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.886996571280841e-07,
+      "loss": 1.7917,
+      "step": 5670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.874907317633988e-07,
+      "loss": 1.6892,
+      "step": 5671
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.862827900519841e-07,
+      "loss": 1.4086,
+      "step": 5672
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.85075832269364e-07,
+      "loss": 1.4293,
+      "step": 5673
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.838698586908355e-07,
+      "loss": 1.5141,
+      "step": 5674
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.82664869591474e-07,
+      "loss": 1.4511,
+      "step": 5675
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.814608652461274e-07,
+      "loss": 1.2923,
+      "step": 5676
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.802578459294235e-07,
+      "loss": 1.6309,
+      "step": 5677
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.790558119157597e-07,
+      "loss": 1.2602,
+      "step": 5678
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.778547634793136e-07,
+      "loss": 1.3521,
+      "step": 5679
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.766547008940344e-07,
+      "loss": 1.3787,
+      "step": 5680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.754556244336496e-07,
+      "loss": 1.7665,
+      "step": 5681
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.742575343716584e-07,
+      "loss": 1.3737,
+      "step": 5682
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.730604309813376e-07,
+      "loss": 1.6533,
+      "step": 5683
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.718643145357401e-07,
+      "loss": 1.3438,
+      "step": 5684
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.70669185307688e-07,
+      "loss": 1.7489,
+      "step": 5685
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.694750435697855e-07,
+      "loss": 1.5938,
+      "step": 5686
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.682818895944049e-07,
+      "loss": 1.5572,
+      "step": 5687
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.670897236536989e-07,
+      "loss": 1.0751,
+      "step": 5688
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.658985460195899e-07,
+      "loss": 1.6245,
+      "step": 5689
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.647083569637797e-07,
+      "loss": 1.3903,
+      "step": 5690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.635191567577393e-07,
+      "loss": 1.3994,
+      "step": 5691
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.623309456727195e-07,
+      "loss": 1.1577,
+      "step": 5692
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.611437239797413e-07,
+      "loss": 1.4041,
+      "step": 5693
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.599574919496022e-07,
+      "loss": 1.4415,
+      "step": 5694
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.587722498528743e-07,
+      "loss": 1.2855,
+      "step": 5695
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.575879979599015e-07,
+      "loss": 1.3394,
+      "step": 5696
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.564047365408049e-07,
+      "loss": 1.4851,
+      "step": 5697
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.552224658654765e-07,
+      "loss": 1.4417,
+      "step": 5698
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.540411862035862e-07,
+      "loss": 1.5818,
+      "step": 5699
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.528608978245732e-07,
+      "loss": 1.7251,
+      "step": 5700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.516816009976556e-07,
+      "loss": 1.1074,
+      "step": 5701
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.5050329599182e-07,
+      "loss": 1.1681,
+      "step": 5702
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.493259830758325e-07,
+      "loss": 1.2221,
+      "step": 5703
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.481496625182271e-07,
+      "loss": 1.1933,
+      "step": 5704
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.469743345873153e-07,
+      "loss": 1.3992,
+      "step": 5705
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.457999995511827e-07,
+      "loss": 1.4784,
+      "step": 5706
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.44626657677685e-07,
+      "loss": 1.6808,
+      "step": 5707
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.434543092344552e-07,
+      "loss": 1.5965,
+      "step": 5708
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.422829544888953e-07,
+      "loss": 1.5102,
+      "step": 5709
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.411125937081858e-07,
+      "loss": 1.2285,
+      "step": 5710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.399432271592748e-07,
+      "loss": 1.7514,
+      "step": 5711
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.387748551088902e-07,
+      "loss": 1.492,
+      "step": 5712
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.376074778235264e-07,
+      "loss": 1.533,
+      "step": 5713
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.364410955694561e-07,
+      "loss": 1.0457,
+      "step": 5714
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.352757086127215e-07,
+      "loss": 1.4534,
+      "step": 5715
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.341113172191399e-07,
+      "loss": 1.2279,
+      "step": 5716
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.329479216543016e-07,
+      "loss": 1.2339,
+      "step": 5717
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.317855221835672e-07,
+      "loss": 1.2629,
+      "step": 5718
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.306241190720741e-07,
+      "loss": 1.2758,
+      "step": 5719
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.294637125847281e-07,
+      "loss": 1.6761,
+      "step": 5720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.283043029862118e-07,
+      "loss": 1.2585,
+      "step": 5721
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.271458905409767e-07,
+      "loss": 1.3338,
+      "step": 5722
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.2598847551325e-07,
+      "loss": 1.3748,
+      "step": 5723
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.248320581670281e-07,
+      "loss": 1.0309,
+      "step": 5724
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.236766387660848e-07,
+      "loss": 1.1852,
+      "step": 5725
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.225222175739598e-07,
+      "loss": 1.8743,
+      "step": 5726
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.213687948539698e-07,
+      "loss": 1.503,
+      "step": 5727
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.202163708692043e-07,
+      "loss": 1.2592,
+      "step": 5728
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.190649458825204e-07,
+      "loss": 1.4158,
+      "step": 5729
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.179145201565529e-07,
+      "loss": 1.362,
+      "step": 5730
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.167650939537028e-07,
+      "loss": 1.5889,
+      "step": 5731
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.156166675361485e-07,
+      "loss": 1.2908,
+      "step": 5732
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.144692411658371e-07,
+      "loss": 1.6664,
+      "step": 5733
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.13322815104489e-07,
+      "loss": 1.4939,
+      "step": 5734
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.121773896135947e-07,
+      "loss": 1.5448,
+      "step": 5735
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.110329649544189e-07,
+      "loss": 1.272,
+      "step": 5736
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.09889541387998e-07,
+      "loss": 1.3774,
+      "step": 5737
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.087471191751365e-07,
+      "loss": 1.388,
+      "step": 5738
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.076056985764151e-07,
+      "loss": 1.5432,
+      "step": 5739
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.064652798521819e-07,
+      "loss": 1.111,
+      "step": 5740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.053258632625603e-07,
+      "loss": 1.3028,
+      "step": 5741
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.041874490674416e-07,
+      "loss": 1.34,
+      "step": 5742
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.03050037526492e-07,
+      "loss": 1.5044,
+      "step": 5743
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.019136288991451e-07,
+      "loss": 1.2641,
+      "step": 5744
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.007782234446096e-07,
+      "loss": 1.3381,
+      "step": 5745
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.996438214218613e-07,
+      "loss": 1.4432,
+      "step": 5746
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.985104230896515e-07,
+      "loss": 1.028,
+      "step": 5747
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.973780287065007e-07,
+      "loss": 1.1794,
+      "step": 5748
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.962466385306981e-07,
+      "loss": 1.1663,
+      "step": 5749
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.95116252820308e-07,
+      "loss": 1.4281,
+      "step": 5750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.939868718331615e-07,
+      "loss": 1.5028,
+      "step": 5751
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.92858495826864e-07,
+      "loss": 1.5744,
+      "step": 5752
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.917311250587893e-07,
+      "loss": 1.3695,
+      "step": 5753
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.906047597860842e-07,
+      "loss": 1.3826,
+      "step": 5754
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.894794002656628e-07,
+      "loss": 1.3376,
+      "step": 5755
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.883550467542137e-07,
+      "loss": 1.4769,
+      "step": 5756
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.872316995081922e-07,
+      "loss": 1.4517,
+      "step": 5757
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.861093587838268e-07,
+      "loss": 1.0184,
+      "step": 5758
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.84988024837117e-07,
+      "loss": 1.6424,
+      "step": 5759
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.838676979238284e-07,
+      "loss": 1.616,
+      "step": 5760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.827483782995025e-07,
+      "loss": 1.1023,
+      "step": 5761
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.816300662194463e-07,
+      "loss": 1.1011,
+      "step": 5762
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.805127619387407e-07,
+      "loss": 1.4666,
+      "step": 5763
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.793964657122331e-07,
+      "loss": 1.2132,
+      "step": 5764
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.782811777945446e-07,
+      "loss": 1.4738,
+      "step": 5765
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.77166898440063e-07,
+      "loss": 1.6203,
+      "step": 5766
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.760536279029494e-07,
+      "loss": 1.3207,
+      "step": 5767
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.749413664371312e-07,
+      "loss": 1.4919,
+      "step": 5768
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.738301142963082e-07,
+      "loss": 1.4402,
+      "step": 5769
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.727198717339511e-07,
+      "loss": 1.508,
+      "step": 5770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.71610639003295e-07,
+      "loss": 1.1696,
+      "step": 5771
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.705024163573519e-07,
+      "loss": 1.1777,
+      "step": 5772
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.693952040488959e-07,
+      "loss": 0.9323,
+      "step": 5773
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.682890023304777e-07,
+      "loss": 1.5662,
+      "step": 5774
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.671838114544115e-07,
+      "loss": 1.2491,
+      "step": 5775
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.660796316727862e-07,
+      "loss": 1.348,
+      "step": 5776
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.649764632374549e-07,
+      "loss": 1.4429,
+      "step": 5777
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.638743064000452e-07,
+      "loss": 1.4583,
+      "step": 5778
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.627731614119492e-07,
+      "loss": 1.3537,
+      "step": 5779
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.616730285243305e-07,
+      "loss": 1.7146,
+      "step": 5780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.60573907988124e-07,
+      "loss": 1.4129,
+      "step": 5781
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.59475800054029e-07,
+      "loss": 1.6678,
+      "step": 5782
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.583787049725181e-07,
+      "loss": 1.3187,
+      "step": 5783
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.572826229938288e-07,
+      "loss": 1.3346,
+      "step": 5784
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.561875543679728e-07,
+      "loss": 1.353,
+      "step": 5785
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.550934993447243e-07,
+      "loss": 1.2554,
+      "step": 5786
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.540004581736324e-07,
+      "loss": 1.5077,
+      "step": 5787
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.529084311040095e-07,
+      "loss": 1.3413,
+      "step": 5788
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.518174183849423e-07,
+      "loss": 1.5223,
+      "step": 5789
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.507274202652807e-07,
+      "loss": 1.47,
+      "step": 5790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.496384369936464e-07,
+      "loss": 1.2517,
+      "step": 5791
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.485504688184307e-07,
+      "loss": 1.2918,
+      "step": 5792
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.474635159877884e-07,
+      "loss": 1.3233,
+      "step": 5793
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.463775787496484e-07,
+      "loss": 1.2806,
+      "step": 5794
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.452926573517037e-07,
+      "loss": 1.3055,
+      "step": 5795
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.442087520414191e-07,
+      "loss": 1.3249,
+      "step": 5796
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.431258630660241e-07,
+      "loss": 1.3769,
+      "step": 5797
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.420439906725195e-07,
+      "loss": 1.6913,
+      "step": 5798
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.409631351076711e-07,
+      "loss": 1.4326,
+      "step": 5799
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.398832966180162e-07,
+      "loss": 1.5081,
+      "step": 5800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.388044754498573e-07,
+      "loss": 1.3851,
+      "step": 5801
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.377266718492663e-07,
+      "loss": 1.1552,
+      "step": 5802
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.366498860620839e-07,
+      "loss": 1.4465,
+      "step": 5803
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.355741183339158e-07,
+      "loss": 1.2258,
+      "step": 5804
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.34499368910138e-07,
+      "loss": 1.3321,
+      "step": 5805
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.334256380358927e-07,
+      "loss": 1.3579,
+      "step": 5806
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.323529259560911e-07,
+      "loss": 1.0919,
+      "step": 5807
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.312812329154104e-07,
+      "loss": 1.7495,
+      "step": 5808
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.302105591582979e-07,
+      "loss": 1.2979,
+      "step": 5809
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.291409049289647e-07,
+      "loss": 1.2781,
+      "step": 5810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.280722704713936e-07,
+      "loss": 1.4915,
+      "step": 5811
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.270046560293307e-07,
+      "loss": 1.8488,
+      "step": 5812
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.259380618462922e-07,
+      "loss": 1.309,
+      "step": 5813
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.24872488165562e-07,
+      "loss": 1.6191,
+      "step": 5814
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.238079352301878e-07,
+      "loss": 1.4174,
+      "step": 5815
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.227444032829887e-07,
+      "loss": 1.6493,
+      "step": 5816
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.216818925665479e-07,
+      "loss": 1.5412,
+      "step": 5817
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.20620403323217e-07,
+      "loss": 1.2636,
+      "step": 5818
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.195599357951136e-07,
+      "loss": 1.623,
+      "step": 5819
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.185004902241241e-07,
+      "loss": 1.2488,
+      "step": 5820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.174420668518992e-07,
+      "loss": 1.5244,
+      "step": 5821
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.163846659198601e-07,
+      "loss": 1.4979,
+      "step": 5822
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.153282876691896e-07,
+      "loss": 1.2346,
+      "step": 5823
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.142729323408419e-07,
+      "loss": 1.286,
+      "step": 5824
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.132186001755374e-07,
+      "loss": 1.3109,
+      "step": 5825
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.121652914137592e-07,
+      "loss": 1.4457,
+      "step": 5826
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.111130062957625e-07,
+      "loss": 1.7201,
+      "step": 5827
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.100617450615636e-07,
+      "loss": 1.465,
+      "step": 5828
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.090115079509494e-07,
+      "loss": 1.4433,
+      "step": 5829
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.079622952034713e-07,
+      "loss": 1.3927,
+      "step": 5830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.069141070584477e-07,
+      "loss": 1.3548,
+      "step": 5831
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.058669437549618e-07,
+      "loss": 1.6646,
+      "step": 5832
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.04820805531866e-07,
+      "loss": 1.6242,
+      "step": 5833
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.037756926277754e-07,
+      "loss": 1.1897,
+      "step": 5834
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.027316052810737e-07,
+      "loss": 1.5044,
+      "step": 5835
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.016885437299113e-07,
+      "loss": 1.811,
+      "step": 5836
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.006465082122002e-07,
+      "loss": 1.3943,
+      "step": 5837
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.996054989656252e-07,
+      "loss": 1.4854,
+      "step": 5838
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.985655162276299e-07,
+      "loss": 1.6199,
+      "step": 5839
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.975265602354295e-07,
+      "loss": 1.4093,
+      "step": 5840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.964886312260003e-07,
+      "loss": 1.2305,
+      "step": 5841
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.954517294360894e-07,
+      "loss": 1.5745,
+      "step": 5842
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.944158551022044e-07,
+      "loss": 1.3667,
+      "step": 5843
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.933810084606216e-07,
+      "loss": 1.2544,
+      "step": 5844
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.923471897473842e-07,
+      "loss": 1.3679,
+      "step": 5845
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.91314399198296e-07,
+      "loss": 1.5028,
+      "step": 5846
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.902826370489323e-07,
+      "loss": 1.2863,
+      "step": 5847
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.892519035346283e-07,
+      "loss": 1.0496,
+      "step": 5848
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.882221988904895e-07,
+      "loss": 1.5656,
+      "step": 5849
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.871935233513813e-07,
+      "loss": 1.7552,
+      "step": 5850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.861658771519412e-07,
+      "loss": 1.6262,
+      "step": 5851
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.851392605265648e-07,
+      "loss": 1.5884,
+      "step": 5852
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.841136737094181e-07,
+      "loss": 1.6701,
+      "step": 5853
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.830891169344287e-07,
+      "loss": 1.7533,
+      "step": 5854
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.820655904352923e-07,
+      "loss": 1.6267,
+      "step": 5855
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.810430944454686e-07,
+      "loss": 1.2305,
+      "step": 5856
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.800216291981802e-07,
+      "loss": 1.6336,
+      "step": 5857
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.790011949264173e-07,
+      "loss": 1.6541,
+      "step": 5858
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.779817918629326e-07,
+      "loss": 1.5946,
+      "step": 5859
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.769634202402467e-07,
+      "loss": 1.2451,
+      "step": 5860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7594608029064124e-07,
+      "loss": 1.5328,
+      "step": 5861
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.74929772246166e-07,
+      "loss": 1.2647,
+      "step": 5862
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7391449633863217e-07,
+      "loss": 1.1598,
+      "step": 5863
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.729002527996185e-07,
+      "loss": 1.4391,
+      "step": 5864
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.718870418604654e-07,
+      "loss": 1.1406,
+      "step": 5865
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.708748637522803e-07,
+      "loss": 1.3194,
+      "step": 5866
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.698637187059341e-07,
+      "loss": 1.4815,
+      "step": 5867
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.688536069520605e-07,
+      "loss": 1.2134,
+      "step": 5868
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6784452872106147e-07,
+      "loss": 1.5593,
+      "step": 5869
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.668364842430978e-07,
+      "loss": 1.3536,
+      "step": 5870
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6582947374810006e-07,
+      "loss": 1.2931,
+      "step": 5871
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6482349746575783e-07,
+      "loss": 1.4396,
+      "step": 5872
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6381855562552926e-07,
+      "loss": 1.0954,
+      "step": 5873
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6281464845663316e-07,
+      "loss": 1.2415,
+      "step": 5874
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.618117761880553e-07,
+      "loss": 1.4156,
+      "step": 5875
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.608099390485421e-07,
+      "loss": 1.3573,
+      "step": 5876
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5980913726660746e-07,
+      "loss": 1.23,
+      "step": 5877
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5880937107052493e-07,
+      "loss": 1.175,
+      "step": 5878
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.57810640688337e-07,
+      "loss": 1.4842,
+      "step": 5879
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5681294634784437e-07,
+      "loss": 1.3017,
+      "step": 5880
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.558162882766154e-07,
+      "loss": 1.687,
+      "step": 5881
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5482066670198223e-07,
+      "loss": 1.4516,
+      "step": 5882
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.538260818510365e-07,
+      "loss": 1.404,
+      "step": 5883
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.52832533950639e-07,
+      "loss": 1.4646,
+      "step": 5884
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5184002322740784e-07,
+      "loss": 1.329,
+      "step": 5885
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5084854990773087e-07,
+      "loss": 1.5813,
+      "step": 5886
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.498581142177538e-07,
+      "loss": 1.5159,
+      "step": 5887
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.488687163833899e-07,
+      "loss": 1.4985,
+      "step": 5888
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.478803566303119e-07,
+      "loss": 1.5381,
+      "step": 5889
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.468930351839607e-07,
+      "loss": 1.6227,
+      "step": 5890
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4590675226953427e-07,
+      "loss": 1.3222,
+      "step": 5891
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.44921508111999e-07,
+      "loss": 1.3863,
+      "step": 5892
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4393730293608104e-07,
+      "loss": 1.3569,
+      "step": 5893
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.429541369662721e-07,
+      "loss": 1.5763,
+      "step": 5894
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.419720104268238e-07,
+      "loss": 1.3625,
+      "step": 5895
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4099092354175364e-07,
+      "loss": 1.3879,
+      "step": 5896
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.400108765348393e-07,
+      "loss": 1.2939,
+      "step": 5897
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.390318696296247e-07,
+      "loss": 1.3231,
+      "step": 5898
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.380539030494124e-07,
+      "loss": 1.0015,
+      "step": 5899
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3707697701727056e-07,
+      "loss": 1.2267,
+      "step": 5900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3610109175603046e-07,
+      "loss": 1.7621,
+      "step": 5901
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3512624748828225e-07,
+      "loss": 1.4339,
+      "step": 5902
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3415244443638373e-07,
+      "loss": 1.4566,
+      "step": 5903
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.331796828224505e-07,
+      "loss": 1.3515,
+      "step": 5904
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.32207962868364e-07,
+      "loss": 1.1502,
+      "step": 5905
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.312372847957652e-07,
+      "loss": 1.4825,
+      "step": 5906
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.302676488260615e-07,
+      "loss": 1.4775,
+      "step": 5907
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2929905518041714e-07,
+      "loss": 1.6092,
+      "step": 5908
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.283315040797642e-07,
+      "loss": 1.4389,
+      "step": 5909
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.273649957447923e-07,
+      "loss": 1.359,
+      "step": 5910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2639953039595725e-07,
+      "loss": 1.4914,
+      "step": 5911
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.25435108253473e-07,
+      "loss": 1.1636,
+      "step": 5912
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2447172953731907e-07,
+      "loss": 1.3499,
+      "step": 5913
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2350939446723363e-07,
+      "loss": 1.6484,
+      "step": 5914
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.22548103262721e-07,
+      "loss": 1.2748,
+      "step": 5915
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2158785614304254e-07,
+      "loss": 1.4848,
+      "step": 5916
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2062865332722525e-07,
+      "loss": 1.7293,
+      "step": 5917
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1967049503405686e-07,
+      "loss": 1.2979,
+      "step": 5918
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.187133814820854e-07,
+      "loss": 1.6891,
+      "step": 5919
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.17757312889624e-07,
+      "loss": 1.3714,
+      "step": 5920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.16802289474742e-07,
+      "loss": 1.4589,
+      "step": 5921
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1584831145527695e-07,
+      "loss": 1.8314,
+      "step": 5922
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1489537904882193e-07,
+      "loss": 1.7471,
+      "step": 5923
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.139434924727359e-07,
+      "loss": 1.2341,
+      "step": 5924
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.129926519441357e-07,
+      "loss": 1.4202,
+      "step": 5925
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1204285767990405e-07,
+      "loss": 1.2288,
+      "step": 5926
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1109410989667976e-07,
+      "loss": 1.3206,
+      "step": 5927
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1014640881086755e-07,
+      "loss": 1.3574,
+      "step": 5928
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0919975463863007e-07,
+      "loss": 1.5292,
+      "step": 5929
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.082541475958934e-07,
+      "loss": 1.3786,
+      "step": 5930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0730958789834297e-07,
+      "loss": 1.3182,
+      "step": 5931
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0636607576142683e-07,
+      "loss": 1.3602,
+      "step": 5932
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.054236114003546e-07,
+      "loss": 1.6776,
+      "step": 5933
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0448219503009423e-07,
+      "loss": 1.4565,
+      "step": 5934
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.035418268653768e-07,
+      "loss": 1.3448,
+      "step": 5935
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0260250712069347e-07,
+      "loss": 1.5343,
+      "step": 5936
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0166423601029735e-07,
+      "loss": 1.4785,
+      "step": 5937
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.007270137482e-07,
+      "loss": 1.7343,
+      "step": 5938
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9979084054817763e-07,
+      "loss": 1.3677,
+      "step": 5939
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.988557166237622e-07,
+      "loss": 1.2665,
+      "step": 5940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.979216421882498e-07,
+      "loss": 1.0773,
+      "step": 5941
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9698861745469774e-07,
+      "loss": 1.5091,
+      "step": 5942
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9605664263592026e-07,
+      "loss": 1.4756,
+      "step": 5943
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9512571794449673e-07,
+      "loss": 1.5735,
+      "step": 5944
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.941958435927623e-07,
+      "loss": 1.3432,
+      "step": 5945
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9326701979281623e-07,
+      "loss": 1.9776,
+      "step": 5946
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9233924675651567e-07,
+      "loss": 1.534,
+      "step": 5947
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9141252469548084e-07,
+      "loss": 1.7203,
+      "step": 5948
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.904868538210882e-07,
+      "loss": 1.3077,
+      "step": 5949
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8956223434447936e-07,
+      "loss": 1.477,
+      "step": 5950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.886386664765512e-07,
+      "loss": 1.5799,
+      "step": 5951
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.877161504279647e-07,
+      "loss": 1.6372,
+      "step": 5952
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.867946864091393e-07,
+      "loss": 1.1154,
+      "step": 5953
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.858742746302535e-07,
+      "loss": 1.2749,
+      "step": 5954
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.849549153012483e-07,
+      "loss": 1.6444,
+      "step": 5955
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.840366086318215e-07,
+      "loss": 1.3566,
+      "step": 5956
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.831193548314338e-07,
+      "loss": 1.273,
+      "step": 5957
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8220315410930296e-07,
+      "loss": 1.2183,
+      "step": 5958
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8128800667440955e-07,
+      "loss": 1.7328,
+      "step": 5959
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8037391273549106e-07,
+      "loss": 1.6522,
+      "step": 5960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7946087250104733e-07,
+      "loss": 1.5603,
+      "step": 5961
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.785488861793341e-07,
+      "loss": 1.3063,
+      "step": 5962
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.776379539783709e-07,
+      "loss": 1.4954,
+      "step": 5963
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.767280761059355e-07,
+      "loss": 1.451,
+      "step": 5964
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7581925276956313e-07,
+      "loss": 1.4876,
+      "step": 5965
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.749114841765517e-07,
+      "loss": 1.2652,
+      "step": 5966
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.74004770533955e-07,
+      "loss": 1.2181,
+      "step": 5967
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7309911204858997e-07,
+      "loss": 1.4041,
+      "step": 5968
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.72194508927029e-07,
+      "loss": 1.2206,
+      "step": 5969
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7129096137560827e-07,
+      "loss": 1.4542,
+      "step": 5970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7038846960041785e-07,
+      "loss": 1.4006,
+      "step": 5971
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6948703380731196e-07,
+      "loss": 1.2731,
+      "step": 5972
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6858665420190066e-07,
+      "loss": 1.5489,
+      "step": 5973
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6768733098955453e-07,
+      "loss": 1.6747,
+      "step": 5974
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6678906437540353e-07,
+      "loss": 1.2096,
+      "step": 5975
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.658918545643353e-07,
+      "loss": 1.2269,
+      "step": 5976
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.649957017609979e-07,
+      "loss": 1.1465,
+      "step": 5977
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.64100606169796e-07,
+      "loss": 1.4419,
+      "step": 5978
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.632065679948965e-07,
+      "loss": 1.3792,
+      "step": 5979
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.623135874402217e-07,
+      "loss": 1.4551,
+      "step": 5980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.614216647094554e-07,
+      "loss": 1.258,
+      "step": 5981
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6053080000603813e-07,
+      "loss": 1.4006,
+      "step": 5982
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.596409935331707e-07,
+      "loss": 1.2243,
+      "step": 5983
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5875224549381084e-07,
+      "loss": 1.618,
+      "step": 5984
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5786455609067627e-07,
+      "loss": 1.6063,
+      "step": 5985
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5697792552624347e-07,
+      "loss": 1.8506,
+      "step": 5986
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.56092354002745e-07,
+      "loss": 1.473,
+      "step": 5987
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.552078417221755e-07,
+      "loss": 1.4739,
+      "step": 5988
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.543243888862841e-07,
+      "loss": 1.4273,
+      "step": 5989
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.534419956965823e-07,
+      "loss": 1.6091,
+      "step": 5990
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5256066235433595e-07,
+      "loss": 1.3813,
+      "step": 5991
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5168038906057187e-07,
+      "loss": 1.416,
+      "step": 5992
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.50801176016074e-07,
+      "loss": 1.3824,
+      "step": 5993
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.499230234213857e-07,
+      "loss": 1.2084,
+      "step": 5994
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.490459314768058e-07,
+      "loss": 1.3658,
+      "step": 5995
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4816990038239353e-07,
+      "loss": 1.2922,
+      "step": 5996
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4729493033796633e-07,
+      "loss": 1.056,
+      "step": 5997
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.464210215430974e-07,
+      "loss": 1.2666,
+      "step": 5998
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.455481741971206e-07,
+      "loss": 1.3904,
+      "step": 5999
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.44676388499125e-07,
+      "loss": 1.6735,
+      "step": 6000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.438056646479604e-07,
+      "loss": 1.63,
+      "step": 6001
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.429360028422307e-07,
+      "loss": 1.1351,
+      "step": 6002
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.420674032803017e-07,
+      "loss": 1.4857,
+      "step": 6003
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.411998661602933e-07,
+      "loss": 1.5577,
+      "step": 6004
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.403333916800866e-07,
+      "loss": 1.1397,
+      "step": 6005
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3946798003731576e-07,
+      "loss": 1.2573,
+      "step": 6006
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3860363142937727e-07,
+      "loss": 1.4045,
+      "step": 6007
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.377403460534223e-07,
+      "loss": 1.77,
+      "step": 6008
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3687812410636013e-07,
+      "loss": 1.1863,
+      "step": 6009
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.360169657848583e-07,
+      "loss": 1.4392,
+      "step": 6010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.351568712853398e-07,
+      "loss": 1.4561,
+      "step": 6011
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3429784080398765e-07,
+      "loss": 1.567,
+      "step": 6012
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3343987453673843e-07,
+      "loss": 1.2706,
+      "step": 6013
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3258297267929116e-07,
+      "loss": 1.3411,
+      "step": 6014
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.317271354270968e-07,
+      "loss": 1.4843,
+      "step": 6015
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3087236297536685e-07,
+      "loss": 1.7301,
+      "step": 6016
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3001865551906876e-07,
+      "loss": 1.8122,
+      "step": 6017
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.291660132529273e-07,
+      "loss": 1.7042,
+      "step": 6018
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.283144363714247e-07,
+      "loss": 1.3737,
+      "step": 6019
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2746392506879887e-07,
+      "loss": 1.3922,
+      "step": 6020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2661447953904635e-07,
+      "loss": 1.5787,
+      "step": 6021
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2576609997591824e-07,
+      "loss": 1.2766,
+      "step": 6022
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2491878657292643e-07,
+      "loss": 1.6462,
+      "step": 6023
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.240725395233346e-07,
+      "loss": 1.3807,
+      "step": 6024
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2322735902016787e-07,
+      "loss": 1.4687,
+      "step": 6025
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2238324525620425e-07,
+      "loss": 1.5334,
+      "step": 6026
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.215401984239819e-07,
+      "loss": 1.8218,
+      "step": 6027
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2069821871579255e-07,
+      "loss": 1.3307,
+      "step": 6028
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.198573063236865e-07,
+      "loss": 1.0261,
+      "step": 6029
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1901746143947035e-07,
+      "loss": 1.0859,
+      "step": 6030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1817868425470577e-07,
+      "loss": 1.4214,
+      "step": 6031
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.173409749607137e-07,
+      "loss": 1.497,
+      "step": 6032
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.165043337485679e-07,
+      "loss": 1.4334,
+      "step": 6033
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1566876080910193e-07,
+      "loss": 1.8182,
+      "step": 6034
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1483425633290264e-07,
+      "loss": 1.7901,
+      "step": 6035
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.140008205103162e-07,
+      "loss": 1.1251,
+      "step": 6036
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.131684535314422e-07,
+      "loss": 1.3645,
+      "step": 6037
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.123371555861382e-07,
+      "loss": 1.2529,
+      "step": 6038
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1150692686401864e-07,
+      "loss": 1.583,
+      "step": 6039
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.106777675544509e-07,
+      "loss": 1.5736,
+      "step": 6040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.098496778465621e-07,
+      "loss": 1.5737,
+      "step": 6041
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.090226579292327e-07,
+      "loss": 1.562,
+      "step": 6042
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.081967079911008e-07,
+      "loss": 1.5642,
+      "step": 6043
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.073718282205584e-07,
+      "loss": 1.5137,
+      "step": 6044
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0654801880575736e-07,
+      "loss": 1.2965,
+      "step": 6045
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0572527993460054e-07,
+      "loss": 1.3769,
+      "step": 6046
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.049036117947507e-07,
+      "loss": 1.5618,
+      "step": 6047
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0408301457362234e-07,
+      "loss": 1.3726,
+      "step": 6048
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.032634884583902e-07,
+      "loss": 1.5118,
+      "step": 6049
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0244503363598187e-07,
+      "loss": 1.286,
+      "step": 6050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0162765029307984e-07,
+      "loss": 1.1824,
+      "step": 6051
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.00811338616126e-07,
+      "loss": 1.2684,
+      "step": 6052
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.999960987913131e-07,
+      "loss": 1.5269,
+      "step": 6053
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.991819310045929e-07,
+      "loss": 1.344,
+      "step": 6054
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.983688354416708e-07,
+      "loss": 1.6904,
+      "step": 6055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9755681228800904e-07,
+      "loss": 1.6299,
+      "step": 6056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9674586172882323e-07,
+      "loss": 1.3709,
+      "step": 6057
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9593598394908663e-07,
+      "loss": 1.5336,
+      "step": 6058
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.951271791335253e-07,
+      "loss": 1.7873,
+      "step": 6059
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.943194474666233e-07,
+      "loss": 1.3902,
+      "step": 6060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9351278913261893e-07,
+      "loss": 0.9836,
+      "step": 6061
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9270720431550384e-07,
+      "loss": 1.4568,
+      "step": 6062
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9190269319902764e-07,
+      "loss": 1.3697,
+      "step": 6063
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9109925596669255e-07,
+      "loss": 1.4299,
+      "step": 6064
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.90296892801758e-07,
+      "loss": 1.4548,
+      "step": 6065
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.894956038872365e-07,
+      "loss": 1.1175,
+      "step": 6066
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.88695389405898e-07,
+      "loss": 1.3254,
+      "step": 6067
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8789624954026374e-07,
+      "loss": 1.3174,
+      "step": 6068
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.870981844726134e-07,
+      "loss": 1.074,
+      "step": 6069
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.863011943849792e-07,
+      "loss": 1.7107,
+      "step": 6070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.855052794591495e-07,
+      "loss": 1.3375,
+      "step": 6071
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.847104398766676e-07,
+      "loss": 1.8766,
+      "step": 6072
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8391667581882995e-07,
+      "loss": 1.3305,
+      "step": 6073
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8312398746668913e-07,
+      "loss": 1.3375,
+      "step": 6074
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.823323750010504e-07,
+      "loss": 1.7315,
+      "step": 6075
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.815418386024771e-07,
+      "loss": 1.1248,
+      "step": 6076
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.807523784512839e-07,
+      "loss": 1.402,
+      "step": 6077
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.799639947275412e-07,
+      "loss": 1.4416,
+      "step": 6078
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7917668761107407e-07,
+      "loss": 1.4373,
+      "step": 6079
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.783904572814622e-07,
+      "loss": 1.4655,
+      "step": 6080
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.776053039180376e-07,
+      "loss": 1.285,
+      "step": 6081
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7682122769988995e-07,
+      "loss": 1.5565,
+      "step": 6082
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.760382288058616e-07,
+      "loss": 1.121,
+      "step": 6083
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7525630741454755e-07,
+      "loss": 1.2194,
+      "step": 6084
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7447546370430066e-07,
+      "loss": 1.4991,
+      "step": 6085
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7369569785322405e-07,
+      "loss": 1.7369,
+      "step": 6086
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.729170100391781e-07,
+      "loss": 1.5051,
+      "step": 6087
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7213940043977537e-07,
+      "loss": 1.5702,
+      "step": 6088
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7136286923238385e-07,
+      "loss": 1.5685,
+      "step": 6089
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7058741659412414e-07,
+      "loss": 1.3245,
+      "step": 6090
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.698130427018725e-07,
+      "loss": 1.3629,
+      "step": 6091
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6903974773225703e-07,
+      "loss": 1.2297,
+      "step": 6092
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.682675318616618e-07,
+      "loss": 1.8703,
+      "step": 6093
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.674963952662241e-07,
+      "loss": 1.422,
+      "step": 6094
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.667263381218338e-07,
+      "loss": 1.6925,
+      "step": 6095
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6595736060413655e-07,
+      "loss": 1.4813,
+      "step": 6096
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.651894628885293e-07,
+      "loss": 1.4067,
+      "step": 6097
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6442264515016634e-07,
+      "loss": 1.3116,
+      "step": 6098
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.636569075639511e-07,
+      "loss": 1.5509,
+      "step": 6099
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6289225030454556e-07,
+      "loss": 1.2074,
+      "step": 6100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6212867354636016e-07,
+      "loss": 1.3586,
+      "step": 6101
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.613661774635634e-07,
+      "loss": 1.1145,
+      "step": 6102
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6060476223007337e-07,
+      "loss": 1.6515,
+      "step": 6103
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.59844428019565e-07,
+      "loss": 1.3866,
+      "step": 6104
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.590851750054657e-07,
+      "loss": 1.3812,
+      "step": 6105
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.583270033609536e-07,
+      "loss": 1.6102,
+      "step": 6106
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.575699132589654e-07,
+      "loss": 1.2692,
+      "step": 6107
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.568139048721852e-07,
+      "loss": 1.283,
+      "step": 6108
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5605897837305494e-07,
+      "loss": 1.4526,
+      "step": 6109
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.55305133933767e-07,
+      "loss": 1.2782,
+      "step": 6110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.545523717262699e-07,
+      "loss": 1.2144,
+      "step": 6111
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.538006919222608e-07,
+      "loss": 1.3298,
+      "step": 6112
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5305009469319464e-07,
+      "loss": 1.2024,
+      "step": 6113
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5230058021027624e-07,
+      "loss": 1.2483,
+      "step": 6114
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5155214864446556e-07,
+      "loss": 1.3958,
+      "step": 6115
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5080480016647426e-07,
+      "loss": 1.5124,
+      "step": 6116
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5005853494676714e-07,
+      "loss": 1.2885,
+      "step": 6117
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.493133531555625e-07,
+      "loss": 1.5869,
+      "step": 6118
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4856925496283045e-07,
+      "loss": 1.5179,
+      "step": 6119
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.478262405382953e-07,
+      "loss": 1.2343,
+      "step": 6120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.470843100514325e-07,
+      "loss": 1.2988,
+      "step": 6121
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4634346367147233e-07,
+      "loss": 1.3799,
+      "step": 6122
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4560370156739567e-07,
+      "loss": 1.6106,
+      "step": 6123
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4486502390793863e-07,
+      "loss": 1.3801,
+      "step": 6124
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.44127430861586e-07,
+      "loss": 1.4917,
+      "step": 6125
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.433909225965797e-07,
+      "loss": 1.4082,
+      "step": 6126
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.426554992809116e-07,
+      "loss": 1.4642,
+      "step": 6127
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.419211610823263e-07,
+      "loss": 1.2463,
+      "step": 6128
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.411879081683216e-07,
+      "loss": 1.4966,
+      "step": 6129
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.404557407061464e-07,
+      "loss": 1.5542,
+      "step": 6130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3972465886280495e-07,
+      "loss": 1.5045,
+      "step": 6131
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3899466280504936e-07,
+      "loss": 1.191,
+      "step": 6132
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3826575269938923e-07,
+      "loss": 1.9231,
+      "step": 6133
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.375379287120816e-07,
+      "loss": 1.159,
+      "step": 6134
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3681119100913974e-07,
+      "loss": 1.4959,
+      "step": 6135
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3608553975632608e-07,
+      "loss": 1.503,
+      "step": 6136
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3536097511915824e-07,
+      "loss": 1.0983,
+      "step": 6137
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3463749726290287e-07,
+      "loss": 1.1429,
+      "step": 6138
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3391510635258075e-07,
+      "loss": 1.3375,
+      "step": 6139
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3319380255296398e-07,
+      "loss": 1.1994,
+      "step": 6140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3247358602857818e-07,
+      "loss": 1.3606,
+      "step": 6141
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3175445694369747e-07,
+      "loss": 1.2259,
+      "step": 6142
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3103641546235233e-07,
+      "loss": 1.6079,
+      "step": 6143
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.303194617483212e-07,
+      "loss": 1.4863,
+      "step": 6144
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2960359596513714e-07,
+      "loss": 1.5983,
+      "step": 6145
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.288888182760851e-07,
+      "loss": 1.8299,
+      "step": 6146
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2817512884419912e-07,
+      "loss": 1.244,
+      "step": 6147
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2746252783226785e-07,
+      "loss": 1.4186,
+      "step": 6148
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2675101540282962e-07,
+      "loss": 1.2677,
+      "step": 6149
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2604059171817738e-07,
+      "loss": 1.3904,
+      "step": 6150
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2533125694035152e-07,
+      "loss": 1.2073,
+      "step": 6151
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2462301123114817e-07,
+      "loss": 1.5362,
+      "step": 6152
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2391585475211142e-07,
+      "loss": 1.5675,
+      "step": 6153
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2320978766454115e-07,
+      "loss": 1.3363,
+      "step": 6154
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2250481012948356e-07,
+      "loss": 1.5537,
+      "step": 6155
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2180092230774165e-07,
+      "loss": 1.2987,
+      "step": 6156
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2109812435986532e-07,
+      "loss": 1.2907,
+      "step": 6157
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.203964164461597e-07,
+      "loss": 1.2291,
+      "step": 6158
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1969579872667734e-07,
+      "loss": 1.5367,
+      "step": 6159
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1899627136122648e-07,
+      "loss": 1.2215,
+      "step": 6160
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.182978345093628e-07,
+      "loss": 1.2311,
+      "step": 6161
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1760048833039503e-07,
+      "loss": 1.5051,
+      "step": 6162
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1690423298338425e-07,
+      "loss": 1.6949,
+      "step": 6163
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1620906862714008e-07,
+      "loss": 0.9662,
+      "step": 6164
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1551499542022513e-07,
+      "loss": 1.3775,
+      "step": 6165
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1482201352095277e-07,
+      "loss": 1.1692,
+      "step": 6166
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1413012308738767e-07,
+      "loss": 1.4662,
+      "step": 6167
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1343932427734415e-07,
+      "loss": 1.7077,
+      "step": 6168
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1274961724838948e-07,
+      "loss": 1.1041,
+      "step": 6169
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1206100215784064e-07,
+      "loss": 1.5483,
+      "step": 6170
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.113734791627664e-07,
+      "loss": 1.5547,
+      "step": 6171
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1068704841998522e-07,
+      "loss": 1.5469,
+      "step": 6172
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1000171008606795e-07,
+      "loss": 1.1418,
+      "step": 6173
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0931746431733458e-07,
+      "loss": 1.3765,
+      "step": 6174
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.08634311269858e-07,
+      "loss": 1.2909,
+      "step": 6175
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0795225109945916e-07,
+      "loss": 1.301,
+      "step": 6176
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0727128396171247e-07,
+      "loss": 1.5471,
+      "step": 6177
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0659141001194095e-07,
+      "loss": 1.1757,
+      "step": 6178
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.059126294052194e-07,
+      "loss": 1.2986,
+      "step": 6179
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0523494229637397e-07,
+      "loss": 1.5968,
+      "step": 6180
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0455834883997826e-07,
+      "loss": 1.5764,
+      "step": 6181
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.038828491903605e-07,
+      "loss": 1.3566,
+      "step": 6182
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.032084435015963e-07,
+      "loss": 1.476,
+      "step": 6183
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0253513192751374e-07,
+      "loss": 1.371,
+      "step": 6184
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0186291462168882e-07,
+      "loss": 1.5053,
+      "step": 6185
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.011917917374523e-07,
+      "loss": 1.2781,
+      "step": 6186
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0052176342787942e-07,
+      "loss": 1.582,
+      "step": 6187
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.998528298458019e-07,
+      "loss": 1.5339,
+      "step": 6188
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.991849911437971e-07,
+      "loss": 1.5457,
+      "step": 6189
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9851824747419424e-07,
+      "loss": 1.6649,
+      "step": 6190
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9785259898907449e-07,
+      "loss": 1.2419,
+      "step": 6191
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9718804584026585e-07,
+      "loss": 1.5211,
+      "step": 6192
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9652458817934984e-07,
+      "loss": 1.2168,
+      "step": 6193
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.958622261576548e-07,
+      "loss": 1.3431,
+      "step": 6194
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.952009599262622e-07,
+      "loss": 1.0645,
+      "step": 6195
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9454078963600186e-07,
+      "loss": 1.4374,
+      "step": 6196
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9388171543745394e-07,
+      "loss": 1.3065,
+      "step": 6197
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.932237374809487e-07,
+      "loss": 1.262,
+      "step": 6198
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9256685591656666e-07,
+      "loss": 1.1609,
+      "step": 6199
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9191107089413687e-07,
+      "loss": 1.508,
+      "step": 6200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.912563825632402e-07,
+      "loss": 1.1915,
+      "step": 6201
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9060279107320666e-07,
+      "loss": 1.3016,
+      "step": 6202
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8995029657311537e-07,
+      "loss": 1.4853,
+      "step": 6203
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8929889921179668e-07,
+      "loss": 1.3463,
+      "step": 6204
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8864859913782784e-07,
+      "loss": 1.3408,
+      "step": 6205
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8799939649954024e-07,
+      "loss": 1.5324,
+      "step": 6206
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.873512914450104e-07,
+      "loss": 1.3795,
+      "step": 6207
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8670428412206788e-07,
+      "loss": 1.3242,
+      "step": 6208
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8605837467828958e-07,
+      "loss": 1.2946,
+      "step": 6209
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8541356326100436e-07,
+      "loss": 1.3613,
+      "step": 6210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.847698500172873e-07,
+      "loss": 1.5235,
+      "step": 6211
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8412723509396546e-07,
+      "loss": 1.589,
+      "step": 6212
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8348571863761656e-07,
+      "loss": 1.2747,
+      "step": 6213
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8284530079456354e-07,
+      "loss": 1.5332,
+      "step": 6214
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8220598171088343e-07,
+      "loss": 1.6092,
+      "step": 6215
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8156776153239852e-07,
+      "loss": 1.3425,
+      "step": 6216
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.80930640404684e-07,
+      "loss": 1.4637,
+      "step": 6217
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.802946184730614e-07,
+      "loss": 1.3561,
+      "step": 6218
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7965969588260358e-07,
+      "loss": 1.3636,
+      "step": 6219
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.790258727781319e-07,
+      "loss": 1.6514,
+      "step": 6220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7839314930421735e-07,
+      "loss": 1.3054,
+      "step": 6221
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7776152560517902e-07,
+      "loss": 1.6115,
+      "step": 6222
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7713100182508604e-07,
+      "loss": 1.5116,
+      "step": 6223
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7650157810775726e-07,
+      "loss": 1.6303,
+      "step": 6224
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7587325459675843e-07,
+      "loss": 1.561,
+      "step": 6225
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7524603143540708e-07,
+      "loss": 1.5188,
+      "step": 6226
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.746199087667677e-07,
+      "loss": 1.2507,
+      "step": 6227
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7399488673365494e-07,
+      "loss": 1.6427,
+      "step": 6228
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.733709654786314e-07,
+      "loss": 1.5154,
+      "step": 6229
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7274814514400995e-07,
+      "loss": 1.3616,
+      "step": 6230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7212642587185024e-07,
+      "loss": 1.4923,
+      "step": 6231
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7150580780396385e-07,
+      "loss": 1.3558,
+      "step": 6232
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7088629108190812e-07,
+      "loss": 1.434,
+      "step": 6233
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7026787584699057e-07,
+      "loss": 1.6699,
+      "step": 6234
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6965056224026888e-07,
+      "loss": 1.0951,
+      "step": 6235
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6903435040254545e-07,
+      "loss": 1.1809,
+      "step": 6236
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6841924047437674e-07,
+      "loss": 1.3565,
+      "step": 6237
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6780523259606217e-07,
+      "loss": 1.3569,
+      "step": 6238
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.671923269076553e-07,
+      "loss": 1.3542,
+      "step": 6239
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.665805235489537e-07,
+      "loss": 1.1259,
+      "step": 6240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6596982265950635e-07,
+      "loss": 1.2192,
+      "step": 6241
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6536022437860955e-07,
+      "loss": 1.2476,
+      "step": 6242
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6475172884530821e-07,
+      "loss": 1.6277,
+      "step": 6243
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6414433619839687e-07,
+      "loss": 1.5162,
+      "step": 6244
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6353804657641638e-07,
+      "loss": 1.2067,
+      "step": 6245
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.629328601176583e-07,
+      "loss": 1.2741,
+      "step": 6246
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6232877696016058e-07,
+      "loss": 1.6151,
+      "step": 6247
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6172579724171078e-07,
+      "loss": 1.3989,
+      "step": 6248
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6112392109984386e-07,
+      "loss": 1.4774,
+      "step": 6249
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6052314867184448e-07,
+      "loss": 1.3551,
+      "step": 6250
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5992348009474356e-07,
+      "loss": 1.6232,
+      "step": 6251
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5932491550532224e-07,
+      "loss": 1.3467,
+      "step": 6252
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5872745504010801e-07,
+      "loss": 1.499,
+      "step": 6253
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5813109883537792e-07,
+      "loss": 1.5963,
+      "step": 6254
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5753584702715764e-07,
+      "loss": 1.2299,
+      "step": 6255
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5694169975121798e-07,
+      "loss": 1.6297,
+      "step": 6256
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5634865714308167e-07,
+      "loss": 1.5578,
+      "step": 6257
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5575671933801606e-07,
+      "loss": 1.5397,
+      "step": 6258
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5516588647103924e-07,
+      "loss": 1.4894,
+      "step": 6259
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.545761586769151e-07,
+      "loss": 1.3113,
+      "step": 6260
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.539875360901577e-07,
+      "loss": 1.764,
+      "step": 6261
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5340001884502577e-07,
+      "loss": 1.3496,
+      "step": 6262
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5281360707552983e-07,
+      "loss": 1.9321,
+      "step": 6263
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5222830091542517e-07,
+      "loss": 1.4066,
+      "step": 6264
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.516441004982161e-07,
+      "loss": 1.593,
+      "step": 6265
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5106100595715602e-07,
+      "loss": 1.6583,
+      "step": 6266
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.504790174252424e-07,
+      "loss": 1.4114,
+      "step": 6267
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.498981350352252e-07,
+      "loss": 1.4425,
+      "step": 6268
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4931835891959735e-07,
+      "loss": 1.3491,
+      "step": 6269
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4873968921060355e-07,
+      "loss": 1.2758,
+      "step": 6270
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4816212604023328e-07,
+      "loss": 1.6735,
+      "step": 6271
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4758566954022557e-07,
+      "loss": 1.1094,
+      "step": 6272
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.470103198420647e-07,
+      "loss": 0.7863,
+      "step": 6273
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.464360770769857e-07,
+      "loss": 1.4565,
+      "step": 6274
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4586294137596768e-07,
+      "loss": 1.4221,
+      "step": 6275
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4529091286973994e-07,
+      "loss": 1.5858,
+      "step": 6276
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4471999168877814e-07,
+      "loss": 1.5864,
+      "step": 6277
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4415017796330477e-07,
+      "loss": 1.6311,
+      "step": 6278
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4358147182329086e-07,
+      "loss": 1.3748,
+      "step": 6279
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.430138733984543e-07,
+      "loss": 1.4747,
+      "step": 6280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4244738281826042e-07,
+      "loss": 1.4325,
+      "step": 6281
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.418820002119209e-07,
+      "loss": 1.1291,
+      "step": 6282
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4131772570839698e-07,
+      "loss": 1.3846,
+      "step": 6283
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.407545594363946e-07,
+      "loss": 1.46,
+      "step": 6284
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.401925015243688e-07,
+      "loss": 1.6218,
+      "step": 6285
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3963155210052037e-07,
+      "loss": 1.457,
+      "step": 6286
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3907171129279806e-07,
+      "loss": 1.3295,
+      "step": 6287
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.385129792288986e-07,
+      "loss": 1.559,
+      "step": 6288
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3795535603626398e-07,
+      "loss": 1.2764,
+      "step": 6289
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3739884184208464e-07,
+      "loss": 1.4537,
+      "step": 6290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3684343677329682e-07,
+      "loss": 1.6967,
+      "step": 6291
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.362891409565853e-07,
+      "loss": 1.3426,
+      "step": 6292
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3573595451838063e-07,
+      "loss": 1.1051,
+      "step": 6293
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.351838775848613e-07,
+      "loss": 1.4754,
+      "step": 6294
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3463291028195157e-07,
+      "loss": 1.3597,
+      "step": 6295
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3408305273532373e-07,
+      "loss": 1.4537,
+      "step": 6296
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3353430507039577e-07,
+      "loss": 1.4239,
+      "step": 6297
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3298666741233424e-07,
+      "loss": 1.3976,
+      "step": 6298
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3244013988605087e-07,
+      "loss": 1.2801,
+      "step": 6299
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3189472261620485e-07,
+      "loss": 1.1161,
+      "step": 6300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.313504157272022e-07,
+      "loss": 0.9667,
+      "step": 6301
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3080721934319474e-07,
+      "loss": 1.4575,
+      "step": 6302
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.302651335880828e-07,
+      "loss": 1.4845,
+      "step": 6303
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2972415858551192e-07,
+      "loss": 1.6434,
+      "step": 6304
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2918429445887503e-07,
+      "loss": 1.4839,
+      "step": 6305
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2864554133131035e-07,
+      "loss": 1.2326,
+      "step": 6306
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.281078993257051e-07,
+      "loss": 1.4951,
+      "step": 6307
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2757136856469064e-07,
+      "loss": 1.3667,
+      "step": 6308
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2703594917064633e-07,
+      "loss": 1.6984,
+      "step": 6309
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2650164126569774e-07,
+      "loss": 1.4855,
+      "step": 6310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2596844497171633e-07,
+      "loss": 1.5723,
+      "step": 6311
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.254363604103209e-07,
+      "loss": 1.1164,
+      "step": 6312
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2490538770287609e-07,
+      "loss": 1.3979,
+      "step": 6313
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2437552697049327e-07,
+      "loss": 1.6286,
+      "step": 6314
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2384677833402914e-07,
+      "loss": 1.5967,
+      "step": 6315
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2331914191408834e-07,
+      "loss": 1.0758,
+      "step": 6316
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.227926178310207e-07,
+      "loss": 1.5627,
+      "step": 6317
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.222672062049235e-07,
+      "loss": 1.1098,
+      "step": 6318
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2174290715563808e-07,
+      "loss": 1.4127,
+      "step": 6319
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2121972080275378e-07,
+      "loss": 1.3836,
+      "step": 6320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2069764726560685e-07,
+      "loss": 1.4869,
+      "step": 6321
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2017668666327752e-07,
+      "loss": 1.2314,
+      "step": 6322
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1965683911459413e-07,
+      "loss": 1.4589,
+      "step": 6323
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1913810473812959e-07,
+      "loss": 1.4105,
+      "step": 6324
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1862048365220369e-07,
+      "loss": 1.2427,
+      "step": 6325
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.18103975974882e-07,
+      "loss": 1.4389,
+      "step": 6326
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1758858182397692e-07,
+      "loss": 1.4934,
+      "step": 6327
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1707430131704556e-07,
+      "loss": 1.4207,
+      "step": 6328
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1656113457139296e-07,
+      "loss": 1.4606,
+      "step": 6329
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1604908170406715e-07,
+      "loss": 1.5484,
+      "step": 6330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.155381428318647e-07,
+      "loss": 1.6358,
+      "step": 6331
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1502831807132742e-07,
+      "loss": 1.6167,
+      "step": 6332
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1451960753874226e-07,
+      "loss": 1.3505,
+      "step": 6333
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1401201135014305e-07,
+      "loss": 1.5217,
+      "step": 6334
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.135055296213089e-07,
+      "loss": 1.437,
+      "step": 6335
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1300016246776402e-07,
+      "loss": 1.6541,
+      "step": 6336
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.124959100047801e-07,
+      "loss": 1.1695,
+      "step": 6337
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1199277234737293e-07,
+      "loss": 1.0951,
+      "step": 6338
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1149074961030515e-07,
+      "loss": 1.5539,
+      "step": 6339
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1098984190808403e-07,
+      "loss": 1.3229,
+      "step": 6340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1049004935496377e-07,
+      "loss": 1.5688,
+      "step": 6341
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0999137206494315e-07,
+      "loss": 1.456,
+      "step": 6342
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0949381015176786e-07,
+      "loss": 1.2265,
+      "step": 6343
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0899736372892655e-07,
+      "loss": 1.489,
+      "step": 6344
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0850203290965699e-07,
+      "loss": 1.2266,
+      "step": 6345
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0800781780693992e-07,
+      "loss": 1.2088,
+      "step": 6346
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0751471853350181e-07,
+      "loss": 1.5116,
+      "step": 6347
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0702273520181605e-07,
+      "loss": 1.043,
+      "step": 6348
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0653186792410063e-07,
+      "loss": 1.439,
+      "step": 6349
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0604211681231769e-07,
+      "loss": 1.3015,
+      "step": 6350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0555348197817728e-07,
+      "loss": 1.4484,
+      "step": 6351
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0506596353313358e-07,
+      "loss": 1.6222,
+      "step": 6352
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0457956158838545e-07,
+      "loss": 1.2966,
+      "step": 6353
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0409427625487855e-07,
+      "loss": 1.8942,
+      "step": 6354
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0361010764330215e-07,
+      "loss": 1.6204,
+      "step": 6355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0312705586409289e-07,
+      "loss": 1.4783,
+      "step": 6356
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.026451210274304e-07,
+      "loss": 1.4115,
+      "step": 6357
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0216430324324122e-07,
+      "loss": 1.7279,
+      "step": 6358
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.016846026211965e-07,
+      "loss": 1.3849,
+      "step": 6359
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.012060192707126e-07,
+      "loss": 1.5224,
+      "step": 6360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0072855330095055e-07,
+      "loss": 1.8696,
+      "step": 6361
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0025220482081765e-07,
+      "loss": 1.2964,
+      "step": 6362
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.977697393896591e-08,
+      "loss": 1.3976,
+      "step": 6363
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.93028607637908e-08,
+      "loss": 1.2804,
+      "step": 6364
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.882986540343642e-08,
+      "loss": 1.2324,
+      "step": 6365
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.835798796578755e-08,
+      "loss": 1.5331,
+      "step": 6366
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.788722855847699e-08,
+      "loss": 1.5157,
+      "step": 6367
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.741758728888218e-08,
+      "loss": 1.7076,
+      "step": 6368
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.694906426412409e-08,
+      "loss": 1.4742,
+      "step": 6369
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.648165959107059e-08,
+      "loss": 1.1469,
+      "step": 6370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.601537337633248e-08,
+      "loss": 1.4087,
+      "step": 6371
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.555020572626639e-08,
+      "loss": 1.6413,
+      "step": 6372
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.508615674697463e-08,
+      "loss": 1.138,
+      "step": 6373
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.462322654430312e-08,
+      "loss": 1.5896,
+      "step": 6374
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.416141522384292e-08,
+      "loss": 1.0253,
+      "step": 6375
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.370072289093035e-08,
+      "loss": 1.1449,
+      "step": 6376
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.324114965064524e-08,
+      "loss": 1.3614,
+      "step": 6377
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.278269560781373e-08,
+      "loss": 1.316,
+      "step": 6378
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.232536086700605e-08,
+      "loss": 1.2124,
+      "step": 6379
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.186914553253712e-08,
+      "loss": 1.4513,
+      "step": 6380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.141404970846535e-08,
+      "loss": 1.4802,
+      "step": 6381
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.096007349859659e-08,
+      "loss": 1.4815,
+      "step": 6382
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.050721700647858e-08,
+      "loss": 1.4825,
+      "step": 6383
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.005548033540479e-08,
+      "loss": 1.3687,
+      "step": 6384
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.960486358841336e-08,
+      "loss": 1.3347,
+      "step": 6385
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.915536686828764e-08,
+      "loss": 1.5361,
+      "step": 6386
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.870699027755281e-08,
+      "loss": 1.6313,
+      "step": 6387
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.825973391848263e-08,
+      "loss": 1.0718,
+      "step": 6388
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.781359789309162e-08,
+      "loss": 1.415,
+      "step": 6389
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.736858230314005e-08,
+      "loss": 1.1999,
+      "step": 6390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.692468725013448e-08,
+      "loss": 1.2398,
+      "step": 6391
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.648191283532337e-08,
+      "loss": 1.43,
+      "step": 6392
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.604025915970038e-08,
+      "loss": 1.4207,
+      "step": 6393
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.559972632400327e-08,
+      "loss": 1.2608,
+      "step": 6394
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.516031442871553e-08,
+      "loss": 1.6886,
+      "step": 6395
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.472202357406312e-08,
+      "loss": 1.4979,
+      "step": 6396
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.428485386001717e-08,
+      "loss": 1.4291,
+      "step": 6397
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.384880538629347e-08,
+      "loss": 1.2058,
+      "step": 6398
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.341387825235192e-08,
+      "loss": 1.5129,
+      "step": 6399
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.298007255739482e-08,
+      "loss": 1.1461,
+      "step": 6400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.25473884003719e-08,
+      "loss": 1.3367,
+      "step": 6401
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.211582587997479e-08,
+      "loss": 1.1475,
+      "step": 6402
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.168538509463975e-08,
+      "loss": 1.8151,
+      "step": 6403
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.12560661425471e-08,
+      "loss": 1.2205,
+      "step": 6404
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.082786912162243e-08,
+      "loss": 1.8443,
+      "step": 6405
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.040079412953372e-08,
+      "loss": 1.3296,
+      "step": 6406
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.997484126369359e-08,
+      "loss": 1.5278,
+      "step": 6407
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.955001062125989e-08,
+      "loss": 1.3602,
+      "step": 6408
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.912630229913232e-08,
+      "loss": 1.5327,
+      "step": 6409
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.87037163939569e-08,
+      "loss": 1.6071,
+      "step": 6410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.828225300212156e-08,
+      "loss": 1.4804,
+      "step": 6411
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.786191221975992e-08,
+      "loss": 1.4085,
+      "step": 6412
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.744269414274863e-08,
+      "loss": 1.2085,
+      "step": 6413
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.702459886670788e-08,
+      "loss": 1.5633,
+      "step": 6414
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.660762648700304e-08,
+      "loss": 1.3219,
+      "step": 6415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.619177709874192e-08,
+      "loss": 1.0782,
+      "step": 6416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.577705079677756e-08,
+      "loss": 1.5316,
+      "step": 6417
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.536344767570536e-08,
+      "loss": 1.3332,
+      "step": 6418
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.495096782986544e-08,
+      "loss": 1.4857,
+      "step": 6419
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.453961135334253e-08,
+      "loss": 1.1585,
+      "step": 6420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.412937833996325e-08,
+      "loss": 1.6196,
+      "step": 6421
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.372026888329887e-08,
+      "loss": 1.398,
+      "step": 6422
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.331228307666471e-08,
+      "loss": 1.7148,
+      "step": 6423
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.290542101311915e-08,
+      "loss": 1.1527,
+      "step": 6424
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.249968278546571e-08,
+      "loss": 1.2232,
+      "step": 6425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.209506848624869e-08,
+      "loss": 1.4709,
+      "step": 6426
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.16915782077593e-08,
+      "loss": 1.2731,
+      "step": 6427
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.128921204202943e-08,
+      "loss": 1.4163,
+      "step": 6428
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.088797008083792e-08,
+      "loss": 1.2035,
+      "step": 6429
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.048785241570321e-08,
+      "loss": 1.5405,
+      "step": 6430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.008885913789066e-08,
+      "loss": 1.1361,
+      "step": 6431
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.969099033840743e-08,
+      "loss": 1.1921,
+      "step": 6432
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.929424610800483e-08,
+      "loss": 1.1915,
+      "step": 6433
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.889862653717716e-08,
+      "loss": 1.3064,
+      "step": 6434
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.850413171616277e-08,
+      "loss": 1.1549,
+      "step": 6435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.811076173494247e-08,
+      "loss": 1.4935,
+      "step": 6436
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.771851668324226e-08,
+      "loss": 1.9301,
+      "step": 6437
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.732739665053001e-08,
+      "loss": 1.0939,
+      "step": 6438
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.693740172601715e-08,
+      "loss": 1.3145,
+      "step": 6439
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.654853199865973e-08,
+      "loss": 1.4369,
+      "step": 6440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.616078755715461e-08,
+      "loss": 1.5211,
+      "step": 6441
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.577416848994545e-08,
+      "loss": 1.4487,
+      "step": 6442
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.538867488521616e-08,
+      "loss": 1.5043,
+      "step": 6443
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.500430683089532e-08,
+      "loss": 1.4763,
+      "step": 6444
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.462106441465499e-08,
+      "loss": 1.4816,
+      "step": 6445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.423894772390971e-08,
+      "loss": 1.0921,
+      "step": 6446
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.385795684581752e-08,
+      "loss": 1.0808,
+      "step": 6447
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.347809186728005e-08,
+      "loss": 1.4165,
+      "step": 6448
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.30993528749424e-08,
+      "loss": 1.3954,
+      "step": 6449
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.27217399551916e-08,
+      "loss": 1.2608,
+      "step": 6450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.234525319415819e-08,
+      "loss": 1.1825,
+      "step": 6451
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.19698926777168e-08,
+      "loss": 1.3755,
+      "step": 6452
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.159565849148508e-08,
+      "loss": 1.45,
+      "step": 6453
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.12225507208214e-08,
+      "loss": 1.2985,
+      "step": 6454
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.085056945083101e-08,
+      "loss": 1.2825,
+      "step": 6455
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.047971476635883e-08,
+      "loss": 1.4824,
+      "step": 6456
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.010998675199554e-08,
+      "loss": 1.5247,
+      "step": 6457
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.974138549207198e-08,
+      "loss": 1.5958,
+      "step": 6458
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.937391107066426e-08,
+      "loss": 1.2803,
+      "step": 6459
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.900756357159143e-08,
+      "loss": 1.1303,
+      "step": 6460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.864234307841388e-08,
+      "loss": 1.4483,
+      "step": 6461
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.827824967443607e-08,
+      "loss": 1.5093,
+      "step": 6462
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.791528344270547e-08,
+      "loss": 1.4436,
+      "step": 6463
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.7553444466011966e-08,
+      "loss": 1.314,
+      "step": 6464
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.7192732826887865e-08,
+      "loss": 1.6472,
+      "step": 6465
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.6833148607610136e-08,
+      "loss": 1.2391,
+      "step": 6466
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.64746918901965e-08,
+      "loss": 1.1123,
+      "step": 6467
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.6117362756408775e-08,
+      "loss": 1.3202,
+      "step": 6468
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.576116128775122e-08,
+      "loss": 1.5755,
+      "step": 6469
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.5406087565471054e-08,
+      "loss": 1.2569,
+      "step": 6470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.5052141670557945e-08,
+      "loss": 0.8808,
+      "step": 6471
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.469932368374453e-08,
+      "loss": 1.7174,
+      "step": 6472
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.434763368550644e-08,
+      "loss": 1.8424,
+      "step": 6473
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.399707175606117e-08,
+      "loss": 1.6544,
+      "step": 6474
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.364763797536976e-08,
+      "loss": 1.5373,
+      "step": 6475
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.3299332423135117e-08,
+      "loss": 1.5099,
+      "step": 6476
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.295215517880425e-08,
+      "loss": 1.4521,
+      "step": 6477
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.2606106321565486e-08,
+      "loss": 1.4122,
+      "step": 6478
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.2261185930349566e-08,
+      "loss": 1.4408,
+      "step": 6479
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1917394083830785e-08,
+      "loss": 1.4876,
+      "step": 6480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.157473086042586e-08,
+      "loss": 1.4446,
+      "step": 6481
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.123319633829393e-08,
+      "loss": 1.7051,
+      "step": 6482
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.089279059533658e-08,
+      "loss": 1.4646,
+      "step": 6483
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.055351370919781e-08,
+      "loss": 1.6984,
+      "step": 6484
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.021536575726404e-08,
+      "loss": 1.1472,
+      "step": 6485
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.987834681666526e-08,
+      "loss": 1.4071,
+      "step": 6486
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.954245696427218e-08,
+      "loss": 1.2784,
+      "step": 6487
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.920769627669963e-08,
+      "loss": 1.1204,
+      "step": 6488
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.887406483030432e-08,
+      "loss": 1.1834,
+      "step": 6489
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.854156270118482e-08,
+      "loss": 1.6286,
+      "step": 6490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.821018996518212e-08,
+      "loss": 1.8713,
+      "step": 6491
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.787994669788076e-08,
+      "loss": 1.798,
+      "step": 6492
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7550832974607166e-08,
+      "loss": 1.551,
+      "step": 6493
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.722284887042905e-08,
+      "loss": 1.4019,
+      "step": 6494
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.689599446015769e-08,
+      "loss": 1.5256,
+      "step": 6495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.657026981834623e-08,
+      "loss": 1.6689,
+      "step": 6496
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.624567501929078e-08,
+      "loss": 1.2175,
+      "step": 6497
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.592221013702769e-08,
+      "loss": 1.497,
+      "step": 6498
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5599875245339045e-08,
+      "loss": 1.2621,
+      "step": 6499
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.527867041774547e-08,
+      "loss": 1.3634,
+      "step": 6500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.495859572751227e-08,
+      "loss": 1.3291,
+      "step": 6501
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.463965124764602e-08,
+      "loss": 1.3614,
+      "step": 6502
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4321837050895764e-08,
+      "loss": 1.511,
+      "step": 6503
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.400515320975296e-08,
+      "loss": 1.0439,
+      "step": 6504
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.368959979645093e-08,
+      "loss": 1.6567,
+      "step": 6505
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.337517688296544e-08,
+      "loss": 1.5143,
+      "step": 6506
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.306188454101301e-08,
+      "loss": 1.8145,
+      "step": 6507
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2749722842055364e-08,
+      "loss": 1.4202,
+      "step": 6508
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2438691857292215e-08,
+      "loss": 1.5853,
+      "step": 6509
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.212879165766959e-08,
+      "loss": 1.11,
+      "step": 6510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.18200223138715e-08,
+      "loss": 1.4329,
+      "step": 6511
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.151238389632828e-08,
+      "loss": 1.3311,
+      "step": 6512
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.120587647520824e-08,
+      "loss": 1.1646,
+      "step": 6513
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0900500120424345e-08,
+      "loss": 1.0625,
+      "step": 6514
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.059625490163144e-08,
+      "loss": 1.451,
+      "step": 6515
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.029314088822511e-08,
+      "loss": 1.7612,
+      "step": 6516
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.99911581493434e-08,
+      "loss": 1.7322,
+      "step": 6517
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9690306753866204e-08,
+      "loss": 1.396,
+      "step": 6518
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.939058677041641e-08,
+      "loss": 1.2608,
+      "step": 6519
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9091998267357654e-08,
+      "loss": 1.5633,
+      "step": 6520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.879454131279603e-08,
+      "loss": 1.0621,
+      "step": 6521
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.849821597457892e-08,
+      "loss": 1.3622,
+      "step": 6522
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.820302232029671e-08,
+      "loss": 1.5824,
+      "step": 6523
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.790896041728109e-08,
+      "loss": 1.272,
+      "step": 6524
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.761603033260453e-08,
+      "loss": 1.4662,
+      "step": 6525
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.732423213308356e-08,
+      "loss": 1.8304,
+      "step": 6526
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.70335658852744e-08,
+      "loss": 1.5522,
+      "step": 6527
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.674403165547624e-08,
+      "loss": 1.0505,
+      "step": 6528
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.645562950973014e-08,
+      "loss": 1.3929,
+      "step": 6529
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.616835951381847e-08,
+      "loss": 1.5132,
+      "step": 6530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5882221733264945e-08,
+      "loss": 1.3619,
+      "step": 6531
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5597216233336785e-08,
+      "loss": 1.3942,
+      "step": 6532
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.531334307904033e-08,
+      "loss": 1.4662,
+      "step": 6533
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.503060233512601e-08,
+      "loss": 1.2051,
+      "step": 6534
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.474899406608501e-08,
+      "loss": 1.3685,
+      "step": 6535
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.446851833614984e-08,
+      "loss": 1.6825,
+      "step": 6536
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.418917520929543e-08,
+      "loss": 1.5488,
+      "step": 6537
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3910964749237495e-08,
+      "loss": 1.1066,
+      "step": 6538
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.363388701943471e-08,
+      "loss": 1.5146,
+      "step": 6539
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3357942083085404e-08,
+      "loss": 1.5334,
+      "step": 6540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3083130003131455e-08,
+      "loss": 1.6193,
+      "step": 6541
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.280945084225551e-08,
+      "loss": 1.7338,
+      "step": 6542
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.253690466288206e-08,
+      "loss": 1.2603,
+      "step": 6543
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.22654915271764e-08,
+      "loss": 1.5466,
+      "step": 6544
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.199521149704676e-08,
+      "loss": 1.4796,
+      "step": 6545
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.172606463414163e-08,
+      "loss": 1.6566,
+      "step": 6546
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.145805099985133e-08,
+      "loss": 1.6059,
+      "step": 6547
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.119117065530808e-08,
+      "loss": 1.3927,
+      "step": 6548
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.09254236613854e-08,
+      "loss": 1.3415,
+      "step": 6549
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.066081007869925e-08,
+      "loss": 1.4653,
+      "step": 6550
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.039732996760469e-08,
+      "loss": 0.8721,
+      "step": 6551
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.013498338820031e-08,
+      "loss": 1.3444,
+      "step": 6552
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9873770400325465e-08,
+      "loss": 1.4883,
+      "step": 6553
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.961369106356138e-08,
+      "loss": 1.732,
+      "step": 6554
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9354745437230048e-08,
+      "loss": 1.1068,
+      "step": 6555
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.909693358039478e-08,
+      "loss": 1.201,
+      "step": 6556
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.884025555186132e-08,
+      "loss": 1.1733,
+      "step": 6557
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.858471141017616e-08,
+      "loss": 1.5891,
+      "step": 6558
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8330301213626566e-08,
+      "loss": 1.2876,
+      "step": 6559
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8077025020241677e-08,
+      "loss": 1.2538,
+      "step": 6560
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7824882887793058e-08,
+      "loss": 1.5647,
+      "step": 6561
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7573874873791372e-08,
+      "loss": 1.4473,
+      "step": 6562
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.732400103549082e-08,
+      "loss": 1.5993,
+      "step": 6563
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7075261429884702e-08,
+      "loss": 1.4875,
+      "step": 6564
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6827656113709855e-08,
+      "loss": 1.1383,
+      "step": 6565
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.658118514344221e-08,
+      "loss": 1.4831,
+      "step": 6566
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6335848575301247e-08,
+      "loss": 1.6907,
+      "step": 6567
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6091646465246646e-08,
+      "loss": 1.4108,
+      "step": 6568
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5848578868977738e-08,
+      "loss": 1.5429,
+      "step": 6569
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.560664584193795e-08,
+      "loss": 1.3079,
+      "step": 6570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5365847439309255e-08,
+      "loss": 1.3192,
+      "step": 6571
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5126183716017717e-08,
+      "loss": 1.5782,
+      "step": 6572
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4887654726727385e-08,
+      "loss": 1.5325,
+      "step": 6573
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4650260525846404e-08,
+      "loss": 1.303,
+      "step": 6574
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.441400116752146e-08,
+      "loss": 1.6058,
+      "step": 6575
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.417887670564223e-08,
+      "loss": 1.4004,
+      "step": 6576
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.394488719383914e-08,
+      "loss": 1.4069,
+      "step": 6577
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3712032685483387e-08,
+      "loss": 1.7192,
+      "step": 6578
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3480313233688045e-08,
+      "loss": 1.2919,
+      "step": 6579
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3249728891305277e-08,
+      "loss": 1.5826,
+      "step": 6580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.302027971093135e-08,
+      "loss": 1.6592,
+      "step": 6581
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2791965744900502e-08,
+      "loss": 1.3638,
+      "step": 6582
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2564787045291082e-08,
+      "loss": 1.3561,
+      "step": 6583
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.233874366391997e-08,
+      "loss": 1.5773,
+      "step": 6584
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2113835652346483e-08,
+      "loss": 1.5255,
+      "step": 6585
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1890063061870138e-08,
+      "loss": 1.2242,
+      "step": 6586
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1667425943532884e-08,
+      "loss": 1.6444,
+      "step": 6587
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.144592434811521e-08,
+      "loss": 1.5711,
+      "step": 6588
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1225558326141148e-08,
+      "loss": 1.5214,
+      "step": 6589
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1006327927874938e-08,
+      "loss": 1.2196,
+      "step": 6590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0788233203320464e-08,
+      "loss": 1.8077,
+      "step": 6591
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0571274202224045e-08,
+      "loss": 1.3186,
+      "step": 6592
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0355450974072766e-08,
+      "loss": 1.5316,
+      "step": 6593
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0140763568093912e-08,
+      "loss": 1.4328,
+      "step": 6594
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.992721203325665e-08,
+      "loss": 1.1719,
+      "step": 6595
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9714796418270897e-08,
+      "loss": 1.5251,
+      "step": 6596
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9503516771585683e-08,
+      "loss": 1.4263,
+      "step": 6597
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9293373141394124e-08,
+      "loss": 1.4749,
+      "step": 6598
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9084365575627872e-08,
+      "loss": 1.0616,
+      "step": 6599
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8876494121959908e-08,
+      "loss": 1.4088,
+      "step": 6600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.866975882780453e-08,
+      "loss": 1.5003,
+      "step": 6601
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8464159740316236e-08,
+      "loss": 1.149,
+      "step": 6602
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.82596969063914e-08,
+      "loss": 1.6997,
+      "step": 6603
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8056370372665498e-08,
+      "loss": 1.4694,
+      "step": 6604
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.785418018551699e-08,
+      "loss": 1.8376,
+      "step": 6605
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7653126391063425e-08,
+      "loss": 1.6066,
+      "step": 6606
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.745320903516423e-08,
+      "loss": 1.4789,
+      "step": 6607
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7254428163419047e-08,
+      "loss": 1.2538,
+      "step": 6608
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.705678382116771e-08,
+      "loss": 1.6407,
+      "step": 6609
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.68602760534925e-08,
+      "loss": 0.99,
+      "step": 6610
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6664904905214773e-08,
+      "loss": 1.2928,
+      "step": 6611
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6470670420897782e-08,
+      "loss": 1.1248,
+      "step": 6612
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.627757264484442e-08,
+      "loss": 1.864,
+      "step": 6613
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6085611621100005e-08,
+      "loss": 1.4386,
+      "step": 6614
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5894787393448408e-08,
+      "loss": 1.186,
+      "step": 6615
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5705100005415918e-08,
+      "loss": 1.5388,
+      "step": 6616
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5516549500268486e-08,
+      "loss": 1.5501,
+      "step": 6617
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5329135921013928e-08,
+      "loss": 1.4996,
+      "step": 6618
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.514285931039916e-08,
+      "loss": 1.9336,
+      "step": 6619
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4957719710912976e-08,
+      "loss": 1.2969,
+      "step": 6620
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4773717164784373e-08,
+      "loss": 1.0071,
+      "step": 6621
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4590851713983112e-08,
+      "loss": 1.3557,
+      "step": 6622
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.440912340021916e-08,
+      "loss": 1.3737,
+      "step": 6623
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4228532264943805e-08,
+      "loss": 1.2258,
+      "step": 6624
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4049078349348544e-08,
+      "loss": 1.7971,
+      "step": 6625
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.387076169436563e-08,
+      "loss": 1.3741,
+      "step": 6626
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3693582340668088e-08,
+      "loss": 1.6421,
+      "step": 6627
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3517540328669143e-08,
+      "loss": 1.6139,
+      "step": 6628
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3342635698522232e-08,
+      "loss": 1.5872,
+      "step": 6629
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3168868490122112e-08,
+      "loss": 1.1523,
+      "step": 6630
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2996238743104295e-08,
+      "loss": 1.3545,
+      "step": 6631
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2824746496844509e-08,
+      "loss": 1.5319,
+      "step": 6632
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2654391790458132e-08,
+      "loss": 1.1471,
+      "step": 6633
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2485174662802413e-08,
+      "loss": 1.8656,
+      "step": 6634
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2317095152474256e-08,
+      "loss": 1.3867,
+      "step": 6635
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2150153297811884e-08,
+      "loss": 1.4013,
+      "step": 6636
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1984349136893725e-08,
+      "loss": 1.3165,
+      "step": 6637
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1819682707537861e-08,
+      "loss": 1.4139,
+      "step": 6638
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1656154047303691e-08,
+      "loss": 1.6285,
+      "step": 6639
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1493763193491936e-08,
+      "loss": 1.5067,
+      "step": 6640
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1332510183141299e-08,
+      "loss": 1.2422,
+      "step": 6641
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1172395053033468e-08,
+      "loss": 1.5666,
+      "step": 6642
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1013417839689232e-08,
+      "loss": 1.6052,
+      "step": 6643
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0855578579370696e-08,
+      "loss": 1.491,
+      "step": 6644
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0698877308079059e-08,
+      "loss": 1.3373,
+      "step": 6645
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0543314061557398e-08,
+      "loss": 1.7546,
+      "step": 6646
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0388888875288438e-08,
+      "loss": 1.2018,
+      "step": 6647
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.023560178449512e-08,
+      "loss": 1.0826,
+      "step": 6648
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0083452824141693e-08,
+      "loss": 1.2349,
+      "step": 6649
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.93244202893262e-09,
+      "loss": 1.2742,
+      "step": 6650
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.782569433310906e-09,
+      "loss": 1.5237,
+      "step": 6651
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.633835071463094e-09,
+      "loss": 1.2664,
+      "step": 6652
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.486238977313711e-09,
+      "loss": 1.6848,
+      "step": 6653
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.339781184528053e-09,
+      "loss": 1.1613,
+      "step": 6654
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.194461726513282e-09,
+      "loss": 1.5205,
+      "step": 6655
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.050280636414555e-09,
+      "loss": 1.7565,
+      "step": 6656
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.907237947118896e-09,
+      "loss": 1.3331,
+      "step": 6657
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.765333691253541e-09,
+      "loss": 1.3895,
+      "step": 6658
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.624567901185931e-09,
+      "loss": 1.4146,
+      "step": 6659
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.484940609023163e-09,
+      "loss": 1.3584,
+      "step": 6660
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.346451846614201e-09,
+      "loss": 1.5532,
+      "step": 6661
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.209101645546558e-09,
+      "loss": 1.5897,
+      "step": 6662
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.072890037149616e-09,
+      "loss": 1.4117,
+      "step": 6663
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.937817052491303e-09,
+      "loss": 1.6237,
+      "step": 6664
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.803882722381417e-09,
+      "loss": 1.3766,
+      "step": 6665
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.67108707736941e-09,
+      "loss": 1.5397,
+      "step": 6666
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.539430147745497e-09,
+      "loss": 1.2422,
+      "step": 6667
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.408911963538989e-09,
+      "loss": 1.4103,
+      "step": 6668
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.279532554519963e-09,
+      "loss": 1.5404,
+      "step": 6669
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.151291950199812e-09,
+      "loss": 1.8361,
+      "step": 6670
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.024190179828472e-09,
+      "loss": 1.5258,
+      "step": 6671
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.898227272398306e-09,
+      "loss": 0.9932,
+      "step": 6672
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.7734032566391104e-09,
+      "loss": 1.3952,
+      "step": 6673
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.649718161022556e-09,
+      "loss": 1.4845,
+      "step": 6674
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.527172013761074e-09,
+      "loss": 1.5097,
+      "step": 6675
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.40576484280675e-09,
+      "loss": 1.3972,
+      "step": 6676
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.2854966758507665e-09,
+      "loss": 1.7297,
+      "step": 6677
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.166367540325624e-09,
+      "loss": 1.5337,
+      "step": 6678
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.048377463404587e-09,
+      "loss": 1.3989,
+      "step": 6679
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.931526472000015e-09,
+      "loss": 1.6226,
+      "step": 6680
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.815814592763924e-09,
+      "loss": 1.3609,
+      "step": 6681
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.7012418520907555e-09,
+      "loss": 1.044,
+      "step": 6682
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.587808276112938e-09,
+      "loss": 1.3741,
+      "step": 6683
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.475513890704221e-09,
+      "loss": 1.1859,
+      "step": 6684
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.364358721478002e-09,
+      "loss": 1.6587,
+      "step": 6685
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.254342793787892e-09,
+      "loss": 1.3144,
+      "step": 6686
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.145466132728261e-09,
+      "loss": 1.2457,
+      "step": 6687
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.037728763132576e-09,
+      "loss": 1.1826,
+      "step": 6688
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.931130709575071e-09,
+      "loss": 1.7846,
+      "step": 6689
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.825671996370184e-09,
+      "loss": 1.2211,
+      "step": 6690
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.721352647572564e-09,
+      "loss": 1.4937,
+      "step": 6691
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.618172686975953e-09,
+      "loss": 1.5942,
+      "step": 6692
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.516132138115414e-09,
+      "loss": 1.5878,
+      "step": 6693
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.415231024265665e-09,
+      "loss": 1.4889,
+      "step": 6694
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.315469368441627e-09,
+      "loss": 1.1014,
+      "step": 6695
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.216847193398432e-09,
+      "loss": 1.584,
+      "step": 6696
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.119364521630864e-09,
+      "loss": 1.3963,
+      "step": 6697
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.023021375373914e-09,
+      "loss": 1.3797,
+      "step": 6698
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.927817776602782e-09,
+      "loss": 1.56,
+      "step": 6699
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.833753747032876e-09,
+      "loss": 1.5333,
+      "step": 6700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.740829308119254e-09,
+      "loss": 1.3018,
+      "step": 6701
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6490444810582946e-09,
+      "loss": 1.1352,
+      "step": 6702
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5583992867843643e-09,
+      "loss": 1.53,
+      "step": 6703
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4688937459737004e-09,
+      "loss": 1.2755,
+      "step": 6704
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3805278790416395e-09,
+      "loss": 1.4059,
+      "step": 6705
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2933017061431705e-09,
+      "loss": 1.5789,
+      "step": 6706
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2072152471751548e-09,
+      "loss": 1.2106,
+      "step": 6707
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1222685217729974e-09,
+      "loss": 1.1076,
+      "step": 6708
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0384615493117555e-09,
+      "loss": 1.5762,
+      "step": 6709
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9557943489078035e-09,
+      "loss": 1.342,
+      "step": 6710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8742669394171696e-09,
+      "loss": 1.254,
+      "step": 6711
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7938793394349796e-09,
+      "loss": 1.2096,
+      "step": 6712
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.714631567297121e-09,
+      "loss": 1.6827,
+      "step": 6713
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6365236410802466e-09,
+      "loss": 1.2002,
+      "step": 6714
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5595555786001035e-09,
+      "loss": 1.6999,
+      "step": 6715
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4837273974115393e-09,
+      "loss": 1.4557,
+      "step": 6716
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4090391148112734e-09,
+      "loss": 1.4419,
+      "step": 6717
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.335490747835678e-09,
+      "loss": 1.4646,
+      "step": 6718
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.263082313259113e-09,
+      "loss": 1.0542,
+      "step": 6719
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.191813827598921e-09,
+      "loss": 1.2819,
+      "step": 6720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1216853071098776e-09,
+      "loss": 1.324,
+      "step": 6721
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.052696767788631e-09,
+      "loss": 1.3638,
+      "step": 6722
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9848482253703727e-09,
+      "loss": 1.3243,
+      "step": 6723
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.918139695331611e-09,
+      "loss": 1.4931,
+      "step": 6724
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8525711928873979e-09,
+      "loss": 1.3363,
+      "step": 6725
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7881427329935476e-09,
+      "loss": 1.3195,
+      "step": 6726
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7248543303455267e-09,
+      "loss": 1.6032,
+      "step": 6727
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6627059993795658e-09,
+      "loss": 1.5024,
+      "step": 6728
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6016977542715473e-09,
+      "loss": 1.6413,
+      "step": 6729
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5418296089358964e-09,
+      "loss": 1.5714,
+      "step": 6730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4831015770289114e-09,
+      "loss": 1.3315,
+      "step": 6731
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4255136719454332e-09,
+      "loss": 1.2921,
+      "step": 6732
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3690659068216206e-09,
+      "loss": 1.4385,
+      "step": 6733
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3137582945327298e-09,
+      "loss": 1.3803,
+      "step": 6734
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2595908476936702e-09,
+      "loss": 1.2225,
+      "step": 6735
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2065635786595586e-09,
+      "loss": 1.8473,
+      "step": 6736
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1546764995257198e-09,
+      "loss": 1.3572,
+      "step": 6737
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1039296221276863e-09,
+      "loss": 1.135,
+      "step": 6738
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.054322958039533e-09,
+      "loss": 1.1447,
+      "step": 6739
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0058565185772085e-09,
+      "loss": 1.3126,
+      "step": 6740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.58530314794648e-10,
+      "loss": 1.4366,
+      "step": 6741
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.12344357487105e-10,
+      "loss": 1.2632,
+      "step": 6742
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.672986571894859e-10,
+      "loss": 1.1657,
+      "step": 6743
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.233932241757947e-10,
+      "loss": 1.1616,
+      "step": 6744
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.806280684613532e-10,
+      "loss": 1.3858,
+      "step": 6745
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.390031998000258e-10,
+      "loss": 1.5523,
+      "step": 6746
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.985186276858847e-10,
+      "loss": 1.4408,
+      "step": 6747
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.591743613537649e-10,
+      "loss": 1.3533,
+      "step": 6748
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.209704097781543e-10,
+      "loss": 1.6099,
+      "step": 6749
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.839067816720834e-10,
+      "loss": 0.87,
+      "step": 6750
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.479834854899002e-10,
+      "loss": 1.2589,
+      "step": 6751
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.132005294256059e-10,
+      "loss": 1.4909,
+      "step": 6752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.795579214128542e-10,
+      "loss": 1.4575,
+      "step": 6753
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.470556691255068e-10,
+      "loss": 1.1994,
+      "step": 6754
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.156937799770777e-10,
+      "loss": 1.337,
+      "step": 6755
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.854722611201789e-10,
+      "loss": 1.7917,
+      "step": 6756
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.563911194492953e-10,
+      "loss": 1.0346,
+      "step": 6757
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2845036159745434e-10,
+      "loss": 1.1701,
+      "step": 6758
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.016499939367812e-10,
+      "loss": 1.5362,
+      "step": 6759
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7599002258127395e-10,
+      "loss": 1.5253,
+      "step": 6760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.514704533829182e-10,
+      "loss": 1.0947,
+      "step": 6761
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2809129193557267e-10,
+      "loss": 1.4242,
+      "step": 6762
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0585254357108342e-10,
+      "loss": 1.1974,
+      "step": 6763
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8475421336205947e-10,
+      "loss": 1.4117,
+      "step": 6764
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6479630612076247e-10,
+      "loss": 1.4132,
+      "step": 6765
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4597882639966198e-10,
+      "loss": 1.4279,
+      "step": 6766
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2830177849088022e-10,
+      "loss": 1.2807,
+      "step": 6767
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1176516642674718e-10,
+      "loss": 1.4748,
+      "step": 6768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.636899397813537e-11,
+      "loss": 1.2146,
+      "step": 6769
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.211326465834556e-11,
+      "loss": 1.4016,
+      "step": 6770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.899798171755567e-11,
+      "loss": 1.0281,
+      "step": 6771
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.702314814781673e-11,
+      "loss": 1.536,
+      "step": 6772
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.61887666808325e-11,
+      "loss": 1.3019,
+      "step": 6773
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.649483978740431e-11,
+      "loss": 1.5718,
+      "step": 6774
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.794136967854133e-11,
+      "loss": 1.5401,
+      "step": 6775
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0528358306015627e-11,
+      "loss": 1.2957,
+      "step": 6776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.425580735958665e-11,
+      "loss": 1.3858,
+      "step": 6777
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.123718270886983e-12,
+      "loss": 1.4912,
+      "step": 6778
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.132092210091699e-12,
+      "loss": 1.5786,
+      "step": 6779
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2809300881387887e-12,
+      "loss": 1.1262,
+      "step": 6780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.702325545087206e-13,
+      "loss": 1.5859,
+      "step": 6781
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.5494,
+      "step": 6782
+    },
+    {
+      "epoch": 1.0,
+      "step": 6782,
+      "total_flos": 9.132175545499386e+18,
+      "train_loss": 1.466076029784839,
+      "train_runtime": 230106.0482,
+      "train_samples_per_second": 0.472,
+      "train_steps_per_second": 0.029
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 6782,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 9.132175545499386e+18,
+  "trial_name": null,
+  "trial_params": null
+}