diff --git "a/ReCogDrive_VLM/trainer_state.json" "b/ReCogDrive_VLM/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/ReCogDrive_VLM/trainer_state.json"
@@ -0,0 +1,12270 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9977957384276266,
+  "eval_steps": 500,
+  "global_step": 2040,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9607843137254904e-07,
+      "loss": 1.2901,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.921568627450981e-07,
+      "loss": 1.2562,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.882352941176471e-07,
+      "loss": 1.264,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.843137254901962e-07,
+      "loss": 1.3028,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.80392156862745e-07,
+      "loss": 1.2955,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1764705882352942e-06,
+      "loss": 1.2551,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3725490196078434e-06,
+      "loss": 1.2509,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5686274509803923e-06,
+      "loss": 1.2005,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7647058823529414e-06,
+      "loss": 1.1993,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.96078431372549e-06,
+      "loss": 1.1764,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1568627450980393e-06,
+      "loss": 1.1002,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3529411764705885e-06,
+      "loss": 1.0689,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.549019607843137e-06,
+      "loss": 1.0329,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7450980392156867e-06,
+      "loss": 1.0154,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9411764705882355e-06,
+      "loss": 0.9664,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1372549019607846e-06,
+      "loss": 0.9411,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.9469,
+      "step": 17
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.529411764705883e-06,
+      "loss": 0.9171,
+      "step": 18
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7254901960784316e-06,
+      "loss": 0.9206,
+      "step": 19
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.92156862745098e-06,
+      "loss": 0.8955,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.11764705882353e-06,
+      "loss": 0.8713,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.313725490196079e-06,
+      "loss": 0.8927,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.509803921568628e-06,
+      "loss": 0.8487,
+      "step": 23
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.705882352941177e-06,
+      "loss": 0.8399,
+      "step": 24
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.901960784313726e-06,
+      "loss": 0.823,
+      "step": 25
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.098039215686274e-06,
+      "loss": 0.8275,
+      "step": 26
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.294117647058824e-06,
+      "loss": 0.8102,
+      "step": 27
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.4901960784313735e-06,
+      "loss": 0.7984,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.686274509803922e-06,
+      "loss": 0.7987,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.882352941176471e-06,
+      "loss": 0.7935,
+      "step": 30
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.07843137254902e-06,
+      "loss": 0.8004,
+      "step": 31
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.274509803921569e-06,
+      "loss": 0.7816,
+      "step": 32
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.470588235294119e-06,
+      "loss": 0.7891,
+      "step": 33
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.7771,
+      "step": 34
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.862745098039216e-06,
+      "loss": 0.7613,
+      "step": 35
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.058823529411766e-06,
+      "loss": 0.7558,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.2549019607843145e-06,
+      "loss": 0.7547,
+      "step": 37
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.450980392156863e-06,
+      "loss": 0.7595,
+      "step": 38
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.647058823529411e-06,
+      "loss": 0.7511,
+      "step": 39
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.84313725490196e-06,
+      "loss": 0.7504,
+      "step": 40
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.03921568627451e-06,
+      "loss": 0.7401,
+      "step": 41
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.23529411764706e-06,
+      "loss": 0.7699,
+      "step": 42
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.43137254901961e-06,
+      "loss": 0.7405,
+      "step": 43
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.627450980392157e-06,
+      "loss": 0.7506,
+      "step": 44
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.823529411764707e-06,
+      "loss": 0.7427,
+      "step": 45
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.019607843137256e-06,
+      "loss": 0.7575,
+      "step": 46
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.215686274509804e-06,
+      "loss": 0.7624,
+      "step": 47
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.411764705882354e-06,
+      "loss": 0.754,
+      "step": 48
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.607843137254903e-06,
+      "loss": 0.7201,
+      "step": 49
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.803921568627451e-06,
+      "loss": 0.7409,
+      "step": 50
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1e-05,
+      "loss": 0.726,
+      "step": 51
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0196078431372549e-05,
+      "loss": 0.6998,
+      "step": 52
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.03921568627451e-05,
+      "loss": 0.7247,
+      "step": 53
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0588235294117648e-05,
+      "loss": 0.7288,
+      "step": 54
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0784313725490196e-05,
+      "loss": 0.7298,
+      "step": 55
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0980392156862747e-05,
+      "loss": 0.7276,
+      "step": 56
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1176470588235295e-05,
+      "loss": 0.7124,
+      "step": 57
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1372549019607844e-05,
+      "loss": 0.6967,
+      "step": 58
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1568627450980394e-05,
+      "loss": 0.7298,
+      "step": 59
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1764705882352942e-05,
+      "loss": 0.7364,
+      "step": 60
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1960784313725491e-05,
+      "loss": 0.7027,
+      "step": 61
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.215686274509804e-05,
+      "loss": 0.7188,
+      "step": 62
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.235294117647059e-05,
+      "loss": 0.7033,
+      "step": 63
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2549019607843138e-05,
+      "loss": 0.7342,
+      "step": 64
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2745098039215686e-05,
+      "loss": 0.7016,
+      "step": 65
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2941176470588238e-05,
+      "loss": 0.7062,
+      "step": 66
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3137254901960785e-05,
+      "loss": 0.6794,
+      "step": 67
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.7153,
+      "step": 68
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3529411764705885e-05,
+      "loss": 0.7007,
+      "step": 69
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3725490196078432e-05,
+      "loss": 0.7179,
+      "step": 70
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.392156862745098e-05,
+      "loss": 0.6999,
+      "step": 71
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4117647058823532e-05,
+      "loss": 0.7049,
+      "step": 72
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.431372549019608e-05,
+      "loss": 0.7057,
+      "step": 73
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4509803921568629e-05,
+      "loss": 0.696,
+      "step": 74
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4705882352941179e-05,
+      "loss": 0.684,
+      "step": 75
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4901960784313726e-05,
+      "loss": 0.6987,
+      "step": 76
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5098039215686276e-05,
+      "loss": 0.6965,
+      "step": 77
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5294117647058822e-05,
+      "loss": 0.6937,
+      "step": 78
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5490196078431373e-05,
+      "loss": 0.6952,
+      "step": 79
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.568627450980392e-05,
+      "loss": 0.6863,
+      "step": 80
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5882352941176473e-05,
+      "loss": 0.6711,
+      "step": 81
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.607843137254902e-05,
+      "loss": 0.7013,
+      "step": 82
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.627450980392157e-05,
+      "loss": 0.6935,
+      "step": 83
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.647058823529412e-05,
+      "loss": 0.6905,
+      "step": 84
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.6942,
+      "step": 85
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.686274509803922e-05,
+      "loss": 0.6869,
+      "step": 86
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7058823529411767e-05,
+      "loss": 0.6925,
+      "step": 87
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7254901960784314e-05,
+      "loss": 0.7018,
+      "step": 88
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7450980392156866e-05,
+      "loss": 0.6966,
+      "step": 89
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7647058823529414e-05,
+      "loss": 0.7022,
+      "step": 90
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7843137254901965e-05,
+      "loss": 0.6755,
+      "step": 91
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8039215686274513e-05,
+      "loss": 0.68,
+      "step": 92
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.823529411764706e-05,
+      "loss": 0.6702,
+      "step": 93
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.843137254901961e-05,
+      "loss": 0.6986,
+      "step": 94
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.862745098039216e-05,
+      "loss": 0.6945,
+      "step": 95
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8823529411764708e-05,
+      "loss": 0.6928,
+      "step": 96
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9019607843137255e-05,
+      "loss": 0.6962,
+      "step": 97
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9215686274509807e-05,
+      "loss": 0.6755,
+      "step": 98
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9411764705882355e-05,
+      "loss": 0.687,
+      "step": 99
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9607843137254903e-05,
+      "loss": 0.6884,
+      "step": 100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9803921568627454e-05,
+      "loss": 0.6792,
+      "step": 101
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.7079,
+      "step": 102
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.019607843137255e-05,
+      "loss": 0.667,
+      "step": 103
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0392156862745097e-05,
+      "loss": 0.6911,
+      "step": 104
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0588235294117645e-05,
+      "loss": 0.6788,
+      "step": 105
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.07843137254902e-05,
+      "loss": 0.6863,
+      "step": 106
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.0980392156862748e-05,
+      "loss": 0.6765,
+      "step": 107
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1176470588235296e-05,
+      "loss": 0.6902,
+      "step": 108
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1372549019607844e-05,
+      "loss": 0.6833,
+      "step": 109
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.156862745098039e-05,
+      "loss": 0.6907,
+      "step": 110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1764705882352943e-05,
+      "loss": 0.7065,
+      "step": 111
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1960784313725494e-05,
+      "loss": 0.6817,
+      "step": 112
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2156862745098042e-05,
+      "loss": 0.679,
+      "step": 113
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.235294117647059e-05,
+      "loss": 0.6694,
+      "step": 114
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2549019607843138e-05,
+      "loss": 0.6801,
+      "step": 115
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.274509803921569e-05,
+      "loss": 0.6605,
+      "step": 116
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2941176470588237e-05,
+      "loss": 0.6765,
+      "step": 117
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.3137254901960788e-05,
+      "loss": 0.6992,
+      "step": 118
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.3333333333333336e-05,
+      "loss": 0.6726,
+      "step": 119
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3529411764705884e-05,
+      "loss": 0.6842,
+      "step": 120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3725490196078435e-05,
+      "loss": 0.6739,
+      "step": 121
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3921568627450983e-05,
+      "loss": 0.6925,
+      "step": 122
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.411764705882353e-05,
+      "loss": 0.6978,
+      "step": 123
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.431372549019608e-05,
+      "loss": 0.6843,
+      "step": 124
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.450980392156863e-05,
+      "loss": 0.6778,
+      "step": 125
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.470588235294118e-05,
+      "loss": 0.6738,
+      "step": 126
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.490196078431373e-05,
+      "loss": 0.7039,
+      "step": 127
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5098039215686277e-05,
+      "loss": 0.6818,
+      "step": 128
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5294117647058825e-05,
+      "loss": 0.6855,
+      "step": 129
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5490196078431373e-05,
+      "loss": 0.6922,
+      "step": 130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5686274509803927e-05,
+      "loss": 0.6869,
+      "step": 131
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5882352941176475e-05,
+      "loss": 0.6961,
+      "step": 132
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6078431372549023e-05,
+      "loss": 0.6835,
+      "step": 133
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.627450980392157e-05,
+      "loss": 0.6737,
+      "step": 134
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.647058823529412e-05,
+      "loss": 0.6853,
+      "step": 135
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.6987,
+      "step": 136
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6862745098039214e-05,
+      "loss": 0.68,
+      "step": 137
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.705882352941177e-05,
+      "loss": 0.6501,
+      "step": 138
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7254901960784317e-05,
+      "loss": 0.6555,
+      "step": 139
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7450980392156865e-05,
+      "loss": 0.6653,
+      "step": 140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7647058823529413e-05,
+      "loss": 0.689,
+      "step": 141
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.784313725490196e-05,
+      "loss": 0.6933,
+      "step": 142
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8039215686274512e-05,
+      "loss": 0.6738,
+      "step": 143
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8235294117647063e-05,
+      "loss": 0.6704,
+      "step": 144
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.843137254901961e-05,
+      "loss": 0.671,
+      "step": 145
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.862745098039216e-05,
+      "loss": 0.6635,
+      "step": 146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8823529411764707e-05,
+      "loss": 0.6946,
+      "step": 147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9019607843137258e-05,
+      "loss": 0.7075,
+      "step": 148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9215686274509806e-05,
+      "loss": 0.6759,
+      "step": 149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9411764705882357e-05,
+      "loss": 0.6625,
+      "step": 150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9607843137254905e-05,
+      "loss": 0.682,
+      "step": 151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9803921568627453e-05,
+      "loss": 0.6556,
+      "step": 152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.6779,
+      "step": 153
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.0196078431372552e-05,
+      "loss": 0.6641,
+      "step": 154
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.03921568627451e-05,
+      "loss": 0.6733,
+      "step": 155
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.0588235294117644e-05,
+      "loss": 0.6632,
+      "step": 156
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.07843137254902e-05,
+      "loss": 0.6679,
+      "step": 157
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.098039215686275e-05,
+      "loss": 0.6577,
+      "step": 158
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.11764705882353e-05,
+      "loss": 0.6884,
+      "step": 159
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.137254901960784e-05,
+      "loss": 0.6706,
+      "step": 160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.1568627450980394e-05,
+      "loss": 0.6742,
+      "step": 161
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.1764705882352945e-05,
+      "loss": 0.6653,
+      "step": 162
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.1960784313725496e-05,
+      "loss": 0.6656,
+      "step": 163
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.215686274509804e-05,
+      "loss": 0.6581,
+      "step": 164
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.235294117647059e-05,
+      "loss": 0.6972,
+      "step": 165
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.254901960784314e-05,
+      "loss": 0.6697,
+      "step": 166
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.274509803921569e-05,
+      "loss": 0.6818,
+      "step": 167
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.294117647058824e-05,
+      "loss": 0.6646,
+      "step": 168
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.313725490196079e-05,
+      "loss": 0.6866,
+      "step": 169
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.6941,
+      "step": 170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.3529411764705886e-05,
+      "loss": 0.6736,
+      "step": 171
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.372549019607844e-05,
+      "loss": 0.6691,
+      "step": 172
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.392156862745098e-05,
+      "loss": 0.665,
+      "step": 173
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.411764705882353e-05,
+      "loss": 0.676,
+      "step": 174
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.431372549019608e-05,
+      "loss": 0.6718,
+      "step": 175
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.450980392156863e-05,
+      "loss": 0.6728,
+      "step": 176
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.470588235294118e-05,
+      "loss": 0.6781,
+      "step": 177
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.490196078431373e-05,
+      "loss": 0.6865,
+      "step": 178
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.5098039215686276e-05,
+      "loss": 0.6737,
+      "step": 179
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.529411764705883e-05,
+      "loss": 0.6742,
+      "step": 180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.549019607843137e-05,
+      "loss": 0.6431,
+      "step": 181
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.568627450980393e-05,
+      "loss": 0.6789,
+      "step": 182
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.5882352941176474e-05,
+      "loss": 0.6949,
+      "step": 183
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6078431372549025e-05,
+      "loss": 0.6718,
+      "step": 184
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.627450980392157e-05,
+      "loss": 0.6892,
+      "step": 185
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.647058823529412e-05,
+      "loss": 0.6821,
+      "step": 186
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6666666666666666e-05,
+      "loss": 0.6678,
+      "step": 187
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.686274509803922e-05,
+      "loss": 0.6739,
+      "step": 188
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.705882352941177e-05,
+      "loss": 0.661,
+      "step": 189
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.725490196078432e-05,
+      "loss": 0.6759,
+      "step": 190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.7450980392156864e-05,
+      "loss": 0.6697,
+      "step": 191
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.7647058823529415e-05,
+      "loss": 0.6613,
+      "step": 192
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.784313725490196e-05,
+      "loss": 0.6628,
+      "step": 193
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.803921568627451e-05,
+      "loss": 0.6667,
+      "step": 194
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.823529411764706e-05,
+      "loss": 0.6802,
+      "step": 195
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.8431372549019614e-05,
+      "loss": 0.6609,
+      "step": 196
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.862745098039216e-05,
+      "loss": 0.6925,
+      "step": 197
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.882352941176471e-05,
+      "loss": 0.6721,
+      "step": 198
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.901960784313726e-05,
+      "loss": 0.6749,
+      "step": 199
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.9215686274509805e-05,
+      "loss": 0.6639,
+      "step": 200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9411764705882356e-05,
+      "loss": 0.6783,
+      "step": 201
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.960784313725491e-05,
+      "loss": 0.6738,
+      "step": 202
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.980392156862745e-05,
+      "loss": 0.67,
+      "step": 203
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4e-05,
+      "loss": 0.6952,
+      "step": 204
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.999997072113173e-05,
+      "loss": 0.6984,
+      "step": 205
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9999882884612655e-05,
+      "loss": 0.6686,
+      "step": 206
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.999973649069994e-05,
+      "loss": 0.684,
+      "step": 207
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9999531539822216e-05,
+      "loss": 0.6674,
+      "step": 208
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999926803257955e-05,
+      "loss": 0.6586,
+      "step": 209
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999894596974347e-05,
+      "loss": 0.6395,
+      "step": 210
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999856535225693e-05,
+      "loss": 0.6564,
+      "step": 211
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999812618123434e-05,
+      "loss": 0.6539,
+      "step": 212
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999762845796154e-05,
+      "loss": 0.653,
+      "step": 213
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9997072183895813e-05,
+      "loss": 0.6854,
+      "step": 214
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999645736066587e-05,
+      "loss": 0.6837,
+      "step": 215
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999578399007183e-05,
+      "loss": 0.678,
+      "step": 216
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999505207408526e-05,
+      "loss": 0.6673,
+      "step": 217
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999426161484911e-05,
+      "loss": 0.6626,
+      "step": 218
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999341261467777e-05,
+      "loss": 0.6721,
+      "step": 219
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999250507605702e-05,
+      "loss": 0.6794,
+      "step": 220
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9991539001644015e-05,
+      "loss": 0.6743,
+      "step": 221
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.999051439426732e-05,
+      "loss": 0.6405,
+      "step": 222
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.998943125692687e-05,
+      "loss": 0.6674,
+      "step": 223
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.998828959279397e-05,
+      "loss": 0.6789,
+      "step": 224
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.998708940521128e-05,
+      "loss": 0.6754,
+      "step": 225
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.998583069769281e-05,
+      "loss": 0.6633,
+      "step": 226
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.998451347392391e-05,
+      "loss": 0.6753,
+      "step": 227
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.998313773776128e-05,
+      "loss": 0.6588,
+      "step": 228
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.998170349323291e-05,
+      "loss": 0.6438,
+      "step": 229
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.99802107445381e-05,
+      "loss": 0.6719,
+      "step": 230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9978659496047456e-05,
+      "loss": 0.6463,
+      "step": 231
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.997704975230285e-05,
+      "loss": 0.6778,
+      "step": 232
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.997538151801744e-05,
+      "loss": 0.6789,
+      "step": 233
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.997365479807562e-05,
+      "loss": 0.6653,
+      "step": 234
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.997186959753304e-05,
+      "loss": 0.6585,
+      "step": 235
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.997002592161655e-05,
+      "loss": 0.6763,
+      "step": 236
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9968123775724236e-05,
+      "loss": 0.6929,
+      "step": 237
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.996616316542537e-05,
+      "loss": 0.6729,
+      "step": 238
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.996414409646038e-05,
+      "loss": 0.6599,
+      "step": 239
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9962066574740886e-05,
+      "loss": 0.6835,
+      "step": 240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9959930606349625e-05,
+      "loss": 0.6777,
+      "step": 241
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.995773619754049e-05,
+      "loss": 0.6577,
+      "step": 242
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.995548335473844e-05,
+      "loss": 0.6561,
+      "step": 243
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.995317208453955e-05,
+      "loss": 0.6865,
+      "step": 244
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9950802393710966e-05,
+      "loss": 0.6643,
+      "step": 245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9948374289190866e-05,
+      "loss": 0.6708,
+      "step": 246
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9945887778088474e-05,
+      "loss": 0.6565,
+      "step": 247
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9943342867684e-05,
+      "loss": 0.6524,
+      "step": 248
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.994073956542866e-05,
+      "loss": 0.661,
+      "step": 249
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.993807787894463e-05,
+      "loss": 0.6676,
+      "step": 250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.993535781602503e-05,
+      "loss": 0.6739,
+      "step": 251
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.993257938463389e-05,
+      "loss": 0.6642,
+      "step": 252
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.992974259290615e-05,
+      "loss": 0.6706,
+      "step": 253
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.99268474491476e-05,
+      "loss": 0.6616,
+      "step": 254
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9923893961834914e-05,
+      "loss": 0.6715,
+      "step": 255
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9920882139615556e-05,
+      "loss": 0.6676,
+      "step": 256
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.99178119913078e-05,
+      "loss": 0.6843,
+      "step": 257
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.991468352590069e-05,
+      "loss": 0.6735,
+      "step": 258
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.991149675255403e-05,
+      "loss": 0.6777,
+      "step": 259
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.990825168059833e-05,
+      "loss": 0.66,
+      "step": 260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9904948319534786e-05,
+      "loss": 0.679,
+      "step": 261
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9901586679035264e-05,
+      "loss": 0.648,
+      "step": 262
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.989816676894227e-05,
+      "loss": 0.6713,
+      "step": 263
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.989468859926893e-05,
+      "loss": 0.6679,
+      "step": 264
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9891152180198904e-05,
+      "loss": 0.6763,
+      "step": 265
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9887557522086446e-05,
+      "loss": 0.6533,
+      "step": 266
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.98839046354563e-05,
+      "loss": 0.6465,
+      "step": 267
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.98801935310037e-05,
+      "loss": 0.6869,
+      "step": 268
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9876424219594344e-05,
+      "loss": 0.6579,
+      "step": 269
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.987259671226436e-05,
+      "loss": 0.6634,
+      "step": 270
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.986871102022024e-05,
+      "loss": 0.6619,
+      "step": 271
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9864767154838864e-05,
+      "loss": 0.6716,
+      "step": 272
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.986076512766741e-05,
+      "loss": 0.6681,
+      "step": 273
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.985670495042338e-05,
+      "loss": 0.6469,
+      "step": 274
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.985258663499449e-05,
+      "loss": 0.6358,
+      "step": 275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.984841019343872e-05,
+      "loss": 0.6582,
+      "step": 276
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9844175637984205e-05,
+      "loss": 0.6588,
+      "step": 277
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.983988298102926e-05,
+      "loss": 0.6625,
+      "step": 278
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.983553223514228e-05,
+      "loss": 0.6568,
+      "step": 279
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9831123413061765e-05,
+      "loss": 0.6693,
+      "step": 280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.982665652769625e-05,
+      "loss": 0.6517,
+      "step": 281
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9822131592124275e-05,
+      "loss": 0.6598,
+      "step": 282
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.981754861959433e-05,
+      "loss": 0.6743,
+      "step": 283
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.981290762352483e-05,
+      "loss": 0.6596,
+      "step": 284
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9808208617504106e-05,
+      "loss": 0.6618,
+      "step": 285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.980345161529031e-05,
+      "loss": 0.6592,
+      "step": 286
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9798636630811396e-05,
+      "loss": 0.649,
+      "step": 287
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.979376367816511e-05,
+      "loss": 0.6846,
+      "step": 288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.978883277161889e-05,
+      "loss": 0.6627,
+      "step": 289
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9783843925609884e-05,
+      "loss": 0.6615,
+      "step": 290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.977879715474487e-05,
+      "loss": 0.6398,
+      "step": 291
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9773692473800205e-05,
+      "loss": 0.6671,
+      "step": 292
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9768529897721835e-05,
+      "loss": 0.6771,
+      "step": 293
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.976330944162519e-05,
+      "loss": 0.6484,
+      "step": 294
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.975803112079518e-05,
+      "loss": 0.6449,
+      "step": 295
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9752694950686124e-05,
+      "loss": 0.6742,
+      "step": 296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9747300946921725e-05,
+      "loss": 0.649,
+      "step": 297
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9741849125295025e-05,
+      "loss": 0.6603,
+      "step": 298
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9736339501768326e-05,
+      "loss": 0.6547,
+      "step": 299
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9730772092473195e-05,
+      "loss": 0.6619,
+      "step": 300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.972514691371037e-05,
+      "loss": 0.6523,
+      "step": 301
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9719463981949734e-05,
+      "loss": 0.6527,
+      "step": 302
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.971372331383028e-05,
+      "loss": 0.6684,
+      "step": 303
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9707924926160026e-05,
+      "loss": 0.6431,
+      "step": 304
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9702068835916e-05,
+      "loss": 0.6614,
+      "step": 305
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9696155060244166e-05,
+      "loss": 0.6518,
+      "step": 306
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.969018361645939e-05,
+      "loss": 0.6365,
+      "step": 307
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.968415452204538e-05,
+      "loss": 0.6353,
+      "step": 308
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9678067794654664e-05,
+      "loss": 0.6748,
+      "step": 309
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.9671923452108455e-05,
+      "loss": 0.6621,
+      "step": 310
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.966572151239673e-05,
+      "loss": 0.6501,
+      "step": 311
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.965946199367804e-05,
+      "loss": 0.6692,
+      "step": 312
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.965314491427956e-05,
+      "loss": 0.6531,
+      "step": 313
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.964677029269697e-05,
+      "loss": 0.6389,
+      "step": 314
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.964033814759446e-05,
+      "loss": 0.6619,
+      "step": 315
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.963384849780461e-05,
+      "loss": 0.64,
+      "step": 316
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.962730136232838e-05,
+      "loss": 0.6441,
+      "step": 317
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.962069676033505e-05,
+      "loss": 0.6683,
+      "step": 318
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.9614034711162144e-05,
+      "loss": 0.6448,
+      "step": 319
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.9607315234315385e-05,
+      "loss": 0.6582,
+      "step": 320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.960053834946864e-05,
+      "loss": 0.6662,
+      "step": 321
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.9593704076463864e-05,
+      "loss": 0.6545,
+      "step": 322
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.958681243531103e-05,
+      "loss": 0.6646,
+      "step": 323
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.957986344618809e-05,
+      "loss": 0.6441,
+      "step": 324
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.95728571294409e-05,
+      "loss": 0.6482,
+      "step": 325
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.956579350558315e-05,
+      "loss": 0.6576,
+      "step": 326
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.955867259529634e-05,
+      "loss": 0.6382,
+      "step": 327
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9551494419429695e-05,
+      "loss": 0.6675,
+      "step": 328
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.954425899900009e-05,
+      "loss": 0.6513,
+      "step": 329
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9536966355192016e-05,
+      "loss": 0.6678,
+      "step": 330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.952961650935752e-05,
+      "loss": 0.6629,
+      "step": 331
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.952220948301612e-05,
+      "loss": 0.6556,
+      "step": 332
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.951474529785473e-05,
+      "loss": 0.6703,
+      "step": 333
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.950722397572766e-05,
+      "loss": 0.6575,
+      "step": 334
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.949964553865648e-05,
+      "loss": 0.657,
+      "step": 335
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.949201000883001e-05,
+      "loss": 0.6697,
+      "step": 336
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.94843174086042e-05,
+      "loss": 0.6516,
+      "step": 337
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.947656776050212e-05,
+      "loss": 0.6687,
+      "step": 338
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.9468761087213864e-05,
+      "loss": 0.6333,
+      "step": 339
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.946089741159648e-05,
+      "loss": 0.6569,
+      "step": 340
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.945297675667393e-05,
+      "loss": 0.6445,
+      "step": 341
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.944499914563699e-05,
+      "loss": 0.6519,
+      "step": 342
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.943696460184321e-05,
+      "loss": 0.6241,
+      "step": 343
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9428873148816815e-05,
+      "loss": 0.6455,
+      "step": 344
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.942072481024867e-05,
+      "loss": 0.6616,
+      "step": 345
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9412519609996186e-05,
+      "loss": 0.6559,
+      "step": 346
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.940425757208325e-05,
+      "loss": 0.6371,
+      "step": 347
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9395938720700196e-05,
+      "loss": 0.6368,
+      "step": 348
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.938756308020366e-05,
+      "loss": 0.6517,
+      "step": 349
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.937913067511658e-05,
+      "loss": 0.6532,
+      "step": 350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.937064153012808e-05,
+      "loss": 0.6663,
+      "step": 351
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9362095670093416e-05,
+      "loss": 0.6593,
+      "step": 352
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.935349312003391e-05,
+      "loss": 0.6483,
+      "step": 353
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.934483390513683e-05,
+      "loss": 0.6531,
+      "step": 354
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.933611805075541e-05,
+      "loss": 0.6511,
+      "step": 355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9327345582408655e-05,
+      "loss": 0.6474,
+      "step": 356
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.931851652578137e-05,
+      "loss": 0.673,
+      "step": 357
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.930963090672403e-05,
+      "loss": 0.6629,
+      "step": 358
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.930068875125274e-05,
+      "loss": 0.6312,
+      "step": 359
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.929169008554911e-05,
+      "loss": 0.638,
+      "step": 360
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.92826349359602e-05,
+      "loss": 0.6588,
+      "step": 361
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.927352332899848e-05,
+      "loss": 0.6591,
+      "step": 362
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.926435529134169e-05,
+      "loss": 0.6238,
+      "step": 363
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.925513084983282e-05,
+      "loss": 0.6456,
+      "step": 364
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.924585003147999e-05,
+      "loss": 0.6723,
+      "step": 365
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.923651286345638e-05,
+      "loss": 0.6668,
+      "step": 366
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.9227119373100166e-05,
+      "loss": 0.6561,
+      "step": 367
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.921766958791442e-05,
+      "loss": 0.6536,
+      "step": 368
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.920816353556705e-05,
+      "loss": 0.6746,
+      "step": 369
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.919860124389068e-05,
+      "loss": 0.6599,
+      "step": 370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.918898274088265e-05,
+      "loss": 0.6446,
+      "step": 371
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9179308054704825e-05,
+      "loss": 0.6229,
+      "step": 372
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.91695772136836e-05,
+      "loss": 0.6437,
+      "step": 373
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.915979024630978e-05,
+      "loss": 0.6498,
+      "step": 374
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9149947181238486e-05,
+      "loss": 0.655,
+      "step": 375
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.914004804728912e-05,
+      "loss": 0.6529,
+      "step": 376
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.91300928734452e-05,
+      "loss": 0.6537,
+      "step": 377
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.912008168885436e-05,
+      "loss": 0.6325,
+      "step": 378
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.9110014522828215e-05,
+      "loss": 0.6536,
+      "step": 379
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.909989140484229e-05,
+      "loss": 0.6619,
+      "step": 380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.908971236453592e-05,
+      "loss": 0.6551,
+      "step": 381
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.90794774317122e-05,
+      "loss": 0.6493,
+      "step": 382
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.906918663633784e-05,
+      "loss": 0.6469,
+      "step": 383
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.9058840008543136e-05,
+      "loss": 0.6469,
+      "step": 384
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9048437578621825e-05,
+      "loss": 0.641,
+      "step": 385
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9037979377031065e-05,
+      "loss": 0.6413,
+      "step": 386
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9027465434391275e-05,
+      "loss": 0.6603,
+      "step": 387
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.90168957814861e-05,
+      "loss": 0.6469,
+      "step": 388
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.900627044926227e-05,
+      "loss": 0.6723,
+      "step": 389
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8995589468829565e-05,
+      "loss": 0.6673,
+      "step": 390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.898485287146068e-05,
+      "loss": 0.6401,
+      "step": 391
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.897406068859117e-05,
+      "loss": 0.6586,
+      "step": 392
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.896321295181932e-05,
+      "loss": 0.6443,
+      "step": 393
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8952309692906074e-05,
+      "loss": 0.6312,
+      "step": 394
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.894135094377494e-05,
+      "loss": 0.6526,
+      "step": 395
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.893033673651189e-05,
+      "loss": 0.6432,
+      "step": 396
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8919267103365296e-05,
+      "loss": 0.6621,
+      "step": 397
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.890814207674576e-05,
+      "loss": 0.6611,
+      "step": 398
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8896961689226125e-05,
+      "loss": 0.6416,
+      "step": 399
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.888572597354129e-05,
+      "loss": 0.6595,
+      "step": 400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8874434962588165e-05,
+      "loss": 0.6465,
+      "step": 401
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.886308868942555e-05,
+      "loss": 0.6484,
+      "step": 402
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.885168718727405e-05,
+      "loss": 0.6476,
+      "step": 403
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.884023048951597e-05,
+      "loss": 0.6322,
+      "step": 404
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.882871862969522e-05,
+      "loss": 0.6627,
+      "step": 405
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.881715164151723e-05,
+      "loss": 0.6519,
+      "step": 406
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.880552955884883e-05,
+      "loss": 0.6346,
+      "step": 407
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.879385241571817e-05,
+      "loss": 0.6543,
+      "step": 408
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8782120246314597e-05,
+      "loss": 0.65,
+      "step": 409
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.877033308498856e-05,
+      "loss": 0.644,
+      "step": 410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8758490966251565e-05,
+      "loss": 0.6325,
+      "step": 411
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.8746593924775976e-05,
+      "loss": 0.6643,
+      "step": 412
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.873464199539498e-05,
+      "loss": 0.6632,
+      "step": 413
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.872263521310248e-05,
+      "loss": 0.6505,
+      "step": 414
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.8710573613052985e-05,
+      "loss": 0.6308,
+      "step": 415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.869845723056147e-05,
+      "loss": 0.6644,
+      "step": 416
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.868628610110337e-05,
+      "loss": 0.6831,
+      "step": 417
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.867406026031433e-05,
+      "loss": 0.6403,
+      "step": 418
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.866177974399026e-05,
+      "loss": 0.621,
+      "step": 419
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.864944458808712e-05,
+      "loss": 0.6408,
+      "step": 420
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.863705482872084e-05,
+      "loss": 0.6491,
+      "step": 421
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.862461050216723e-05,
+      "loss": 0.6326,
+      "step": 422
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8612111644861876e-05,
+      "loss": 0.648,
+      "step": 423
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.859955829340002e-05,
+      "loss": 0.6488,
+      "step": 424
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.858695048453645e-05,
+      "loss": 0.6655,
+      "step": 425
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.857428825518541e-05,
+      "loss": 0.6545,
+      "step": 426
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.8561571642420464e-05,
+      "loss": 0.6239,
+      "step": 427
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.854880068347443e-05,
+      "loss": 0.6588,
+      "step": 428
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.853597541573921e-05,
+      "loss": 0.6629,
+      "step": 429
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.852309587676576e-05,
+      "loss": 0.6873,
+      "step": 430
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.851016210426389e-05,
+      "loss": 0.6568,
+      "step": 431
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.849717413610224e-05,
+      "loss": 0.6456,
+      "step": 432
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.84841320103081e-05,
+      "loss": 0.6527,
+      "step": 433
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.847103576506734e-05,
+      "loss": 0.6554,
+      "step": 434
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.845788543872429e-05,
+      "loss": 0.6478,
+      "step": 435
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.8444681069781606e-05,
+      "loss": 0.6434,
+      "step": 436
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.84314226969002e-05,
+      "loss": 0.6442,
+      "step": 437
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.841811035889908e-05,
+      "loss": 0.6421,
+      "step": 438
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.840474409475526e-05,
+      "loss": 0.6407,
+      "step": 439
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.839132394360365e-05,
+      "loss": 0.6468,
+      "step": 440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.837784994473695e-05,
+      "loss": 0.6566,
+      "step": 441
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.8364322137605484e-05,
+      "loss": 0.6506,
+      "step": 442
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.835074056181715e-05,
+      "loss": 0.6321,
+      "step": 443
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.8337105257137264e-05,
+      "loss": 0.6457,
+      "step": 444
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.8323416263488447e-05,
+      "loss": 0.6525,
+      "step": 445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.830967362095053e-05,
+      "loss": 0.6527,
+      "step": 446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8295877369760426e-05,
+      "loss": 0.6401,
+      "step": 447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.828202755031198e-05,
+      "loss": 0.6447,
+      "step": 448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.82681242031559e-05,
+      "loss": 0.6624,
+      "step": 449
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.825416736899961e-05,
+      "loss": 0.6322,
+      "step": 450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.824015708870715e-05,
+      "loss": 0.6597,
+      "step": 451
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8226093403299035e-05,
+      "loss": 0.6449,
+      "step": 452
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.821197635395213e-05,
+      "loss": 0.6458,
+      "step": 453
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.819780598199958e-05,
+      "loss": 0.6545,
+      "step": 454
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.818358232893061e-05,
+      "loss": 0.6412,
+      "step": 455
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8169305436390474e-05,
+      "loss": 0.6306,
+      "step": 456
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.815497534618031e-05,
+      "loss": 0.654,
+      "step": 457
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.814059210025697e-05,
+      "loss": 0.6424,
+      "step": 458
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.812615574073301e-05,
+      "loss": 0.6404,
+      "step": 459
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8111666309876416e-05,
+      "loss": 0.6452,
+      "step": 460
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.809712385011063e-05,
+      "loss": 0.6479,
+      "step": 461
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8082528404014316e-05,
+      "loss": 0.6516,
+      "step": 462
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.80678800143213e-05,
+      "loss": 0.6532,
+      "step": 463
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.80531787239204e-05,
+      "loss": 0.6356,
+      "step": 464
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8038424575855326e-05,
+      "loss": 0.6317,
+      "step": 465
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.802361761332456e-05,
+      "loss": 0.6457,
+      "step": 466
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.800875787968122e-05,
+      "loss": 0.6287,
+      "step": 467
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.799384541843291e-05,
+      "loss": 0.6686,
+      "step": 468
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.7978880273241636e-05,
+      "loss": 0.6484,
+      "step": 469
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.7963862487923646e-05,
+      "loss": 0.6605,
+      "step": 470
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.794879210644933e-05,
+      "loss": 0.6306,
+      "step": 471
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.7933669172943034e-05,
+      "loss": 0.665,
+      "step": 472
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.791849373168302e-05,
+      "loss": 0.6495,
+      "step": 473
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.790326582710125e-05,
+      "loss": 0.6523,
+      "step": 474
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.788798550378331e-05,
+      "loss": 0.6525,
+      "step": 475
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.787265280646825e-05,
+      "loss": 0.6471,
+      "step": 476
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.785726778004848e-05,
+      "loss": 0.6356,
+      "step": 477
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.784183046956961e-05,
+      "loss": 0.644,
+      "step": 478
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.7826340920230346e-05,
+      "loss": 0.6332,
+      "step": 479
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.781079917738233e-05,
+      "loss": 0.65,
+      "step": 480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7795205286530023e-05,
+      "loss": 0.646,
+      "step": 481
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.777955929333058e-05,
+      "loss": 0.6498,
+      "step": 482
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.77638612435937e-05,
+      "loss": 0.6611,
+      "step": 483
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.774811118328149e-05,
+      "loss": 0.6556,
+      "step": 484
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.773230915850835e-05,
+      "loss": 0.6546,
+      "step": 485
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7716455215540795e-05,
+      "loss": 0.6392,
+      "step": 486
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7700549400797414e-05,
+      "loss": 0.6313,
+      "step": 487
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.768459176084861e-05,
+      "loss": 0.6385,
+      "step": 488
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7668582342416557e-05,
+      "loss": 0.6522,
+      "step": 489
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.765252119237501e-05,
+      "loss": 0.6509,
+      "step": 490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.76364083577492e-05,
+      "loss": 0.6536,
+      "step": 491
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7620243885715695e-05,
+      "loss": 0.6316,
+      "step": 492
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.760402782360222e-05,
+      "loss": 0.6472,
+      "step": 493
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7587760218887595e-05,
+      "loss": 0.6603,
+      "step": 494
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7571441119201506e-05,
+      "loss": 0.6395,
+      "step": 495
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7555070572324435e-05,
+      "loss": 0.6495,
+      "step": 496
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7538648626187495e-05,
+      "loss": 0.6305,
+      "step": 497
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7522175328872274e-05,
+      "loss": 0.6356,
+      "step": 498
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.750565072861074e-05,
+      "loss": 0.6301,
+      "step": 499
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7489074873785035e-05,
+      "loss": 0.6358,
+      "step": 500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7472447812927395e-05,
+      "loss": 0.6287,
+      "step": 501
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.745576959471997e-05,
+      "loss": 0.629,
+      "step": 502
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.74390402679947e-05,
+      "loss": 0.6363,
+      "step": 503
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.742225988173315e-05,
+      "loss": 0.6408,
+      "step": 504
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.740542848506641e-05,
+      "loss": 0.6493,
+      "step": 505
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.738854612727489e-05,
+      "loss": 0.6374,
+      "step": 506
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7371612857788225e-05,
+      "loss": 0.6253,
+      "step": 507
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.735462872618511e-05,
+      "loss": 0.6482,
+      "step": 508
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7337593782193165e-05,
+      "loss": 0.6404,
+      "step": 509
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.732050807568878e-05,
+      "loss": 0.6393,
+      "step": 510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.730337165669695e-05,
+      "loss": 0.6692,
+      "step": 511
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.728618457539121e-05,
+      "loss": 0.6568,
+      "step": 512
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.726894688209335e-05,
+      "loss": 0.6539,
+      "step": 513
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.725165862727341e-05,
+      "loss": 0.6425,
+      "step": 514
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.723431986154943e-05,
+      "loss": 0.6369,
+      "step": 515
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.721693063568736e-05,
+      "loss": 0.617,
+      "step": 516
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.719949100060088e-05,
+      "loss": 0.6575,
+      "step": 517
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.718200100735127e-05,
+      "loss": 0.6432,
+      "step": 518
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7164460707147255e-05,
+      "loss": 0.6433,
+      "step": 519
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7146870151344845e-05,
+      "loss": 0.6348,
+      "step": 520
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7129229391447206e-05,
+      "loss": 0.6502,
+      "step": 521
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7111538479104464e-05,
+      "loss": 0.6486,
+      "step": 522
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.709379746611364e-05,
+      "loss": 0.6553,
+      "step": 523
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.707600640441837e-05,
+      "loss": 0.6376,
+      "step": 524
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.705816534610891e-05,
+      "loss": 0.6346,
+      "step": 525
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7040274343421845e-05,
+      "loss": 0.6414,
+      "step": 526
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.702233344873999e-05,
+      "loss": 0.6388,
+      "step": 527
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.700434271459229e-05,
+      "loss": 0.624,
+      "step": 528
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6986302193653545e-05,
+      "loss": 0.6692,
+      "step": 529
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.696821193874437e-05,
+      "loss": 0.6494,
+      "step": 530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.695007200283098e-05,
+      "loss": 0.6359,
+      "step": 531
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6931882439025066e-05,
+      "loss": 0.6439,
+      "step": 532
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.69136433005836e-05,
+      "loss": 0.625,
+      "step": 533
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.689535464090873e-05,
+      "loss": 0.6332,
+      "step": 534
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6877016513547575e-05,
+      "loss": 0.6151,
+      "step": 535
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6858628972192084e-05,
+      "loss": 0.6438,
+      "step": 536
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.684019207067891e-05,
+      "loss": 0.6268,
+      "step": 537
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.682170586298921e-05,
+      "loss": 0.6373,
+      "step": 538
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.680317040324851e-05,
+      "loss": 0.6322,
+      "step": 539
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.678458574572654e-05,
+      "loss": 0.6306,
+      "step": 540
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.676595194483706e-05,
+      "loss": 0.6242,
+      "step": 541
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6747269055137754e-05,
+      "loss": 0.6377,
+      "step": 542
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.672853713132999e-05,
+      "loss": 0.6512,
+      "step": 543
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6709756228258735e-05,
+      "loss": 0.6267,
+      "step": 544
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6690926400912325e-05,
+      "loss": 0.6383,
+      "step": 545
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.667204770442239e-05,
+      "loss": 0.6301,
+      "step": 546
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.665312019406361e-05,
+      "loss": 0.6306,
+      "step": 547
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.663414392525358e-05,
+      "loss": 0.6286,
+      "step": 548
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.661511895355268e-05,
+      "loss": 0.6457,
+      "step": 549
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.659604533466386e-05,
+      "loss": 0.6343,
+      "step": 550
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.657692312443253e-05,
+      "loss": 0.6415,
+      "step": 551
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.655775237884636e-05,
+      "loss": 0.6478,
+      "step": 552
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6538533154035116e-05,
+      "loss": 0.6385,
+      "step": 553
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6519265506270514e-05,
+      "loss": 0.6542,
+      "step": 554
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6499949491966046e-05,
+      "loss": 0.6365,
+      "step": 555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.648058516767682e-05,
+      "loss": 0.6679,
+      "step": 556
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6461172590099386e-05,
+      "loss": 0.6396,
+      "step": 557
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6441711816071566e-05,
+      "loss": 0.6149,
+      "step": 558
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.642220290257231e-05,
+      "loss": 0.6339,
+      "step": 559
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6402645906721515e-05,
+      "loss": 0.6427,
+      "step": 560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.638304088577984e-05,
+      "loss": 0.6515,
+      "step": 561
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.636338789714857e-05,
+      "loss": 0.6298,
+      "step": 562
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6343686998369435e-05,
+      "loss": 0.638,
+      "step": 563
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.632393824712444e-05,
+      "loss": 0.6153,
+      "step": 564
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.630414170123568e-05,
+      "loss": 0.6375,
+      "step": 565
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.628429741866522e-05,
+      "loss": 0.6235,
+      "step": 566
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6264405457514857e-05,
+      "loss": 0.6548,
+      "step": 567
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.624446587602602e-05,
+      "loss": 0.6533,
+      "step": 568
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6224478732579516e-05,
+      "loss": 0.6397,
+      "step": 569
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.620444408569547e-05,
+      "loss": 0.6445,
+      "step": 570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6184361994033054e-05,
+      "loss": 0.6272,
+      "step": 571
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6164232516390346e-05,
+      "loss": 0.6478,
+      "step": 572
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6144055711704196e-05,
+      "loss": 0.6471,
+      "step": 573
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.612383163904999e-05,
+      "loss": 0.6484,
+      "step": 574
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.610356035764153e-05,
+      "loss": 0.6479,
+      "step": 575
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.6083241926830836e-05,
+      "loss": 0.6379,
+      "step": 576
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.606287640610797e-05,
+      "loss": 0.6489,
+      "step": 577
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.604246385510088e-05,
+      "loss": 0.6306,
+      "step": 578
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.6022004333575194e-05,
+      "loss": 0.6421,
+      "step": 579
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.600149790143408e-05,
+      "loss": 0.6281,
+      "step": 580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.598094461871805e-05,
+      "loss": 0.6576,
+      "step": 581
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5960344545604796e-05,
+      "loss": 0.6203,
+      "step": 582
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5939697742408985e-05,
+      "loss": 0.6469,
+      "step": 583
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.591900426958214e-05,
+      "loss": 0.6567,
+      "step": 584
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.58982641877124e-05,
+      "loss": 0.6335,
+      "step": 585
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5877477557524364e-05,
+      "loss": 0.6379,
+      "step": 586
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.585664443987895e-05,
+      "loss": 0.6577,
+      "step": 587
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.583576489577317e-05,
+      "loss": 0.6512,
+      "step": 588
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.581483898633995e-05,
+      "loss": 0.641,
+      "step": 589
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5793866772847996e-05,
+      "loss": 0.6492,
+      "step": 590
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.577284831670157e-05,
+      "loss": 0.6322,
+      "step": 591
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.575178367944034e-05,
+      "loss": 0.6415,
+      "step": 592
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.573067292273918e-05,
+      "loss": 0.6515,
+      "step": 593
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.570951610840798e-05,
+      "loss": 0.6595,
+      "step": 594
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.568831329839152e-05,
+      "loss": 0.6143,
+      "step": 595
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5667064554769214e-05,
+      "loss": 0.6252,
+      "step": 596
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5645769939754986e-05,
+      "loss": 0.61,
+      "step": 597
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5624429515697047e-05,
+      "loss": 0.6238,
+      "step": 598
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.560304334507777e-05,
+      "loss": 0.6385,
+      "step": 599
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.558161149051341e-05,
+      "loss": 0.631,
+      "step": 600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.556013401475404e-05,
+      "loss": 0.6541,
+      "step": 601
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5538610980683247e-05,
+      "loss": 0.6599,
+      "step": 602
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.551704245131807e-05,
+      "loss": 0.627,
+      "step": 603
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5495428489808696e-05,
+      "loss": 0.6461,
+      "step": 604
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.547376915943838e-05,
+      "loss": 0.6349,
+      "step": 605
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.545206452362318e-05,
+      "loss": 0.6384,
+      "step": 606
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5430314645911815e-05,
+      "loss": 0.6422,
+      "step": 607
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.540851958998547e-05,
+      "loss": 0.6319,
+      "step": 608
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.538667941965758e-05,
+      "loss": 0.6306,
+      "step": 609
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.536479419887372e-05,
+      "loss": 0.6396,
+      "step": 610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.534286399171133e-05,
+      "loss": 0.6305,
+      "step": 611
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.532088886237956e-05,
+      "loss": 0.6329,
+      "step": 612
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.529886887521913e-05,
+      "loss": 0.6397,
+      "step": 613
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.527680409470205e-05,
+      "loss": 0.6389,
+      "step": 614
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5254694585431494e-05,
+      "loss": 0.6278,
+      "step": 615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.523254041214163e-05,
+      "loss": 0.6482,
+      "step": 616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.521034163969735e-05,
+      "loss": 0.6424,
+      "step": 617
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5188098333094145e-05,
+      "loss": 0.6427,
+      "step": 618
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.516581055745791e-05,
+      "loss": 0.6444,
+      "step": 619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.514347837804473e-05,
+      "loss": 0.6127,
+      "step": 620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.512110186024069e-05,
+      "loss": 0.6293,
+      "step": 621
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5098681069561705e-05,
+      "loss": 0.6581,
+      "step": 622
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5076216071653314e-05,
+      "loss": 0.6449,
+      "step": 623
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5053706932290496e-05,
+      "loss": 0.6481,
+      "step": 624
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.503115371737745e-05,
+      "loss": 0.6227,
+      "step": 625
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.500855649294744e-05,
+      "loss": 0.6325,
+      "step": 626
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.498591532516259e-05,
+      "loss": 0.6244,
+      "step": 627
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4963230280313675e-05,
+      "loss": 0.636,
+      "step": 628
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4940501424819927e-05,
+      "loss": 0.6272,
+      "step": 629
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.491772882522887e-05,
+      "loss": 0.6364,
+      "step": 630
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4894912548216105e-05,
+      "loss": 0.6319,
+      "step": 631
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.48720526605851e-05,
+      "loss": 0.637,
+      "step": 632
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.484914922926702e-05,
+      "loss": 0.6263,
+      "step": 633
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.482620232132053e-05,
+      "loss": 0.6507,
+      "step": 634
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4803212003931566e-05,
+      "loss": 0.6285,
+      "step": 635
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.478017834441319e-05,
+      "loss": 0.6472,
+      "step": 636
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4757101410205326e-05,
+      "loss": 0.6537,
+      "step": 637
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4733981268874656e-05,
+      "loss": 0.6205,
+      "step": 638
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4710817988114314e-05,
+      "loss": 0.6526,
+      "step": 639
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.468761163574377e-05,
+      "loss": 0.6462,
+      "step": 640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4664362279708605e-05,
+      "loss": 0.6462,
+      "step": 641
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4641069988080296e-05,
+      "loss": 0.6351,
+      "step": 642
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.461773482905604e-05,
+      "loss": 0.6245,
+      "step": 643
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.459435687095854e-05,
+      "loss": 0.6344,
+      "step": 644
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.457093618223581e-05,
+      "loss": 0.6505,
+      "step": 645
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4547472831460976e-05,
+      "loss": 0.6351,
+      "step": 646
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.452396688733208e-05,
+      "loss": 0.6255,
+      "step": 647
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.450041841867186e-05,
+      "loss": 0.6413,
+      "step": 648
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.447682749442757e-05,
+      "loss": 0.6479,
+      "step": 649
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4453194183670754e-05,
+      "loss": 0.6379,
+      "step": 650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.442951855559709e-05,
+      "loss": 0.627,
+      "step": 651
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.440580067952613e-05,
+      "loss": 0.6107,
+      "step": 652
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4382040624901125e-05,
+      "loss": 0.6333,
+      "step": 653
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.435823846128884e-05,
+      "loss": 0.6282,
+      "step": 654
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.433439425837931e-05,
+      "loss": 0.6202,
+      "step": 655
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.431050808598565e-05,
+      "loss": 0.6264,
+      "step": 656
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4286580014043884e-05,
+      "loss": 0.6564,
+      "step": 657
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.42626101126127e-05,
+      "loss": 0.6407,
+      "step": 658
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.423859845187324e-05,
+      "loss": 0.6402,
+      "step": 659
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.421454510212894e-05,
+      "loss": 0.634,
+      "step": 660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.41904501338053e-05,
+      "loss": 0.622,
+      "step": 661
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4166313617449635e-05,
+      "loss": 0.6433,
+      "step": 662
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4142135623730954e-05,
+      "loss": 0.6242,
+      "step": 663
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.411791622343967e-05,
+      "loss": 0.6373,
+      "step": 664
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.409365548748746e-05,
+      "loss": 0.6426,
+      "step": 665
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4069353486907e-05,
+      "loss": 0.6301,
+      "step": 666
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.404501029285181e-05,
+      "loss": 0.6341,
+      "step": 667
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.402062597659599e-05,
+      "loss": 0.6167,
+      "step": 668
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3996200609534075e-05,
+      "loss": 0.6307,
+      "step": 669
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3971734263180765e-05,
+      "loss": 0.6365,
+      "step": 670
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3947227009170764e-05,
+      "loss": 0.6164,
+      "step": 671
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.392267891925854e-05,
+      "loss": 0.6203,
+      "step": 672
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.38980900653181e-05,
+      "loss": 0.6403,
+      "step": 673
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3873460519342847e-05,
+      "loss": 0.6265,
+      "step": 674
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.384879035344529e-05,
+      "loss": 0.6084,
+      "step": 675
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.38240796398569e-05,
+      "loss": 0.63,
+      "step": 676
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.379932845092783e-05,
+      "loss": 0.6435,
+      "step": 677
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3774536859126774e-05,
+      "loss": 0.6384,
+      "step": 678
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3749704937040705e-05,
+      "loss": 0.6307,
+      "step": 679
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.372483275737468e-05,
+      "loss": 0.6399,
+      "step": 680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3699920392951615e-05,
+      "loss": 0.5265,
+      "step": 681
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.367496791671211e-05,
+      "loss": 0.5803,
+      "step": 682
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.364997540171418e-05,
+      "loss": 0.5753,
+      "step": 683
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.362494292113308e-05,
+      "loss": 0.5597,
+      "step": 684
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.35998705482611e-05,
+      "loss": 0.5715,
+      "step": 685
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3574758356507285e-05,
+      "loss": 0.5686,
+      "step": 686
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.35496064193973e-05,
+      "loss": 0.5782,
+      "step": 687
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.352441481057318e-05,
+      "loss": 0.5777,
+      "step": 688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3499183603793084e-05,
+      "loss": 0.575,
+      "step": 689
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.347391287293115e-05,
+      "loss": 0.5759,
+      "step": 690
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.34486026919772e-05,
+      "loss": 0.5677,
+      "step": 691
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.34232531350366e-05,
+      "loss": 0.5811,
+      "step": 692
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3397864276329966e-05,
+      "loss": 0.5666,
+      "step": 693
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.337243619019301e-05,
+      "loss": 0.5694,
+      "step": 694
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.334696895107629e-05,
+      "loss": 0.5649,
+      "step": 695
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.332146263354501e-05,
+      "loss": 0.5673,
+      "step": 696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3295917312278754e-05,
+      "loss": 0.5587,
+      "step": 697
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.327033306207136e-05,
+      "loss": 0.5635,
+      "step": 698
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.324470995783061e-05,
+      "loss": 0.5858,
+      "step": 699
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.321904807457804e-05,
+      "loss": 0.5621,
+      "step": 700
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.319334748744876e-05,
+      "loss": 0.5719,
+      "step": 701
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.3167608271691164e-05,
+      "loss": 0.5623,
+      "step": 702
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.314183050266676e-05,
+      "loss": 0.5899,
+      "step": 703
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.311601425584996e-05,
+      "loss": 0.5746,
+      "step": 704
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.3090159606827785e-05,
+      "loss": 0.5587,
+      "step": 705
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.306426663129974e-05,
+      "loss": 0.5885,
+      "step": 706
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.3038335405077536e-05,
+      "loss": 0.5666,
+      "step": 707
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.301236600408484e-05,
+      "loss": 0.5731,
+      "step": 708
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.298635850435715e-05,
+      "loss": 0.5707,
+      "step": 709
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.296031298204146e-05,
+      "loss": 0.5783,
+      "step": 710
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2934229513396115e-05,
+      "loss": 0.5673,
+      "step": 711
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.290810817479057e-05,
+      "loss": 0.5676,
+      "step": 712
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.288194904270513e-05,
+      "loss": 0.5638,
+      "step": 713
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.285575219373079e-05,
+      "loss": 0.5685,
+      "step": 714
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.282951770456896e-05,
+      "loss": 0.5935,
+      "step": 715
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.280324565203123e-05,
+      "loss": 0.5623,
+      "step": 716
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.277693611303922e-05,
+      "loss": 0.5668,
+      "step": 717
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.27505891646243e-05,
+      "loss": 0.5716,
+      "step": 718
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.27242048839273e-05,
+      "loss": 0.5721,
+      "step": 719
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.269778334819846e-05,
+      "loss": 0.5904,
+      "step": 720
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.267132463479702e-05,
+      "loss": 0.5718,
+      "step": 721
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2644828821191096e-05,
+      "loss": 0.5688,
+      "step": 722
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.261829598495744e-05,
+      "loss": 0.5534,
+      "step": 723
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2591726203781205e-05,
+      "loss": 0.564,
+      "step": 724
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.256511955545568e-05,
+      "loss": 0.578,
+      "step": 725
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.253847611788214e-05,
+      "loss": 0.5766,
+      "step": 726
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.251179596906953e-05,
+      "loss": 0.5902,
+      "step": 727
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.248507918713434e-05,
+      "loss": 0.5815,
+      "step": 728
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2458325850300254e-05,
+      "loss": 0.577,
+      "step": 729
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.243153603689803e-05,
+      "loss": 0.576,
+      "step": 730
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2404709825365204e-05,
+      "loss": 0.5849,
+      "step": 731
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.237784729424589e-05,
+      "loss": 0.5635,
+      "step": 732
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.235094852219055e-05,
+      "loss": 0.5833,
+      "step": 733
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.2324013587955716e-05,
+      "loss": 0.5889,
+      "step": 734
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.229704257040385e-05,
+      "loss": 0.5855,
+      "step": 735
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.227003554850304e-05,
+      "loss": 0.5814,
+      "step": 736
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.224299260132678e-05,
+      "loss": 0.5636,
+      "step": 737
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.221591380805376e-05,
+      "loss": 0.5794,
+      "step": 738
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.2188799247967625e-05,
+      "loss": 0.569,
+      "step": 739
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.2161649000456735e-05,
+      "loss": 0.5889,
+      "step": 740
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.2134463145013946e-05,
+      "loss": 0.5594,
+      "step": 741
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.210724176123636e-05,
+      "loss": 0.5504,
+      "step": 742
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.207998492882512e-05,
+      "loss": 0.5915,
+      "step": 743
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.205269272758513e-05,
+      "loss": 0.5853,
+      "step": 744
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.2025365237424884e-05,
+      "loss": 0.5787,
+      "step": 745
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1998002538356164e-05,
+      "loss": 0.5605,
+      "step": 746
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1970604710493865e-05,
+      "loss": 0.5657,
+      "step": 747
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.194317183405573e-05,
+      "loss": 0.5639,
+      "step": 748
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.19157039893621e-05,
+      "loss": 0.581,
+      "step": 749
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.188820125683574e-05,
+      "loss": 0.573,
+      "step": 750
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.186066371700152e-05,
+      "loss": 0.5586,
+      "step": 751
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.183309145048625e-05,
+      "loss": 0.5752,
+      "step": 752
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1805484538018396e-05,
+      "loss": 0.5892,
+      "step": 753
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1777843060427876e-05,
+      "loss": 0.5727,
+      "step": 754
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.175016709864582e-05,
+      "loss": 0.57,
+      "step": 755
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1722456733704296e-05,
+      "loss": 0.569,
+      "step": 756
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.169471204673612e-05,
+      "loss": 0.5819,
+      "step": 757
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.166693311897461e-05,
+      "loss": 0.5735,
+      "step": 758
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.163912003175331e-05,
+      "loss": 0.5725,
+      "step": 759
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1611272866505784e-05,
+      "loss": 0.5824,
+      "step": 760
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.15833917047654e-05,
+      "loss": 0.5793,
+      "step": 761
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.155547662816503e-05,
+      "loss": 0.57,
+      "step": 762
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.152752771843686e-05,
+      "loss": 0.5871,
+      "step": 763
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.149954505741214e-05,
+      "loss": 0.5846,
+      "step": 764
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.147152872702092e-05,
+      "loss": 0.5725,
+      "step": 765
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1443478809291874e-05,
+      "loss": 0.5662,
+      "step": 766
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.141539538635196e-05,
+      "loss": 0.5617,
+      "step": 767
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1387278540426275e-05,
+      "loss": 0.5565,
+      "step": 768
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1359128353837754e-05,
+      "loss": 0.5598,
+      "step": 769
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.133094490900697e-05,
+      "loss": 0.5735,
+      "step": 770
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.130272828845184e-05,
+      "loss": 0.5691,
+      "step": 771
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.127447857478745e-05,
+      "loss": 0.5869,
+      "step": 772
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1246195850725766e-05,
+      "loss": 0.5812,
+      "step": 773
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.12178801990754e-05,
+      "loss": 0.5986,
+      "step": 774
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.118953170274137e-05,
+      "loss": 0.591,
+      "step": 775
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.116115044472487e-05,
+      "loss": 0.5621,
+      "step": 776
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.113273650812301e-05,
+      "loss": 0.573,
+      "step": 777
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1104289976128584e-05,
+      "loss": 0.5829,
+      "step": 778
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1075810932029816e-05,
+      "loss": 0.5801,
+      "step": 779
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.104729945921012e-05,
+      "loss": 0.5711,
+      "step": 780
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.1018755641147866e-05,
+      "loss": 0.5764,
+      "step": 781
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0990179561416124e-05,
+      "loss": 0.5856,
+      "step": 782
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.096157130368242e-05,
+      "loss": 0.567,
+      "step": 783
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.093293095170849e-05,
+      "loss": 0.5697,
+      "step": 784
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.090425858935006e-05,
+      "loss": 0.5647,
+      "step": 785
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0875554300556537e-05,
+      "loss": 0.5681,
+      "step": 786
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0846818169370845e-05,
+      "loss": 0.5624,
+      "step": 787
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.081805027992912e-05,
+      "loss": 0.5832,
+      "step": 788
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0789250716460504e-05,
+      "loss": 0.5772,
+      "step": 789
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.076041956328682e-05,
+      "loss": 0.5775,
+      "step": 790
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.073155690482246e-05,
+      "loss": 0.5808,
+      "step": 791
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.070266282557401e-05,
+      "loss": 0.5858,
+      "step": 792
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.067373741014006e-05,
+      "loss": 0.5645,
+      "step": 793
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.064478074321095e-05,
+      "loss": 0.5877,
+      "step": 794
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.061579290956852e-05,
+      "loss": 0.5619,
+      "step": 795
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0586773994085886e-05,
+      "loss": 0.5862,
+      "step": 796
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.055772408172713e-05,
+      "loss": 0.5713,
+      "step": 797
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.052864325754712e-05,
+      "loss": 0.5562,
+      "step": 798
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0499531606691204e-05,
+      "loss": 0.5808,
+      "step": 799
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0470389214395015e-05,
+      "loss": 0.5688,
+      "step": 800
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0441216165984166e-05,
+      "loss": 0.5854,
+      "step": 801
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0412012546874062e-05,
+      "loss": 0.5812,
+      "step": 802
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0382778442569568e-05,
+      "loss": 0.5894,
+      "step": 803
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0353513938664855e-05,
+      "loss": 0.5887,
+      "step": 804
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0324219120843062e-05,
+      "loss": 0.574,
+      "step": 805
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.029489407487611e-05,
+      "loss": 0.568,
+      "step": 806
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0265538886624413e-05,
+      "loss": 0.5802,
+      "step": 807
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0236153642036644e-05,
+      "loss": 0.5822,
+      "step": 808
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0206738427149463e-05,
+      "loss": 0.5853,
+      "step": 809
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0177293328087303e-05,
+      "loss": 0.572,
+      "step": 810
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0147818431062076e-05,
+      "loss": 0.5856,
+      "step": 811
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.011831382237293e-05,
+      "loss": 0.5576,
+      "step": 812
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0088779588406045e-05,
+      "loss": 0.5729,
+      "step": 813
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.00592158156343e-05,
+      "loss": 0.5738,
+      "step": 814
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0029622590617085e-05,
+      "loss": 0.5671,
+      "step": 815
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.5601,
+      "step": 816
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9970348130514656e-05,
+      "loss": 0.5774,
+      "step": 817
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9940667068978358e-05,
+      "loss": 0.5648,
+      "step": 818
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.99109569022939e-05,
+      "loss": 0.5789,
+      "step": 819
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.988121771744928e-05,
+      "loss": 0.5837,
+      "step": 820
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.985144960151748e-05,
+      "loss": 0.5835,
+      "step": 821
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.982165264165617e-05,
+      "loss": 0.587,
+      "step": 822
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9791826925107467e-05,
+      "loss": 0.5815,
+      "step": 823
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.97619725391977e-05,
+      "loss": 0.5786,
+      "step": 824
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9732089571337126e-05,
+      "loss": 0.5672,
+      "step": 825
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9702178109019705e-05,
+      "loss": 0.5568,
+      "step": 826
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.967223823982281e-05,
+      "loss": 0.5737,
+      "step": 827
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9642270051406982e-05,
+      "loss": 0.5816,
+      "step": 828
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.961227363151569e-05,
+      "loss": 0.5705,
+      "step": 829
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9582249067975058e-05,
+      "loss": 0.5884,
+      "step": 830
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9552196448693603e-05,
+      "loss": 0.586,
+      "step": 831
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9522115861662003e-05,
+      "loss": 0.5617,
+      "step": 832
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9492007394952812e-05,
+      "loss": 0.5672,
+      "step": 833
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9461871136720205e-05,
+      "loss": 0.5708,
+      "step": 834
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9431707175199743e-05,
+      "loss": 0.5804,
+      "step": 835
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.940151559870808e-05,
+      "loss": 0.5624,
+      "step": 836
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9371296495642752e-05,
+      "loss": 0.5705,
+      "step": 837
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9341049954481858e-05,
+      "loss": 0.5811,
+      "step": 838
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9310776063783863e-05,
+      "loss": 0.568,
+      "step": 839
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.928047491218727e-05,
+      "loss": 0.569,
+      "step": 840
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9250146588410443e-05,
+      "loss": 0.5621,
+      "step": 841
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.921979118125127e-05,
+      "loss": 0.5493,
+      "step": 842
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9189408779586953e-05,
+      "loss": 0.5887,
+      "step": 843
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9158999472373727e-05,
+      "loss": 0.5734,
+      "step": 844
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9128563348646594e-05,
+      "loss": 0.5786,
+      "step": 845
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.909810049751909e-05,
+      "loss": 0.581,
+      "step": 846
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9067611008182988e-05,
+      "loss": 0.5714,
+      "step": 847
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9037094969908067e-05,
+      "loss": 0.5771,
+      "step": 848
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9006552472041825e-05,
+      "loss": 0.5714,
+      "step": 849
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.897598360400925e-05,
+      "loss": 0.5566,
+      "step": 850
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.8945388455312518e-05,
+      "loss": 0.5708,
+      "step": 851
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.891476711553077e-05,
+      "loss": 0.5685,
+      "step": 852
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.8884119674319818e-05,
+      "loss": 0.5836,
+      "step": 853
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.88534462214119e-05,
+      "loss": 0.5793,
+      "step": 854
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8822746846615425e-05,
+      "loss": 0.5781,
+      "step": 855
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.879202163981468e-05,
+      "loss": 0.5731,
+      "step": 856
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.87612706909696e-05,
+      "loss": 0.5881,
+      "step": 857
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8730494090115465e-05,
+      "loss": 0.5809,
+      "step": 858
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8699691927362704e-05,
+      "loss": 0.5838,
+      "step": 859
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8668864292896543e-05,
+      "loss": 0.5867,
+      "step": 860
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8638011276976825e-05,
+      "loss": 0.5723,
+      "step": 861
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8607132969937672e-05,
+      "loss": 0.5969,
+      "step": 862
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8576229462187277e-05,
+      "loss": 0.5783,
+      "step": 863
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.854530084420762e-05,
+      "loss": 0.5789,
+      "step": 864
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.851434720655419e-05,
+      "loss": 0.5706,
+      "step": 865
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8483368639855737e-05,
+      "loss": 0.5797,
+      "step": 866
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8452365234813992e-05,
+      "loss": 0.5687,
+      "step": 867
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8421337082203426e-05,
+      "loss": 0.5691,
+      "step": 868
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8390284272870943e-05,
+      "loss": 0.572,
+      "step": 869
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8359206897735673e-05,
+      "loss": 0.5679,
+      "step": 870
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8328105047788638e-05,
+      "loss": 0.5715,
+      "step": 871
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8296978814092548e-05,
+      "loss": 0.5706,
+      "step": 872
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8265828287781483e-05,
+      "loss": 0.5777,
+      "step": 873
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.823465356006066e-05,
+      "loss": 0.5798,
+      "step": 874
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8203454722206155e-05,
+      "loss": 0.5689,
+      "step": 875
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.817223186556463e-05,
+      "loss": 0.58,
+      "step": 876
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8140985081553087e-05,
+      "loss": 0.5724,
+      "step": 877
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.810971446165857e-05,
+      "loss": 0.5722,
+      "step": 878
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.80784200974379e-05,
+      "loss": 0.5935,
+      "step": 879
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.804710208051745e-05,
+      "loss": 0.5616,
+      "step": 880
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.801576050259283e-05,
+      "loss": 0.5678,
+      "step": 881
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.798439545542862e-05,
+      "loss": 0.5712,
+      "step": 882
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.795300703085814e-05,
+      "loss": 0.5615,
+      "step": 883
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.792159532078314e-05,
+      "loss": 0.5662,
+      "step": 884
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.7890160417173552e-05,
+      "loss": 0.5738,
+      "step": 885
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.7858702412067214e-05,
+      "loss": 0.5828,
+      "step": 886
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.7827221397569616e-05,
+      "loss": 0.5835,
+      "step": 887
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.7795717465853588e-05,
+      "loss": 0.5674,
+      "step": 888
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.776419070915909e-05,
+      "loss": 0.5823,
+      "step": 889
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.7732641219792898e-05,
+      "loss": 0.5708,
+      "step": 890
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.7701069090128334e-05,
+      "loss": 0.5738,
+      "step": 891
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.7669474412605038e-05,
+      "loss": 0.5785,
+      "step": 892
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.763785727972863e-05,
+      "loss": 0.5855,
+      "step": 893
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.7606217784070518e-05,
+      "loss": 0.5767,
+      "step": 894
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.757455601826755e-05,
+      "loss": 0.5854,
+      "step": 895
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7542872075021793e-05,
+      "loss": 0.5705,
+      "step": 896
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7511166047100255e-05,
+      "loss": 0.5685,
+      "step": 897
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7479438027334594e-05,
+      "loss": 0.569,
+      "step": 898
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7447688108620858e-05,
+      "loss": 0.56,
+      "step": 899
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7415916383919218e-05,
+      "loss": 0.5883,
+      "step": 900
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.738412294625369e-05,
+      "loss": 0.5696,
+      "step": 901
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.735230788871186e-05,
+      "loss": 0.5818,
+      "step": 902
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.732047130444462e-05,
+      "loss": 0.5806,
+      "step": 903
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7288613286665877e-05,
+      "loss": 0.5754,
+      "step": 904
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7256733928652307e-05,
+      "loss": 0.5849,
+      "step": 905
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7224833323743064e-05,
+      "loss": 0.5821,
+      "step": 906
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.71929115653395e-05,
+      "loss": 0.5694,
+      "step": 907
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7160968746904924e-05,
+      "loss": 0.5679,
+      "step": 908
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7129004961964285e-05,
+      "loss": 0.57,
+      "step": 909
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7097020304103927e-05,
+      "loss": 0.5752,
+      "step": 910
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.706501486697131e-05,
+      "loss": 0.5598,
+      "step": 911
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7032988744274736e-05,
+      "loss": 0.578,
+      "step": 912
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7000942029783063e-05,
+      "loss": 0.5761,
+      "step": 913
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.696887481732545e-05,
+      "loss": 0.5684,
+      "step": 914
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.693678720079105e-05,
+      "loss": 0.5699,
+      "step": 915
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.69046792741288e-05,
+      "loss": 0.5795,
+      "step": 916
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6872551131347048e-05,
+      "loss": 0.598,
+      "step": 917
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.684040286651338e-05,
+      "loss": 0.5663,
+      "step": 918
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6808234573754265e-05,
+      "loss": 0.5713,
+      "step": 919
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6776046347254826e-05,
+      "loss": 0.5587,
+      "step": 920
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.674383828125855e-05,
+      "loss": 0.5804,
+      "step": 921
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6711610470067007e-05,
+      "loss": 0.5537,
+      "step": 922
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.667936300803958e-05,
+      "loss": 0.5901,
+      "step": 923
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6647095989593194e-05,
+      "loss": 0.575,
+      "step": 924
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6614809509202028e-05,
+      "loss": 0.5613,
+      "step": 925
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.658250366139723e-05,
+      "loss": 0.5847,
+      "step": 926
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.655017854076668e-05,
+      "loss": 0.5818,
+      "step": 927
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6517834241954655e-05,
+      "loss": 0.5786,
+      "step": 928
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.648547085966163e-05,
+      "loss": 0.5684,
+      "step": 929
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.64530884886439e-05,
+      "loss": 0.5844,
+      "step": 930
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6420687223713396e-05,
+      "loss": 0.5701,
+      "step": 931
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6388267159737345e-05,
+      "loss": 0.5715,
+      "step": 932
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6355828391638036e-05,
+      "loss": 0.5726,
+      "step": 933
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6323371014392503e-05,
+      "loss": 0.5725,
+      "step": 934
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6290895123032277e-05,
+      "loss": 0.5789,
+      "step": 935
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.625840081264309e-05,
+      "loss": 0.5787,
+      "step": 936
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6225888178364606e-05,
+      "loss": 0.562,
+      "step": 937
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6193357315390144e-05,
+      "loss": 0.574,
+      "step": 938
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6160808318966378e-05,
+      "loss": 0.5556,
+      "step": 939
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6128241284393098e-05,
+      "loss": 0.5723,
+      "step": 940
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.609565630702289e-05,
+      "loss": 0.5829,
+      "step": 941
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.606305348226087e-05,
+      "loss": 0.5757,
+      "step": 942
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.603043290556444e-05,
+      "loss": 0.5744,
+      "step": 943
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.599779467244294e-05,
+      "loss": 0.5827,
+      "step": 944
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.5965138878457426e-05,
+      "loss": 0.5764,
+      "step": 945
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.5932465619220377e-05,
+      "loss": 0.5847,
+      "step": 946
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.589977499039539e-05,
+      "loss": 0.5714,
+      "step": 947
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.5867067087696924e-05,
+      "loss": 0.5829,
+      "step": 948
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.5834342006890026e-05,
+      "loss": 0.5693,
+      "step": 949
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5801599843790008e-05,
+      "loss": 0.5836,
+      "step": 950
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5768840694262247e-05,
+      "loss": 0.5722,
+      "step": 951
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5736064654221808e-05,
+      "loss": 0.5663,
+      "step": 952
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.570327181963323e-05,
+      "loss": 0.5625,
+      "step": 953
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.567046228651022e-05,
+      "loss": 0.5684,
+      "step": 954
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5637636150915378e-05,
+      "loss": 0.5826,
+      "step": 955
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.5604793508959916e-05,
+      "loss": 0.5571,
+      "step": 956
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5571934456803376e-05,
+      "loss": 0.5669,
+      "step": 957
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5539059090653337e-05,
+      "loss": 0.5666,
+      "step": 958
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.550616750676515e-05,
+      "loss": 0.5591,
+      "step": 959
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5473259801441663e-05,
+      "loss": 0.5799,
+      "step": 960
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5440336071032894e-05,
+      "loss": 0.5782,
+      "step": 961
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.5407396411935815e-05,
+      "loss": 0.5754,
+      "step": 962
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.537444092059401e-05,
+      "loss": 0.5682,
+      "step": 963
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5341469693497434e-05,
+      "loss": 0.5783,
+      "step": 964
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5308482827182108e-05,
+      "loss": 0.5723,
+      "step": 965
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5275480418229837e-05,
+      "loss": 0.5731,
+      "step": 966
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.524246256326795e-05,
+      "loss": 0.5708,
+      "step": 967
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.520942935896898e-05,
+      "loss": 0.5696,
+      "step": 968
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.5176380902050418e-05,
+      "loss": 0.571,
+      "step": 969
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.51433172892744e-05,
+      "loss": 0.593,
+      "step": 970
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5110238617447448e-05,
+      "loss": 0.5817,
+      "step": 971
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5077144983420157e-05,
+      "loss": 0.5606,
+      "step": 972
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.504403648408696e-05,
+      "loss": 0.5754,
+      "step": 973
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.5010913216385778e-05,
+      "loss": 0.5675,
+      "step": 974
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.4977775277297816e-05,
+      "loss": 0.5531,
+      "step": 975
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.494462276384718e-05,
+      "loss": 0.5739,
+      "step": 976
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.4911455773100693e-05,
+      "loss": 0.5727,
+      "step": 977
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.4878274402167544e-05,
+      "loss": 0.5738,
+      "step": 978
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.4845078748199046e-05,
+      "loss": 0.5747,
+      "step": 979
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.481186890838829e-05,
+      "loss": 0.5708,
+      "step": 980
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.477864497996995e-05,
+      "loss": 0.567,
+      "step": 981
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.4745407060219934e-05,
+      "loss": 0.5879,
+      "step": 982
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.4712155246455085e-05,
+      "loss": 0.5856,
+      "step": 983
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.4678889636032976e-05,
+      "loss": 0.5938,
+      "step": 984
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.4645610326351522e-05,
+      "loss": 0.5846,
+      "step": 985
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.4612317414848804e-05,
+      "loss": 0.5819,
+      "step": 986
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.4579010999002683e-05,
+      "loss": 0.5822,
+      "step": 987
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.454569117633058e-05,
+      "loss": 0.5841,
+      "step": 988
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.4512358044389156e-05,
+      "loss": 0.5635,
+      "step": 989
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.4479011700774065e-05,
+      "loss": 0.5798,
+      "step": 990
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.4445652243119612e-05,
+      "loss": 0.5833,
+      "step": 991
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.441227976909852e-05,
+      "loss": 0.5631,
+      "step": 992
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.4378894376421616e-05,
+      "loss": 0.5698,
+      "step": 993
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.434549616283755e-05,
+      "loss": 0.5762,
+      "step": 994
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.4312085226132518e-05,
+      "loss": 0.5602,
+      "step": 995
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.427866166412995e-05,
+      "loss": 0.5787,
+      "step": 996
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.4245225574690263e-05,
+      "loss": 0.5791,
+      "step": 997
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.421177705571054e-05,
+      "loss": 0.5566,
+      "step": 998
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.4178316205124253e-05,
+      "loss": 0.569,
+      "step": 999
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.4144843120900994e-05,
+      "loss": 0.5687,
+      "step": 1000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.4111357901046168e-05,
+      "loss": 0.5651,
+      "step": 1001
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.407786064360069e-05,
+      "loss": 0.5602,
+      "step": 1002
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.4044351446640763e-05,
+      "loss": 0.5763,
+      "step": 1003
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.4010830408277507e-05,
+      "loss": 0.5911,
+      "step": 1004
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.3977297626656736e-05,
+      "loss": 0.5878,
+      "step": 1005
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.394375319995864e-05,
+      "loss": 0.5585,
+      "step": 1006
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.3910197226397492e-05,
+      "loss": 0.5689,
+      "step": 1007
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.3876629804221402e-05,
+      "loss": 0.5663,
+      "step": 1008
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.3843051031711973e-05,
+      "loss": 0.5781,
+      "step": 1009
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.380946100718406e-05,
+      "loss": 0.5785,
+      "step": 1010
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.3775859828985438e-05,
+      "loss": 0.5715,
+      "step": 1011
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.374224759549657e-05,
+      "loss": 0.5585,
+      "step": 1012
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.370862440513026e-05,
+      "loss": 0.5689,
+      "step": 1013
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.367499035633141e-05,
+      "loss": 0.5669,
+      "step": 1014
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.3641345547576707e-05,
+      "loss": 0.565,
+      "step": 1015
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.3607690077374342e-05,
+      "loss": 0.5659,
+      "step": 1016
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.3574024044263727e-05,
+      "loss": 0.565,
+      "step": 1017
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.3540347546815188e-05,
+      "loss": 0.5581,
+      "step": 1018
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.3506660683629714e-05,
+      "loss": 0.5592,
+      "step": 1019
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.3472963553338614e-05,
+      "loss": 0.573,
+      "step": 1020
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.3439256254603268e-05,
+      "loss": 0.5811,
+      "step": 1021
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.340553888611485e-05,
+      "loss": 0.5913,
+      "step": 1022
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.3371811546593985e-05,
+      "loss": 0.5566,
+      "step": 1023
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.3338074334790513e-05,
+      "loss": 0.5658,
+      "step": 1024
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.3304327349483172e-05,
+      "loss": 0.5811,
+      "step": 1025
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.32705706894793e-05,
+      "loss": 0.5712,
+      "step": 1026
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.3236804453614602e-05,
+      "loss": 0.5699,
+      "step": 1027
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.320302874075278e-05,
+      "loss": 0.5873,
+      "step": 1028
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.3169243649785306e-05,
+      "loss": 0.5753,
+      "step": 1029
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.313544927963109e-05,
+      "loss": 0.5767,
+      "step": 1030
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.3101645729236237e-05,
+      "loss": 0.5915,
+      "step": 1031
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.3067833097573713e-05,
+      "loss": 0.5698,
+      "step": 1032
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.3034011483643074e-05,
+      "loss": 0.5798,
+      "step": 1033
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.3000180986470175e-05,
+      "loss": 0.5762,
+      "step": 1034
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.2966341705106895e-05,
+      "loss": 0.5785,
+      "step": 1035
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.2932493738630814e-05,
+      "loss": 0.5822,
+      "step": 1036
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.2898637186144935e-05,
+      "loss": 0.5678,
+      "step": 1037
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2864772146777435e-05,
+      "loss": 0.5765,
+      "step": 1038
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.28308987196813e-05,
+      "loss": 0.5632,
+      "step": 1039
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2797017004034095e-05,
+      "loss": 0.5632,
+      "step": 1040
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2763127099037646e-05,
+      "loss": 0.5406,
+      "step": 1041
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.272922910391776e-05,
+      "loss": 0.5735,
+      "step": 1042
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2695323117923946e-05,
+      "loss": 0.5874,
+      "step": 1043
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.266140924032908e-05,
+      "loss": 0.5554,
+      "step": 1044
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.262748757042915e-05,
+      "loss": 0.5683,
+      "step": 1045
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.259355820754298e-05,
+      "loss": 0.5626,
+      "step": 1046
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.25596212510119e-05,
+      "loss": 0.565,
+      "step": 1047
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.2525676800199467e-05,
+      "loss": 0.5672,
+      "step": 1048
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.2491724954491215e-05,
+      "loss": 0.5613,
+      "step": 1049
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.2457765813294285e-05,
+      "loss": 0.5649,
+      "step": 1050
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.242379947603721e-05,
+      "loss": 0.5683,
+      "step": 1051
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.238982604216958e-05,
+      "loss": 0.5733,
+      "step": 1052
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.2355845611161756e-05,
+      "loss": 0.5716,
+      "step": 1053
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.2321858282504606e-05,
+      "loss": 0.5511,
+      "step": 1054
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.2287864155709178e-05,
+      "loss": 0.573,
+      "step": 1055
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.2253863330306426e-05,
+      "loss": 0.5629,
+      "step": 1056
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.2219855905846918e-05,
+      "loss": 0.5837,
+      "step": 1057
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.218584198190055e-05,
+      "loss": 0.569,
+      "step": 1058
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2151821658056232e-05,
+      "loss": 0.5448,
+      "step": 1059
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2117795033921633e-05,
+      "loss": 0.5764,
+      "step": 1060
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2083762209122846e-05,
+      "loss": 0.5581,
+      "step": 1061
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.204972328330414e-05,
+      "loss": 0.5683,
+      "step": 1062
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.2015678356127633e-05,
+      "loss": 0.5674,
+      "step": 1063
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.1981627527273023e-05,
+      "loss": 0.5544,
+      "step": 1064
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1947570896437277e-05,
+      "loss": 0.5746,
+      "step": 1065
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1913508563334364e-05,
+      "loss": 0.576,
+      "step": 1066
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1879440627694926e-05,
+      "loss": 0.576,
+      "step": 1067
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.184536718926604e-05,
+      "loss": 0.5586,
+      "step": 1068
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1811288347810874e-05,
+      "loss": 0.5503,
+      "step": 1069
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1777204203108404e-05,
+      "loss": 0.5694,
+      "step": 1070
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.174311485495317e-05,
+      "loss": 0.5626,
+      "step": 1071
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1709020403154902e-05,
+      "loss": 0.5554,
+      "step": 1072
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.167492094753832e-05,
+      "loss": 0.5783,
+      "step": 1073
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.164081658794276e-05,
+      "loss": 0.5766,
+      "step": 1074
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1606707424221917e-05,
+      "loss": 0.5596,
+      "step": 1075
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1572593556243573e-05,
+      "loss": 0.5657,
+      "step": 1076
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1538475083889278e-05,
+      "loss": 0.5786,
+      "step": 1077
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.150435210705405e-05,
+      "loss": 0.587,
+      "step": 1078
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1470224725646105e-05,
+      "loss": 0.5757,
+      "step": 1079
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1436093039586553e-05,
+      "loss": 0.5583,
+      "step": 1080
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1401957148809105e-05,
+      "loss": 0.5779,
+      "step": 1081
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.13678171532598e-05,
+      "loss": 0.559,
+      "step": 1082
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1333673152896655e-05,
+      "loss": 0.5743,
+      "step": 1083
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1299525247689466e-05,
+      "loss": 0.5601,
+      "step": 1084
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.126537353761942e-05,
+      "loss": 0.5755,
+      "step": 1085
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1231218122678862e-05,
+      "loss": 0.5579,
+      "step": 1086
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.119705910287098e-05,
+      "loss": 0.5477,
+      "step": 1087
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1162896578209517e-05,
+      "loss": 0.5776,
+      "step": 1088
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.112873064871849e-05,
+      "loss": 0.5665,
+      "step": 1089
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.109456141443186e-05,
+      "loss": 0.5868,
+      "step": 1090
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1060388975393282e-05,
+      "loss": 0.5629,
+      "step": 1091
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1026213431655794e-05,
+      "loss": 0.5677,
+      "step": 1092
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.0992034883281524e-05,
+      "loss": 0.5556,
+      "step": 1093
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.0957853430341373e-05,
+      "loss": 0.5679,
+      "step": 1094
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.0923669172914796e-05,
+      "loss": 0.5679,
+      "step": 1095
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.088948221108941e-05,
+      "loss": 0.5542,
+      "step": 1096
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.0855292644960774e-05,
+      "loss": 0.578,
+      "step": 1097
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.082110057463207e-05,
+      "loss": 0.5763,
+      "step": 1098
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.0786906100213812e-05,
+      "loss": 0.5654,
+      "step": 1099
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.0752709321823545e-05,
+      "loss": 0.5688,
+      "step": 1100
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.071851033958557e-05,
+      "loss": 0.5724,
+      "step": 1101
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.0684309253630642e-05,
+      "loss": 0.5773,
+      "step": 1102
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.0650106164095663e-05,
+      "loss": 0.5767,
+      "step": 1103
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.061590117112341e-05,
+      "loss": 0.5549,
+      "step": 1104
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.058169437486223e-05,
+      "loss": 0.5541,
+      "step": 1105
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0547485875465755e-05,
+      "loss": 0.5685,
+      "step": 1106
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.05132757730926e-05,
+      "loss": 0.5612,
+      "step": 1107
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0479064167906064e-05,
+      "loss": 0.5705,
+      "step": 1108
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0444851160073863e-05,
+      "loss": 0.5607,
+      "step": 1109
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0410636849767813e-05,
+      "loss": 0.5805,
+      "step": 1110
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0376421337163536e-05,
+      "loss": 0.5591,
+      "step": 1111
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.0342204722440185e-05,
+      "loss": 0.5583,
+      "step": 1112
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.0307987105780138e-05,
+      "loss": 0.5839,
+      "step": 1113
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.02737685873687e-05,
+      "loss": 0.5735,
+      "step": 1114
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.0239549267393826e-05,
+      "loss": 0.5677,
+      "step": 1115
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.02053292460458e-05,
+      "loss": 0.5794,
+      "step": 1116
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.0171108623516984e-05,
+      "loss": 0.5555,
+      "step": 1117
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.013688750000148e-05,
+      "loss": 0.5671,
+      "step": 1118
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.0102665975694874e-05,
+      "loss": 0.5691,
+      "step": 1119
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.0068444150793907e-05,
+      "loss": 0.5753,
+      "step": 1120
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.0034222125496207e-05,
+      "loss": 0.5606,
+      "step": 1121
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2e-05,
+      "loss": 0.5639,
+      "step": 1122
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.99657778745038e-05,
+      "loss": 0.5684,
+      "step": 1123
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9931555849206106e-05,
+      "loss": 0.5676,
+      "step": 1124
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9897334024305132e-05,
+      "loss": 0.5631,
+      "step": 1125
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9863112499998523e-05,
+      "loss": 0.5597,
+      "step": 1126
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9828891376483022e-05,
+      "loss": 0.5737,
+      "step": 1127
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9794670753954208e-05,
+      "loss": 0.5591,
+      "step": 1128
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.976045073260618e-05,
+      "loss": 0.5538,
+      "step": 1129
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9726231412631303e-05,
+      "loss": 0.5555,
+      "step": 1130
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.969201289421987e-05,
+      "loss": 0.5644,
+      "step": 1131
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9657795277559818e-05,
+      "loss": 0.5562,
+      "step": 1132
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9623578662836464e-05,
+      "loss": 0.5468,
+      "step": 1133
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9589363150232194e-05,
+      "loss": 0.5625,
+      "step": 1134
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9555148839926144e-05,
+      "loss": 0.5665,
+      "step": 1135
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9520935832093943e-05,
+      "loss": 0.578,
+      "step": 1136
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.948672422690741e-05,
+      "loss": 0.566,
+      "step": 1137
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.945251412453425e-05,
+      "loss": 0.5675,
+      "step": 1138
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9418305625137773e-05,
+      "loss": 0.5746,
+      "step": 1139
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9384098828876597e-05,
+      "loss": 0.5616,
+      "step": 1140
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9349893835904344e-05,
+      "loss": 0.5723,
+      "step": 1141
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.931569074636936e-05,
+      "loss": 0.5726,
+      "step": 1142
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9281489660414433e-05,
+      "loss": 0.5738,
+      "step": 1143
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.924729067817646e-05,
+      "loss": 0.5683,
+      "step": 1144
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9213093899786198e-05,
+      "loss": 0.5678,
+      "step": 1145
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.917889942536794e-05,
+      "loss": 0.5614,
+      "step": 1146
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9144707355039232e-05,
+      "loss": 0.5668,
+      "step": 1147
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9110517788910596e-05,
+      "loss": 0.5651,
+      "step": 1148
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9076330827085214e-05,
+      "loss": 0.5779,
+      "step": 1149
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.904214656965863e-05,
+      "loss": 0.5571,
+      "step": 1150
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9007965116718483e-05,
+      "loss": 0.5609,
+      "step": 1151
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8973786568344206e-05,
+      "loss": 0.5717,
+      "step": 1152
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8939611024606725e-05,
+      "loss": 0.567,
+      "step": 1153
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8905438585568148e-05,
+      "loss": 0.565,
+      "step": 1154
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8871269351281513e-05,
+      "loss": 0.5718,
+      "step": 1155
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8837103421790486e-05,
+      "loss": 0.5559,
+      "step": 1156
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8802940897129026e-05,
+      "loss": 0.5523,
+      "step": 1157
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8768781877321145e-05,
+      "loss": 0.5581,
+      "step": 1158
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.873462646238059e-05,
+      "loss": 0.557,
+      "step": 1159
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8700474752310537e-05,
+      "loss": 0.5594,
+      "step": 1160
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8666326847103345e-05,
+      "loss": 0.5765,
+      "step": 1161
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.863218284674021e-05,
+      "loss": 0.5668,
+      "step": 1162
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.85980428511909e-05,
+      "loss": 0.5653,
+      "step": 1163
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.856390696041345e-05,
+      "loss": 0.5636,
+      "step": 1164
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8529775274353902e-05,
+      "loss": 0.5589,
+      "step": 1165
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8495647892945956e-05,
+      "loss": 0.5594,
+      "step": 1166
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8461524916110725e-05,
+      "loss": 0.5689,
+      "step": 1167
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8427406443756433e-05,
+      "loss": 0.5669,
+      "step": 1168
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8393292575778087e-05,
+      "loss": 0.588,
+      "step": 1169
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.835918341205725e-05,
+      "loss": 0.5749,
+      "step": 1170
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8325079052461685e-05,
+      "loss": 0.582,
+      "step": 1171
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.82909795968451e-05,
+      "loss": 0.5894,
+      "step": 1172
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8256885145046837e-05,
+      "loss": 0.5637,
+      "step": 1173
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8222795796891596e-05,
+      "loss": 0.5618,
+      "step": 1174
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8188711652189136e-05,
+      "loss": 0.5606,
+      "step": 1175
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.815463281073396e-05,
+      "loss": 0.5672,
+      "step": 1176
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.812055937230507e-05,
+      "loss": 0.5654,
+      "step": 1177
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8086491436665643e-05,
+      "loss": 0.5594,
+      "step": 1178
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8052429103562727e-05,
+      "loss": 0.5596,
+      "step": 1179
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.801837247272698e-05,
+      "loss": 0.5553,
+      "step": 1180
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7984321643872373e-05,
+      "loss": 0.5601,
+      "step": 1181
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7950276716695862e-05,
+      "loss": 0.5781,
+      "step": 1182
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7916237790877157e-05,
+      "loss": 0.5772,
+      "step": 1183
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7882204966078377e-05,
+      "loss": 0.5711,
+      "step": 1184
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7848178341943775e-05,
+      "loss": 0.5602,
+      "step": 1185
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7814158018099456e-05,
+      "loss": 0.56,
+      "step": 1186
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.7780144094153092e-05,
+      "loss": 0.5656,
+      "step": 1187
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.774613666969358e-05,
+      "loss": 0.5521,
+      "step": 1188
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.771213584429083e-05,
+      "loss": 0.5607,
+      "step": 1189
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7678141717495394e-05,
+      "loss": 0.5517,
+      "step": 1190
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7644154388838247e-05,
+      "loss": 0.5482,
+      "step": 1191
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.761017395783043e-05,
+      "loss": 0.5652,
+      "step": 1192
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7576200523962793e-05,
+      "loss": 0.5743,
+      "step": 1193
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.7542234186705722e-05,
+      "loss": 0.577,
+      "step": 1194
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.750827504550879e-05,
+      "loss": 0.5657,
+      "step": 1195
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7474323199800533e-05,
+      "loss": 0.5717,
+      "step": 1196
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.744037874898811e-05,
+      "loss": 0.5743,
+      "step": 1197
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.740644179245703e-05,
+      "loss": 0.5623,
+      "step": 1198
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7372512429570855e-05,
+      "loss": 0.5743,
+      "step": 1199
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.7338590759670934e-05,
+      "loss": 0.5638,
+      "step": 1200
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.730467688207606e-05,
+      "loss": 0.55,
+      "step": 1201
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.727077089608224e-05,
+      "loss": 0.5847,
+      "step": 1202
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7236872900962364e-05,
+      "loss": 0.5682,
+      "step": 1203
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7202982995965912e-05,
+      "loss": 0.5685,
+      "step": 1204
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7169101280318707e-05,
+      "loss": 0.5573,
+      "step": 1205
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7135227853222575e-05,
+      "loss": 0.562,
+      "step": 1206
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7101362813855068e-05,
+      "loss": 0.5707,
+      "step": 1207
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.7067506261369192e-05,
+      "loss": 0.5294,
+      "step": 1208
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.703365829489311e-05,
+      "loss": 0.564,
+      "step": 1209
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6999819013529828e-05,
+      "loss": 0.5529,
+      "step": 1210
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6965988516356936e-05,
+      "loss": 0.5681,
+      "step": 1211
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.693216690242629e-05,
+      "loss": 0.5544,
+      "step": 1212
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6898354270763773e-05,
+      "loss": 0.5526,
+      "step": 1213
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6864550720368915e-05,
+      "loss": 0.5632,
+      "step": 1214
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6830756350214704e-05,
+      "loss": 0.5646,
+      "step": 1215
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6796971259247226e-05,
+      "loss": 0.5623,
+      "step": 1216
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.67631955463854e-05,
+      "loss": 0.5747,
+      "step": 1217
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.67294293105207e-05,
+      "loss": 0.5799,
+      "step": 1218
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.669567265051684e-05,
+      "loss": 0.5637,
+      "step": 1219
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6661925665209494e-05,
+      "loss": 0.5625,
+      "step": 1220
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.6628188453406015e-05,
+      "loss": 0.5694,
+      "step": 1221
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.659446111388516e-05,
+      "loss": 0.5698,
+      "step": 1222
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6560743745396735e-05,
+      "loss": 0.578,
+      "step": 1223
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6527036446661396e-05,
+      "loss": 0.5876,
+      "step": 1224
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.64933393163703e-05,
+      "loss": 0.5668,
+      "step": 1225
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.645965245318482e-05,
+      "loss": 0.563,
+      "step": 1226
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.6425975955736277e-05,
+      "loss": 0.5852,
+      "step": 1227
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.639230992262566e-05,
+      "loss": 0.5595,
+      "step": 1228
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6358654452423303e-05,
+      "loss": 0.5502,
+      "step": 1229
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6325009643668592e-05,
+      "loss": 0.5449,
+      "step": 1230
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6291375594869744e-05,
+      "loss": 0.573,
+      "step": 1231
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6257752404503438e-05,
+      "loss": 0.5582,
+      "step": 1232
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6224140171014565e-05,
+      "loss": 0.5559,
+      "step": 1233
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6190538992815945e-05,
+      "loss": 0.5744,
+      "step": 1234
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.6156948968288034e-05,
+      "loss": 0.5707,
+      "step": 1235
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.61233701957786e-05,
+      "loss": 0.5505,
+      "step": 1236
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.608980277360251e-05,
+      "loss": 0.5631,
+      "step": 1237
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.605624680004137e-05,
+      "loss": 0.552,
+      "step": 1238
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.6022702373343274e-05,
+      "loss": 0.5769,
+      "step": 1239
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.5989169591722497e-05,
+      "loss": 0.5617,
+      "step": 1240
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.5955648553359247e-05,
+      "loss": 0.5521,
+      "step": 1241
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.5922139356399315e-05,
+      "loss": 0.5663,
+      "step": 1242
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.588864209895384e-05,
+      "loss": 0.5684,
+      "step": 1243
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.5855156879099012e-05,
+      "loss": 0.5655,
+      "step": 1244
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.582168379487575e-05,
+      "loss": 0.5437,
+      "step": 1245
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.5788222944289466e-05,
+      "loss": 0.5695,
+      "step": 1246
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.5754774425309736e-05,
+      "loss": 0.5518,
+      "step": 1247
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.5721338335870057e-05,
+      "loss": 0.565,
+      "step": 1248
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.568791477386749e-05,
+      "loss": 0.5602,
+      "step": 1249
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.565450383716245e-05,
+      "loss": 0.565,
+      "step": 1250
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.562110562357839e-05,
+      "loss": 0.5729,
+      "step": 1251
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5587720230901488e-05,
+      "loss": 0.5709,
+      "step": 1252
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.555434775688039e-05,
+      "loss": 0.5666,
+      "step": 1253
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5520988299225942e-05,
+      "loss": 0.5596,
+      "step": 1254
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.5487641955610848e-05,
+      "loss": 0.5538,
+      "step": 1255
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5454308823669424e-05,
+      "loss": 0.5764,
+      "step": 1256
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5420989000997324e-05,
+      "loss": 0.5786,
+      "step": 1257
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.53876825851512e-05,
+      "loss": 0.5577,
+      "step": 1258
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5354389673648474e-05,
+      "loss": 0.5546,
+      "step": 1259
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5321110363967034e-05,
+      "loss": 0.5666,
+      "step": 1260
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5287844753544918e-05,
+      "loss": 0.5632,
+      "step": 1261
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.5254592939780071e-05,
+      "loss": 0.5741,
+      "step": 1262
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.522135502003005e-05,
+      "loss": 0.5601,
+      "step": 1263
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5188131091611715e-05,
+      "loss": 0.5687,
+      "step": 1264
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5154921251800962e-05,
+      "loss": 0.5498,
+      "step": 1265
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5121725597832457e-05,
+      "loss": 0.561,
+      "step": 1266
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5088544226899313e-05,
+      "loss": 0.5649,
+      "step": 1267
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5055377236152827e-05,
+      "loss": 0.5553,
+      "step": 1268
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.5022224722702187e-05,
+      "loss": 0.572,
+      "step": 1269
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4989086783614224e-05,
+      "loss": 0.5841,
+      "step": 1270
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4955963515913043e-05,
+      "loss": 0.5641,
+      "step": 1271
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4922855016579844e-05,
+      "loss": 0.5532,
+      "step": 1272
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4889761382552564e-05,
+      "loss": 0.5677,
+      "step": 1273
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4856682710725605e-05,
+      "loss": 0.558,
+      "step": 1274
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4823619097949584e-05,
+      "loss": 0.553,
+      "step": 1275
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4790570641031026e-05,
+      "loss": 0.5735,
+      "step": 1276
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4757537436732055e-05,
+      "loss": 0.5597,
+      "step": 1277
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.472451958177016e-05,
+      "loss": 0.5514,
+      "step": 1278
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4691517172817903e-05,
+      "loss": 0.5599,
+      "step": 1279
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.465853030650257e-05,
+      "loss": 0.5732,
+      "step": 1280
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4625559079405994e-05,
+      "loss": 0.5614,
+      "step": 1281
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4592603588064194e-05,
+      "loss": 0.5665,
+      "step": 1282
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4559663928967114e-05,
+      "loss": 0.5606,
+      "step": 1283
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4526740198558345e-05,
+      "loss": 0.5587,
+      "step": 1284
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4493832493234852e-05,
+      "loss": 0.5665,
+      "step": 1285
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.446094090934667e-05,
+      "loss": 0.5728,
+      "step": 1286
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4428065543196633e-05,
+      "loss": 0.5606,
+      "step": 1287
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4395206491040084e-05,
+      "loss": 0.5741,
+      "step": 1288
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4362363849084626e-05,
+      "loss": 0.5723,
+      "step": 1289
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4329537713489785e-05,
+      "loss": 0.5592,
+      "step": 1290
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4296728180366772e-05,
+      "loss": 0.5613,
+      "step": 1291
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4263935345778202e-05,
+      "loss": 0.5541,
+      "step": 1292
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4231159305737757e-05,
+      "loss": 0.5469,
+      "step": 1293
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.419840015620999e-05,
+      "loss": 0.5541,
+      "step": 1294
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4165657993109987e-05,
+      "loss": 0.5514,
+      "step": 1295
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4132932912303083e-05,
+      "loss": 0.5706,
+      "step": 1296
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4100225009604613e-05,
+      "loss": 0.5717,
+      "step": 1297
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4067534380779628e-05,
+      "loss": 0.5481,
+      "step": 1298
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4034861121542575e-05,
+      "loss": 0.545,
+      "step": 1299
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4002205327557066e-05,
+      "loss": 0.5552,
+      "step": 1300
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3969567094435572e-05,
+      "loss": 0.5637,
+      "step": 1301
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3936946517739132e-05,
+      "loss": 0.5624,
+      "step": 1302
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.390434369297712e-05,
+      "loss": 0.5609,
+      "step": 1303
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3871758715606905e-05,
+      "loss": 0.567,
+      "step": 1304
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.383919168103363e-05,
+      "loss": 0.5621,
+      "step": 1305
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.380664268460986e-05,
+      "loss": 0.5571,
+      "step": 1306
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3774111821635394e-05,
+      "loss": 0.5514,
+      "step": 1307
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3741599187356918e-05,
+      "loss": 0.5523,
+      "step": 1308
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.3709104876967732e-05,
+      "loss": 0.5679,
+      "step": 1309
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.36766289856075e-05,
+      "loss": 0.5629,
+      "step": 1310
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.364417160836197e-05,
+      "loss": 0.5535,
+      "step": 1311
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.361173284026266e-05,
+      "loss": 0.5579,
+      "step": 1312
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3579312776286611e-05,
+      "loss": 0.5374,
+      "step": 1313
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3546911511356109e-05,
+      "loss": 0.5633,
+      "step": 1314
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3514529140338377e-05,
+      "loss": 0.5542,
+      "step": 1315
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3482165758045342e-05,
+      "loss": 0.5605,
+      "step": 1316
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.344982145923333e-05,
+      "loss": 0.5591,
+      "step": 1317
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3417496338602778e-05,
+      "loss": 0.5465,
+      "step": 1318
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3385190490797979e-05,
+      "loss": 0.5362,
+      "step": 1319
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3352904010406811e-05,
+      "loss": 0.5641,
+      "step": 1320
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3320636991960423e-05,
+      "loss": 0.5648,
+      "step": 1321
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3288389529932998e-05,
+      "loss": 0.5463,
+      "step": 1322
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3256161718741454e-05,
+      "loss": 0.5607,
+      "step": 1323
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.322395365274518e-05,
+      "loss": 0.5527,
+      "step": 1324
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3191765426245745e-05,
+      "loss": 0.5587,
+      "step": 1325
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3159597133486628e-05,
+      "loss": 0.5489,
+      "step": 1326
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3127448868652959e-05,
+      "loss": 0.5418,
+      "step": 1327
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3095320725871206e-05,
+      "loss": 0.5546,
+      "step": 1328
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3063212799208947e-05,
+      "loss": 0.5606,
+      "step": 1329
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3031125182674557e-05,
+      "loss": 0.5499,
+      "step": 1330
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2999057970216942e-05,
+      "loss": 0.5599,
+      "step": 1331
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2967011255725264e-05,
+      "loss": 0.5427,
+      "step": 1332
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2934985133028692e-05,
+      "loss": 0.5644,
+      "step": 1333
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2902979695896078e-05,
+      "loss": 0.5538,
+      "step": 1334
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2870995038035724e-05,
+      "loss": 0.5553,
+      "step": 1335
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2839031253095086e-05,
+      "loss": 0.5461,
+      "step": 1336
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2807088434660502e-05,
+      "loss": 0.5337,
+      "step": 1337
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2775166676256942e-05,
+      "loss": 0.5679,
+      "step": 1338
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.27432660713477e-05,
+      "loss": 0.545,
+      "step": 1339
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2711386713334131e-05,
+      "loss": 0.5544,
+      "step": 1340
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2679528695555385e-05,
+      "loss": 0.5695,
+      "step": 1341
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.264769211128814e-05,
+      "loss": 0.5725,
+      "step": 1342
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2615877053746315e-05,
+      "loss": 0.5551,
+      "step": 1343
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2584083616080787e-05,
+      "loss": 0.5649,
+      "step": 1344
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2552311891379146e-05,
+      "loss": 0.54,
+      "step": 1345
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2520561972665414e-05,
+      "loss": 0.551,
+      "step": 1346
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.248883395289975e-05,
+      "loss": 0.5497,
+      "step": 1347
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2457127924978212e-05,
+      "loss": 0.5518,
+      "step": 1348
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.242544398173246e-05,
+      "loss": 0.5644,
+      "step": 1349
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2393782215929488e-05,
+      "loss": 0.5557,
+      "step": 1350
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2362142720271371e-05,
+      "loss": 0.5621,
+      "step": 1351
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2330525587394972e-05,
+      "loss": 0.5489,
+      "step": 1352
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.229893090987167e-05,
+      "loss": 0.5472,
+      "step": 1353
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2267358780207105e-05,
+      "loss": 0.5582,
+      "step": 1354
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2235809290840913e-05,
+      "loss": 0.548,
+      "step": 1355
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2204282534146414e-05,
+      "loss": 0.5604,
+      "step": 1356
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2172778602430393e-05,
+      "loss": 0.5504,
+      "step": 1357
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2141297587932792e-05,
+      "loss": 0.554,
+      "step": 1358
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2109839582826453e-05,
+      "loss": 0.5592,
+      "step": 1359
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2078404679216864e-05,
+      "loss": 0.5548,
+      "step": 1360
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2046992969141869e-05,
+      "loss": 0.615,
+      "step": 1361
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2015604544571387e-05,
+      "loss": 0.492,
+      "step": 1362
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1984239497407175e-05,
+      "loss": 0.4816,
+      "step": 1363
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1952897919482549e-05,
+      "loss": 0.4765,
+      "step": 1364
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1921579902562103e-05,
+      "loss": 0.4776,
+      "step": 1365
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.189028553834144e-05,
+      "loss": 0.476,
+      "step": 1366
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1859014918446911e-05,
+      "loss": 0.4879,
+      "step": 1367
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1827768134435372e-05,
+      "loss": 0.4846,
+      "step": 1368
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1796545277793851e-05,
+      "loss": 0.4739,
+      "step": 1369
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1765346439939346e-05,
+      "loss": 0.4695,
+      "step": 1370
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1734171712218529e-05,
+      "loss": 0.4866,
+      "step": 1371
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1703021185907457e-05,
+      "loss": 0.4791,
+      "step": 1372
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.167189495221136e-05,
+      "loss": 0.4647,
+      "step": 1373
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.164079310226434e-05,
+      "loss": 0.4677,
+      "step": 1374
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1609715727129063e-05,
+      "loss": 0.4803,
+      "step": 1375
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1578662917796581e-05,
+      "loss": 0.4699,
+      "step": 1376
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1547634765186016e-05,
+      "loss": 0.4703,
+      "step": 1377
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1516631360144275e-05,
+      "loss": 0.4759,
+      "step": 1378
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1485652793445814e-05,
+      "loss": 0.4678,
+      "step": 1379
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1454699155792388e-05,
+      "loss": 0.4714,
+      "step": 1380
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1423770537812726e-05,
+      "loss": 0.4712,
+      "step": 1381
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1392867030062338e-05,
+      "loss": 0.4743,
+      "step": 1382
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1361988723023183e-05,
+      "loss": 0.4752,
+      "step": 1383
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.133113570710346e-05,
+      "loss": 0.4778,
+      "step": 1384
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1300308072637298e-05,
+      "loss": 0.4825,
+      "step": 1385
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1269505909884528e-05,
+      "loss": 0.4757,
+      "step": 1386
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1238729309030414e-05,
+      "loss": 0.4699,
+      "step": 1387
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1207978360185323e-05,
+      "loss": 0.4646,
+      "step": 1388
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1177253153384574e-05,
+      "loss": 0.4735,
+      "step": 1389
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.11465537785881e-05,
+      "loss": 0.4659,
+      "step": 1390
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1115880325680192e-05,
+      "loss": 0.4705,
+      "step": 1391
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1085232884469236e-05,
+      "loss": 0.4716,
+      "step": 1392
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.105461154468749e-05,
+      "loss": 0.4759,
+      "step": 1393
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1024016395990758e-05,
+      "loss": 0.4623,
+      "step": 1394
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0993447527958185e-05,
+      "loss": 0.462,
+      "step": 1395
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0962905030091948e-05,
+      "loss": 0.4664,
+      "step": 1396
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.093238899181702e-05,
+      "loss": 0.4663,
+      "step": 1397
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0901899502480912e-05,
+      "loss": 0.4649,
+      "step": 1398
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0871436651353409e-05,
+      "loss": 0.4747,
+      "step": 1399
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0841000527626283e-05,
+      "loss": 0.4677,
+      "step": 1400
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.081059122041305e-05,
+      "loss": 0.4719,
+      "step": 1401
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0780208818748727e-05,
+      "loss": 0.4699,
+      "step": 1402
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0749853411589562e-05,
+      "loss": 0.4754,
+      "step": 1403
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0719525087812737e-05,
+      "loss": 0.4735,
+      "step": 1404
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0689223936216147e-05,
+      "loss": 0.4648,
+      "step": 1405
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0658950045518149e-05,
+      "loss": 0.471,
+      "step": 1406
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0628703504357251e-05,
+      "loss": 0.4866,
+      "step": 1407
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0598484401291918e-05,
+      "loss": 0.4871,
+      "step": 1408
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.056829282480027e-05,
+      "loss": 0.4768,
+      "step": 1409
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0538128863279801e-05,
+      "loss": 0.4761,
+      "step": 1410
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0507992605047193e-05,
+      "loss": 0.4798,
+      "step": 1411
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0477884138338e-05,
+      "loss": 0.4718,
+      "step": 1412
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0447803551306403e-05,
+      "loss": 0.4652,
+      "step": 1413
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.041775093202495e-05,
+      "loss": 0.4723,
+      "step": 1414
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0387726368484321e-05,
+      "loss": 0.4738,
+      "step": 1415
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0357729948593023e-05,
+      "loss": 0.4702,
+      "step": 1416
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0327761760177199e-05,
+      "loss": 0.4725,
+      "step": 1417
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0297821890980304e-05,
+      "loss": 0.4661,
+      "step": 1418
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0267910428662878e-05,
+      "loss": 0.4725,
+      "step": 1419
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0238027460802305e-05,
+      "loss": 0.4542,
+      "step": 1420
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0208173074892533e-05,
+      "loss": 0.4667,
+      "step": 1421
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0178347358343843e-05,
+      "loss": 0.4737,
+      "step": 1422
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0148550398482521e-05,
+      "loss": 0.4683,
+      "step": 1423
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0118782282550716e-05,
+      "loss": 0.4922,
+      "step": 1424
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0089043097706106e-05,
+      "loss": 0.4793,
+      "step": 1425
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.005933293102165e-05,
+      "loss": 0.4822,
+      "step": 1426
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0029651869485349e-05,
+      "loss": 0.4617,
+      "step": 1427
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0000000000000006e-05,
+      "loss": 0.4718,
+      "step": 1428
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.970377409382922e-06,
+      "loss": 0.4745,
+      "step": 1429
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.940784184365706e-06,
+      "loss": 0.4676,
+      "step": 1430
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.911220411593967e-06,
+      "loss": 0.465,
+      "step": 1431
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.881686177627074e-06,
+      "loss": 0.4635,
+      "step": 1432
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.852181568937933e-06,
+      "loss": 0.4698,
+      "step": 1433
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.822706671912705e-06,
+      "loss": 0.4693,
+      "step": 1434
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.793261572850545e-06,
+      "loss": 0.4674,
+      "step": 1435
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.763846357963363e-06,
+      "loss": 0.47,
+      "step": 1436
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.734461113375593e-06,
+      "loss": 0.4753,
+      "step": 1437
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.705105925123893e-06,
+      "loss": 0.4754,
+      "step": 1438
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.675780879156946e-06,
+      "loss": 0.4681,
+      "step": 1439
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.646486061335151e-06,
+      "loss": 0.4685,
+      "step": 1440
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.617221557430439e-06,
+      "loss": 0.4773,
+      "step": 1441
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.587987453125945e-06,
+      "loss": 0.4804,
+      "step": 1442
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.558783834015834e-06,
+      "loss": 0.4656,
+      "step": 1443
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.529610785604995e-06,
+      "loss": 0.461,
+      "step": 1444
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.5004683933088e-06,
+      "loss": 0.4727,
+      "step": 1445
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.471356742452881e-06,
+      "loss": 0.4773,
+      "step": 1446
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.442275918272871e-06,
+      "loss": 0.4613,
+      "step": 1447
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.41322600591412e-06,
+      "loss": 0.4643,
+      "step": 1448
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.384207090431479e-06,
+      "loss": 0.4888,
+      "step": 1449
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.355219256789058e-06,
+      "loss": 0.4647,
+      "step": 1450
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.326262589859946e-06,
+      "loss": 0.4677,
+      "step": 1451
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.297337174425996e-06,
+      "loss": 0.4645,
+      "step": 1452
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.268443095177543e-06,
+      "loss": 0.4726,
+      "step": 1453
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.239580436713178e-06,
+      "loss": 0.4686,
+      "step": 1454
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.210749283539504e-06,
+      "loss": 0.4711,
+      "step": 1455
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.181949720070879e-06,
+      "loss": 0.4731,
+      "step": 1456
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.153181830629164e-06,
+      "loss": 0.4713,
+      "step": 1457
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.12444569944347e-06,
+      "loss": 0.4792,
+      "step": 1458
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.095741410649947e-06,
+      "loss": 0.4726,
+      "step": 1459
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.067069048291512e-06,
+      "loss": 0.4739,
+      "step": 1460
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.03842869631759e-06,
+      "loss": 0.4773,
+      "step": 1461
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.009820438583881e-06,
+      "loss": 0.4655,
+      "step": 1462
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.981244358852142e-06,
+      "loss": 0.4753,
+      "step": 1463
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.952700540789884e-06,
+      "loss": 0.4697,
+      "step": 1464
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.924189067970193e-06,
+      "loss": 0.4838,
+      "step": 1465
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.895710023871427e-06,
+      "loss": 0.4697,
+      "step": 1466
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.867263491876994e-06,
+      "loss": 0.4649,
+      "step": 1467
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.83884955527513e-06,
+      "loss": 0.4778,
+      "step": 1468
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.810468297258636e-06,
+      "loss": 0.466,
+      "step": 1469
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.78211980092461e-06,
+      "loss": 0.4707,
+      "step": 1470
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.753804149274236e-06,
+      "loss": 0.4822,
+      "step": 1471
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.725521425212556e-06,
+      "loss": 0.4754,
+      "step": 1472
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.697271711548163e-06,
+      "loss": 0.4734,
+      "step": 1473
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.669055090993041e-06,
+      "loss": 0.4627,
+      "step": 1474
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.640871646162255e-06,
+      "loss": 0.4566,
+      "step": 1475
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.612721459573733e-06,
+      "loss": 0.4548,
+      "step": 1476
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.584604613648042e-06,
+      "loss": 0.4863,
+      "step": 1477
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.556521190708134e-06,
+      "loss": 0.4507,
+      "step": 1478
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.528471272979083e-06,
+      "loss": 0.4559,
+      "step": 1479
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.500454942587871e-06,
+      "loss": 0.4776,
+      "step": 1480
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.472472281563144e-06,
+      "loss": 0.4631,
+      "step": 1481
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.444523371834978e-06,
+      "loss": 0.4769,
+      "step": 1482
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.416608295234611e-06,
+      "loss": 0.4698,
+      "step": 1483
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.388727133494219e-06,
+      "loss": 0.4601,
+      "step": 1484
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.3608799682467e-06,
+      "loss": 0.476,
+      "step": 1485
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.333066881025393e-06,
+      "loss": 0.4755,
+      "step": 1486
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.30528795326388e-06,
+      "loss": 0.4603,
+      "step": 1487
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.277543266295713e-06,
+      "loss": 0.4766,
+      "step": 1488
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.24983290135419e-06,
+      "loss": 0.4655,
+      "step": 1489
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.222156939572124e-06,
+      "loss": 0.4679,
+      "step": 1490
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.194515461981612e-06,
+      "loss": 0.4741,
+      "step": 1491
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.166908549513761e-06,
+      "loss": 0.4575,
+      "step": 1492
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.139336282998485e-06,
+      "loss": 0.4604,
+      "step": 1493
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.111798743164267e-06,
+      "loss": 0.4587,
+      "step": 1494
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.0842960106379e-06,
+      "loss": 0.4563,
+      "step": 1495
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.056828165944282e-06,
+      "loss": 0.4746,
+      "step": 1496
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.029395289506139e-06,
+      "loss": 0.4713,
+      "step": 1497
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.001997461643845e-06,
+      "loss": 0.4733,
+      "step": 1498
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.974634762575122e-06,
+      "loss": 0.4617,
+      "step": 1499
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.947307272414874e-06,
+      "loss": 0.4711,
+      "step": 1500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.92001507117489e-06,
+      "loss": 0.4616,
+      "step": 1501
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.892758238763645e-06,
+      "loss": 0.4701,
+      "step": 1502
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.865536854986058e-06,
+      "loss": 0.4574,
+      "step": 1503
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.83835099954327e-06,
+      "loss": 0.4762,
+      "step": 1504
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.811200752032384e-06,
+      "loss": 0.479,
+      "step": 1505
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.784086191946245e-06,
+      "loss": 0.477,
+      "step": 1506
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.757007398673229e-06,
+      "loss": 0.4766,
+      "step": 1507
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.729964451496962e-06,
+      "loss": 0.4669,
+      "step": 1508
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.702957429596152e-06,
+      "loss": 0.4699,
+      "step": 1509
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.675986412044292e-06,
+      "loss": 0.4664,
+      "step": 1510
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.649051477809461e-06,
+      "loss": 0.4633,
+      "step": 1511
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.622152705754109e-06,
+      "loss": 0.4751,
+      "step": 1512
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.595290174634802e-06,
+      "loss": 0.4607,
+      "step": 1513
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.568463963101982e-06,
+      "loss": 0.4524,
+      "step": 1514
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.541674149699754e-06,
+      "loss": 0.4584,
+      "step": 1515
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.514920812865667e-06,
+      "loss": 0.4657,
+      "step": 1516
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.488204030930473e-06,
+      "loss": 0.4655,
+      "step": 1517
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.461523882117876e-06,
+      "loss": 0.4705,
+      "step": 1518
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.434880444544323e-06,
+      "loss": 0.4707,
+      "step": 1519
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.408273796218806e-06,
+      "loss": 0.4621,
+      "step": 1520
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.3817040150425565e-06,
+      "loss": 0.4601,
+      "step": 1521
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.355171178808909e-06,
+      "loss": 0.4685,
+      "step": 1522
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.328675365202993e-06,
+      "loss": 0.4527,
+      "step": 1523
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.302216651801546e-06,
+      "loss": 0.481,
+      "step": 1524
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.275795116072694e-06,
+      "loss": 0.4789,
+      "step": 1525
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.249410835375712e-06,
+      "loss": 0.4595,
+      "step": 1526
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.223063886960779e-06,
+      "loss": 0.4595,
+      "step": 1527
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.196754347968771e-06,
+      "loss": 0.464,
+      "step": 1528
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.170482295431053e-06,
+      "loss": 0.4569,
+      "step": 1529
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.1442478062692135e-06,
+      "loss": 0.4645,
+      "step": 1530
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.118050957294877e-06,
+      "loss": 0.4686,
+      "step": 1531
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.091891825209445e-06,
+      "loss": 0.4747,
+      "step": 1532
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.065770486603891e-06,
+      "loss": 0.4622,
+      "step": 1533
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.039687017958545e-06,
+      "loss": 0.4659,
+      "step": 1534
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.013641495642858e-06,
+      "loss": 0.4696,
+      "step": 1535
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.987633995915164e-06,
+      "loss": 0.4536,
+      "step": 1536
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.96166459492247e-06,
+      "loss": 0.456,
+      "step": 1537
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.935733368700255e-06,
+      "loss": 0.4632,
+      "step": 1538
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.909840393172218e-06,
+      "loss": 0.4684,
+      "step": 1539
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.883985744150055e-06,
+      "loss": 0.4726,
+      "step": 1540
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.858169497333244e-06,
+      "loss": 0.4726,
+      "step": 1541
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.832391728308852e-06,
+      "loss": 0.4749,
+      "step": 1542
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.806652512551248e-06,
+      "loss": 0.461,
+      "step": 1543
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.7809519254219655e-06,
+      "loss": 0.4657,
+      "step": 1544
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.755290042169402e-06,
+      "loss": 0.4728,
+      "step": 1545
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.729666937928643e-06,
+      "loss": 0.4666,
+      "step": 1546
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.704082687721243e-06,
+      "loss": 0.4527,
+      "step": 1547
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.678537366454998e-06,
+      "loss": 0.4679,
+      "step": 1548
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.653031048923713e-06,
+      "loss": 0.465,
+      "step": 1549
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.627563809806991e-06,
+      "loss": 0.46,
+      "step": 1550
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.60213572367004e-06,
+      "loss": 0.4634,
+      "step": 1551
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.576746864963406e-06,
+      "loss": 0.4585,
+      "step": 1552
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.551397308022802e-06,
+      "loss": 0.4644,
+      "step": 1553
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.526087127068857e-06,
+      "loss": 0.4675,
+      "step": 1554
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.500816396206921e-06,
+      "loss": 0.4767,
+      "step": 1555
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.475585189426827e-06,
+      "loss": 0.4713,
+      "step": 1556
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.450393580602703e-06,
+      "loss": 0.4705,
+      "step": 1557
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.425241643492724e-06,
+      "loss": 0.4623,
+      "step": 1558
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.400129451738908e-06,
+      "loss": 0.4742,
+      "step": 1559
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.375057078866915e-06,
+      "loss": 0.4616,
+      "step": 1560
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.350024598285824e-06,
+      "loss": 0.4575,
+      "step": 1561
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.325032083287901e-06,
+      "loss": 0.4675,
+      "step": 1562
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.300079607048388e-06,
+      "loss": 0.4732,
+      "step": 1563
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.275167242625331e-06,
+      "loss": 0.4599,
+      "step": 1564
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.2502950629593e-06,
+      "loss": 0.4682,
+      "step": 1565
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.225463140873229e-06,
+      "loss": 0.4546,
+      "step": 1566
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.200671549072177e-06,
+      "loss": 0.4663,
+      "step": 1567
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.175920360143106e-06,
+      "loss": 0.4716,
+      "step": 1568
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.151209646554708e-06,
+      "loss": 0.4661,
+      "step": 1569
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.126539480657158e-06,
+      "loss": 0.4586,
+      "step": 1570
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.101909934681906e-06,
+      "loss": 0.4614,
+      "step": 1571
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.077321080741469e-06,
+      "loss": 0.4625,
+      "step": 1572
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.052772990829232e-06,
+      "loss": 0.4697,
+      "step": 1573
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.028265736819234e-06,
+      "loss": 0.4636,
+      "step": 1574
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.003799390465932e-06,
+      "loss": 0.467,
+      "step": 1575
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.979374023404015e-06,
+      "loss": 0.4674,
+      "step": 1576
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.954989707148202e-06,
+      "loss": 0.4709,
+      "step": 1577
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.9306465130930035e-06,
+      "loss": 0.4656,
+      "step": 1578
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.90634451251255e-06,
+      "loss": 0.4582,
+      "step": 1579
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.8820837765603364e-06,
+      "loss": 0.47,
+      "step": 1580
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.857864376269051e-06,
+      "loss": 0.4697,
+      "step": 1581
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.833686382550361e-06,
+      "loss": 0.461,
+      "step": 1582
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.8095498661947035e-06,
+      "loss": 0.4809,
+      "step": 1583
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.785454897871059e-06,
+      "loss": 0.4744,
+      "step": 1584
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.7614015481267616e-06,
+      "loss": 0.4877,
+      "step": 1585
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.73738988738731e-06,
+      "loss": 0.4661,
+      "step": 1586
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.713419985956119e-06,
+      "loss": 0.4616,
+      "step": 1587
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.689491914014356e-06,
+      "loss": 0.4682,
+      "step": 1588
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.665605741620703e-06,
+      "loss": 0.4715,
+      "step": 1589
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.641761538711164e-06,
+      "loss": 0.4604,
+      "step": 1590
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.6179593750988716e-06,
+      "loss": 0.4695,
+      "step": 1591
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.594199320473874e-06,
+      "loss": 0.4616,
+      "step": 1592
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.570481444402918e-06,
+      "loss": 0.4643,
+      "step": 1593
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.54680581632925e-06,
+      "loss": 0.457,
+      "step": 1594
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.523172505572438e-06,
+      "loss": 0.4685,
+      "step": 1595
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.499581581328148e-06,
+      "loss": 0.4677,
+      "step": 1596
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.476033112667929e-06,
+      "loss": 0.485,
+      "step": 1597
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.452527168539026e-06,
+      "loss": 0.4506,
+      "step": 1598
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.429063817764197e-06,
+      "loss": 0.4666,
+      "step": 1599
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.405643129041464e-06,
+      "loss": 0.4651,
+      "step": 1600
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.3822651709439655e-06,
+      "loss": 0.4692,
+      "step": 1601
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.358930011919712e-06,
+      "loss": 0.4758,
+      "step": 1602
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.3356377202914e-06,
+      "loss": 0.4607,
+      "step": 1603
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.312388364256229e-06,
+      "loss": 0.4561,
+      "step": 1604
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.289182011885692e-06,
+      "loss": 0.4617,
+      "step": 1605
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.266018731125355e-06,
+      "loss": 0.4537,
+      "step": 1606
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.242898589794674e-06,
+      "loss": 0.4668,
+      "step": 1607
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.219821655586821e-06,
+      "loss": 0.4691,
+      "step": 1608
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.196787996068435e-06,
+      "loss": 0.4628,
+      "step": 1609
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.173797678679475e-06,
+      "loss": 0.4697,
+      "step": 1610
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.150850770732978e-06,
+      "loss": 0.4694,
+      "step": 1611
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.127947339414905e-06,
+      "loss": 0.4658,
+      "step": 1612
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.1050874517838984e-06,
+      "loss": 0.4575,
+      "step": 1613
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.0822711747711365e-06,
+      "loss": 0.451,
+      "step": 1614
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.059498575180084e-06,
+      "loss": 0.4625,
+      "step": 1615
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.036769719686336e-06,
+      "loss": 0.456,
+      "step": 1616
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.014084674837414e-06,
+      "loss": 0.4565,
+      "step": 1617
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.991443507052565e-06,
+      "loss": 0.4542,
+      "step": 1618
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.968846282622559e-06,
+      "loss": 0.4728,
+      "step": 1619
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.946293067709511e-06,
+      "loss": 0.4705,
+      "step": 1620
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.92378392834669e-06,
+      "loss": 0.4508,
+      "step": 1621
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.901318930438297e-06,
+      "loss": 0.4714,
+      "step": 1622
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.8788981397593186e-06,
+      "loss": 0.4595,
+      "step": 1623
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.856521621955282e-06,
+      "loss": 0.4608,
+      "step": 1624
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.834189442542094e-06,
+      "loss": 0.461,
+      "step": 1625
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.811901666905856e-06,
+      "loss": 0.4655,
+      "step": 1626
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.789658360302658e-06,
+      "loss": 0.4703,
+      "step": 1627
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.767459587858379e-06,
+      "loss": 0.4612,
+      "step": 1628
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.7453054145685044e-06,
+      "loss": 0.4534,
+      "step": 1629
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.723195905297955e-06,
+      "loss": 0.4637,
+      "step": 1630
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.701131124780876e-06,
+      "loss": 0.4753,
+      "step": 1631
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.679111137620442e-06,
+      "loss": 0.46,
+      "step": 1632
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.657136008288678e-06,
+      "loss": 0.4601,
+      "step": 1633
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.635205801126286e-06,
+      "loss": 0.4643,
+      "step": 1634
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.613320580342422e-06,
+      "loss": 0.4673,
+      "step": 1635
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.5914804100145434e-06,
+      "loss": 0.4632,
+      "step": 1636
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.5696853540881936e-06,
+      "loss": 0.458,
+      "step": 1637
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.547935476376823e-06,
+      "loss": 0.4648,
+      "step": 1638
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.5262308405616184e-06,
+      "loss": 0.4853,
+      "step": 1639
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.504571510191304e-06,
+      "loss": 0.4609,
+      "step": 1640
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.482957548681941e-06,
+      "loss": 0.4486,
+      "step": 1641
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.461389019316755e-06,
+      "loss": 0.4607,
+      "step": 1642
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.439865985245973e-06,
+      "loss": 0.4514,
+      "step": 1643
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.41838850948659e-06,
+      "loss": 0.4632,
+      "step": 1644
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.396956654922238e-06,
+      "loss": 0.4642,
+      "step": 1645
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.375570484302953e-06,
+      "loss": 0.476,
+      "step": 1646
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.3542300602450195e-06,
+      "loss": 0.468,
+      "step": 1647
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.332935445230788e-06,
+      "loss": 0.4659,
+      "step": 1648
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.311686701608486e-06,
+      "loss": 0.4779,
+      "step": 1649
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.290483891592028e-06,
+      "loss": 0.4732,
+      "step": 1650
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.269327077260832e-06,
+      "loss": 0.4756,
+      "step": 1651
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.248216320559664e-06,
+      "loss": 0.4563,
+      "step": 1652
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.2271516832984335e-06,
+      "loss": 0.4615,
+      "step": 1653
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.206133227152014e-06,
+      "loss": 0.4639,
+      "step": 1654
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.185161013660055e-06,
+      "loss": 0.4691,
+      "step": 1655
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.1642351042268415e-06,
+      "loss": 0.4598,
+      "step": 1656
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.143355560121052e-06,
+      "loss": 0.4641,
+      "step": 1657
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.12252244247564e-06,
+      "loss": 0.4565,
+      "step": 1658
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.1017358122876125e-06,
+      "loss": 0.4412,
+      "step": 1659
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.080995730417865e-06,
+      "loss": 0.4621,
+      "step": 1660
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.060302257591014e-06,
+      "loss": 0.4595,
+      "step": 1661
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.03965545439521e-06,
+      "loss": 0.4605,
+      "step": 1662
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.019055381281955e-06,
+      "loss": 0.4547,
+      "step": 1663
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.998502098565924e-06,
+      "loss": 0.4625,
+      "step": 1664
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.9779956664248125e-06,
+      "loss": 0.4392,
+      "step": 1665
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.957536144899123e-06,
+      "loss": 0.4592,
+      "step": 1666
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.937123593892032e-06,
+      "loss": 0.4582,
+      "step": 1667
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.9167580731691645e-06,
+      "loss": 0.46,
+      "step": 1668
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.896439642358474e-06,
+      "loss": 0.4628,
+      "step": 1669
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8761683609500125e-06,
+      "loss": 0.4573,
+      "step": 1670
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.85594428829581e-06,
+      "loss": 0.4629,
+      "step": 1671
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.835767483609658e-06,
+      "loss": 0.4548,
+      "step": 1672
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.8156380059669525e-06,
+      "loss": 0.4658,
+      "step": 1673
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7955559143045296e-06,
+      "loss": 0.4595,
+      "step": 1674
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7755212674204856e-06,
+      "loss": 0.459,
+      "step": 1675
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.755534123973994e-06,
+      "loss": 0.467,
+      "step": 1676
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.735594542485146e-06,
+      "loss": 0.4578,
+      "step": 1677
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.715702581334788e-06,
+      "loss": 0.4601,
+      "step": 1678
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6958582987643212e-06,
+      "loss": 0.4791,
+      "step": 1679
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6760617528755682e-06,
+      "loss": 0.4595,
+      "step": 1680
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6563130016305716e-06,
+      "loss": 0.4531,
+      "step": 1681
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6366121028514334e-06,
+      "loss": 0.4612,
+      "step": 1682
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.616959114220162e-06,
+      "loss": 0.4579,
+      "step": 1683
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.597354093278489e-06,
+      "loss": 0.4602,
+      "step": 1684
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.577797097427691e-06,
+      "loss": 0.4725,
+      "step": 1685
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5582881839284357e-06,
+      "loss": 0.4719,
+      "step": 1686
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.538827409900614e-06,
+      "loss": 0.4693,
+      "step": 1687
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.51941483232318e-06,
+      "loss": 0.4587,
+      "step": 1688
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5000505080339565e-06,
+      "loss": 0.4687,
+      "step": 1689
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4807344937294873e-06,
+      "loss": 0.4737,
+      "step": 1690
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.4614668459648892e-06,
+      "loss": 0.4599,
+      "step": 1691
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4422476211536426e-06,
+      "loss": 0.4558,
+      "step": 1692
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4230768755674727e-06,
+      "loss": 0.459,
+      "step": 1693
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.403954665336149e-06,
+      "loss": 0.4743,
+      "step": 1694
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.3848810464473325e-06,
+      "loss": 0.4518,
+      "step": 1695
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.3658560747464253e-06,
+      "loss": 0.4727,
+      "step": 1696
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.3468798059363983e-06,
+      "loss": 0.4674,
+      "step": 1697
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.327952295577612e-06,
+      "loss": 0.463,
+      "step": 1698
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.3090735990876733e-06,
+      "loss": 0.4662,
+      "step": 1699
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.290243771741275e-06,
+      "loss": 0.4583,
+      "step": 1700
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.27146286867001e-06,
+      "loss": 0.464,
+      "step": 1701
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.25273094486225e-06,
+      "loss": 0.4513,
+      "step": 1702
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.234048055162944e-06,
+      "loss": 0.4707,
+      "step": 1703
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.215414254273468e-06,
+      "loss": 0.4679,
+      "step": 1704
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.196829596751494e-06,
+      "loss": 0.4629,
+      "step": 1705
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.178294137010796e-06,
+      "loss": 0.4799,
+      "step": 1706
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.159807929321097e-06,
+      "loss": 0.4557,
+      "step": 1707
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.14137102780792e-06,
+      "loss": 0.47,
+      "step": 1708
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.1229834864524287e-06,
+      "loss": 0.4599,
+      "step": 1709
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.1046453590912697e-06,
+      "loss": 0.4559,
+      "step": 1710
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.0863566994163975e-06,
+      "loss": 0.4585,
+      "step": 1711
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.0681175609749325e-06,
+      "loss": 0.4762,
+      "step": 1712
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.04992799716902e-06,
+      "loss": 0.4648,
+      "step": 1713
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.031788061255634e-06,
+      "loss": 0.4547,
+      "step": 1714
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.0136978063464627e-06,
+      "loss": 0.4507,
+      "step": 1715
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.9956572854077205e-06,
+      "loss": 0.4661,
+      "step": 1716
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.9776665512600054e-06,
+      "loss": 0.4659,
+      "step": 1717
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.9597256565781585e-06,
+      "loss": 0.4529,
+      "step": 1718
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.9418346538910915e-06,
+      "loss": 0.4535,
+      "step": 1719
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.9239935955816313e-06,
+      "loss": 0.4656,
+      "step": 1720
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.906202533886371e-06,
+      "loss": 0.4574,
+      "step": 1721
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.888461520895538e-06,
+      "loss": 0.4567,
+      "step": 1722
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.870770608552802e-06,
+      "loss": 0.4561,
+      "step": 1723
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.8531298486551564e-06,
+      "loss": 0.466,
+      "step": 1724
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.835539292852745e-06,
+      "loss": 0.466,
+      "step": 1725
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.8179989926487316e-06,
+      "loss": 0.4676,
+      "step": 1726
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.800508999399123e-06,
+      "loss": 0.464,
+      "step": 1727
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.783069364312647e-06,
+      "loss": 0.4739,
+      "step": 1728
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.7656801384505773e-06,
+      "loss": 0.4714,
+      "step": 1729
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.748341372726595e-06,
+      "loss": 0.46,
+      "step": 1730
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.731053117906648e-06,
+      "loss": 0.4545,
+      "step": 1731
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.7138154246087946e-06,
+      "loss": 0.469,
+      "step": 1732
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.696628343303047e-06,
+      "loss": 0.4418,
+      "step": 1733
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.679491924311226e-06,
+      "loss": 0.4621,
+      "step": 1734
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.662406217806841e-06,
+      "loss": 0.4755,
+      "step": 1735
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.645371273814894e-06,
+      "loss": 0.4594,
+      "step": 1736
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.6283871422117814e-06,
+      "loss": 0.4563,
+      "step": 1737
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.611453872725118e-06,
+      "loss": 0.4659,
+      "step": 1738
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.594571514933595e-06,
+      "loss": 0.4722,
+      "step": 1739
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.577740118266845e-06,
+      "loss": 0.4646,
+      "step": 1740
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.560959732005308e-06,
+      "loss": 0.4518,
+      "step": 1741
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.544230405280035e-06,
+      "loss": 0.4565,
+      "step": 1742
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5275521870726107e-06,
+      "loss": 0.4491,
+      "step": 1743
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.510925126214969e-06,
+      "loss": 0.4855,
+      "step": 1744
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.494349271389267e-06,
+      "loss": 0.4743,
+      "step": 1745
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.477824671127731e-06,
+      "loss": 0.4591,
+      "step": 1746
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.461351373812513e-06,
+      "loss": 0.4513,
+      "step": 1747
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.444929427675571e-06,
+      "loss": 0.4606,
+      "step": 1748
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4285588807984974e-06,
+      "loss": 0.464,
+      "step": 1749
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4122397811124086e-06,
+      "loss": 0.4662,
+      "step": 1750
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.3959721763977805e-06,
+      "loss": 0.4505,
+      "step": 1751
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.3797561142843107e-06,
+      "loss": 0.4584,
+      "step": 1752
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.3635916422507998e-06,
+      "loss": 0.4544,
+      "step": 1753
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.3474788076249988e-06,
+      "loss": 0.4466,
+      "step": 1754
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.33141765758345e-06,
+      "loss": 0.4631,
+      "step": 1755
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.315408239151391e-06,
+      "loss": 0.4607,
+      "step": 1756
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.2994505992025906e-06,
+      "loss": 0.4482,
+      "step": 1757
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.2835447844592016e-06,
+      "loss": 0.4623,
+      "step": 1758
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.2676908414916633e-06,
+      "loss": 0.4619,
+      "step": 1759
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.251888816718517e-06,
+      "loss": 0.4507,
+      "step": 1760
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.236138756406303e-06,
+      "loss": 0.4547,
+      "step": 1761
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.2204407066694178e-06,
+      "loss": 0.4487,
+      "step": 1762
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.2047947134699777e-06,
+      "loss": 0.4621,
+      "step": 1763
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.1892008226176763e-06,
+      "loss": 0.4562,
+      "step": 1764
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.173659079769659e-06,
+      "loss": 0.4593,
+      "step": 1765
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.158169530430392e-06,
+      "loss": 0.4535,
+      "step": 1766
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.1427322199515267e-06,
+      "loss": 0.4747,
+      "step": 1767
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.127347193531757e-06,
+      "loss": 0.4586,
+      "step": 1768
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.1120144962166968e-06,
+      "loss": 0.4616,
+      "step": 1769
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.0967341728987554e-06,
+      "loss": 0.4616,
+      "step": 1770
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.081506268316982e-06,
+      "loss": 0.4582,
+      "step": 1771
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.066330827056966e-06,
+      "loss": 0.4475,
+      "step": 1772
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.0512078935506795e-06,
+      "loss": 0.4557,
+      "step": 1773
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.036137512076353e-06,
+      "loss": 0.4678,
+      "step": 1774
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.021119726758365e-06,
+      "loss": 0.46,
+      "step": 1775
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.006154581567097e-06,
+      "loss": 0.4635,
+      "step": 1776
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.991242120318788e-06,
+      "loss": 0.4579,
+      "step": 1777
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9763823866754394e-06,
+      "loss": 0.4591,
+      "step": 1778
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9615754241446794e-06,
+      "loss": 0.4645,
+      "step": 1779
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9468212760796044e-06,
+      "loss": 0.4681,
+      "step": 1780
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9321199856787043e-06,
+      "loss": 0.4647,
+      "step": 1781
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9174715959856828e-06,
+      "loss": 0.4507,
+      "step": 1782
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9028761498893744e-06,
+      "loss": 0.4588,
+      "step": 1783
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8883336901235849e-06,
+      "loss": 0.464,
+      "step": 1784
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8738442592670014e-06,
+      "loss": 0.4662,
+      "step": 1785
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.859407899743031e-06,
+      "loss": 0.4454,
+      "step": 1786
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8450246538197003e-06,
+      "loss": 0.4501,
+      "step": 1787
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8306945636095253e-06,
+      "loss": 0.457,
+      "step": 1788
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8164176710693971e-06,
+      "loss": 0.4652,
+      "step": 1789
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.802194018000427e-06,
+      "loss": 0.4631,
+      "step": 1790
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.7880236460478695e-06,
+      "loss": 0.4443,
+      "step": 1791
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.7739065967009717e-06,
+      "loss": 0.4645,
+      "step": 1792
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.7598429112928527e-06,
+      "loss": 0.4777,
+      "step": 1793
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.745832631000395e-06,
+      "loss": 0.4633,
+      "step": 1794
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7318757968441113e-06,
+      "loss": 0.4809,
+      "step": 1795
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.717972449688028e-06,
+      "loss": 0.4604,
+      "step": 1796
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7041226302395797e-06,
+      "loss": 0.4573,
+      "step": 1797
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.690326379049474e-06,
+      "loss": 0.4615,
+      "step": 1798
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6765837365115588e-06,
+      "loss": 0.4597,
+      "step": 1799
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.662894742862744e-06,
+      "loss": 0.4642,
+      "step": 1800
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.649259438182853e-06,
+      "loss": 0.4687,
+      "step": 1801
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6356778623945223e-06,
+      "loss": 0.4731,
+      "step": 1802
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6221500552630566e-06,
+      "loss": 0.4551,
+      "step": 1803
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6086760563963478e-06,
+      "loss": 0.4461,
+      "step": 1804
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5952559052447458e-06,
+      "loss": 0.4549,
+      "step": 1805
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.5818896411009266e-06,
+      "loss": 0.4543,
+      "step": 1806
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5685773030998053e-06,
+      "loss": 0.4498,
+      "step": 1807
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.555318930218399e-06,
+      "loss": 0.4719,
+      "step": 1808
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.542114561275716e-06,
+      "loss": 0.4641,
+      "step": 1809
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5289642349326616e-06,
+      "loss": 0.4557,
+      "step": 1810
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5158679896919082e-06,
+      "loss": 0.4506,
+      "step": 1811
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.5028258638977678e-06,
+      "loss": 0.4554,
+      "step": 1812
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.4898378957361125e-06,
+      "loss": 0.4604,
+      "step": 1813
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4769041232342464e-06,
+      "loss": 0.4458,
+      "step": 1814
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4640245842607925e-06,
+      "loss": 0.4547,
+      "step": 1815
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4511993165255755e-06,
+      "loss": 0.4547,
+      "step": 1816
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4384283575795377e-06,
+      "loss": 0.4564,
+      "step": 1817
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4257117448145952e-06,
+      "loss": 0.4621,
+      "step": 1818
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4130495154635494e-06,
+      "loss": 0.4682,
+      "step": 1819
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.4004417065999865e-06,
+      "loss": 0.4539,
+      "step": 1820
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3878883551381296e-06,
+      "loss": 0.4546,
+      "step": 1821
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3753894978327754e-06,
+      "loss": 0.4627,
+      "step": 1822
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3629451712791664e-06,
+      "loss": 0.4614,
+      "step": 1823
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3505554119128861e-06,
+      "loss": 0.4394,
+      "step": 1824
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3382202560097391e-06,
+      "loss": 0.4598,
+      "step": 1825
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3259397396856665e-06,
+      "loss": 0.4739,
+      "step": 1826
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.3137138988966403e-06,
+      "loss": 0.4564,
+      "step": 1827
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.3015427694385286e-06,
+      "loss": 0.4623,
+      "step": 1828
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2894263869470236e-06,
+      "loss": 0.4535,
+      "step": 1829
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2773647868975237e-06,
+      "loss": 0.4569,
+      "step": 1830
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2653580046050262e-06,
+      "loss": 0.4625,
+      "step": 1831
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2534060752240307e-06,
+      "loss": 0.4562,
+      "step": 1832
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.2415090337484425e-06,
+      "loss": 0.451,
+      "step": 1833
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2296669150114381e-06,
+      "loss": 0.4599,
+      "step": 1834
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2178797536854093e-06,
+      "loss": 0.467,
+      "step": 1835
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.2061475842818337e-06,
+      "loss": 0.4593,
+      "step": 1836
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.1944704411511697e-06,
+      "loss": 0.4615,
+      "step": 1837
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.182848358482771e-06,
+      "loss": 0.4601,
+      "step": 1838
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.1712813703047844e-06,
+      "loss": 0.4556,
+      "step": 1839
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.1597695104840368e-06,
+      "loss": 0.4617,
+      "step": 1840
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1483128127259557e-06,
+      "loss": 0.4621,
+      "step": 1841
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1369113105744512e-06,
+      "loss": 0.4613,
+      "step": 1842
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1255650374118376e-06,
+      "loss": 0.453,
+      "step": 1843
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1142740264587104e-06,
+      "loss": 0.4567,
+      "step": 1844
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1030383107738763e-06,
+      "loss": 0.4512,
+      "step": 1845
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.0918579232542448e-06,
+      "loss": 0.4751,
+      "step": 1846
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.0807328966347152e-06,
+      "loss": 0.4661,
+      "step": 1847
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0696632634881099e-06,
+      "loss": 0.4595,
+      "step": 1848
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0586490562250651e-06,
+      "loss": 0.4495,
+      "step": 1849
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0476903070939315e-06,
+      "loss": 0.4537,
+      "step": 1850
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.036787048180683e-06,
+      "loss": 0.4645,
+      "step": 1851
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0259393114088346e-06,
+      "loss": 0.4625,
+      "step": 1852
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0151471285393223e-06,
+      "loss": 0.4631,
+      "step": 1853
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0044105311704455e-06,
+      "loss": 0.4468,
+      "step": 1854
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.937295507377409e-07,
+      "loss": 0.4525,
+      "step": 1855
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.831042185139106e-07,
+      "loss": 0.4678,
+      "step": 1856
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.725345656087271e-07,
+      "loss": 0.4481,
+      "step": 1857
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.620206229689376e-07,
+      "loss": 0.4566,
+      "step": 1858
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.51562421378176e-07,
+      "loss": 0.4581,
+      "step": 1859
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.411599914568703e-07,
+      "loss": 0.4407,
+      "step": 1860
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.308133636621597e-07,
+      "loss": 0.4425,
+      "step": 1861
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.205225682878028e-07,
+      "loss": 0.4607,
+      "step": 1862
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.1028763546408e-07,
+      "loss": 0.4628,
+      "step": 1863
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.00108595157716e-07,
+      "loss": 0.4551,
+      "step": 1864
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.899854771717909e-07,
+      "loss": 0.4612,
+      "step": 1865
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.799183111456466e-07,
+      "loss": 0.4694,
+      "step": 1866
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.699071265548054e-07,
+      "loss": 0.4486,
+      "step": 1867
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.599519527108913e-07,
+      "loss": 0.4605,
+      "step": 1868
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.500528187615131e-07,
+      "loss": 0.4705,
+      "step": 1869
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.402097536902221e-07,
+      "loss": 0.4683,
+      "step": 1870
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.304227863164005e-07,
+      "loss": 0.4477,
+      "step": 1871
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.206919452951778e-07,
+      "loss": 0.4601,
+      "step": 1872
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.110172591173548e-07,
+      "loss": 0.4465,
+      "step": 1873
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.013987561093217e-07,
+      "loss": 0.4576,
+      "step": 1874
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.918364644329602e-07,
+      "loss": 0.4662,
+      "step": 1875
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.823304120855857e-07,
+      "loss": 0.4644,
+      "step": 1876
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.72880626899839e-07,
+      "loss": 0.4627,
+      "step": 1877
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.634871365436192e-07,
+      "loss": 0.4448,
+      "step": 1878
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.541499685200082e-07,
+      "loss": 0.4598,
+      "step": 1879
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.448691501671757e-07,
+      "loss": 0.4667,
+      "step": 1880
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.356447086583096e-07,
+      "loss": 0.4521,
+      "step": 1881
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.264766710015258e-07,
+      "loss": 0.4594,
+      "step": 1882
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.173650640398034e-07,
+      "loss": 0.4652,
+      "step": 1883
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.083099144508976e-07,
+      "loss": 0.4539,
+      "step": 1884
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.993112487472609e-07,
+      "loss": 0.4424,
+      "step": 1885
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.903690932759666e-07,
+      "loss": 0.4651,
+      "step": 1886
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.814834742186361e-07,
+      "loss": 0.4647,
+      "step": 1887
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.726544175913519e-07,
+      "loss": 0.4463,
+      "step": 1888
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.638819492446002e-07,
+      "loss": 0.4608,
+      "step": 1889
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.551660948631711e-07,
+      "loss": 0.472,
+      "step": 1890
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.465068799660956e-07,
+      "loss": 0.4502,
+      "step": 1891
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.379043299065823e-07,
+      "loss": 0.4544,
+      "step": 1892
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.293584698719213e-07,
+      "loss": 0.4583,
+      "step": 1893
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.208693248834241e-07,
+      "loss": 0.466,
+      "step": 1894
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.124369197963442e-07,
+      "loss": 0.4674,
+      "step": 1895
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.040612792998124e-07,
+      "loss": 0.4525,
+      "step": 1896
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.957424279167523e-07,
+      "loss": 0.4515,
+      "step": 1897
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.874803900038251e-07,
+      "loss": 0.4635,
+      "step": 1898
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.79275189751336e-07,
+      "loss": 0.4561,
+      "step": 1899
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.711268511831903e-07,
+      "loss": 0.4661,
+      "step": 1900
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.630353981567948e-07,
+      "loss": 0.4767,
+      "step": 1901
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.550008543630103e-07,
+      "loss": 0.4592,
+      "step": 1902
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.470232433260747e-07,
+      "loss": 0.4482,
+      "step": 1903
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.391025884035239e-07,
+      "loss": 0.4642,
+      "step": 1904
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.312389127861428e-07,
+      "loss": 0.4704,
+      "step": 1905
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.234322394978852e-07,
+      "loss": 0.4724,
+      "step": 1906
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.156825913958074e-07,
+      "loss": 0.4617,
+      "step": 1907
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.07989991169997e-07,
+      "loss": 0.4511,
+      "step": 1908
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.003544613435196e-07,
+      "loss": 0.4731,
+      "step": 1909
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.927760242723434e-07,
+      "loss": 0.4604,
+      "step": 1910
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.852547021452747e-07,
+      "loss": 0.4532,
+      "step": 1911
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.777905169838915e-07,
+      "loss": 0.4594,
+      "step": 1912
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.7038349064248313e-07,
+      "loss": 0.4632,
+      "step": 1913
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.630336448079864e-07,
+      "loss": 0.4653,
+      "step": 1914
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.5574100099991636e-07,
+      "loss": 0.4534,
+      "step": 1915
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.4850558057031315e-07,
+      "loss": 0.4615,
+      "step": 1916
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.4132740470365976e-07,
+      "loss": 0.4577,
+      "step": 1917
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.3420649441685113e-07,
+      "loss": 0.4523,
+      "step": 1918
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.271428705591052e-07,
+      "loss": 0.4573,
+      "step": 1919
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.2013655381190955e-07,
+      "loss": 0.4534,
+      "step": 1920
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.1318756468897047e-07,
+      "loss": 0.4524,
+      "step": 1921
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.062959235361397e-07,
+      "loss": 0.4585,
+      "step": 1922
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.994616505313631e-07,
+      "loss": 0.4575,
+      "step": 1923
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.926847656846189e-07,
+      "loss": 0.4607,
+      "step": 1924
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.8596528883785954e-07,
+      "loss": 0.4659,
+      "step": 1925
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.793032396649499e-07,
+      "loss": 0.4587,
+      "step": 1926
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7269863767161796e-07,
+      "loss": 0.4585,
+      "step": 1927
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.661515021953932e-07,
+      "loss": 0.4392,
+      "step": 1928
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.5966185240554394e-07,
+      "loss": 0.4546,
+      "step": 1929
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5322970730303106e-07,
+      "loss": 0.4577,
+      "step": 1930
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.4685508572044777e-07,
+      "loss": 0.4502,
+      "step": 1931
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.4053800632196434e-07,
+      "loss": 0.4583,
+      "step": 1932
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.3427848760327673e-07,
+      "loss": 0.4439,
+      "step": 1933
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.280765478915449e-07,
+      "loss": 0.46,
+      "step": 1934
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.219322053453433e-07,
+      "loss": 0.4637,
+      "step": 1935
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.1584547795461716e-07,
+      "loss": 0.4671,
+      "step": 1936
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.098163835406154e-07,
+      "loss": 0.4576,
+      "step": 1937
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.038449397558396e-07,
+      "loss": 0.4583,
+      "step": 1938
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.9793116408400434e-07,
+      "loss": 0.4532,
+      "step": 1939
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.920750738399747e-07,
+      "loss": 0.4575,
+      "step": 1940
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.8627668616972194e-07,
+      "loss": 0.4625,
+      "step": 1941
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.8053601805026585e-07,
+      "loss": 0.4694,
+      "step": 1942
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.748530862896348e-07,
+      "loss": 0.4591,
+      "step": 1943
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.6922790752681004e-07,
+      "loss": 0.4608,
+      "step": 1944
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.6366049823167704e-07,
+      "loss": 0.4614,
+      "step": 1945
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.5815087470498326e-07,
+      "loss": 0.4588,
+      "step": 1946
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.52699053078278e-07,
+      "loss": 0.4679,
+      "step": 1947
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.4730504931388177e-07,
+      "loss": 0.4557,
+      "step": 1948
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.419688792048258e-07,
+      "loss": 0.4524,
+      "step": 1949
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.3669055837481247e-07,
+      "loss": 0.4447,
+      "step": 1950
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.3147010227816847e-07,
+      "loss": 0.4609,
+      "step": 1951
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.26307526199796e-07,
+      "loss": 0.456,
+      "step": 1952
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2120284525513735e-07,
+      "loss": 0.448,
+      "step": 1953
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1615607439011698e-07,
+      "loss": 0.4503,
+      "step": 1954
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.111672283811106e-07,
+      "loss": 0.4687,
+      "step": 1955
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0623632183489395e-07,
+      "loss": 0.4719,
+      "step": 1956
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0136336918860522e-07,
+      "loss": 0.4429,
+      "step": 1957
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9654838470969607e-07,
+      "loss": 0.4554,
+      "step": 1958
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9179138249589836e-07,
+      "loss": 0.4571,
+      "step": 1959
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8709237647517308e-07,
+      "loss": 0.4666,
+      "step": 1960
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8245138040567934e-07,
+      "loss": 0.454,
+      "step": 1961
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.778684078757298e-07,
+      "loss": 0.4579,
+      "step": 1962
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.733434723037486e-07,
+      "loss": 0.4681,
+      "step": 1963
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.6887658693823583e-07,
+      "loss": 0.4588,
+      "step": 1964
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.6446776485772531e-07,
+      "loss": 0.4561,
+      "step": 1965
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.6011701897074906e-07,
+      "loss": 0.4578,
+      "step": 1966
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5582436201579732e-07,
+      "loss": 0.4647,
+      "step": 1967
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.515898065612853e-07,
+      "loss": 0.4489,
+      "step": 1968
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.4741336500551318e-07,
+      "loss": 0.451,
+      "step": 1969
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.432950495766261e-07,
+      "loss": 0.4545,
+      "step": 1970
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.3923487233258982e-07,
+      "loss": 0.459,
+      "step": 1971
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.3523284516113955e-07,
+      "loss": 0.4559,
+      "step": 1972
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.312889797797623e-07,
+      "loss": 0.4624,
+      "step": 1973
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.274032877356457e-07,
+      "loss": 0.4638,
+      "step": 1974
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.235757804056581e-07,
+      "loss": 0.4654,
+      "step": 1975
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1980646899630632e-07,
+      "loss": 0.4554,
+      "step": 1976
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.1609536454371129e-07,
+      "loss": 0.4694,
+      "step": 1977
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.1244247791356133e-07,
+      "loss": 0.4513,
+      "step": 1978
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0884781980109893e-07,
+      "loss": 0.4601,
+      "step": 1979
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0531140073107626e-07,
+      "loss": 0.452,
+      "step": 1980
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0183323105772635e-07,
+      "loss": 0.4716,
+      "step": 1981
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.841332096473866e-08,
+      "loss": 0.4573,
+      "step": 1982
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.505168046521907e-08,
+      "loss": 0.4412,
+      "step": 1983
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.17483194016744e-08,
+      "loss": 0.4558,
+      "step": 1984
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.850324744597017e-08,
+      "loss": 0.4628,
+      "step": 1985
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.531647409931065e-08,
+      "loss": 0.4512,
+      "step": 1986
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.218800869220555e-08,
+      "loss": 0.4528,
+      "step": 1987
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.911786038445002e-08,
+      "loss": 0.461,
+      "step": 1988
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.61060381650891e-08,
+      "loss": 0.4533,
+      "step": 1989
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.315255085240003e-08,
+      "loss": 0.4653,
+      "step": 1990
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.025740709385665e-08,
+      "loss": 0.4517,
+      "step": 1991
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.742061536611166e-08,
+      "loss": 0.4365,
+      "step": 1992
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.464218397497224e-08,
+      "loss": 0.4528,
+      "step": 1993
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.192212105537109e-08,
+      "loss": 0.4616,
+      "step": 1994
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.926043457134212e-08,
+      "loss": 0.4523,
+      "step": 1995
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.6657132316004783e-08,
+      "loss": 0.4553,
+      "step": 1996
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.411222191153309e-08,
+      "loss": 0.4503,
+      "step": 1997
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.162571080913781e-08,
+      "loss": 0.469,
+      "step": 1998
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.919760628903758e-08,
+      "loss": 0.4645,
+      "step": 1999
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.682791546045229e-08,
+      "loss": 0.4627,
+      "step": 2000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.4516645261567516e-08,
+      "loss": 0.4639,
+      "step": 2001
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.226380245951678e-08,
+      "loss": 0.4621,
+      "step": 2002
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.0069393650374875e-08,
+      "loss": 0.4658,
+      "step": 2003
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.793342525911792e-08,
+      "loss": 0.4707,
+      "step": 2004
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5855903539623315e-08,
+      "loss": 0.457,
+      "step": 2005
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.383683457463649e-08,
+      "loss": 0.4625,
+      "step": 2006
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.18762242757642e-08,
+      "loss": 0.4695,
+      "step": 2007
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.997407838345012e-08,
+      "loss": 0.461,
+      "step": 2008
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.8130402466963745e-08,
+      "loss": 0.451,
+      "step": 2009
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.6345201924380393e-08,
+      "loss": 0.4548,
+      "step": 2010
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.4618481982563447e-08,
+      "loss": 0.4587,
+      "step": 2011
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.295024769715326e-08,
+      "loss": 0.445,
+      "step": 2012
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.1340503952551606e-08,
+      "loss": 0.471,
+      "step": 2013
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.9789255461906132e-08,
+      "loss": 0.4599,
+      "step": 2014
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8296506767097045e-08,
+      "loss": 0.4571,
+      "step": 2015
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6862262238723782e-08,
+      "loss": 0.4499,
+      "step": 2016
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.548652607608947e-08,
+      "loss": 0.4755,
+      "step": 2017
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4169302307198706e-08,
+      "loss": 0.4492,
+      "step": 2018
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2910594788728692e-08,
+      "loss": 0.4567,
+      "step": 2019
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.171040720603589e-08,
+      "loss": 0.4592,
+      "step": 2020
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0568743073133825e-08,
+      "loss": 0.4712,
+      "step": 2021
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.48560573268198e-09,
+      "loss": 0.4603,
+      "step": 2022
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.460998355988014e-09,
+      "loss": 0.4451,
+      "step": 2023
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.494923942985566e-09,
+      "loss": 0.445,
+      "step": 2024
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.5873853222298e-09,
+      "loss": 0.4718,
+      "step": 2025
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.738385150890757e-09,
+      "loss": 0.4686,
+      "step": 2026
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.947925914748908e-09,
+      "loss": 0.4545,
+      "step": 2027
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.216009928172948e-09,
+      "loss": 0.4487,
+      "step": 2028
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.542639334135345e-09,
+      "loss": 0.4602,
+      "step": 2029
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9278161041879082e-09,
+      "loss": 0.467,
+      "step": 2030
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.371542038461794e-09,
+      "loss": 0.4471,
+      "step": 2031
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8738187656652805e-09,
+      "loss": 0.4626,
+      "step": 2032
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4346477430771111e-09,
+      "loss": 0.4438,
+      "step": 2033
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0540302565376081e-09,
+      "loss": 0.4651,
+      "step": 2034
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.319674204553373e-10,
+      "loss": 0.4615,
+      "step": 2035
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.684601777893427e-10,
+      "loss": 0.4696,
+      "step": 2036
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.6350930006247e-10,
+      "loss": 0.4466,
+      "step": 2037
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1711538734804352e-10,
+      "loss": 0.4566,
+      "step": 2038
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.927886826764592e-11,
+      "loss": 0.4599,
+      "step": 2039
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.4608,
+      "step": 2040
+    },
+    {
+      "epoch": 3.0,
+      "step": 2040,
+      "total_flos": 7.044039609589174e+23,
+      "train_loss": 0.5695364278321172,
+      "train_runtime": 108001.0924,
+      "train_samples_per_second": 19.355,
+      "train_steps_per_second": 0.019
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2040,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 200,
+  "total_flos": 7.044039609589174e+23,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}