diff --git "a/checkpoint-5634/trainer_state.json" "b/checkpoint-5634/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-5634/trainer_state.json"
@@ -0,0 +1,33820 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 5634,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1764705882352942e-06,
+      "loss": 1.2688,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3529411764705885e-06,
+      "loss": 1.2504,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5294117647058825e-06,
+      "loss": 1.225,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.705882352941177e-06,
+      "loss": 1.2719,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.882352941176471e-06,
+      "loss": 0.3014,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.058823529411765e-06,
+      "loss": 1.2515,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.23529411764706e-06,
+      "loss": 1.2199,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.411764705882354e-06,
+      "loss": 1.3506,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0588235294117648e-05,
+      "loss": 1.2771,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1764705882352942e-05,
+      "loss": 0.2694,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2941176470588238e-05,
+      "loss": 1.2069,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.411764705882353e-05,
+      "loss": 1.1212,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5294117647058826e-05,
+      "loss": 1.1811,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.647058823529412e-05,
+      "loss": 1.1972,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7647058823529414e-05,
+      "loss": 1.1753,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8823529411764708e-05,
+      "loss": 1.1165,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 1.1001,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1176470588235296e-05,
+      "loss": 1.1655,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.235294117647059e-05,
+      "loss": 1.1392,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3529411764705884e-05,
+      "loss": 1.1299,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4705882352941178e-05,
+      "loss": 1.1601,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5882352941176475e-05,
+      "loss": 1.0794,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7058823529411766e-05,
+      "loss": 1.1012,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.823529411764706e-05,
+      "loss": 1.0361,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9411764705882354e-05,
+      "loss": 1.0554,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.058823529411765e-05,
+      "loss": 1.0684,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.176470588235294e-05,
+      "loss": 1.0014,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.294117647058824e-05,
+      "loss": 1.083,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.411764705882353e-05,
+      "loss": 1.0448,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.529411764705883e-05,
+      "loss": 1.0212,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6470588235294114e-05,
+      "loss": 1.0086,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7647058823529415e-05,
+      "loss": 0.9803,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.882352941176471e-05,
+      "loss": 1.1466,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4e-05,
+      "loss": 1.0603,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.11764705882353e-05,
+      "loss": 1.002,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.235294117647059e-05,
+      "loss": 1.0081,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3529411764705885e-05,
+      "loss": 0.9772,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.470588235294118e-05,
+      "loss": 1.064,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.588235294117647e-05,
+      "loss": 1.0832,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.705882352941177e-05,
+      "loss": 1.0444,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.823529411764706e-05,
+      "loss": 1.0374,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9411764705882355e-05,
+      "loss": 0.9721,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.058823529411765e-05,
+      "loss": 1.0113,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.176470588235295e-05,
+      "loss": 0.323,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.294117647058824e-05,
+      "loss": 1.0028,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.411764705882353e-05,
+      "loss": 1.0171,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.529411764705883e-05,
+      "loss": 1.0066,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.647058823529412e-05,
+      "loss": 1.0142,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.764705882352941e-05,
+      "loss": 0.9798,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.882352941176471e-05,
+      "loss": 0.9921,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-05,
+      "loss": 1.0466,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.11764705882353e-05,
+      "loss": 0.9956,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.23529411764706e-05,
+      "loss": 0.9524,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.352941176470588e-05,
+      "loss": 0.9727,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.470588235294118e-05,
+      "loss": 0.9322,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.588235294117648e-05,
+      "loss": 0.9948,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.705882352941176e-05,
+      "loss": 0.9715,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.823529411764707e-05,
+      "loss": 0.9428,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.941176470588236e-05,
+      "loss": 1.0618,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.058823529411765e-05,
+      "loss": 1.032,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.176470588235295e-05,
+      "loss": 1.0185,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.294117647058823e-05,
+      "loss": 0.936,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.411764705882354e-05,
+      "loss": 0.9932,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.529411764705883e-05,
+      "loss": 0.9524,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.647058823529411e-05,
+      "loss": 1.0313,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.764705882352942e-05,
+      "loss": 0.9176,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.882352941176471e-05,
+      "loss": 0.9543,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8e-05,
+      "loss": 1.0212,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.11764705882353e-05,
+      "loss": 0.3062,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.23529411764706e-05,
+      "loss": 1.0062,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.352941176470589e-05,
+      "loss": 0.2933,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.470588235294118e-05,
+      "loss": 0.9966,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.588235294117646e-05,
+      "loss": 0.9769,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.705882352941177e-05,
+      "loss": 0.9738,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.823529411764706e-05,
+      "loss": 0.9751,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.941176470588236e-05,
+      "loss": 0.9251,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.058823529411765e-05,
+      "loss": 0.9724,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.176470588235295e-05,
+      "loss": 0.971,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.294117647058824e-05,
+      "loss": 0.952,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.411764705882353e-05,
+      "loss": 0.9376,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.529411764705883e-05,
+      "loss": 0.9655,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.647058823529412e-05,
+      "loss": 0.9348,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.764705882352942e-05,
+      "loss": 0.9216,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.882352941176471e-05,
+      "loss": 0.9185,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001,
+      "loss": 0.9265,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001011764705882353,
+      "loss": 0.9164,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010235294117647058,
+      "loss": 0.9694,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001035294117647059,
+      "loss": 1.0046,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010470588235294118,
+      "loss": 0.9894,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010588235294117647,
+      "loss": 0.9176,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010705882352941177,
+      "loss": 0.8976,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010823529411764706,
+      "loss": 0.8502,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010941176470588237,
+      "loss": 0.9479,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011058823529411766,
+      "loss": 0.9745,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011176470588235294,
+      "loss": 0.9509,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011294117647058824,
+      "loss": 0.3086,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011411764705882353,
+      "loss": 0.9264,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011529411764705881,
+      "loss": 0.9866,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011647058823529413,
+      "loss": 0.8725,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011764705882352942,
+      "loss": 0.9604,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011882352941176471,
+      "loss": 0.9903,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012,
+      "loss": 0.9824,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001211764705882353,
+      "loss": 1.0026,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001223529411764706,
+      "loss": 0.8374,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001235294117647059,
+      "loss": 0.8776,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001247058823529412,
+      "loss": 0.9502,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001258823529411765,
+      "loss": 0.8804,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012705882352941175,
+      "loss": 1.0031,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012823529411764707,
+      "loss": 0.9907,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012941176470588237,
+      "loss": 0.948,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013058823529411766,
+      "loss": 0.9477,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013176470588235296,
+      "loss": 0.9618,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013294117647058822,
+      "loss": 0.9109,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013411764705882352,
+      "loss": 0.914,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013529411764705884,
+      "loss": 0.9686,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013647058823529413,
+      "loss": 0.8838,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013764705882352943,
+      "loss": 0.9375,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013882352941176472,
+      "loss": 0.8848,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014,
+      "loss": 0.9557,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001411764705882353,
+      "loss": 0.9564,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001423529411764706,
+      "loss": 0.9535,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001435294117647059,
+      "loss": 0.9509,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001447058823529412,
+      "loss": 0.933,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014588235294117646,
+      "loss": 0.9747,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014705882352941178,
+      "loss": 0.8997,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014823529411764707,
+      "loss": 0.9811,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014941176470588237,
+      "loss": 0.9517,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015058823529411766,
+      "loss": 1.0035,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015176470588235295,
+      "loss": 0.9089,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015294117647058822,
+      "loss": 0.93,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015411764705882354,
+      "loss": 0.9629,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015529411764705884,
+      "loss": 0.882,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015647058823529413,
+      "loss": 0.9793,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015764705882352943,
+      "loss": 0.9243,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001588235294117647,
+      "loss": 0.9624,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016,
+      "loss": 0.3099,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001611764705882353,
+      "loss": 0.9456,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001623529411764706,
+      "loss": 0.9403,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001635294117647059,
+      "loss": 0.9373,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001647058823529412,
+      "loss": 0.9781,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016588235294117648,
+      "loss": 0.9119,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016705882352941178,
+      "loss": 0.9547,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016823529411764707,
+      "loss": 0.9523,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016941176470588237,
+      "loss": 0.8786,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017058823529411766,
+      "loss": 1.0202,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017176470588235293,
+      "loss": 0.9094,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017294117647058825,
+      "loss": 0.9048,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017411764705882354,
+      "loss": 0.9292,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017529411764705884,
+      "loss": 0.9405,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017647058823529413,
+      "loss": 0.8884,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017764705882352942,
+      "loss": 0.9282,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017882352941176472,
+      "loss": 0.9642,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018,
+      "loss": 0.9344,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001811764705882353,
+      "loss": 0.9972,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001823529411764706,
+      "loss": 0.264,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001835294117647059,
+      "loss": 0.3281,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001847058823529412,
+      "loss": 0.9375,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018588235294117648,
+      "loss": 0.9473,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018705882352941178,
+      "loss": 0.9365,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018823529411764707,
+      "loss": 0.9844,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018941176470588236,
+      "loss": 0.9461,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019058823529411766,
+      "loss": 0.9059,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019176470588235295,
+      "loss": 0.9582,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019294117647058825,
+      "loss": 0.3353,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019411764705882354,
+      "loss": 0.9396,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019529411764705883,
+      "loss": 0.9781,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019647058823529413,
+      "loss": 0.9599,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019764705882352942,
+      "loss": 0.9924,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019882352941176472,
+      "loss": 0.9006,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.9024,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999998347093237,
+      "loss": 0.9141,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999993388373499,
+      "loss": 0.2918,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999998512384242,
+      "loss": 0.9596,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999973553502733,
+      "loss": 0.8994,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999958677358265,
+      "loss": 0.9226,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999940495413936,
+      "loss": 0.9263,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999919007675752,
+      "loss": 0.9456,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999894214150818,
+      "loss": 0.9084,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999986611484733,
+      "loss": 0.8917,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999834709774576,
+      "loss": 0.9745,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999799998942944,
+      "loss": 0.9051,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000199997619823639,
+      "loss": 0.95,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999720660050015,
+      "loss": 1.0082,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999676032014953,
+      "loss": 0.9214,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999628098273464,
+      "loss": 0.9739,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999576858841395,
+      "loss": 0.8807,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999522313735683,
+      "loss": 0.9916,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999946446297436,
+      "loss": 0.3132,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999940330657655,
+      "loss": 0.9753,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999338844562477,
+      "loss": 0.9591,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999271076953442,
+      "loss": 0.9092,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999920000377185,
+      "loss": 0.9109,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999125625041198,
+      "loss": 0.9882,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999047940786073,
+      "loss": 0.9048,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998966951032156,
+      "loss": 0.8864,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998882655806224,
+      "loss": 0.9412,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998795055136139,
+      "loss": 0.9315,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998704149050864,
+      "loss": 0.9577,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998609937580447,
+      "loss": 0.9097,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998512420756032,
+      "loss": 0.9334,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998411598609863,
+      "loss": 0.9469,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998307471175264,
+      "loss": 0.929,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998200038486656,
+      "loss": 0.9898,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998089300579558,
+      "loss": 0.905,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997975257490583,
+      "loss": 0.9354,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999785790925742,
+      "loss": 0.917,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997737255918869,
+      "loss": 0.9026,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997613297514816,
+      "loss": 0.9729,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997486034086236,
+      "loss": 0.9404,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019997355465675205,
+      "loss": 0.9578,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999722159232488,
+      "loss": 0.8983,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999708441407952,
+      "loss": 0.9254,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996943930984475,
+      "loss": 0.945,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996800143086188,
+      "loss": 0.3203,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996653050432186,
+      "loss": 0.2944,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199965026530711,
+      "loss": 0.88,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996348951052646,
+      "loss": 0.923,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996191944427638,
+      "loss": 0.9529,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996031633247977,
+      "loss": 0.8942,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999586801756666,
+      "loss": 0.9375,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995701097437775,
+      "loss": 0.9674,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995530872916501,
+      "loss": 0.8702,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995357344059116,
+      "loss": 0.908,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999518051092298,
+      "loss": 0.9567,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999500037356655,
+      "loss": 0.9552,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994816932049383,
+      "loss": 0.9127,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994630186432116,
+      "loss": 0.8833,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994440136776484,
+      "loss": 0.8393,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994246783145317,
+      "loss": 0.9056,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999405012560253,
+      "loss": 0.8949,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993850164213138,
+      "loss": 0.8563,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993646899043238,
+      "loss": 0.9048,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993440330160037,
+      "loss": 0.9036,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999323045763181,
+      "loss": 0.8698,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993017281527947,
+      "loss": 0.9343,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992800801918914,
+      "loss": 0.9074,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992581018876278,
+      "loss": 0.9249,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992357932472693,
+      "loss": 0.915,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992131542781908,
+      "loss": 0.3391,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991901849878766,
+      "loss": 0.9112,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991668853839198,
+      "loss": 0.9525,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991432554740225,
+      "loss": 0.9152,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991192952659967,
+      "loss": 0.9166,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999095004767763,
+      "loss": 0.9243,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990703839873516,
+      "loss": 0.9136,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999045432932901,
+      "loss": 1.0022,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199902015161266,
+      "loss": 0.8966,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989945400349866,
+      "loss": 0.9421,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001998968598208347,
+      "loss": 0.8899,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001998942326141317,
+      "loss": 0.9061,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989157238425818,
+      "loss": 0.9614,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019988887913209355,
+      "loss": 0.9341,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019988615285852822,
+      "loss": 0.9008,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019988339356446334,
+      "loss": 0.956,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019988060125081112,
+      "loss": 0.9075,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987777591849468,
+      "loss": 0.8904,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000199874917568448,
+      "loss": 0.8967,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000199872026201616,
+      "loss": 0.9571,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986910181895448,
+      "loss": 0.8654,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986614442143023,
+      "loss": 0.8893,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998631540100209,
+      "loss": 0.9644,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986013058571504,
+      "loss": 0.3024,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985707414951217,
+      "loss": 0.8882,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985398470242268,
+      "loss": 0.9105,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985086224546788,
+      "loss": 0.8936,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984770677968,
+      "loss": 0.8842,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984451830610213,
+      "loss": 0.893,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984129682578842,
+      "loss": 0.9249,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983804233980376,
+      "loss": 0.9369,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983475484922406,
+      "loss": 0.9459,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983143435513604,
+      "loss": 0.9609,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019982808085863745,
+      "loss": 0.9004,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019982469436083687,
+      "loss": 0.9039,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019982127486285384,
+      "loss": 0.929,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019981782236581875,
+      "loss": 0.8776,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019981433687087295,
+      "loss": 0.9194,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019981081837916866,
+      "loss": 0.9156,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980726689186907,
+      "loss": 0.9027,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980368241014816,
+      "loss": 0.338,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000199800064935191,
+      "loss": 0.9048,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019979641446819338,
+      "loss": 1.0194,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997927310103621,
+      "loss": 0.919,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997890145629148,
+      "loss": 0.9026,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019978526512708013,
+      "loss": 0.9195,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997814827040976,
+      "loss": 0.9053,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019977766729521753,
+      "loss": 0.9171,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019977381890170126,
+      "loss": 0.8686,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000199769937524821,
+      "loss": 0.9071,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997660231658599,
+      "loss": 0.9556,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019976207582611189,
+      "loss": 0.9047,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019975809550688194,
+      "loss": 0.8971,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019975408220948584,
+      "loss": 0.8999,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019975003593525036,
+      "loss": 0.8962,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997459566855131,
+      "loss": 0.9958,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019974184446162253,
+      "loss": 0.9171,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997376992649382,
+      "loss": 0.9073,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997335210968303,
+      "loss": 0.287,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019972930995868014,
+      "loss": 0.9688,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997250658518798,
+      "loss": 0.8313,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019972078877783232,
+      "loss": 0.9455,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019971647873795164,
+      "loss": 0.9233,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997121357336625,
+      "loss": 0.9523,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019970775976640076,
+      "loss": 0.9033,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997033508376129,
+      "loss": 0.9485,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019969890894875646,
+      "loss": 0.939,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001996944341012999,
+      "loss": 0.8683,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019968992629672245,
+      "loss": 0.9165,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019968538553651437,
+      "loss": 0.882,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019968081182217672,
+      "loss": 0.9104,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019967620515522146,
+      "loss": 0.9181,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019967156553717152,
+      "loss": 0.8445,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019966689296956064,
+      "loss": 0.9868,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019966218745393345,
+      "loss": 0.8836,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996574489918456,
+      "loss": 0.8852,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019965267758486347,
+      "loss": 0.8653,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019964787323456436,
+      "loss": 0.8844,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996430359425366,
+      "loss": 0.8834,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019963816571037923,
+      "loss": 0.9464,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996332625397023,
+      "loss": 0.3105,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019962832643212667,
+      "loss": 0.8941,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996233573892842,
+      "loss": 0.8778,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019961835541281746,
+      "loss": 1.03,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019961332050438008,
+      "loss": 0.9788,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019960825266563648,
+      "loss": 0.9662,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000199603151898262,
+      "loss": 0.9428,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019959801820394285,
+      "loss": 0.8569,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019959285158437616,
+      "loss": 0.9473,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019958765204126987,
+      "loss": 0.9431,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019958241957634292,
+      "loss": 0.9541,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019957715419132498,
+      "loss": 0.97,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019957185588795678,
+      "loss": 0.893,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019956652466798978,
+      "loss": 0.9134,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019956116053318642,
+      "loss": 0.9279,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019955576348531994,
+      "loss": 0.936,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995503335261745,
+      "loss": 0.8786,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019954487065754518,
+      "loss": 0.2736,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995393748812379,
+      "loss": 0.9651,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019953384619906945,
+      "loss": 0.867,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995282846128675,
+      "loss": 0.9798,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019952269012447064,
+      "loss": 0.9679,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019951706273572825,
+      "loss": 0.8963,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995114024485007,
+      "loss": 0.8813,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019950570926465912,
+      "loss": 0.9208,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019949998318608561,
+      "loss": 0.8673,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019949422421467307,
+      "loss": 0.9188,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019948843235232535,
+      "loss": 0.8964,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994826076009571,
+      "loss": 0.8255,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019947674996249393,
+      "loss": 0.8778,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019947085943887216,
+      "loss": 0.2906,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019946493603203918,
+      "loss": 0.9063,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994589797439531,
+      "loss": 0.9522,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000199452990576583,
+      "loss": 0.8977,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994469685319088,
+      "loss": 0.8638,
+      "step": 353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994409136119212,
+      "loss": 0.8772,
+      "step": 354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019943482581862192,
+      "loss": 0.9007,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019942870515402345,
+      "loss": 0.9491,
+      "step": 356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019942255162014912,
+      "loss": 0.8825,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019941636521903321,
+      "loss": 0.8983,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019941014595272085,
+      "loss": 0.8626,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019940389382326802,
+      "loss": 0.7946,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019939760883274147,
+      "loss": 0.8905,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019939129098321904,
+      "loss": 0.8895,
+      "step": 362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019938494027678917,
+      "loss": 0.9351,
+      "step": 363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019937855671555132,
+      "loss": 0.9064,
+      "step": 364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019937214030161583,
+      "loss": 0.9537,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019936569103710377,
+      "loss": 0.8936,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019935920892414716,
+      "loss": 0.8949,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019935269396488894,
+      "loss": 0.9504,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019934614616148275,
+      "loss": 0.8792,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019933956551609322,
+      "loss": 0.9513,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019933295203089575,
+      "loss": 0.9527,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019932630570807666,
+      "loss": 1.0036,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019931962654983311,
+      "loss": 0.9097,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019931291455837306,
+      "loss": 0.918,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019930616973591546,
+      "loss": 0.9322,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019929939208468991,
+      "loss": 0.875,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019929258160693704,
+      "loss": 0.9534,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019928573830490826,
+      "loss": 0.8712,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019927886218086584,
+      "loss": 0.8896,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001992719532370829,
+      "loss": 0.8667,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019926501147584337,
+      "loss": 0.9749,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019925803689944212,
+      "loss": 0.8544,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019925102951018478,
+      "loss": 0.8933,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019924398931038786,
+      "loss": 0.8868,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019923691630237874,
+      "loss": 0.8392,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019922981048849564,
+      "loss": 0.8743,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019922267187108756,
+      "loss": 0.9887,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019921550045251443,
+      "loss": 0.9397,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019920829623514696,
+      "loss": 0.8723,
+      "step": 389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019920105922136678,
+      "loss": 0.9092,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019919378941356627,
+      "loss": 0.9217,
+      "step": 391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019918648681414868,
+      "loss": 0.8281,
+      "step": 392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019917915142552816,
+      "loss": 0.9495,
+      "step": 393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019917178325012963,
+      "loss": 0.9613,
+      "step": 394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019916438229038884,
+      "loss": 0.8999,
+      "step": 395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019915694854875246,
+      "loss": 0.8869,
+      "step": 396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019914948202767795,
+      "loss": 0.8369,
+      "step": 397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019914198272963352,
+      "loss": 0.9108,
+      "step": 398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019913445065709837,
+      "loss": 0.8951,
+      "step": 399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019912688581256248,
+      "loss": 0.89,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019911928819852658,
+      "loss": 0.8557,
+      "step": 401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019911165781750237,
+      "loss": 0.9351,
+      "step": 402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019910399467201227,
+      "loss": 0.8719,
+      "step": 403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019909629876458954,
+      "loss": 0.9104,
+      "step": 404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019908857009777838,
+      "loss": 0.9172,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019908080867413368,
+      "loss": 0.8932,
+      "step": 406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019907301449622128,
+      "loss": 0.9456,
+      "step": 407
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990651875666177,
+      "loss": 0.8991,
+      "step": 408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019905732788791052,
+      "loss": 0.8715,
+      "step": 409
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019904943546269785,
+      "loss": 0.8472,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019904151029358889,
+      "loss": 0.9459,
+      "step": 411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019903355238320346,
+      "loss": 0.8942,
+      "step": 412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990255617341724,
+      "loss": 0.9287,
+      "step": 413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001990175383491372,
+      "loss": 0.9173,
+      "step": 414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019900948223075022,
+      "loss": 0.8221,
+      "step": 415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019900139338167473,
+      "loss": 0.9501,
+      "step": 416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001989932718045847,
+      "loss": 0.838,
+      "step": 417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019898511750216505,
+      "loss": 0.9673,
+      "step": 418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019897693047711134,
+      "loss": 0.891,
+      "step": 419
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019896871073213007,
+      "loss": 0.8828,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001989604582699386,
+      "loss": 0.87,
+      "step": 421
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000198952173093265,
+      "loss": 0.9732,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019894385520484816,
+      "loss": 0.904,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019893550460743788,
+      "loss": 0.9552,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019892712130379465,
+      "loss": 0.8491,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001989187052966899,
+      "loss": 0.9757,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019891025658890575,
+      "loss": 0.9846,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001989017751832352,
+      "loss": 0.8932,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019889326108248208,
+      "loss": 0.9535,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019888471428946094,
+      "loss": 0.8844,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019887613480699723,
+      "loss": 0.8562,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019886752263792714,
+      "loss": 0.8739,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001988588777850977,
+      "loss": 0.9218,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019885020025136677,
+      "loss": 0.8724,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019884149003960292,
+      "loss": 0.8532,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019883274715268564,
+      "loss": 0.9342,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019882397159350513,
+      "loss": 0.8824,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019881516336496243,
+      "loss": 0.3096,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019880632246996937,
+      "loss": 0.9278,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019879744891144864,
+      "loss": 0.3265,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019878854269233357,
+      "loss": 0.9275,
+      "step": 441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987796038155685,
+      "loss": 0.9381,
+      "step": 442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987706322841084,
+      "loss": 0.8607,
+      "step": 443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019876162810091908,
+      "loss": 0.9197,
+      "step": 444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987525912689772,
+      "loss": 0.9242,
+      "step": 445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019874352179127014,
+      "loss": 0.905,
+      "step": 446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019873441967079608,
+      "loss": 0.9207,
+      "step": 447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019872528491056405,
+      "loss": 0.8933,
+      "step": 448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987161175135938,
+      "loss": 0.8746,
+      "step": 449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987069174829159,
+      "loss": 0.9272,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019869768482157175,
+      "loss": 0.8379,
+      "step": 451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986884195326135,
+      "loss": 0.8572,
+      "step": 452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000198679121619104,
+      "loss": 0.9076,
+      "step": 453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000198669791084117,
+      "loss": 0.8832,
+      "step": 454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019866042793073707,
+      "loss": 0.8811,
+      "step": 455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986510321620594,
+      "loss": 0.889,
+      "step": 456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019864160378119014,
+      "loss": 0.9211,
+      "step": 457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019863214279124608,
+      "loss": 0.8825,
+      "step": 458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019862264919535486,
+      "loss": 0.9427,
+      "step": 459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986131229966549,
+      "loss": 0.8621,
+      "step": 460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019860356419829534,
+      "loss": 0.8785,
+      "step": 461
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985939728034362,
+      "loss": 0.8941,
+      "step": 462
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019858434881524816,
+      "loss": 0.9145,
+      "step": 463
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019857469223691276,
+      "loss": 0.8599,
+      "step": 464
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019856500307162232,
+      "loss": 0.9049,
+      "step": 465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019855528132257984,
+      "loss": 0.9626,
+      "step": 466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019854552699299918,
+      "loss": 0.9065,
+      "step": 467
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985357400861049,
+      "loss": 0.9038,
+      "step": 468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019852592060513242,
+      "loss": 0.8398,
+      "step": 469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019851606855332787,
+      "loss": 0.8391,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985061839339481,
+      "loss": 0.8842,
+      "step": 471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019849626675026087,
+      "loss": 0.9216,
+      "step": 472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019848631700554453,
+      "loss": 0.8808,
+      "step": 473
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019847633470308833,
+      "loss": 0.9023,
+      "step": 474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019846631984619224,
+      "loss": 0.844,
+      "step": 475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019845627243816693,
+      "loss": 0.9268,
+      "step": 476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019844619248233394,
+      "loss": 0.9371,
+      "step": 477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001984360799820255,
+      "loss": 0.8778,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001984259349405846,
+      "loss": 0.9037,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019841575736136502,
+      "loss": 0.9363,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019840554724773126,
+      "loss": 0.8774,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019839530460305862,
+      "loss": 0.8917,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019838502943073308,
+      "loss": 0.9097,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019837472173415147,
+      "loss": 0.8818,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019836438151672132,
+      "loss": 0.9094,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001983540087818609,
+      "loss": 0.9224,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001983436035329992,
+      "loss": 0.9041,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019833316577357607,
+      "loss": 0.8845,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019832269550704198,
+      "loss": 0.9597,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019831219273685826,
+      "loss": 0.8906,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019830165746649692,
+      "loss": 0.8309,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019829108969944068,
+      "loss": 0.9038,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019828048943918306,
+      "loss": 0.9544,
+      "step": 493
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019826985668922834,
+      "loss": 0.9113,
+      "step": 494
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982591914530915,
+      "loss": 0.9209,
+      "step": 495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019824849373429825,
+      "loss": 0.8845,
+      "step": 496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019823776353638506,
+      "loss": 0.948,
+      "step": 497
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019822700086289915,
+      "loss": 0.8092,
+      "step": 498
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019821620571739844,
+      "loss": 0.3087,
+      "step": 499
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019820537810345164,
+      "loss": 0.8902,
+      "step": 500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981945180246381,
+      "loss": 0.9107,
+      "step": 501
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000198183625484548,
+      "loss": 0.8675,
+      "step": 502
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981727004867822,
+      "loss": 0.9212,
+      "step": 503
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981617430349523,
+      "loss": 0.9151,
+      "step": 504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019815075313268064,
+      "loss": 0.8742,
+      "step": 505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019813973078360025,
+      "loss": 0.868,
+      "step": 506
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981286759913549,
+      "loss": 0.9056,
+      "step": 507
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981175887595992,
+      "loss": 0.9007,
+      "step": 508
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019810646909199827,
+      "loss": 0.9157,
+      "step": 509
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001980953169922281,
+      "loss": 0.8351,
+      "step": 510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019808413246397537,
+      "loss": 0.921,
+      "step": 511
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019807291551093747,
+      "loss": 0.9492,
+      "step": 512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019806166613682254,
+      "loss": 0.8591,
+      "step": 513
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001980503843453494,
+      "loss": 0.8864,
+      "step": 514
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019803907014024757,
+      "loss": 0.882,
+      "step": 515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019802772352525735,
+      "loss": 0.87,
+      "step": 516
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019801634450412973,
+      "loss": 0.8874,
+      "step": 517
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019800493308062635,
+      "loss": 0.8967,
+      "step": 518
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019799348925851967,
+      "loss": 0.9312,
+      "step": 519
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019798201304159282,
+      "loss": 0.9393,
+      "step": 520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001979705044336395,
+      "loss": 0.9097,
+      "step": 521
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019795896343846437,
+      "loss": 0.8806,
+      "step": 522
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019794739005988263,
+      "loss": 0.8996,
+      "step": 523
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019793578430172022,
+      "loss": 0.922,
+      "step": 524
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019792414616781376,
+      "loss": 0.8764,
+      "step": 525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019791247566201063,
+      "loss": 0.9025,
+      "step": 526
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019790077278816886,
+      "loss": 0.8887,
+      "step": 527
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019788903755015724,
+      "loss": 0.8919,
+      "step": 528
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019787726995185517,
+      "loss": 0.9073,
+      "step": 529
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019786546999715285,
+      "loss": 0.9098,
+      "step": 530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019785363768995111,
+      "loss": 0.9026,
+      "step": 531
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019784177303416148,
+      "loss": 0.9069,
+      "step": 532
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019782987603370617,
+      "loss": 0.8807,
+      "step": 533
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019781794669251817,
+      "loss": 0.8695,
+      "step": 534
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019780598501454102,
+      "loss": 0.8886,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001977939910037291,
+      "loss": 0.8432,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019778196466404736,
+      "loss": 0.9363,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019776990599947147,
+      "loss": 0.853,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019775781501398784,
+      "loss": 0.8992,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019774569171159353,
+      "loss": 0.9194,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019773353609629625,
+      "loss": 0.8939,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019772134817211442,
+      "loss": 0.8868,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019770912794307715,
+      "loss": 0.8862,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019769687541322422,
+      "loss": 0.9114,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019768459058660612,
+      "loss": 0.8786,
+      "step": 545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019767227346728392,
+      "loss": 0.9281,
+      "step": 546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019765992405932945,
+      "loss": 0.8866,
+      "step": 547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019764754236682524,
+      "loss": 0.8965,
+      "step": 548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019763512839386438,
+      "loss": 0.8885,
+      "step": 549
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019762268214455072,
+      "loss": 0.3043,
+      "step": 550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001976102036229988,
+      "loss": 0.9249,
+      "step": 551
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019759769283333377,
+      "loss": 0.9367,
+      "step": 552
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001975851497796914,
+      "loss": 0.8883,
+      "step": 553
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019757257446621827,
+      "loss": 0.871,
+      "step": 554
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019755996689707148,
+      "loss": 0.8865,
+      "step": 555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001975473270764189,
+      "loss": 0.8887,
+      "step": 556
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019753465500843903,
+      "loss": 0.8811,
+      "step": 557
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000197521950697321,
+      "loss": 0.9492,
+      "step": 558
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001975092141472646,
+      "loss": 0.8959,
+      "step": 559
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019749644536248031,
+      "loss": 0.8991,
+      "step": 560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019748364434718925,
+      "loss": 0.9095,
+      "step": 561
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019747081110562322,
+      "loss": 0.9298,
+      "step": 562
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019745794564202462,
+      "loss": 0.8752,
+      "step": 563
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019744504796064653,
+      "loss": 0.9383,
+      "step": 564
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019743211806575272,
+      "loss": 0.9039,
+      "step": 565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019741915596161756,
+      "loss": 0.9152,
+      "step": 566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019740616165252601,
+      "loss": 0.8626,
+      "step": 567
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019739313514277384,
+      "loss": 0.9142,
+      "step": 568
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019738007643666734,
+      "loss": 0.9184,
+      "step": 569
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001973669855385235,
+      "loss": 0.9058,
+      "step": 570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019735386245266985,
+      "loss": 0.8383,
+      "step": 571
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019734070718344468,
+      "loss": 0.9013,
+      "step": 572
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019732751973519693,
+      "loss": 0.7868,
+      "step": 573
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019731430011228604,
+      "loss": 0.9347,
+      "step": 574
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001973010483190822,
+      "loss": 0.8752,
+      "step": 575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019728776435996625,
+      "loss": 0.9397,
+      "step": 576
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019727444823932957,
+      "loss": 0.8759,
+      "step": 577
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019726109996157424,
+      "loss": 0.9069,
+      "step": 578
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001972477195311129,
+      "loss": 0.8683,
+      "step": 579
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019723430695236895,
+      "loss": 0.8695,
+      "step": 580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001972208622297763,
+      "loss": 0.9089,
+      "step": 581
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019720738536777951,
+      "loss": 0.8956,
+      "step": 582
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001971938763708338,
+      "loss": 0.8982,
+      "step": 583
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019718033524340504,
+      "loss": 0.8711,
+      "step": 584
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019716676198996958,
+      "loss": 0.8824,
+      "step": 585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001971531566150145,
+      "loss": 0.8785,
+      "step": 586
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001971395191230376,
+      "loss": 0.9675,
+      "step": 587
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019712584951854701,
+      "loss": 0.8636,
+      "step": 588
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001971121478060618,
+      "loss": 0.9293,
+      "step": 589
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001970984139901114,
+      "loss": 0.9258,
+      "step": 590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019708464807523596,
+      "loss": 0.9003,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019707085006598628,
+      "loss": 0.9227,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019705701996692367,
+      "loss": 0.8891,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019704315778262016,
+      "loss": 0.9172,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001970292635176583,
+      "loss": 0.9257,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019701533717663133,
+      "loss": 0.8777,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019700137876414294,
+      "loss": 0.9132,
+      "step": 597
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019698738828480758,
+      "loss": 0.9205,
+      "step": 598
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001969733657432502,
+      "loss": 0.8906,
+      "step": 599
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019695931114410646,
+      "loss": 0.8928,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001969452244920225,
+      "loss": 0.8598,
+      "step": 601
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019693110579165513,
+      "loss": 0.8689,
+      "step": 602
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001969169550476717,
+      "loss": 0.928,
+      "step": 603
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001969027722647502,
+      "loss": 0.8447,
+      "step": 604
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019688855744757916,
+      "loss": 0.8518,
+      "step": 605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001968743106008578,
+      "loss": 0.321,
+      "step": 606
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019686003172929582,
+      "loss": 0.8691,
+      "step": 607
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019684572083761352,
+      "loss": 0.9114,
+      "step": 608
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019683137793054188,
+      "loss": 0.879,
+      "step": 609
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019681700301282234,
+      "loss": 0.882,
+      "step": 610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019680259608920704,
+      "loss": 0.8676,
+      "step": 611
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019678815716445857,
+      "loss": 0.943,
+      "step": 612
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019677368624335025,
+      "loss": 0.8995,
+      "step": 613
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001967591833306658,
+      "loss": 0.8039,
+      "step": 614
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001967446484311997,
+      "loss": 0.8662,
+      "step": 615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019673008154975685,
+      "loss": 0.858,
+      "step": 616
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019671548269115287,
+      "loss": 0.9085,
+      "step": 617
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019670085186021375,
+      "loss": 0.9514,
+      "step": 618
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001966861890617763,
+      "loss": 0.8718,
+      "step": 619
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019667149430068766,
+      "loss": 0.9328,
+      "step": 620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001966567675818057,
+      "loss": 0.9091,
+      "step": 621
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019664200890999882,
+      "loss": 0.8655,
+      "step": 622
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019662721829014591,
+      "loss": 0.3406,
+      "step": 623
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001966123957271365,
+      "loss": 0.9033,
+      "step": 624
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019659754122587067,
+      "loss": 0.9266,
+      "step": 625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000196582654791259,
+      "loss": 0.3353,
+      "step": 626
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001965677364282227,
+      "loss": 0.8673,
+      "step": 627
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019655278614169345,
+      "loss": 0.8647,
+      "step": 628
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019653780393661362,
+      "loss": 0.8839,
+      "step": 629
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019652278981793596,
+      "loss": 0.9084,
+      "step": 630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019650774379062393,
+      "loss": 0.8721,
+      "step": 631
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019649266585965145,
+      "loss": 0.9574,
+      "step": 632
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019647755603000296,
+      "loss": 0.8916,
+      "step": 633
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019646241430667353,
+      "loss": 0.898,
+      "step": 634
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019644724069466872,
+      "loss": 0.8684,
+      "step": 635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019643203519900465,
+      "loss": 0.931,
+      "step": 636
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019641679782470792,
+      "loss": 0.9213,
+      "step": 637
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001964015285768158,
+      "loss": 0.8937,
+      "step": 638
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000196386227460376,
+      "loss": 0.8358,
+      "step": 639
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019637089448044676,
+      "loss": 0.8557,
+      "step": 640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019635552964209686,
+      "loss": 0.882,
+      "step": 641
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001963401329504057,
+      "loss": 0.9002,
+      "step": 642
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019632470441046306,
+      "loss": 0.9595,
+      "step": 643
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001963092440273694,
+      "loss": 0.8813,
+      "step": 644
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019629375180623554,
+      "loss": 0.8698,
+      "step": 645
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019627822775218303,
+      "loss": 0.8776,
+      "step": 646
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019626267187034376,
+      "loss": 0.8782,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019624708416586021,
+      "loss": 0.9242,
+      "step": 648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019623146464388543,
+      "loss": 0.8541,
+      "step": 649
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019621581330958295,
+      "loss": 0.8776,
+      "step": 650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019620013016812674,
+      "loss": 0.9006,
+      "step": 651
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001961844152247014,
+      "loss": 0.892,
+      "step": 652
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000196168668484502,
+      "loss": 0.8957,
+      "step": 653
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019615288995273412,
+      "loss": 0.9281,
+      "step": 654
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001961370796346138,
+      "loss": 0.8216,
+      "step": 655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001961212375353677,
+      "loss": 0.8984,
+      "step": 656
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001961053636602329,
+      "loss": 0.8705,
+      "step": 657
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000196089458014457,
+      "loss": 0.8649,
+      "step": 658
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001960735206032981,
+      "loss": 0.9344,
+      "step": 659
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019605755143202488,
+      "loss": 0.957,
+      "step": 660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019604155050591634,
+      "loss": 0.8682,
+      "step": 661
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019602551783026216,
+      "loss": 0.9528,
+      "step": 662
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019600945341036242,
+      "loss": 0.899,
+      "step": 663
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019599335725152775,
+      "loss": 0.8736,
+      "step": 664
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019597722935907918,
+      "loss": 0.8831,
+      "step": 665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019596106973834835,
+      "loss": 0.9373,
+      "step": 666
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001959448783946773,
+      "loss": 0.8571,
+      "step": 667
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019592865533341858,
+      "loss": 0.9156,
+      "step": 668
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019591240055993523,
+      "loss": 0.8633,
+      "step": 669
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001958961140796008,
+      "loss": 0.8363,
+      "step": 670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019587979589779928,
+      "loss": 0.8867,
+      "step": 671
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019586344601992515,
+      "loss": 0.9139,
+      "step": 672
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019584706445138344,
+      "loss": 0.8179,
+      "step": 673
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001958306511975895,
+      "loss": 0.9151,
+      "step": 674
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019581420626396928,
+      "loss": 0.8976,
+      "step": 675
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019579772965595918,
+      "loss": 0.826,
+      "step": 676
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019578122137900602,
+      "loss": 0.902,
+      "step": 677
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019576468143856719,
+      "loss": 0.3204,
+      "step": 678
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019574810984011043,
+      "loss": 0.855,
+      "step": 679
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019573150658911404,
+      "loss": 0.9234,
+      "step": 680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019571487169106673,
+      "loss": 0.8792,
+      "step": 681
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019569820515146768,
+      "loss": 0.3167,
+      "step": 682
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019568150697582657,
+      "loss": 0.9447,
+      "step": 683
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019566477716966344,
+      "loss": 0.3512,
+      "step": 684
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001956480157385089,
+      "loss": 0.9035,
+      "step": 685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000195631222687904,
+      "loss": 0.8274,
+      "step": 686
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019561439802340014,
+      "loss": 0.8947,
+      "step": 687
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019559754175055925,
+      "loss": 0.8881,
+      "step": 688
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019558065387495374,
+      "loss": 0.9015,
+      "step": 689
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001955637344021664,
+      "loss": 0.9029,
+      "step": 690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001955467833377905,
+      "loss": 0.8893,
+      "step": 691
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019552980068742977,
+      "loss": 0.9332,
+      "step": 692
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001955127864566983,
+      "loss": 0.8688,
+      "step": 693
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001954957406512207,
+      "loss": 0.9393,
+      "step": 694
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000195478663276632,
+      "loss": 0.8795,
+      "step": 695
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001954615543385777,
+      "loss": 0.9201,
+      "step": 696
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019544441384271365,
+      "loss": 0.9098,
+      "step": 697
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019542724179470616,
+      "loss": 0.854,
+      "step": 698
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019541003820023204,
+      "loss": 0.892,
+      "step": 699
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019539280306497844,
+      "loss": 0.9137,
+      "step": 700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019537553639464302,
+      "loss": 0.8952,
+      "step": 701
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019535823819493374,
+      "loss": 0.8842,
+      "step": 702
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019534090847156917,
+      "loss": 0.3211,
+      "step": 703
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001953235472302781,
+      "loss": 0.3614,
+      "step": 704
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019530615447679989,
+      "loss": 0.8442,
+      "step": 705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001952887302168842,
+      "loss": 0.8662,
+      "step": 706
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019527127445629122,
+      "loss": 0.8218,
+      "step": 707
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019525378720079147,
+      "loss": 0.8422,
+      "step": 708
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019523626845616593,
+      "loss": 0.8747,
+      "step": 709
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019521871822820598,
+      "loss": 0.8684,
+      "step": 710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019520113652271337,
+      "loss": 0.9023,
+      "step": 711
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001951835233455003,
+      "loss": 0.9003,
+      "step": 712
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019516587870238933,
+      "loss": 0.8683,
+      "step": 713
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019514820259921352,
+      "loss": 0.8547,
+      "step": 714
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019513049504181614,
+      "loss": 0.8965,
+      "step": 715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001951127560360511,
+      "loss": 0.8518,
+      "step": 716
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019509498558778254,
+      "loss": 0.9266,
+      "step": 717
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019507718370288503,
+      "loss": 0.8489,
+      "step": 718
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019505935038724356,
+      "loss": 0.348,
+      "step": 719
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001950414856467534,
+      "loss": 0.9085,
+      "step": 720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019502358948732048,
+      "loss": 0.8231,
+      "step": 721
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019500566191486075,
+      "loss": 0.8977,
+      "step": 722
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019498770293530085,
+      "loss": 0.8716,
+      "step": 723
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019496971255457765,
+      "loss": 0.8667,
+      "step": 724
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019495169077863843,
+      "loss": 0.8505,
+      "step": 725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019493363761344086,
+      "loss": 0.9099,
+      "step": 726
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019491555306495295,
+      "loss": 0.9382,
+      "step": 727
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019489743713915316,
+      "loss": 0.8404,
+      "step": 728
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019487928984203028,
+      "loss": 0.8394,
+      "step": 729
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019486111117958342,
+      "loss": 0.8748,
+      "step": 730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019484290115782214,
+      "loss": 0.8864,
+      "step": 731
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001948246597827663,
+      "loss": 0.8658,
+      "step": 732
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019480638706044623,
+      "loss": 0.3394,
+      "step": 733
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019478808299690247,
+      "loss": 0.9283,
+      "step": 734
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019476974759818604,
+      "loss": 0.8824,
+      "step": 735
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001947513808703583,
+      "loss": 0.9029,
+      "step": 736
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019473298281949092,
+      "loss": 0.8741,
+      "step": 737
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019471455345166595,
+      "loss": 0.8379,
+      "step": 738
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001946960927729758,
+      "loss": 0.8814,
+      "step": 739
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019467760078952325,
+      "loss": 0.8643,
+      "step": 740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019465907750742137,
+      "loss": 0.8724,
+      "step": 741
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019464052293279363,
+      "loss": 0.9513,
+      "step": 742
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019462193707177382,
+      "loss": 0.8953,
+      "step": 743
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019460331993050609,
+      "loss": 0.9,
+      "step": 744
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019458467151514487,
+      "loss": 0.9549,
+      "step": 745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019456599183185507,
+      "loss": 0.9562,
+      "step": 746
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001945472808868118,
+      "loss": 0.3273,
+      "step": 747
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001945285386862005,
+      "loss": 0.8983,
+      "step": 748
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019450976523621708,
+      "loss": 0.893,
+      "step": 749
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019449096054306763,
+      "loss": 0.8162,
+      "step": 750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019447212461296863,
+      "loss": 0.913,
+      "step": 751
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019445325745214695,
+      "loss": 0.8795,
+      "step": 752
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019443435906683966,
+      "loss": 0.8513,
+      "step": 753
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019441542946329422,
+      "loss": 0.9243,
+      "step": 754
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019439646864776843,
+      "loss": 0.9043,
+      "step": 755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001943774766265304,
+      "loss": 0.8723,
+      "step": 756
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019435845340585846,
+      "loss": 0.844,
+      "step": 757
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019433939899204142,
+      "loss": 0.8505,
+      "step": 758
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019432031339137822,
+      "loss": 0.9312,
+      "step": 759
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001943011966101783,
+      "loss": 0.8793,
+      "step": 760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019428204865476123,
+      "loss": 0.91,
+      "step": 761
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019426286953145704,
+      "loss": 0.8834,
+      "step": 762
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019424365924660593,
+      "loss": 0.888,
+      "step": 763
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001942244178065585,
+      "loss": 0.9679,
+      "step": 764
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001942051452176756,
+      "loss": 0.8776,
+      "step": 765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019418584148632836,
+      "loss": 0.8447,
+      "step": 766
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001941665066188983,
+      "loss": 0.8009,
+      "step": 767
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019414714062177712,
+      "loss": 0.8472,
+      "step": 768
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019412774350136686,
+      "loss": 0.867,
+      "step": 769
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019410831526407984,
+      "loss": 0.8579,
+      "step": 770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001940888559163387,
+      "loss": 0.8994,
+      "step": 771
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001940693654645763,
+      "loss": 0.8698,
+      "step": 772
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019404984391523585,
+      "loss": 0.8551,
+      "step": 773
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001940302912747708,
+      "loss": 0.8672,
+      "step": 774
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019401070754964493,
+      "loss": 0.8699,
+      "step": 775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019399109274633215,
+      "loss": 0.9256,
+      "step": 776
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019397144687131686,
+      "loss": 0.8351,
+      "step": 777
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019395176993109356,
+      "loss": 0.3508,
+      "step": 778
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001939320619321671,
+      "loss": 0.9408,
+      "step": 779
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019391232288105254,
+      "loss": 0.89,
+      "step": 780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019389255278427528,
+      "loss": 0.8944,
+      "step": 781
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019387275164837098,
+      "loss": 0.8726,
+      "step": 782
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019385291947988543,
+      "loss": 0.9026,
+      "step": 783
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019383305628537485,
+      "loss": 0.8698,
+      "step": 784
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001938131620714056,
+      "loss": 0.8418,
+      "step": 785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001937932368445544,
+      "loss": 0.9278,
+      "step": 786
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019377328061140806,
+      "loss": 0.8678,
+      "step": 787
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019375329337856383,
+      "loss": 0.9196,
+      "step": 788
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019373327515262906,
+      "loss": 0.8403,
+      "step": 789
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001937132259402214,
+      "loss": 0.9464,
+      "step": 790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001936931457479688,
+      "loss": 0.8072,
+      "step": 791
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019367303458250938,
+      "loss": 0.8522,
+      "step": 792
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001936528924504915,
+      "loss": 0.8511,
+      "step": 793
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019363271935857372,
+      "loss": 0.3261,
+      "step": 794
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000193612515313425,
+      "loss": 0.8338,
+      "step": 795
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019359228032172433,
+      "loss": 0.8621,
+      "step": 796
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019357201439016106,
+      "loss": 0.8352,
+      "step": 797
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019355171752543472,
+      "loss": 0.8356,
+      "step": 798
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019353138973425508,
+      "loss": 0.8217,
+      "step": 799
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019351103102334212,
+      "loss": 0.3243,
+      "step": 800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019349064139942605,
+      "loss": 0.8891,
+      "step": 801
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019347022086924732,
+      "loss": 0.9383,
+      "step": 802
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019344976943955658,
+      "loss": 0.3142,
+      "step": 803
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019342928711711465,
+      "loss": 0.7721,
+      "step": 804
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019340877390869264,
+      "loss": 0.9008,
+      "step": 805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001933882298210718,
+      "loss": 0.8637,
+      "step": 806
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019336765486104365,
+      "loss": 0.8836,
+      "step": 807
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001933470490354099,
+      "loss": 0.3391,
+      "step": 808
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019332641235098242,
+      "loss": 0.8957,
+      "step": 809
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019330574481458333,
+      "loss": 0.8726,
+      "step": 810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019328504643304493,
+      "loss": 0.8751,
+      "step": 811
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019326431721320973,
+      "loss": 0.8938,
+      "step": 812
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019324355716193037,
+      "loss": 0.334,
+      "step": 813
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001932227662860698,
+      "loss": 0.8468,
+      "step": 814
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019320194459250108,
+      "loss": 0.2899,
+      "step": 815
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019318109208810746,
+      "loss": 0.9029,
+      "step": 816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001931602087797824,
+      "loss": 0.8882,
+      "step": 817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019313929467442952,
+      "loss": 0.8989,
+      "step": 818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019311834977896265,
+      "loss": 0.8826,
+      "step": 819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019309737410030578,
+      "loss": 0.8997,
+      "step": 820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019307636764539306,
+      "loss": 0.8857,
+      "step": 821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019305533042116883,
+      "loss": 0.9011,
+      "step": 822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019303426243458765,
+      "loss": 0.8987,
+      "step": 823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019301316369261414,
+      "loss": 0.866,
+      "step": 824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001929920342022232,
+      "loss": 0.8787,
+      "step": 825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019297087397039984,
+      "loss": 0.9215,
+      "step": 826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019294968300413918,
+      "loss": 0.8791,
+      "step": 827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019292846131044664,
+      "loss": 0.8864,
+      "step": 828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019290720889633768,
+      "loss": 0.8749,
+      "step": 829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019288592576883793,
+      "loss": 0.852,
+      "step": 830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019286461193498324,
+      "loss": 0.8278,
+      "step": 831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019284326740181952,
+      "loss": 0.8429,
+      "step": 832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001928218921764029,
+      "loss": 0.8991,
+      "step": 833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019280048626579962,
+      "loss": 0.8841,
+      "step": 834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019277904967708614,
+      "loss": 0.7942,
+      "step": 835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019275758241734886,
+      "loss": 0.9371,
+      "step": 836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001927360844936846,
+      "loss": 0.9089,
+      "step": 837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019271455591320007,
+      "loss": 0.9133,
+      "step": 838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019269299668301228,
+      "loss": 0.9089,
+      "step": 839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001926714068102483,
+      "loss": 0.8698,
+      "step": 840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019264978630204526,
+      "loss": 0.8699,
+      "step": 841
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001926281351655506,
+      "loss": 0.8131,
+      "step": 842
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019260645340792175,
+      "loss": 0.8369,
+      "step": 843
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019258474103632625,
+      "loss": 0.9608,
+      "step": 844
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001925629980579419,
+      "loss": 0.8668,
+      "step": 845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019254122447995645,
+      "loss": 0.9349,
+      "step": 846
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019251942030956785,
+      "loss": 0.9356,
+      "step": 847
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019249758555398412,
+      "loss": 0.8697,
+      "step": 848
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019247572022042348,
+      "loss": 0.8503,
+      "step": 849
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001924538243161142,
+      "loss": 0.8729,
+      "step": 850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019243189784829464,
+      "loss": 0.886,
+      "step": 851
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019240994082421326,
+      "loss": 0.8277,
+      "step": 852
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001923879532511287,
+      "loss": 0.9004,
+      "step": 853
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001923659351363096,
+      "loss": 0.8243,
+      "step": 854
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001923438864870347,
+      "loss": 0.8166,
+      "step": 855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019232180731059293,
+      "loss": 0.8637,
+      "step": 856
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019229969761428323,
+      "loss": 0.8372,
+      "step": 857
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001922775574054147,
+      "loss": 0.9382,
+      "step": 858
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019225538669130643,
+      "loss": 0.8635,
+      "step": 859
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019223318547928762,
+      "loss": 0.9304,
+      "step": 860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019221095377669765,
+      "loss": 0.8914,
+      "step": 861
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001921886915908859,
+      "loss": 0.8619,
+      "step": 862
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019216639892921173,
+      "loss": 0.9036,
+      "step": 863
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001921440757990448,
+      "loss": 0.8627,
+      "step": 864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001921217222077647,
+      "loss": 0.8789,
+      "step": 865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019209933816276102,
+      "loss": 0.8464,
+      "step": 866
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019207692367143362,
+      "loss": 0.908,
+      "step": 867
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019205447874119224,
+      "loss": 0.3077,
+      "step": 868
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019203200337945678,
+      "loss": 0.8645,
+      "step": 869
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019200949759365718,
+      "loss": 0.8124,
+      "step": 870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019198696139123342,
+      "loss": 0.8465,
+      "step": 871
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019196439477963556,
+      "loss": 0.8992,
+      "step": 872
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001919417977663237,
+      "loss": 0.8387,
+      "step": 873
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019191917035876798,
+      "loss": 0.882,
+      "step": 874
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019189651256444862,
+      "loss": 0.8522,
+      "step": 875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019187382439085586,
+      "loss": 0.8753,
+      "step": 876
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019185110584548997,
+      "loss": 0.8939,
+      "step": 877
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019182835693586127,
+      "loss": 0.9037,
+      "step": 878
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019180557766949018,
+      "loss": 0.8887,
+      "step": 879
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019178276805390703,
+      "loss": 0.8092,
+      "step": 880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001917599280966523,
+      "loss": 0.8505,
+      "step": 881
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019173705780527642,
+      "loss": 0.3412,
+      "step": 882
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019171415718733991,
+      "loss": 0.9581,
+      "step": 883
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019169122625041328,
+      "loss": 0.8596,
+      "step": 884
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001916682650020771,
+      "loss": 0.8768,
+      "step": 885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019164527344992186,
+      "loss": 0.3229,
+      "step": 886
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019162225160154818,
+      "loss": 0.8711,
+      "step": 887
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019159919946456667,
+      "loss": 0.8394,
+      "step": 888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019157611704659792,
+      "loss": 0.971,
+      "step": 889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019155300435527256,
+      "loss": 0.8733,
+      "step": 890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019152986139823116,
+      "loss": 0.8876,
+      "step": 891
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001915066881831244,
+      "loss": 0.9015,
+      "step": 892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001914834847176129,
+      "loss": 0.8975,
+      "step": 893
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019146025100936733,
+      "loss": 0.8815,
+      "step": 894
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019143698706606825,
+      "loss": 0.8991,
+      "step": 895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019141369289540637,
+      "loss": 0.807,
+      "step": 896
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001913903685050822,
+      "loss": 0.8473,
+      "step": 897
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019136701390280644,
+      "loss": 0.815,
+      "step": 898
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019134362909629967,
+      "loss": 0.3468,
+      "step": 899
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019132021409329242,
+      "loss": 0.8942,
+      "step": 900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001912967689015253,
+      "loss": 0.8948,
+      "step": 901
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019127329352874886,
+      "loss": 0.9245,
+      "step": 902
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019124978798272358,
+      "loss": 0.8824,
+      "step": 903
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019122625227122002,
+      "loss": 0.823,
+      "step": 904
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019120268640201857,
+      "loss": 0.907,
+      "step": 905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019117909038290974,
+      "loss": 0.8569,
+      "step": 906
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019115546422169386,
+      "loss": 0.8958,
+      "step": 907
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019113180792618132,
+      "loss": 0.8295,
+      "step": 908
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001911081215041925,
+      "loss": 0.9222,
+      "step": 909
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019108440496355767,
+      "loss": 0.8376,
+      "step": 910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000191060658312117,
+      "loss": 0.8906,
+      "step": 911
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019103688155772082,
+      "loss": 0.8868,
+      "step": 912
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019101307470822918,
+      "loss": 0.8732,
+      "step": 913
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019098923777151222,
+      "loss": 0.9215,
+      "step": 914
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019096537075544997,
+      "loss": 0.9051,
+      "step": 915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019094147366793243,
+      "loss": 0.857,
+      "step": 916
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019091754651685952,
+      "loss": 0.903,
+      "step": 917
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019089358931014114,
+      "loss": 0.8842,
+      "step": 918
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019086960205569708,
+      "loss": 0.9173,
+      "step": 919
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019084558476145706,
+      "loss": 0.9066,
+      "step": 920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001908215374353608,
+      "loss": 0.8942,
+      "step": 921
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019079746008535784,
+      "loss": 0.8357,
+      "step": 922
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019077335271940773,
+      "loss": 0.8551,
+      "step": 923
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001907492153454799,
+      "loss": 0.9085,
+      "step": 924
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019072504797155375,
+      "loss": 0.8813,
+      "step": 925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019070085060561852,
+      "loss": 0.8416,
+      "step": 926
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019067662325567345,
+      "loss": 0.8518,
+      "step": 927
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001906523659297276,
+      "loss": 0.9137,
+      "step": 928
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019062807863580005,
+      "loss": 0.3324,
+      "step": 929
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001906037613819197,
+      "loss": 0.8734,
+      "step": 930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019057941417612535,
+      "loss": 0.9157,
+      "step": 931
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019055503702646576,
+      "loss": 0.899,
+      "step": 932
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001905306299409996,
+      "loss": 0.9592,
+      "step": 933
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001905061929277953,
+      "loss": 0.8459,
+      "step": 934
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019048172599493138,
+      "loss": 0.8281,
+      "step": 935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019045722915049607,
+      "loss": 0.8723,
+      "step": 936
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019043270240258763,
+      "loss": 0.886,
+      "step": 937
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019040814575931413,
+      "loss": 0.8654,
+      "step": 938
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019038355922879353,
+      "loss": 0.8572,
+      "step": 939
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019035894281915368,
+      "loss": 0.8902,
+      "step": 940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019033429653853232,
+      "loss": 0.8608,
+      "step": 941
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019030962039507702,
+      "loss": 0.8422,
+      "step": 942
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019028491439694527,
+      "loss": 0.8479,
+      "step": 943
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019026017855230444,
+      "loss": 0.9019,
+      "step": 944
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019023541286933168,
+      "loss": 0.931,
+      "step": 945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019021061735621412,
+      "loss": 0.9189,
+      "step": 946
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019018579202114867,
+      "loss": 0.9145,
+      "step": 947
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001901609368723421,
+      "loss": 0.3139,
+      "step": 948
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001901360519180111,
+      "loss": 0.845,
+      "step": 949
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019011113716638217,
+      "loss": 0.8963,
+      "step": 950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001900861926256916,
+      "loss": 0.896,
+      "step": 951
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019006121830418565,
+      "loss": 0.9051,
+      "step": 952
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019003621421012038,
+      "loss": 0.8773,
+      "step": 953
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019001118035176162,
+      "loss": 0.8826,
+      "step": 954
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018998611673738515,
+      "loss": 0.8744,
+      "step": 955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018996102337527648,
+      "loss": 0.8959,
+      "step": 956
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018993590027373104,
+      "loss": 0.8202,
+      "step": 957
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018991074744105407,
+      "loss": 0.9362,
+      "step": 958
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001898855648855606,
+      "loss": 0.8728,
+      "step": 959
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018986035261557552,
+      "loss": 0.8987,
+      "step": 960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018983511063943356,
+      "loss": 0.8538,
+      "step": 961
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001898098389654792,
+      "loss": 0.9099,
+      "step": 962
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018978453760206684,
+      "loss": 0.893,
+      "step": 963
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001897592065575606,
+      "loss": 0.9331,
+      "step": 964
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018973384584033447,
+      "loss": 0.9088,
+      "step": 965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001897084554587722,
+      "loss": 0.8428,
+      "step": 966
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018968303542126743,
+      "loss": 0.8265,
+      "step": 967
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001896575857362235,
+      "loss": 0.9125,
+      "step": 968
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001896321064120536,
+      "loss": 0.8514,
+      "step": 969
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001896065974571808,
+      "loss": 0.8309,
+      "step": 970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018958105888003778,
+      "loss": 0.8223,
+      "step": 971
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018955549068906717,
+      "loss": 0.8558,
+      "step": 972
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018952989289272134,
+      "loss": 0.8753,
+      "step": 973
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001895042654994624,
+      "loss": 0.8841,
+      "step": 974
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018947860851776236,
+      "loss": 0.8787,
+      "step": 975
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018945292195610288,
+      "loss": 0.8482,
+      "step": 976
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001894272058229755,
+      "loss": 0.8842,
+      "step": 977
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018940146012688146,
+      "loss": 0.8782,
+      "step": 978
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018937568487633184,
+      "loss": 0.8443,
+      "step": 979
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001893498800798474,
+      "loss": 0.8912,
+      "step": 980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001893240457459588,
+      "loss": 0.357,
+      "step": 981
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018929818188320635,
+      "loss": 0.8782,
+      "step": 982
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018927228850014013,
+      "loss": 0.9298,
+      "step": 983
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018924636560532006,
+      "loss": 0.8921,
+      "step": 984
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018922041320731578,
+      "loss": 0.8136,
+      "step": 985
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018919443131470658,
+      "loss": 0.8497,
+      "step": 986
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001891684199360817,
+      "loss": 0.9151,
+      "step": 987
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001891423790800399,
+      "loss": 0.8811,
+      "step": 988
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018911630875518992,
+      "loss": 0.9558,
+      "step": 989
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018909020897015004,
+      "loss": 0.868,
+      "step": 990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001890640797335484,
+      "loss": 0.884,
+      "step": 991
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018903792105402282,
+      "loss": 0.875,
+      "step": 992
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018901173294022088,
+      "loss": 0.8697,
+      "step": 993
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018898551540079989,
+      "loss": 0.8568,
+      "step": 994
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018895926844442686,
+      "loss": 0.8815,
+      "step": 995
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018893299207977857,
+      "loss": 0.9471,
+      "step": 996
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018890668631554148,
+      "loss": 0.9122,
+      "step": 997
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018888035116041175,
+      "loss": 0.8878,
+      "step": 998
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018885398662309537,
+      "loss": 0.3447,
+      "step": 999
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001888275927123079,
+      "loss": 0.8534,
+      "step": 1000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001888011694367747,
+      "loss": 0.912,
+      "step": 1001
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018877471680523082,
+      "loss": 0.3276,
+      "step": 1002
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000188748234826421,
+      "loss": 0.8963,
+      "step": 1003
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018872172350909968,
+      "loss": 0.8958,
+      "step": 1004
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000188695182862031,
+      "loss": 0.8512,
+      "step": 1005
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018866861289398883,
+      "loss": 0.8317,
+      "step": 1006
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018864201361375667,
+      "loss": 0.7918,
+      "step": 1007
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001886153850301278,
+      "loss": 0.809,
+      "step": 1008
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018858872715190507,
+      "loss": 0.8229,
+      "step": 1009
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018856203998790112,
+      "loss": 0.8552,
+      "step": 1010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018853532354693822,
+      "loss": 0.8093,
+      "step": 1011
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001885085778378483,
+      "loss": 0.9041,
+      "step": 1012
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018848180286947303,
+      "loss": 0.8259,
+      "step": 1013
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018845499865066372,
+      "loss": 0.8894,
+      "step": 1014
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001884281651902813,
+      "loss": 0.8565,
+      "step": 1015
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018840130249719644,
+      "loss": 0.8674,
+      "step": 1016
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018837441058028946,
+      "loss": 0.8473,
+      "step": 1017
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018834748944845028,
+      "loss": 0.9202,
+      "step": 1018
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001883205391105786,
+      "loss": 0.8964,
+      "step": 1019
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018829355957558362,
+      "loss": 0.8647,
+      "step": 1020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001882665508523843,
+      "loss": 0.2963,
+      "step": 1021
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018823951294990923,
+      "loss": 0.8654,
+      "step": 1022
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018821244587709663,
+      "loss": 0.8354,
+      "step": 1023
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001881853496428944,
+      "loss": 0.8222,
+      "step": 1024
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018815822425625998,
+      "loss": 0.8775,
+      "step": 1025
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018813106972616055,
+      "loss": 0.7821,
+      "step": 1026
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018810388606157292,
+      "loss": 0.8561,
+      "step": 1027
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018807667327148345,
+      "loss": 0.7904,
+      "step": 1028
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018804943136488826,
+      "loss": 0.9139,
+      "step": 1029
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018802216035079293,
+      "loss": 0.8633,
+      "step": 1030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001879948602382128,
+      "loss": 0.8736,
+      "step": 1031
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018796753103617278,
+      "loss": 0.8094,
+      "step": 1032
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018794017275370736,
+      "loss": 0.8969,
+      "step": 1033
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001879127853998607,
+      "loss": 0.9061,
+      "step": 1034
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018788536898368654,
+      "loss": 0.8382,
+      "step": 1035
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018785792351424827,
+      "loss": 0.8435,
+      "step": 1036
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001878304490006188,
+      "loss": 0.8541,
+      "step": 1037
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001878029454518807,
+      "loss": 0.8256,
+      "step": 1038
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018777541287712618,
+      "loss": 0.948,
+      "step": 1039
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018774785128545694,
+      "loss": 0.8558,
+      "step": 1040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018772026068598433,
+      "loss": 0.8546,
+      "step": 1041
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018769264108782933,
+      "loss": 0.8851,
+      "step": 1042
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018766499250012244,
+      "loss": 0.9036,
+      "step": 1043
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018763731493200375,
+      "loss": 0.8484,
+      "step": 1044
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018760960839262297,
+      "loss": 0.9394,
+      "step": 1045
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018758187289113937,
+      "loss": 0.3397,
+      "step": 1046
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018755410843672177,
+      "loss": 0.9114,
+      "step": 1047
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018752631503854864,
+      "loss": 0.8966,
+      "step": 1048
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018749849270580785,
+      "loss": 0.8593,
+      "step": 1049
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018747064144769703,
+      "loss": 0.8453,
+      "step": 1050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018744276127342327,
+      "loss": 0.8862,
+      "step": 1051
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001874148521922032,
+      "loss": 0.9112,
+      "step": 1052
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018738691421326306,
+      "loss": 0.8273,
+      "step": 1053
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018735894734583867,
+      "loss": 0.9041,
+      "step": 1054
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001873309515991753,
+      "loss": 0.832,
+      "step": 1055
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018730292698252785,
+      "loss": 0.8616,
+      "step": 1056
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018727487350516072,
+      "loss": 0.903,
+      "step": 1057
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001872467911763479,
+      "loss": 0.8869,
+      "step": 1058
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018721868000537282,
+      "loss": 0.8528,
+      "step": 1059
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018719054000152855,
+      "loss": 0.8549,
+      "step": 1060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018716237117411767,
+      "loss": 0.8843,
+      "step": 1061
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018713417353245223,
+      "loss": 0.8168,
+      "step": 1062
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018710594708585383,
+      "loss": 0.9161,
+      "step": 1063
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018707769184365367,
+      "loss": 0.8539,
+      "step": 1064
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018704940781519235,
+      "loss": 0.8255,
+      "step": 1065
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001870210950098201,
+      "loss": 0.8377,
+      "step": 1066
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018699275343689656,
+      "loss": 0.905,
+      "step": 1067
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018696438310579093,
+      "loss": 0.8946,
+      "step": 1068
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018693598402588194,
+      "loss": 0.8629,
+      "step": 1069
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018690755620655774,
+      "loss": 0.9052,
+      "step": 1070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018687909965721606,
+      "loss": 0.9146,
+      "step": 1071
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018685061438726414,
+      "loss": 0.937,
+      "step": 1072
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018682210040611868,
+      "loss": 0.8508,
+      "step": 1073
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018679355772320585,
+      "loss": 0.9442,
+      "step": 1074
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018676498634796128,
+      "loss": 0.8272,
+      "step": 1075
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018673638628983018,
+      "loss": 0.9507,
+      "step": 1076
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018670775755826722,
+      "loss": 0.8431,
+      "step": 1077
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018667910016273648,
+      "loss": 0.8501,
+      "step": 1078
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018665041411271158,
+      "loss": 0.9209,
+      "step": 1079
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018662169941767562,
+      "loss": 0.8723,
+      "step": 1080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001865929560871211,
+      "loss": 0.9022,
+      "step": 1081
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018656418413055007,
+      "loss": 0.9228,
+      "step": 1082
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018653538355747394,
+      "loss": 0.9264,
+      "step": 1083
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018650655437741368,
+      "loss": 0.8002,
+      "step": 1084
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001864776965998997,
+      "loss": 0.873,
+      "step": 1085
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018644881023447177,
+      "loss": 0.856,
+      "step": 1086
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018641989529067933,
+      "loss": 0.8569,
+      "step": 1087
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018639095177808095,
+      "loss": 0.8494,
+      "step": 1088
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001863619797062449,
+      "loss": 0.8578,
+      "step": 1089
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001863329790847488,
+      "loss": 0.8425,
+      "step": 1090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018630394992317977,
+      "loss": 0.8264,
+      "step": 1091
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018627489223113422,
+      "loss": 0.8037,
+      "step": 1092
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001862458060182181,
+      "loss": 0.8856,
+      "step": 1093
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001862166912940468,
+      "loss": 0.9131,
+      "step": 1094
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018618754806824507,
+      "loss": 0.8876,
+      "step": 1095
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018615837635044716,
+      "loss": 0.8487,
+      "step": 1096
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018612917615029665,
+      "loss": 0.8896,
+      "step": 1097
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001860999474774466,
+      "loss": 0.9137,
+      "step": 1098
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018607069034155948,
+      "loss": 0.8735,
+      "step": 1099
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018604140475230715,
+      "loss": 0.8763,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018601209071937087,
+      "loss": 0.91,
+      "step": 1101
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001859827482524413,
+      "loss": 0.8246,
+      "step": 1102
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018595337736121854,
+      "loss": 0.8694,
+      "step": 1103
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018592397805541205,
+      "loss": 0.8865,
+      "step": 1104
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018589455034474068,
+      "loss": 0.8986,
+      "step": 1105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018586509423893267,
+      "loss": 0.9047,
+      "step": 1106
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018583560974772572,
+      "loss": 0.8036,
+      "step": 1107
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018580609688086678,
+      "loss": 0.7817,
+      "step": 1108
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018577655564811228,
+      "loss": 0.8435,
+      "step": 1109
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000185746986059228,
+      "loss": 0.9147,
+      "step": 1110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018571738812398912,
+      "loss": 0.8838,
+      "step": 1111
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018568776185218016,
+      "loss": 0.843,
+      "step": 1112
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000185658107253595,
+      "loss": 0.9026,
+      "step": 1113
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018562842433803687,
+      "loss": 0.8315,
+      "step": 1114
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018559871311531843,
+      "loss": 0.8416,
+      "step": 1115
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018556897359526162,
+      "loss": 0.8668,
+      "step": 1116
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001855392057876978,
+      "loss": 0.8862,
+      "step": 1117
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001855094097024677,
+      "loss": 0.8392,
+      "step": 1118
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018547958534942123,
+      "loss": 0.8421,
+      "step": 1119
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018544973273841784,
+      "loss": 0.8799,
+      "step": 1120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018541985187932626,
+      "loss": 0.9488,
+      "step": 1121
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018538994278202448,
+      "loss": 0.8732,
+      "step": 1122
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018536000545639995,
+      "loss": 0.931,
+      "step": 1123
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018533003991234937,
+      "loss": 0.8451,
+      "step": 1124
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001853000461597788,
+      "loss": 0.9258,
+      "step": 1125
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018527002420860362,
+      "loss": 0.3445,
+      "step": 1126
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018523997406874848,
+      "loss": 0.867,
+      "step": 1127
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018520989575014746,
+      "loss": 0.8925,
+      "step": 1128
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018517978926274386,
+      "loss": 0.9227,
+      "step": 1129
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001851496546164903,
+      "loss": 0.9315,
+      "step": 1130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001851194918213488,
+      "loss": 0.8691,
+      "step": 1131
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018508930088729052,
+      "loss": 0.8802,
+      "step": 1132
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001850590818242961,
+      "loss": 0.8725,
+      "step": 1133
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001850288346423554,
+      "loss": 0.825,
+      "step": 1134
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001849985593514675,
+      "loss": 0.8152,
+      "step": 1135
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018496825596164094,
+      "loss": 0.7978,
+      "step": 1136
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018493792448289338,
+      "loss": 0.9123,
+      "step": 1137
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018490756492525187,
+      "loss": 0.9021,
+      "step": 1138
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001848771772987527,
+      "loss": 0.8641,
+      "step": 1139
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001848467616134415,
+      "loss": 0.8582,
+      "step": 1140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001848163178793731,
+      "loss": 0.8443,
+      "step": 1141
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001847858461066116,
+      "loss": 0.8333,
+      "step": 1142
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018475534630523047,
+      "loss": 0.9588,
+      "step": 1143
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018472481848531226,
+      "loss": 0.8896,
+      "step": 1144
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018469426265694904,
+      "loss": 0.9175,
+      "step": 1145
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018466367883024186,
+      "loss": 0.8617,
+      "step": 1146
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001846330670153013,
+      "loss": 0.8796,
+      "step": 1147
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018460242722224694,
+      "loss": 0.8488,
+      "step": 1148
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018457175946120778,
+      "loss": 0.8207,
+      "step": 1149
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018454106374232197,
+      "loss": 0.8561,
+      "step": 1150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018451034007573703,
+      "loss": 0.8292,
+      "step": 1151
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018447958847160953,
+      "loss": 0.8723,
+      "step": 1152
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018444880894010545,
+      "loss": 0.907,
+      "step": 1153
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018441800149139988,
+      "loss": 0.8508,
+      "step": 1154
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001843871661356772,
+      "loss": 0.8552,
+      "step": 1155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000184356302883131,
+      "loss": 0.8034,
+      "step": 1156
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018432541174396411,
+      "loss": 0.9139,
+      "step": 1157
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001842944927283886,
+      "loss": 0.8934,
+      "step": 1158
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018426354584662565,
+      "loss": 0.7996,
+      "step": 1159
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018423257110890574,
+      "loss": 0.8499,
+      "step": 1160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018420156852546859,
+      "loss": 0.8345,
+      "step": 1161
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018417053810656302,
+      "loss": 0.8274,
+      "step": 1162
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018413947986244713,
+      "loss": 0.9107,
+      "step": 1163
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018410839380338817,
+      "loss": 0.8662,
+      "step": 1164
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018407727993966267,
+      "loss": 0.9166,
+      "step": 1165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018404613828155623,
+      "loss": 0.852,
+      "step": 1166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018401496883936375,
+      "loss": 0.8672,
+      "step": 1167
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018398377162338924,
+      "loss": 0.8953,
+      "step": 1168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018395254664394593,
+      "loss": 0.9201,
+      "step": 1169
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001839212939113562,
+      "loss": 0.8358,
+      "step": 1170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018389001343595163,
+      "loss": 0.8883,
+      "step": 1171
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018385870522807295,
+      "loss": 0.8321,
+      "step": 1172
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018382736929807008,
+      "loss": 0.8778,
+      "step": 1173
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018379600565630213,
+      "loss": 0.8922,
+      "step": 1174
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001837646143131373,
+      "loss": 0.8825,
+      "step": 1175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018373319527895294,
+      "loss": 0.925,
+      "step": 1176
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018370174856413568,
+      "loss": 0.8757,
+      "step": 1177
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018367027417908117,
+      "loss": 0.9165,
+      "step": 1178
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018363877213419425,
+      "loss": 0.8743,
+      "step": 1179
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001836072424398889,
+      "loss": 0.9016,
+      "step": 1180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018357568510658833,
+      "loss": 0.8766,
+      "step": 1181
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001835441001447247,
+      "loss": 0.8357,
+      "step": 1182
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018351248756473946,
+      "loss": 0.8992,
+      "step": 1183
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001834808473770832,
+      "loss": 0.9222,
+      "step": 1184
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018344917959221545,
+      "loss": 0.9322,
+      "step": 1185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018341748422060503,
+      "loss": 0.8509,
+      "step": 1186
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018338576127272988,
+      "loss": 0.8657,
+      "step": 1187
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000183354010759077,
+      "loss": 0.8527,
+      "step": 1188
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018332223269014252,
+      "loss": 0.8484,
+      "step": 1189
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018329042707643164,
+      "loss": 0.8291,
+      "step": 1190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001832585939284588,
+      "loss": 0.9436,
+      "step": 1191
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001832267332567473,
+      "loss": 0.8751,
+      "step": 1192
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001831948450718298,
+      "loss": 0.8812,
+      "step": 1193
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018316292938424787,
+      "loss": 0.8594,
+      "step": 1194
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001831309862045523,
+      "loss": 0.8306,
+      "step": 1195
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018309901554330288,
+      "loss": 0.8667,
+      "step": 1196
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018306701741106852,
+      "loss": 0.9316,
+      "step": 1197
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001830349918184272,
+      "loss": 0.9205,
+      "step": 1198
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018300293877596596,
+      "loss": 0.3403,
+      "step": 1199
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000182970858294281,
+      "loss": 0.8762,
+      "step": 1200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018293875038397745,
+      "loss": 0.89,
+      "step": 1201
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018290661505566963,
+      "loss": 0.8862,
+      "step": 1202
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001828744523199809,
+      "loss": 0.8998,
+      "step": 1203
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018284226218754363,
+      "loss": 0.9072,
+      "step": 1204
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018281004466899931,
+      "loss": 0.8287,
+      "step": 1205
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001827777997749984,
+      "loss": 0.8416,
+      "step": 1206
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018274552751620048,
+      "loss": 0.9062,
+      "step": 1207
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001827132279032742,
+      "loss": 0.8496,
+      "step": 1208
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018268090094689717,
+      "loss": 0.8949,
+      "step": 1209
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018264854665775605,
+      "loss": 0.8965,
+      "step": 1210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018261616504654663,
+      "loss": 0.3235,
+      "step": 1211
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018258375612397365,
+      "loss": 0.8909,
+      "step": 1212
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018255131990075084,
+      "loss": 0.8829,
+      "step": 1213
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018251885638760105,
+      "loss": 0.8818,
+      "step": 1214
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018248636559525615,
+      "loss": 0.8545,
+      "step": 1215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018245384753445693,
+      "loss": 0.8192,
+      "step": 1216
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001824213022159533,
+      "loss": 0.8829,
+      "step": 1217
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001823887296505041,
+      "loss": 0.883,
+      "step": 1218
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018235612984887723,
+      "loss": 0.9073,
+      "step": 1219
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018232350282184956,
+      "loss": 0.8315,
+      "step": 1220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018229084858020696,
+      "loss": 0.8739,
+      "step": 1221
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001822581671347444,
+      "loss": 0.833,
+      "step": 1222
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018222545849626566,
+      "loss": 0.8857,
+      "step": 1223
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001821927226755837,
+      "loss": 0.8599,
+      "step": 1224
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018215995968352028,
+      "loss": 0.8871,
+      "step": 1225
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018212716953090624,
+      "loss": 0.9458,
+      "step": 1226
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018209435222858147,
+      "loss": 0.9069,
+      "step": 1227
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001820615077873947,
+      "loss": 0.8994,
+      "step": 1228
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001820286362182037,
+      "loss": 0.8396,
+      "step": 1229
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001819957375318752,
+      "loss": 0.8684,
+      "step": 1230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018196281173928492,
+      "loss": 0.9027,
+      "step": 1231
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018192985885131743,
+      "loss": 0.8581,
+      "step": 1232
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001818968788788664,
+      "loss": 0.7835,
+      "step": 1233
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018186387183283443,
+      "loss": 0.8677,
+      "step": 1234
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000181830837724133,
+      "loss": 0.9271,
+      "step": 1235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018179777656368253,
+      "loss": 0.815,
+      "step": 1236
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018176468836241248,
+      "loss": 0.8144,
+      "step": 1237
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018173157313126114,
+      "loss": 0.8801,
+      "step": 1238
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018169843088117584,
+      "loss": 0.8529,
+      "step": 1239
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018166526162311276,
+      "loss": 0.9089,
+      "step": 1240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018163206536803706,
+      "loss": 0.8714,
+      "step": 1241
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018159884212692274,
+      "loss": 0.3608,
+      "step": 1242
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018156559191075287,
+      "loss": 0.8663,
+      "step": 1243
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018153231473051933,
+      "loss": 0.8759,
+      "step": 1244
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018149901059722292,
+      "loss": 0.855,
+      "step": 1245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018146567952187333,
+      "loss": 0.9089,
+      "step": 1246
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018143232151548923,
+      "loss": 0.8003,
+      "step": 1247
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018139893658909817,
+      "loss": 0.8914,
+      "step": 1248
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018136552475373657,
+      "loss": 0.8583,
+      "step": 1249
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018133208602044972,
+      "loss": 0.8267,
+      "step": 1250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018129862040029192,
+      "loss": 0.9048,
+      "step": 1251
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001812651279043262,
+      "loss": 0.3333,
+      "step": 1252
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001812316085436246,
+      "loss": 0.8824,
+      "step": 1253
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000181198062329268,
+      "loss": 0.8834,
+      "step": 1254
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018116448927234612,
+      "loss": 0.9013,
+      "step": 1255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018113088938395762,
+      "loss": 0.852,
+      "step": 1256
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018109726267520997,
+      "loss": 0.8373,
+      "step": 1257
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018106360915721956,
+      "loss": 0.8794,
+      "step": 1258
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018102992884111161,
+      "loss": 0.8867,
+      "step": 1259
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018099622173802014,
+      "loss": 0.8668,
+      "step": 1260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018096248785908823,
+      "loss": 0.8489,
+      "step": 1261
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018092872721546754,
+      "loss": 0.9099,
+      "step": 1262
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018089493981831876,
+      "loss": 0.8485,
+      "step": 1263
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018086112567881137,
+      "loss": 0.3121,
+      "step": 1264
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001808272848081237,
+      "loss": 0.3369,
+      "step": 1265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001807934172174429,
+      "loss": 0.3354,
+      "step": 1266
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018075952291796496,
+      "loss": 0.897,
+      "step": 1267
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001807256019208947,
+      "loss": 0.8439,
+      "step": 1268
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018069165423744582,
+      "loss": 0.8845,
+      "step": 1269
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018065767987884073,
+      "loss": 0.819,
+      "step": 1270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018062367885631072,
+      "loss": 0.8909,
+      "step": 1271
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018058965118109593,
+      "loss": 0.8348,
+      "step": 1272
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001805555968644453,
+      "loss": 0.8944,
+      "step": 1273
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018052151591761644,
+      "loss": 0.8636,
+      "step": 1274
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018048740835187598,
+      "loss": 0.9138,
+      "step": 1275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018045327417849923,
+      "loss": 0.8516,
+      "step": 1276
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018041911340877027,
+      "loss": 0.8651,
+      "step": 1277
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018038492605398205,
+      "loss": 0.3255,
+      "step": 1278
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018035071212543625,
+      "loss": 0.8454,
+      "step": 1279
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018031647163444339,
+      "loss": 0.8705,
+      "step": 1280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001802822045923227,
+      "loss": 0.8589,
+      "step": 1281
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001802479110104022,
+      "loss": 0.8711,
+      "step": 1282
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018021359090001878,
+      "loss": 0.8452,
+      "step": 1283
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000180179244272518,
+      "loss": 0.8866,
+      "step": 1284
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001801448711392542,
+      "loss": 0.8338,
+      "step": 1285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018011047151159052,
+      "loss": 0.9103,
+      "step": 1286
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018007604540089882,
+      "loss": 0.886,
+      "step": 1287
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018004159281855974,
+      "loss": 0.8666,
+      "step": 1288
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018000711377596268,
+      "loss": 0.8548,
+      "step": 1289
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001799726082845057,
+      "loss": 0.8266,
+      "step": 1290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017993807635559569,
+      "loss": 0.8788,
+      "step": 1291
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017990351800064834,
+      "loss": 0.8514,
+      "step": 1292
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017986893323108794,
+      "loss": 0.8766,
+      "step": 1293
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017983432205834755,
+      "loss": 0.8105,
+      "step": 1294
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017979968449386902,
+      "loss": 0.8873,
+      "step": 1295
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017976502054910286,
+      "loss": 0.9033,
+      "step": 1296
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001797303302355083,
+      "loss": 0.8857,
+      "step": 1297
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017969561356455336,
+      "loss": 0.8858,
+      "step": 1298
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017966087054771475,
+      "loss": 0.8827,
+      "step": 1299
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017962610119647777,
+      "loss": 0.8878,
+      "step": 1300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017959130552233662,
+      "loss": 0.8066,
+      "step": 1301
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017955648353679398,
+      "loss": 0.8309,
+      "step": 1302
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017952163525136142,
+      "loss": 0.3437,
+      "step": 1303
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017948676067755916,
+      "loss": 0.8909,
+      "step": 1304
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017945185982691606,
+      "loss": 0.8318,
+      "step": 1305
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017941693271096966,
+      "loss": 0.8822,
+      "step": 1306
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017938197934126625,
+      "loss": 0.8448,
+      "step": 1307
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017934699972936075,
+      "loss": 0.8755,
+      "step": 1308
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017931199388681676,
+      "loss": 0.8897,
+      "step": 1309
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017927696182520658,
+      "loss": 0.9133,
+      "step": 1310
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017924190355611116,
+      "loss": 0.9064,
+      "step": 1311
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017920681909112008,
+      "loss": 0.8647,
+      "step": 1312
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001791717084418316,
+      "loss": 0.8988,
+      "step": 1313
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017913657161985268,
+      "loss": 0.808,
+      "step": 1314
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017910140863679886,
+      "loss": 0.8324,
+      "step": 1315
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017906621950429443,
+      "loss": 0.8449,
+      "step": 1316
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001790310042339722,
+      "loss": 0.7559,
+      "step": 1317
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017899576283747373,
+      "loss": 0.878,
+      "step": 1318
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001789604953264491,
+      "loss": 0.816,
+      "step": 1319
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001789252017125572,
+      "loss": 0.8428,
+      "step": 1320
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017888988200746533,
+      "loss": 0.8494,
+      "step": 1321
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001788545362228496,
+      "loss": 0.921,
+      "step": 1322
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001788191643703946,
+      "loss": 0.8453,
+      "step": 1323
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017878376646179368,
+      "loss": 0.8771,
+      "step": 1324
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001787483425087487,
+      "loss": 0.3333,
+      "step": 1325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017871289252297011,
+      "loss": 0.8767,
+      "step": 1326
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001786774165161771,
+      "loss": 0.8288,
+      "step": 1327
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001786419145000973,
+      "loss": 0.9207,
+      "step": 1328
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017860638648646706,
+      "loss": 0.8583,
+      "step": 1329
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017857083248703126,
+      "loss": 0.8699,
+      "step": 1330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001785352525135434,
+      "loss": 0.8818,
+      "step": 1331
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017849964657776552,
+      "loss": 0.9206,
+      "step": 1332
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017846401469146834,
+      "loss": 0.8762,
+      "step": 1333
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017842835686643108,
+      "loss": 0.9065,
+      "step": 1334
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001783926731144415,
+      "loss": 0.9138,
+      "step": 1335
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017835696344729605,
+      "loss": 0.8421,
+      "step": 1336
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001783212278767996,
+      "loss": 0.8761,
+      "step": 1337
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017828546641476578,
+      "loss": 0.8811,
+      "step": 1338
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017824967907301655,
+      "loss": 0.8485,
+      "step": 1339
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001782138658633826,
+      "loss": 0.9329,
+      "step": 1340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017817802679770308,
+      "loss": 0.8659,
+      "step": 1341
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017814216188782577,
+      "loss": 0.7723,
+      "step": 1342
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017810627114560687,
+      "loss": 0.8929,
+      "step": 1343
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017807035458291122,
+      "loss": 0.8124,
+      "step": 1344
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017803441221161216,
+      "loss": 0.8135,
+      "step": 1345
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001779984440435916,
+      "loss": 0.8659,
+      "step": 1346
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001779624500907399,
+      "loss": 0.8625,
+      "step": 1347
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000177926430364956,
+      "loss": 0.7906,
+      "step": 1348
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017789038487814738,
+      "loss": 0.8141,
+      "step": 1349
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017785431364222997,
+      "loss": 0.8376,
+      "step": 1350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017781821666912827,
+      "loss": 0.8217,
+      "step": 1351
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017778209397077528,
+      "loss": 0.8215,
+      "step": 1352
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017774594555911244,
+      "loss": 0.9511,
+      "step": 1353
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017770977144608978,
+      "loss": 0.8253,
+      "step": 1354
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017767357164366577,
+      "loss": 0.8548,
+      "step": 1355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001776373461638074,
+      "loss": 0.8614,
+      "step": 1356
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017760109501849014,
+      "loss": 0.3316,
+      "step": 1357
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017756481821969798,
+      "loss": 0.8439,
+      "step": 1358
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017752851577942325,
+      "loss": 0.8702,
+      "step": 1359
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017749218770966692,
+      "loss": 0.8627,
+      "step": 1360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001774558340224384,
+      "loss": 0.8994,
+      "step": 1361
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001774194547297555,
+      "loss": 0.9099,
+      "step": 1362
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017738304984364457,
+      "loss": 0.8789,
+      "step": 1363
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017734661937614035,
+      "loss": 0.8087,
+      "step": 1364
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001773101633392861,
+      "loss": 0.8936,
+      "step": 1365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017727368174513347,
+      "loss": 0.959,
+      "step": 1366
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017723717460574266,
+      "loss": 0.8716,
+      "step": 1367
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001772006419331822,
+      "loss": 0.8599,
+      "step": 1368
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017716408373952912,
+      "loss": 0.7967,
+      "step": 1369
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017712750003686883,
+      "loss": 0.9021,
+      "step": 1370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017709089083729534,
+      "loss": 0.8223,
+      "step": 1371
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017705425615291084,
+      "loss": 0.8295,
+      "step": 1372
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017701759599582614,
+      "loss": 0.8892,
+      "step": 1373
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017698091037816042,
+      "loss": 0.7881,
+      "step": 1374
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017694419931204121,
+      "loss": 0.9226,
+      "step": 1375
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017690746280960454,
+      "loss": 0.9415,
+      "step": 1376
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001768707008829948,
+      "loss": 0.8693,
+      "step": 1377
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001768339135443648,
+      "loss": 0.328,
+      "step": 1378
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017679710080587574,
+      "loss": 0.8052,
+      "step": 1379
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017676026267969728,
+      "loss": 0.8747,
+      "step": 1380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017672339917800732,
+      "loss": 0.8698,
+      "step": 1381
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001766865103129923,
+      "loss": 0.8665,
+      "step": 1382
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017664959609684698,
+      "loss": 0.8415,
+      "step": 1383
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017661265654177454,
+      "loss": 0.8554,
+      "step": 1384
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017657569165998647,
+      "loss": 0.3087,
+      "step": 1385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017653870146370267,
+      "loss": 0.8844,
+      "step": 1386
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017650168596515146,
+      "loss": 0.8805,
+      "step": 1387
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017646464517656943,
+      "loss": 0.8737,
+      "step": 1388
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001764275791102016,
+      "loss": 0.8438,
+      "step": 1389
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001763904877783013,
+      "loss": 0.3093,
+      "step": 1390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001763533711931302,
+      "loss": 0.9278,
+      "step": 1391
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001763162293669584,
+      "loss": 0.7984,
+      "step": 1392
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001762790623120643,
+      "loss": 0.828,
+      "step": 1393
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017624187004073463,
+      "loss": 0.8227,
+      "step": 1394
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017620465256526445,
+      "loss": 0.8955,
+      "step": 1395
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001761674098979571,
+      "loss": 0.8354,
+      "step": 1396
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017613014205112443,
+      "loss": 0.8013,
+      "step": 1397
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017609284903708644,
+      "loss": 0.8418,
+      "step": 1398
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001760555308681715,
+      "loss": 0.8737,
+      "step": 1399
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001760181875567163,
+      "loss": 0.8602,
+      "step": 1400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017598081911506582,
+      "loss": 0.8218,
+      "step": 1401
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001759434255555734,
+      "loss": 0.8373,
+      "step": 1402
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001759060068906007,
+      "loss": 0.9228,
+      "step": 1403
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017586856313251756,
+      "loss": 0.9106,
+      "step": 1404
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001758310942937022,
+      "loss": 0.812,
+      "step": 1405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017579360038654114,
+      "loss": 0.7977,
+      "step": 1406
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017575608142342915,
+      "loss": 0.8999,
+      "step": 1407
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017571853741676932,
+      "loss": 0.8269,
+      "step": 1408
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017568096837897298,
+      "loss": 0.8658,
+      "step": 1409
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017564337432245976,
+      "loss": 0.7861,
+      "step": 1410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017560575525965753,
+      "loss": 0.8555,
+      "step": 1411
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017556811120300253,
+      "loss": 0.8847,
+      "step": 1412
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017553044216493907,
+      "loss": 0.8699,
+      "step": 1413
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017549274815791994,
+      "loss": 0.8401,
+      "step": 1414
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000175455029194406,
+      "loss": 0.9245,
+      "step": 1415
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017541728528686645,
+      "loss": 0.3101,
+      "step": 1416
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017537951644777872,
+      "loss": 0.8484,
+      "step": 1417
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017534172268962852,
+      "loss": 0.8512,
+      "step": 1418
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017530390402490975,
+      "loss": 0.3219,
+      "step": 1419
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017526606046612452,
+      "loss": 0.8779,
+      "step": 1420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017522819202578325,
+      "loss": 0.8866,
+      "step": 1421
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001751902987164045,
+      "loss": 0.8617,
+      "step": 1422
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001751523805505151,
+      "loss": 0.8804,
+      "step": 1423
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017511443754065012,
+      "loss": 0.8327,
+      "step": 1424
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017507646969935282,
+      "loss": 0.9355,
+      "step": 1425
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017503847703917455,
+      "loss": 0.8694,
+      "step": 1426
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017500045957267513,
+      "loss": 0.8823,
+      "step": 1427
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001749624173124223,
+      "loss": 0.8797,
+      "step": 1428
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001749243502709922,
+      "loss": 0.8382,
+      "step": 1429
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017488625846096904,
+      "loss": 0.3355,
+      "step": 1430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001748481418949453,
+      "loss": 0.8158,
+      "step": 1431
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017481000058552156,
+      "loss": 0.8646,
+      "step": 1432
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017477183454530666,
+      "loss": 0.8141,
+      "step": 1433
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001747336437869176,
+      "loss": 0.9443,
+      "step": 1434
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017469542832297945,
+      "loss": 0.8538,
+      "step": 1435
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017465718816612563,
+      "loss": 0.9118,
+      "step": 1436
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017461892332899754,
+      "loss": 0.8437,
+      "step": 1437
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017458063382424488,
+      "loss": 0.7703,
+      "step": 1438
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017454231966452542,
+      "loss": 0.8111,
+      "step": 1439
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017450398086250513,
+      "loss": 0.8285,
+      "step": 1440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017446561743085805,
+      "loss": 0.8783,
+      "step": 1441
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017442722938226647,
+      "loss": 0.9083,
+      "step": 1442
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017438881672942073,
+      "loss": 0.8979,
+      "step": 1443
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017435037948501935,
+      "loss": 0.8905,
+      "step": 1444
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000174311917661769,
+      "loss": 0.9238,
+      "step": 1445
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017427343127238439,
+      "loss": 0.8332,
+      "step": 1446
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017423492032958836,
+      "loss": 0.341,
+      "step": 1447
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017419638484611206,
+      "loss": 0.8132,
+      "step": 1448
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017415782483469445,
+      "loss": 0.8606,
+      "step": 1449
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017411924030808284,
+      "loss": 0.8734,
+      "step": 1450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001740806312790325,
+      "loss": 0.8637,
+      "step": 1451
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001740419977603069,
+      "loss": 0.8283,
+      "step": 1452
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017400333976467754,
+      "loss": 0.9274,
+      "step": 1453
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017396465730492406,
+      "loss": 0.8791,
+      "step": 1454
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001739259503938341,
+      "loss": 0.9095,
+      "step": 1455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017388721904420352,
+      "loss": 0.8371,
+      "step": 1456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001738484632688361,
+      "loss": 0.8179,
+      "step": 1457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017380968308054385,
+      "loss": 0.8515,
+      "step": 1458
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017377087849214672,
+      "loss": 0.7949,
+      "step": 1459
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001737320495164728,
+      "loss": 0.8518,
+      "step": 1460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017369319616635824,
+      "loss": 0.9256,
+      "step": 1461
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017365431845464723,
+      "loss": 0.8983,
+      "step": 1462
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000173615416394192,
+      "loss": 0.8594,
+      "step": 1463
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001735764899978529,
+      "loss": 0.8712,
+      "step": 1464
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001735375392784982,
+      "loss": 0.8407,
+      "step": 1465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001734985642490043,
+      "loss": 0.8366,
+      "step": 1466
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017345956492225564,
+      "loss": 0.8249,
+      "step": 1467
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017342054131114465,
+      "loss": 0.9082,
+      "step": 1468
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001733814934285718,
+      "loss": 0.8547,
+      "step": 1469
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017334242128744568,
+      "loss": 0.9582,
+      "step": 1470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017330332490068267,
+      "loss": 0.7946,
+      "step": 1471
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001732642042812074,
+      "loss": 0.7926,
+      "step": 1472
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001732250594419524,
+      "loss": 0.832,
+      "step": 1473
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017318589039585816,
+      "loss": 0.8792,
+      "step": 1474
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017314669715587337,
+      "loss": 0.8679,
+      "step": 1475
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017310747973495446,
+      "loss": 0.8225,
+      "step": 1476
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017306823814606606,
+      "loss": 0.8512,
+      "step": 1477
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017302897240218065,
+      "loss": 0.8456,
+      "step": 1478
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017298968251627876,
+      "loss": 0.9132,
+      "step": 1479
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017295036850134893,
+      "loss": 0.3166,
+      "step": 1480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017291103037038764,
+      "loss": 0.8784,
+      "step": 1481
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001728716681363993,
+      "loss": 0.8467,
+      "step": 1482
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017283228181239638,
+      "loss": 0.8628,
+      "step": 1483
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017279287141139918,
+      "loss": 0.8306,
+      "step": 1484
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017275343694643616,
+      "loss": 0.9134,
+      "step": 1485
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017271397843054352,
+      "loss": 0.7657,
+      "step": 1486
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017267449587676556,
+      "loss": 0.8823,
+      "step": 1487
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017263498929815448,
+      "loss": 0.8256,
+      "step": 1488
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017259545870777039,
+      "loss": 0.8562,
+      "step": 1489
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017255590411868136,
+      "loss": 0.8612,
+      "step": 1490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017251632554396348,
+      "loss": 0.8966,
+      "step": 1491
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017247672299670053,
+      "loss": 0.3252,
+      "step": 1492
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017243709648998457,
+      "loss": 0.8806,
+      "step": 1493
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017239744603691524,
+      "loss": 0.3569,
+      "step": 1494
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001723577716506003,
+      "loss": 0.9195,
+      "step": 1495
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017231807334415532,
+      "loss": 0.9217,
+      "step": 1496
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017227835113070388,
+      "loss": 0.92,
+      "step": 1497
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017223860502337733,
+      "loss": 0.8585,
+      "step": 1498
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017219883503531507,
+      "loss": 0.8095,
+      "step": 1499
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017215904117966427,
+      "loss": 0.3411,
+      "step": 1500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017211922346958004,
+      "loss": 0.8462,
+      "step": 1501
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001720793819182254,
+      "loss": 0.3421,
+      "step": 1502
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017203951653877117,
+      "loss": 0.9114,
+      "step": 1503
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017199962734439618,
+      "loss": 0.809,
+      "step": 1504
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017195971434828694,
+      "loss": 0.9317,
+      "step": 1505
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017191977756363808,
+      "loss": 0.8887,
+      "step": 1506
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017187981700365182,
+      "loss": 0.8288,
+      "step": 1507
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017183983268153849,
+      "loss": 0.7604,
+      "step": 1508
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001717998246105161,
+      "loss": 0.8307,
+      "step": 1509
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017175979280381056,
+      "loss": 0.8624,
+      "step": 1510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001717197372746557,
+      "loss": 0.8554,
+      "step": 1511
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017167965803629307,
+      "loss": 0.8779,
+      "step": 1512
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001716395551019722,
+      "loss": 0.3639,
+      "step": 1513
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017159942848495025,
+      "loss": 0.8635,
+      "step": 1514
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001715592781984924,
+      "loss": 0.9198,
+      "step": 1515
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017151910425587162,
+      "loss": 0.8069,
+      "step": 1516
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017147890667036859,
+      "loss": 0.8934,
+      "step": 1517
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017143868545527196,
+      "loss": 0.8983,
+      "step": 1518
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017139844062387803,
+      "loss": 0.9061,
+      "step": 1519
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017135817218949108,
+      "loss": 0.8582,
+      "step": 1520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017131788016542306,
+      "loss": 0.8946,
+      "step": 1521
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017127756456499372,
+      "loss": 0.8512,
+      "step": 1522
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017123722540153072,
+      "loss": 0.8492,
+      "step": 1523
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001711968626883694,
+      "loss": 0.7956,
+      "step": 1524
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017115647643885294,
+      "loss": 0.8696,
+      "step": 1525
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017111606666633225,
+      "loss": 0.9015,
+      "step": 1526
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017107563338416605,
+      "loss": 0.8125,
+      "step": 1527
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017103517660572087,
+      "loss": 0.9196,
+      "step": 1528
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001709946963443709,
+      "loss": 0.8428,
+      "step": 1529
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001709541926134982,
+      "loss": 0.8677,
+      "step": 1530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017091366542649255,
+      "loss": 0.883,
+      "step": 1531
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017087311479675147,
+      "loss": 0.8533,
+      "step": 1532
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017083254073768027,
+      "loss": 0.8496,
+      "step": 1533
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017079194326269194,
+      "loss": 0.8508,
+      "step": 1534
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017075132238520725,
+      "loss": 0.3429,
+      "step": 1535
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017071067811865476,
+      "loss": 0.9232,
+      "step": 1536
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017067001047647065,
+      "loss": 0.795,
+      "step": 1537
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001706293194720989,
+      "loss": 0.9046,
+      "step": 1538
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017058860511899118,
+      "loss": 0.9118,
+      "step": 1539
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017054786743060692,
+      "loss": 0.9113,
+      "step": 1540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017050710642041325,
+      "loss": 0.7613,
+      "step": 1541
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017046632210188496,
+      "loss": 0.8362,
+      "step": 1542
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017042551448850462,
+      "loss": 0.308,
+      "step": 1543
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017038468359376245,
+      "loss": 0.9056,
+      "step": 1544
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001703438294311564,
+      "loss": 0.9124,
+      "step": 1545
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017030295201419206,
+      "loss": 0.8666,
+      "step": 1546
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017026205135638276,
+      "loss": 0.3867,
+      "step": 1547
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001702211274712495,
+      "loss": 0.8891,
+      "step": 1548
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000170180180372321,
+      "loss": 0.9253,
+      "step": 1549
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017013921007313348,
+      "loss": 0.8347,
+      "step": 1550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001700982165872311,
+      "loss": 0.8458,
+      "step": 1551
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017005719992816546,
+      "loss": 0.8496,
+      "step": 1552
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017001616010949589,
+      "loss": 0.8329,
+      "step": 1553
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016997509714478944,
+      "loss": 0.8412,
+      "step": 1554
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016993401104762072,
+      "loss": 0.3275,
+      "step": 1555
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016989290183157206,
+      "loss": 0.9324,
+      "step": 1556
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001698517695102334,
+      "loss": 0.9236,
+      "step": 1557
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001698106140972023,
+      "loss": 0.8794,
+      "step": 1558
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016976943560608395,
+      "loss": 0.858,
+      "step": 1559
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016972823405049124,
+      "loss": 0.8128,
+      "step": 1560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016968700944404463,
+      "loss": 0.8418,
+      "step": 1561
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016964576180037217,
+      "loss": 0.8062,
+      "step": 1562
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001696044911331096,
+      "loss": 0.8548,
+      "step": 1563
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016956319745590017,
+      "loss": 0.8132,
+      "step": 1564
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001695218807823949,
+      "loss": 0.852,
+      "step": 1565
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016948054112625222,
+      "loss": 0.8587,
+      "step": 1566
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001694391785011383,
+      "loss": 0.8671,
+      "step": 1567
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016939779292072683,
+      "loss": 0.9015,
+      "step": 1568
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016935638439869913,
+      "loss": 0.8868,
+      "step": 1569
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016931495294874408,
+      "loss": 0.8575,
+      "step": 1570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001692734985845581,
+      "loss": 0.8777,
+      "step": 1571
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001692320213198453,
+      "loss": 0.9439,
+      "step": 1572
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001691905211683173,
+      "loss": 0.8628,
+      "step": 1573
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001691489981436932,
+      "loss": 0.7991,
+      "step": 1574
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016910745225969976,
+      "loss": 0.8732,
+      "step": 1575
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016906588353007132,
+      "loss": 0.9143,
+      "step": 1576
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016902429196854967,
+      "loss": 0.8771,
+      "step": 1577
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016898267758888423,
+      "loss": 0.8512,
+      "step": 1578
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016894104040483194,
+      "loss": 0.8493,
+      "step": 1579
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016889938043015726,
+      "loss": 0.8749,
+      "step": 1580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016885769767863225,
+      "loss": 0.8612,
+      "step": 1581
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001688159921640364,
+      "loss": 0.8884,
+      "step": 1582
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016877426390015678,
+      "loss": 0.9379,
+      "step": 1583
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000168732512900788,
+      "loss": 0.9136,
+      "step": 1584
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016869073917973214,
+      "loss": 0.7785,
+      "step": 1585
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016864894275079882,
+      "loss": 0.8885,
+      "step": 1586
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016860712362780516,
+      "loss": 0.8663,
+      "step": 1587
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001685652818245758,
+      "loss": 0.8718,
+      "step": 1588
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016852341735494283,
+      "loss": 0.8013,
+      "step": 1589
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001684815302327459,
+      "loss": 0.8517,
+      "step": 1590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016843962047183205,
+      "loss": 0.8222,
+      "step": 1591
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016839768808605594,
+      "loss": 0.3632,
+      "step": 1592
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001683557330892796,
+      "loss": 0.8376,
+      "step": 1593
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016831375549537252,
+      "loss": 0.8327,
+      "step": 1594
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001682717553182118,
+      "loss": 0.8552,
+      "step": 1595
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016822973257168186,
+      "loss": 0.779,
+      "step": 1596
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016818768726967464,
+      "loss": 0.3523,
+      "step": 1597
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016814561942608957,
+      "loss": 0.8735,
+      "step": 1598
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016810352905483343,
+      "loss": 0.84,
+      "step": 1599
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016806141616982059,
+      "loss": 0.8558,
+      "step": 1600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016801928078497274,
+      "loss": 0.8682,
+      "step": 1601
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016797712291421904,
+      "loss": 0.8321,
+      "step": 1602
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001679349425714961,
+      "loss": 0.8251,
+      "step": 1603
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016789273977074797,
+      "loss": 0.9084,
+      "step": 1604
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016785051452592614,
+      "loss": 0.3262,
+      "step": 1605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016780826685098942,
+      "loss": 0.8576,
+      "step": 1606
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016776599675990417,
+      "loss": 0.8372,
+      "step": 1607
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016772370426664402,
+      "loss": 0.8872,
+      "step": 1608
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016768138938519014,
+      "loss": 0.8168,
+      "step": 1609
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016763905212953102,
+      "loss": 0.8866,
+      "step": 1610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016759669251366257,
+      "loss": 0.8534,
+      "step": 1611
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016755431055158807,
+      "loss": 0.8607,
+      "step": 1612
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016751190625731828,
+      "loss": 0.8999,
+      "step": 1613
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016746947964487116,
+      "loss": 0.849,
+      "step": 1614
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016742703072827224,
+      "loss": 0.8171,
+      "step": 1615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001673845595215543,
+      "loss": 0.853,
+      "step": 1616
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016734206603875756,
+      "loss": 0.8499,
+      "step": 1617
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001672995502939295,
+      "loss": 0.8324,
+      "step": 1618
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016725701230112513,
+      "loss": 0.7571,
+      "step": 1619
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016721445207440664,
+      "loss": 0.8852,
+      "step": 1620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001671718696278437,
+      "loss": 0.8819,
+      "step": 1621
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016712926497551326,
+      "loss": 0.8739,
+      "step": 1622
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016708663813149962,
+      "loss": 0.8721,
+      "step": 1623
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001670439891098944,
+      "loss": 0.802,
+      "step": 1624
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016700131792479658,
+      "loss": 0.8309,
+      "step": 1625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016695862459031248,
+      "loss": 0.869,
+      "step": 1626
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001669159091205557,
+      "loss": 0.8518,
+      "step": 1627
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016687317152964718,
+      "loss": 0.8559,
+      "step": 1628
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001668304118317152,
+      "loss": 0.9053,
+      "step": 1629
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016678763004089527,
+      "loss": 0.8409,
+      "step": 1630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016674482617133024,
+      "loss": 0.3681,
+      "step": 1631
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016670200023717038,
+      "loss": 0.8886,
+      "step": 1632
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016665915225257303,
+      "loss": 0.9331,
+      "step": 1633
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016661628223170295,
+      "loss": 0.8149,
+      "step": 1634
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001665733901887322,
+      "loss": 0.7995,
+      "step": 1635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001665304761378401,
+      "loss": 0.8655,
+      "step": 1636
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016648754009321328,
+      "loss": 0.3445,
+      "step": 1637
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016644458206904546,
+      "loss": 0.7889,
+      "step": 1638
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016640160207953789,
+      "loss": 0.8403,
+      "step": 1639
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016635860013889886,
+      "loss": 0.7947,
+      "step": 1640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016631557626134406,
+      "loss": 0.8751,
+      "step": 1641
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016627253046109638,
+      "loss": 0.8454,
+      "step": 1642
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016622946275238594,
+      "loss": 0.8501,
+      "step": 1643
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016618637314945014,
+      "loss": 0.8563,
+      "step": 1644
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016614326166653357,
+      "loss": 0.7941,
+      "step": 1645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016610012831788813,
+      "loss": 0.8437,
+      "step": 1646
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016605697311777284,
+      "loss": 0.3162,
+      "step": 1647
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016601379608045406,
+      "loss": 0.8591,
+      "step": 1648
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016597059722020532,
+      "loss": 0.8852,
+      "step": 1649
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001659273765513073,
+      "loss": 0.8749,
+      "step": 1650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016588413408804797,
+      "loss": 0.8604,
+      "step": 1651
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001658408698447225,
+      "loss": 0.8097,
+      "step": 1652
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016579758383563324,
+      "loss": 0.8255,
+      "step": 1653
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016575427607508974,
+      "loss": 0.8647,
+      "step": 1654
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016571094657740867,
+      "loss": 0.872,
+      "step": 1655
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016566759535691406,
+      "loss": 0.8765,
+      "step": 1656
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016562422242793695,
+      "loss": 0.9012,
+      "step": 1657
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016558082780481563,
+      "loss": 0.8721,
+      "step": 1658
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016553741150189557,
+      "loss": 0.8823,
+      "step": 1659
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016549397353352938,
+      "loss": 0.8628,
+      "step": 1660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016545051391407683,
+      "loss": 0.8281,
+      "step": 1661
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001654070326579049,
+      "loss": 0.3551,
+      "step": 1662
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016536352977938762,
+      "loss": 0.9085,
+      "step": 1663
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001653200052929063,
+      "loss": 0.8613,
+      "step": 1664
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001652764592128493,
+      "loss": 0.7852,
+      "step": 1665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016523289155361204,
+      "loss": 0.9184,
+      "step": 1666
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016518930232959736,
+      "loss": 0.879,
+      "step": 1667
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016514569155521493,
+      "loss": 0.826,
+      "step": 1668
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016510205924488166,
+      "loss": 0.8421,
+      "step": 1669
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001650584054130216,
+      "loss": 0.8269,
+      "step": 1670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001650147300740659,
+      "loss": 0.8564,
+      "step": 1671
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016497103324245282,
+      "loss": 0.8303,
+      "step": 1672
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016492731493262768,
+      "loss": 0.8779,
+      "step": 1673
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016488357515904295,
+      "loss": 0.808,
+      "step": 1674
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001648398139361582,
+      "loss": 0.8264,
+      "step": 1675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001647960312784401,
+      "loss": 0.8036,
+      "step": 1676
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001647522272003623,
+      "loss": 0.8345,
+      "step": 1677
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001647084017164057,
+      "loss": 0.8525,
+      "step": 1678
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016466455484105813,
+      "loss": 0.8329,
+      "step": 1679
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016462068658881456,
+      "loss": 0.3698,
+      "step": 1680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000164576796974177,
+      "loss": 0.8518,
+      "step": 1681
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001645328860116546,
+      "loss": 0.7851,
+      "step": 1682
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016448895371576341,
+      "loss": 0.8912,
+      "step": 1683
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016444500010102676,
+      "loss": 0.8724,
+      "step": 1684
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016440102518197475,
+      "loss": 0.8544,
+      "step": 1685
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016435702897314478,
+      "loss": 0.8649,
+      "step": 1686
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016431301148908114,
+      "loss": 0.8059,
+      "step": 1687
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016426897274433513,
+      "loss": 0.8565,
+      "step": 1688
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016422491275346524,
+      "loss": 0.8654,
+      "step": 1689
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016418083153103683,
+      "loss": 0.3347,
+      "step": 1690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001641367290916223,
+      "loss": 0.8785,
+      "step": 1691
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016409260544980115,
+      "loss": 0.8737,
+      "step": 1692
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001640484606201598,
+      "loss": 0.829,
+      "step": 1693
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001640042946172917,
+      "loss": 0.8804,
+      "step": 1694
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016396010745579735,
+      "loss": 0.8826,
+      "step": 1695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016391589915028417,
+      "loss": 0.8676,
+      "step": 1696
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001638716697153666,
+      "loss": 0.8438,
+      "step": 1697
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001638274191656661,
+      "loss": 0.8831,
+      "step": 1698
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016378314751581098,
+      "loss": 0.8169,
+      "step": 1699
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016373885478043672,
+      "loss": 0.8915,
+      "step": 1700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016369454097418568,
+      "loss": 0.8678,
+      "step": 1701
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016365020611170712,
+      "loss": 0.9217,
+      "step": 1702
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016360585020765736,
+      "loss": 0.8436,
+      "step": 1703
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001635614732766996,
+      "loss": 0.8864,
+      "step": 1704
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016351707533350402,
+      "loss": 0.8596,
+      "step": 1705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016347265639274778,
+      "loss": 0.7585,
+      "step": 1706
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016342821646911498,
+      "loss": 0.3546,
+      "step": 1707
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016338375557729658,
+      "loss": 0.8047,
+      "step": 1708
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016333927373199053,
+      "loss": 0.8566,
+      "step": 1709
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016329477094790168,
+      "loss": 0.8447,
+      "step": 1710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001632502472397419,
+      "loss": 0.8254,
+      "step": 1711
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016320570262222983,
+      "loss": 0.8658,
+      "step": 1712
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001631611371100911,
+      "loss": 0.8442,
+      "step": 1713
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016311655071805822,
+      "loss": 0.8044,
+      "step": 1714
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016307194346087064,
+      "loss": 0.3404,
+      "step": 1715
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016302731535327474,
+      "loss": 0.844,
+      "step": 1716
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016298266641002367,
+      "loss": 0.7918,
+      "step": 1717
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016293799664587755,
+      "loss": 0.3283,
+      "step": 1718
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016289330607560337,
+      "loss": 0.9274,
+      "step": 1719
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016284859471397503,
+      "loss": 0.8548,
+      "step": 1720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001628038625757732,
+      "loss": 0.8418,
+      "step": 1721
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016275910967578558,
+      "loss": 0.8615,
+      "step": 1722
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016271433602880664,
+      "loss": 0.8253,
+      "step": 1723
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016266954164963763,
+      "loss": 0.775,
+      "step": 1724
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016262472655308678,
+      "loss": 0.8134,
+      "step": 1725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016257989075396916,
+      "loss": 0.8302,
+      "step": 1726
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001625350342671066,
+      "loss": 0.9283,
+      "step": 1727
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016249015710732785,
+      "loss": 0.879,
+      "step": 1728
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016244525928946844,
+      "loss": 0.8546,
+      "step": 1729
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016240034082837078,
+      "loss": 0.7971,
+      "step": 1730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000162355401738884,
+      "loss": 0.8346,
+      "step": 1731
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016231044203586422,
+      "loss": 0.8378,
+      "step": 1732
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016226546173417426,
+      "loss": 0.3438,
+      "step": 1733
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016222046084868373,
+      "loss": 0.8229,
+      "step": 1734
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001621754393942691,
+      "loss": 0.8584,
+      "step": 1735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016213039738581362,
+      "loss": 0.8439,
+      "step": 1736
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016208533483820734,
+      "loss": 0.9178,
+      "step": 1737
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016204025176634712,
+      "loss": 0.7663,
+      "step": 1738
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016199514818513656,
+      "loss": 0.8636,
+      "step": 1739
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001619500241094861,
+      "loss": 0.8817,
+      "step": 1740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016190487955431286,
+      "loss": 0.8909,
+      "step": 1741
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016185971453454078,
+      "loss": 0.7909,
+      "step": 1742
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016181452906510066,
+      "loss": 0.8467,
+      "step": 1743
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001617693231609299,
+      "loss": 0.8633,
+      "step": 1744
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016172409683697274,
+      "loss": 0.8685,
+      "step": 1745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016167885010818017,
+      "loss": 0.8471,
+      "step": 1746
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016163358298950991,
+      "loss": 0.8724,
+      "step": 1747
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016158829549592647,
+      "loss": 0.8186,
+      "step": 1748
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016154298764240097,
+      "loss": 0.8521,
+      "step": 1749
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001614976594439114,
+      "loss": 0.8215,
+      "step": 1750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001614523109154424,
+      "loss": 0.84,
+      "step": 1751
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016140694207198534,
+      "loss": 0.8944,
+      "step": 1752
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001613615529285383,
+      "loss": 0.7886,
+      "step": 1753
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016131614350010614,
+      "loss": 0.8901,
+      "step": 1754
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016127071380170032,
+      "loss": 0.8413,
+      "step": 1755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016122526384833907,
+      "loss": 0.3765,
+      "step": 1756
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001611797936550473,
+      "loss": 0.7562,
+      "step": 1757
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016113430323685658,
+      "loss": 0.8212,
+      "step": 1758
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001610887926088052,
+      "loss": 0.8775,
+      "step": 1759
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016104326178593818,
+      "loss": 0.8379,
+      "step": 1760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016099771078330712,
+      "loss": 0.8478,
+      "step": 1761
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016095213961597033,
+      "loss": 0.8963,
+      "step": 1762
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001609065482989928,
+      "loss": 0.8469,
+      "step": 1763
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001608609368474461,
+      "loss": 0.8562,
+      "step": 1764
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016081530527640867,
+      "loss": 0.8728,
+      "step": 1765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016076965360096535,
+      "loss": 0.8833,
+      "step": 1766
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016072398183620773,
+      "loss": 0.881,
+      "step": 1767
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016067828999723405,
+      "loss": 0.8407,
+      "step": 1768
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016063257809914924,
+      "loss": 0.87,
+      "step": 1769
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016058684615706477,
+      "loss": 0.8606,
+      "step": 1770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016054109418609872,
+      "loss": 0.8417,
+      "step": 1771
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001604953222013759,
+      "loss": 0.8779,
+      "step": 1772
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016044953021802764,
+      "loss": 0.3479,
+      "step": 1773
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001604037182511919,
+      "loss": 0.8758,
+      "step": 1774
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016035788631601334,
+      "loss": 0.862,
+      "step": 1775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016031203442764307,
+      "loss": 0.8058,
+      "step": 1776
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001602661626012389,
+      "loss": 0.9015,
+      "step": 1777
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016022027085196516,
+      "loss": 0.9034,
+      "step": 1778
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016017435919499286,
+      "loss": 0.8422,
+      "step": 1779
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016012842764549952,
+      "loss": 0.83,
+      "step": 1780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016008247621866924,
+      "loss": 0.8697,
+      "step": 1781
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001600365049296927,
+      "loss": 0.8221,
+      "step": 1782
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015999051379376718,
+      "loss": 0.8565,
+      "step": 1783
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001599445028260965,
+      "loss": 0.8277,
+      "step": 1784
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015989847204189093,
+      "loss": 0.8318,
+      "step": 1785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001598524214563675,
+      "loss": 0.8967,
+      "step": 1786
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015980635108474967,
+      "loss": 0.8563,
+      "step": 1787
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001597602609422674,
+      "loss": 0.8629,
+      "step": 1788
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015971415104415723,
+      "loss": 0.9295,
+      "step": 1789
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015966802140566225,
+      "loss": 0.811,
+      "step": 1790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015962187204203207,
+      "loss": 0.8869,
+      "step": 1791
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001595757029685228,
+      "loss": 0.8288,
+      "step": 1792
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015952951420039705,
+      "loss": 0.8864,
+      "step": 1793
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015948330575292401,
+      "loss": 0.8494,
+      "step": 1794
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015943707764137929,
+      "loss": 0.8204,
+      "step": 1795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015939082988104505,
+      "loss": 0.7675,
+      "step": 1796
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015934456248720995,
+      "loss": 0.84,
+      "step": 1797
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015929827547516914,
+      "loss": 0.8693,
+      "step": 1798
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001592519688602242,
+      "loss": 0.8586,
+      "step": 1799
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001592056426576833,
+      "loss": 0.8346,
+      "step": 1800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015915929688286093,
+      "loss": 0.8065,
+      "step": 1801
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001591129315510782,
+      "loss": 0.8687,
+      "step": 1802
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015906654667766264,
+      "loss": 0.7658,
+      "step": 1803
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015902014227794816,
+      "loss": 0.8292,
+      "step": 1804
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015897371836727525,
+      "loss": 0.7925,
+      "step": 1805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015892727496099075,
+      "loss": 0.8556,
+      "step": 1806
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000158880812074448,
+      "loss": 0.8186,
+      "step": 1807
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015883432972300674,
+      "loss": 0.8503,
+      "step": 1808
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015878782792203323,
+      "loss": 0.8977,
+      "step": 1809
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015874130668690003,
+      "loss": 0.8706,
+      "step": 1810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015869476603298623,
+      "loss": 0.8748,
+      "step": 1811
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001586482059756773,
+      "loss": 0.8182,
+      "step": 1812
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015860162653036513,
+      "loss": 0.8206,
+      "step": 1813
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015855502771244798,
+      "loss": 0.8906,
+      "step": 1814
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015850840953733058,
+      "loss": 0.8463,
+      "step": 1815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015846177202042406,
+      "loss": 0.8277,
+      "step": 1816
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015841511517714584,
+      "loss": 0.8656,
+      "step": 1817
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015836843902291984,
+      "loss": 0.8057,
+      "step": 1818
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015832174357317634,
+      "loss": 0.9477,
+      "step": 1819
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000158275028843352,
+      "loss": 0.83,
+      "step": 1820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001582282948488898,
+      "loss": 0.8823,
+      "step": 1821
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015818154160523911,
+      "loss": 0.8921,
+      "step": 1822
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001581347691278557,
+      "loss": 0.3367,
+      "step": 1823
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015808797743220175,
+      "loss": 0.8597,
+      "step": 1824
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001580411665337456,
+      "loss": 0.8925,
+      "step": 1825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015799433644796216,
+      "loss": 0.7198,
+      "step": 1826
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015794748719033253,
+      "loss": 0.8946,
+      "step": 1827
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001579006187763442,
+      "loss": 0.8701,
+      "step": 1828
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015785373122149103,
+      "loss": 0.8572,
+      "step": 1829
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015780682454127312,
+      "loss": 0.9018,
+      "step": 1830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015775989875119702,
+      "loss": 0.8582,
+      "step": 1831
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015771295386677543,
+      "loss": 0.834,
+      "step": 1832
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001576659899035275,
+      "loss": 0.8447,
+      "step": 1833
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015761900687697865,
+      "loss": 0.9196,
+      "step": 1834
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015757200480266058,
+      "loss": 0.8164,
+      "step": 1835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015752498369611133,
+      "loss": 0.78,
+      "step": 1836
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015747794357287513,
+      "loss": 0.8604,
+      "step": 1837
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001574308844485026,
+      "loss": 0.8178,
+      "step": 1838
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015738380633855062,
+      "loss": 0.8638,
+      "step": 1839
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015733670925858237,
+      "loss": 0.805,
+      "step": 1840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001572895932241672,
+      "loss": 0.8954,
+      "step": 1841
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015724245825088086,
+      "loss": 0.3795,
+      "step": 1842
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001571953043543052,
+      "loss": 0.8983,
+      "step": 1843
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001571481315500285,
+      "loss": 0.8456,
+      "step": 1844
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015710093985364516,
+      "loss": 0.8131,
+      "step": 1845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015705372928075594,
+      "loss": 0.8413,
+      "step": 1846
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015700649984696772,
+      "loss": 0.8383,
+      "step": 1847
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015695925156789366,
+      "loss": 0.8664,
+      "step": 1848
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015691198445915317,
+      "loss": 0.8664,
+      "step": 1849
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015686469853637192,
+      "loss": 0.3381,
+      "step": 1850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001568173938151817,
+      "loss": 0.8847,
+      "step": 1851
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001567700703112206,
+      "loss": 0.8527,
+      "step": 1852
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001567227280401329,
+      "loss": 0.8469,
+      "step": 1853
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015667536701756903,
+      "loss": 0.8566,
+      "step": 1854
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015662798725918567,
+      "loss": 0.8393,
+      "step": 1855
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015658058878064573,
+      "loss": 0.9351,
+      "step": 1856
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001565331715976182,
+      "loss": 0.8645,
+      "step": 1857
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015648573572577839,
+      "loss": 0.8871,
+      "step": 1858
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015643828118080762,
+      "loss": 0.8787,
+      "step": 1859
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001563908079783935,
+      "loss": 0.8258,
+      "step": 1860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015634331613422985,
+      "loss": 0.8753,
+      "step": 1861
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015629580566401657,
+      "loss": 0.9092,
+      "step": 1862
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015624827658345972,
+      "loss": 0.8187,
+      "step": 1863
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001562007289082715,
+      "loss": 0.8514,
+      "step": 1864
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001561531626541703,
+      "loss": 0.8195,
+      "step": 1865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001561055778368807,
+      "loss": 0.8144,
+      "step": 1866
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015605797447213325,
+      "loss": 0.8317,
+      "step": 1867
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015601035257566478,
+      "loss": 0.8436,
+      "step": 1868
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015596271216321826,
+      "loss": 0.8582,
+      "step": 1869
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015591505325054258,
+      "loss": 0.8983,
+      "step": 1870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015586737585339305,
+      "loss": 0.8299,
+      "step": 1871
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015581967998753082,
+      "loss": 0.8069,
+      "step": 1872
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015577196566872329,
+      "loss": 0.8917,
+      "step": 1873
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015572423291274393,
+      "loss": 0.882,
+      "step": 1874
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001556764817353723,
+      "loss": 0.8806,
+      "step": 1875
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015562871215239402,
+      "loss": 0.8783,
+      "step": 1876
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015558092417960082,
+      "loss": 0.8753,
+      "step": 1877
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015553311783279055,
+      "loss": 0.8679,
+      "step": 1878
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015548529312776711,
+      "loss": 0.896,
+      "step": 1879
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015543745008034042,
+      "loss": 0.8267,
+      "step": 1880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001553895887063265,
+      "loss": 0.8033,
+      "step": 1881
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015534170902154742,
+      "loss": 0.8846,
+      "step": 1882
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015529381104183133,
+      "loss": 0.8581,
+      "step": 1883
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001552458947830124,
+      "loss": 0.8282,
+      "step": 1884
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015519796026093087,
+      "loss": 0.8068,
+      "step": 1885
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000155150007491433,
+      "loss": 0.9108,
+      "step": 1886
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015510203649037103,
+      "loss": 0.9098,
+      "step": 1887
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015505404727360334,
+      "loss": 0.8763,
+      "step": 1888
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001550060398569942,
+      "loss": 0.8513,
+      "step": 1889
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015495801425641407,
+      "loss": 0.8163,
+      "step": 1890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015490997048773922,
+      "loss": 0.9151,
+      "step": 1891
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015486190856685208,
+      "loss": 0.8457,
+      "step": 1892
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015481382850964103,
+      "loss": 0.8214,
+      "step": 1893
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001547657303320004,
+      "loss": 0.8067,
+      "step": 1894
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015471761404983056,
+      "loss": 0.298,
+      "step": 1895
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015466947967903786,
+      "loss": 0.8927,
+      "step": 1896
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015462132723553464,
+      "loss": 0.8502,
+      "step": 1897
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001545731567352392,
+      "loss": 0.8626,
+      "step": 1898
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001545249681940758,
+      "loss": 0.9039,
+      "step": 1899
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015447676162797465,
+      "loss": 0.8853,
+      "step": 1900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000154428537052872,
+      "loss": 0.8608,
+      "step": 1901
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015438029448470991,
+      "loss": 0.8587,
+      "step": 1902
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015433203393943655,
+      "loss": 0.872,
+      "step": 1903
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015428375543300599,
+      "loss": 0.3561,
+      "step": 1904
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015423545898137807,
+      "loss": 0.8735,
+      "step": 1905
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015418714460051875,
+      "loss": 0.8182,
+      "step": 1906
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015413881230639994,
+      "loss": 0.8829,
+      "step": 1907
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001540904621149993,
+      "loss": 0.899,
+      "step": 1908
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015404209404230057,
+      "loss": 0.8633,
+      "step": 1909
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001539937081042933,
+      "loss": 0.8718,
+      "step": 1910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015394530431697297,
+      "loss": 0.8481,
+      "step": 1911
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015389688269634098,
+      "loss": 0.8131,
+      "step": 1912
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015384844325840459,
+      "loss": 0.7812,
+      "step": 1913
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015379998601917704,
+      "loss": 0.9047,
+      "step": 1914
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015375151099467736,
+      "loss": 0.8427,
+      "step": 1915
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015370301820093042,
+      "loss": 0.8268,
+      "step": 1916
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015365450765396715,
+      "loss": 0.8717,
+      "step": 1917
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015360597936982416,
+      "loss": 0.8362,
+      "step": 1918
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000153557433364544,
+      "loss": 0.7668,
+      "step": 1919
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001535088696541751,
+      "loss": 0.8656,
+      "step": 1920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001534602882547717,
+      "loss": 0.8439,
+      "step": 1921
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001534116891823939,
+      "loss": 0.8491,
+      "step": 1922
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015336307245310767,
+      "loss": 0.8624,
+      "step": 1923
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015331443808298473,
+      "loss": 0.848,
+      "step": 1924
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001532657860881028,
+      "loss": 0.8234,
+      "step": 1925
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015321711648454524,
+      "loss": 0.8405,
+      "step": 1926
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015316842928840134,
+      "loss": 0.8187,
+      "step": 1927
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015311972451576618,
+      "loss": 0.8432,
+      "step": 1928
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015307100218274065,
+      "loss": 0.3748,
+      "step": 1929
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015302226230543147,
+      "loss": 0.9019,
+      "step": 1930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015297350489995107,
+      "loss": 0.7744,
+      "step": 1931
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015292472998241778,
+      "loss": 0.8492,
+      "step": 1932
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015287593756895568,
+      "loss": 0.8862,
+      "step": 1933
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015282712767569463,
+      "loss": 0.8172,
+      "step": 1934
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015277830031877027,
+      "loss": 0.7905,
+      "step": 1935
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015272945551432398,
+      "loss": 0.8204,
+      "step": 1936
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015268059327850302,
+      "loss": 0.9209,
+      "step": 1937
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015263171362746026,
+      "loss": 0.8601,
+      "step": 1938
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015258281657735444,
+      "loss": 0.8383,
+      "step": 1939
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015253390214435,
+      "loss": 0.8079,
+      "step": 1940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015248497034461712,
+      "loss": 0.8302,
+      "step": 1941
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001524360211943318,
+      "loss": 0.8603,
+      "step": 1942
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015238705470967565,
+      "loss": 0.7944,
+      "step": 1943
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001523380709068361,
+      "loss": 0.8573,
+      "step": 1944
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001522890698020063,
+      "loss": 0.7871,
+      "step": 1945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001522400514113851,
+      "loss": 0.8539,
+      "step": 1946
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000152191015751177,
+      "loss": 0.7447,
+      "step": 1947
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015214196283759238,
+      "loss": 0.8132,
+      "step": 1948
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015209289268684715,
+      "loss": 0.8314,
+      "step": 1949
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015204380531516298,
+      "loss": 0.8274,
+      "step": 1950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001519947007387673,
+      "loss": 0.8198,
+      "step": 1951
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001519455789738931,
+      "loss": 0.8453,
+      "step": 1952
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015189644003677917,
+      "loss": 0.81,
+      "step": 1953
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015184728394366988,
+      "loss": 0.8035,
+      "step": 1954
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015179811071081538,
+      "loss": 0.8285,
+      "step": 1955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015174892035447134,
+      "loss": 0.8633,
+      "step": 1956
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001516997128908993,
+      "loss": 0.8528,
+      "step": 1957
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015165048833636616,
+      "loss": 0.8995,
+      "step": 1958
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015160124670714477,
+      "loss": 0.8082,
+      "step": 1959
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001515519880195135,
+      "loss": 0.8283,
+      "step": 1960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015150271228975623,
+      "loss": 0.8119,
+      "step": 1961
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015145341953416271,
+      "loss": 0.8616,
+      "step": 1962
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015140410976902818,
+      "loss": 0.8734,
+      "step": 1963
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015135478301065352,
+      "loss": 0.8706,
+      "step": 1964
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015130543927534525,
+      "loss": 0.7916,
+      "step": 1965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015125607857941547,
+      "loss": 0.9337,
+      "step": 1966
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001512067009391819,
+      "loss": 0.8708,
+      "step": 1967
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001511573063709679,
+      "loss": 0.8837,
+      "step": 1968
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015110789489110234,
+      "loss": 0.8268,
+      "step": 1969
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001510584665159198,
+      "loss": 0.8306,
+      "step": 1970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015100902126176033,
+      "loss": 0.8799,
+      "step": 1971
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015095955914496965,
+      "loss": 0.8632,
+      "step": 1972
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015091008018189897,
+      "loss": 0.9168,
+      "step": 1973
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015086058438890508,
+      "loss": 0.3643,
+      "step": 1974
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001508110717823505,
+      "loss": 0.8454,
+      "step": 1975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015076154237860304,
+      "loss": 0.8227,
+      "step": 1976
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015071199619403624,
+      "loss": 0.8737,
+      "step": 1977
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015066243324502918,
+      "loss": 0.8696,
+      "step": 1978
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015061285354796644,
+      "loss": 0.8908,
+      "step": 1979
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015056325711923808,
+      "loss": 0.8522,
+      "step": 1980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015051364397523982,
+      "loss": 0.902,
+      "step": 1981
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015046401413237282,
+      "loss": 0.8446,
+      "step": 1982
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015041436760704374,
+      "loss": 0.8825,
+      "step": 1983
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015036470441566488,
+      "loss": 0.8332,
+      "step": 1984
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001503150245746539,
+      "loss": 0.8794,
+      "step": 1985
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015026532810043407,
+      "loss": 0.8295,
+      "step": 1986
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015021561500943407,
+      "loss": 0.7997,
+      "step": 1987
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015016588531808816,
+      "loss": 0.8683,
+      "step": 1988
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015011613904283603,
+      "loss": 0.8476,
+      "step": 1989
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015006637620012286,
+      "loss": 0.8327,
+      "step": 1990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015001659680639938,
+      "loss": 0.9288,
+      "step": 1991
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014996680087812165,
+      "loss": 0.8519,
+      "step": 1992
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014991698843175132,
+      "loss": 0.8314,
+      "step": 1993
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014986715948375542,
+      "loss": 0.8996,
+      "step": 1994
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014981731405060652,
+      "loss": 0.7966,
+      "step": 1995
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014976745214878256,
+      "loss": 0.8242,
+      "step": 1996
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014971757379476697,
+      "loss": 0.9219,
+      "step": 1997
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014966767900504856,
+      "loss": 0.8372,
+      "step": 1998
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014961776779612169,
+      "loss": 0.8153,
+      "step": 1999
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014956784018448603,
+      "loss": 0.8444,
+      "step": 2000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001495178961866467,
+      "loss": 0.8922,
+      "step": 2001
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014946793581911428,
+      "loss": 0.8611,
+      "step": 2002
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014941795909840478,
+      "loss": 0.3439,
+      "step": 2003
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014936796604103948,
+      "loss": 0.8544,
+      "step": 2004
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001493179566635452,
+      "loss": 0.8476,
+      "step": 2005
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014926793098245415,
+      "loss": 0.7953,
+      "step": 2006
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014921788901430381,
+      "loss": 0.8819,
+      "step": 2007
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014916783077563716,
+      "loss": 0.8655,
+      "step": 2008
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014911775628300252,
+      "loss": 0.8735,
+      "step": 2009
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014906766555295358,
+      "loss": 0.9141,
+      "step": 2010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014901755860204937,
+      "loss": 0.8566,
+      "step": 2011
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001489674354468544,
+      "loss": 0.809,
+      "step": 2012
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014891729610393836,
+      "loss": 0.8371,
+      "step": 2013
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014886714058987642,
+      "loss": 0.8196,
+      "step": 2014
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014881696892124905,
+      "loss": 0.8808,
+      "step": 2015
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001487667811146421,
+      "loss": 0.864,
+      "step": 2016
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014871657718664667,
+      "loss": 0.8457,
+      "step": 2017
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014866635715385927,
+      "loss": 0.8661,
+      "step": 2018
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014861612103288168,
+      "loss": 0.8798,
+      "step": 2019
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014856586884032108,
+      "loss": 0.8084,
+      "step": 2020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014851560059278985,
+      "loss": 0.8711,
+      "step": 2021
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014846531630690582,
+      "loss": 0.8854,
+      "step": 2022
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014841501599929192,
+      "loss": 0.8999,
+      "step": 2023
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014836469968657659,
+      "loss": 0.838,
+      "step": 2024
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001483143673853934,
+      "loss": 0.8759,
+      "step": 2025
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001482640191123813,
+      "loss": 0.8743,
+      "step": 2026
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001482136548841845,
+      "loss": 0.8598,
+      "step": 2027
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014816327471745244,
+      "loss": 0.8149,
+      "step": 2028
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001481128786288399,
+      "loss": 0.3398,
+      "step": 2029
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014806246663500686,
+      "loss": 0.3793,
+      "step": 2030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014801203875261862,
+      "loss": 0.8632,
+      "step": 2031
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014796159499834568,
+      "loss": 0.8301,
+      "step": 2032
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014791113538886378,
+      "loss": 0.8745,
+      "step": 2033
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014786065994085396,
+      "loss": 0.8237,
+      "step": 2034
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014781016867100244,
+      "loss": 0.872,
+      "step": 2035
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001477596615960007,
+      "loss": 0.8476,
+      "step": 2036
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014770913873254546,
+      "loss": 0.8225,
+      "step": 2037
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014765860009733858,
+      "loss": 0.8638,
+      "step": 2038
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014760804570708723,
+      "loss": 0.7848,
+      "step": 2039
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014755747557850378,
+      "loss": 0.9205,
+      "step": 2040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014750688972830567,
+      "loss": 0.8686,
+      "step": 2041
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014745628817321578,
+      "loss": 0.8239,
+      "step": 2042
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014740567092996193,
+      "loss": 0.8596,
+      "step": 2043
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014735503801527726,
+      "loss": 0.3336,
+      "step": 2044
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014730438944590007,
+      "loss": 0.9115,
+      "step": 2045
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014725372523857386,
+      "loss": 0.8824,
+      "step": 2046
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014720304541004724,
+      "loss": 0.7741,
+      "step": 2047
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014715234997707404,
+      "loss": 0.7515,
+      "step": 2048
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014710163895641318,
+      "loss": 0.884,
+      "step": 2049
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014705091236482887,
+      "loss": 0.8083,
+      "step": 2050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014700017021909028,
+      "loss": 0.9014,
+      "step": 2051
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014694941253597183,
+      "loss": 0.8501,
+      "step": 2052
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014689863933225314,
+      "loss": 0.784,
+      "step": 2053
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014684785062471883,
+      "loss": 0.8431,
+      "step": 2054
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014679704643015868,
+      "loss": 0.7955,
+      "step": 2055
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001467462267653676,
+      "loss": 0.8658,
+      "step": 2056
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001466953916471457,
+      "loss": 0.86,
+      "step": 2057
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014664454109229808,
+      "loss": 0.8894,
+      "step": 2058
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014659367511763496,
+      "loss": 0.8306,
+      "step": 2059
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014654279373997172,
+      "loss": 0.8633,
+      "step": 2060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014649189697612877,
+      "loss": 0.8544,
+      "step": 2061
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014644098484293164,
+      "loss": 0.8627,
+      "step": 2062
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014639005735721093,
+      "loss": 0.905,
+      "step": 2063
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001463391145358023,
+      "loss": 0.7922,
+      "step": 2064
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014628815639554656,
+      "loss": 0.9022,
+      "step": 2065
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014623718295328944,
+      "loss": 0.8785,
+      "step": 2066
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014618619422588185,
+      "loss": 0.8721,
+      "step": 2067
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014613519023017974,
+      "loss": 0.7961,
+      "step": 2068
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000146084170983044,
+      "loss": 0.8928,
+      "step": 2069
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014603313650134075,
+      "loss": 0.8758,
+      "step": 2070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014598208680194092,
+      "loss": 0.8185,
+      "step": 2071
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014593102190172067,
+      "loss": 0.8483,
+      "step": 2072
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014587994181756108,
+      "loss": 0.8516,
+      "step": 2073
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014582884656634827,
+      "loss": 0.7619,
+      "step": 2074
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014577773616497342,
+      "loss": 0.8052,
+      "step": 2075
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001457266106303326,
+      "loss": 0.8843,
+      "step": 2076
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000145675469979327,
+      "loss": 0.7892,
+      "step": 2077
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014562431422886272,
+      "loss": 0.8248,
+      "step": 2078
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014557314339585102,
+      "loss": 0.8667,
+      "step": 2079
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001455219574972079,
+      "loss": 0.7973,
+      "step": 2080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001454707565498545,
+      "loss": 0.8463,
+      "step": 2081
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014541954057071692,
+      "loss": 0.8353,
+      "step": 2082
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014536830957672615,
+      "loss": 0.8815,
+      "step": 2083
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001453170635848183,
+      "loss": 0.8498,
+      "step": 2084
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014526580261193424,
+      "loss": 0.7556,
+      "step": 2085
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014521452667501996,
+      "loss": 0.8727,
+      "step": 2086
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001451632357910263,
+      "loss": 0.8426,
+      "step": 2087
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014511192997690905,
+      "loss": 0.8839,
+      "step": 2088
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000145060609249629,
+      "loss": 0.8939,
+      "step": 2089
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014500927362615177,
+      "loss": 0.8802,
+      "step": 2090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014495792312344802,
+      "loss": 0.8041,
+      "step": 2091
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014490655775849324,
+      "loss": 0.8368,
+      "step": 2092
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014485517754826783,
+      "loss": 0.7869,
+      "step": 2093
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001448037825097572,
+      "loss": 0.8459,
+      "step": 2094
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001447523726599515,
+      "loss": 0.7566,
+      "step": 2095
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000144700948015846,
+      "loss": 0.8446,
+      "step": 2096
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001446495085944406,
+      "loss": 0.8883,
+      "step": 2097
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014459805441274028,
+      "loss": 0.8501,
+      "step": 2098
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001445465854877548,
+      "loss": 0.8757,
+      "step": 2099
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014449510183649886,
+      "loss": 0.8237,
+      "step": 2100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014444360347599198,
+      "loss": 0.8346,
+      "step": 2101
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014439209042325856,
+      "loss": 0.8194,
+      "step": 2102
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014434056269532782,
+      "loss": 0.836,
+      "step": 2103
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014428902030923392,
+      "loss": 0.8109,
+      "step": 2104
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014423746328201577,
+      "loss": 0.8575,
+      "step": 2105
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014418589163071722,
+      "loss": 0.7225,
+      "step": 2106
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001441343053723868,
+      "loss": 0.8105,
+      "step": 2107
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014408270452407807,
+      "loss": 0.7851,
+      "step": 2108
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014403108910284923,
+      "loss": 0.8854,
+      "step": 2109
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001439794591257634,
+      "loss": 0.8345,
+      "step": 2110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014392781460988852,
+      "loss": 0.84,
+      "step": 2111
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014387615557229726,
+      "loss": 0.8359,
+      "step": 2112
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014382448203006714,
+      "loss": 0.8043,
+      "step": 2113
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014377279400028053,
+      "loss": 0.8939,
+      "step": 2114
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001437210915000244,
+      "loss": 0.859,
+      "step": 2115
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014366937454639078,
+      "loss": 0.8831,
+      "step": 2116
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014361764315647623,
+      "loss": 0.897,
+      "step": 2117
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001435658973473822,
+      "loss": 0.8613,
+      "step": 2118
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014351413713621492,
+      "loss": 0.821,
+      "step": 2119
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014346236254008537,
+      "loss": 0.9085,
+      "step": 2120
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014341057357610917,
+      "loss": 0.8357,
+      "step": 2121
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014335877026140688,
+      "loss": 0.8317,
+      "step": 2122
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001433069526131037,
+      "loss": 0.3258,
+      "step": 2123
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014325512064832953,
+      "loss": 0.8567,
+      "step": 2124
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014320327438421907,
+      "loss": 0.9338,
+      "step": 2125
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014315141383791175,
+      "loss": 0.2998,
+      "step": 2126
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014309953902655166,
+      "loss": 0.8293,
+      "step": 2127
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001430476499672877,
+      "loss": 0.8991,
+      "step": 2128
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014299574667727338,
+      "loss": 0.8843,
+      "step": 2129
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000142943829173667,
+      "loss": 0.8984,
+      "step": 2130
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014289189747363143,
+      "loss": 0.7812,
+      "step": 2131
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014283995159433444,
+      "loss": 0.8436,
+      "step": 2132
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001427879915529483,
+      "loss": 0.7787,
+      "step": 2133
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014273601736665,
+      "loss": 0.7995,
+      "step": 2134
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014268402905262136,
+      "loss": 0.7955,
+      "step": 2135
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014263202662804863,
+      "loss": 0.8524,
+      "step": 2136
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014258001011012284,
+      "loss": 0.9035,
+      "step": 2137
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014252797951603977,
+      "loss": 0.8013,
+      "step": 2138
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001424759348629997,
+      "loss": 0.8624,
+      "step": 2139
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014242387616820762,
+      "loss": 0.8172,
+      "step": 2140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001423718034488732,
+      "loss": 0.8478,
+      "step": 2141
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001423197167222107,
+      "loss": 0.8253,
+      "step": 2142
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000142267616005439,
+      "loss": 0.8213,
+      "step": 2143
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014221550131578162,
+      "loss": 0.3819,
+      "step": 2144
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014216337267046675,
+      "loss": 0.8399,
+      "step": 2145
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014211123008672712,
+      "loss": 0.9245,
+      "step": 2146
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001420590735818001,
+      "loss": 0.8772,
+      "step": 2147
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001420069031729276,
+      "loss": 0.8259,
+      "step": 2148
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001419547188773563,
+      "loss": 0.8644,
+      "step": 2149
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014190252071233727,
+      "loss": 0.8438,
+      "step": 2150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014185030869512624,
+      "loss": 0.8374,
+      "step": 2151
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001417980828429836,
+      "loss": 0.8573,
+      "step": 2152
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014174584317317422,
+      "loss": 0.8214,
+      "step": 2153
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001416935897029675,
+      "loss": 0.8201,
+      "step": 2154
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014164132244963751,
+      "loss": 0.8267,
+      "step": 2155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014158904143046286,
+      "loss": 0.8548,
+      "step": 2156
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014153674666272665,
+      "loss": 0.8056,
+      "step": 2157
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001414844381637165,
+      "loss": 0.869,
+      "step": 2158
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014143211595072473,
+      "loss": 0.8041,
+      "step": 2159
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014137978004104802,
+      "loss": 0.8506,
+      "step": 2160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014132743045198765,
+      "loss": 0.7709,
+      "step": 2161
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001412750672008494,
+      "loss": 0.8721,
+      "step": 2162
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014122269030494367,
+      "loss": 0.3663,
+      "step": 2163
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001411702997815852,
+      "loss": 0.8314,
+      "step": 2164
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014111789564809335,
+      "loss": 0.844,
+      "step": 2165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014106547792179196,
+      "loss": 0.8779,
+      "step": 2166
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014101304662000934,
+      "loss": 0.8575,
+      "step": 2167
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014096060176007827,
+      "loss": 0.8699,
+      "step": 2168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001409081433593361,
+      "loss": 0.9272,
+      "step": 2169
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014085567143512457,
+      "loss": 0.8129,
+      "step": 2170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001408031860047899,
+      "loss": 0.819,
+      "step": 2171
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014075068708568284,
+      "loss": 0.3648,
+      "step": 2172
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014069817469515854,
+      "loss": 0.3435,
+      "step": 2173
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014064564885057657,
+      "loss": 0.8733,
+      "step": 2174
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014059310956930106,
+      "loss": 0.7951,
+      "step": 2175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001405405568687005,
+      "loss": 0.8949,
+      "step": 2176
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001404879907661478,
+      "loss": 0.8042,
+      "step": 2177
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014043541127902037,
+      "loss": 0.8845,
+      "step": 2178
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014038281842469996,
+      "loss": 0.8728,
+      "step": 2179
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014033021222057283,
+      "loss": 0.9183,
+      "step": 2180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014027759268402963,
+      "loss": 0.8617,
+      "step": 2181
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014022495983246534,
+      "loss": 0.8094,
+      "step": 2182
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014017231368327945,
+      "loss": 0.8812,
+      "step": 2183
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014011965425387573,
+      "loss": 0.8838,
+      "step": 2184
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014006698156166249,
+      "loss": 0.8397,
+      "step": 2185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014001429562405225,
+      "loss": 0.8031,
+      "step": 2186
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001399615964584621,
+      "loss": 0.7626,
+      "step": 2187
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013990888408231333,
+      "loss": 0.852,
+      "step": 2188
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001398561585130317,
+      "loss": 0.8099,
+      "step": 2189
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013980341976804726,
+      "loss": 0.8449,
+      "step": 2190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013975066786479448,
+      "loss": 0.9235,
+      "step": 2191
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013969790282071217,
+      "loss": 0.8875,
+      "step": 2192
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013964512465324344,
+      "loss": 0.8641,
+      "step": 2193
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013959233337983582,
+      "loss": 0.3509,
+      "step": 2194
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013953952901794102,
+      "loss": 0.8423,
+      "step": 2195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001394867115850153,
+      "loss": 0.8549,
+      "step": 2196
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013943388109851905,
+      "loss": 0.3663,
+      "step": 2197
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013938103757591704,
+      "loss": 0.898,
+      "step": 2198
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013932818103467838,
+      "loss": 0.8123,
+      "step": 2199
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013927531149227645,
+      "loss": 0.8585,
+      "step": 2200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013922242896618892,
+      "loss": 0.7758,
+      "step": 2201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013916953347389776,
+      "loss": 0.8394,
+      "step": 2202
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013911662503288927,
+      "loss": 0.8484,
+      "step": 2203
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013906370366065398,
+      "loss": 0.8927,
+      "step": 2204
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001390107693746867,
+      "loss": 0.8812,
+      "step": 2205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001389578221924865,
+      "loss": 0.8791,
+      "step": 2206
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013890486213155676,
+      "loss": 0.8668,
+      "step": 2207
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013885188920940506,
+      "loss": 0.7987,
+      "step": 2208
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013879890344354332,
+      "loss": 0.7876,
+      "step": 2209
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001387459048514876,
+      "loss": 0.8296,
+      "step": 2210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013869289345075822,
+      "loss": 0.8213,
+      "step": 2211
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013863986925887983,
+      "loss": 0.8422,
+      "step": 2212
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001385868322933812,
+      "loss": 0.8438,
+      "step": 2213
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013853378257179535,
+      "loss": 0.8215,
+      "step": 2214
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013848072011165953,
+      "loss": 0.8966,
+      "step": 2215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013842764493051526,
+      "loss": 0.8387,
+      "step": 2216
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001383745570459081,
+      "loss": 0.83,
+      "step": 2217
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013832145647538799,
+      "loss": 0.7642,
+      "step": 2218
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000138268343236509,
+      "loss": 0.7992,
+      "step": 2219
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013821521734682933,
+      "loss": 0.8401,
+      "step": 2220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013816207882391142,
+      "loss": 0.8996,
+      "step": 2221
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013810892768532186,
+      "loss": 0.8818,
+      "step": 2222
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013805576394863148,
+      "loss": 0.8398,
+      "step": 2223
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013800258763141518,
+      "loss": 0.8133,
+      "step": 2224
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013794939875125205,
+      "loss": 0.7863,
+      "step": 2225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013789619732572538,
+      "loss": 0.8414,
+      "step": 2226
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013784298337242255,
+      "loss": 0.8771,
+      "step": 2227
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013778975690893503,
+      "loss": 0.8251,
+      "step": 2228
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013773651795285862,
+      "loss": 0.8552,
+      "step": 2229
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013768326652179307,
+      "loss": 0.8184,
+      "step": 2230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013763000263334227,
+      "loss": 0.831,
+      "step": 2231
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013757672630511434,
+      "loss": 0.7648,
+      "step": 2232
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001375234375547214,
+      "loss": 0.8828,
+      "step": 2233
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013747013639977973,
+      "loss": 0.8473,
+      "step": 2234
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001374168228579097,
+      "loss": 0.7871,
+      "step": 2235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013736349694673576,
+      "loss": 0.7806,
+      "step": 2236
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013731015868388645,
+      "loss": 0.3499,
+      "step": 2237
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013725680808699444,
+      "loss": 0.8132,
+      "step": 2238
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001372034451736964,
+      "loss": 0.8485,
+      "step": 2239
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013715006996163317,
+      "loss": 0.8491,
+      "step": 2240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013709668246844955,
+      "loss": 0.8162,
+      "step": 2241
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001370432827117945,
+      "loss": 0.7925,
+      "step": 2242
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013698987070932095,
+      "loss": 0.8627,
+      "step": 2243
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013693644647868586,
+      "loss": 0.8803,
+      "step": 2244
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013688301003755038,
+      "loss": 0.8877,
+      "step": 2245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013682956140357954,
+      "loss": 0.836,
+      "step": 2246
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001367761005944425,
+      "loss": 0.8778,
+      "step": 2247
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013672262762781242,
+      "loss": 0.8282,
+      "step": 2248
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013666914252136636,
+      "loss": 0.8883,
+      "step": 2249
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001366156452927856,
+      "loss": 0.9406,
+      "step": 2250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013656213595975532,
+      "loss": 0.8625,
+      "step": 2251
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013650861453996465,
+      "loss": 0.8289,
+      "step": 2252
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013645508105110686,
+      "loss": 0.8886,
+      "step": 2253
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013640153551087902,
+      "loss": 0.8557,
+      "step": 2254
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013634797793698236,
+      "loss": 0.8361,
+      "step": 2255
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000136294408347122,
+      "loss": 0.7781,
+      "step": 2256
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013624082675900705,
+      "loss": 0.8765,
+      "step": 2257
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013618723319035056,
+      "loss": 0.8569,
+      "step": 2258
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001361336276588696,
+      "loss": 0.8941,
+      "step": 2259
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013608001018228512,
+      "loss": 0.8099,
+      "step": 2260
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013602638077832207,
+      "loss": 0.7868,
+      "step": 2261
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013597273946470937,
+      "loss": 0.8621,
+      "step": 2262
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013591908625917978,
+      "loss": 0.8115,
+      "step": 2263
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001358654211794701,
+      "loss": 0.3581,
+      "step": 2264
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013581174424332097,
+      "loss": 0.8639,
+      "step": 2265
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000135758055468477,
+      "loss": 0.8659,
+      "step": 2266
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013570435487268668,
+      "loss": 0.8862,
+      "step": 2267
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013565064247370248,
+      "loss": 0.8682,
+      "step": 2268
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013559691828928062,
+      "loss": 0.8216,
+      "step": 2269
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013554318233718136,
+      "loss": 0.886,
+      "step": 2270
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013548943463516884,
+      "loss": 0.7924,
+      "step": 2271
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013543567520101106,
+      "loss": 0.8532,
+      "step": 2272
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001353819040524798,
+      "loss": 0.8293,
+      "step": 2273
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013532812120735087,
+      "loss": 0.8379,
+      "step": 2274
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013527432668340383,
+      "loss": 0.3332,
+      "step": 2275
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013522052049842216,
+      "loss": 0.3412,
+      "step": 2276
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001351667026701932,
+      "loss": 0.8093,
+      "step": 2277
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001351128732165081,
+      "loss": 0.8427,
+      "step": 2278
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013505903215516192,
+      "loss": 0.7791,
+      "step": 2279
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013500517950395348,
+      "loss": 0.7944,
+      "step": 2280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013495131528068542,
+      "loss": 0.8382,
+      "step": 2281
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001348974395031643,
+      "loss": 0.8555,
+      "step": 2282
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013484355218920045,
+      "loss": 0.8586,
+      "step": 2283
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013478965335660798,
+      "loss": 0.8503,
+      "step": 2284
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013473574302320484,
+      "loss": 0.8438,
+      "step": 2285
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013468182120681278,
+      "loss": 0.8017,
+      "step": 2286
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001346278879252574,
+      "loss": 0.8164,
+      "step": 2287
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000134573943196368,
+      "loss": 0.8661,
+      "step": 2288
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013451998703797765,
+      "loss": 0.7904,
+      "step": 2289
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013446601946792334,
+      "loss": 0.789,
+      "step": 2290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001344120405040457,
+      "loss": 0.9156,
+      "step": 2291
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013435805016418913,
+      "loss": 0.7867,
+      "step": 2292
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001343040484662019,
+      "loss": 0.816,
+      "step": 2293
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013425003542793596,
+      "loss": 0.7978,
+      "step": 2294
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013419601106724692,
+      "loss": 0.8046,
+      "step": 2295
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013414197540199436,
+      "loss": 0.8297,
+      "step": 2296
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013408792845004134,
+      "loss": 0.7996,
+      "step": 2297
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013403387022925488,
+      "loss": 0.7825,
+      "step": 2298
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013397980075750552,
+      "loss": 0.8874,
+      "step": 2299
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001339257200526677,
+      "loss": 0.7835,
+      "step": 2300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013387162813261946,
+      "loss": 0.9031,
+      "step": 2301
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001338175250152426,
+      "loss": 0.7939,
+      "step": 2302
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013376341071842256,
+      "loss": 0.8404,
+      "step": 2303
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013370928526004855,
+      "loss": 0.8396,
+      "step": 2304
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013365514865801344,
+      "loss": 0.8458,
+      "step": 2305
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013360100093021376,
+      "loss": 0.8416,
+      "step": 2306
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013354684209454976,
+      "loss": 0.8133,
+      "step": 2307
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013349267216892529,
+      "loss": 0.8099,
+      "step": 2308
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000133438491171248,
+      "loss": 0.818,
+      "step": 2309
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013338429911942908,
+      "loss": 0.8298,
+      "step": 2310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013333009603138337,
+      "loss": 0.7972,
+      "step": 2311
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013327588192502948,
+      "loss": 0.8698,
+      "step": 2312
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001332216568182895,
+      "loss": 0.783,
+      "step": 2313
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013316742072908927,
+      "loss": 0.8479,
+      "step": 2314
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013311317367535827,
+      "loss": 0.8499,
+      "step": 2315
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013305891567502953,
+      "loss": 0.8485,
+      "step": 2316
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013300464674603974,
+      "loss": 0.9276,
+      "step": 2317
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001329503669063292,
+      "loss": 0.7766,
+      "step": 2318
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013289607617384175,
+      "loss": 0.8718,
+      "step": 2319
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000132841774566525,
+      "loss": 0.8834,
+      "step": 2320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013278746210232997,
+      "loss": 0.8777,
+      "step": 2321
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001327331387992114,
+      "loss": 0.8229,
+      "step": 2322
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013267880467512753,
+      "loss": 0.8371,
+      "step": 2323
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001326244597480402,
+      "loss": 0.8111,
+      "step": 2324
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013257010403591485,
+      "loss": 0.8438,
+      "step": 2325
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013251573755672047,
+      "loss": 0.8065,
+      "step": 2326
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013246136032842956,
+      "loss": 0.881,
+      "step": 2327
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001324069723690183,
+      "loss": 0.8615,
+      "step": 2328
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013235257369646622,
+      "loss": 0.8204,
+      "step": 2329
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013229816432875664,
+      "loss": 0.8123,
+      "step": 2330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001322437442838762,
+      "loss": 0.8436,
+      "step": 2331
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013218931357981514,
+      "loss": 0.8054,
+      "step": 2332
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001321348722345673,
+      "loss": 0.3506,
+      "step": 2333
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001320804202661299,
+      "loss": 0.3403,
+      "step": 2334
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013202595769250378,
+      "loss": 0.8195,
+      "step": 2335
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001319714845316933,
+      "loss": 0.3663,
+      "step": 2336
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013191700080170617,
+      "loss": 0.7965,
+      "step": 2337
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013186250652055378,
+      "loss": 0.7977,
+      "step": 2338
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013180800170625089,
+      "loss": 0.8032,
+      "step": 2339
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013175348637681575,
+      "loss": 0.8736,
+      "step": 2340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013169896055027018,
+      "loss": 0.819,
+      "step": 2341
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013164442424463935,
+      "loss": 0.3379,
+      "step": 2342
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013158987747795192,
+      "loss": 0.7853,
+      "step": 2343
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001315353202682401,
+      "loss": 0.8419,
+      "step": 2344
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013148075263353946,
+      "loss": 0.822,
+      "step": 2345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013142617459188899,
+      "loss": 0.8546,
+      "step": 2346
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013137158616133128,
+      "loss": 0.8728,
+      "step": 2347
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013131698735991217,
+      "loss": 0.8896,
+      "step": 2348
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013126237820568104,
+      "loss": 0.8113,
+      "step": 2349
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001312077587166906,
+      "loss": 0.8235,
+      "step": 2350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013115312891099712,
+      "loss": 0.8344,
+      "step": 2351
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013109848880666014,
+      "loss": 0.8507,
+      "step": 2352
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013104383842174266,
+      "loss": 0.7804,
+      "step": 2353
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001309891777743111,
+      "loss": 0.8269,
+      "step": 2354
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013093450688243525,
+      "loss": 0.8841,
+      "step": 2355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013087982576418823,
+      "loss": 0.8292,
+      "step": 2356
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001308251344376467,
+      "loss": 0.8779,
+      "step": 2357
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013077043292089054,
+      "loss": 0.8274,
+      "step": 2358
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013071572123200302,
+      "loss": 0.8493,
+      "step": 2359
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013066099938907085,
+      "loss": 0.8807,
+      "step": 2360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000130606267410184,
+      "loss": 0.8155,
+      "step": 2361
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013055152531343592,
+      "loss": 0.8419,
+      "step": 2362
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013049677311692326,
+      "loss": 0.8463,
+      "step": 2363
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013044201083874612,
+      "loss": 0.7918,
+      "step": 2364
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013038723849700785,
+      "loss": 0.819,
+      "step": 2365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013033245610981516,
+      "loss": 0.8411,
+      "step": 2366
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013027766369527814,
+      "loss": 0.8575,
+      "step": 2367
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013022286127151007,
+      "loss": 0.8464,
+      "step": 2368
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013016804885662768,
+      "loss": 0.8546,
+      "step": 2369
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013011322646875088,
+      "loss": 0.8446,
+      "step": 2370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013005839412600295,
+      "loss": 0.8809,
+      "step": 2371
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013000355184651045,
+      "loss": 0.8513,
+      "step": 2372
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001299486996484032,
+      "loss": 0.828,
+      "step": 2373
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001298938375498143,
+      "loss": 0.8422,
+      "step": 2374
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012983896556888018,
+      "loss": 0.8295,
+      "step": 2375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012978408372374048,
+      "loss": 0.7739,
+      "step": 2376
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012972919203253804,
+      "loss": 0.8525,
+      "step": 2377
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012967429051341913,
+      "loss": 0.8709,
+      "step": 2378
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012961937918453312,
+      "loss": 0.8058,
+      "step": 2379
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001295644580640327,
+      "loss": 0.7863,
+      "step": 2380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012950952717007374,
+      "loss": 0.823,
+      "step": 2381
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012945458652081535,
+      "loss": 0.7021,
+      "step": 2382
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012939963613441995,
+      "loss": 0.8402,
+      "step": 2383
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012934467602905304,
+      "loss": 0.9096,
+      "step": 2384
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012928970622288347,
+      "loss": 0.8795,
+      "step": 2385
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012923472673408322,
+      "loss": 0.768,
+      "step": 2386
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012917973758082743,
+      "loss": 0.8458,
+      "step": 2387
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012912473878129454,
+      "loss": 0.8342,
+      "step": 2388
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012906973035366613,
+      "loss": 0.7949,
+      "step": 2389
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001290147123161269,
+      "loss": 0.8659,
+      "step": 2390
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012895968468686485,
+      "loss": 0.8332,
+      "step": 2391
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001289046474840711,
+      "loss": 0.8488,
+      "step": 2392
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012884960072593986,
+      "loss": 0.8435,
+      "step": 2393
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001287945444306686,
+      "loss": 0.9362,
+      "step": 2394
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001287394786164579,
+      "loss": 0.8887,
+      "step": 2395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012868440330151152,
+      "loss": 0.829,
+      "step": 2396
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001286293185040363,
+      "loss": 0.8695,
+      "step": 2397
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001285742242422422,
+      "loss": 0.8689,
+      "step": 2398
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012851912053434244,
+      "loss": 0.823,
+      "step": 2399
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012846400739855324,
+      "loss": 0.8315,
+      "step": 2400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000128408884853094,
+      "loss": 0.8246,
+      "step": 2401
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012835375291618716,
+      "loss": 0.799,
+      "step": 2402
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012829861160605833,
+      "loss": 0.8371,
+      "step": 2403
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001282434609409362,
+      "loss": 0.839,
+      "step": 2404
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001281883009390526,
+      "loss": 0.7539,
+      "step": 2405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012813313161864228,
+      "loss": 0.8694,
+      "step": 2406
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012807795299794332,
+      "loss": 0.8501,
+      "step": 2407
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012802276509519666,
+      "loss": 0.9039,
+      "step": 2408
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001279675679286464,
+      "loss": 0.874,
+      "step": 2409
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012791236151653973,
+      "loss": 0.8852,
+      "step": 2410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012785714587712686,
+      "loss": 0.8074,
+      "step": 2411
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012780192102866098,
+      "loss": 0.8633,
+      "step": 2412
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012774668698939848,
+      "loss": 0.7611,
+      "step": 2413
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012769144377759866,
+      "loss": 0.8423,
+      "step": 2414
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012763619141152387,
+      "loss": 0.8748,
+      "step": 2415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012758092990943962,
+      "loss": 0.7678,
+      "step": 2416
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012752565928961422,
+      "loss": 0.8643,
+      "step": 2417
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012747037957031916,
+      "loss": 0.8634,
+      "step": 2418
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012741509076982885,
+      "loss": 0.8343,
+      "step": 2419
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012735979290642076,
+      "loss": 0.7839,
+      "step": 2420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012730448599837532,
+      "loss": 0.8447,
+      "step": 2421
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000127249170063976,
+      "loss": 0.8627,
+      "step": 2422
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012719384512150915,
+      "loss": 0.8463,
+      "step": 2423
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012713851118926426,
+      "loss": 0.8147,
+      "step": 2424
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012708316828553357,
+      "loss": 0.7978,
+      "step": 2425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012702781642861253,
+      "loss": 0.8587,
+      "step": 2426
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012697245563679936,
+      "loss": 0.8467,
+      "step": 2427
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012691708592839533,
+      "loss": 0.8679,
+      "step": 2428
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012686170732170463,
+      "loss": 0.8509,
+      "step": 2429
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012680631983503436,
+      "loss": 0.7952,
+      "step": 2430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012675092348669469,
+      "loss": 0.846,
+      "step": 2431
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012669551829499852,
+      "loss": 0.8638,
+      "step": 2432
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001266401042782618,
+      "loss": 0.8659,
+      "step": 2433
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012658468145480337,
+      "loss": 0.8589,
+      "step": 2434
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012652924984294503,
+      "loss": 0.8496,
+      "step": 2435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001264738094610114,
+      "loss": 0.8149,
+      "step": 2436
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012641836032732998,
+      "loss": 0.8188,
+      "step": 2437
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001263629024602313,
+      "loss": 0.8175,
+      "step": 2438
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012630743587804867,
+      "loss": 0.8832,
+      "step": 2439
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012625196059911834,
+      "loss": 0.8724,
+      "step": 2440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012619647664177932,
+      "loss": 0.7779,
+      "step": 2441
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012614098402437366,
+      "loss": 0.7806,
+      "step": 2442
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012608548276524614,
+      "loss": 0.8885,
+      "step": 2443
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012602997288274444,
+      "loss": 0.8481,
+      "step": 2444
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012597445439521912,
+      "loss": 0.7599,
+      "step": 2445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001259189273210235,
+      "loss": 0.8161,
+      "step": 2446
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012586339167851385,
+      "loss": 0.7991,
+      "step": 2447
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012580784748604922,
+      "loss": 0.844,
+      "step": 2448
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012575229476199145,
+      "loss": 0.8815,
+      "step": 2449
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012569673352470523,
+      "loss": 0.8275,
+      "step": 2450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012564116379255811,
+      "loss": 0.8841,
+      "step": 2451
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012558558558392038,
+      "loss": 0.7678,
+      "step": 2452
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012552999891716517,
+      "loss": 0.901,
+      "step": 2453
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001254744038106684,
+      "loss": 0.8079,
+      "step": 2454
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012541880028280874,
+      "loss": 0.7923,
+      "step": 2455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012536318835196773,
+      "loss": 0.3545,
+      "step": 2456
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012530756803652955,
+      "loss": 0.8471,
+      "step": 2457
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012525193935488137,
+      "loss": 0.8639,
+      "step": 2458
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001251963023254129,
+      "loss": 0.7829,
+      "step": 2459
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012514065696651674,
+      "loss": 0.9179,
+      "step": 2460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012508500329658822,
+      "loss": 0.8695,
+      "step": 2461
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012502934133402533,
+      "loss": 0.9035,
+      "step": 2462
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012497367109722896,
+      "loss": 0.8835,
+      "step": 2463
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012491799260460265,
+      "loss": 0.7645,
+      "step": 2464
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001248623058745526,
+      "loss": 0.7751,
+      "step": 2465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012480661092548786,
+      "loss": 0.8353,
+      "step": 2466
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012475090777582017,
+      "loss": 0.7973,
+      "step": 2467
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012469519644396385,
+      "loss": 0.8745,
+      "step": 2468
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012463947694833614,
+      "loss": 0.7758,
+      "step": 2469
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001245837493073568,
+      "loss": 0.3244,
+      "step": 2470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012452801353944837,
+      "loss": 0.8102,
+      "step": 2471
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012447226966303605,
+      "loss": 0.7661,
+      "step": 2472
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001244165176965477,
+      "loss": 0.7846,
+      "step": 2473
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012436075765841396,
+      "loss": 0.9018,
+      "step": 2474
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012430498956706799,
+      "loss": 0.7901,
+      "step": 2475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012424921344094566,
+      "loss": 0.8636,
+      "step": 2476
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001241934292984856,
+      "loss": 0.8404,
+      "step": 2477
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001241376371581289,
+      "loss": 0.7971,
+      "step": 2478
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012408183703831955,
+      "loss": 0.7998,
+      "step": 2479
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001240260289575039,
+      "loss": 0.7588,
+      "step": 2480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001239702129341311,
+      "loss": 0.8292,
+      "step": 2481
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012391438898665287,
+      "loss": 0.8106,
+      "step": 2482
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012385855713352363,
+      "loss": 0.8084,
+      "step": 2483
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012380271739320027,
+      "loss": 0.8546,
+      "step": 2484
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012374686978414243,
+      "loss": 0.8355,
+      "step": 2485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012369101432481224,
+      "loss": 0.8578,
+      "step": 2486
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001236351510336745,
+      "loss": 0.8254,
+      "step": 2487
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012357927992919657,
+      "loss": 0.8173,
+      "step": 2488
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012352340102984838,
+      "loss": 0.3535,
+      "step": 2489
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012346751435410248,
+      "loss": 0.8389,
+      "step": 2490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012341161992043394,
+      "loss": 0.8092,
+      "step": 2491
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012335571774732044,
+      "loss": 0.76,
+      "step": 2492
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012329980785324217,
+      "loss": 0.8466,
+      "step": 2493
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001232438902566819,
+      "loss": 0.8481,
+      "step": 2494
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012318796497612494,
+      "loss": 0.8104,
+      "step": 2495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001231320320300592,
+      "loss": 0.8666,
+      "step": 2496
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012307609143697497,
+      "loss": 0.8137,
+      "step": 2497
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001230201432153653,
+      "loss": 0.8344,
+      "step": 2498
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001229641873837255,
+      "loss": 0.8607,
+      "step": 2499
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012290822396055355,
+      "loss": 0.8459,
+      "step": 2500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012285225296435,
+      "loss": 0.8849,
+      "step": 2501
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012279627441361772,
+      "loss": 0.8673,
+      "step": 2502
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012274028832686223,
+      "loss": 0.7707,
+      "step": 2503
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012268429472259143,
+      "loss": 0.8163,
+      "step": 2504
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012262829361931588,
+      "loss": 0.7824,
+      "step": 2505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012257228503554835,
+      "loss": 0.8081,
+      "step": 2506
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012251626898980434,
+      "loss": 0.8157,
+      "step": 2507
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012246024550060166,
+      "loss": 0.8408,
+      "step": 2508
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012240421458646066,
+      "loss": 0.8695,
+      "step": 2509
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001223481762659041,
+      "loss": 0.8122,
+      "step": 2510
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001222921305574572,
+      "loss": 0.8459,
+      "step": 2511
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012223607747964766,
+      "loss": 0.8181,
+      "step": 2512
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012218001705100552,
+      "loss": 0.8498,
+      "step": 2513
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012212394929006336,
+      "loss": 0.8339,
+      "step": 2514
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012206787421535612,
+      "loss": 0.772,
+      "step": 2515
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012201179184542115,
+      "loss": 0.8102,
+      "step": 2516
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001219557021987983,
+      "loss": 0.8846,
+      "step": 2517
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012189960529402971,
+      "loss": 0.7772,
+      "step": 2518
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012184350114965998,
+      "loss": 0.8578,
+      "step": 2519
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012178738978423612,
+      "loss": 0.8396,
+      "step": 2520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012173127121630745,
+      "loss": 0.3559,
+      "step": 2521
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012167514546442576,
+      "loss": 0.8248,
+      "step": 2522
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012161901254714516,
+      "loss": 0.848,
+      "step": 2523
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012156287248302219,
+      "loss": 0.8191,
+      "step": 2524
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012150672529061562,
+      "loss": 0.8353,
+      "step": 2525
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012145057098848673,
+      "loss": 0.8659,
+      "step": 2526
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001213944095951991,
+      "loss": 0.8776,
+      "step": 2527
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012133824112931858,
+      "loss": 0.7983,
+      "step": 2528
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012128206560941346,
+      "loss": 0.9326,
+      "step": 2529
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012122588305405434,
+      "loss": 0.8526,
+      "step": 2530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012116969348181407,
+      "loss": 0.8366,
+      "step": 2531
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012111349691126785,
+      "loss": 0.8745,
+      "step": 2532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012105729336099334,
+      "loss": 0.7847,
+      "step": 2533
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012100108284957028,
+      "loss": 0.8053,
+      "step": 2534
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012094486539558085,
+      "loss": 0.834,
+      "step": 2535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001208886410176095,
+      "loss": 0.8753,
+      "step": 2536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012083240973424292,
+      "loss": 0.8419,
+      "step": 2537
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001207761715640702,
+      "loss": 0.8894,
+      "step": 2538
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012071992652568256,
+      "loss": 0.8019,
+      "step": 2539
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012066367463767361,
+      "loss": 0.3667,
+      "step": 2540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012060741591863913,
+      "loss": 0.8513,
+      "step": 2541
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012055115038717722,
+      "loss": 0.8416,
+      "step": 2542
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012049487806188822,
+      "loss": 0.8452,
+      "step": 2543
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012043859896137472,
+      "loss": 0.366,
+      "step": 2544
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012038231310424155,
+      "loss": 0.807,
+      "step": 2545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012032602050909574,
+      "loss": 0.8337,
+      "step": 2546
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012026972119454656,
+      "loss": 0.7296,
+      "step": 2547
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012021341517920555,
+      "loss": 0.7914,
+      "step": 2548
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012015710248168641,
+      "loss": 0.8897,
+      "step": 2549
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012010078312060504,
+      "loss": 0.8144,
+      "step": 2550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012004445711457966,
+      "loss": 0.8511,
+      "step": 2551
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011998812448223049,
+      "loss": 0.8561,
+      "step": 2552
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011993178524218012,
+      "loss": 0.8213,
+      "step": 2553
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011987543941305321,
+      "loss": 0.8372,
+      "step": 2554
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011981908701347668,
+      "loss": 0.9105,
+      "step": 2555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011976272806207956,
+      "loss": 0.8269,
+      "step": 2556
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011970636257749306,
+      "loss": 0.8338,
+      "step": 2557
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011964999057835055,
+      "loss": 0.8666,
+      "step": 2558
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011959361208328761,
+      "loss": 0.8307,
+      "step": 2559
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011953722711094189,
+      "loss": 0.8706,
+      "step": 2560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011948083567995319,
+      "loss": 0.8009,
+      "step": 2561
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011942443780896351,
+      "loss": 0.7942,
+      "step": 2562
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011936803351661688,
+      "loss": 0.828,
+      "step": 2563
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011931162282155953,
+      "loss": 0.8385,
+      "step": 2564
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001192552057424398,
+      "loss": 0.8265,
+      "step": 2565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011919878229790813,
+      "loss": 0.868,
+      "step": 2566
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011914235250661705,
+      "loss": 0.8189,
+      "step": 2567
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011908591638722115,
+      "loss": 0.8765,
+      "step": 2568
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011902947395837723,
+      "loss": 0.7791,
+      "step": 2569
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011897302523874405,
+      "loss": 0.8291,
+      "step": 2570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011891657024698255,
+      "loss": 0.8045,
+      "step": 2571
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011886010900175564,
+      "loss": 0.8448,
+      "step": 2572
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011880364152172842,
+      "loss": 0.8287,
+      "step": 2573
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011874716782556794,
+      "loss": 0.8109,
+      "step": 2574
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011869068793194336,
+      "loss": 0.8658,
+      "step": 2575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001186342018595259,
+      "loss": 0.7962,
+      "step": 2576
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001185777096269888,
+      "loss": 0.7898,
+      "step": 2577
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001185212112530073,
+      "loss": 0.8347,
+      "step": 2578
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011846470675625876,
+      "loss": 0.7847,
+      "step": 2579
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011840819615542247,
+      "loss": 0.3408,
+      "step": 2580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011835167946917979,
+      "loss": 0.8443,
+      "step": 2581
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011829515671621412,
+      "loss": 0.8419,
+      "step": 2582
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011823862791521078,
+      "loss": 0.7914,
+      "step": 2583
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011818209308485717,
+      "loss": 0.8269,
+      "step": 2584
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011812555224384261,
+      "loss": 0.8009,
+      "step": 2585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001180690054108585,
+      "loss": 0.8467,
+      "step": 2586
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011801245260459815,
+      "loss": 0.8947,
+      "step": 2587
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011795589384375686,
+      "loss": 0.8427,
+      "step": 2588
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011789932914703187,
+      "loss": 0.8372,
+      "step": 2589
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011784275853312245,
+      "loss": 0.8311,
+      "step": 2590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001177861820207298,
+      "loss": 0.834,
+      "step": 2591
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011772959962855704,
+      "loss": 0.7952,
+      "step": 2592
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011767301137530926,
+      "loss": 0.8378,
+      "step": 2593
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011761641727969343,
+      "loss": 0.796,
+      "step": 2594
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011755981736041861,
+      "loss": 0.3425,
+      "step": 2595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001175032116361956,
+      "loss": 0.8197,
+      "step": 2596
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001174466001257372,
+      "loss": 0.777,
+      "step": 2597
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011738998284775815,
+      "loss": 0.8442,
+      "step": 2598
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011733335982097504,
+      "loss": 0.885,
+      "step": 2599
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011727673106410642,
+      "loss": 0.8542,
+      "step": 2600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011722009659587265,
+      "loss": 0.8504,
+      "step": 2601
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011716345643499608,
+      "loss": 0.8629,
+      "step": 2602
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011710681060020085,
+      "loss": 0.8224,
+      "step": 2603
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011705015911021307,
+      "loss": 0.8044,
+      "step": 2604
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011699350198376063,
+      "loss": 0.8232,
+      "step": 2605
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011693683923957328,
+      "loss": 0.781,
+      "step": 2606
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011688017089638272,
+      "loss": 0.7816,
+      "step": 2607
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011682349697292245,
+      "loss": 0.8255,
+      "step": 2608
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011676681748792781,
+      "loss": 0.8358,
+      "step": 2609
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011671013246013596,
+      "loss": 0.8129,
+      "step": 2610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011665344190828591,
+      "loss": 0.7958,
+      "step": 2611
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001165967458511185,
+      "loss": 0.7858,
+      "step": 2612
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011654004430737642,
+      "loss": 0.8208,
+      "step": 2613
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011648333729580412,
+      "loss": 0.8404,
+      "step": 2614
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001164266248351479,
+      "loss": 0.8044,
+      "step": 2615
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001163699069441558,
+      "loss": 0.8379,
+      "step": 2616
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011631318364157773,
+      "loss": 0.897,
+      "step": 2617
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011625645494616535,
+      "loss": 0.7899,
+      "step": 2618
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011619972087667211,
+      "loss": 0.7876,
+      "step": 2619
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011614298145185323,
+      "loss": 0.8139,
+      "step": 2620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011608623669046572,
+      "loss": 0.8209,
+      "step": 2621
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011602948661126828,
+      "loss": 0.8499,
+      "step": 2622
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011597273123302154,
+      "loss": 0.8969,
+      "step": 2623
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011591597057448769,
+      "loss": 0.3871,
+      "step": 2624
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011585920465443076,
+      "loss": 0.8178,
+      "step": 2625
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001158024334916165,
+      "loss": 0.8721,
+      "step": 2626
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011574565710481243,
+      "loss": 0.889,
+      "step": 2627
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011568887551278768,
+      "loss": 0.855,
+      "step": 2628
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011563208873431331,
+      "loss": 0.8431,
+      "step": 2629
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011557529678816188,
+      "loss": 0.7651,
+      "step": 2630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011551849969310777,
+      "loss": 0.7542,
+      "step": 2631
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011546169746792705,
+      "loss": 0.8137,
+      "step": 2632
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011540489013139746,
+      "loss": 0.834,
+      "step": 2633
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011534807770229845,
+      "loss": 0.8069,
+      "step": 2634
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011529126019941116,
+      "loss": 0.8066,
+      "step": 2635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011523443764151842,
+      "loss": 0.7995,
+      "step": 2636
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011517761004740467,
+      "loss": 0.8306,
+      "step": 2637
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011512077743585603,
+      "loss": 0.8134,
+      "step": 2638
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011506393982566038,
+      "loss": 0.8608,
+      "step": 2639
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001150070972356071,
+      "loss": 0.7953,
+      "step": 2640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011495024968448731,
+      "loss": 0.8332,
+      "step": 2641
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011489339719109378,
+      "loss": 0.7903,
+      "step": 2642
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011483653977422087,
+      "loss": 0.8295,
+      "step": 2643
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011477967745266453,
+      "loss": 0.8514,
+      "step": 2644
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001147228102452225,
+      "loss": 0.8739,
+      "step": 2645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011466593817069391,
+      "loss": 0.8377,
+      "step": 2646
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011460906124787965,
+      "loss": 0.8096,
+      "step": 2647
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011455217949558217,
+      "loss": 0.8228,
+      "step": 2648
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011449529293260548,
+      "loss": 0.8239,
+      "step": 2649
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011443840157775527,
+      "loss": 0.8544,
+      "step": 2650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011438150544983876,
+      "loss": 0.8098,
+      "step": 2651
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011432460456766471,
+      "loss": 0.8213,
+      "step": 2652
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011426769895004352,
+      "loss": 0.8177,
+      "step": 2653
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011421078861578709,
+      "loss": 0.8302,
+      "step": 2654
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011415387358370897,
+      "loss": 0.8349,
+      "step": 2655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011409695387262416,
+      "loss": 0.8109,
+      "step": 2656
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011404002950134929,
+      "loss": 0.811,
+      "step": 2657
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011398310048870247,
+      "loss": 0.8112,
+      "step": 2658
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011392616685350336,
+      "loss": 0.7711,
+      "step": 2659
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011386922861457319,
+      "loss": 0.8157,
+      "step": 2660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011381228579073467,
+      "loss": 0.8632,
+      "step": 2661
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011375533840081202,
+      "loss": 0.8346,
+      "step": 2662
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000113698386463631,
+      "loss": 0.8535,
+      "step": 2663
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011364142999801887,
+      "loss": 0.8703,
+      "step": 2664
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011358446902280435,
+      "loss": 0.3316,
+      "step": 2665
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011352750355681772,
+      "loss": 0.781,
+      "step": 2666
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011347053361889063,
+      "loss": 0.8279,
+      "step": 2667
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011341355922785634,
+      "loss": 0.9281,
+      "step": 2668
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011335658040254951,
+      "loss": 0.779,
+      "step": 2669
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011329959716180622,
+      "loss": 0.7764,
+      "step": 2670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011324260952446417,
+      "loss": 0.8349,
+      "step": 2671
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011318561750936232,
+      "loss": 0.8698,
+      "step": 2672
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011312862113534122,
+      "loss": 0.8175,
+      "step": 2673
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011307162042124277,
+      "loss": 0.8268,
+      "step": 2674
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011301461538591039,
+      "loss": 0.86,
+      "step": 2675
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011295760604818882,
+      "loss": 0.8394,
+      "step": 2676
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011290059242692432,
+      "loss": 0.8693,
+      "step": 2677
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011284357454096457,
+      "loss": 0.8091,
+      "step": 2678
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011278655240915854,
+      "loss": 0.8861,
+      "step": 2679
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011272952605035674,
+      "loss": 0.8684,
+      "step": 2680
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011267249548341097,
+      "loss": 0.7263,
+      "step": 2681
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011261546072717454,
+      "loss": 0.8402,
+      "step": 2682
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011255842180050203,
+      "loss": 0.8605,
+      "step": 2683
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011250137872224946,
+      "loss": 0.8283,
+      "step": 2684
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011244433151127418,
+      "loss": 0.8341,
+      "step": 2685
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011238728018643499,
+      "loss": 0.8763,
+      "step": 2686
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011233022476659197,
+      "loss": 0.7387,
+      "step": 2687
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011227316527060651,
+      "loss": 0.8482,
+      "step": 2688
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011221610171734154,
+      "loss": 0.8291,
+      "step": 2689
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011215903412566111,
+      "loss": 0.3553,
+      "step": 2690
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011210196251443076,
+      "loss": 0.8618,
+      "step": 2691
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011204488690251725,
+      "loss": 0.831,
+      "step": 2692
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011198780730878875,
+      "loss": 0.7418,
+      "step": 2693
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011193072375211468,
+      "loss": 0.7969,
+      "step": 2694
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011187363625136585,
+      "loss": 0.8248,
+      "step": 2695
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011181654482541428,
+      "loss": 0.8017,
+      "step": 2696
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001117594494931333,
+      "loss": 0.821,
+      "step": 2697
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011170235027339766,
+      "loss": 0.8387,
+      "step": 2698
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001116452471850832,
+      "loss": 0.8346,
+      "step": 2699
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001115881402470672,
+      "loss": 0.7586,
+      "step": 2700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011153102947822813,
+      "loss": 0.8531,
+      "step": 2701
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001114739148974457,
+      "loss": 0.7696,
+      "step": 2702
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000111416796523601,
+      "loss": 0.8573,
+      "step": 2703
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011135967437557626,
+      "loss": 0.754,
+      "step": 2704
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011130254847225501,
+      "loss": 0.8138,
+      "step": 2705
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011124541883252198,
+      "loss": 0.836,
+      "step": 2706
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011118828547526319,
+      "loss": 0.8131,
+      "step": 2707
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011113114841936584,
+      "loss": 0.828,
+      "step": 2708
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011107400768371842,
+      "loss": 0.9061,
+      "step": 2709
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011101686328721053,
+      "loss": 0.3624,
+      "step": 2710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001109597152487331,
+      "loss": 0.8198,
+      "step": 2711
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011090256358717819,
+      "loss": 0.7845,
+      "step": 2712
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011084540832143904,
+      "loss": 0.8597,
+      "step": 2713
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011078824947041016,
+      "loss": 0.8017,
+      "step": 2714
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011073108705298715,
+      "loss": 0.7366,
+      "step": 2715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011067392108806692,
+      "loss": 0.7763,
+      "step": 2716
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011061675159454738,
+      "loss": 0.7573,
+      "step": 2717
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011055957859132773,
+      "loss": 0.7586,
+      "step": 2718
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011050240209730834,
+      "loss": 0.7909,
+      "step": 2719
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011044522213139064,
+      "loss": 0.3309,
+      "step": 2720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011038803871247729,
+      "loss": 0.8211,
+      "step": 2721
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011033085185947208,
+      "loss": 0.3623,
+      "step": 2722
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011027366159127985,
+      "loss": 0.8237,
+      "step": 2723
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011021646792680667,
+      "loss": 0.8647,
+      "step": 2724
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011015927088495974,
+      "loss": 0.8388,
+      "step": 2725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011010207048464729,
+      "loss": 0.8295,
+      "step": 2726
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011004486674477872,
+      "loss": 0.8216,
+      "step": 2727
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010998765968426449,
+      "loss": 0.803,
+      "step": 2728
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010993044932201624,
+      "loss": 0.7649,
+      "step": 2729
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010987323567694661,
+      "loss": 0.8409,
+      "step": 2730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010981601876796939,
+      "loss": 0.7822,
+      "step": 2731
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010975879861399938,
+      "loss": 0.8008,
+      "step": 2732
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010970157523395257,
+      "loss": 0.8141,
+      "step": 2733
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010964434864674584,
+      "loss": 0.8069,
+      "step": 2734
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010958711887129733,
+      "loss": 0.8054,
+      "step": 2735
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010952988592652611,
+      "loss": 0.7864,
+      "step": 2736
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001094726498313523,
+      "loss": 0.7501,
+      "step": 2737
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010941541060469712,
+      "loss": 0.8195,
+      "step": 2738
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010935816826548273,
+      "loss": 0.8923,
+      "step": 2739
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010930092283263243,
+      "loss": 0.8032,
+      "step": 2740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010924367432507048,
+      "loss": 0.826,
+      "step": 2741
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010918642276172218,
+      "loss": 0.7949,
+      "step": 2742
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010912916816151381,
+      "loss": 0.8589,
+      "step": 2743
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010907191054337269,
+      "loss": 0.7948,
+      "step": 2744
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010901464992622709,
+      "loss": 0.8689,
+      "step": 2745
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010895738632900636,
+      "loss": 0.7695,
+      "step": 2746
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010890011977064072,
+      "loss": 0.7952,
+      "step": 2747
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010884285027006147,
+      "loss": 0.8478,
+      "step": 2748
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001087855778462008,
+      "loss": 0.3431,
+      "step": 2749
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001087283025179919,
+      "loss": 0.8176,
+      "step": 2750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000108671024304369,
+      "loss": 0.7986,
+      "step": 2751
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010861374322426714,
+      "loss": 0.831,
+      "step": 2752
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010855645929662239,
+      "loss": 0.8756,
+      "step": 2753
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010849917254037174,
+      "loss": 0.8121,
+      "step": 2754
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010844188297445314,
+      "loss": 0.819,
+      "step": 2755
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010838459061780546,
+      "loss": 0.8261,
+      "step": 2756
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010832729548936848,
+      "loss": 0.8498,
+      "step": 2757
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001082699976080829,
+      "loss": 0.8114,
+      "step": 2758
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010821269699289031,
+      "loss": 0.8436,
+      "step": 2759
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010815539366273327,
+      "loss": 0.8363,
+      "step": 2760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010809808763655511,
+      "loss": 0.8107,
+      "step": 2761
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010804077893330023,
+      "loss": 0.7876,
+      "step": 2762
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010798346757191375,
+      "loss": 0.8112,
+      "step": 2763
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001079261535713418,
+      "loss": 0.79,
+      "step": 2764
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010786883695053125,
+      "loss": 0.829,
+      "step": 2765
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010781151772842993,
+      "loss": 0.7997,
+      "step": 2766
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010775419592398655,
+      "loss": 0.7752,
+      "step": 2767
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010769687155615055,
+      "loss": 0.831,
+      "step": 2768
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010763954464387238,
+      "loss": 0.7952,
+      "step": 2769
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010758221520610321,
+      "loss": 0.8296,
+      "step": 2770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010752488326179508,
+      "loss": 0.7932,
+      "step": 2771
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010746754882990082,
+      "loss": 0.8719,
+      "step": 2772
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010741021192937421,
+      "loss": 0.8282,
+      "step": 2773
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010735287257916972,
+      "loss": 0.8368,
+      "step": 2774
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010729553079824266,
+      "loss": 0.8639,
+      "step": 2775
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010723818660554913,
+      "loss": 0.8033,
+      "step": 2776
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010718084002004613,
+      "loss": 0.8313,
+      "step": 2777
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010712349106069131,
+      "loss": 0.8895,
+      "step": 2778
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010706613974644317,
+      "loss": 0.8642,
+      "step": 2779
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010700878609626102,
+      "loss": 0.8286,
+      "step": 2780
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010695143012910487,
+      "loss": 0.8367,
+      "step": 2781
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010689407186393552,
+      "loss": 0.7968,
+      "step": 2782
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001068367113197146,
+      "loss": 0.838,
+      "step": 2783
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001067793485154044,
+      "loss": 0.8501,
+      "step": 2784
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010672198346996802,
+      "loss": 0.8477,
+      "step": 2785
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010666461620236922,
+      "loss": 0.8643,
+      "step": 2786
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010660724673157261,
+      "loss": 0.8184,
+      "step": 2787
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010654987507654341,
+      "loss": 0.8342,
+      "step": 2788
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010649250125624767,
+      "loss": 0.8177,
+      "step": 2789
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010643512528965207,
+      "loss": 0.8474,
+      "step": 2790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010637774719572406,
+      "loss": 0.7921,
+      "step": 2791
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010632036699343178,
+      "loss": 0.7624,
+      "step": 2792
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010626298470174398,
+      "loss": 0.829,
+      "step": 2793
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010620560033963025,
+      "loss": 0.8123,
+      "step": 2794
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010614821392606079,
+      "loss": 0.7951,
+      "step": 2795
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010609082548000642,
+      "loss": 0.836,
+      "step": 2796
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010603343502043875,
+      "loss": 0.8893,
+      "step": 2797
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010597604256632994,
+      "loss": 0.8774,
+      "step": 2798
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010591864813665293,
+      "loss": 0.8264,
+      "step": 2799
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001058612517503812,
+      "loss": 0.3308,
+      "step": 2800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001058038534264889,
+      "loss": 0.7464,
+      "step": 2801
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010574645318395095,
+      "loss": 0.8306,
+      "step": 2802
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010568905104174272,
+      "loss": 0.8594,
+      "step": 2803
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010563164701884027,
+      "loss": 0.8492,
+      "step": 2804
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010557424113422037,
+      "loss": 0.8311,
+      "step": 2805
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010551683340686027,
+      "loss": 0.7568,
+      "step": 2806
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010545942385573795,
+      "loss": 0.8013,
+      "step": 2807
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010540201249983188,
+      "loss": 0.8234,
+      "step": 2808
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001053445993581212,
+      "loss": 0.768,
+      "step": 2809
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010528718444958567,
+      "loss": 0.7889,
+      "step": 2810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010522976779320555,
+      "loss": 0.7658,
+      "step": 2811
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010517234940796173,
+      "loss": 0.8897,
+      "step": 2812
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010511492931283563,
+      "loss": 0.8129,
+      "step": 2813
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010505750752680926,
+      "loss": 0.9116,
+      "step": 2814
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010500008406886524,
+      "loss": 0.8727,
+      "step": 2815
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010494265895798665,
+      "loss": 0.8409,
+      "step": 2816
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010488523221315719,
+      "loss": 0.7959,
+      "step": 2817
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010482780385336106,
+      "loss": 0.8113,
+      "step": 2818
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000104770373897583,
+      "loss": 0.8421,
+      "step": 2819
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010471294236480827,
+      "loss": 0.8307,
+      "step": 2820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010465550927402267,
+      "loss": 0.8316,
+      "step": 2821
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010459807464421257,
+      "loss": 0.8032,
+      "step": 2822
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010454063849436468,
+      "loss": 0.8396,
+      "step": 2823
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001044832008434664,
+      "loss": 0.4043,
+      "step": 2824
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001044257617105055,
+      "loss": 0.7913,
+      "step": 2825
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010436832111447034,
+      "loss": 0.8153,
+      "step": 2826
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010431087907434966,
+      "loss": 0.8074,
+      "step": 2827
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010425343560913277,
+      "loss": 0.818,
+      "step": 2828
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010419599073780937,
+      "loss": 0.7965,
+      "step": 2829
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010413854447936966,
+      "loss": 0.8084,
+      "step": 2830
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010408109685280434,
+      "loss": 0.8462,
+      "step": 2831
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010402364787710451,
+      "loss": 0.816,
+      "step": 2832
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010396619757126169,
+      "loss": 0.8405,
+      "step": 2833
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010390874595426794,
+      "loss": 0.8075,
+      "step": 2834
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010385129304511567,
+      "loss": 0.8512,
+      "step": 2835
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001037938388627977,
+      "loss": 0.799,
+      "step": 2836
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010373638342630738,
+      "loss": 0.8448,
+      "step": 2837
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010367892675463837,
+      "loss": 0.8067,
+      "step": 2838
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010362146886678477,
+      "loss": 0.3772,
+      "step": 2839
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001035640097817411,
+      "loss": 0.8451,
+      "step": 2840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010350654951850224,
+      "loss": 0.7818,
+      "step": 2841
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010344908809606353,
+      "loss": 0.8165,
+      "step": 2842
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001033916255334206,
+      "loss": 0.3467,
+      "step": 2843
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001033341618495695,
+      "loss": 0.8563,
+      "step": 2844
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010327669706350667,
+      "loss": 0.8394,
+      "step": 2845
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001032192311942289,
+      "loss": 0.363,
+      "step": 2846
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001031617642607333,
+      "loss": 0.8046,
+      "step": 2847
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010310429628201743,
+      "loss": 0.8478,
+      "step": 2848
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010304682727707908,
+      "loss": 0.8289,
+      "step": 2849
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010298935726491648,
+      "loss": 0.8236,
+      "step": 2850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010293188626452808,
+      "loss": 0.8119,
+      "step": 2851
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010287441429491274,
+      "loss": 0.7652,
+      "step": 2852
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010281694137506965,
+      "loss": 0.8463,
+      "step": 2853
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001027594675239983,
+      "loss": 0.8103,
+      "step": 2854
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010270199276069843,
+      "loss": 0.8158,
+      "step": 2855
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010264451710417011,
+      "loss": 0.7769,
+      "step": 2856
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010258704057341377,
+      "loss": 0.7821,
+      "step": 2857
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010252956318743006,
+      "loss": 0.8549,
+      "step": 2858
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010247208496521992,
+      "loss": 0.772,
+      "step": 2859
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001024146059257846,
+      "loss": 0.8588,
+      "step": 2860
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001023571260881256,
+      "loss": 0.8406,
+      "step": 2861
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010229964547124464,
+      "loss": 0.833,
+      "step": 2862
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010224216409414378,
+      "loss": 0.8149,
+      "step": 2863
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001021846819758253,
+      "loss": 0.7978,
+      "step": 2864
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010212719913529168,
+      "loss": 0.833,
+      "step": 2865
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001020697155915457,
+      "loss": 0.8521,
+      "step": 2866
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010201223136359033,
+      "loss": 0.8001,
+      "step": 2867
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001019547464704288,
+      "loss": 0.8288,
+      "step": 2868
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001018972609310645,
+      "loss": 0.7914,
+      "step": 2869
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010183977476450117,
+      "loss": 0.8388,
+      "step": 2870
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001017822879897426,
+      "loss": 0.7589,
+      "step": 2871
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010172480062579287,
+      "loss": 0.8023,
+      "step": 2872
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010166731269165617,
+      "loss": 0.8377,
+      "step": 2873
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000101609824206337,
+      "loss": 0.8439,
+      "step": 2874
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010155233518884,
+      "loss": 0.7965,
+      "step": 2875
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010149484565816992,
+      "loss": 0.8055,
+      "step": 2876
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010143735563333171,
+      "loss": 0.874,
+      "step": 2877
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010137986513333055,
+      "loss": 0.8073,
+      "step": 2878
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010132237417717171,
+      "loss": 0.7913,
+      "step": 2879
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010126488278386063,
+      "loss": 0.7106,
+      "step": 2880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001012073909724029,
+      "loss": 0.7735,
+      "step": 2881
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010114989876180423,
+      "loss": 0.8095,
+      "step": 2882
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010109240617107045,
+      "loss": 0.7923,
+      "step": 2883
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010103491321920757,
+      "loss": 0.7991,
+      "step": 2884
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010097741992522168,
+      "loss": 0.7962,
+      "step": 2885
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000100919926308119,
+      "loss": 0.91,
+      "step": 2886
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010086243238690583,
+      "loss": 0.3517,
+      "step": 2887
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010080493818058859,
+      "loss": 0.814,
+      "step": 2888
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010074744370817378,
+      "loss": 0.8379,
+      "step": 2889
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010068994898866804,
+      "loss": 0.8261,
+      "step": 2890
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010063245404107799,
+      "loss": 0.8941,
+      "step": 2891
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010057495888441046,
+      "loss": 0.8174,
+      "step": 2892
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010051746353767226,
+      "loss": 0.3445,
+      "step": 2893
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010045996801987023,
+      "loss": 0.7897,
+      "step": 2894
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010040247235001139,
+      "loss": 0.8037,
+      "step": 2895
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010034497654710266,
+      "loss": 0.893,
+      "step": 2896
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010028748063015114,
+      "loss": 0.7634,
+      "step": 2897
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010022998461816389,
+      "loss": 0.8372,
+      "step": 2898
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000100172488530148,
+      "loss": 0.7967,
+      "step": 2899
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010011499238511062,
+      "loss": 0.7075,
+      "step": 2900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010005749620205889,
+      "loss": 0.825,
+      "step": 2901
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001,
+      "loss": 0.3674,
+      "step": 2902
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.994250379794112e-05,
+      "loss": 0.3513,
+      "step": 2903
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.988500761488941e-05,
+      "loss": 0.7914,
+      "step": 2904
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.982751146985203e-05,
+      "loss": 0.8436,
+      "step": 2905
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.977001538183616e-05,
+      "loss": 0.8405,
+      "step": 2906
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97125193698489e-05,
+      "loss": 0.8183,
+      "step": 2907
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.965502345289733e-05,
+      "loss": 0.8081,
+      "step": 2908
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.959752764998862e-05,
+      "loss": 0.7327,
+      "step": 2909
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.954003198012977e-05,
+      "loss": 0.8175,
+      "step": 2910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.948253646232776e-05,
+      "loss": 0.846,
+      "step": 2911
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.942504111558956e-05,
+      "loss": 0.794,
+      "step": 2912
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.936754595892203e-05,
+      "loss": 0.8816,
+      "step": 2913
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.9310051011332e-05,
+      "loss": 0.7839,
+      "step": 2914
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.925255629182626e-05,
+      "loss": 0.8113,
+      "step": 2915
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.919506181941146e-05,
+      "loss": 0.7935,
+      "step": 2916
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.913756761309423e-05,
+      "loss": 0.8271,
+      "step": 2917
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.908007369188105e-05,
+      "loss": 0.7927,
+      "step": 2918
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.902258007477832e-05,
+      "loss": 0.8157,
+      "step": 2919
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.896508678079244e-05,
+      "loss": 0.8486,
+      "step": 2920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.890759382892956e-05,
+      "loss": 0.7948,
+      "step": 2921
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.88501012381958e-05,
+      "loss": 0.7637,
+      "step": 2922
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.879260902759711e-05,
+      "loss": 0.3202,
+      "step": 2923
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.873511721613938e-05,
+      "loss": 0.8394,
+      "step": 2924
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.86776258228283e-05,
+      "loss": 0.8055,
+      "step": 2925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.862013486666947e-05,
+      "loss": 0.8682,
+      "step": 2926
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.856264436666831e-05,
+      "loss": 0.8329,
+      "step": 2927
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.850515434183013e-05,
+      "loss": 0.8537,
+      "step": 2928
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.844766481116e-05,
+      "loss": 0.8715,
+      "step": 2929
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.839017579366299e-05,
+      "loss": 0.8152,
+      "step": 2930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.833268730834383e-05,
+      "loss": 0.7796,
+      "step": 2931
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.827519937420716e-05,
+      "loss": 0.7856,
+      "step": 2932
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.821771201025742e-05,
+      "loss": 0.8398,
+      "step": 2933
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.816022523549885e-05,
+      "loss": 0.8221,
+      "step": 2934
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.810273906893552e-05,
+      "loss": 0.8761,
+      "step": 2935
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.804525352957124e-05,
+      "loss": 0.8385,
+      "step": 2936
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.798776863640971e-05,
+      "loss": 0.83,
+      "step": 2937
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.793028440845434e-05,
+      "loss": 0.7708,
+      "step": 2938
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.787280086470837e-05,
+      "loss": 0.8233,
+      "step": 2939
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.781531802417473e-05,
+      "loss": 0.7743,
+      "step": 2940
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.775783590585622e-05,
+      "loss": 0.7822,
+      "step": 2941
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.770035452875537e-05,
+      "loss": 0.7979,
+      "step": 2942
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.764287391187442e-05,
+      "loss": 0.8106,
+      "step": 2943
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.758539407421542e-05,
+      "loss": 0.7951,
+      "step": 2944
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.752791503478009e-05,
+      "loss": 0.8155,
+      "step": 2945
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.747043681256996e-05,
+      "loss": 0.7646,
+      "step": 2946
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.741295942658625e-05,
+      "loss": 0.8498,
+      "step": 2947
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.735548289582992e-05,
+      "loss": 0.8076,
+      "step": 2948
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.729800723930162e-05,
+      "loss": 0.8118,
+      "step": 2949
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.724053247600175e-05,
+      "loss": 0.8021,
+      "step": 2950
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.718305862493033e-05,
+      "loss": 0.7799,
+      "step": 2951
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.712558570508726e-05,
+      "loss": 0.876,
+      "step": 2952
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.706811373547194e-05,
+      "loss": 0.8408,
+      "step": 2953
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.701064273508356e-05,
+      "loss": 0.795,
+      "step": 2954
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.695317272292093e-05,
+      "loss": 0.9054,
+      "step": 2955
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.68957037179826e-05,
+      "loss": 0.8658,
+      "step": 2956
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.683823573926673e-05,
+      "loss": 0.8806,
+      "step": 2957
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.678076880577114e-05,
+      "loss": 0.793,
+      "step": 2958
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.672330293649337e-05,
+      "loss": 0.8634,
+      "step": 2959
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.666583815043054e-05,
+      "loss": 0.7991,
+      "step": 2960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.660837446657943e-05,
+      "loss": 0.885,
+      "step": 2961
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.65509119039365e-05,
+      "loss": 0.3616,
+      "step": 2962
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.649345048149777e-05,
+      "loss": 0.874,
+      "step": 2963
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.643599021825892e-05,
+      "loss": 0.8544,
+      "step": 2964
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.637853113321524e-05,
+      "loss": 0.8471,
+      "step": 2965
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.632107324536165e-05,
+      "loss": 0.8449,
+      "step": 2966
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.626361657369265e-05,
+      "loss": 0.7906,
+      "step": 2967
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.620616113720232e-05,
+      "loss": 0.8647,
+      "step": 2968
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.614870695488438e-05,
+      "loss": 0.8419,
+      "step": 2969
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.609125404573211e-05,
+      "loss": 0.847,
+      "step": 2970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.603380242873834e-05,
+      "loss": 0.8045,
+      "step": 2971
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.59763521228955e-05,
+      "loss": 0.7966,
+      "step": 2972
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.591890314719567e-05,
+      "loss": 0.8202,
+      "step": 2973
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.586145552063035e-05,
+      "loss": 0.7549,
+      "step": 2974
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.580400926219066e-05,
+      "loss": 0.8373,
+      "step": 2975
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.574656439086725e-05,
+      "loss": 0.8194,
+      "step": 2976
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.568912092565035e-05,
+      "loss": 0.7876,
+      "step": 2977
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.563167888552968e-05,
+      "loss": 0.8327,
+      "step": 2978
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.557423828949452e-05,
+      "loss": 0.7921,
+      "step": 2979
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.551679915653362e-05,
+      "loss": 0.819,
+      "step": 2980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.545936150563534e-05,
+      "loss": 0.3554,
+      "step": 2981
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.540192535578748e-05,
+      "loss": 0.8371,
+      "step": 2982
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.534449072597734e-05,
+      "loss": 0.8833,
+      "step": 2983
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.528705763519176e-05,
+      "loss": 0.8464,
+      "step": 2984
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.522962610241703e-05,
+      "loss": 0.7722,
+      "step": 2985
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.517219614663896e-05,
+      "loss": 0.798,
+      "step": 2986
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.511476778684282e-05,
+      "loss": 0.814,
+      "step": 2987
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.505734104201336e-05,
+      "loss": 0.8008,
+      "step": 2988
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.49999159311348e-05,
+      "loss": 0.8337,
+      "step": 2989
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.494249247319077e-05,
+      "loss": 0.7989,
+      "step": 2990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.488507068716443e-05,
+      "loss": 0.8454,
+      "step": 2991
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.482765059203834e-05,
+      "loss": 0.8541,
+      "step": 2992
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.477023220679446e-05,
+      "loss": 0.8022,
+      "step": 2993
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.471281555041432e-05,
+      "loss": 0.7576,
+      "step": 2994
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.465540064187878e-05,
+      "loss": 0.9046,
+      "step": 2995
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.459798750016813e-05,
+      "loss": 0.7743,
+      "step": 2996
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.454057614426209e-05,
+      "loss": 0.7963,
+      "step": 2997
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.448316659313975e-05,
+      "loss": 0.734,
+      "step": 2998
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.442575886577967e-05,
+      "loss": 0.8665,
+      "step": 2999
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.436835298115975e-05,
+      "loss": 0.8771,
+      "step": 3000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.431094895825733e-05,
+      "loss": 0.7868,
+      "step": 3001
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.425354681604907e-05,
+      "loss": 0.7985,
+      "step": 3002
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41961465735111e-05,
+      "loss": 0.822,
+      "step": 3003
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.413874824961883e-05,
+      "loss": 0.7789,
+      "step": 3004
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.40813518633471e-05,
+      "loss": 0.8038,
+      "step": 3005
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.402395743367008e-05,
+      "loss": 0.8478,
+      "step": 3006
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.396656497956128e-05,
+      "loss": 0.7487,
+      "step": 3007
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.390917451999359e-05,
+      "loss": 0.805,
+      "step": 3008
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.385178607393924e-05,
+      "loss": 0.8074,
+      "step": 3009
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.379439966036977e-05,
+      "loss": 0.3434,
+      "step": 3010
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.373701529825604e-05,
+      "loss": 0.8223,
+      "step": 3011
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.367963300656827e-05,
+      "loss": 0.808,
+      "step": 3012
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.362225280427597e-05,
+      "loss": 0.7818,
+      "step": 3013
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.356487471034796e-05,
+      "loss": 0.8691,
+      "step": 3014
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.350749874375234e-05,
+      "loss": 0.8818,
+      "step": 3015
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34501249234566e-05,
+      "loss": 0.8403,
+      "step": 3016
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.339275326842741e-05,
+      "loss": 0.8326,
+      "step": 3017
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.333538379763079e-05,
+      "loss": 0.8369,
+      "step": 3018
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.3278016530032e-05,
+      "loss": 0.8307,
+      "step": 3019
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.32206514845956e-05,
+      "loss": 0.3582,
+      "step": 3020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.316328868028542e-05,
+      "loss": 0.858,
+      "step": 3021
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.310592813606449e-05,
+      "loss": 0.8027,
+      "step": 3022
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.304856987089518e-05,
+      "loss": 0.8296,
+      "step": 3023
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.2991213903739e-05,
+      "loss": 0.8552,
+      "step": 3024
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.293386025355682e-05,
+      "loss": 0.8553,
+      "step": 3025
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.28765089393087e-05,
+      "loss": 0.8788,
+      "step": 3026
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.281915997995388e-05,
+      "loss": 0.8529,
+      "step": 3027
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.276181339445088e-05,
+      "loss": 0.7378,
+      "step": 3028
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.270446920175736e-05,
+      "loss": 0.7868,
+      "step": 3029
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.26471274208303e-05,
+      "loss": 0.7982,
+      "step": 3030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.258978807062581e-05,
+      "loss": 0.8143,
+      "step": 3031
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.253245117009919e-05,
+      "loss": 0.807,
+      "step": 3032
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.247511673820497e-05,
+      "loss": 0.8186,
+      "step": 3033
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.241778479389683e-05,
+      "loss": 0.8269,
+      "step": 3034
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.236045535612766e-05,
+      "loss": 0.8023,
+      "step": 3035
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.230312844384943e-05,
+      "loss": 0.7658,
+      "step": 3036
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.224580407601346e-05,
+      "loss": 0.7884,
+      "step": 3037
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.218848227157007e-05,
+      "loss": 0.8385,
+      "step": 3038
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.213116304946877e-05,
+      "loss": 0.8451,
+      "step": 3039
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.207384642865824e-05,
+      "loss": 0.7825,
+      "step": 3040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.201653242808626e-05,
+      "loss": 0.8381,
+      "step": 3041
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.195922106669981e-05,
+      "loss": 0.8359,
+      "step": 3042
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.190191236344493e-05,
+      "loss": 0.8359,
+      "step": 3043
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.18446063372668e-05,
+      "loss": 0.7919,
+      "step": 3044
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.178730300710973e-05,
+      "loss": 0.3774,
+      "step": 3045
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.173000239191713e-05,
+      "loss": 0.8012,
+      "step": 3046
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.167270451063153e-05,
+      "loss": 0.8407,
+      "step": 3047
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.161540938219454e-05,
+      "loss": 0.7548,
+      "step": 3048
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.155811702554687e-05,
+      "loss": 0.855,
+      "step": 3049
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.150082745962828e-05,
+      "loss": 0.7829,
+      "step": 3050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.144354070337763e-05,
+      "loss": 0.8425,
+      "step": 3051
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.138625677573289e-05,
+      "loss": 0.3455,
+      "step": 3052
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.132897569563102e-05,
+      "loss": 0.8167,
+      "step": 3053
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.127169748200812e-05,
+      "loss": 0.7917,
+      "step": 3054
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.121442215379923e-05,
+      "loss": 0.7736,
+      "step": 3055
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.115714972993858e-05,
+      "loss": 0.8116,
+      "step": 3056
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.109988022935928e-05,
+      "loss": 0.8656,
+      "step": 3057
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.104261367099365e-05,
+      "loss": 0.8577,
+      "step": 3058
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.098535007377292e-05,
+      "loss": 0.8473,
+      "step": 3059
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.092808945662733e-05,
+      "loss": 0.8372,
+      "step": 3060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.087083183848621e-05,
+      "loss": 0.7527,
+      "step": 3061
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.081357723827785e-05,
+      "loss": 0.7684,
+      "step": 3062
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.075632567492954e-05,
+      "loss": 0.8458,
+      "step": 3063
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.069907716736761e-05,
+      "loss": 0.8197,
+      "step": 3064
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.064183173451731e-05,
+      "loss": 0.7768,
+      "step": 3065
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.058458939530295e-05,
+      "loss": 0.8098,
+      "step": 3066
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.052735016864774e-05,
+      "loss": 0.8762,
+      "step": 3067
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.047011407347389e-05,
+      "loss": 0.8156,
+      "step": 3068
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.041288112870265e-05,
+      "loss": 0.8348,
+      "step": 3069
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.035565135325414e-05,
+      "loss": 0.8643,
+      "step": 3070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.029842476604746e-05,
+      "loss": 0.832,
+      "step": 3071
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.024120138600063e-05,
+      "loss": 0.7948,
+      "step": 3072
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.018398123203063e-05,
+      "loss": 0.8271,
+      "step": 3073
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.01267643230534e-05,
+      "loss": 0.7411,
+      "step": 3074
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.00695506779838e-05,
+      "loss": 0.7639,
+      "step": 3075
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.001234031573553e-05,
+      "loss": 0.8442,
+      "step": 3076
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.995513325522134e-05,
+      "loss": 0.3611,
+      "step": 3077
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.989792951535276e-05,
+      "loss": 0.8163,
+      "step": 3078
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.984072911504027e-05,
+      "loss": 0.8172,
+      "step": 3079
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.978353207319332e-05,
+      "loss": 0.8587,
+      "step": 3080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.972633840872017e-05,
+      "loss": 0.808,
+      "step": 3081
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.966914814052796e-05,
+      "loss": 0.7887,
+      "step": 3082
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.961196128752272e-05,
+      "loss": 0.8453,
+      "step": 3083
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.955477786860937e-05,
+      "loss": 0.8076,
+      "step": 3084
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.949759790269169e-05,
+      "loss": 0.8086,
+      "step": 3085
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.944042140867229e-05,
+      "loss": 0.804,
+      "step": 3086
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.938324840545267e-05,
+      "loss": 0.7461,
+      "step": 3087
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.932607891193315e-05,
+      "loss": 0.8736,
+      "step": 3088
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.926891294701284e-05,
+      "loss": 0.8526,
+      "step": 3089
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.921175052958985e-05,
+      "loss": 0.8097,
+      "step": 3090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.915459167856097e-05,
+      "loss": 0.7459,
+      "step": 3091
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.909743641282183e-05,
+      "loss": 0.8815,
+      "step": 3092
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.90402847512669e-05,
+      "loss": 0.8026,
+      "step": 3093
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.898313671278948e-05,
+      "loss": 0.8098,
+      "step": 3094
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89259923162816e-05,
+      "loss": 0.7891,
+      "step": 3095
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.886885158063416e-05,
+      "loss": 0.7958,
+      "step": 3096
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.881171452473685e-05,
+      "loss": 0.8843,
+      "step": 3097
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.875458116747806e-05,
+      "loss": 0.8612,
+      "step": 3098
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.869745152774505e-05,
+      "loss": 0.7632,
+      "step": 3099
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.864032562442374e-05,
+      "loss": 0.9154,
+      "step": 3100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.858320347639901e-05,
+      "loss": 0.8296,
+      "step": 3101
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.852608510255429e-05,
+      "loss": 0.7947,
+      "step": 3102
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.84689705217719e-05,
+      "loss": 0.831,
+      "step": 3103
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.841185975293282e-05,
+      "loss": 0.8664,
+      "step": 3104
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.835475281491681e-05,
+      "loss": 0.8102,
+      "step": 3105
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.829764972660237e-05,
+      "loss": 0.7712,
+      "step": 3106
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.82405505068667e-05,
+      "loss": 0.8291,
+      "step": 3107
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.818345517458576e-05,
+      "loss": 0.7717,
+      "step": 3108
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.812636374863419e-05,
+      "loss": 0.3595,
+      "step": 3109
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.806927624788535e-05,
+      "loss": 0.8246,
+      "step": 3110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.801219269121126e-05,
+      "loss": 0.7683,
+      "step": 3111
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.795511309748276e-05,
+      "loss": 0.8017,
+      "step": 3112
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.789803748556926e-05,
+      "loss": 0.8304,
+      "step": 3113
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.78409658743389e-05,
+      "loss": 0.8358,
+      "step": 3114
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.77838982826585e-05,
+      "loss": 0.7827,
+      "step": 3115
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.772683472939351e-05,
+      "loss": 0.8362,
+      "step": 3116
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.766977523340807e-05,
+      "loss": 0.8209,
+      "step": 3117
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.761271981356504e-05,
+      "loss": 0.8503,
+      "step": 3118
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.755566848872584e-05,
+      "loss": 0.7936,
+      "step": 3119
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.749862127775058e-05,
+      "loss": 0.8167,
+      "step": 3120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.744157819949798e-05,
+      "loss": 0.8371,
+      "step": 3121
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.738453927282548e-05,
+      "loss": 0.7756,
+      "step": 3122
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.732750451658901e-05,
+      "loss": 0.8001,
+      "step": 3123
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.727047394964328e-05,
+      "loss": 0.8401,
+      "step": 3124
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.721344759084149e-05,
+      "loss": 0.7797,
+      "step": 3125
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.715642545903546e-05,
+      "loss": 0.8093,
+      "step": 3126
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.709940757307569e-05,
+      "loss": 0.7517,
+      "step": 3127
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.704239395181121e-05,
+      "loss": 0.7954,
+      "step": 3128
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.698538461408966e-05,
+      "loss": 0.3699,
+      "step": 3129
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.692837957875725e-05,
+      "loss": 0.8784,
+      "step": 3130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.687137886465883e-05,
+      "loss": 0.7895,
+      "step": 3131
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.681438249063767e-05,
+      "loss": 0.7665,
+      "step": 3132
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.675739047553584e-05,
+      "loss": 0.3755,
+      "step": 3133
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.670040283819376e-05,
+      "loss": 0.8335,
+      "step": 3134
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.664341959745051e-05,
+      "loss": 0.8293,
+      "step": 3135
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.658644077214368e-05,
+      "loss": 0.8072,
+      "step": 3136
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.652946638110938e-05,
+      "loss": 0.7548,
+      "step": 3137
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.647249644318232e-05,
+      "loss": 0.8437,
+      "step": 3138
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.641553097719568e-05,
+      "loss": 0.8739,
+      "step": 3139
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.635857000198114e-05,
+      "loss": 0.7732,
+      "step": 3140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.630161353636901e-05,
+      "loss": 0.8459,
+      "step": 3141
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6244661599188e-05,
+      "loss": 0.7895,
+      "step": 3142
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.618771420926534e-05,
+      "loss": 0.8264,
+      "step": 3143
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.613077138542684e-05,
+      "loss": 0.7985,
+      "step": 3144
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.607383314649666e-05,
+      "loss": 0.8254,
+      "step": 3145
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.601689951129757e-05,
+      "loss": 0.8444,
+      "step": 3146
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.595997049865074e-05,
+      "loss": 0.8029,
+      "step": 3147
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.590304612737587e-05,
+      "loss": 0.8427,
+      "step": 3148
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.584612641629107e-05,
+      "loss": 0.821,
+      "step": 3149
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.578921138421294e-05,
+      "loss": 0.7963,
+      "step": 3150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.573230104995651e-05,
+      "loss": 0.8616,
+      "step": 3151
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.567539543233532e-05,
+      "loss": 0.7934,
+      "step": 3152
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.561849455016125e-05,
+      "loss": 0.8526,
+      "step": 3153
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.556159842224472e-05,
+      "loss": 0.3325,
+      "step": 3154
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.55047070673945e-05,
+      "loss": 0.3638,
+      "step": 3155
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.544782050441785e-05,
+      "loss": 0.831,
+      "step": 3156
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.539093875212037e-05,
+      "loss": 0.3385,
+      "step": 3157
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.533406182930613e-05,
+      "loss": 0.8606,
+      "step": 3158
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.527718975477753e-05,
+      "loss": 0.8116,
+      "step": 3159
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.522032254733548e-05,
+      "loss": 0.7567,
+      "step": 3160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.516346022577918e-05,
+      "loss": 0.8797,
+      "step": 3161
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.510660280890625e-05,
+      "loss": 0.8382,
+      "step": 3162
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.504975031551272e-05,
+      "loss": 0.7675,
+      "step": 3163
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.499290276439293e-05,
+      "loss": 0.8087,
+      "step": 3164
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.493606017433966e-05,
+      "loss": 0.78,
+      "step": 3165
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.4879222564144e-05,
+      "loss": 0.8124,
+      "step": 3166
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.482238995259536e-05,
+      "loss": 0.8564,
+      "step": 3167
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.47655623584816e-05,
+      "loss": 0.8175,
+      "step": 3168
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.470873980058885e-05,
+      "loss": 0.8396,
+      "step": 3169
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.465192229770156e-05,
+      "loss": 0.8431,
+      "step": 3170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.459510986860258e-05,
+      "loss": 0.3625,
+      "step": 3171
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.4538302532073e-05,
+      "loss": 0.8137,
+      "step": 3172
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.448150030689228e-05,
+      "loss": 0.829,
+      "step": 3173
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.442470321183817e-05,
+      "loss": 0.8421,
+      "step": 3174
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.436791126568669e-05,
+      "loss": 0.8221,
+      "step": 3175
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.43111244872123e-05,
+      "loss": 0.8673,
+      "step": 3176
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.42543428951876e-05,
+      "loss": 0.7975,
+      "step": 3177
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.41975665083835e-05,
+      "loss": 0.8121,
+      "step": 3178
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.414079534556926e-05,
+      "loss": 0.8114,
+      "step": 3179
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.408402942551234e-05,
+      "loss": 0.8266,
+      "step": 3180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.402726876697848e-05,
+      "loss": 0.8292,
+      "step": 3181
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.397051338873172e-05,
+      "loss": 0.8406,
+      "step": 3182
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.391376330953433e-05,
+      "loss": 0.8394,
+      "step": 3183
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.38570185481468e-05,
+      "loss": 0.8572,
+      "step": 3184
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.38002791233279e-05,
+      "loss": 0.778,
+      "step": 3185
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.374354505383467e-05,
+      "loss": 0.8377,
+      "step": 3186
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.36868163584223e-05,
+      "loss": 0.8207,
+      "step": 3187
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.363009305584424e-05,
+      "loss": 0.8085,
+      "step": 3188
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.357337516485213e-05,
+      "loss": 0.851,
+      "step": 3189
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.351666270419589e-05,
+      "loss": 0.7909,
+      "step": 3190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.345995569262359e-05,
+      "loss": 0.8041,
+      "step": 3191
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.340325414888152e-05,
+      "loss": 0.8348,
+      "step": 3192
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.334655809171413e-05,
+      "loss": 0.7784,
+      "step": 3193
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.328986753986409e-05,
+      "loss": 0.7796,
+      "step": 3194
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.323318251207224e-05,
+      "loss": 0.8443,
+      "step": 3195
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.317650302707754e-05,
+      "loss": 0.7727,
+      "step": 3196
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.311982910361728e-05,
+      "loss": 0.7535,
+      "step": 3197
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.306316076042673e-05,
+      "loss": 0.8201,
+      "step": 3198
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.300649801623941e-05,
+      "loss": 0.8276,
+      "step": 3199
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.294984088978694e-05,
+      "loss": 0.8191,
+      "step": 3200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.289318939979916e-05,
+      "loss": 0.7768,
+      "step": 3201
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.283654356500394e-05,
+      "loss": 0.7795,
+      "step": 3202
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.277990340412737e-05,
+      "loss": 0.7945,
+      "step": 3203
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.272326893589362e-05,
+      "loss": 0.8121,
+      "step": 3204
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.2666640179025e-05,
+      "loss": 0.8724,
+      "step": 3205
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.261001715224188e-05,
+      "loss": 0.8563,
+      "step": 3206
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.25533998742628e-05,
+      "loss": 0.8188,
+      "step": 3207
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.249678836380442e-05,
+      "loss": 0.7746,
+      "step": 3208
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.244018263958141e-05,
+      "loss": 0.8418,
+      "step": 3209
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.238358272030658e-05,
+      "loss": 0.8711,
+      "step": 3210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.232698862469078e-05,
+      "loss": 0.8141,
+      "step": 3211
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.227040037144297e-05,
+      "loss": 0.8731,
+      "step": 3212
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.221381797927022e-05,
+      "loss": 0.8321,
+      "step": 3213
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.215724146687756e-05,
+      "loss": 0.7971,
+      "step": 3214
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.210067085296815e-05,
+      "loss": 0.795,
+      "step": 3215
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.204410615624318e-05,
+      "loss": 0.8061,
+      "step": 3216
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.198754739540185e-05,
+      "loss": 0.8349,
+      "step": 3217
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.193099458914148e-05,
+      "loss": 0.8307,
+      "step": 3218
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.187444775615738e-05,
+      "loss": 0.8267,
+      "step": 3219
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.181790691514284e-05,
+      "loss": 0.7876,
+      "step": 3220
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.176137208478923e-05,
+      "loss": 0.8027,
+      "step": 3221
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.17048432837859e-05,
+      "loss": 0.8396,
+      "step": 3222
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.164832053082022e-05,
+      "loss": 0.8071,
+      "step": 3223
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.159180384457757e-05,
+      "loss": 0.8031,
+      "step": 3224
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.153529324374127e-05,
+      "loss": 0.8443,
+      "step": 3225
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.147878874699274e-05,
+      "loss": 0.7801,
+      "step": 3226
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.142229037301125e-05,
+      "loss": 0.7383,
+      "step": 3227
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.136579814047409e-05,
+      "loss": 0.8174,
+      "step": 3228
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.130931206805663e-05,
+      "loss": 0.8407,
+      "step": 3229
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.125283217443207e-05,
+      "loss": 0.8317,
+      "step": 3230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.11963584782716e-05,
+      "loss": 0.7912,
+      "step": 3231
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.113989099824438e-05,
+      "loss": 0.7604,
+      "step": 3232
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.108342975301748e-05,
+      "loss": 0.7919,
+      "step": 3233
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.102697476125597e-05,
+      "loss": 0.8742,
+      "step": 3234
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.097052604162281e-05,
+      "loss": 0.8787,
+      "step": 3235
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.091408361277888e-05,
+      "loss": 0.8464,
+      "step": 3236
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.085764749338301e-05,
+      "loss": 0.859,
+      "step": 3237
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.080121770209191e-05,
+      "loss": 0.7794,
+      "step": 3238
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.074479425756019e-05,
+      "loss": 0.7518,
+      "step": 3239
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.068837717844047e-05,
+      "loss": 0.8233,
+      "step": 3240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.063196648338313e-05,
+      "loss": 0.8341,
+      "step": 3241
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.057556219103653e-05,
+      "loss": 0.7707,
+      "step": 3242
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.051916432004682e-05,
+      "loss": 0.796,
+      "step": 3243
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.046277288905814e-05,
+      "loss": 0.8194,
+      "step": 3244
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.040638791671241e-05,
+      "loss": 0.8164,
+      "step": 3245
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.035000942164947e-05,
+      "loss": 0.8418,
+      "step": 3246
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.029363742250697e-05,
+      "loss": 0.734,
+      "step": 3247
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.023727193792048e-05,
+      "loss": 0.8191,
+      "step": 3248
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.018091298652332e-05,
+      "loss": 0.7975,
+      "step": 3249
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.012456058694678e-05,
+      "loss": 0.8036,
+      "step": 3250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.006821475781989e-05,
+      "loss": 0.8793,
+      "step": 3251
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.001187551776952e-05,
+      "loss": 0.8195,
+      "step": 3252
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.995554288542038e-05,
+      "loss": 0.8036,
+      "step": 3253
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.989921687939497e-05,
+      "loss": 0.8728,
+      "step": 3254
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.984289751831362e-05,
+      "loss": 0.8085,
+      "step": 3255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.978658482079449e-05,
+      "loss": 0.7849,
+      "step": 3256
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.973027880545348e-05,
+      "loss": 0.7698,
+      "step": 3257
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.967397949090431e-05,
+      "loss": 0.8795,
+      "step": 3258
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.96176868957585e-05,
+      "loss": 0.7232,
+      "step": 3259
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.956140103862527e-05,
+      "loss": 0.8245,
+      "step": 3260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.950512193811178e-05,
+      "loss": 0.8191,
+      "step": 3261
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.944884961282279e-05,
+      "loss": 0.7763,
+      "step": 3262
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93925840813609e-05,
+      "loss": 0.8482,
+      "step": 3263
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.933632536232642e-05,
+      "loss": 0.7796,
+      "step": 3264
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.928007347431746e-05,
+      "loss": 0.8189,
+      "step": 3265
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.922382843592984e-05,
+      "loss": 0.8027,
+      "step": 3266
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.91675902657571e-05,
+      "loss": 0.8108,
+      "step": 3267
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.911135898239055e-05,
+      "loss": 0.861,
+      "step": 3268
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.905513460441919e-05,
+      "loss": 0.7898,
+      "step": 3269
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.899891715042976e-05,
+      "loss": 0.8695,
+      "step": 3270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.894270663900667e-05,
+      "loss": 0.7967,
+      "step": 3271
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.888650308873213e-05,
+      "loss": 0.7651,
+      "step": 3272
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.883030651818596e-05,
+      "loss": 0.8272,
+      "step": 3273
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.87741169459457e-05,
+      "loss": 0.8057,
+      "step": 3274
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.871793439058655e-05,
+      "loss": 0.8809,
+      "step": 3275
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.866175887068143e-05,
+      "loss": 0.8433,
+      "step": 3276
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.86055904048009e-05,
+      "loss": 0.7962,
+      "step": 3277
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.854942901151328e-05,
+      "loss": 0.8115,
+      "step": 3278
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.84932747093844e-05,
+      "loss": 0.3206,
+      "step": 3279
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.843712751697786e-05,
+      "loss": 0.841,
+      "step": 3280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.838098745285485e-05,
+      "loss": 0.3386,
+      "step": 3281
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.832485453557424e-05,
+      "loss": 0.7826,
+      "step": 3282
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.826872878369256e-05,
+      "loss": 0.79,
+      "step": 3283
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.821261021576391e-05,
+      "loss": 0.756,
+      "step": 3284
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.815649885034003e-05,
+      "loss": 0.3404,
+      "step": 3285
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.81003947059703e-05,
+      "loss": 0.8289,
+      "step": 3286
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.804429780120172e-05,
+      "loss": 0.8267,
+      "step": 3287
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.798820815457886e-05,
+      "loss": 0.8015,
+      "step": 3288
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.793212578464393e-05,
+      "loss": 0.7841,
+      "step": 3289
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.787605070993668e-05,
+      "loss": 0.7362,
+      "step": 3290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.781998294899453e-05,
+      "loss": 0.7768,
+      "step": 3291
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.776392252035237e-05,
+      "loss": 0.7929,
+      "step": 3292
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.77078694425428e-05,
+      "loss": 0.8605,
+      "step": 3293
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.765182373409591e-05,
+      "loss": 0.7831,
+      "step": 3294
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.759578541353935e-05,
+      "loss": 0.8332,
+      "step": 3295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.753975449939835e-05,
+      "loss": 0.8187,
+      "step": 3296
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.748373101019568e-05,
+      "loss": 0.8258,
+      "step": 3297
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.742771496445167e-05,
+      "loss": 0.8082,
+      "step": 3298
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.737170638068416e-05,
+      "loss": 0.8682,
+      "step": 3299
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.731570527740856e-05,
+      "loss": 0.8142,
+      "step": 3300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.725971167313782e-05,
+      "loss": 0.7623,
+      "step": 3301
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.720372558638233e-05,
+      "loss": 0.8077,
+      "step": 3302
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.714774703565004e-05,
+      "loss": 0.808,
+      "step": 3303
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.709177603944645e-05,
+      "loss": 0.3668,
+      "step": 3304
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.703581261627453e-05,
+      "loss": 0.865,
+      "step": 3305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.697985678463476e-05,
+      "loss": 0.7597,
+      "step": 3306
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.692390856302502e-05,
+      "loss": 0.8528,
+      "step": 3307
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.686796796994084e-05,
+      "loss": 0.8656,
+      "step": 3308
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.681203502387507e-05,
+      "loss": 0.7806,
+      "step": 3309
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.675610974331813e-05,
+      "loss": 0.7906,
+      "step": 3310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.670019214675788e-05,
+      "loss": 0.817,
+      "step": 3311
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.66442822526796e-05,
+      "loss": 0.7627,
+      "step": 3312
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.658838007956605e-05,
+      "loss": 0.7667,
+      "step": 3313
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.653248564589751e-05,
+      "loss": 0.8457,
+      "step": 3314
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.647659897015161e-05,
+      "loss": 0.8102,
+      "step": 3315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.642072007080343e-05,
+      "loss": 0.8599,
+      "step": 3316
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.63648489663255e-05,
+      "loss": 0.8289,
+      "step": 3317
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.630898567518778e-05,
+      "loss": 0.7476,
+      "step": 3318
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.62531302158576e-05,
+      "loss": 0.8617,
+      "step": 3319
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.619728260679975e-05,
+      "loss": 0.8389,
+      "step": 3320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.61414428664764e-05,
+      "loss": 0.8075,
+      "step": 3321
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.608561101334714e-05,
+      "loss": 0.8503,
+      "step": 3322
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.602978706586894e-05,
+      "loss": 0.8251,
+      "step": 3323
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.597397104249613e-05,
+      "loss": 0.7801,
+      "step": 3324
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.591816296168049e-05,
+      "loss": 0.7925,
+      "step": 3325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.586236284187106e-05,
+      "loss": 0.8506,
+      "step": 3326
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.580657070151442e-05,
+      "loss": 0.8214,
+      "step": 3327
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.575078655905434e-05,
+      "loss": 0.7929,
+      "step": 3328
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.569501043293205e-05,
+      "loss": 0.8141,
+      "step": 3329
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.563924234158607e-05,
+      "loss": 0.8157,
+      "step": 3330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.558348230345231e-05,
+      "loss": 0.8021,
+      "step": 3331
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.552773033696398e-05,
+      "loss": 0.3585,
+      "step": 3332
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.547198646055167e-05,
+      "loss": 0.7805,
+      "step": 3333
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.541625069264324e-05,
+      "loss": 0.7941,
+      "step": 3334
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.536052305166387e-05,
+      "loss": 0.3112,
+      "step": 3335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.530480355603615e-05,
+      "loss": 0.8207,
+      "step": 3336
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.524909222417987e-05,
+      "loss": 0.8006,
+      "step": 3337
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.519338907451215e-05,
+      "loss": 0.842,
+      "step": 3338
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.513769412544742e-05,
+      "loss": 0.7796,
+      "step": 3339
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.508200739539739e-05,
+      "loss": 0.765,
+      "step": 3340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.502632890277106e-05,
+      "loss": 0.7606,
+      "step": 3341
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.49706586659747e-05,
+      "loss": 0.3557,
+      "step": 3342
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.491499670341182e-05,
+      "loss": 0.8038,
+      "step": 3343
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.485934303348327e-05,
+      "loss": 0.8711,
+      "step": 3344
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.48036976745871e-05,
+      "loss": 0.327,
+      "step": 3345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.474806064511863e-05,
+      "loss": 0.8678,
+      "step": 3346
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.469243196347044e-05,
+      "loss": 0.8023,
+      "step": 3347
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.46368116480323e-05,
+      "loss": 0.803,
+      "step": 3348
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.458119971719128e-05,
+      "loss": 0.7625,
+      "step": 3349
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.452559618933164e-05,
+      "loss": 0.8124,
+      "step": 3350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.447000108283485e-05,
+      "loss": 0.7478,
+      "step": 3351
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.441441441607964e-05,
+      "loss": 0.8168,
+      "step": 3352
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.435883620744192e-05,
+      "loss": 0.7982,
+      "step": 3353
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.43032664752948e-05,
+      "loss": 0.7984,
+      "step": 3354
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.424770523800859e-05,
+      "loss": 0.7624,
+      "step": 3355
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.419215251395078e-05,
+      "loss": 0.7774,
+      "step": 3356
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.413660832148614e-05,
+      "loss": 0.846,
+      "step": 3357
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.408107267897651e-05,
+      "loss": 0.8764,
+      "step": 3358
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.402554560478092e-05,
+      "loss": 0.8145,
+      "step": 3359
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.397002711725558e-05,
+      "loss": 0.8055,
+      "step": 3360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.391451723475389e-05,
+      "loss": 0.8313,
+      "step": 3361
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.385901597562637e-05,
+      "loss": 0.8686,
+      "step": 3362
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.380352335822069e-05,
+      "loss": 0.7557,
+      "step": 3363
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.374803940088171e-05,
+      "loss": 0.8292,
+      "step": 3364
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.369256412195134e-05,
+      "loss": 0.8469,
+      "step": 3365
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.36370975397687e-05,
+      "loss": 0.8549,
+      "step": 3366
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.358163967267002e-05,
+      "loss": 0.7976,
+      "step": 3367
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.352619053898864e-05,
+      "loss": 0.8063,
+      "step": 3368
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.347075015705499e-05,
+      "loss": 0.8224,
+      "step": 3369
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.341531854519664e-05,
+      "loss": 0.8241,
+      "step": 3370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.335989572173821e-05,
+      "loss": 0.8104,
+      "step": 3371
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.33044817050015e-05,
+      "loss": 0.8285,
+      "step": 3372
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.324907651330535e-05,
+      "loss": 0.8393,
+      "step": 3373
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.319368016496564e-05,
+      "loss": 0.8243,
+      "step": 3374
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.313829267829542e-05,
+      "loss": 0.7745,
+      "step": 3375
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.308291407160472e-05,
+      "loss": 0.8407,
+      "step": 3376
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.302754436320065e-05,
+      "loss": 0.849,
+      "step": 3377
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.29721835713875e-05,
+      "loss": 0.338,
+      "step": 3378
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.291683171446643e-05,
+      "loss": 0.827,
+      "step": 3379
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.286148881073578e-05,
+      "loss": 0.7991,
+      "step": 3380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.280615487849084e-05,
+      "loss": 0.7949,
+      "step": 3381
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.275082993602402e-05,
+      "loss": 0.8747,
+      "step": 3382
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.269551400162469e-05,
+      "loss": 0.8071,
+      "step": 3383
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.264020709357927e-05,
+      "loss": 0.7703,
+      "step": 3384
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.258490923017118e-05,
+      "loss": 0.8011,
+      "step": 3385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.25296204296809e-05,
+      "loss": 0.8123,
+      "step": 3386
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.247434071038582e-05,
+      "loss": 0.8153,
+      "step": 3387
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.241907009056039e-05,
+      "loss": 0.805,
+      "step": 3388
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.236380858847611e-05,
+      "loss": 0.8029,
+      "step": 3389
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.230855622240136e-05,
+      "loss": 0.8172,
+      "step": 3390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.225331301060154e-05,
+      "loss": 0.801,
+      "step": 3391
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.219807897133906e-05,
+      "loss": 0.7851,
+      "step": 3392
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.214285412287318e-05,
+      "loss": 0.8432,
+      "step": 3393
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.208763848346029e-05,
+      "loss": 0.815,
+      "step": 3394
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.203243207135362e-05,
+      "loss": 0.7947,
+      "step": 3395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.197723490480338e-05,
+      "loss": 0.8471,
+      "step": 3396
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.192204700205674e-05,
+      "loss": 0.8177,
+      "step": 3397
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.186686838135774e-05,
+      "loss": 0.8298,
+      "step": 3398
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.181169906094743e-05,
+      "loss": 0.7947,
+      "step": 3399
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.17565390590638e-05,
+      "loss": 0.8025,
+      "step": 3400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.170138839394167e-05,
+      "loss": 0.8093,
+      "step": 3401
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.164624708381285e-05,
+      "loss": 0.8266,
+      "step": 3402
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.159111514690602e-05,
+      "loss": 0.8002,
+      "step": 3403
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.153599260144677e-05,
+      "loss": 0.8255,
+      "step": 3404
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.148087946565757e-05,
+      "loss": 0.8613,
+      "step": 3405
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.142577575775782e-05,
+      "loss": 0.8268,
+      "step": 3406
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.137068149596375e-05,
+      "loss": 0.8015,
+      "step": 3407
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.13155966984885e-05,
+      "loss": 0.8329,
+      "step": 3408
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.126052138354207e-05,
+      "loss": 0.7894,
+      "step": 3409
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.120545556933138e-05,
+      "loss": 0.7945,
+      "step": 3410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.115039927406013e-05,
+      "loss": 0.8566,
+      "step": 3411
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.109535251592892e-05,
+      "loss": 0.7429,
+      "step": 3412
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.104031531313515e-05,
+      "loss": 0.8535,
+      "step": 3413
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.098528768387311e-05,
+      "loss": 0.8078,
+      "step": 3414
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.093026964633391e-05,
+      "loss": 0.8169,
+      "step": 3415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.087526121870548e-05,
+      "loss": 0.7948,
+      "step": 3416
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.082026241917259e-05,
+      "loss": 0.8346,
+      "step": 3417
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.076527326591682e-05,
+      "loss": 0.824,
+      "step": 3418
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.071029377711656e-05,
+      "loss": 0.8045,
+      "step": 3419
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.065532397094695e-05,
+      "loss": 0.7916,
+      "step": 3420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.060036386558005e-05,
+      "loss": 0.8145,
+      "step": 3421
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.054541347918464e-05,
+      "loss": 0.7795,
+      "step": 3422
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.049047282992629e-05,
+      "loss": 0.81,
+      "step": 3423
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.043554193596732e-05,
+      "loss": 0.8279,
+      "step": 3424
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.038062081546689e-05,
+      "loss": 0.8421,
+      "step": 3425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.03257094865809e-05,
+      "loss": 0.8198,
+      "step": 3426
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.027080796746198e-05,
+      "loss": 0.8822,
+      "step": 3427
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.021591627625958e-05,
+      "loss": 0.7893,
+      "step": 3428
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.016103443111986e-05,
+      "loss": 0.7924,
+      "step": 3429
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.010616245018573e-05,
+      "loss": 0.7505,
+      "step": 3430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.005130035159681e-05,
+      "loss": 0.8435,
+      "step": 3431
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.999644815348956e-05,
+      "loss": 0.8368,
+      "step": 3432
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.994160587399705e-05,
+      "loss": 0.8079,
+      "step": 3433
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.988677353124913e-05,
+      "loss": 0.7993,
+      "step": 3434
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.983195114337236e-05,
+      "loss": 0.8117,
+      "step": 3435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.977713872848995e-05,
+      "loss": 0.8352,
+      "step": 3436
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.972233630472188e-05,
+      "loss": 0.815,
+      "step": 3437
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.966754389018487e-05,
+      "loss": 0.8333,
+      "step": 3438
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.96127615029922e-05,
+      "loss": 0.8464,
+      "step": 3439
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.955798916125393e-05,
+      "loss": 0.8503,
+      "step": 3440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.950322688307675e-05,
+      "loss": 0.8145,
+      "step": 3441
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.94484746865641e-05,
+      "loss": 0.7536,
+      "step": 3442
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9393732589816e-05,
+      "loss": 0.8004,
+      "step": 3443
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.933900061092919e-05,
+      "loss": 0.7888,
+      "step": 3444
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.928427876799702e-05,
+      "loss": 0.7761,
+      "step": 3445
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.92295670791095e-05,
+      "loss": 0.7841,
+      "step": 3446
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.917486556235333e-05,
+      "loss": 0.8103,
+      "step": 3447
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.912017423581179e-05,
+      "loss": 0.7886,
+      "step": 3448
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.906549311756481e-05,
+      "loss": 0.7734,
+      "step": 3449
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.901082222568895e-05,
+      "loss": 0.8388,
+      "step": 3450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.895616157825739e-05,
+      "loss": 0.3211,
+      "step": 3451
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.890151119333988e-05,
+      "loss": 0.8486,
+      "step": 3452
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.88468710890029e-05,
+      "loss": 0.7828,
+      "step": 3453
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.87922412833094e-05,
+      "loss": 0.7935,
+      "step": 3454
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8737621794319e-05,
+      "loss": 0.8215,
+      "step": 3455
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.868301264008785e-05,
+      "loss": 0.8336,
+      "step": 3456
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.862841383866873e-05,
+      "loss": 0.3357,
+      "step": 3457
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.857382540811101e-05,
+      "loss": 0.7967,
+      "step": 3458
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.851924736646058e-05,
+      "loss": 0.7741,
+      "step": 3459
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.846467973175993e-05,
+      "loss": 0.7789,
+      "step": 3460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.841012252204811e-05,
+      "loss": 0.8147,
+      "step": 3461
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.835557575536071e-05,
+      "loss": 0.8159,
+      "step": 3462
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.830103944972986e-05,
+      "loss": 0.8704,
+      "step": 3463
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.824651362318425e-05,
+      "loss": 0.8457,
+      "step": 3464
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.819199829374914e-05,
+      "loss": 0.333,
+      "step": 3465
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.813749347944625e-05,
+      "loss": 0.8741,
+      "step": 3466
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.808299919829384e-05,
+      "loss": 0.7888,
+      "step": 3467
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.802851546830674e-05,
+      "loss": 0.8261,
+      "step": 3468
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.797404230749623e-05,
+      "loss": 0.8429,
+      "step": 3469
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.791957973387013e-05,
+      "loss": 0.7756,
+      "step": 3470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.786512776543274e-05,
+      "loss": 0.8198,
+      "step": 3471
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.781068642018488e-05,
+      "loss": 0.8034,
+      "step": 3472
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.77562557161238e-05,
+      "loss": 0.7628,
+      "step": 3473
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.770183567124337e-05,
+      "loss": 0.8277,
+      "step": 3474
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.764742630353376e-05,
+      "loss": 0.7617,
+      "step": 3475
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.759302763098172e-05,
+      "loss": 0.3365,
+      "step": 3476
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.753863967157045e-05,
+      "loss": 0.7945,
+      "step": 3477
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.748426244327957e-05,
+      "loss": 0.8242,
+      "step": 3478
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.742989596408517e-05,
+      "loss": 0.3731,
+      "step": 3479
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.737554025195984e-05,
+      "loss": 0.8001,
+      "step": 3480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.73211953248725e-05,
+      "loss": 0.7759,
+      "step": 3481
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.726686120078862e-05,
+      "loss": 0.8112,
+      "step": 3482
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.721253789767006e-05,
+      "loss": 0.7647,
+      "step": 3483
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.715822543347502e-05,
+      "loss": 0.7964,
+      "step": 3484
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.710392382615826e-05,
+      "loss": 0.7736,
+      "step": 3485
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.704963309367083e-05,
+      "loss": 0.8377,
+      "step": 3486
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.699535325396027e-05,
+      "loss": 0.8078,
+      "step": 3487
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.694108432497048e-05,
+      "loss": 0.778,
+      "step": 3488
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.688682632464174e-05,
+      "loss": 0.7979,
+      "step": 3489
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.683257927091074e-05,
+      "loss": 0.829,
+      "step": 3490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.677834318171054e-05,
+      "loss": 0.8293,
+      "step": 3491
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.672411807497057e-05,
+      "loss": 0.7861,
+      "step": 3492
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.666990396861667e-05,
+      "loss": 0.8436,
+      "step": 3493
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.661570088057097e-05,
+      "loss": 0.8164,
+      "step": 3494
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6561508828752e-05,
+      "loss": 0.7789,
+      "step": 3495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.65073278310747e-05,
+      "loss": 0.3397,
+      "step": 3496
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.645315790545025e-05,
+      "loss": 0.8531,
+      "step": 3497
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.639899906978626e-05,
+      "loss": 0.8581,
+      "step": 3498
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.634485134198659e-05,
+      "loss": 0.9016,
+      "step": 3499
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.629071473995147e-05,
+      "loss": 0.8249,
+      "step": 3500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.623658928157747e-05,
+      "loss": 0.7402,
+      "step": 3501
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.618247498475744e-05,
+      "loss": 0.8266,
+      "step": 3502
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.612837186738054e-05,
+      "loss": 0.776,
+      "step": 3503
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.60742799473323e-05,
+      "loss": 0.7966,
+      "step": 3504
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.602019924249447e-05,
+      "loss": 0.8039,
+      "step": 3505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.596612977074515e-05,
+      "loss": 0.7824,
+      "step": 3506
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.591207154995867e-05,
+      "loss": 0.8068,
+      "step": 3507
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.585802459800566e-05,
+      "loss": 0.8324,
+      "step": 3508
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.580398893275308e-05,
+      "loss": 0.8069,
+      "step": 3509
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.574996457206408e-05,
+      "loss": 0.8567,
+      "step": 3510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.569595153379811e-05,
+      "loss": 0.7575,
+      "step": 3511
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.564194983581089e-05,
+      "loss": 0.8782,
+      "step": 3512
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.558795949595435e-05,
+      "loss": 0.7672,
+      "step": 3513
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.553398053207671e-05,
+      "loss": 0.8437,
+      "step": 3514
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.548001296202238e-05,
+      "loss": 0.8145,
+      "step": 3515
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.542605680363204e-05,
+      "loss": 0.7689,
+      "step": 3516
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.53721120747426e-05,
+      "loss": 0.7343,
+      "step": 3517
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.53181787931872e-05,
+      "loss": 0.7873,
+      "step": 3518
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.526425697679518e-05,
+      "loss": 0.8142,
+      "step": 3519
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.521034664339204e-05,
+      "loss": 0.8203,
+      "step": 3520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.515644781079958e-05,
+      "loss": 0.8089,
+      "step": 3521
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.510256049683571e-05,
+      "loss": 0.7457,
+      "step": 3522
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.50486847193146e-05,
+      "loss": 0.8574,
+      "step": 3523
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.499482049604656e-05,
+      "loss": 0.7368,
+      "step": 3524
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.494096784483809e-05,
+      "loss": 0.8219,
+      "step": 3525
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.488712678349189e-05,
+      "loss": 0.8579,
+      "step": 3526
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.483329732980678e-05,
+      "loss": 0.8106,
+      "step": 3527
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.477947950157785e-05,
+      "loss": 0.7953,
+      "step": 3528
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.47256733165962e-05,
+      "loss": 0.8762,
+      "step": 3529
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.467187879264916e-05,
+      "loss": 0.8638,
+      "step": 3530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.461809594752022e-05,
+      "loss": 0.7342,
+      "step": 3531
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.456432479898897e-05,
+      "loss": 0.7731,
+      "step": 3532
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.451056536483115e-05,
+      "loss": 0.8354,
+      "step": 3533
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.445681766281863e-05,
+      "loss": 0.3718,
+      "step": 3534
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.440308171071942e-05,
+      "loss": 0.8566,
+      "step": 3535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.434935752629758e-05,
+      "loss": 0.8272,
+      "step": 3536
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.429564512731331e-05,
+      "loss": 0.8771,
+      "step": 3537
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4241944531523e-05,
+      "loss": 0.8202,
+      "step": 3538
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.418825575667904e-05,
+      "loss": 0.7954,
+      "step": 3539
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.413457882052991e-05,
+      "loss": 0.8155,
+      "step": 3540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.408091374082021e-05,
+      "loss": 0.7731,
+      "step": 3541
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.402726053529065e-05,
+      "loss": 0.8332,
+      "step": 3542
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.397361922167793e-05,
+      "loss": 0.7536,
+      "step": 3543
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.391998981771492e-05,
+      "loss": 0.8643,
+      "step": 3544
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.386637234113044e-05,
+      "loss": 0.7719,
+      "step": 3545
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.381276680964947e-05,
+      "loss": 0.8287,
+      "step": 3546
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.375917324099297e-05,
+      "loss": 0.8118,
+      "step": 3547
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3705591652878e-05,
+      "loss": 0.7424,
+      "step": 3548
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.365202206301765e-05,
+      "loss": 0.8672,
+      "step": 3549
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.359846448912099e-05,
+      "loss": 0.8307,
+      "step": 3550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.354491894889319e-05,
+      "loss": 0.8532,
+      "step": 3551
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.349138546003534e-05,
+      "loss": 0.8368,
+      "step": 3552
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.34378640402447e-05,
+      "loss": 0.8279,
+      "step": 3553
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.338435470721442e-05,
+      "loss": 0.8018,
+      "step": 3554
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.333085747863368e-05,
+      "loss": 0.8367,
+      "step": 3555
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.327737237218765e-05,
+      "loss": 0.7675,
+      "step": 3556
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.322389940555753e-05,
+      "loss": 0.8087,
+      "step": 3557
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.317043859642049e-05,
+      "loss": 0.8004,
+      "step": 3558
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.311698996244962e-05,
+      "loss": 0.8198,
+      "step": 3559
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.306355352131414e-05,
+      "loss": 0.7774,
+      "step": 3560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.301012929067908e-05,
+      "loss": 0.8139,
+      "step": 3561
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.295671728820553e-05,
+      "loss": 0.3723,
+      "step": 3562
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.290331753155046e-05,
+      "loss": 0.8449,
+      "step": 3563
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.284993003836686e-05,
+      "loss": 0.7716,
+      "step": 3564
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.279655482630361e-05,
+      "loss": 0.8674,
+      "step": 3565
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.27431919130056e-05,
+      "loss": 0.778,
+      "step": 3566
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.268984131611359e-05,
+      "loss": 0.7071,
+      "step": 3567
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.263650305326429e-05,
+      "loss": 0.8312,
+      "step": 3568
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.258317714209031e-05,
+      "loss": 0.8137,
+      "step": 3569
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.252986360022029e-05,
+      "loss": 0.799,
+      "step": 3570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.247656244527861e-05,
+      "loss": 0.7766,
+      "step": 3571
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.242327369488568e-05,
+      "loss": 0.8626,
+      "step": 3572
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.236999736665774e-05,
+      "loss": 0.793,
+      "step": 3573
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.231673347820694e-05,
+      "loss": 0.8254,
+      "step": 3574
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.22634820471414e-05,
+      "loss": 0.8049,
+      "step": 3575
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.221024309106498e-05,
+      "loss": 0.8139,
+      "step": 3576
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.215701662757751e-05,
+      "loss": 0.8,
+      "step": 3577
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.210380267427467e-05,
+      "loss": 0.7435,
+      "step": 3578
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.205060124874799e-05,
+      "loss": 0.7875,
+      "step": 3579
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.199741236858483e-05,
+      "loss": 0.8287,
+      "step": 3580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.194423605136852e-05,
+      "loss": 0.763,
+      "step": 3581
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.189107231467814e-05,
+      "loss": 0.8747,
+      "step": 3582
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.18379211760886e-05,
+      "loss": 0.8038,
+      "step": 3583
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.17847826531707e-05,
+      "loss": 0.8002,
+      "step": 3584
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.173165676349103e-05,
+      "loss": 0.3644,
+      "step": 3585
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.167854352461202e-05,
+      "loss": 0.7801,
+      "step": 3586
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.162544295409192e-05,
+      "loss": 0.7926,
+      "step": 3587
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.15723550694848e-05,
+      "loss": 0.7985,
+      "step": 3588
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.15192798883405e-05,
+      "loss": 0.8507,
+      "step": 3589
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.146621742820471e-05,
+      "loss": 0.754,
+      "step": 3590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.141316770661882e-05,
+      "loss": 0.8426,
+      "step": 3591
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.136013074112018e-05,
+      "loss": 0.7843,
+      "step": 3592
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.130710654924179e-05,
+      "loss": 0.8263,
+      "step": 3593
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.125409514851243e-05,
+      "loss": 0.7832,
+      "step": 3594
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.12010965564567e-05,
+      "loss": 0.8251,
+      "step": 3595
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.114811079059495e-05,
+      "loss": 0.7867,
+      "step": 3596
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.109513786844325e-05,
+      "loss": 0.36,
+      "step": 3597
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.104217780751353e-05,
+      "loss": 0.8197,
+      "step": 3598
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.098923062531334e-05,
+      "loss": 0.8145,
+      "step": 3599
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0936296339346054e-05,
+      "loss": 0.7842,
+      "step": 3600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.088337496711076e-05,
+      "loss": 0.8344,
+      "step": 3601
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.083046652610224e-05,
+      "loss": 0.818,
+      "step": 3602
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0777571033811095e-05,
+      "loss": 0.7978,
+      "step": 3603
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.072468850772357e-05,
+      "loss": 0.773,
+      "step": 3604
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0671818965321634e-05,
+      "loss": 0.7975,
+      "step": 3605
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.061896242408298e-05,
+      "loss": 0.7941,
+      "step": 3606
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.056611890148098e-05,
+      "loss": 0.8105,
+      "step": 3607
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.051328841498473e-05,
+      "loss": 0.7645,
+      "step": 3608
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.046047098205899e-05,
+      "loss": 0.883,
+      "step": 3609
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.040766662016424e-05,
+      "loss": 0.8331,
+      "step": 3610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.035487534675659e-05,
+      "loss": 0.8123,
+      "step": 3611
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0302097179287844e-05,
+      "loss": 0.853,
+      "step": 3612
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.024933213520553e-05,
+      "loss": 0.7603,
+      "step": 3613
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.019658023195276e-05,
+      "loss": 0.7932,
+      "step": 3614
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.014384148696833e-05,
+      "loss": 0.8108,
+      "step": 3615
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.009111591768668e-05,
+      "loss": 0.8239,
+      "step": 3616
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.003840354153791e-05,
+      "loss": 0.8059,
+      "step": 3617
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.998570437594775e-05,
+      "loss": 0.7585,
+      "step": 3618
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.993301843833752e-05,
+      "loss": 0.7892,
+      "step": 3619
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9880345746124265e-05,
+      "loss": 0.7855,
+      "step": 3620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9827686316720586e-05,
+      "loss": 0.8335,
+      "step": 3621
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.977504016753468e-05,
+      "loss": 0.8287,
+      "step": 3622
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.972240731597038e-05,
+      "loss": 0.8324,
+      "step": 3623
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9669787779427155e-05,
+      "loss": 0.8026,
+      "step": 3624
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.961718157530004e-05,
+      "loss": 0.7788,
+      "step": 3625
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9564588720979655e-05,
+      "loss": 0.8487,
+      "step": 3626
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.951200923385223e-05,
+      "loss": 0.844,
+      "step": 3627
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.945944313129953e-05,
+      "loss": 0.7906,
+      "step": 3628
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9406890430698955e-05,
+      "loss": 0.7963,
+      "step": 3629
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.935435114942345e-05,
+      "loss": 0.8567,
+      "step": 3630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9301825304841497e-05,
+      "loss": 0.8086,
+      "step": 3631
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.924931291431719e-05,
+      "loss": 0.759,
+      "step": 3632
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.919681399521013e-05,
+      "loss": 0.8166,
+      "step": 3633
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.914432856487544e-05,
+      "loss": 0.3652,
+      "step": 3634
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.909185664066391e-05,
+      "loss": 0.8216,
+      "step": 3635
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.903939823992174e-05,
+      "loss": 0.7951,
+      "step": 3636
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.89869533799907e-05,
+      "loss": 0.8549,
+      "step": 3637
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8934522078208066e-05,
+      "loss": 0.7681,
+      "step": 3638
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.888210435190666e-05,
+      "loss": 0.8184,
+      "step": 3639
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.882970021841483e-05,
+      "loss": 0.8355,
+      "step": 3640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8777309695056346e-05,
+      "loss": 0.8056,
+      "step": 3641
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8724932799150586e-05,
+      "loss": 0.7799,
+      "step": 3642
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.867256954801238e-05,
+      "loss": 0.7918,
+      "step": 3643
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8620219958952e-05,
+      "loss": 0.8038,
+      "step": 3644
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.856788404927529e-05,
+      "loss": 0.7665,
+      "step": 3645
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.851556183628348e-05,
+      "loss": 0.7658,
+      "step": 3646
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.846325333727337e-05,
+      "loss": 0.7688,
+      "step": 3647
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8410958569537146e-05,
+      "loss": 0.7938,
+      "step": 3648
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.835867755036251e-05,
+      "loss": 0.8167,
+      "step": 3649
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.830641029703254e-05,
+      "loss": 0.7445,
+      "step": 3650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8254156826825846e-05,
+      "loss": 0.8231,
+      "step": 3651
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.82019171570164e-05,
+      "loss": 0.7862,
+      "step": 3652
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.814969130487375e-05,
+      "loss": 0.7949,
+      "step": 3653
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8097479287662756e-05,
+      "loss": 0.8137,
+      "step": 3654
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.804528112264372e-05,
+      "loss": 0.7757,
+      "step": 3655
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.79930968270724e-05,
+      "loss": 0.7923,
+      "step": 3656
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.79409264181999e-05,
+      "loss": 0.8372,
+      "step": 3657
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.788876991327288e-05,
+      "loss": 0.8306,
+      "step": 3658
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.783662732953323e-05,
+      "loss": 0.8019,
+      "step": 3659
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.778449868421836e-05,
+      "loss": 0.878,
+      "step": 3660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7732383994561004e-05,
+      "loss": 0.832,
+      "step": 3661
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.768028327778932e-05,
+      "loss": 0.7934,
+      "step": 3662
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.76281965511268e-05,
+      "loss": 0.7986,
+      "step": 3663
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.757612383179238e-05,
+      "loss": 0.8285,
+      "step": 3664
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7524065137000327e-05,
+      "loss": 0.7596,
+      "step": 3665
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.747202048396023e-05,
+      "loss": 0.3352,
+      "step": 3666
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.741998988987716e-05,
+      "loss": 0.7979,
+      "step": 3667
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.73679733719514e-05,
+      "loss": 0.8299,
+      "step": 3668
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.731597094737867e-05,
+      "loss": 0.7461,
+      "step": 3669
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.726398263335e-05,
+      "loss": 0.8626,
+      "step": 3670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.721200844705175e-05,
+      "loss": 0.7495,
+      "step": 3671
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.71600484056656e-05,
+      "loss": 0.8203,
+      "step": 3672
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.710810252636859e-05,
+      "loss": 0.7662,
+      "step": 3673
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.705617082633306e-05,
+      "loss": 0.7814,
+      "step": 3674
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.700425332272665e-05,
+      "loss": 0.7511,
+      "step": 3675
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.695235003271231e-05,
+      "loss": 0.8229,
+      "step": 3676
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.690046097344833e-05,
+      "loss": 0.802,
+      "step": 3677
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.684858616208826e-05,
+      "loss": 0.8115,
+      "step": 3678
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.679672561578093e-05,
+      "loss": 0.8488,
+      "step": 3679
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.674487935167049e-05,
+      "loss": 0.8225,
+      "step": 3680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.669304738689633e-05,
+      "loss": 0.8189,
+      "step": 3681
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.664122973859313e-05,
+      "loss": 0.7277,
+      "step": 3682
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6589426423890844e-05,
+      "loss": 0.8499,
+      "step": 3683
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.653763745991467e-05,
+      "loss": 0.8316,
+      "step": 3684
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.64858628637851e-05,
+      "loss": 0.7536,
+      "step": 3685
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.643410265261784e-05,
+      "loss": 0.8207,
+      "step": 3686
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.638235684352379e-05,
+      "loss": 0.8208,
+      "step": 3687
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.633062545360925e-05,
+      "loss": 0.8052,
+      "step": 3688
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.627890849997561e-05,
+      "loss": 0.8572,
+      "step": 3689
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.622720599971952e-05,
+      "loss": 0.7809,
+      "step": 3690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.617551796993288e-05,
+      "loss": 0.7952,
+      "step": 3691
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6123844427702775e-05,
+      "loss": 0.8208,
+      "step": 3692
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.607218539011152e-05,
+      "loss": 0.8007,
+      "step": 3693
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.602054087423663e-05,
+      "loss": 0.7887,
+      "step": 3694
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5968910897150816e-05,
+      "loss": 0.8654,
+      "step": 3695
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.591729547592195e-05,
+      "loss": 0.8026,
+      "step": 3696
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.586569462761321e-05,
+      "loss": 0.7741,
+      "step": 3697
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5814108369282824e-05,
+      "loss": 0.7907,
+      "step": 3698
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.576253671798424e-05,
+      "loss": 0.7583,
+      "step": 3699
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.571097969076611e-05,
+      "loss": 0.8506,
+      "step": 3700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5659437304672166e-05,
+      "loss": 0.7564,
+      "step": 3701
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5607909576741445e-05,
+      "loss": 0.8097,
+      "step": 3702
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.555639652400801e-05,
+      "loss": 0.8501,
+      "step": 3703
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.550489816350113e-05,
+      "loss": 0.771,
+      "step": 3704
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.545341451224519e-05,
+      "loss": 0.7792,
+      "step": 3705
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.540194558725973e-05,
+      "loss": 0.7823,
+      "step": 3706
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5350491405559416e-05,
+      "loss": 0.8101,
+      "step": 3707
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5299051984153995e-05,
+      "loss": 0.8758,
+      "step": 3708
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.524762734004845e-05,
+      "loss": 0.8449,
+      "step": 3709
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5196217490242793e-05,
+      "loss": 0.832,
+      "step": 3710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.514482245173216e-05,
+      "loss": 0.778,
+      "step": 3711
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5093442241506784e-05,
+      "loss": 0.3368,
+      "step": 3712
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5042076876552005e-05,
+      "loss": 0.7844,
+      "step": 3713
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4990726373848243e-05,
+      "loss": 0.8259,
+      "step": 3714
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.493939075037104e-05,
+      "loss": 0.7934,
+      "step": 3715
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.488807002309098e-05,
+      "loss": 0.8444,
+      "step": 3716
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4836764208973746e-05,
+      "loss": 0.8151,
+      "step": 3717
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.478547332498007e-05,
+      "loss": 0.74,
+      "step": 3718
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.473419738806575e-05,
+      "loss": 0.8224,
+      "step": 3719
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.468293641518172e-05,
+      "loss": 0.3991,
+      "step": 3720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.463169042327383e-05,
+      "loss": 0.7876,
+      "step": 3721
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.458045942928309e-05,
+      "loss": 0.7909,
+      "step": 3722
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.452924345014552e-05,
+      "loss": 0.3489,
+      "step": 3723
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.447804250279213e-05,
+      "loss": 0.7614,
+      "step": 3724
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.442685660414901e-05,
+      "loss": 0.762,
+      "step": 3725
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.437568577113727e-05,
+      "loss": 0.8034,
+      "step": 3726
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.432453002067304e-05,
+      "loss": 0.8167,
+      "step": 3727
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4273389369667436e-05,
+      "loss": 0.8189,
+      "step": 3728
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4222263835026645e-05,
+      "loss": 0.3707,
+      "step": 3729
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.417115343365171e-05,
+      "loss": 0.3617,
+      "step": 3730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4120058182438925e-05,
+      "loss": 0.8305,
+      "step": 3731
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4068978098279336e-05,
+      "loss": 0.7901,
+      "step": 3732
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.401791319805909e-05,
+      "loss": 0.3298,
+      "step": 3733
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.396686349865929e-05,
+      "loss": 0.8432,
+      "step": 3734
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.391582901695602e-05,
+      "loss": 0.8367,
+      "step": 3735
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3864809769820315e-05,
+      "loss": 0.86,
+      "step": 3736
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3813805774118186e-05,
+      "loss": 0.7793,
+      "step": 3737
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.37628170467106e-05,
+      "loss": 0.7389,
+      "step": 3738
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3711843604453506e-05,
+      "loss": 0.7619,
+      "step": 3739
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.366088546419771e-05,
+      "loss": 0.8093,
+      "step": 3740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3609942642789104e-05,
+      "loss": 0.8284,
+      "step": 3741
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3559015157068404e-05,
+      "loss": 0.8599,
+      "step": 3742
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.350810302387127e-05,
+      "loss": 0.7977,
+      "step": 3743
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3457206260028324e-05,
+      "loss": 0.782,
+      "step": 3744
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.340632488236503e-05,
+      "loss": 0.8014,
+      "step": 3745
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3355458907701925e-05,
+      "loss": 0.8105,
+      "step": 3746
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.33046083528543e-05,
+      "loss": 0.8241,
+      "step": 3747
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.325377323463239e-05,
+      "loss": 0.8316,
+      "step": 3748
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.320295356984135e-05,
+      "loss": 0.8054,
+      "step": 3749
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.315214937528121e-05,
+      "loss": 0.8248,
+      "step": 3750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.310136066774685e-05,
+      "loss": 0.8149,
+      "step": 3751
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3050587464028136e-05,
+      "loss": 0.8112,
+      "step": 3752
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.299982978090971e-05,
+      "loss": 0.8093,
+      "step": 3753
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2949087635171144e-05,
+      "loss": 0.7715,
+      "step": 3754
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2898361043586806e-05,
+      "loss": 0.7188,
+      "step": 3755
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.284765002292598e-05,
+      "loss": 0.7671,
+      "step": 3756
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2796954589952787e-05,
+      "loss": 0.7794,
+      "step": 3757
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2746274761426176e-05,
+      "loss": 0.8571,
+      "step": 3758
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.269561055409996e-05,
+      "loss": 0.7413,
+      "step": 3759
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2644961984722796e-05,
+      "loss": 0.8409,
+      "step": 3760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.259432907003813e-05,
+      "loss": 0.848,
+      "step": 3761
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.254371182678424e-05,
+      "loss": 0.7975,
+      "step": 3762
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.249311027169431e-05,
+      "loss": 0.7822,
+      "step": 3763
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.244252442149624e-05,
+      "loss": 0.7906,
+      "step": 3764
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.239195429291277e-05,
+      "loss": 0.797,
+      "step": 3765
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.234139990266143e-05,
+      "loss": 0.8501,
+      "step": 3766
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2290861267454574e-05,
+      "loss": 0.7929,
+      "step": 3767
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.224033840399931e-05,
+      "loss": 0.7385,
+      "step": 3768
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.218983132899758e-05,
+      "loss": 0.8277,
+      "step": 3769
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.213934005914607e-05,
+      "loss": 0.7293,
+      "step": 3770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.208886461113625e-05,
+      "loss": 0.7659,
+      "step": 3771
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.203840500165434e-05,
+      "loss": 0.3381,
+      "step": 3772
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.198796124738138e-05,
+      "loss": 0.7294,
+      "step": 3773
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1937533364993143e-05,
+      "loss": 0.8355,
+      "step": 3774
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.188712137116012e-05,
+      "loss": 0.8269,
+      "step": 3775
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1836725282547585e-05,
+      "loss": 0.842,
+      "step": 3776
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1786345115815546e-05,
+      "loss": 0.7521,
+      "step": 3777
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.173598088761874e-05,
+      "loss": 0.8307,
+      "step": 3778
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.168563261460665e-05,
+      "loss": 0.7789,
+      "step": 3779
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.163530031342347e-05,
+      "loss": 0.8187,
+      "step": 3780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1584984000708124e-05,
+      "loss": 0.7624,
+      "step": 3781
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.153468369309424e-05,
+      "loss": 0.8018,
+      "step": 3782
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.148439940721015e-05,
+      "loss": 0.8074,
+      "step": 3783
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1434131159678945e-05,
+      "loss": 0.7319,
+      "step": 3784
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.138387896711834e-05,
+      "loss": 0.7919,
+      "step": 3785
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.133364284614077e-05,
+      "loss": 0.3128,
+      "step": 3786
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.128342281335338e-05,
+      "loss": 0.7847,
+      "step": 3787
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.123321888535795e-05,
+      "loss": 0.7358,
+      "step": 3788
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1183031078750956e-05,
+      "loss": 0.8479,
+      "step": 3789
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.113285941012358e-05,
+      "loss": 0.7866,
+      "step": 3790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1082703896061646e-05,
+      "loss": 0.8146,
+      "step": 3791
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.103256455314562e-05,
+      "loss": 0.7698,
+      "step": 3792
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.098244139795063e-05,
+      "loss": 0.8306,
+      "step": 3793
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.093233444704641e-05,
+      "loss": 0.8932,
+      "step": 3794
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.088224371699748e-05,
+      "loss": 0.7902,
+      "step": 3795
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.083216922436284e-05,
+      "loss": 0.8053,
+      "step": 3796
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.078211098569619e-05,
+      "loss": 0.823,
+      "step": 3797
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.073206901754586e-05,
+      "loss": 0.7975,
+      "step": 3798
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0682043336454786e-05,
+      "loss": 0.7861,
+      "step": 3799
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.063203395896052e-05,
+      "loss": 0.7452,
+      "step": 3800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.058204090159525e-05,
+      "loss": 0.8264,
+      "step": 3801
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.053206418088572e-05,
+      "loss": 0.7923,
+      "step": 3802
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.048210381335332e-05,
+      "loss": 0.7642,
+      "step": 3803
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.043215981551398e-05,
+      "loss": 0.7565,
+      "step": 3804
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0382232203878324e-05,
+      "loss": 0.798,
+      "step": 3805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.033232099495144e-05,
+      "loss": 0.8361,
+      "step": 3806
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.028242620523306e-05,
+      "loss": 0.7827,
+      "step": 3807
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.023254785121746e-05,
+      "loss": 0.8262,
+      "step": 3808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0182685949393507e-05,
+      "loss": 0.8284,
+      "step": 3809
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0132840516244604e-05,
+      "loss": 0.7568,
+      "step": 3810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.008301156824873e-05,
+      "loss": 0.7928,
+      "step": 3811
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.003319912187838e-05,
+      "loss": 0.8202,
+      "step": 3812
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9983403193600664e-05,
+      "loss": 0.7801,
+      "step": 3813
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.993362379987716e-05,
+      "loss": 0.8092,
+      "step": 3814
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.988386095716398e-05,
+      "loss": 0.8474,
+      "step": 3815
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9834114681911846e-05,
+      "loss": 0.7556,
+      "step": 3816
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.978438499056593e-05,
+      "loss": 0.8956,
+      "step": 3817
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9734671899565955e-05,
+      "loss": 0.7344,
+      "step": 3818
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.968497542534611e-05,
+      "loss": 0.8331,
+      "step": 3819
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.963529558433514e-05,
+      "loss": 0.7955,
+      "step": 3820
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.958563239295627e-05,
+      "loss": 0.8407,
+      "step": 3821
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.953598586762722e-05,
+      "loss": 0.8211,
+      "step": 3822
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.948635602476023e-05,
+      "loss": 0.7574,
+      "step": 3823
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9436742880761964e-05,
+      "loss": 0.7391,
+      "step": 3824
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9387146452033626e-05,
+      "loss": 0.8437,
+      "step": 3825
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.933756675497082e-05,
+      "loss": 0.7378,
+      "step": 3826
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.928800380596377e-05,
+      "loss": 0.8186,
+      "step": 3827
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.923845762139699e-05,
+      "loss": 0.8591,
+      "step": 3828
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.918892821764954e-05,
+      "loss": 0.7685,
+      "step": 3829
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.913941561109493e-05,
+      "loss": 0.7419,
+      "step": 3830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.908991981810109e-05,
+      "loss": 0.3483,
+      "step": 3831
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.904044085503041e-05,
+      "loss": 0.8066,
+      "step": 3832
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.899097873823971e-05,
+      "loss": 0.8327,
+      "step": 3833
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.894153348408021e-05,
+      "loss": 0.826,
+      "step": 3834
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.889210510889766e-05,
+      "loss": 0.7903,
+      "step": 3835
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.884269362903212e-05,
+      "loss": 0.8417,
+      "step": 3836
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.879329906081812e-05,
+      "loss": 0.8109,
+      "step": 3837
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.874392142058456e-05,
+      "loss": 0.822,
+      "step": 3838
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8694560724654735e-05,
+      "loss": 0.7476,
+      "step": 3839
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8645216989346466e-05,
+      "loss": 0.8277,
+      "step": 3840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.859589023097181e-05,
+      "loss": 0.835,
+      "step": 3841
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8546580465837274e-05,
+      "loss": 0.833,
+      "step": 3842
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8497287710243765e-05,
+      "loss": 0.8034,
+      "step": 3843
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.844801198048654e-05,
+      "loss": 0.7657,
+      "step": 3844
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.839875329285524e-05,
+      "loss": 0.8154,
+      "step": 3845
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.834951166363385e-05,
+      "loss": 0.3171,
+      "step": 3846
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.830028710910073e-05,
+      "loss": 0.7784,
+      "step": 3847
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.825107964552864e-05,
+      "loss": 0.7592,
+      "step": 3848
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.820188928918463e-05,
+      "loss": 0.753,
+      "step": 3849
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.815271605633012e-05,
+      "loss": 0.8182,
+      "step": 3850
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.810355996322085e-05,
+      "loss": 0.3298,
+      "step": 3851
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8054421026106913e-05,
+      "loss": 0.8393,
+      "step": 3852
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.800529926123273e-05,
+      "loss": 0.7967,
+      "step": 3853
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7956194684837045e-05,
+      "loss": 0.7912,
+      "step": 3854
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7907107313152885e-05,
+      "loss": 0.8427,
+      "step": 3855
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.785803716240767e-05,
+      "loss": 0.8058,
+      "step": 3856
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.780898424882303e-05,
+      "loss": 0.7892,
+      "step": 3857
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.775994858861492e-05,
+      "loss": 0.8534,
+      "step": 3858
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.771093019799371e-05,
+      "loss": 0.3864,
+      "step": 3859
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7661929093163905e-05,
+      "loss": 0.8426,
+      "step": 3860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7612945290324374e-05,
+      "loss": 0.7985,
+      "step": 3861
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.756397880566823e-05,
+      "loss": 0.769,
+      "step": 3862
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7515029655382894e-05,
+      "loss": 0.8484,
+      "step": 3863
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7466097855650025e-05,
+      "loss": 0.8044,
+      "step": 3864
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7417183422645586e-05,
+      "loss": 0.8586,
+      "step": 3865
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7368286372539775e-05,
+      "loss": 0.8101,
+      "step": 3866
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7319406721497016e-05,
+      "loss": 0.8246,
+      "step": 3867
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.727054448567601e-05,
+      "loss": 0.798,
+      "step": 3868
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.722169968122975e-05,
+      "loss": 0.7646,
+      "step": 3869
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7172872324305395e-05,
+      "loss": 0.8006,
+      "step": 3870
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.712406243104435e-05,
+      "loss": 0.794,
+      "step": 3871
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7075270017582254e-05,
+      "loss": 0.8314,
+      "step": 3872
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.702649510004897e-05,
+      "loss": 0.759,
+      "step": 3873
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.697773769456859e-05,
+      "loss": 0.7691,
+      "step": 3874
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.692899781725939e-05,
+      "loss": 0.3588,
+      "step": 3875
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.688027548423386e-05,
+      "loss": 0.7614,
+      "step": 3876
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6831570711598707e-05,
+      "loss": 0.7706,
+      "step": 3877
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.678288351545478e-05,
+      "loss": 0.7882,
+      "step": 3878
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.673421391189723e-05,
+      "loss": 0.7815,
+      "step": 3879
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6685561917015276e-05,
+      "loss": 0.7995,
+      "step": 3880
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.663692754689237e-05,
+      "loss": 0.7809,
+      "step": 3881
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.658831081760614e-05,
+      "loss": 0.8124,
+      "step": 3882
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.653971174522831e-05,
+      "loss": 0.7518,
+      "step": 3883
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6491130345824906e-05,
+      "loss": 0.7928,
+      "step": 3884
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6442566635456e-05,
+      "loss": 0.8307,
+      "step": 3885
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.639402063017585e-05,
+      "loss": 0.7306,
+      "step": 3886
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.634549234603286e-05,
+      "loss": 0.3192,
+      "step": 3887
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.629698179906958e-05,
+      "loss": 0.8199,
+      "step": 3888
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.624848900532268e-05,
+      "loss": 0.8055,
+      "step": 3889
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6200013980822954e-05,
+      "loss": 0.7381,
+      "step": 3890
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.615155674159539e-05,
+      "loss": 0.7995,
+      "step": 3891
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.610311730365904e-05,
+      "loss": 0.8548,
+      "step": 3892
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.605469568302705e-05,
+      "loss": 0.7893,
+      "step": 3893
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.600629189570672e-05,
+      "loss": 0.7779,
+      "step": 3894
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5957905957699445e-05,
+      "loss": 0.3481,
+      "step": 3895
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.590953788500071e-05,
+      "loss": 0.7449,
+      "step": 3896
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5861187693600097e-05,
+      "loss": 0.8288,
+      "step": 3897
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5812855399481256e-05,
+      "loss": 0.3577,
+      "step": 3898
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5764541018621976e-05,
+      "loss": 0.8237,
+      "step": 3899
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.571624456699404e-05,
+      "loss": 0.8506,
+      "step": 3900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.566796606056343e-05,
+      "loss": 0.8314,
+      "step": 3901
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.561970551529008e-05,
+      "loss": 0.7985,
+      "step": 3902
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.557146294712802e-05,
+      "loss": 0.3811,
+      "step": 3903
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5523238372025356e-05,
+      "loss": 0.8581,
+      "step": 3904
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.547503180592423e-05,
+      "loss": 0.8447,
+      "step": 3905
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.542684326476082e-05,
+      "loss": 0.8086,
+      "step": 3906
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.537867276446538e-05,
+      "loss": 0.8068,
+      "step": 3907
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.533052032096217e-05,
+      "loss": 0.8176,
+      "step": 3908
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.528238595016948e-05,
+      "loss": 0.7895,
+      "step": 3909
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.523426966799965e-05,
+      "loss": 0.7682,
+      "step": 3910
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5186171490358986e-05,
+      "loss": 0.7933,
+      "step": 3911
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5138091433147925e-05,
+      "loss": 0.7907,
+      "step": 3912
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.509002951226079e-05,
+      "loss": 0.7684,
+      "step": 3913
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.504198574358596e-05,
+      "loss": 0.8306,
+      "step": 3914
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4993960143005806e-05,
+      "loss": 0.8046,
+      "step": 3915
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4945952726396714e-05,
+      "loss": 0.7878,
+      "step": 3916
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4897963509629016e-05,
+      "loss": 0.8225,
+      "step": 3917
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.484999250856706e-05,
+      "loss": 0.8289,
+      "step": 3918
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.480203973906918e-05,
+      "loss": 0.7642,
+      "step": 3919
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.475410521698764e-05,
+      "loss": 0.847,
+      "step": 3920
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.470618895816873e-05,
+      "loss": 0.7965,
+      "step": 3921
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.465829097845261e-05,
+      "loss": 0.8255,
+      "step": 3922
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.461041129367354e-05,
+      "loss": 0.7479,
+      "step": 3923
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4562549919659625e-05,
+      "loss": 0.8816,
+      "step": 3924
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.451470687223293e-05,
+      "loss": 0.8736,
+      "step": 3925
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4466882167209464e-05,
+      "loss": 0.3563,
+      "step": 3926
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.441907582039917e-05,
+      "loss": 0.7684,
+      "step": 3927
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4371287847606e-05,
+      "loss": 0.8443,
+      "step": 3928
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.432351826462773e-05,
+      "loss": 0.8256,
+      "step": 3929
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.427576708725609e-05,
+      "loss": 0.8026,
+      "step": 3930
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4228034331276724e-05,
+      "loss": 0.6958,
+      "step": 3931
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.418032001246917e-05,
+      "loss": 0.8065,
+      "step": 3932
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4132624146606947e-05,
+      "loss": 0.8671,
+      "step": 3933
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.408494674945739e-05,
+      "loss": 0.7637,
+      "step": 3934
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.403728783678176e-05,
+      "loss": 0.778,
+      "step": 3935
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3989647424335214e-05,
+      "loss": 0.8322,
+      "step": 3936
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.394202552786676e-05,
+      "loss": 0.824,
+      "step": 3937
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.389442216311933e-05,
+      "loss": 0.8307,
+      "step": 3938
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.384683734582969e-05,
+      "loss": 0.336,
+      "step": 3939
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3799271091728525e-05,
+      "loss": 0.3286,
+      "step": 3940
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3751723416540317e-05,
+      "loss": 0.8199,
+      "step": 3941
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.370419433598345e-05,
+      "loss": 0.8473,
+      "step": 3942
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.365668386577012e-05,
+      "loss": 0.798,
+      "step": 3943
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.360919202160648e-05,
+      "loss": 0.786,
+      "step": 3944
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.356171881919241e-05,
+      "loss": 0.8173,
+      "step": 3945
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.351426427422165e-05,
+      "loss": 0.7845,
+      "step": 3946
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.346682840238181e-05,
+      "loss": 0.8358,
+      "step": 3947
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.341941121935429e-05,
+      "loss": 0.8538,
+      "step": 3948
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.337201274081434e-05,
+      "loss": 0.804,
+      "step": 3949
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.332463298243099e-05,
+      "loss": 0.8814,
+      "step": 3950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.327727195986713e-05,
+      "loss": 0.7874,
+      "step": 3951
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3229929688779414e-05,
+      "loss": 0.8158,
+      "step": 3952
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.318260618481833e-05,
+      "loss": 0.7993,
+      "step": 3953
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.313530146362809e-05,
+      "loss": 0.7837,
+      "step": 3954
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3088015540846826e-05,
+      "loss": 0.7817,
+      "step": 3955
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.304074843210637e-05,
+      "loss": 0.829,
+      "step": 3956
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.299350015303233e-05,
+      "loss": 0.8583,
+      "step": 3957
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.29462707192441e-05,
+      "loss": 0.8449,
+      "step": 3958
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.289906014635487e-05,
+      "loss": 0.822,
+      "step": 3959
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.285186844997154e-05,
+      "loss": 0.7757,
+      "step": 3960
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2804695645694835e-05,
+      "loss": 0.8362,
+      "step": 3961
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.275754174911921e-05,
+      "loss": 0.7831,
+      "step": 3962
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2710406775832835e-05,
+      "loss": 0.8044,
+      "step": 3963
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.266329074141764e-05,
+      "loss": 0.8047,
+      "step": 3964
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.261619366144938e-05,
+      "loss": 0.7757,
+      "step": 3965
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.256911555149742e-05,
+      "loss": 0.8403,
+      "step": 3966
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.252205642712491e-05,
+      "loss": 0.8418,
+      "step": 3967
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.247501630388873e-05,
+      "loss": 0.3241,
+      "step": 3968
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.242799519733944e-05,
+      "loss": 0.8591,
+      "step": 3969
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2380993123021385e-05,
+      "loss": 0.8023,
+      "step": 3970
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.23340100964725e-05,
+      "loss": 0.8657,
+      "step": 3971
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2287046133224584e-05,
+      "loss": 0.8312,
+      "step": 3972
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.224010124880301e-05,
+      "loss": 0.8688,
+      "step": 3973
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.219317545872689e-05,
+      "loss": 0.7666,
+      "step": 3974
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2146268778509004e-05,
+      "loss": 0.8116,
+      "step": 3975
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.209938122365579e-05,
+      "loss": 0.7553,
+      "step": 3976
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2052512809667475e-05,
+      "loss": 0.7807,
+      "step": 3977
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.200566355203784e-05,
+      "loss": 0.8019,
+      "step": 3978
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.195883346625439e-05,
+      "loss": 0.8213,
+      "step": 3979
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.191202256779827e-05,
+      "loss": 0.7818,
+      "step": 3980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.186523087214429e-05,
+      "loss": 0.7713,
+      "step": 3981
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.181845839476091e-05,
+      "loss": 0.8538,
+      "step": 3982
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.177170515111024e-05,
+      "loss": 0.8061,
+      "step": 3983
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.172497115664803e-05,
+      "loss": 0.8707,
+      "step": 3984
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1678256426823666e-05,
+      "loss": 0.8503,
+      "step": 3985
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.163156097708014e-05,
+      "loss": 0.8519,
+      "step": 3986
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1584884822854167e-05,
+      "loss": 0.843,
+      "step": 3987
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.153822797957596e-05,
+      "loss": 0.7798,
+      "step": 3988
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.149159046266941e-05,
+      "loss": 0.8046,
+      "step": 3989
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.144497228755203e-05,
+      "loss": 0.8275,
+      "step": 3990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1398373469634896e-05,
+      "loss": 0.8401,
+      "step": 3991
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1351794024322724e-05,
+      "loss": 0.793,
+      "step": 3992
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.130523396701379e-05,
+      "loss": 0.8228,
+      "step": 3993
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1258693313099996e-05,
+      "loss": 0.7824,
+      "step": 3994
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.12121720779668e-05,
+      "loss": 0.741,
+      "step": 3995
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1165670276993254e-05,
+      "loss": 0.8165,
+      "step": 3996
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1119187925552016e-05,
+      "loss": 0.7388,
+      "step": 3997
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1072725039009275e-05,
+      "loss": 0.7225,
+      "step": 3998
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1026281632724774e-05,
+      "loss": 0.8381,
+      "step": 3999
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.097985772205186e-05,
+      "loss": 0.8212,
+      "step": 4000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.09334533223374e-05,
+      "loss": 0.3358,
+      "step": 4001
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.088706844892182e-05,
+      "loss": 0.8046,
+      "step": 4002
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.08407031171391e-05,
+      "loss": 0.3391,
+      "step": 4003
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.079435734231676e-05,
+      "loss": 0.7927,
+      "step": 4004
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0748031139775834e-05,
+      "loss": 0.7669,
+      "step": 4005
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.070172452483091e-05,
+      "loss": 0.8152,
+      "step": 4006
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0655437512790065e-05,
+      "loss": 0.8208,
+      "step": 4007
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0609170118954965e-05,
+      "loss": 0.7949,
+      "step": 4008
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.056292235862074e-05,
+      "loss": 0.8394,
+      "step": 4009
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.051669424707602e-05,
+      "loss": 0.3133,
+      "step": 4010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0470485799602975e-05,
+      "loss": 0.7385,
+      "step": 4011
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.042429703147723e-05,
+      "loss": 0.7669,
+      "step": 4012
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.037812795796796e-05,
+      "loss": 0.727,
+      "step": 4013
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.033197859433777e-05,
+      "loss": 0.8548,
+      "step": 4014
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0285848955842763e-05,
+      "loss": 0.8503,
+      "step": 4015
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0239739057732614e-05,
+      "loss": 0.793,
+      "step": 4016
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.019364891525034e-05,
+      "loss": 0.7766,
+      "step": 4017
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.014757854363249e-05,
+      "loss": 0.8119,
+      "step": 4018
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0101527958109076e-05,
+      "loss": 0.7846,
+      "step": 4019
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.005549717390352e-05,
+      "loss": 0.7677,
+      "step": 4020
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.000948620623283e-05,
+      "loss": 0.7881,
+      "step": 4021
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.996349507030731e-05,
+      "loss": 0.789,
+      "step": 4022
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.991752378133078e-05,
+      "loss": 0.8153,
+      "step": 4023
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.987157235450051e-05,
+      "loss": 0.778,
+      "step": 4024
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982564080500716e-05,
+      "loss": 0.8183,
+      "step": 4025
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.977972914803486e-05,
+      "loss": 0.8163,
+      "step": 4026
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.973383739876114e-05,
+      "loss": 0.785,
+      "step": 4027
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9687965572356935e-05,
+      "loss": 0.8948,
+      "step": 4028
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.964211368398666e-05,
+      "loss": 0.8046,
+      "step": 4029
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9596281748808086e-05,
+      "loss": 0.824,
+      "step": 4030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9550469781972376e-05,
+      "loss": 0.7976,
+      "step": 4031
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.950467779862411e-05,
+      "loss": 0.7669,
+      "step": 4032
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.945890581390129e-05,
+      "loss": 0.7898,
+      "step": 4033
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9413153842935255e-05,
+      "loss": 0.7191,
+      "step": 4034
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9367421900850774e-05,
+      "loss": 0.7574,
+      "step": 4035
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9321710002765956e-05,
+      "loss": 0.3779,
+      "step": 4036
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.92760181637923e-05,
+      "loss": 0.8118,
+      "step": 4037
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.92303463990347e-05,
+      "loss": 0.7969,
+      "step": 4038
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.918469472359134e-05,
+      "loss": 0.775,
+      "step": 4039
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9139063152553864e-05,
+      "loss": 0.8081,
+      "step": 4040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9093451701007225e-05,
+      "loss": 0.7164,
+      "step": 4041
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9047860384029675e-05,
+      "loss": 0.79,
+      "step": 4042
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.900228921669289e-05,
+      "loss": 0.7842,
+      "step": 4043
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.895673821406183e-05,
+      "loss": 0.7836,
+      "step": 4044
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8911207391194794e-05,
+      "loss": 0.7721,
+      "step": 4045
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8865696763143447e-05,
+      "loss": 0.7871,
+      "step": 4046
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8820206344952746e-05,
+      "loss": 0.7792,
+      "step": 4047
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.877473615166097e-05,
+      "loss": 0.3632,
+      "step": 4048
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.872928619829972e-05,
+      "loss": 0.7483,
+      "step": 4049
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.868385649989388e-05,
+      "loss": 0.8183,
+      "step": 4050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8638447071461704e-05,
+      "loss": 0.7968,
+      "step": 4051
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.859305792801469e-05,
+      "loss": 0.8146,
+      "step": 4052
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.854768908455764e-05,
+      "loss": 0.7955,
+      "step": 4053
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.850234055608863e-05,
+      "loss": 0.819,
+      "step": 4054
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.845701235759906e-05,
+      "loss": 0.8089,
+      "step": 4055
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.841170450407358e-05,
+      "loss": 0.7569,
+      "step": 4056
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8366417010490116e-05,
+      "loss": 0.8365,
+      "step": 4057
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.832114989181988e-05,
+      "loss": 0.8273,
+      "step": 4058
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.827590316302732e-05,
+      "loss": 0.3331,
+      "step": 4059
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8230676839070134e-05,
+      "loss": 0.7866,
+      "step": 4060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.818547093489937e-05,
+      "loss": 0.8466,
+      "step": 4061
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.814028546545924e-05,
+      "loss": 0.7791,
+      "step": 4062
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.809512044568719e-05,
+      "loss": 0.8034,
+      "step": 4063
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804997589051394e-05,
+      "loss": 0.8031,
+      "step": 4064
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.800485181486343e-05,
+      "loss": 0.7538,
+      "step": 4065
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.795974823365287e-05,
+      "loss": 0.7944,
+      "step": 4066
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.791466516179265e-05,
+      "loss": 0.8319,
+      "step": 4067
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7869602614186395e-05,
+      "loss": 0.824,
+      "step": 4068
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.782456060573092e-05,
+      "loss": 0.8086,
+      "step": 4069
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.77795391513163e-05,
+      "loss": 0.7711,
+      "step": 4070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.773453826582574e-05,
+      "loss": 0.8475,
+      "step": 4071
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.768955796413577e-05,
+      "loss": 0.8371,
+      "step": 4072
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.764459826111598e-05,
+      "loss": 0.8336,
+      "step": 4073
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.759965917162925e-05,
+      "loss": 0.7877,
+      "step": 4074
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.755474071053158e-05,
+      "loss": 0.8063,
+      "step": 4075
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.750984289267217e-05,
+      "loss": 0.7902,
+      "step": 4076
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.746496573289342e-05,
+      "loss": 0.3333,
+      "step": 4077
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7420109246030866e-05,
+      "loss": 0.8487,
+      "step": 4078
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.737527344691324e-05,
+      "loss": 0.3446,
+      "step": 4079
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.733045835036241e-05,
+      "loss": 0.8156,
+      "step": 4080
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.728566397119341e-05,
+      "loss": 0.7499,
+      "step": 4081
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.724089032421441e-05,
+      "loss": 0.8211,
+      "step": 4082
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7196137424226784e-05,
+      "loss": 0.8387,
+      "step": 4083
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7151405286025e-05,
+      "loss": 0.3503,
+      "step": 4084
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.710669392439665e-05,
+      "loss": 0.8291,
+      "step": 4085
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.706200335412248e-05,
+      "loss": 0.7722,
+      "step": 4086
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.701733358997636e-05,
+      "loss": 0.8318,
+      "step": 4087
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6972684646725283e-05,
+      "loss": 0.777,
+      "step": 4088
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6928056539129365e-05,
+      "loss": 0.8192,
+      "step": 4089
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.688344928194181e-05,
+      "loss": 0.7856,
+      "step": 4090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6838862889908955e-05,
+      "loss": 0.876,
+      "step": 4091
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6794297377770196e-05,
+      "loss": 0.8029,
+      "step": 4092
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6749752760258115e-05,
+      "loss": 0.7977,
+      "step": 4093
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.670522905209832e-05,
+      "loss": 0.7835,
+      "step": 4094
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.66607262680095e-05,
+      "loss": 0.8491,
+      "step": 4095
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.661624442270346e-05,
+      "loss": 0.7247,
+      "step": 4096
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6571783530885063e-05,
+      "loss": 0.8226,
+      "step": 4097
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.652734360725224e-05,
+      "loss": 0.7479,
+      "step": 4098
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.648292466649601e-05,
+      "loss": 0.8313,
+      "step": 4099
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6438526723300446e-05,
+      "loss": 0.7186,
+      "step": 4100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.639414979234269e-05,
+      "loss": 0.7918,
+      "step": 4101
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6349793888292915e-05,
+      "loss": 0.7651,
+      "step": 4102
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6305459025814326e-05,
+      "loss": 0.7802,
+      "step": 4103
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.626114521956327e-05,
+      "loss": 0.8006,
+      "step": 4104
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6216852484189035e-05,
+      "loss": 0.3873,
+      "step": 4105
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.617258083433396e-05,
+      "loss": 0.7987,
+      "step": 4106
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.612833028463344e-05,
+      "loss": 0.8169,
+      "step": 4107
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6084100849715876e-05,
+      "loss": 0.8103,
+      "step": 4108
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6039892544202655e-05,
+      "loss": 0.8538,
+      "step": 4109
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.59957053827083e-05,
+      "loss": 0.816,
+      "step": 4110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5951539379840226e-05,
+      "loss": 0.7481,
+      "step": 4111
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.590739455019888e-05,
+      "loss": 0.7785,
+      "step": 4112
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.586327090837772e-05,
+      "loss": 0.7895,
+      "step": 4113
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.581916846896318e-05,
+      "loss": 0.7821,
+      "step": 4114
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5775087246534754e-05,
+      "loss": 0.7337,
+      "step": 4115
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.573102725566485e-05,
+      "loss": 0.7801,
+      "step": 4116
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5686988510918876e-05,
+      "loss": 0.8167,
+      "step": 4117
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.564297102685522e-05,
+      "loss": 0.7484,
+      "step": 4118
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5598974818025245e-05,
+      "loss": 0.8676,
+      "step": 4119
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.555499989897326e-05,
+      "loss": 0.3347,
+      "step": 4120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.551104628423658e-05,
+      "loss": 0.8268,
+      "step": 4121
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.546711398834543e-05,
+      "loss": 0.8002,
+      "step": 4122
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.542320302582302e-05,
+      "loss": 0.8183,
+      "step": 4123
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5379313411185453e-05,
+      "loss": 0.7946,
+      "step": 4124
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.533544515894188e-05,
+      "loss": 0.7464,
+      "step": 4125
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5291598283594316e-05,
+      "loss": 0.7617,
+      "step": 4126
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.52477727996377e-05,
+      "loss": 0.8625,
+      "step": 4127
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.520396872155992e-05,
+      "loss": 0.8155,
+      "step": 4128
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5160186063841806e-05,
+      "loss": 0.8376,
+      "step": 4129
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5116424840957065e-05,
+      "loss": 0.7913,
+      "step": 4130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.507268506737235e-05,
+      "loss": 0.328,
+      "step": 4131
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.502896675754722e-05,
+      "loss": 0.8223,
+      "step": 4132
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.498526992593413e-05,
+      "loss": 0.7369,
+      "step": 4133
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.494159458697843e-05,
+      "loss": 0.7737,
+      "step": 4134
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.489794075511834e-05,
+      "loss": 0.7885,
+      "step": 4135
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.485430844478509e-05,
+      "loss": 0.8007,
+      "step": 4136
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.481069767040266e-05,
+      "loss": 0.7866,
+      "step": 4137
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.476710844638795e-05,
+      "loss": 0.7865,
+      "step": 4138
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.472354078715076e-05,
+      "loss": 0.805,
+      "step": 4139
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.467999470709373e-05,
+      "loss": 0.8099,
+      "step": 4140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.463647022061239e-05,
+      "loss": 0.7723,
+      "step": 4141
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.459296734209514e-05,
+      "loss": 0.8338,
+      "step": 4142
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.45494860859232e-05,
+      "loss": 0.8052,
+      "step": 4143
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.450602646647066e-05,
+      "loss": 0.7907,
+      "step": 4144
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.446258849810448e-05,
+      "loss": 0.7768,
+      "step": 4145
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.441917219518438e-05,
+      "loss": 0.7387,
+      "step": 4146
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4375777572063075e-05,
+      "loss": 0.7949,
+      "step": 4147
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.433240464308597e-05,
+      "loss": 0.8344,
+      "step": 4148
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4289053422591346e-05,
+      "loss": 0.7829,
+      "step": 4149
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4245723924910315e-05,
+      "loss": 0.7883,
+      "step": 4150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4202416164366805e-05,
+      "loss": 0.8304,
+      "step": 4151
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.415913015527753e-05,
+      "loss": 0.8374,
+      "step": 4152
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4115865911952026e-05,
+      "loss": 0.8405,
+      "step": 4153
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.407262344869272e-05,
+      "loss": 0.7241,
+      "step": 4154
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.402940277979471e-05,
+      "loss": 0.8214,
+      "step": 4155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3986203919545945e-05,
+      "loss": 0.8136,
+      "step": 4156
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.394302688222717e-05,
+      "loss": 0.8305,
+      "step": 4157
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.389987168211187e-05,
+      "loss": 0.7617,
+      "step": 4158
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.385673833346643e-05,
+      "loss": 0.8392,
+      "step": 4159
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.381362685054987e-05,
+      "loss": 0.8124,
+      "step": 4160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3770537247614076e-05,
+      "loss": 0.3539,
+      "step": 4161
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3727469538903646e-05,
+      "loss": 0.802,
+      "step": 4162
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.368442373865596e-05,
+      "loss": 0.7394,
+      "step": 4163
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3641399861101165e-05,
+      "loss": 0.8207,
+      "step": 4164
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3598397920462144e-05,
+      "loss": 0.7901,
+      "step": 4165
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.355541793095456e-05,
+      "loss": 0.8158,
+      "step": 4166
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.351245990678673e-05,
+      "loss": 0.7858,
+      "step": 4167
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3469523862159856e-05,
+      "loss": 0.7616,
+      "step": 4168
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3426609811267775e-05,
+      "loss": 0.8546,
+      "step": 4169
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.338371776829705e-05,
+      "loss": 0.8291,
+      "step": 4170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3340847747426995e-05,
+      "loss": 0.8281,
+      "step": 4171
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3297999762829655e-05,
+      "loss": 0.7914,
+      "step": 4172
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.325517382866976e-05,
+      "loss": 0.8218,
+      "step": 4173
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3212369959104774e-05,
+      "loss": 0.8047,
+      "step": 4174
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.316958816828485e-05,
+      "loss": 0.835,
+      "step": 4175
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.312682847035284e-05,
+      "loss": 0.79,
+      "step": 4176
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.308409087944433e-05,
+      "loss": 0.7994,
+      "step": 4177
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3041375409687526e-05,
+      "loss": 0.763,
+      "step": 4178
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2998682075203424e-05,
+      "loss": 0.7519,
+      "step": 4179
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.295601089010562e-05,
+      "loss": 0.8023,
+      "step": 4180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.29133618685004e-05,
+      "loss": 0.8033,
+      "step": 4181
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.287073502448675e-05,
+      "loss": 0.844,
+      "step": 4182
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.282813037215632e-05,
+      "loss": 0.8154,
+      "step": 4183
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.278554792559337e-05,
+      "loss": 0.7645,
+      "step": 4184
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.274298769887492e-05,
+      "loss": 0.7845,
+      "step": 4185
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2700449706070534e-05,
+      "loss": 0.7504,
+      "step": 4186
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.26579339612425e-05,
+      "loss": 0.7997,
+      "step": 4187
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2615440478445715e-05,
+      "loss": 0.7815,
+      "step": 4188
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.257296927172778e-05,
+      "loss": 0.8265,
+      "step": 4189
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2530520355128854e-05,
+      "loss": 0.816,
+      "step": 4190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.248809374268176e-05,
+      "loss": 0.7589,
+      "step": 4191
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2445689448411934e-05,
+      "loss": 0.7893,
+      "step": 4192
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.240330748633745e-05,
+      "loss": 0.7825,
+      "step": 4193
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.236094787046901e-05,
+      "loss": 0.782,
+      "step": 4194
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.23186106148099e-05,
+      "loss": 0.822,
+      "step": 4195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2276295733356024e-05,
+      "loss": 0.8084,
+      "step": 4196
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2234003240095865e-05,
+      "loss": 0.7974,
+      "step": 4197
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2191733149010594e-05,
+      "loss": 0.724,
+      "step": 4198
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.214948547407389e-05,
+      "loss": 0.8089,
+      "step": 4199
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2107260229252036e-05,
+      "loss": 0.7942,
+      "step": 4200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.206505742850392e-05,
+      "loss": 0.8062,
+      "step": 4201
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.202287708578097e-05,
+      "loss": 0.7967,
+      "step": 4202
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.198071921502728e-05,
+      "loss": 0.8448,
+      "step": 4203
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.193858383017942e-05,
+      "loss": 0.8113,
+      "step": 4204
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.189647094516656e-05,
+      "loss": 0.7706,
+      "step": 4205
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.185438057391045e-05,
+      "loss": 0.7036,
+      "step": 4206
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1812312730325366e-05,
+      "loss": 0.8198,
+      "step": 4207
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1770267428318154e-05,
+      "loss": 0.8107,
+      "step": 4208
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.172824468178822e-05,
+      "loss": 0.7436,
+      "step": 4209
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.168624450462746e-05,
+      "loss": 0.733,
+      "step": 4210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1644266910720424e-05,
+      "loss": 0.3337,
+      "step": 4211
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.160231191394407e-05,
+      "loss": 0.844,
+      "step": 4212
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.156037952816795e-05,
+      "loss": 0.771,
+      "step": 4213
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.151846976725412e-05,
+      "loss": 0.8364,
+      "step": 4214
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1476582645057176e-05,
+      "loss": 0.7488,
+      "step": 4215
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.143471817542422e-05,
+      "loss": 0.7889,
+      "step": 4216
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1392876372194846e-05,
+      "loss": 0.7835,
+      "step": 4217
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.13510572492012e-05,
+      "loss": 0.752,
+      "step": 4218
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.130926082026789e-05,
+      "loss": 0.8155,
+      "step": 4219
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1267487099212e-05,
+      "loss": 0.7962,
+      "step": 4220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.122573609984323e-05,
+      "loss": 0.7808,
+      "step": 4221
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.118400783596361e-05,
+      "loss": 0.817,
+      "step": 4222
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.114230232136777e-05,
+      "loss": 0.7477,
+      "step": 4223
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.110061956984275e-05,
+      "loss": 0.7913,
+      "step": 4224
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.105895959516808e-05,
+      "loss": 0.8207,
+      "step": 4225
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.10173224111158e-05,
+      "loss": 0.7877,
+      "step": 4226
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0975708031450365e-05,
+      "loss": 0.8351,
+      "step": 4227
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.093411646992873e-05,
+      "loss": 0.8142,
+      "step": 4228
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.089254774030027e-05,
+      "loss": 0.3667,
+      "step": 4229
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.085100185630685e-05,
+      "loss": 0.6959,
+      "step": 4230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0809478831682715e-05,
+      "loss": 0.8358,
+      "step": 4231
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0767978680154684e-05,
+      "loss": 0.8069,
+      "step": 4232
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0726501415441885e-05,
+      "loss": 0.7933,
+      "step": 4233
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0685047051255946e-05,
+      "loss": 0.8555,
+      "step": 4234
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.064361560130089e-05,
+      "loss": 0.7822,
+      "step": 4235
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.060220707927319e-05,
+      "loss": 0.8217,
+      "step": 4236
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.056082149886173e-05,
+      "loss": 0.8154,
+      "step": 4237
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.051945887374782e-05,
+      "loss": 0.8052,
+      "step": 4238
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.047811921760515e-05,
+      "loss": 0.8298,
+      "step": 4239
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0436802544099862e-05,
+      "loss": 0.7684,
+      "step": 4240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.039550886689043e-05,
+      "loss": 0.8189,
+      "step": 4241
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.035423819962785e-05,
+      "loss": 0.3194,
+      "step": 4242
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.03129905559554e-05,
+      "loss": 0.7944,
+      "step": 4243
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.027176594950878e-05,
+      "loss": 0.802,
+      "step": 4244
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0230564393916074e-05,
+      "loss": 0.8174,
+      "step": 4245
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0189385902797705e-05,
+      "loss": 0.7689,
+      "step": 4246
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0148230489766606e-05,
+      "loss": 0.8125,
+      "step": 4247
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0107098168427937e-05,
+      "loss": 0.8137,
+      "step": 4248
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0065988952379286e-05,
+      "loss": 0.7844,
+      "step": 4249
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.002490285521059e-05,
+      "loss": 0.7972,
+      "step": 4250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.998383989050414e-05,
+      "loss": 0.7969,
+      "step": 4251
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9942800071834554e-05,
+      "loss": 0.8189,
+      "step": 4252
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9901783412768912e-05,
+      "loss": 0.7509,
+      "step": 4253
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9860789926866504e-05,
+      "loss": 0.7756,
+      "step": 4254
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9819819627679024e-05,
+      "loss": 0.3414,
+      "step": 4255
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.977887252875049e-05,
+      "loss": 0.7746,
+      "step": 4256
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9737948643617243e-05,
+      "loss": 0.8575,
+      "step": 4257
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9697047985807958e-05,
+      "loss": 0.7907,
+      "step": 4258
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9656170568843635e-05,
+      "loss": 0.7816,
+      "step": 4259
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.961531640623757e-05,
+      "loss": 0.3295,
+      "step": 4260
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.957448551149541e-05,
+      "loss": 0.8206,
+      "step": 4261
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9533677898115063e-05,
+      "loss": 0.8227,
+      "step": 4262
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9492893579586754e-05,
+      "loss": 0.8323,
+      "step": 4263
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9452132569393077e-05,
+      "loss": 0.7615,
+      "step": 4264
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9411394881008826e-05,
+      "loss": 0.7884,
+      "step": 4265
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9370680527901116e-05,
+      "loss": 0.7802,
+      "step": 4266
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9329989523529366e-05,
+      "loss": 0.8032,
+      "step": 4267
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9289321881345254e-05,
+      "loss": 0.7953,
+      "step": 4268
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9248677614792753e-05,
+      "loss": 0.776,
+      "step": 4269
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9208056737308074e-05,
+      "loss": 0.8157,
+      "step": 4270
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9167459262319764e-05,
+      "loss": 0.7642,
+      "step": 4271
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9126885203248554e-05,
+      "loss": 0.7798,
+      "step": 4272
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.908633457350749e-05,
+      "loss": 0.7354,
+      "step": 4273
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.904580738650181e-05,
+      "loss": 0.8245,
+      "step": 4274
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9005303655629122e-05,
+      "loss": 0.7852,
+      "step": 4275
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8964823394279174e-05,
+      "loss": 0.7408,
+      "step": 4276
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8924366615833977e-05,
+      "loss": 0.8134,
+      "step": 4277
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.888393333366778e-05,
+      "loss": 0.3301,
+      "step": 4278
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.884352356114709e-05,
+      "loss": 0.7623,
+      "step": 4279
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.880313731163061e-05,
+      "loss": 0.7883,
+      "step": 4280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.876277459846929e-05,
+      "loss": 0.8135,
+      "step": 4281
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.872243543500629e-05,
+      "loss": 0.7832,
+      "step": 4282
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8682119834576982e-05,
+      "loss": 0.363,
+      "step": 4283
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.864182781050895e-05,
+      "loss": 0.7719,
+      "step": 4284
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8601559376121957e-05,
+      "loss": 0.812,
+      "step": 4285
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.856131454472807e-05,
+      "loss": 0.8545,
+      "step": 4286
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.852109332963142e-05,
+      "loss": 0.8497,
+      "step": 4287
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8480895744128422e-05,
+      "loss": 0.7872,
+      "step": 4288
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.844072180150762e-05,
+      "loss": 0.3373,
+      "step": 4289
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.840057151504979e-05,
+      "loss": 0.7947,
+      "step": 4290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8360444898027838e-05,
+      "loss": 0.8071,
+      "step": 4291
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.832034196370693e-05,
+      "loss": 0.752,
+      "step": 4292
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8280262725344308e-05,
+      "loss": 0.7636,
+      "step": 4293
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.824020719618944e-05,
+      "loss": 0.8459,
+      "step": 4294
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.820017538948393e-05,
+      "loss": 0.8165,
+      "step": 4295
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8160167318461506e-05,
+      "loss": 0.7804,
+      "step": 4296
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8120182996348166e-05,
+      "loss": 0.7908,
+      "step": 4297
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8080222436361934e-05,
+      "loss": 0.7695,
+      "step": 4298
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.804028565171305e-05,
+      "loss": 0.8014,
+      "step": 4299
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8000372655603847e-05,
+      "loss": 0.8409,
+      "step": 4300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7960483461228838e-05,
+      "loss": 0.7406,
+      "step": 4301
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7920618081774618e-05,
+      "loss": 0.7971,
+      "step": 4302
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7880776530419973e-05,
+      "loss": 0.7826,
+      "step": 4303
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.784095882033575e-05,
+      "loss": 0.7855,
+      "step": 4304
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7801164964684955e-05,
+      "loss": 0.7657,
+      "step": 4305
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7761394976622658e-05,
+      "loss": 0.7863,
+      "step": 4306
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7721648869296135e-05,
+      "loss": 0.7934,
+      "step": 4307
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.768192665584468e-05,
+      "loss": 0.7794,
+      "step": 4308
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7642228349399723e-05,
+      "loss": 0.7793,
+      "step": 4309
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7602553963084776e-05,
+      "loss": 0.7517,
+      "step": 4310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.756290351001546e-05,
+      "loss": 0.7672,
+      "step": 4311
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7523277003299463e-05,
+      "loss": 0.7363,
+      "step": 4312
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7483674456036567e-05,
+      "loss": 0.807,
+      "step": 4313
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7444095881318656e-05,
+      "loss": 0.7949,
+      "step": 4314
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7404541292229647e-05,
+      "loss": 0.8275,
+      "step": 4315
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.736501070184556e-05,
+      "loss": 0.3588,
+      "step": 4316
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7325504123234446e-05,
+      "loss": 0.7942,
+      "step": 4317
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.728602156945649e-05,
+      "loss": 0.8387,
+      "step": 4318
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7246563053563866e-05,
+      "loss": 0.7914,
+      "step": 4319
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.720712858860083e-05,
+      "loss": 0.7867,
+      "step": 4320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7167718187603663e-05,
+      "loss": 0.7757,
+      "step": 4321
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.712833186360072e-05,
+      "loss": 0.7402,
+      "step": 4322
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7088969629612393e-05,
+      "loss": 0.7793,
+      "step": 4323
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7049631498651085e-05,
+      "loss": 0.7719,
+      "step": 4324
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.701031748372126e-05,
+      "loss": 0.7914,
+      "step": 4325
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.69710275978194e-05,
+      "loss": 0.8036,
+      "step": 4326
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6931761853933967e-05,
+      "loss": 0.7861,
+      "step": 4327
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6892520265045552e-05,
+      "loss": 0.7685,
+      "step": 4328
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.685330284412666e-05,
+      "loss": 0.7581,
+      "step": 4329
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6814109604141848e-05,
+      "loss": 0.7487,
+      "step": 4330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6774940558047657e-05,
+      "loss": 0.7794,
+      "step": 4331
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6735795718792646e-05,
+      "loss": 0.7627,
+      "step": 4332
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.669667509931737e-05,
+      "loss": 0.8042,
+      "step": 4333
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.665757871255439e-05,
+      "loss": 0.3346,
+      "step": 4334
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.661850657142819e-05,
+      "loss": 0.8025,
+      "step": 4335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6579458688855362e-05,
+      "loss": 0.7753,
+      "step": 4336
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6540435077744387e-05,
+      "loss": 0.7137,
+      "step": 4337
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6501435750995727e-05,
+      "loss": 0.7828,
+      "step": 4338
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6462460721501834e-05,
+      "loss": 0.7741,
+      "step": 4339
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6423510002147113e-05,
+      "loss": 0.76,
+      "step": 4340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6384583605807988e-05,
+      "loss": 0.7477,
+      "step": 4341
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6345681545352773e-05,
+      "loss": 0.8095,
+      "step": 4342
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6306803833641768e-05,
+      "loss": 0.7718,
+      "step": 4343
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6267950483527216e-05,
+      "loss": 0.7914,
+      "step": 4344
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6229121507853305e-05,
+      "loss": 0.7709,
+      "step": 4345
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.619031691945618e-05,
+      "loss": 0.7027,
+      "step": 4346
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6151536731163916e-05,
+      "loss": 0.7484,
+      "step": 4347
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.611278095579651e-05,
+      "loss": 0.7799,
+      "step": 4348
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6074049606165883e-05,
+      "loss": 0.851,
+      "step": 4349
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6035342695075937e-05,
+      "loss": 0.7809,
+      "step": 4350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5996660235322445e-05,
+      "loss": 0.8475,
+      "step": 4351
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5958002239693092e-05,
+      "loss": 0.8244,
+      "step": 4352
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5919368720967496e-05,
+      "loss": 0.7915,
+      "step": 4353
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.588075969191718e-05,
+      "loss": 0.8355,
+      "step": 4354
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.584217516530557e-05,
+      "loss": 0.8433,
+      "step": 4355
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5803615153887983e-05,
+      "loss": 0.7888,
+      "step": 4356
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.576507967041164e-05,
+      "loss": 0.8059,
+      "step": 4357
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5726568727615662e-05,
+      "loss": 0.7937,
+      "step": 4358
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5688082338231023e-05,
+      "loss": 0.7153,
+      "step": 4359
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5649620514980644e-05,
+      "loss": 0.7666,
+      "step": 4360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5611183270579274e-05,
+      "loss": 0.7746,
+      "step": 4361
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5572770617733544e-05,
+      "loss": 0.8096,
+      "step": 4362
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5534382569141967e-05,
+      "loss": 0.3264,
+      "step": 4363
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5496019137494908e-05,
+      "loss": 0.8347,
+      "step": 4364
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.54576803354746e-05,
+      "loss": 0.8076,
+      "step": 4365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5419366175755145e-05,
+      "loss": 0.7556,
+      "step": 4366
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.538107667100249e-05,
+      "loss": 0.812,
+      "step": 4367
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5342811833874423e-05,
+      "loss": 0.8562,
+      "step": 4368
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5304571677020582e-05,
+      "loss": 0.8517,
+      "step": 4369
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5266356213082433e-05,
+      "loss": 0.3369,
+      "step": 4370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5228165454693352e-05,
+      "loss": 0.8309,
+      "step": 4371
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.518999941447846e-05,
+      "loss": 0.735,
+      "step": 4372
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.515185810505474e-05,
+      "loss": 0.7748,
+      "step": 4373
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5113741539030987e-05,
+      "loss": 0.8801,
+      "step": 4374
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5075649729007846e-05,
+      "loss": 0.7941,
+      "step": 4375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.503758268757773e-05,
+      "loss": 0.8112,
+      "step": 4376
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.499954042732493e-05,
+      "loss": 0.7905,
+      "step": 4377
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.496152296082548e-05,
+      "loss": 0.7993,
+      "step": 4378
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4923530300647224e-05,
+      "loss": 0.3366,
+      "step": 4379
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4885562459349888e-05,
+      "loss": 0.7793,
+      "step": 4380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4847619449484906e-05,
+      "loss": 0.8284,
+      "step": 4381
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.480970128359552e-05,
+      "loss": 0.7836,
+      "step": 4382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.477180797421679e-05,
+      "loss": 0.811,
+      "step": 4383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4733939533875472e-05,
+      "loss": 0.7824,
+      "step": 4384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.469609597509026e-05,
+      "loss": 0.8352,
+      "step": 4385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.465827731037147e-05,
+      "loss": 0.8452,
+      "step": 4386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4620483552221273e-05,
+      "loss": 0.3226,
+      "step": 4387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.458271471313357e-05,
+      "loss": 0.7704,
+      "step": 4388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4544970805594026e-05,
+      "loss": 0.7981,
+      "step": 4389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4507251842080092e-05,
+      "loss": 0.8278,
+      "step": 4390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.446955783506091e-05,
+      "loss": 0.7442,
+      "step": 4391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.443188879699747e-05,
+      "loss": 0.7959,
+      "step": 4392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4394244740342444e-05,
+      "loss": 0.7912,
+      "step": 4393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4356625677540233e-05,
+      "loss": 0.8195,
+      "step": 4394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4319031621027024e-05,
+      "loss": 0.783,
+      "step": 4395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4281462583230686e-05,
+      "loss": 0.7625,
+      "step": 4396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4243918576570858e-05,
+      "loss": 0.7875,
+      "step": 4397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4206399613458875e-05,
+      "loss": 0.7949,
+      "step": 4398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.416890570629782e-05,
+      "loss": 0.3218,
+      "step": 4399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.413143686748247e-05,
+      "loss": 0.8179,
+      "step": 4400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4093993109399327e-05,
+      "loss": 0.7842,
+      "step": 4401
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.405657444442657e-05,
+      "loss": 0.3513,
+      "step": 4402
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.401918088493418e-05,
+      "loss": 0.7893,
+      "step": 4403
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3981812443283723e-05,
+      "loss": 0.7801,
+      "step": 4404
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3944469131828528e-05,
+      "loss": 0.7854,
+      "step": 4405
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3907150962913584e-05,
+      "loss": 0.8221,
+      "step": 4406
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3869857948875583e-05,
+      "loss": 0.8206,
+      "step": 4407
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3832590102042895e-05,
+      "loss": 0.7937,
+      "step": 4408
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3795347434735592e-05,
+      "loss": 0.8141,
+      "step": 4409
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3758129959265407e-05,
+      "loss": 0.7418,
+      "step": 4410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3720937687935717e-05,
+      "loss": 0.765,
+      "step": 4411
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3683770633041613e-05,
+      "loss": 0.7848,
+      "step": 4412
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3646628806869797e-05,
+      "loss": 0.8517,
+      "step": 4413
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3609512221698725e-05,
+      "loss": 0.7984,
+      "step": 4414
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.357242088979842e-05,
+      "loss": 0.7593,
+      "step": 4415
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3535354823430577e-05,
+      "loss": 0.7702,
+      "step": 4416
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3498314034848545e-05,
+      "loss": 0.7828,
+      "step": 4417
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3461298536297328e-05,
+      "loss": 0.7535,
+      "step": 4418
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3424308340013557e-05,
+      "loss": 0.709,
+      "step": 4419
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.33873434582255e-05,
+      "loss": 0.8173,
+      "step": 4420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.335040390315305e-05,
+      "loss": 0.7839,
+      "step": 4421
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.331348968700775e-05,
+      "loss": 0.7766,
+      "step": 4422
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3276600821992712e-05,
+      "loss": 0.3309,
+      "step": 4423
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3239737320302756e-05,
+      "loss": 0.8331,
+      "step": 4424
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3202899194124263e-05,
+      "loss": 0.7415,
+      "step": 4425
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3166086455635218e-05,
+      "loss": 0.8464,
+      "step": 4426
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.312929911700522e-05,
+      "loss": 0.8024,
+      "step": 4427
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3092537190395457e-05,
+      "loss": 0.8229,
+      "step": 4428
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3055800687958785e-05,
+      "loss": 0.8453,
+      "step": 4429
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3019089621839597e-05,
+      "loss": 0.7525,
+      "step": 4430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2982404004173863e-05,
+      "loss": 0.8065,
+      "step": 4431
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2945743847089174e-05,
+      "loss": 0.7579,
+      "step": 4432
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.29091091627047e-05,
+      "loss": 0.809,
+      "step": 4433
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2872499963131155e-05,
+      "loss": 0.8757,
+      "step": 4434
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.28359162604709e-05,
+      "loss": 0.8443,
+      "step": 4435
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.279935806681782e-05,
+      "loss": 0.7084,
+      "step": 4436
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2762825394257347e-05,
+      "loss": 0.8636,
+      "step": 4437
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.272631825486653e-05,
+      "loss": 0.7301,
+      "step": 4438
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2689836660713915e-05,
+      "loss": 0.3505,
+      "step": 4439
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2653380623859665e-05,
+      "loss": 0.8194,
+      "step": 4440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2616950156355455e-05,
+      "loss": 0.8227,
+      "step": 4441
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.258054527024451e-05,
+      "loss": 0.7604,
+      "step": 4442
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2544165977561627e-05,
+      "loss": 0.7685,
+      "step": 4443
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2507812290333097e-05,
+      "loss": 0.3236,
+      "step": 4444
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2471484220576767e-05,
+      "loss": 0.7201,
+      "step": 4445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.243518178030206e-05,
+      "loss": 0.7473,
+      "step": 4446
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.239890498150986e-05,
+      "loss": 0.8055,
+      "step": 4447
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2362653836192603e-05,
+      "loss": 0.3092,
+      "step": 4448
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.232642835633424e-05,
+      "loss": 0.7887,
+      "step": 4449
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2290228553910242e-05,
+      "loss": 0.8219,
+      "step": 4450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2254054440887583e-05,
+      "loss": 0.7691,
+      "step": 4451
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2217906029224757e-05,
+      "loss": 0.7887,
+      "step": 4452
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2181783330871742e-05,
+      "loss": 0.7777,
+      "step": 4453
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2145686357770046e-05,
+      "loss": 0.7842,
+      "step": 4454
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2109615121852622e-05,
+      "loss": 0.7321,
+      "step": 4455
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2073569635044e-05,
+      "loss": 0.7873,
+      "step": 4456
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2037549909260112e-05,
+      "loss": 0.7758,
+      "step": 4457
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2001555956408428e-05,
+      "loss": 0.7628,
+      "step": 4458
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1965587788387855e-05,
+      "loss": 0.334,
+      "step": 4459
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1929645417088805e-05,
+      "loss": 0.8183,
+      "step": 4460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.189372885439316e-05,
+      "loss": 0.7939,
+      "step": 4461
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1857838112174267e-05,
+      "loss": 0.8025,
+      "step": 4462
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.182197320229694e-05,
+      "loss": 0.7491,
+      "step": 4463
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.178613413661743e-05,
+      "loss": 0.7413,
+      "step": 4464
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.175032092698348e-05,
+      "loss": 0.8157,
+      "step": 4465
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1714533585234244e-05,
+      "loss": 0.8339,
+      "step": 4466
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1678772123200397e-05,
+      "loss": 0.3416,
+      "step": 4467
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.164303655270399e-05,
+      "loss": 0.7924,
+      "step": 4468
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1607326885558533e-05,
+      "loss": 0.7853,
+      "step": 4469
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1571643133568964e-05,
+      "loss": 0.7718,
+      "step": 4470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.153598530853168e-05,
+      "loss": 0.8086,
+      "step": 4471
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1500353422234475e-05,
+      "loss": 0.8449,
+      "step": 4472
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1464747486456625e-05,
+      "loss": 0.8068,
+      "step": 4473
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.142916751296876e-05,
+      "loss": 0.7966,
+      "step": 4474
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1393613513532962e-05,
+      "loss": 0.7771,
+      "step": 4475
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1358085499902725e-05,
+      "loss": 0.7286,
+      "step": 4476
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.132258348382291e-05,
+      "loss": 0.8048,
+      "step": 4477
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1287107477029878e-05,
+      "loss": 0.8096,
+      "step": 4478
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1251657491251307e-05,
+      "loss": 0.8047,
+      "step": 4479
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.121623353820632e-05,
+      "loss": 0.7823,
+      "step": 4480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1180835629605388e-05,
+      "loss": 0.8598,
+      "step": 4481
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.114546377715042e-05,
+      "loss": 0.7908,
+      "step": 4482
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1110117992534674e-05,
+      "loss": 0.8055,
+      "step": 4483
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.107479828744282e-05,
+      "loss": 0.7738,
+      "step": 4484
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.103950467355089e-05,
+      "loss": 0.7521,
+      "step": 4485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1004237162526296e-05,
+      "loss": 0.8075,
+      "step": 4486
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0968995766027786e-05,
+      "loss": 0.8628,
+      "step": 4487
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.093378049570558e-05,
+      "loss": 0.7778,
+      "step": 4488
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0898591363201137e-05,
+      "loss": 0.7578,
+      "step": 4489
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0863428380147344e-05,
+      "loss": 0.7928,
+      "step": 4490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.082829155816842e-05,
+      "loss": 0.8312,
+      "step": 4491
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.079318090887996e-05,
+      "loss": 0.7617,
+      "step": 4492
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0758096443888876e-05,
+      "loss": 0.7749,
+      "step": 4493
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.072303817479343e-05,
+      "loss": 0.8338,
+      "step": 4494
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0688006113183245e-05,
+      "loss": 0.3379,
+      "step": 4495
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0653000270639268e-05,
+      "loss": 0.7952,
+      "step": 4496
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0618020658733772e-05,
+      "loss": 0.7295,
+      "step": 4497
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0583067289030335e-05,
+      "loss": 0.7872,
+      "step": 4498
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0548140173083953e-05,
+      "loss": 0.7839,
+      "step": 4499
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0513239322440847e-05,
+      "loss": 0.7812,
+      "step": 4500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.047836474863858e-05,
+      "loss": 0.7355,
+      "step": 4501
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0443516463206048e-05,
+      "loss": 0.766,
+      "step": 4502
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0408694477663438e-05,
+      "loss": 0.7414,
+      "step": 4503
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.037389880352225e-05,
+      "loss": 0.7442,
+      "step": 4504
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0339129452285276e-05,
+      "loss": 0.8484,
+      "step": 4505
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.030438643544663e-05,
+      "loss": 0.8643,
+      "step": 4506
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0269669764491706e-05,
+      "loss": 0.7834,
+      "step": 4507
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0234979450897184e-05,
+      "loss": 0.8113,
+      "step": 4508
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0200315506130997e-05,
+      "loss": 0.3301,
+      "step": 4509
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.016567794165246e-05,
+      "loss": 0.3251,
+      "step": 4510
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.013106676891209e-05,
+      "loss": 0.3415,
+      "step": 4511
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0096481999351678e-05,
+      "loss": 0.8379,
+      "step": 4512
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.006192364440431e-05,
+      "loss": 0.7933,
+      "step": 4513
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0027391715494347e-05,
+      "loss": 0.8137,
+      "step": 4514
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9992886224037377e-05,
+      "loss": 0.8085,
+      "step": 4515
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9958407181440286e-05,
+      "loss": 0.8085,
+      "step": 4516
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.992395459910118e-05,
+      "loss": 0.7691,
+      "step": 4517
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.988952848840948e-05,
+      "loss": 0.7635,
+      "step": 4518
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9855128860745796e-05,
+      "loss": 0.8028,
+      "step": 4519
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.982075572748201e-05,
+      "loss": 0.8273,
+      "step": 4520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.978640909998123e-05,
+      "loss": 0.8156,
+      "step": 4521
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9752088989597795e-05,
+      "loss": 0.8089,
+      "step": 4522
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.971779540767733e-05,
+      "loss": 0.8516,
+      "step": 4523
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9683528365556637e-05,
+      "loss": 0.7887,
+      "step": 4524
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.964928787456376e-05,
+      "loss": 0.788,
+      "step": 4525
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.961507394601797e-05,
+      "loss": 0.3397,
+      "step": 4526
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9580886591229752e-05,
+      "loss": 0.843,
+      "step": 4527
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.95467258215008e-05,
+      "loss": 0.8036,
+      "step": 4528
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9512591648124033e-05,
+      "loss": 0.776,
+      "step": 4529
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9478484082383562e-05,
+      "loss": 0.7469,
+      "step": 4530
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9444403135554733e-05,
+      "loss": 0.7578,
+      "step": 4531
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9410348818904078e-05,
+      "loss": 0.786,
+      "step": 4532
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.937632114368929e-05,
+      "loss": 0.7725,
+      "step": 4533
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9342320121159295e-05,
+      "loss": 0.3396,
+      "step": 4534
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9308345762554214e-05,
+      "loss": 0.7672,
+      "step": 4535
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9274398079105316e-05,
+      "loss": 0.8071,
+      "step": 4536
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.924047708203507e-05,
+      "loss": 0.7769,
+      "step": 4537
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9206582782557136e-05,
+      "loss": 0.7456,
+      "step": 4538
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9172715191876335e-05,
+      "loss": 0.7772,
+      "step": 4539
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.913887432118866e-05,
+      "loss": 0.7717,
+      "step": 4540
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9105060181681244e-05,
+      "loss": 0.8119,
+      "step": 4541
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9071272784532468e-05,
+      "loss": 0.7815,
+      "step": 4542
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9037512140911783e-05,
+      "loss": 0.7951,
+      "step": 4543
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9003778261979843e-05,
+      "loss": 0.7225,
+      "step": 4544
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8970071158888413e-05,
+      "loss": 0.8246,
+      "step": 4545
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.893639084278046e-05,
+      "loss": 0.762,
+      "step": 4546
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8902737324790043e-05,
+      "loss": 0.8149,
+      "step": 4547
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8869110616042407e-05,
+      "loss": 0.7388,
+      "step": 4548
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8835510727653906e-05,
+      "loss": 0.825,
+      "step": 4549
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.880193767073204e-05,
+      "loss": 0.7915,
+      "step": 4550
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8768391456375413e-05,
+      "loss": 0.829,
+      "step": 4551
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8734872095673817e-05,
+      "loss": 0.7838,
+      "step": 4552
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.870137959970811e-05,
+      "loss": 0.7806,
+      "step": 4553
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.86679139795503e-05,
+      "loss": 0.755,
+      "step": 4554
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8634475246263473e-05,
+      "loss": 0.814,
+      "step": 4555
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8601063410901852e-05,
+      "loss": 0.7592,
+      "step": 4556
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8567678484510786e-05,
+      "loss": 0.7782,
+      "step": 4557
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.853432047812671e-05,
+      "loss": 0.7547,
+      "step": 4558
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8500989402777137e-05,
+      "loss": 0.8287,
+      "step": 4559
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8467685269480705e-05,
+      "loss": 0.8056,
+      "step": 4560
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8434408089247124e-05,
+      "loss": 0.8052,
+      "step": 4561
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8401157873077257e-05,
+      "loss": 0.8037,
+      "step": 4562
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8367934631962968e-05,
+      "loss": 0.7598,
+      "step": 4563
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8334738376887262e-05,
+      "loss": 0.7246,
+      "step": 4564
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.830156911882418e-05,
+      "loss": 0.7891,
+      "step": 4565
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.826842686873885e-05,
+      "loss": 0.7933,
+      "step": 4566
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.823531163758754e-05,
+      "loss": 0.8025,
+      "step": 4567
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.820222343631748e-05,
+      "loss": 0.7154,
+      "step": 4568
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8169162275867025e-05,
+      "loss": 0.7611,
+      "step": 4569
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8136128167165578e-05,
+      "loss": 0.7576,
+      "step": 4570
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8103121121133593e-05,
+      "loss": 0.7998,
+      "step": 4571
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8070141148682584e-05,
+      "loss": 0.7875,
+      "step": 4572
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.80371882607151e-05,
+      "loss": 0.8259,
+      "step": 4573
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.80042624681248e-05,
+      "loss": 0.8312,
+      "step": 4574
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7971363781796304e-05,
+      "loss": 0.7357,
+      "step": 4575
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7938492212605306e-05,
+      "loss": 0.7106,
+      "step": 4576
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7905647771418542e-05,
+      "loss": 0.7814,
+      "step": 4577
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.787283046909376e-05,
+      "loss": 0.7795,
+      "step": 4578
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7840040316479757e-05,
+      "loss": 0.7695,
+      "step": 4579
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7807277324416338e-05,
+      "loss": 0.809,
+      "step": 4580
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7774541503734342e-05,
+      "loss": 0.8298,
+      "step": 4581
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7741832865255625e-05,
+      "loss": 0.721,
+      "step": 4582
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7709151419793024e-05,
+      "loss": 0.7392,
+      "step": 4583
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7676497178150464e-05,
+      "loss": 0.8645,
+      "step": 4584
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7643870151122797e-05,
+      "loss": 0.7465,
+      "step": 4585
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7611270349495924e-05,
+      "loss": 0.812,
+      "step": 4586
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7578697784046727e-05,
+      "loss": 0.7888,
+      "step": 4587
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7546152465543088e-05,
+      "loss": 0.8551,
+      "step": 4588
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7513634404743873e-05,
+      "loss": 0.7905,
+      "step": 4589
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7481143612398955e-05,
+      "loss": 0.831,
+      "step": 4590
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.744868009924918e-05,
+      "loss": 0.7848,
+      "step": 4591
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7416243876026396e-05,
+      "loss": 0.8042,
+      "step": 4592
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7383834953453392e-05,
+      "loss": 0.7856,
+      "step": 4593
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.735145334224394e-05,
+      "loss": 0.8071,
+      "step": 4594
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.731909905310285e-05,
+      "loss": 0.8177,
+      "step": 4595
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.728677209672581e-05,
+      "loss": 0.3434,
+      "step": 4596
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7254472483799523e-05,
+      "loss": 0.8229,
+      "step": 4597
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7222200225001616e-05,
+      "loss": 0.7583,
+      "step": 4598
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7189955331000716e-05,
+      "loss": 0.8174,
+      "step": 4599
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7157737812456386e-05,
+      "loss": 0.8741,
+      "step": 4600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7125547680019127e-05,
+      "loss": 0.7709,
+      "step": 4601
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7093384944330393e-05,
+      "loss": 0.8551,
+      "step": 4602
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706124961602259e-05,
+      "loss": 0.7472,
+      "step": 4603
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7029141705719064e-05,
+      "loss": 0.8479,
+      "step": 4604
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6997061224034062e-05,
+      "loss": 0.811,
+      "step": 4605
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.696500818157284e-05,
+      "loss": 0.7823,
+      "step": 4606
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.693298258893151e-05,
+      "loss": 0.7014,
+      "step": 4607
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6900984456697145e-05,
+      "loss": 0.8179,
+      "step": 4608
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.686901379544772e-05,
+      "loss": 0.7721,
+      "step": 4609
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6837070615752115e-05,
+      "loss": 0.768,
+      "step": 4610
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6805154928170207e-05,
+      "loss": 0.8333,
+      "step": 4611
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6773266743252703e-05,
+      "loss": 0.6893,
+      "step": 4612
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.674140607154123e-05,
+      "loss": 0.8082,
+      "step": 4613
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.670957292356835e-05,
+      "loss": 0.2961,
+      "step": 4614
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6677767309857474e-05,
+      "loss": 0.7433,
+      "step": 4615
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6645989240922987e-05,
+      "loss": 0.7984,
+      "step": 4616
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6614238727270113e-05,
+      "loss": 0.7983,
+      "step": 4617
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6582515779394968e-05,
+      "loss": 0.3343,
+      "step": 4618
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6550820407784584e-05,
+      "loss": 0.7588,
+      "step": 4619
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6519152622916843e-05,
+      "loss": 0.7964,
+      "step": 4620
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.648751243526053e-05,
+      "loss": 0.8333,
+      "step": 4621
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6455899855275303e-05,
+      "loss": 0.7526,
+      "step": 4622
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6424314893411697e-05,
+      "loss": 0.7995,
+      "step": 4623
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6392757560111093e-05,
+      "loss": 0.7215,
+      "step": 4624
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.636122786580577e-05,
+      "loss": 0.8134,
+      "step": 4625
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.632972582091884e-05,
+      "loss": 0.8164,
+      "step": 4626
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6298251435864332e-05,
+      "loss": 0.803,
+      "step": 4627
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6266804721047058e-05,
+      "loss": 0.7787,
+      "step": 4628
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6235385686862726e-05,
+      "loss": 0.8053,
+      "step": 4629
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6203994343697882e-05,
+      "loss": 0.3258,
+      "step": 4630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6172630701929915e-05,
+      "loss": 0.7737,
+      "step": 4631
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6141294771927062e-05,
+      "loss": 0.7882,
+      "step": 4632
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6109986564048396e-05,
+      "loss": 0.7686,
+      "step": 4633
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6078706088643836e-05,
+      "loss": 0.8469,
+      "step": 4634
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.604745335605411e-05,
+      "loss": 0.795,
+      "step": 4635
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.60162283766108e-05,
+      "loss": 0.7534,
+      "step": 4636
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5985031160636256e-05,
+      "loss": 0.7863,
+      "step": 4637
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5953861718443774e-05,
+      "loss": 0.7552,
+      "step": 4638
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.592272006033735e-05,
+      "loss": 0.7906,
+      "step": 4639
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5891606196611843e-05,
+      "loss": 0.8244,
+      "step": 4640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5860520137552902e-05,
+      "loss": 0.7492,
+      "step": 4641
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5829461893437015e-05,
+      "loss": 0.7686,
+      "step": 4642
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5798431474531438e-05,
+      "loss": 0.6846,
+      "step": 4643
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.576742889109427e-05,
+      "loss": 0.7724,
+      "step": 4644
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5736454153374382e-05,
+      "loss": 0.8096,
+      "step": 4645
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.570550727161144e-05,
+      "loss": 0.8365,
+      "step": 4646
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5674588256035882e-05,
+      "loss": 0.7776,
+      "step": 4647
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5643697116869004e-05,
+      "loss": 0.7938,
+      "step": 4648
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.561283386432283e-05,
+      "loss": 0.8013,
+      "step": 4649
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.558199850860016e-05,
+      "loss": 0.8092,
+      "step": 4650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5551191059894587e-05,
+      "loss": 0.3327,
+      "step": 4651
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.55204115283905e-05,
+      "loss": 0.8112,
+      "step": 4652
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5489659924263012e-05,
+      "loss": 0.7963,
+      "step": 4653
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5458936257678014e-05,
+      "loss": 0.8675,
+      "step": 4654
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.542824053879224e-05,
+      "loss": 0.763,
+      "step": 4655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.539757277775308e-05,
+      "loss": 0.3268,
+      "step": 4656
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5366932984698733e-05,
+      "loss": 0.8695,
+      "step": 4657
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.533632116975814e-05,
+      "loss": 0.8015,
+      "step": 4658
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5305737343050976e-05,
+      "loss": 0.3194,
+      "step": 4659
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.527518151468773e-05,
+      "loss": 0.3462,
+      "step": 4660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5244653694769562e-05,
+      "loss": 0.8034,
+      "step": 4661
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5214153893388405e-05,
+      "loss": 0.3274,
+      "step": 4662
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5183682120626918e-05,
+      "loss": 0.7752,
+      "step": 4663
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.51532383865585e-05,
+      "loss": 0.8311,
+      "step": 4664
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5122822701247297e-05,
+      "loss": 0.7736,
+      "step": 4665
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5092435074748146e-05,
+      "loss": 0.7462,
+      "step": 4666
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5062075517106644e-05,
+      "loss": 0.7542,
+      "step": 4667
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5031744038359097e-05,
+      "loss": 0.756,
+      "step": 4668
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5001440648532495e-05,
+      "loss": 0.3522,
+      "step": 4669
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4971165357644613e-05,
+      "loss": 0.8288,
+      "step": 4670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.49409181757039e-05,
+      "loss": 0.8704,
+      "step": 4671
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.491069911270948e-05,
+      "loss": 0.8252,
+      "step": 4672
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4880508178651232e-05,
+      "loss": 0.8133,
+      "step": 4673
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.48503453835097e-05,
+      "loss": 0.8033,
+      "step": 4674
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4820210737256157e-05,
+      "loss": 0.7997,
+      "step": 4675
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4790104249852554e-05,
+      "loss": 0.7859,
+      "step": 4676
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4760025931251532e-05,
+      "loss": 0.7609,
+      "step": 4677
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4729975791396411e-05,
+      "loss": 0.7823,
+      "step": 4678
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4699953840221193e-05,
+      "loss": 0.8288,
+      "step": 4679
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4669960087650625e-05,
+      "loss": 0.7175,
+      "step": 4680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4639994543600056e-05,
+      "loss": 0.7725,
+      "step": 4681
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4610057217975526e-05,
+      "loss": 0.4041,
+      "step": 4682
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4580148120673776e-05,
+      "loss": 0.7992,
+      "step": 4683
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4550267261582173e-05,
+      "loss": 0.3156,
+      "step": 4684
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4520414650578796e-05,
+      "loss": 0.8098,
+      "step": 4685
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4490590297532346e-05,
+      "loss": 0.8093,
+      "step": 4686
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4460794212302197e-05,
+      "loss": 0.8058,
+      "step": 4687
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4431026404738391e-05,
+      "loss": 0.7276,
+      "step": 4688
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4401286884681609e-05,
+      "loss": 0.7355,
+      "step": 4689
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4371575661963143e-05,
+      "loss": 0.7744,
+      "step": 4690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4341892746405029e-05,
+      "loss": 0.7598,
+      "step": 4691
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4312238147819857e-05,
+      "loss": 0.7092,
+      "step": 4692
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4282611876010887e-05,
+      "loss": 0.3207,
+      "step": 4693
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.425301394077201e-05,
+      "loss": 0.7775,
+      "step": 4694
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.422344435188775e-05,
+      "loss": 0.8081,
+      "step": 4695
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4193903119133256e-05,
+      "loss": 0.34,
+      "step": 4696
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4164390252274329e-05,
+      "loss": 0.8532,
+      "step": 4697
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4134905761067329e-05,
+      "loss": 0.7811,
+      "step": 4698
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.410544965525934e-05,
+      "loss": 0.8042,
+      "step": 4699
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.407602194458797e-05,
+      "loss": 0.7928,
+      "step": 4700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.404662263878147e-05,
+      "loss": 0.7254,
+      "step": 4701
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4017251747558712e-05,
+      "loss": 0.3588,
+      "step": 4702
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.398790928062913e-05,
+      "loss": 0.8156,
+      "step": 4703
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.395859524769284e-05,
+      "loss": 0.7948,
+      "step": 4704
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3929309658440504e-05,
+      "loss": 0.7632,
+      "step": 4705
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3900052522553397e-05,
+      "loss": 0.769,
+      "step": 4706
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3870823849703363e-05,
+      "loss": 0.8562,
+      "step": 4707
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.384162364955286e-05,
+      "loss": 0.7822,
+      "step": 4708
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3812451931754944e-05,
+      "loss": 0.7322,
+      "step": 4709
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3783308705953224e-05,
+      "loss": 0.7913,
+      "step": 4710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3754193981781905e-05,
+      "loss": 0.7961,
+      "step": 4711
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3725107768865787e-05,
+      "loss": 0.806,
+      "step": 4712
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3696050076820233e-05,
+      "loss": 0.802,
+      "step": 4713
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3667020915251173e-05,
+      "loss": 0.7981,
+      "step": 4714
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3638020293755094e-05,
+      "loss": 0.8101,
+      "step": 4715
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3609048221919064e-05,
+      "loss": 0.7499,
+      "step": 4716
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3580104709320706e-05,
+      "loss": 0.7713,
+      "step": 4717
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3551189765528217e-05,
+      "loss": 0.7558,
+      "step": 4718
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3522303400100333e-05,
+      "loss": 0.7674,
+      "step": 4719
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3493445622586343e-05,
+      "loss": 0.7801,
+      "step": 4720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.34646164425261e-05,
+      "loss": 0.8421,
+      "step": 4721
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3435815869449964e-05,
+      "loss": 0.7678,
+      "step": 4722
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3407043912878903e-05,
+      "loss": 0.795,
+      "step": 4723
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3378300582324387e-05,
+      "loss": 0.7274,
+      "step": 4724
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3349585887288417e-05,
+      "loss": 0.8493,
+      "step": 4725
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3320899837263524e-05,
+      "loss": 0.8232,
+      "step": 4726
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3292242441732804e-05,
+      "loss": 0.7815,
+      "step": 4727
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3263613710169831e-05,
+      "loss": 0.7655,
+      "step": 4728
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3235013652038753e-05,
+      "loss": 0.7502,
+      "step": 4729
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3206442276794207e-05,
+      "loss": 0.7954,
+      "step": 4730
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3177899593881349e-05,
+      "loss": 0.7861,
+      "step": 4731
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3149385612735876e-05,
+      "loss": 0.3441,
+      "step": 4732
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3120900342783937e-05,
+      "loss": 0.7807,
+      "step": 4733
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3092443793442277e-05,
+      "loss": 0.7958,
+      "step": 4734
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3064015974118093e-05,
+      "loss": 0.7628,
+      "step": 4735
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.303561689420909e-05,
+      "loss": 0.7522,
+      "step": 4736
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3007246563103458e-05,
+      "loss": 0.7005,
+      "step": 4737
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.297890499017992e-05,
+      "loss": 0.7867,
+      "step": 4738
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2950592184807653e-05,
+      "loss": 0.75,
+      "step": 4739
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2922308156346353e-05,
+      "loss": 0.7987,
+      "step": 4740
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2894052914146193e-05,
+      "loss": 0.8113,
+      "step": 4741
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2865826467547825e-05,
+      "loss": 0.3162,
+      "step": 4742
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2837628825882364e-05,
+      "loss": 0.7797,
+      "step": 4743
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2809459998471462e-05,
+      "loss": 0.8566,
+      "step": 4744
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2781319994627206e-05,
+      "loss": 0.8186,
+      "step": 4745
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2753208823652141e-05,
+      "loss": 0.8361,
+      "step": 4746
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.272512649483929e-05,
+      "loss": 0.8319,
+      "step": 4747
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.269707301747215e-05,
+      "loss": 0.8262,
+      "step": 4748
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.266904840082469e-05,
+      "loss": 0.8135,
+      "step": 4749
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2641052654161333e-05,
+      "loss": 0.8061,
+      "step": 4750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2613085786736934e-05,
+      "loss": 0.8277,
+      "step": 4751
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2585147807796815e-05,
+      "loss": 0.7937,
+      "step": 4752
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2557238726576759e-05,
+      "loss": 0.7467,
+      "step": 4753
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2529358552302972e-05,
+      "loss": 0.7933,
+      "step": 4754
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.250150729419215e-05,
+      "loss": 0.8068,
+      "step": 4755
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2473684961451381e-05,
+      "loss": 0.8209,
+      "step": 4756
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2445891563278212e-05,
+      "loss": 0.7362,
+      "step": 4757
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2418127108860623e-05,
+      "loss": 0.8445,
+      "step": 4758
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2390391607377028e-05,
+      "loss": 0.7981,
+      "step": 4759
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.236268506799625e-05,
+      "loss": 0.7999,
+      "step": 4760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2335007499877582e-05,
+      "loss": 0.8407,
+      "step": 4761
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2307358912170686e-05,
+      "loss": 0.7481,
+      "step": 4762
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2279739314015682e-05,
+      "loss": 0.7654,
+      "step": 4763
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2252148714543088e-05,
+      "loss": 0.7976,
+      "step": 4764
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2224587122873833e-05,
+      "loss": 0.7196,
+      "step": 4765
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2197054548119302e-05,
+      "loss": 0.8239,
+      "step": 4766
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.216955099938122e-05,
+      "loss": 0.8254,
+      "step": 4767
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2142076485751751e-05,
+      "loss": 0.837,
+      "step": 4768
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.211463101631346e-05,
+      "loss": 0.7849,
+      "step": 4769
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2087214600139308e-05,
+      "loss": 0.3512,
+      "step": 4770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2059827246292654e-05,
+      "loss": 0.802,
+      "step": 4771
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2032468963827249e-05,
+      "loss": 0.7373,
+      "step": 4772
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2005139761787209e-05,
+      "loss": 0.7324,
+      "step": 4773
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.197783964920709e-05,
+      "loss": 0.8004,
+      "step": 4774
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1950568635111758e-05,
+      "loss": 0.7597,
+      "step": 4775
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1923326728516549e-05,
+      "loss": 0.7938,
+      "step": 4776
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1896113938427101e-05,
+      "loss": 0.8376,
+      "step": 4777
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1868930273839473e-05,
+      "loss": 0.8109,
+      "step": 4778
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1841775743740058e-05,
+      "loss": 0.7596,
+      "step": 4779
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.181465035710565e-05,
+      "loss": 0.6815,
+      "step": 4780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.17875541229034e-05,
+      "loss": 0.7384,
+      "step": 4781
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1760487050090796e-05,
+      "loss": 0.7153,
+      "step": 4782
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1733449147615738e-05,
+      "loss": 0.7884,
+      "step": 4783
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.170644042441642e-05,
+      "loss": 0.7612,
+      "step": 4784
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1679460889421445e-05,
+      "loss": 0.331,
+      "step": 4785
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1652510551549723e-05,
+      "loss": 0.3539,
+      "step": 4786
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1625589419710559e-05,
+      "loss": 0.3294,
+      "step": 4787
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1598697502803568e-05,
+      "loss": 0.8037,
+      "step": 4788
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1571834809718718e-05,
+      "loss": 0.7555,
+      "step": 4789
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1545001349336315e-05,
+      "loss": 0.8105,
+      "step": 4790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1518197130526987e-05,
+      "loss": 0.7521,
+      "step": 4791
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.14914221621517e-05,
+      "loss": 0.7217,
+      "step": 4792
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1464676453061796e-05,
+      "loss": 0.7958,
+      "step": 4793
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1437960012098892e-05,
+      "loss": 0.763,
+      "step": 4794
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1411272848094934e-05,
+      "loss": 0.7863,
+      "step": 4795
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1384614969872221e-05,
+      "loss": 0.8064,
+      "step": 4796
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1357986386243313e-05,
+      "loss": 0.3722,
+      "step": 4797
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1331387106011172e-05,
+      "loss": 0.8094,
+      "step": 4798
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1304817137968993e-05,
+      "loss": 0.7883,
+      "step": 4799
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1278276490900319e-05,
+      "loss": 0.7802,
+      "step": 4800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1251765173578998e-05,
+      "loss": 0.7532,
+      "step": 4801
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1225283194769176e-05,
+      "loss": 0.7872,
+      "step": 4802
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1198830563225292e-05,
+      "loss": 0.7927,
+      "step": 4803
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1172407287692099e-05,
+      "loss": 0.7785,
+      "step": 4804
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.114601337690464e-05,
+      "loss": 0.8037,
+      "step": 4805
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1119648839588258e-05,
+      "loss": 0.8384,
+      "step": 4806
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1093313684458539e-05,
+      "loss": 0.833,
+      "step": 4807
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1067007920221439e-05,
+      "loss": 0.7886,
+      "step": 4808
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1040731555573148e-05,
+      "loss": 0.779,
+      "step": 4809
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1014484599200125e-05,
+      "loss": 0.8349,
+      "step": 4810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0988267059779122e-05,
+      "loss": 0.7676,
+      "step": 4811
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0962078945977195e-05,
+      "loss": 0.7813,
+      "step": 4812
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0935920266451616e-05,
+      "loss": 0.7732,
+      "step": 4813
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.090979102984998e-05,
+      "loss": 0.3163,
+      "step": 4814
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0883691244810102e-05,
+      "loss": 0.3306,
+      "step": 4815
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.085762091996011e-05,
+      "loss": 0.776,
+      "step": 4816
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0831580063918345e-05,
+      "loss": 0.7571,
+      "step": 4817
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0805568685293422e-05,
+      "loss": 0.815,
+      "step": 4818
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0779586792684248e-05,
+      "loss": 0.7751,
+      "step": 4819
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0753634394679934e-05,
+      "loss": 0.8523,
+      "step": 4820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0727711499859872e-05,
+      "loss": 0.7987,
+      "step": 4821
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0701818116793672e-05,
+      "loss": 0.8461,
+      "step": 4822
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.067595425404121e-05,
+      "loss": 0.8013,
+      "step": 4823
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.06501199201526e-05,
+      "loss": 0.8017,
+      "step": 4824
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0624315123668183e-05,
+      "loss": 0.8078,
+      "step": 4825
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0598539873118552e-05,
+      "loss": 0.7305,
+      "step": 4826
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0572794177024526e-05,
+      "loss": 0.7497,
+      "step": 4827
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.054707804389713e-05,
+      "loss": 0.8159,
+      "step": 4828
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0521391482237642e-05,
+      "loss": 0.3288,
+      "step": 4829
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0495734500537591e-05,
+      "loss": 0.8065,
+      "step": 4830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0470107107278682e-05,
+      "loss": 0.7753,
+      "step": 4831
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0444509310932848e-05,
+      "loss": 0.793,
+      "step": 4832
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0418941119962244e-05,
+      "loss": 0.8272,
+      "step": 4833
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0393402542819231e-05,
+      "loss": 0.8227,
+      "step": 4834
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0367893587946398e-05,
+      "loss": 0.811,
+      "step": 4835
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0342414263776512e-05,
+      "loss": 0.739,
+      "step": 4836
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0316964578732579e-05,
+      "loss": 0.79,
+      "step": 4837
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0291544541227804e-05,
+      "loss": 0.6871,
+      "step": 4838
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0266154159665552e-05,
+      "loss": 0.8081,
+      "step": 4839
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0240793442439411e-05,
+      "loss": 0.7792,
+      "step": 4840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0215462397933162e-05,
+      "loss": 0.8082,
+      "step": 4841
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0190161034520795e-05,
+      "loss": 0.7824,
+      "step": 4842
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0164889360566454e-05,
+      "loss": 0.726,
+      "step": 4843
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0139647384424477e-05,
+      "loss": 0.7693,
+      "step": 4844
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0114435114439413e-05,
+      "loss": 0.7428,
+      "step": 4845
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.008925255894595e-05,
+      "loss": 0.7564,
+      "step": 4846
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0064099726268971e-05,
+      "loss": 0.823,
+      "step": 4847
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0038976624723539e-05,
+      "loss": 0.7959,
+      "step": 4848
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0013883262614887e-05,
+      "loss": 0.7863,
+      "step": 4849
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.988819648238379e-06,
+      "loss": 0.7366,
+      "step": 4850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.963785789879631e-06,
+      "loss": 0.7225,
+      "step": 4851
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.938781695814337e-06,
+      "loss": 0.7906,
+      "step": 4852
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.913807374308404e-06,
+      "loss": 0.7849,
+      "step": 4853
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.888862833617862e-06,
+      "loss": 0.8076,
+      "step": 4854
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.863948081988905e-06,
+      "loss": 0.7357,
+      "step": 4855
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.83906312765791e-06,
+      "loss": 0.8122,
+      "step": 4856
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.814207978851353e-06,
+      "loss": 0.8003,
+      "step": 4857
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.789382643785895e-06,
+      "loss": 0.814,
+      "step": 4858
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.764587130668323e-06,
+      "loss": 0.792,
+      "step": 4859
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.739821447695585e-06,
+      "loss": 0.7969,
+      "step": 4860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.715085603054719e-06,
+      "loss": 0.8323,
+      "step": 4861
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.690379604922983e-06,
+      "loss": 0.7061,
+      "step": 4862
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.66570346146769e-06,
+      "loss": 0.8316,
+      "step": 4863
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.641057180846324e-06,
+      "loss": 0.3854,
+      "step": 4864
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.616440771206481e-06,
+      "loss": 0.7244,
+      "step": 4865
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.591854240685882e-06,
+      "loss": 0.7365,
+      "step": 4866
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.567297597412384e-06,
+      "loss": 0.3485,
+      "step": 4867
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.542770849503946e-06,
+      "loss": 0.7929,
+      "step": 4868
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.518274005068661e-06,
+      "loss": 0.857,
+      "step": 4869
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.493807072204718e-06,
+      "loss": 0.7598,
+      "step": 4870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.469370059000426e-06,
+      "loss": 0.7913,
+      "step": 4871
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.444962973534244e-06,
+      "loss": 0.7998,
+      "step": 4872
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.420585823874661e-06,
+      "loss": 0.7996,
+      "step": 4873
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.396238618080322e-06,
+      "loss": 0.7933,
+      "step": 4874
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.371921364199964e-06,
+      "loss": 0.7652,
+      "step": 4875
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.347634070272404e-06,
+      "loss": 0.7894,
+      "step": 4876
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.323376744326573e-06,
+      "loss": 0.8065,
+      "step": 4877
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.299149394381501e-06,
+      "loss": 0.7597,
+      "step": 4878
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.274952028446282e-06,
+      "loss": 0.7982,
+      "step": 4879
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.250784654520106e-06,
+      "loss": 0.8028,
+      "step": 4880
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.226647280592294e-06,
+      "loss": 0.8033,
+      "step": 4881
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.202539914642182e-06,
+      "loss": 0.8019,
+      "step": 4882
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.178462564639223e-06,
+      "loss": 0.8081,
+      "step": 4883
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.154415238542946e-06,
+      "loss": 0.748,
+      "step": 4884
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.130397944302926e-06,
+      "loss": 0.69,
+      "step": 4885
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.106410689858857e-06,
+      "loss": 0.704,
+      "step": 4886
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.082453483140484e-06,
+      "loss": 0.7767,
+      "step": 4887
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.058526332067586e-06,
+      "loss": 0.8041,
+      "step": 4888
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.034629244550053e-06,
+      "loss": 0.8347,
+      "step": 4889
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.010762228487813e-06,
+      "loss": 0.8332,
+      "step": 4890
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.986925291770843e-06,
+      "loss": 0.8062,
+      "step": 4891
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.963118442279205e-06,
+      "loss": 0.8132,
+      "step": 4892
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.939341687882985e-06,
+      "loss": 0.7931,
+      "step": 4893
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.915595036442349e-06,
+      "loss": 0.7886,
+      "step": 4894
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.891878495807504e-06,
+      "loss": 0.8266,
+      "step": 4895
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.868192073818671e-06,
+      "loss": 0.8196,
+      "step": 4896
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.844535778306161e-06,
+      "loss": 0.7792,
+      "step": 4897
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.820909617090289e-06,
+      "loss": 0.8642,
+      "step": 4898
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.797313597981437e-06,
+      "loss": 0.7778,
+      "step": 4899
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.773747728780001e-06,
+      "loss": 0.812,
+      "step": 4900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.750212017276415e-06,
+      "loss": 0.8091,
+      "step": 4901
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.726706471251156e-06,
+      "loss": 0.7697,
+      "step": 4902
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.703231098474695e-06,
+      "loss": 0.8124,
+      "step": 4903
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.679785906707582e-06,
+      "loss": 0.7778,
+      "step": 4904
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.656370903700361e-06,
+      "loss": 0.831,
+      "step": 4905
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.632986097193573e-06,
+      "loss": 0.3154,
+      "step": 4906
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.609631494917813e-06,
+      "loss": 0.8062,
+      "step": 4907
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.586307104593672e-06,
+      "loss": 0.79,
+      "step": 4908
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.563012933931758e-06,
+      "loss": 0.7772,
+      "step": 4909
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.539748990632701e-06,
+      "loss": 0.7437,
+      "step": 4910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.516515282387105e-06,
+      "loss": 0.8019,
+      "step": 4911
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.493311816875615e-06,
+      "loss": 0.7843,
+      "step": 4912
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.470138601768873e-06,
+      "loss": 0.8629,
+      "step": 4913
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.446995644727473e-06,
+      "loss": 0.7685,
+      "step": 4914
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.423882953402096e-06,
+      "loss": 0.8121,
+      "step": 4915
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.40080053543334e-06,
+      "loss": 0.7597,
+      "step": 4916
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.377748398451824e-06,
+      "loss": 0.7926,
+      "step": 4917
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.354726550078152e-06,
+      "loss": 0.8023,
+      "step": 4918
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.331734997922936e-06,
+      "loss": 0.7743,
+      "step": 4919
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.308773749586728e-06,
+      "loss": 0.7747,
+      "step": 4920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.285842812660106e-06,
+      "loss": 0.7913,
+      "step": 4921
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.2629421947236e-06,
+      "loss": 0.3701,
+      "step": 4922
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.240071903347734e-06,
+      "loss": 0.7907,
+      "step": 4923
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.217231946092984e-06,
+      "loss": 0.7484,
+      "step": 4924
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.194422330509843e-06,
+      "loss": 0.3473,
+      "step": 4925
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.171643064138735e-06,
+      "loss": 0.7863,
+      "step": 4926
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.148894154510056e-06,
+      "loss": 0.8265,
+      "step": 4927
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.12617560914416e-06,
+      "loss": 0.7752,
+      "step": 4928
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.103487435551371e-06,
+      "loss": 0.832,
+      "step": 4929
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.080829641232013e-06,
+      "loss": 0.8085,
+      "step": 4930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.058202233676304e-06,
+      "loss": 0.7413,
+      "step": 4931
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.03560522036445e-06,
+      "loss": 0.7367,
+      "step": 4932
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.013038608766587e-06,
+      "loss": 0.7682,
+      "step": 4933
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.990502406342836e-06,
+      "loss": 0.8048,
+      "step": 4934
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.967996620543228e-06,
+      "loss": 0.8153,
+      "step": 4935
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.945521258807776e-06,
+      "loss": 0.8467,
+      "step": 4936
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.923076328566392e-06,
+      "loss": 0.8457,
+      "step": 4937
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.900661837238977e-06,
+      "loss": 0.8047,
+      "step": 4938
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.878277792235334e-06,
+      "loss": 0.7777,
+      "step": 4939
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.8559242009552e-06,
+      "loss": 0.7158,
+      "step": 4940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.833601070788266e-06,
+      "loss": 0.3265,
+      "step": 4941
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.811308409114138e-06,
+      "loss": 0.8078,
+      "step": 4942
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.789046223302354e-06,
+      "loss": 0.783,
+      "step": 4943
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.766814520712384e-06,
+      "loss": 0.791,
+      "step": 4944
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.744613308693605e-06,
+      "loss": 0.7309,
+      "step": 4945
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.72244259458531e-06,
+      "loss": 0.7724,
+      "step": 4946
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.700302385716763e-06,
+      "loss": 0.7512,
+      "step": 4947
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.678192689407082e-06,
+      "loss": 0.8327,
+      "step": 4948
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.656113512965325e-06,
+      "loss": 0.7918,
+      "step": 4949
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.634064863690448e-06,
+      "loss": 0.81,
+      "step": 4950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.612046748871327e-06,
+      "loss": 0.7929,
+      "step": 4951
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.590059175786746e-06,
+      "loss": 0.7509,
+      "step": 4952
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.5681021517053805e-06,
+      "loss": 0.7864,
+      "step": 4953
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.546175683885814e-06,
+      "loss": 0.8116,
+      "step": 4954
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.524279779576526e-06,
+      "loss": 0.836,
+      "step": 4955
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.502414446015893e-06,
+      "loss": 0.8023,
+      "step": 4956
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.480579690432176e-06,
+      "loss": 0.7705,
+      "step": 4957
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.45877552004357e-06,
+      "loss": 0.7875,
+      "step": 4958
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.437001942058108e-06,
+      "loss": 0.3234,
+      "step": 4959
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.415258963673732e-06,
+      "loss": 0.7661,
+      "step": 4960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.393546592078271e-06,
+      "loss": 0.7455,
+      "step": 4961
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.371864834449405e-06,
+      "loss": 0.311,
+      "step": 4962
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.350213697954756e-06,
+      "loss": 0.7081,
+      "step": 4963
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.328593189751754e-06,
+      "loss": 0.7622,
+      "step": 4964
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.307003316987748e-06,
+      "loss": 0.7902,
+      "step": 4965
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.285444086799942e-06,
+      "loss": 0.7734,
+      "step": 4966
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.263915506315433e-06,
+      "loss": 0.8307,
+      "step": 4967
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.2424175826511286e-06,
+      "loss": 0.7745,
+      "step": 4968
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.220950322913888e-06,
+      "loss": 0.8324,
+      "step": 4969
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.199513734200369e-06,
+      "loss": 0.8354,
+      "step": 4970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.178107823597113e-06,
+      "loss": 0.7597,
+      "step": 4971
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.156732598180505e-06,
+      "loss": 0.7196,
+      "step": 4972
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.1353880650167904e-06,
+      "loss": 0.7361,
+      "step": 4973
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.114074231162082e-06,
+      "loss": 0.3528,
+      "step": 4974
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.092791103662344e-06,
+      "loss": 0.8061,
+      "step": 4975
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.071538689553381e-06,
+      "loss": 0.7688,
+      "step": 4976
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.0503169958608286e-06,
+      "loss": 0.7789,
+      "step": 4977
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.029126029600197e-06,
+      "loss": 0.7783,
+      "step": 4978
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.0079657977768054e-06,
+      "loss": 0.7942,
+      "step": 4979
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.986836307385858e-06,
+      "loss": 0.7147,
+      "step": 4980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.965737565412367e-06,
+      "loss": 0.7781,
+      "step": 4981
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.944669578831176e-06,
+      "loss": 0.7736,
+      "step": 4982
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.923632354606957e-06,
+      "loss": 0.7365,
+      "step": 4983
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.902625899694237e-06,
+      "loss": 0.7944,
+      "step": 4984
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.881650221037361e-06,
+      "loss": 0.744,
+      "step": 4985
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.860705325570494e-06,
+      "loss": 0.7677,
+      "step": 4986
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.839791220217617e-06,
+      "loss": 0.7759,
+      "step": 4987
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.818907911892558e-06,
+      "loss": 0.7991,
+      "step": 4988
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.798055407498927e-06,
+      "loss": 0.7875,
+      "step": 4989
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.777233713930198e-06,
+      "loss": 0.7693,
+      "step": 4990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.756442838069632e-06,
+      "loss": 0.7653,
+      "step": 4991
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.7356827867902984e-06,
+      "loss": 0.7518,
+      "step": 4992
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.714953566955084e-06,
+      "loss": 0.8442,
+      "step": 4993
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.694255185416687e-06,
+      "loss": 0.7792,
+      "step": 4994
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.673587649017598e-06,
+      "loss": 0.754,
+      "step": 4995
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.652950964590121e-06,
+      "loss": 0.8103,
+      "step": 4996
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.632345138956364e-06,
+      "loss": 0.7496,
+      "step": 4997
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.611770178928223e-06,
+      "loss": 0.8049,
+      "step": 4998
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.5912260913073895e-06,
+      "loss": 0.8218,
+      "step": 4999
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.570712882885355e-06,
+      "loss": 0.7423,
+      "step": 5000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.550230560443438e-06,
+      "loss": 0.7822,
+      "step": 5001
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.529779130752678e-06,
+      "loss": 0.7716,
+      "step": 5002
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.509358600573945e-06,
+      "loss": 0.7311,
+      "step": 5003
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.488968976657894e-06,
+      "loss": 0.7661,
+      "step": 5004
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.468610265744935e-06,
+      "loss": 0.7781,
+      "step": 5005
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.448282474565303e-06,
+      "loss": 0.7689,
+      "step": 5006
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.427985609838972e-06,
+      "loss": 0.8335,
+      "step": 5007
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.407719678275703e-06,
+      "loss": 0.8259,
+      "step": 5008
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.3874846865750405e-06,
+      "loss": 0.8048,
+      "step": 5009
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.3672806414262765e-06,
+      "loss": 0.7972,
+      "step": 5010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.347107549508536e-06,
+      "loss": 0.7315,
+      "step": 5011
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.326965417490638e-06,
+      "loss": 0.8029,
+      "step": 5012
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.306854252031202e-06,
+      "loss": 0.8419,
+      "step": 5013
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.286774059778599e-06,
+      "loss": 0.327,
+      "step": 5014
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.266724847370964e-06,
+      "loss": 0.7836,
+      "step": 5015
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.246706621436205e-06,
+      "loss": 0.8227,
+      "step": 5016
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.226719388591961e-06,
+      "loss": 0.7211,
+      "step": 5017
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.206763155445627e-06,
+      "loss": 0.8795,
+      "step": 5018
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.186837928594402e-06,
+      "loss": 0.738,
+      "step": 5019
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.166943714625173e-06,
+      "loss": 0.7778,
+      "step": 5020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.14708052011459e-06,
+      "loss": 0.7512,
+      "step": 5021
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.127248351629056e-06,
+      "loss": 0.7593,
+      "step": 5022
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.1074472157247045e-06,
+      "loss": 0.7681,
+      "step": 5023
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.087677118947455e-06,
+      "loss": 0.7764,
+      "step": 5024
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.067938067832912e-06,
+      "loss": 0.7616,
+      "step": 5025
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.0482300689064466e-06,
+      "loss": 0.8525,
+      "step": 5026
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.028553128683145e-06,
+      "loss": 0.8175,
+      "step": 5027
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.008907253667839e-06,
+      "loss": 0.7829,
+      "step": 5028
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.9892924503551e-06,
+      "loss": 0.3612,
+      "step": 5029
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.969708725229195e-06,
+      "loss": 0.7999,
+      "step": 5030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.950156084764147e-06,
+      "loss": 0.8268,
+      "step": 5031
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.930634535423696e-06,
+      "loss": 0.7199,
+      "step": 5032
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.911144083661313e-06,
+      "loss": 0.8383,
+      "step": 5033
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.891684735920167e-06,
+      "loss": 0.8013,
+      "step": 5034
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8722564986331555e-06,
+      "loss": 0.3199,
+      "step": 5035
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.852859378222897e-06,
+      "loss": 0.7106,
+      "step": 5036
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.833493381101706e-06,
+      "loss": 0.7521,
+      "step": 5037
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.81415851367163e-06,
+      "loss": 0.8024,
+      "step": 5038
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.794854782324421e-06,
+      "loss": 0.7843,
+      "step": 5039
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.77558219344152e-06,
+      "loss": 0.7888,
+      "step": 5040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.756340753394085e-06,
+      "loss": 0.3287,
+      "step": 5041
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.737130468542972e-06,
+      "loss": 0.7189,
+      "step": 5042
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.71795134523877e-06,
+      "loss": 0.8184,
+      "step": 5043
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.698803389821728e-06,
+      "loss": 0.7945,
+      "step": 5044
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.679686608621792e-06,
+      "loss": 0.7515,
+      "step": 5045
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.6606010079586215e-06,
+      "loss": 0.7625,
+      "step": 5046
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.641546594141556e-06,
+      "loss": 0.7897,
+      "step": 5047
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.622523373469635e-06,
+      "loss": 0.814,
+      "step": 5048
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.603531352231572e-06,
+      "loss": 0.7935,
+      "step": 5049
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.58457053670578e-06,
+      "loss": 0.7875,
+      "step": 5050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.565640933160354e-06,
+      "loss": 0.7758,
+      "step": 5051
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.546742547853067e-06,
+      "loss": 0.8273,
+      "step": 5052
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.527875387031356e-06,
+      "loss": 0.7192,
+      "step": 5053
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.509039456932385e-06,
+      "loss": 0.7882,
+      "step": 5054
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.490234763782931e-06,
+      "loss": 0.8189,
+      "step": 5055
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.471461313799497e-06,
+      "loss": 0.7682,
+      "step": 5056
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.4527191131882316e-06,
+      "loss": 0.8129,
+      "step": 5057
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.434008168144944e-06,
+      "loss": 0.7911,
+      "step": 5058
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.4153284848551315e-06,
+      "loss": 0.8239,
+      "step": 5059
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.396680069493953e-06,
+      "loss": 0.7971,
+      "step": 5060
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.378062928226213e-06,
+      "loss": 0.8577,
+      "step": 5061
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.359477067206397e-06,
+      "loss": 0.7904,
+      "step": 5062
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.34092249257866e-06,
+      "loss": 0.7886,
+      "step": 5063
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.322399210476781e-06,
+      "loss": 0.8054,
+      "step": 5064
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.303907227024219e-06,
+      "loss": 0.7986,
+      "step": 5065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2854465483340725e-06,
+      "loss": 0.8178,
+      "step": 5066
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.267017180509093e-06,
+      "loss": 0.7497,
+      "step": 5067
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.248619129641707e-06,
+      "loss": 0.8005,
+      "step": 5068
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.230252401813962e-06,
+      "loss": 0.7866,
+      "step": 5069
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.211917003097544e-06,
+      "loss": 0.7501,
+      "step": 5070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.193612939553794e-06,
+      "loss": 0.685,
+      "step": 5071
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.175340217233704e-06,
+      "loss": 0.7682,
+      "step": 5072
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.157098842177888e-06,
+      "loss": 0.7533,
+      "step": 5073
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1388888204165875e-06,
+      "loss": 0.7976,
+      "step": 5074
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.120710157969733e-06,
+      "loss": 0.7974,
+      "step": 5075
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.102562860846827e-06,
+      "loss": 0.3151,
+      "step": 5076
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.084446935047049e-06,
+      "loss": 0.793,
+      "step": 5077
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.066362386559154e-06,
+      "loss": 0.7768,
+      "step": 5078
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0483092213615805e-06,
+      "loss": 0.7891,
+      "step": 5079
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.030287445422366e-06,
+      "loss": 0.7872,
+      "step": 5080
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.012297064699156e-06,
+      "loss": 0.7868,
+      "step": 5081
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9943380851392604e-06,
+      "loss": 0.8129,
+      "step": 5082
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.976410512679564e-06,
+      "loss": 0.8234,
+      "step": 5083
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.958514353246602e-06,
+      "loss": 0.7602,
+      "step": 5084
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.940649612756476e-06,
+      "loss": 0.7716,
+      "step": 5085
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9228162971149846e-06,
+      "loss": 0.7406,
+      "step": 5086
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.905014412217468e-06,
+      "loss": 0.8158,
+      "step": 5087
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.887243963948895e-06,
+      "loss": 0.8207,
+      "step": 5088
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.869504958183846e-06,
+      "loss": 0.8092,
+      "step": 5089
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.851797400786506e-06,
+      "loss": 0.8247,
+      "step": 5090
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.834121297610672e-06,
+      "loss": 0.8093,
+      "step": 5091
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.816476654499713e-06,
+      "loss": 0.758,
+      "step": 5092
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.798863477286641e-06,
+      "loss": 0.8546,
+      "step": 5093
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.781281771794033e-06,
+      "loss": 0.8389,
+      "step": 5094
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.763731543834071e-06,
+      "loss": 0.7835,
+      "step": 5095
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.746212799208527e-06,
+      "loss": 0.7787,
+      "step": 5096
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.72872554370879e-06,
+      "loss": 0.7477,
+      "step": 5097
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7112697831158126e-06,
+      "loss": 0.7959,
+      "step": 5098
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.693845523200146e-06,
+      "loss": 0.7257,
+      "step": 5099
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.676452769721917e-06,
+      "loss": 0.779,
+      "step": 5100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.659091528430859e-06,
+      "loss": 0.8176,
+      "step": 5101
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.641761805066258e-06,
+      "loss": 0.317,
+      "step": 5102
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.62446360535701e-06,
+      "loss": 0.7742,
+      "step": 5103
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.607196935021574e-06,
+      "loss": 0.7613,
+      "step": 5104
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.589961799767984e-06,
+      "loss": 0.3308,
+      "step": 5105
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.572758205293848e-06,
+      "loss": 0.8417,
+      "step": 5106
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.555586157286384e-06,
+      "loss": 0.76,
+      "step": 5107
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.53844566142232e-06,
+      "loss": 0.7739,
+      "step": 5108
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5213367233680084e-06,
+      "loss": 0.8294,
+      "step": 5109
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.504259348779316e-06,
+      "loss": 0.8062,
+      "step": 5110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.487213543301716e-06,
+      "loss": 0.7734,
+      "step": 5111
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.470199312570256e-06,
+      "loss": 0.7828,
+      "step": 5112
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.453216662209492e-06,
+      "loss": 0.7792,
+      "step": 5113
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4362655978336e-06,
+      "loss": 0.82,
+      "step": 5114
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.419346125046265e-06,
+      "loss": 0.7834,
+      "step": 5115
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4024582494407556e-06,
+      "loss": 0.7701,
+      "step": 5116
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.385601976599873e-06,
+      "loss": 0.7609,
+      "step": 5117
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.368777312096006e-06,
+      "loss": 0.7832,
+      "step": 5118
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.351984261491082e-06,
+      "loss": 0.7737,
+      "step": 5119
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3352228303365605e-06,
+      "loss": 0.8018,
+      "step": 5120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.318493024173453e-06,
+      "loss": 0.7949,
+      "step": 5121
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3017948485323255e-06,
+      "loss": 0.7714,
+      "step": 5122
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.285128308933284e-06,
+      "loss": 0.6907,
+      "step": 5123
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2684934108859765e-06,
+      "loss": 0.7984,
+      "step": 5124
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.251890159889582e-06,
+      "loss": 0.8214,
+      "step": 5125
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.235318561432844e-06,
+      "loss": 0.7567,
+      "step": 5126
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.218778620994002e-06,
+      "loss": 0.8134,
+      "step": 5127
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2022703440408486e-06,
+      "loss": 0.7705,
+      "step": 5128
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1857937360307405e-06,
+      "loss": 0.8726,
+      "step": 5129
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.169348802410522e-06,
+      "loss": 0.7956,
+      "step": 5130
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.152935548616588e-06,
+      "loss": 0.7382,
+      "step": 5131
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.136553980074842e-06,
+      "loss": 0.7738,
+      "step": 5132
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.120204102200731e-06,
+      "loss": 0.754,
+      "step": 5133
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.10388592039922e-06,
+      "loss": 0.7638,
+      "step": 5134
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.087599440064793e-06,
+      "loss": 0.3475,
+      "step": 5135
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.071344666581456e-06,
+      "loss": 0.7851,
+      "step": 5136
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.055121605322742e-06,
+      "loss": 0.8024,
+      "step": 5137
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.038930261651674e-06,
+      "loss": 0.8211,
+      "step": 5138
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.022770640920837e-06,
+      "loss": 0.7909,
+      "step": 5139
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.006642748472278e-06,
+      "loss": 0.7999,
+      "step": 5140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9905465896376e-06,
+      "loss": 0.8165,
+      "step": 5141
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.974482169737859e-06,
+      "loss": 0.7482,
+      "step": 5142
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.958449494083683e-06,
+      "loss": 0.8093,
+      "step": 5143
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9424485679751546e-06,
+      "loss": 0.822,
+      "step": 5144
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.926479396701899e-06,
+      "loss": 0.8179,
+      "step": 5145
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.910541985543014e-06,
+      "loss": 0.8123,
+      "step": 5146
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.894636339767122e-06,
+      "loss": 0.7397,
+      "step": 5147
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.878762464632313e-06,
+      "loss": 0.7393,
+      "step": 5148
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.862920365386202e-06,
+      "loss": 0.7793,
+      "step": 5149
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.847110047265911e-06,
+      "loss": 0.81,
+      "step": 5150
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.831331515498016e-06,
+      "loss": 0.7571,
+      "step": 5151
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.81558477529862e-06,
+      "loss": 0.7513,
+      "step": 5152
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.799869831873282e-06,
+      "loss": 0.7889,
+      "step": 5153
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7841866904170798e-06,
+      "loss": 0.7583,
+      "step": 5154
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7685353561145662e-06,
+      "loss": 0.7415,
+      "step": 5155
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.752915834139781e-06,
+      "loss": 0.7335,
+      "step": 5156
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.737328129656259e-06,
+      "loss": 0.7718,
+      "step": 5157
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7217722478169903e-06,
+      "loss": 0.7949,
+      "step": 5158
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.706248193764461e-06,
+      "loss": 0.7797,
+      "step": 5159
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.690755972630622e-06,
+      "loss": 0.7972,
+      "step": 5160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6752955895369424e-06,
+      "loss": 0.7948,
+      "step": 5161
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6598670495943123e-06,
+      "loss": 0.8212,
+      "step": 5162
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6444703579031293e-06,
+      "loss": 0.7932,
+      "step": 5163
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.629105519553255e-06,
+      "loss": 0.8126,
+      "step": 5164
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6137725396240164e-06,
+      "loss": 0.8297,
+      "step": 5165
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.598471423184202e-06,
+      "loss": 0.7968,
+      "step": 5166
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5832021752920885e-06,
+      "loss": 0.7683,
+      "step": 5167
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5679648009953935e-06,
+      "loss": 0.8287,
+      "step": 5168
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5527593053313103e-06,
+      "loss": 0.7625,
+      "step": 5169
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.537585693326484e-06,
+      "loss": 0.7617,
+      "step": 5170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.522443969997058e-06,
+      "loss": 0.8441,
+      "step": 5171
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5073341403485727e-06,
+      "loss": 0.3375,
+      "step": 5172
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.492256209376077e-06,
+      "loss": 0.7353,
+      "step": 5173
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.477210182064039e-06,
+      "loss": 0.8176,
+      "step": 5174
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4621960633864028e-06,
+      "loss": 0.7487,
+      "step": 5175
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.447213858306564e-06,
+      "loss": 0.784,
+      "step": 5176
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.432263571777339e-06,
+      "loss": 0.7728,
+      "step": 5177
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4173452087410187e-06,
+      "loss": 0.8116,
+      "step": 5178
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.402458774129358e-06,
+      "loss": 0.7361,
+      "step": 5179
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3876042728635092e-06,
+      "loss": 0.7449,
+      "step": 5180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3727817098540893e-06,
+      "loss": 0.8052,
+      "step": 5181
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.357991090001189e-06,
+      "loss": 0.8099,
+      "step": 5182
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3432324181942866e-06,
+      "loss": 0.7666,
+      "step": 5183
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3285056993123455e-06,
+      "loss": 0.7713,
+      "step": 5184
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.313810938223727e-06,
+      "loss": 0.7384,
+      "step": 5185
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2991481397862568e-06,
+      "loss": 0.7621,
+      "step": 5186
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2845173088471682e-06,
+      "loss": 0.3183,
+      "step": 5187
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.269918450243159e-06,
+      "loss": 0.7553,
+      "step": 5188
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2553515688003244e-06,
+      "loss": 0.773,
+      "step": 5189
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2408166693342123e-06,
+      "loss": 0.7612,
+      "step": 5190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.22631375664979e-06,
+      "loss": 0.7159,
+      "step": 5191
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.211842835541423e-06,
+      "loss": 0.3276,
+      "step": 5192
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1974039107929733e-06,
+      "loss": 0.7482,
+      "step": 5193
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1829969871776555e-06,
+      "loss": 0.7358,
+      "step": 5194
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.168622069458138e-06,
+      "loss": 0.726,
+      "step": 5195
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1542791623864863e-06,
+      "loss": 0.8022,
+      "step": 5196
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.139968270704219e-06,
+      "loss": 0.8041,
+      "step": 5197
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.125689399142229e-06,
+      "loss": 0.3295,
+      "step": 5198
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1114425524208515e-06,
+      "loss": 0.7655,
+      "step": 5199
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0972277352498303e-06,
+      "loss": 0.7866,
+      "step": 5200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.083044952328329e-06,
+      "loss": 0.7871,
+      "step": 5201
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0688942083448967e-06,
+      "loss": 0.3636,
+      "step": 5202
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0547755079775143e-06,
+      "loss": 0.7433,
+      "step": 5203
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0406888558935476e-06,
+      "loss": 0.8511,
+      "step": 5204
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.026634256749794e-06,
+      "loss": 0.7507,
+      "step": 5205
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.012611715192437e-06,
+      "loss": 0.7572,
+      "step": 5206
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9986212358570796e-06,
+      "loss": 0.7979,
+      "step": 5207
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.984662823368689e-06,
+      "loss": 0.8343,
+      "step": 5208
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.970736482341685e-06,
+      "loss": 0.7812,
+      "step": 5209
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9568422173798294e-06,
+      "loss": 0.8336,
+      "step": 5210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9429800330763256e-06,
+      "loss": 0.7877,
+      "step": 5211
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.929149934013742e-06,
+      "loss": 0.7138,
+      "step": 5212
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9153519247640425e-06,
+      "loss": 0.7919,
+      "step": 5213
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.901586009888624e-06,
+      "loss": 0.7711,
+      "step": 5214
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.887852193938223e-06,
+      "loss": 0.8355,
+      "step": 5215
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.874150481452975e-06,
+      "loss": 0.3442,
+      "step": 5216
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8604808769624235e-06,
+      "loss": 0.7332,
+      "step": 5217
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.846843384985476e-06,
+      "loss": 0.8066,
+      "step": 5218
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8332380100304367e-06,
+      "loss": 0.7913,
+      "step": 5219
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8196647565949864e-06,
+      "loss": 0.7931,
+      "step": 5220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8061236291661906e-06,
+      "loss": 0.7787,
+      "step": 5221
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7926146322204914e-06,
+      "loss": 0.7831,
+      "step": 5222
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7791377702237274e-06,
+      "loss": 0.8432,
+      "step": 5223
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7656930476310683e-06,
+      "loss": 0.7375,
+      "step": 5224
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.752280468887114e-06,
+      "loss": 0.7908,
+      "step": 5225
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7389000384257955e-06,
+      "loss": 0.7724,
+      "step": 5226
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.725551760670453e-06,
+      "loss": 0.8033,
+      "step": 5227
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7122356400337667e-06,
+      "loss": 0.7703,
+      "step": 5228
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.698951680917794e-06,
+      "loss": 0.8048,
+      "step": 5229
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6856998877139773e-06,
+      "loss": 0.7881,
+      "step": 5230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.67248026480309e-06,
+      "loss": 0.7615,
+      "step": 5231
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6592928165553143e-06,
+      "loss": 0.7632,
+      "step": 5232
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.646137547330174e-06,
+      "loss": 0.7877,
+      "step": 5233
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.633014461476524e-06,
+      "loss": 0.7878,
+      "step": 5234
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.61992356333266e-06,
+      "loss": 0.7693,
+      "step": 5235
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6068648572261543e-06,
+      "loss": 0.8154,
+      "step": 5236
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.593838347473987e-06,
+      "loss": 0.8104,
+      "step": 5237
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5808440383824796e-06,
+      "loss": 0.3127,
+      "step": 5238
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.567881934247296e-06,
+      "loss": 0.7555,
+      "step": 5239
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.554952039353475e-06,
+      "loss": 0.6975,
+      "step": 5240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5420543579753988e-06,
+      "loss": 0.7833,
+      "step": 5241
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5291888943767992e-06,
+      "loss": 0.7989,
+      "step": 5242
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5163556528107645e-06,
+      "loss": 0.3432,
+      "step": 5243
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5035546375197006e-06,
+      "loss": 0.7536,
+      "step": 5244
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4907858527354112e-06,
+      "loss": 0.803,
+      "step": 5245
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.47804930267902e-06,
+      "loss": 0.8104,
+      "step": 5246
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4653449915609694e-06,
+      "loss": 0.7872,
+      "step": 5247
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4526729235810896e-06,
+      "loss": 0.7713,
+      "step": 5248
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4400331029285073e-06,
+      "loss": 0.8103,
+      "step": 5249
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.427425533781746e-06,
+      "loss": 0.8613,
+      "step": 5250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4148502203086065e-06,
+      "loss": 0.7725,
+      "step": 5251
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4023071666662624e-06,
+      "loss": 0.7728,
+      "step": 5252
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3897963770011967e-06,
+      "loss": 0.8109,
+      "step": 5253
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.377317855449268e-06,
+      "loss": 0.832,
+      "step": 5254
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3648716061356326e-06,
+      "loss": 0.7658,
+      "step": 5255
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3524576331747762e-06,
+      "loss": 0.7855,
+      "step": 5256
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3400759406705496e-06,
+      "loss": 0.7417,
+      "step": 5257
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3277265327160904e-06,
+      "loss": 0.8202,
+      "step": 5258
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3154094133938987e-06,
+      "loss": 0.7813,
+      "step": 5259
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3031245867757734e-06,
+      "loss": 0.8242,
+      "step": 5260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2908720569228435e-06,
+      "loss": 0.7413,
+      "step": 5261
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2786518278855807e-06,
+      "loss": 0.7739,
+      "step": 5262
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.266463903703764e-06,
+      "loss": 0.7568,
+      "step": 5263
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2543082884064815e-06,
+      "loss": 0.8127,
+      "step": 5264
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2421849860121637e-06,
+      "loss": 0.8075,
+      "step": 5265
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2300940005285374e-06,
+      "loss": 0.8176,
+      "step": 5266
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.218035335952673e-06,
+      "loss": 0.7811,
+      "step": 5267
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2060089962709253e-06,
+      "loss": 0.8261,
+      "step": 5268
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.194014985458992e-06,
+      "loss": 0.754,
+      "step": 5269
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.182053307481857e-06,
+      "loss": 0.8226,
+      "step": 5270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1701239662938354e-06,
+      "loss": 0.7993,
+      "step": 5271
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.158226965838539e-06,
+      "loss": 0.7706,
+      "step": 5272
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1463623100488994e-06,
+      "loss": 0.7605,
+      "step": 5273
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.134530002847146e-06,
+      "loss": 0.7328,
+      "step": 5274
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1227300481448276e-06,
+      "loss": 0.8128,
+      "step": 5275
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1109624498427794e-06,
+      "loss": 0.8103,
+      "step": 5276
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.099227211831145e-06,
+      "loss": 0.7879,
+      "step": 5277
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0875243379893883e-06,
+      "loss": 0.783,
+      "step": 5278
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0758538321862584e-06,
+      "loss": 0.7544,
+      "step": 5279
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0642156982798144e-06,
+      "loss": 0.7484,
+      "step": 5280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.052609940117389e-06,
+      "loss": 0.7515,
+      "step": 5281
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0410365615356365e-06,
+      "loss": 0.8099,
+      "step": 5282
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0294955663605066e-06,
+      "loss": 0.7614,
+      "step": 5283
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0179869584072254e-06,
+      "loss": 0.8413,
+      "step": 5284
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0065107414803275e-06,
+      "loss": 0.7255,
+      "step": 5285
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.995066919373645e-06,
+      "loss": 0.7645,
+      "step": 5286
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9836554958702956e-06,
+      "loss": 0.8096,
+      "step": 5287
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9722764747426515e-06,
+      "loss": 0.8095,
+      "step": 5288
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.960929859752436e-06,
+      "loss": 0.3728,
+      "step": 5289
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9496156546506274e-06,
+      "loss": 0.7877,
+      "step": 5290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9383338631774774e-06,
+      "loss": 0.7814,
+      "step": 5291
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.927084489062547e-06,
+      "loss": 0.7885,
+      "step": 5292
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9158675360246513e-06,
+      "loss": 0.8184,
+      "step": 5293
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9046830077719236e-06,
+      "loss": 0.8428,
+      "step": 5294
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8935309080017617e-06,
+      "loss": 0.7782,
+      "step": 5295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8824112404008275e-06,
+      "loss": 0.8489,
+      "step": 5296
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8713240086450923e-06,
+      "loss": 0.8245,
+      "step": 5297
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8602692163997681e-06,
+      "loss": 0.7794,
+      "step": 5298
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8492468673193874e-06,
+      "loss": 0.7698,
+      "step": 5299
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8382569650477133e-06,
+      "loss": 0.7533,
+      "step": 5300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8272995132178061e-06,
+      "loss": 0.7738,
+      "step": 5301
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8163745154520129e-06,
+      "loss": 0.8111,
+      "step": 5302
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8054819753619112e-06,
+      "loss": 0.7825,
+      "step": 5303
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7946218965483763e-06,
+      "loss": 0.8114,
+      "step": 5304
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7837942826015585e-06,
+      "loss": 0.76,
+      "step": 5305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7729991371008502e-06,
+      "loss": 0.7257,
+      "step": 5306
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7622364636149413e-06,
+      "loss": 0.7771,
+      "step": 5307
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7515062657017632e-06,
+      "loss": 0.7584,
+      "step": 5308
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.740808546908501e-06,
+      "loss": 0.8037,
+      "step": 5309
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7301433107716592e-06,
+      "loss": 0.8143,
+      "step": 5310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7195105608169393e-06,
+      "loss": 0.7431,
+      "step": 5311
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.708910300559341e-06,
+      "loss": 0.7795,
+      "step": 5312
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6983425335031055e-06,
+      "loss": 0.8061,
+      "step": 5313
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6878072631417386e-06,
+      "loss": 0.7973,
+      "step": 5314
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6773044929580205e-06,
+      "loss": 0.8171,
+      "step": 5315
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6668342264239522e-06,
+      "loss": 0.8586,
+      "step": 5316
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6563964670008093e-06,
+      "loss": 0.8176,
+      "step": 5317
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6459912181391312e-06,
+      "loss": 0.7335,
+      "step": 5318
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6356184832786892e-06,
+      "loss": 0.784,
+      "step": 5319
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6252782658485178e-06,
+      "loss": 0.7735,
+      "step": 5320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6149705692669158e-06,
+      "loss": 0.8045,
+      "step": 5321
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6046953969413915e-06,
+      "loss": 0.764,
+      "step": 5322
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5944527522687492e-06,
+      "loss": 0.8399,
+      "step": 5323
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5842426386349917e-06,
+      "loss": 0.7708,
+      "step": 5324
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5740650594154194e-06,
+      "loss": 0.8036,
+      "step": 5325
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5639200179745184e-06,
+      "loss": 0.7712,
+      "step": 5326
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5538075176660727e-06,
+      "loss": 0.8045,
+      "step": 5327
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.543727561833086e-06,
+      "loss": 0.3251,
+      "step": 5328
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5336801538077928e-06,
+      "loss": 0.8485,
+      "step": 5329
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5236652969116804e-06,
+      "loss": 0.3409,
+      "step": 5330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5136829944554788e-06,
+      "loss": 0.7511,
+      "step": 5331
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5037332497391588e-06,
+      "loss": 0.7829,
+      "step": 5332
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4938160660519007e-06,
+      "loss": 0.3329,
+      "step": 5333
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4839314466721599e-06,
+      "loss": 0.7954,
+      "step": 5334
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.474079394867589e-06,
+      "loss": 0.7932,
+      "step": 5335
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4642599138951163e-06,
+      "loss": 0.7991,
+      "step": 5336
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.454473007000845e-06,
+      "loss": 0.7867,
+      "step": 5337
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.444718677420176e-06,
+      "loss": 0.7489,
+      "step": 5338
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4349969283776965e-06,
+      "loss": 0.7792,
+      "step": 5339
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4253077630872357e-06,
+      "loss": 0.7807,
+      "step": 5340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.415651184751854e-06,
+      "loss": 0.7176,
+      "step": 5341
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4060271965638194e-06,
+      "loss": 0.3289,
+      "step": 5342
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3964358017046764e-06,
+      "loss": 0.7234,
+      "step": 5343
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3868770033451328e-06,
+      "loss": 0.7724,
+      "step": 5344
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.377350804645161e-06,
+      "loss": 0.7436,
+      "step": 5345
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.367857208753931e-06,
+      "loss": 0.6703,
+      "step": 5346
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3583962188098654e-06,
+      "loss": 0.7466,
+      "step": 5347
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3489678379405956e-06,
+      "loss": 0.809,
+      "step": 5348
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3395720692629398e-06,
+      "loss": 0.7994,
+      "step": 5349
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3302089158829912e-06,
+      "loss": 0.7782,
+      "step": 5350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.320878380896029e-06,
+      "loss": 0.8378,
+      "step": 5351
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3115804673865306e-06,
+      "loss": 0.7821,
+      "step": 5352
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3023151784282484e-06,
+      "loss": 0.7476,
+      "step": 5353
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2930825170840877e-06,
+      "loss": 0.8113,
+      "step": 5354
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.283882486406207e-06,
+      "loss": 0.7743,
+      "step": 5355
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2747150894359738e-06,
+      "loss": 0.8377,
+      "step": 5356
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2655803292039303e-06,
+      "loss": 0.7333,
+      "step": 5357
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.256478208729883e-06,
+      "loss": 0.8134,
+      "step": 5358
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2474087310228143e-06,
+      "loss": 0.728,
+      "step": 5359
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2383718990809146e-06,
+      "loss": 0.8074,
+      "step": 5360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2293677158916162e-06,
+      "loss": 0.741,
+      "step": 5361
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2203961844315048e-06,
+      "loss": 0.7759,
+      "step": 5362
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2114573076664193e-06,
+      "loss": 0.8231,
+      "step": 5363
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2025510885513847e-06,
+      "loss": 0.791,
+      "step": 5364
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1936775300306346e-06,
+      "loss": 0.7872,
+      "step": 5365
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1848366350375895e-06,
+      "loss": 0.7529,
+      "step": 5366
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1760284064949001e-06,
+      "loss": 0.7683,
+      "step": 5367
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1672528473143818e-06,
+      "loss": 0.787,
+      "step": 5368
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1585099603970917e-06,
+      "loss": 0.3426,
+      "step": 5369
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1497997486332513e-06,
+      "loss": 0.8261,
+      "step": 5370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1411222149023015e-06,
+      "loss": 0.8255,
+      "step": 5371
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1324773620728702e-06,
+      "loss": 0.7641,
+      "step": 5372
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.123865193002782e-06,
+      "loss": 0.7951,
+      "step": 5373
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1152857105390602e-06,
+      "loss": 0.7895,
+      "step": 5374
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1067389175179354e-06,
+      "loss": 0.8117,
+      "step": 5375
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0982248167647923e-06,
+      "loss": 0.7222,
+      "step": 5376
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0897434110942574e-06,
+      "loss": 0.8252,
+      "step": 5377
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0812947033101207e-06,
+      "loss": 0.8368,
+      "step": 5378
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0728786962053595e-06,
+      "loss": 0.8243,
+      "step": 5379
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0644953925621482e-06,
+      "loss": 0.8287,
+      "step": 5380
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0561447951518477e-06,
+      "loss": 0.7355,
+      "step": 5381
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0478269067350166e-06,
+      "loss": 0.7816,
+      "step": 5382
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0395417300614107e-06,
+      "loss": 0.782,
+      "step": 5383
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0312892678699281e-06,
+      "loss": 0.7939,
+      "step": 5384
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0230695228886866e-06,
+      "loss": 0.7718,
+      "step": 5385
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0148824978349792e-06,
+      "loss": 0.7857,
+      "step": 5386
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0067281954152851e-06,
+      "loss": 0.8189,
+      "step": 5387
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.986066183252818e-07,
+      "loss": 0.8293,
+      "step": 5388
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.90517769249788e-07,
+      "loss": 0.7798,
+      "step": 5389
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.824616508628315e-07,
+      "loss": 0.8394,
+      "step": 5390
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.744382658276263e-07,
+      "loss": 0.7652,
+      "step": 5391
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.664476167965397e-07,
+      "loss": 0.7434,
+      "step": 5392
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.584897064111364e-07,
+      "loss": 0.8444,
+      "step": 5393
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.505645373021455e-07,
+      "loss": 0.8303,
+      "step": 5394
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.426721120895043e-07,
+      "loss": 0.8222,
+      "step": 5395
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.348124333822706e-07,
+      "loss": 0.7183,
+      "step": 5396
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.269855037787322e-07,
+      "loss": 0.8245,
+      "step": 5397
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.191913258663199e-07,
+      "loss": 0.8177,
+      "step": 5398
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.114299022216388e-07,
+      "loss": 0.7733,
+      "step": 5399
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.03701235410459e-07,
+      "loss": 0.7846,
+      "step": 5400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.960053279877589e-07,
+      "loss": 0.7858,
+      "step": 5401
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.883421824976479e-07,
+      "loss": 0.7386,
+      "step": 5402
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.807118014734217e-07,
+      "loss": 0.7098,
+      "step": 5403
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.731141874375403e-07,
+      "loss": 0.3313,
+      "step": 5404
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.655493429016281e-07,
+      "loss": 0.7808,
+      "step": 5405
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.580172703664846e-07,
+      "loss": 0.7231,
+      "step": 5406
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.505179723220846e-07,
+      "loss": 0.712,
+      "step": 5407
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.430514512475452e-07,
+      "loss": 0.7663,
+      "step": 5408
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.356177096111695e-07,
+      "loss": 0.7952,
+      "step": 5409
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.282167498703918e-07,
+      "loss": 0.7875,
+      "step": 5410
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.208485744718552e-07,
+      "loss": 0.7474,
+      "step": 5411
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.135131858513223e-07,
+      "loss": 0.8046,
+      "step": 5412
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.062105864337533e-07,
+      "loss": 0.7883,
+      "step": 5413
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.989407786332393e-07,
+      "loss": 0.7176,
+      "step": 5414
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.917037648530356e-07,
+      "loss": 0.7181,
+      "step": 5415
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.844995474855843e-07,
+      "loss": 0.8334,
+      "step": 5416
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.773281289124468e-07,
+      "loss": 0.7943,
+      "step": 5417
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.701895115043822e-07,
+      "loss": 0.7645,
+      "step": 5418
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.630836976212696e-07,
+      "loss": 0.7305,
+      "step": 5419
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.560106896121522e-07,
+      "loss": 0.8534,
+      "step": 5420
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.489704898152483e-07,
+      "loss": 0.8332,
+      "step": 5421
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.419631005579075e-07,
+      "loss": 0.7984,
+      "step": 5422
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.349885241566434e-07,
+      "loss": 0.7934,
+      "step": 5423
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.280467629171339e-07,
+      "loss": 0.7654,
+      "step": 5424
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.211378191341766e-07,
+      "loss": 0.7273,
+      "step": 5425
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.142616950917446e-07,
+      "loss": 0.7402,
+      "step": 5426
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.074183930629641e-07,
+      "loss": 0.8173,
+      "step": 5427
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.00607915310103e-07,
+      "loss": 0.8271,
+      "step": 5428
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.938302640845717e-07,
+      "loss": 0.7411,
+      "step": 5429
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.870854416269334e-07,
+      "loss": 0.7854,
+      "step": 5430
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.803734501669046e-07,
+      "loss": 0.8048,
+      "step": 5431
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.736942919233436e-07,
+      "loss": 0.7912,
+      "step": 5432
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.670479691042619e-07,
+      "loss": 0.346,
+      "step": 5433
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.604344839068021e-07,
+      "loss": 0.8152,
+      "step": 5434
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.538538385172599e-07,
+      "loss": 0.7691,
+      "step": 5435
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.473060351110727e-07,
+      "loss": 0.8141,
+      "step": 5436
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.407910758528313e-07,
+      "loss": 0.7819,
+      "step": 5437
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.343089628962462e-07,
+      "loss": 0.3894,
+      "step": 5438
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.278596983842033e-07,
+      "loss": 0.7753,
+      "step": 5439
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.214432844486861e-07,
+      "loss": 0.8043,
+      "step": 5440
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.150597232108535e-07,
+      "loss": 0.7761,
+      "step": 5441
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.087090167809839e-07,
+      "loss": 0.7807,
+      "step": 5442
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.023911672585203e-07,
+      "loss": 0.7854,
+      "step": 5443
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.961061767320142e-07,
+      "loss": 0.8054,
+      "step": 5444
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.898540472791592e-07,
+      "loss": 0.773,
+      "step": 5445
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.836347809668019e-07,
+      "loss": 0.7677,
+      "step": 5446
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.774483798509089e-07,
+      "loss": 0.7841,
+      "step": 5447
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.712948459765887e-07,
+      "loss": 0.8439,
+      "step": 5448
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.651741813781031e-07,
+      "loss": 0.8144,
+      "step": 5449
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.590863880788111e-07,
+      "loss": 0.7223,
+      "step": 5450
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.530314680912252e-07,
+      "loss": 0.7125,
+      "step": 5451
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.470094234169998e-07,
+      "loss": 0.3579,
+      "step": 5452
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.41020256046898e-07,
+      "loss": 0.7002,
+      "step": 5453
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.35063967960836e-07,
+      "loss": 0.7422,
+      "step": 5454
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.291405611278499e-07,
+      "loss": 0.7898,
+      "step": 5455
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.232500375060956e-07,
+      "loss": 0.7983,
+      "step": 5456
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.173923990428931e-07,
+      "loss": 0.7581,
+      "step": 5457
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.115676476746489e-07,
+      "loss": 0.7916,
+      "step": 5458
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.05775785326923e-07,
+      "loss": 0.7792,
+      "step": 5459
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.000168139143946e-07,
+      "loss": 0.79,
+      "step": 5460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.942907353408855e-07,
+      "loss": 0.7722,
+      "step": 5461
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.885975514993147e-07,
+      "loss": 0.7921,
+      "step": 5462
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.829372642717433e-07,
+      "loss": 0.8007,
+      "step": 5463
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.773098755293747e-07,
+      "loss": 0.7737,
+      "step": 5464
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7171538713249865e-07,
+      "loss": 0.7899,
+      "step": 5465
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.661538009305577e-07,
+      "loss": 0.8055,
+      "step": 5466
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.606251187621036e-07,
+      "loss": 0.8161,
+      "step": 5467
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5512934245481865e-07,
+      "loss": 0.7744,
+      "step": 5468
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.496664738255052e-07,
+      "loss": 0.7785,
+      "step": 5469
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.442365146800853e-07,
+      "loss": 0.7659,
+      "step": 5470
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.38839466813612e-07,
+      "loss": 0.8034,
+      "step": 5471
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3347533201022474e-07,
+      "loss": 0.6831,
+      "step": 5472
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.281441120432272e-07,
+      "loss": 0.7877,
+      "step": 5473
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2284580867500976e-07,
+      "loss": 0.7945,
+      "step": 5474
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.175804236570935e-07,
+      "loss": 0.3328,
+      "step": 5475
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1234795873013045e-07,
+      "loss": 0.7862,
+      "step": 5476
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.071484156238592e-07,
+      "loss": 0.7737,
+      "step": 5477
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0198179605716033e-07,
+      "loss": 0.7687,
+      "step": 5478
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9684810173801213e-07,
+      "loss": 0.7303,
+      "step": 5479
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9174733436353475e-07,
+      "loss": 0.775,
+      "step": 5480
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.866794956199349e-07,
+      "loss": 0.7627,
+      "step": 5481
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8164458718255025e-07,
+      "loss": 0.3218,
+      "step": 5482
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7664261071581606e-07,
+      "loss": 0.7967,
+      "step": 5483
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7167356787332073e-07,
+      "loss": 0.7764,
+      "step": 5484
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6673746029770583e-07,
+      "loss": 0.7884,
+      "step": 5485
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6183428962077716e-07,
+      "loss": 0.8379,
+      "step": 5486
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.569640574634159e-07,
+      "loss": 0.7989,
+      "step": 5487
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5212676543563416e-07,
+      "loss": 0.8072,
+      "step": 5488
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4732241513656393e-07,
+      "loss": 0.7361,
+      "step": 5489
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4255100815442365e-07,
+      "loss": 0.7566,
+      "step": 5490
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3781254606655154e-07,
+      "loss": 0.8035,
+      "step": 5491
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3310703043938354e-07,
+      "loss": 0.8184,
+      "step": 5492
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2843446282849745e-07,
+      "loss": 0.7127,
+      "step": 5493
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.237948447785466e-07,
+      "loss": 0.8035,
+      "step": 5494
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.191881778233041e-07,
+      "loss": 0.8177,
+      "step": 5495
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.146144634856407e-07,
+      "loss": 0.7259,
+      "step": 5496
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.100737032775469e-07,
+      "loss": 0.8163,
+      "step": 5497
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0556589870012196e-07,
+      "loss": 0.8081,
+      "step": 5498
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0109105124354056e-07,
+      "loss": 0.8106,
+      "step": 5499
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.966491623871193e-07,
+      "loss": 0.7166,
+      "step": 5500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9224023359926133e-07,
+      "loss": 0.7719,
+      "step": 5501
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8786426633747863e-07,
+      "loss": 0.7777,
+      "step": 5502
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.835212620483807e-07,
+      "loss": 0.8109,
+      "step": 5503
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.792112221676857e-07,
+      "loss": 0.3788,
+      "step": 5504
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7493414812020943e-07,
+      "loss": 0.7394,
+      "step": 5505
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7069004131987653e-07,
+      "loss": 0.7412,
+      "step": 5506
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.66478903169709e-07,
+      "loss": 0.8008,
+      "step": 5507
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.623007350618267e-07,
+      "loss": 0.7503,
+      "step": 5508
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.581555383774581e-07,
+      "loss": 0.7981,
+      "step": 5509
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.540433144869292e-07,
+      "loss": 0.708,
+      "step": 5510
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.499640647496526e-07,
+      "loss": 0.7154,
+      "step": 5511
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4591779051416075e-07,
+      "loss": 0.7855,
+      "step": 5512
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.419044931180836e-07,
+      "loss": 0.7611,
+      "step": 5513
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.379241738881377e-07,
+      "loss": 0.7735,
+      "step": 5514
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3397683414013717e-07,
+      "loss": 0.8133,
+      "step": 5515
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.300624751790048e-07,
+      "loss": 0.7586,
+      "step": 5516
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.261810982987611e-07,
+      "loss": 0.747,
+      "step": 5517
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.223327047824908e-07,
+      "loss": 0.7453,
+      "step": 5518
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1851729590242065e-07,
+      "loss": 0.8102,
+      "step": 5519
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1473487291986395e-07,
+      "loss": 0.7197,
+      "step": 5520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1098543708519826e-07,
+      "loss": 0.739,
+      "step": 5521
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0726898963793205e-07,
+      "loss": 0.7835,
+      "step": 5522
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0358553180663818e-07,
+      "loss": 0.7796,
+      "step": 5523
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9993506480900926e-07,
+      "loss": 0.7814,
+      "step": 5524
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9631758985182436e-07,
+      "loss": 0.8303,
+      "step": 5525
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9273310813093804e-07,
+      "loss": 0.7743,
+      "step": 5526
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8918162083133574e-07,
+      "loss": 0.7617,
+      "step": 5527
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8566312912706718e-07,
+      "loss": 0.8244,
+      "step": 5528
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8217763418125756e-07,
+      "loss": 0.8312,
+      "step": 5529
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.78725137146174e-07,
+      "loss": 0.7462,
+      "step": 5530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7530563916313691e-07,
+      "loss": 0.7881,
+      "step": 5531
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7191914136256427e-07,
+      "loss": 0.8707,
+      "step": 5532
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6856564486397163e-07,
+      "loss": 0.8044,
+      "step": 5533
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6524515077597224e-07,
+      "loss": 0.8562,
+      "step": 5534
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6195766019625469e-07,
+      "loss": 0.7642,
+      "step": 5535
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.58703174211583e-07,
+      "loss": 0.8038,
+      "step": 5536
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5548169389786317e-07,
+      "loss": 0.7794,
+      "step": 5537
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5229322032002115e-07,
+      "loss": 0.7336,
+      "step": 5538
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4913775453213596e-07,
+      "loss": 0.7748,
+      "step": 5539
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4601529757732878e-07,
+      "loss": 0.8115,
+      "step": 5540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4292585048784058e-07,
+      "loss": 0.8064,
+      "step": 5541
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3986941428496548e-07,
+      "loss": 0.8074,
+      "step": 5542
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3684598997911746e-07,
+      "loss": 0.787,
+      "step": 5543
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3385557856977483e-07,
+      "loss": 0.8115,
+      "step": 5544
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3089818104552453e-07,
+      "loss": 0.7883,
+      "step": 5545
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.27973798384029e-07,
+      "loss": 0.7948,
+      "step": 5546
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2508243155201493e-07,
+      "loss": 0.8204,
+      "step": 5547
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2222408150532882e-07,
+      "loss": 0.7301,
+      "step": 5548
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.193987491888815e-07,
+      "loss": 0.7484,
+      "step": 5549
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1660643553668138e-07,
+      "loss": 0.7803,
+      "step": 5550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.138471414718123e-07,
+      "loss": 0.7933,
+      "step": 5551
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.111208679064446e-07,
+      "loss": 0.738,
+      "step": 5552
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0842761574183514e-07,
+      "loss": 0.7962,
+      "step": 5553
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0576738586831614e-07,
+      "loss": 0.815,
+      "step": 5554
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0314017916531748e-07,
+      "loss": 0.7325,
+      "step": 5555
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0054599650135555e-07,
+      "loss": 0.7781,
+      "step": 5556
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.798483873398879e-08,
+      "loss": 0.2958,
+      "step": 5557
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.545670670991769e-08,
+      "loss": 0.7757,
+      "step": 5558
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.296160126487374e-08,
+      "loss": 0.8205,
+      "step": 5559
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.049952322370824e-08,
+      "loss": 0.315,
+      "step": 5560
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.807047340033681e-08,
+      "loss": 0.8393,
+      "step": 5561
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.567445259775042e-08,
+      "loss": 0.7773,
+      "step": 5562
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.33114616080266e-08,
+      "loss": 0.7546,
+      "step": 5563
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.09815012123294e-08,
+      "loss": 0.3439,
+      "step": 5564
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.86845721809093e-08,
+      "loss": 0.7196,
+      "step": 5565
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.642067527308116e-08,
+      "loss": 0.8455,
+      "step": 5566
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.418981123723523e-08,
+      "loss": 0.8487,
+      "step": 5567
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.199198081087044e-08,
+      "loss": 0.7783,
+      "step": 5568
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.982718472055006e-08,
+      "loss": 0.7978,
+      "step": 5569
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.769542368190162e-08,
+      "loss": 0.8172,
+      "step": 5570
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.559669839966142e-08,
+      "loss": 0.8218,
+      "step": 5571
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.353100956761893e-08,
+      "loss": 0.7515,
+      "step": 5572
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.1498357868639e-08,
+      "loss": 0.7846,
+      "step": 5573
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.949874397470634e-08,
+      "loss": 0.7767,
+      "step": 5574
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.753216854683663e-08,
+      "loss": 0.7916,
+      "step": 5575
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.559863223515427e-08,
+      "loss": 0.74,
+      "step": 5576
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.3698135678836856e-08,
+      "loss": 0.8456,
+      "step": 5577
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.183067950617071e-08,
+      "loss": 0.7827,
+      "step": 5578
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.999626433448423e-08,
+      "loss": 0.8691,
+      "step": 5579
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.819489077021455e-08,
+      "loss": 0.7328,
+      "step": 5580
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.642655940886309e-08,
+      "loss": 0.7759,
+      "step": 5581
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.469127083498448e-08,
+      "loss": 0.8127,
+      "step": 5582
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.298902562226426e-08,
+      "loss": 0.7885,
+      "step": 5583
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1319824333407864e-08,
+      "loss": 0.7862,
+      "step": 5584
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.968366752024055e-08,
+      "loss": 0.7821,
+      "step": 5585
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.808055572362967e-08,
+      "loss": 0.7477,
+      "step": 5586
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.651048947354019e-08,
+      "loss": 0.7559,
+      "step": 5587
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4973469289012465e-08,
+      "loss": 0.7897,
+      "step": 5588
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.346949567815116e-08,
+      "loss": 0.7663,
+      "step": 5589
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.199856913813637e-08,
+      "loss": 0.7548,
+      "step": 5590
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0560690155245766e-08,
+      "loss": 0.7197,
+      "step": 5591
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.915585920479913e-08,
+      "loss": 0.8701,
+      "step": 5592
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.778407675121386e-08,
+      "loss": 0.7474,
+      "step": 5593
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6445343247982755e-08,
+      "loss": 0.7736,
+      "step": 5594
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5139659137651817e-08,
+      "loss": 0.3362,
+      "step": 5595
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3867024851853546e-08,
+      "loss": 0.779,
+      "step": 5596
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2627440811318067e-08,
+      "loss": 0.8209,
+      "step": 5597
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.142090742580649e-08,
+      "loss": 0.7478,
+      "step": 5598
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0247425094199744e-08,
+      "loss": 0.8049,
+      "step": 5599
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9106994204409755e-08,
+      "loss": 0.7875,
+      "step": 5600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7999615133446056e-08,
+      "loss": 0.8019,
+      "step": 5601
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6925288247393588e-08,
+      "loss": 0.3015,
+      "step": 5602
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5884013901390492e-08,
+      "loss": 0.8465,
+      "step": 5603
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4875792439683623e-08,
+      "loss": 0.742,
+      "step": 5604
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3900624195550826e-08,
+      "loss": 0.8143,
+      "step": 5605
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2958509491389769e-08,
+      "loss": 0.7666,
+      "step": 5606
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.204944863861801e-08,
+      "loss": 0.7577,
+      "step": 5607
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1173441937772922e-08,
+      "loss": 0.7289,
+      "step": 5608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0330489678433975e-08,
+      "loss": 0.7376,
+      "step": 5609
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.52059213927825e-09,
+      "loss": 0.8229,
+      "step": 5610
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.743749588036032e-09,
+      "loss": 0.8424,
+      "step": 5611
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.999962281513006e-09,
+      "loss": 0.3534,
+      "step": 5612
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.289230465601371e-09,
+      "loss": 0.8056,
+      "step": 5613
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.61155437524652e-09,
+      "loss": 0.7621,
+      "step": 5614
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.966934234480359e-09,
+      "loss": 0.722,
+      "step": 5615
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.355370256410197e-09,
+      "loss": 0.7836,
+      "step": 5616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7768626431854425e-09,
+      "loss": 0.817,
+      "step": 5617
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.231411586064216e-09,
+      "loss": 0.8515,
+      "step": 5618
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7190172653689403e-09,
+      "loss": 0.8027,
+      "step": 5619
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2396798504752414e-09,
+      "loss": 0.7285,
+      "step": 5620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7933994998452508e-09,
+      "loss": 0.3402,
+      "step": 5621
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3801763610165064e-09,
+      "loss": 0.861,
+      "step": 5622
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000105705908488e-09,
+      "loss": 0.7849,
+      "step": 5623
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6529022542455252e-09,
+      "loss": 0.7679,
+      "step": 5624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3388515267220846e-09,
+      "loss": 0.8098,
+      "step": 5625
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0578584918374823e-09,
+      "loss": 0.8184,
+      "step": 5626
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.099232424951808e-10,
+      "loss": 0.8045,
+      "step": 5627
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.950458606518439e-10,
+      "loss": 0.7977,
+      "step": 5628
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.132264173395406e-10,
+      "loss": 0.8033,
+      "step": 5629
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6446497266574555e-10,
+      "loss": 0.7923,
+      "step": 5630
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4876157581333872e-10,
+      "loss": 0.7803,
+      "step": 5631
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.611626501840107e-11,
+      "loss": 0.7647,
+      "step": 5632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6529067625725703e-11,
+      "loss": 0.742,
+      "step": 5633
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.3695,
+      "step": 5634
+    }
+  ],
+  "max_steps": 5634,
+  "num_train_epochs": 1,
+  "total_flos": 8920288866992128.0,
+  "trial_name": null,
+  "trial_params": null
+}