diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,13476 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 2241,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.352941176470589e-07,
+      "loss": 3.2565,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4705882352941177e-06,
+      "loss": 2.883,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2058823529411767e-06,
+      "loss": 3.1085,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9411764705882355e-06,
+      "loss": 3.2674,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6764705882352942e-06,
+      "loss": 2.7946,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.411764705882353e-06,
+      "loss": 2.7028,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.147058823529412e-06,
+      "loss": 2.1117,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.882352941176471e-06,
+      "loss": 2.1564,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.61764705882353e-06,
+      "loss": 2.139,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.3529411764705884e-06,
+      "loss": 1.9931,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.088235294117648e-06,
+      "loss": 2.0554,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.823529411764707e-06,
+      "loss": 2.068,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.558823529411764e-06,
+      "loss": 1.9334,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0294117647058824e-05,
+      "loss": 2.0691,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1029411764705883e-05,
+      "loss": 1.937,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1764705882352942e-05,
+      "loss": 1.9159,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.25e-05,
+      "loss": 1.8625,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.323529411764706e-05,
+      "loss": 1.8466,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3970588235294118e-05,
+      "loss": 1.828,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4705882352941177e-05,
+      "loss": 1.8082,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5441176470588237e-05,
+      "loss": 1.7946,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6176470588235296e-05,
+      "loss": 1.8418,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6911764705882355e-05,
+      "loss": 1.8296,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7647058823529414e-05,
+      "loss": 1.6518,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8382352941176472e-05,
+      "loss": 1.8011,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9117647058823528e-05,
+      "loss": 1.7562,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9852941176470586e-05,
+      "loss": 1.7733,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.058823529411765e-05,
+      "loss": 1.6487,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1323529411764707e-05,
+      "loss": 1.7625,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2058823529411766e-05,
+      "loss": 1.6253,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2794117647058825e-05,
+      "loss": 1.7045,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3529411764705884e-05,
+      "loss": 1.7601,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4264705882352942e-05,
+      "loss": 1.6038,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5e-05,
+      "loss": 1.7157,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5735294117647057e-05,
+      "loss": 1.6013,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.647058823529412e-05,
+      "loss": 1.7038,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7205882352941174e-05,
+      "loss": 1.6517,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7941176470588236e-05,
+      "loss": 1.7225,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8676470588235295e-05,
+      "loss": 1.5917,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9411764705882354e-05,
+      "loss": 1.6841,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0147058823529413e-05,
+      "loss": 1.6012,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0882352941176475e-05,
+      "loss": 1.6254,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.161764705882353e-05,
+      "loss": 1.6999,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.235294117647059e-05,
+      "loss": 1.547,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.308823529411765e-05,
+      "loss": 1.7753,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.382352941176471e-05,
+      "loss": 1.5224,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.455882352941177e-05,
+      "loss": 1.6402,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.529411764705883e-05,
+      "loss": 1.6917,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6029411764705886e-05,
+      "loss": 1.5667,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6764705882352945e-05,
+      "loss": 1.6037,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 1.6097,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8235294117647055e-05,
+      "loss": 1.6434,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.897058823529412e-05,
+      "loss": 1.5276,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.970588235294117e-05,
+      "loss": 1.6643,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.044117647058824e-05,
+      "loss": 1.5208,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.11764705882353e-05,
+      "loss": 1.6364,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.1911764705882356e-05,
+      "loss": 1.639,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2647058823529415e-05,
+      "loss": 1.5488,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3382352941176474e-05,
+      "loss": 1.6252,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.411764705882353e-05,
+      "loss": 1.5263,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.485294117647059e-05,
+      "loss": 1.5395,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.558823529411765e-05,
+      "loss": 1.627,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.632352941176471e-05,
+      "loss": 1.6452,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.705882352941177e-05,
+      "loss": 1.5227,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7794117647058826e-05,
+      "loss": 1.6233,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8529411764705885e-05,
+      "loss": 1.4477,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9264705882352944e-05,
+      "loss": 1.6069,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5e-05,
+      "loss": 1.59,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9999973872958805e-05,
+      "loss": 1.5516,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9999895491889815e-05,
+      "loss": 1.6256,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999976485695686e-05,
+      "loss": 1.5784,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9999581968433e-05,
+      "loss": 1.6478,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99993468267005e-05,
+      "loss": 1.5027,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999905943225083e-05,
+      "loss": 1.6159,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99987197856847e-05,
+      "loss": 1.5175,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999832788771203e-05,
+      "loss": 1.5941,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.999788373915195e-05,
+      "loss": 1.5668,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99973873409328e-05,
+      "loss": 1.5394,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999683869409214e-05,
+      "loss": 1.5895,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9996237799776724e-05,
+      "loss": 1.5458,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999558465924252e-05,
+      "loss": 1.5013,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99948792738547e-05,
+      "loss": 1.617,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9994121645087635e-05,
+      "loss": 1.5596,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9993311774524885e-05,
+      "loss": 1.4589,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9992449663859226e-05,
+      "loss": 1.6172,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99915353148926e-05,
+      "loss": 1.4953,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.999056872953615e-05,
+      "loss": 1.5553,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.99895499098102e-05,
+      "loss": 1.5104,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998847885784424e-05,
+      "loss": 1.5164,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998735557587695e-05,
+      "loss": 1.5678,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998618006625617e-05,
+      "loss": 1.6039,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9984952331438916e-05,
+      "loss": 1.556,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998367237399134e-05,
+      "loss": 1.5301,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.998234019658877e-05,
+      "loss": 1.6128,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9980955802015685e-05,
+      "loss": 1.4674,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997951919316567e-05,
+      "loss": 1.554,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997803037304149e-05,
+      "loss": 1.5161,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997648934475503e-05,
+      "loss": 1.4822,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997489611152727e-05,
+      "loss": 1.5932,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9973250676688335e-05,
+      "loss": 1.5728,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9971553043677456e-05,
+      "loss": 1.4865,
+      "step": 101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9969803216042955e-05,
+      "loss": 1.5268,
+      "step": 102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996800119744227e-05,
+      "loss": 1.5735,
+      "step": 103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996614699164191e-05,
+      "loss": 1.4462,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996424060251747e-05,
+      "loss": 1.616,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9962282034053596e-05,
+      "loss": 1.4571,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.996027129034405e-05,
+      "loss": 1.5886,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995820837559158e-05,
+      "loss": 1.6094,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995609329410804e-05,
+      "loss": 1.4322,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.995392605031428e-05,
+      "loss": 1.5582,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.99517066487402e-05,
+      "loss": 1.5643,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994943509402471e-05,
+      "loss": 1.4594,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994711139091573e-05,
+      "loss": 1.4349,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994473554427019e-05,
+      "loss": 1.6046,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.994230755905398e-05,
+      "loss": 1.4707,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993982744034199e-05,
+      "loss": 1.5674,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993729519331808e-05,
+      "loss": 1.4875,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993471082327506e-05,
+      "loss": 1.5457,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.993207433561468e-05,
+      "loss": 1.4774,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9929385735847624e-05,
+      "loss": 1.5594,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992664502959351e-05,
+      "loss": 1.5138,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.992385222258087e-05,
+      "loss": 1.5603,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9921007320647125e-05,
+      "loss": 1.5725,
+      "step": 123
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.991811032973857e-05,
+      "loss": 1.4637,
+      "step": 124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9915161255910405e-05,
+      "loss": 1.5539,
+      "step": 125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.991216010532667e-05,
+      "loss": 1.4468,
+      "step": 126
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.990910688426025e-05,
+      "loss": 1.5316,
+      "step": 127
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.99060015990929e-05,
+      "loss": 1.5733,
+      "step": 128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9902844256315144e-05,
+      "loss": 1.453,
+      "step": 129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9899634862526364e-05,
+      "loss": 1.5325,
+      "step": 130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989637342443472e-05,
+      "loss": 1.6102,
+      "step": 131
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.989305994885712e-05,
+      "loss": 1.4727,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.988969444271931e-05,
+      "loss": 1.4271,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.988627691305572e-05,
+      "loss": 1.554,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.988280736700955e-05,
+      "loss": 1.5001,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.987928581183273e-05,
+      "loss": 1.5589,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.987571225488587e-05,
+      "loss": 1.4295,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9872086703638294e-05,
+      "loss": 1.57,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.986840916566799e-05,
+      "loss": 1.5687,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9864679648661625e-05,
+      "loss": 1.4409,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.986089816041449e-05,
+      "loss": 1.5557,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9857064708830525e-05,
+      "loss": 1.5628,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9853179301922246e-05,
+      "loss": 1.5352,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.984924194781081e-05,
+      "loss": 1.4871,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.984525265472592e-05,
+      "loss": 1.5497,
+      "step": 145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9841211431005854e-05,
+      "loss": 1.4796,
+      "step": 146
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9837118285097426e-05,
+      "loss": 1.5338,
+      "step": 147
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.983297322555598e-05,
+      "loss": 1.5793,
+      "step": 148
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9828776261045364e-05,
+      "loss": 1.5001,
+      "step": 149
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.982452740033793e-05,
+      "loss": 1.5168,
+      "step": 150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.982022665231448e-05,
+      "loss": 1.5871,
+      "step": 151
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9815874025964285e-05,
+      "loss": 1.4673,
+      "step": 152
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.981146953038503e-05,
+      "loss": 1.5134,
+      "step": 153
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9807013174782865e-05,
+      "loss": 1.3908,
+      "step": 154
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.980250496847227e-05,
+      "loss": 1.5338,
+      "step": 155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.979794492087614e-05,
+      "loss": 1.5405,
+      "step": 156
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.979333304152572e-05,
+      "loss": 1.4948,
+      "step": 157
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.97886693400606e-05,
+      "loss": 1.5194,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.978395382622866e-05,
+      "loss": 1.4859,
+      "step": 159
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9779186509886114e-05,
+      "loss": 1.4378,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.977436740099741e-05,
+      "loss": 1.5812,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.976949650963528e-05,
+      "loss": 1.4284,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.97645738459807e-05,
+      "loss": 1.4992,
+      "step": 163
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.975959942032282e-05,
+      "loss": 1.5542,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.975457324305901e-05,
+      "loss": 1.5415,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9749495324694795e-05,
+      "loss": 1.4295,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9744365675843864e-05,
+      "loss": 1.552,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9739184307228014e-05,
+      "loss": 1.5359,
+      "step": 168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9733951229677145e-05,
+      "loss": 1.4142,
+      "step": 169
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.972866645412926e-05,
+      "loss": 1.5704,
+      "step": 170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9723329991630394e-05,
+      "loss": 1.3892,
+      "step": 171
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.971794185333463e-05,
+      "loss": 1.5,
+      "step": 172
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.971250205050404e-05,
+      "loss": 1.5624,
+      "step": 173
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.970701059450872e-05,
+      "loss": 1.4853,
+      "step": 174
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9701467496826696e-05,
+      "loss": 1.4895,
+      "step": 175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9695872769043964e-05,
+      "loss": 1.5435,
+      "step": 176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.969022642285439e-05,
+      "loss": 1.5246,
+      "step": 177
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9684528470059794e-05,
+      "loss": 1.3165,
+      "step": 178
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.96787789225698e-05,
+      "loss": 1.5143,
+      "step": 179
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.967297779240192e-05,
+      "loss": 1.4732,
+      "step": 180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.966712509168146e-05,
+      "loss": 1.4995,
+      "step": 181
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.966122083264151e-05,
+      "loss": 1.5193,
+      "step": 182
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.965526502762295e-05,
+      "loss": 1.4636,
+      "step": 183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9649257689074355e-05,
+      "loss": 1.4632,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9643198829552085e-05,
+      "loss": 1.4715,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.963708846172012e-05,
+      "loss": 1.5037,
+      "step": 186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9630926598350136e-05,
+      "loss": 1.5715,
+      "step": 187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.962471325232143e-05,
+      "loss": 1.5111,
+      "step": 188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9618448436620904e-05,
+      "loss": 1.4312,
+      "step": 189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.961213216434305e-05,
+      "loss": 1.5174,
+      "step": 190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.960576444868992e-05,
+      "loss": 1.537,
+      "step": 191
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.959934530297107e-05,
+      "loss": 1.4463,
+      "step": 192
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9592874740603555e-05,
+      "loss": 1.5609,
+      "step": 193
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.958635277511192e-05,
+      "loss": 1.4145,
+      "step": 194
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.957977942012813e-05,
+      "loss": 1.5078,
+      "step": 195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9573154689391573e-05,
+      "loss": 1.5478,
+      "step": 196
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.956647859674902e-05,
+      "loss": 1.4246,
+      "step": 197
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9559751156154595e-05,
+      "loss": 1.54,
+      "step": 198
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.955297238166974e-05,
+      "loss": 1.5592,
+      "step": 199
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.954614228746321e-05,
+      "loss": 1.472,
+      "step": 200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.953926088781101e-05,
+      "loss": 1.475,
+      "step": 201
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.953232819709639e-05,
+      "loss": 1.414,
+      "step": 202
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.952534422980981e-05,
+      "loss": 1.5544,
+      "step": 203
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.95183090005489e-05,
+      "loss": 1.3353,
+      "step": 204
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.951122252401844e-05,
+      "loss": 1.4659,
+      "step": 205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.950408481503032e-05,
+      "loss": 1.5543,
+      "step": 206
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9496895888503514e-05,
+      "loss": 1.5307,
+      "step": 207
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.948965575946406e-05,
+      "loss": 1.4686,
+      "step": 208
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9482364443045e-05,
+      "loss": 1.5332,
+      "step": 209
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.947502195448639e-05,
+      "loss": 1.4894,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9467628309135215e-05,
+      "loss": 1.4585,
+      "step": 211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.946018352244542e-05,
+      "loss": 1.4921,
+      "step": 212
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9452687609977813e-05,
+      "loss": 1.5562,
+      "step": 213
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9445140587400064e-05,
+      "loss": 1.4537,
+      "step": 214
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.94375424704867e-05,
+      "loss": 1.5181,
+      "step": 215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.942989327511902e-05,
+      "loss": 1.5334,
+      "step": 216
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.942219301728509e-05,
+      "loss": 1.4258,
+      "step": 217
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.941444171307972e-05,
+      "loss": 1.51,
+      "step": 218
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.940663937870438e-05,
+      "loss": 1.5319,
+      "step": 219
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.939878603046723e-05,
+      "loss": 1.458,
+      "step": 220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.939088168478305e-05,
+      "loss": 1.533,
+      "step": 221
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9382926358173224e-05,
+      "loss": 1.4652,
+      "step": 222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.937492006726567e-05,
+      "loss": 1.4763,
+      "step": 223
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.936686282879486e-05,
+      "loss": 1.5218,
+      "step": 224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.935875465960172e-05,
+      "loss": 1.4077,
+      "step": 225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.935059557663365e-05,
+      "loss": 1.4505,
+      "step": 226
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.934238559694448e-05,
+      "loss": 1.5778,
+      "step": 227
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9334124737694395e-05,
+      "loss": 1.2911,
+      "step": 228
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.932581301614994e-05,
+      "loss": 1.5432,
+      "step": 229
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.931745044968398e-05,
+      "loss": 1.5291,
+      "step": 230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.930903705577563e-05,
+      "loss": 1.4299,
+      "step": 231
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.930057285201027e-05,
+      "loss": 1.5582,
+      "step": 232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.929205785607947e-05,
+      "loss": 1.4146,
+      "step": 233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9283492085780946e-05,
+      "loss": 1.4921,
+      "step": 234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.927487555901857e-05,
+      "loss": 1.4607,
+      "step": 235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.926620829380228e-05,
+      "loss": 1.5377,
+      "step": 236
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.925749030824809e-05,
+      "loss": 1.3965,
+      "step": 237
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.924872162057801e-05,
+      "loss": 1.4986,
+      "step": 238
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9239902249120015e-05,
+      "loss": 1.4337,
+      "step": 239
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.923103221230804e-05,
+      "loss": 1.4982,
+      "step": 240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.922211152868191e-05,
+      "loss": 1.5202,
+      "step": 241
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.921314021688732e-05,
+      "loss": 1.4415,
+      "step": 242
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9204118295675764e-05,
+      "loss": 1.5424,
+      "step": 243
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.919504578390454e-05,
+      "loss": 1.4675,
+      "step": 244
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.918592270053667e-05,
+      "loss": 1.4414,
+      "step": 245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.917674906464089e-05,
+      "loss": 1.5496,
+      "step": 246
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.91675248953916e-05,
+      "loss": 1.508,
+      "step": 247
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9158250212068817e-05,
+      "loss": 1.4451,
+      "step": 248
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.914892503405815e-05,
+      "loss": 1.4014,
+      "step": 249
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.913954938085073e-05,
+      "loss": 1.5232,
+      "step": 250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.913012327204323e-05,
+      "loss": 1.3619,
+      "step": 251
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.912064672733773e-05,
+      "loss": 1.5114,
+      "step": 252
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.911111976654177e-05,
+      "loss": 1.4462,
+      "step": 253
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.910154240956824e-05,
+      "loss": 1.5089,
+      "step": 254
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.909191467643539e-05,
+      "loss": 1.5276,
+      "step": 255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.908223658726676e-05,
+      "loss": 1.4794,
+      "step": 256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.907250816229113e-05,
+      "loss": 1.4311,
+      "step": 257
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.90627294218425e-05,
+      "loss": 1.4991,
+      "step": 258
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9052900386360024e-05,
+      "loss": 1.4712,
+      "step": 259
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9043021076387995e-05,
+      "loss": 1.484,
+      "step": 260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.90330915125758e-05,
+      "loss": 1.5606,
+      "step": 261
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.902311171567783e-05,
+      "loss": 1.4648,
+      "step": 262
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9013081706553497e-05,
+      "loss": 1.4823,
+      "step": 263
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.900300150616716e-05,
+      "loss": 1.5159,
+      "step": 264
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.899287113558808e-05,
+      "loss": 1.4352,
+      "step": 265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8982690615990386e-05,
+      "loss": 1.516,
+      "step": 266
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.897245996865303e-05,
+      "loss": 1.4969,
+      "step": 267
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.896217921495972e-05,
+      "loss": 1.3826,
+      "step": 268
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8951848376398944e-05,
+      "loss": 1.4228,
+      "step": 269
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.894146747456381e-05,
+      "loss": 1.4415,
+      "step": 270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8931036531152116e-05,
+      "loss": 1.5009,
+      "step": 271
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.892055556796622e-05,
+      "loss": 1.5099,
+      "step": 272
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.891002460691306e-05,
+      "loss": 1.4112,
+      "step": 273
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.889944367000406e-05,
+      "loss": 1.5301,
+      "step": 274
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8888812779355114e-05,
+      "loss": 1.4626,
+      "step": 275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8878131957186506e-05,
+      "loss": 1.4526,
+      "step": 276
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.886740122582291e-05,
+      "loss": 1.5097,
+      "step": 277
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8856620607693295e-05,
+      "loss": 1.5032,
+      "step": 278
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8845790125330924e-05,
+      "loss": 1.4411,
+      "step": 279
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8834909801373264e-05,
+      "loss": 1.502,
+      "step": 280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.882397965856198e-05,
+      "loss": 1.5253,
+      "step": 281
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.881299971974285e-05,
+      "loss": 1.3615,
+      "step": 282
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.880197000786574e-05,
+      "loss": 1.5476,
+      "step": 283
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8790890545984536e-05,
+      "loss": 1.5046,
+      "step": 284
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.877976135725715e-05,
+      "loss": 1.4307,
+      "step": 285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.876858246494538e-05,
+      "loss": 1.5138,
+      "step": 286
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.875735389241494e-05,
+      "loss": 1.4925,
+      "step": 287
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.874607566313539e-05,
+      "loss": 1.4487,
+      "step": 288
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.873474780068007e-05,
+      "loss": 1.5032,
+      "step": 289
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8723370328726046e-05,
+      "loss": 1.4531,
+      "step": 290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.871194327105411e-05,
+      "loss": 1.5292,
+      "step": 291
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.870046665154867e-05,
+      "loss": 1.4371,
+      "step": 292
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8688940494197744e-05,
+      "loss": 1.4174,
+      "step": 293
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.867736482309287e-05,
+      "loss": 1.519,
+      "step": 294
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.866573966242911e-05,
+      "loss": 1.441,
+      "step": 295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.865406503650492e-05,
+      "loss": 1.4788,
+      "step": 296
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.86423409697222e-05,
+      "loss": 1.5159,
+      "step": 297
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8630567486586145e-05,
+      "loss": 1.3679,
+      "step": 298
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8618744611705276e-05,
+      "loss": 1.3981,
+      "step": 299
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8606872369791316e-05,
+      "loss": 1.531,
+      "step": 300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.85949507856592e-05,
+      "loss": 1.3788,
+      "step": 301
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.858297988422698e-05,
+      "loss": 1.5267,
+      "step": 302
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8570959690515795e-05,
+      "loss": 1.5267,
+      "step": 303
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.855889022964981e-05,
+      "loss": 1.4311,
+      "step": 304
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.854677152685618e-05,
+      "loss": 1.4613,
+      "step": 305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.853460360746497e-05,
+      "loss": 1.5154,
+      "step": 306
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8522386496909103e-05,
+      "loss": 1.4939,
+      "step": 307
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8510120220724355e-05,
+      "loss": 1.3653,
+      "step": 308
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.849780480454924e-05,
+      "loss": 1.5009,
+      "step": 309
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.848544027412497e-05,
+      "loss": 1.4517,
+      "step": 310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8473026655295465e-05,
+      "loss": 1.4596,
+      "step": 311
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.84605639740072e-05,
+      "loss": 1.5651,
+      "step": 312
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8448052256309205e-05,
+      "loss": 1.4188,
+      "step": 313
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8435491528353026e-05,
+      "loss": 1.4759,
+      "step": 314
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.842288181639263e-05,
+      "loss": 1.459,
+      "step": 315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8410223146784375e-05,
+      "loss": 1.5034,
+      "step": 316
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.839751554598695e-05,
+      "loss": 1.2514,
+      "step": 317
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.838475904056131e-05,
+      "loss": 1.4847,
+      "step": 318
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.837195365717064e-05,
+      "loss": 1.4648,
+      "step": 319
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8359099422580285e-05,
+      "loss": 1.4689,
+      "step": 320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.834619636365769e-05,
+      "loss": 1.4292,
+      "step": 321
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.833324450737235e-05,
+      "loss": 1.5108,
+      "step": 322
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8320243880795765e-05,
+      "loss": 1.4846,
+      "step": 323
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.830719451110137e-05,
+      "loss": 1.4446,
+      "step": 324
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8294096425564475e-05,
+      "loss": 1.5137,
+      "step": 325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.828094965156221e-05,
+      "loss": 1.4435,
+      "step": 326
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8267754216573504e-05,
+      "loss": 1.4847,
+      "step": 327
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8254510148178945e-05,
+      "loss": 1.5375,
+      "step": 328
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.824121747406081e-05,
+      "loss": 1.4065,
+      "step": 329
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.822787622200295e-05,
+      "loss": 1.3803,
+      "step": 330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.821448641989078e-05,
+      "loss": 1.4778,
+      "step": 331
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.820104809571115e-05,
+      "loss": 1.5333,
+      "step": 332
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8187561277552374e-05,
+      "loss": 1.3852,
+      "step": 333
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8174025993604076e-05,
+      "loss": 1.5276,
+      "step": 334
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.816044227215724e-05,
+      "loss": 1.4704,
+      "step": 335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8146810141604046e-05,
+      "loss": 1.4479,
+      "step": 336
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.813312963043787e-05,
+      "loss": 1.4775,
+      "step": 337
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.811940076725322e-05,
+      "loss": 1.4432,
+      "step": 338
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.810562358074566e-05,
+      "loss": 1.4761,
+      "step": 339
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.809179809971177e-05,
+      "loss": 1.5084,
+      "step": 340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.807792435304904e-05,
+      "loss": 1.2421,
+      "step": 341
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8064002369755894e-05,
+      "loss": 1.4824,
+      "step": 342
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.805003217893153e-05,
+      "loss": 1.4938,
+      "step": 343
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.803601380977594e-05,
+      "loss": 1.4094,
+      "step": 344
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.802194729158979e-05,
+      "loss": 1.4963,
+      "step": 345
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.800783265377441e-05,
+      "loss": 1.4535,
+      "step": 346
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.799366992583171e-05,
+      "loss": 1.4461,
+      "step": 347
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7979459137364077e-05,
+      "loss": 1.5395,
+      "step": 348
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7965200318074397e-05,
+      "loss": 1.352,
+      "step": 349
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.795089349776593e-05,
+      "loss": 1.4936,
+      "step": 350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.793653870634226e-05,
+      "loss": 1.5035,
+      "step": 351
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.792213597380726e-05,
+      "loss": 1.477,
+      "step": 352
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.790768533026497e-05,
+      "loss": 1.4166,
+      "step": 353
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.789318680591961e-05,
+      "loss": 1.5194,
+      "step": 354
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.787864043107546e-05,
+      "loss": 1.4726,
+      "step": 355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.786404623613683e-05,
+      "loss": 1.4004,
+      "step": 356
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7849404251607946e-05,
+      "loss": 1.5028,
+      "step": 357
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.783471450809297e-05,
+      "loss": 1.4042,
+      "step": 358
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7819977036295846e-05,
+      "loss": 1.4973,
+      "step": 359
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.780519186702031e-05,
+      "loss": 1.467,
+      "step": 360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.779035903116977e-05,
+      "loss": 1.4469,
+      "step": 361
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.777547855974729e-05,
+      "loss": 1.4644,
+      "step": 362
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.776055048385548e-05,
+      "loss": 1.4814,
+      "step": 363
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.774557483469644e-05,
+      "loss": 1.2477,
+      "step": 364
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.773055164357174e-05,
+      "loss": 1.5122,
+      "step": 365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.77154809418823e-05,
+      "loss": 1.3997,
+      "step": 366
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.770036276112835e-05,
+      "loss": 1.4706,
+      "step": 367
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7685197132909336e-05,
+      "loss": 1.4681,
+      "step": 368
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.766998408892392e-05,
+      "loss": 1.4111,
+      "step": 369
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.765472366096985e-05,
+      "loss": 1.4748,
+      "step": 370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7639415880943895e-05,
+      "loss": 1.4591,
+      "step": 371
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7624060780841816e-05,
+      "loss": 1.4189,
+      "step": 372
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.76086583927583e-05,
+      "loss": 1.4974,
+      "step": 373
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.759320874888683e-05,
+      "loss": 1.4495,
+      "step": 374
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7577711881519704e-05,
+      "loss": 1.4156,
+      "step": 375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.756216782304789e-05,
+      "loss": 1.5191,
+      "step": 376
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.754657660596101e-05,
+      "loss": 1.4669,
+      "step": 377
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7530938262847266e-05,
+      "loss": 1.4153,
+      "step": 378
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.751525282639334e-05,
+      "loss": 1.4781,
+      "step": 379
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7499520329384363e-05,
+      "loss": 1.4652,
+      "step": 380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.748374080470381e-05,
+      "loss": 1.4466,
+      "step": 381
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.746791428533347e-05,
+      "loss": 1.4702,
+      "step": 382
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.745204080435335e-05,
+      "loss": 1.4743,
+      "step": 383
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.743612039494163e-05,
+      "loss": 1.4378,
+      "step": 384
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7420153090374555e-05,
+      "loss": 1.4894,
+      "step": 385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.740413892402639e-05,
+      "loss": 1.4387,
+      "step": 386
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.738807792936938e-05,
+      "loss": 1.4437,
+      "step": 387
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.737197013997361e-05,
+      "loss": 1.4668,
+      "step": 388
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.735581558950699e-05,
+      "loss": 1.4398,
+      "step": 389
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7339614311735184e-05,
+      "loss": 1.4094,
+      "step": 390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.732336634052149e-05,
+      "loss": 1.5099,
+      "step": 391
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.730707170982683e-05,
+      "loss": 1.4212,
+      "step": 392
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.729073045370964e-05,
+      "loss": 1.481,
+      "step": 393
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.727434260632582e-05,
+      "loss": 1.4022,
+      "step": 394
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.725790820192864e-05,
+      "loss": 1.3832,
+      "step": 395
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.724142727486869e-05,
+      "loss": 1.4982,
+      "step": 396
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7224899859593795e-05,
+      "loss": 1.4624,
+      "step": 397
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.720832599064896e-05,
+      "loss": 1.4822,
+      "step": 398
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.719170570267628e-05,
+      "loss": 1.4087,
+      "step": 399
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.717503903041486e-05,
+      "loss": 1.4927,
+      "step": 400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.715832600870077e-05,
+      "loss": 1.4497,
+      "step": 401
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.714156667246696e-05,
+      "loss": 1.4265,
+      "step": 402
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.712476105674318e-05,
+      "loss": 1.5374,
+      "step": 403
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.71079091966559e-05,
+      "loss": 1.4549,
+      "step": 404
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7091011127428264e-05,
+      "loss": 1.434,
+      "step": 405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.707406688438001e-05,
+      "loss": 1.4033,
+      "step": 406
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.705707650292735e-05,
+      "loss": 1.4429,
+      "step": 407
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.704004001858297e-05,
+      "loss": 1.5416,
+      "step": 408
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7022957466955904e-05,
+      "loss": 1.3618,
+      "step": 409
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7005828883751466e-05,
+      "loss": 1.4505,
+      "step": 410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.698865430477121e-05,
+      "loss": 1.5145,
+      "step": 411
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6971433765912784e-05,
+      "loss": 1.3776,
+      "step": 412
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6954167303169946e-05,
+      "loss": 1.3783,
+      "step": 413
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6936854952632425e-05,
+      "loss": 1.4341,
+      "step": 414
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.691949675048585e-05,
+      "loss": 1.35,
+      "step": 415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.690209273301169e-05,
+      "loss": 1.4821,
+      "step": 416
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6884642936587196e-05,
+      "loss": 1.3226,
+      "step": 417
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.68671473976853e-05,
+      "loss": 1.5035,
+      "step": 418
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.684960615287453e-05,
+      "loss": 1.4017,
+      "step": 419
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6832019238818934e-05,
+      "loss": 1.4027,
+      "step": 420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6814386692278054e-05,
+      "loss": 1.4867,
+      "step": 421
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.679670855010678e-05,
+      "loss": 1.4668,
+      "step": 422
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.677898484925533e-05,
+      "loss": 1.4594,
+      "step": 423
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.676121562676912e-05,
+      "loss": 1.3911,
+      "step": 424
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.674340091978874e-05,
+      "loss": 1.4997,
+      "step": 425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6725540765549816e-05,
+      "loss": 1.4054,
+      "step": 426
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.670763520138301e-05,
+      "loss": 1.5007,
+      "step": 427
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6689684264713846e-05,
+      "loss": 1.4464,
+      "step": 428
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.667168799306274e-05,
+      "loss": 1.4159,
+      "step": 429
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.665364642404483e-05,
+      "loss": 1.4357,
+      "step": 430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.663555959536994e-05,
+      "loss": 1.4496,
+      "step": 431
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.66174275448425e-05,
+      "loss": 1.4359,
+      "step": 432
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.659925031036145e-05,
+      "loss": 1.491,
+      "step": 433
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.658102792992018e-05,
+      "loss": 1.2767,
+      "step": 434
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.656276044160643e-05,
+      "loss": 1.411,
+      "step": 435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.654444788360226e-05,
+      "loss": 1.4758,
+      "step": 436
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.652609029418389e-05,
+      "loss": 1.382,
+      "step": 437
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.650768771172168e-05,
+      "loss": 1.4852,
+      "step": 438
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.648924017468003e-05,
+      "loss": 1.5023,
+      "step": 439
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6470747721617316e-05,
+      "loss": 1.39,
+      "step": 440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.645221039118577e-05,
+      "loss": 1.4708,
+      "step": 441
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.643362822213145e-05,
+      "loss": 1.4391,
+      "step": 442
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6415001253294115e-05,
+      "loss": 1.4081,
+      "step": 443
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.639632952360719e-05,
+      "loss": 1.4492,
+      "step": 444
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.637761307209761e-05,
+      "loss": 1.4711,
+      "step": 445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.635885193788584e-05,
+      "loss": 1.3838,
+      "step": 446
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.63400461601857e-05,
+      "loss": 1.4615,
+      "step": 447
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.632119577830435e-05,
+      "loss": 1.4221,
+      "step": 448
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6302300831642154e-05,
+      "loss": 1.3975,
+      "step": 449
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.628336135969264e-05,
+      "loss": 1.4829,
+      "step": 450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.62643774020424e-05,
+      "loss": 1.463,
+      "step": 451
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6245348998370996e-05,
+      "loss": 1.4066,
+      "step": 452
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.622627618845091e-05,
+      "loss": 1.5207,
+      "step": 453
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.620715901214742e-05,
+      "loss": 1.4469,
+      "step": 454
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6187997509418554e-05,
+      "loss": 1.3726,
+      "step": 455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6168791720314975e-05,
+      "loss": 1.4778,
+      "step": 456
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.614954168497993e-05,
+      "loss": 1.4319,
+      "step": 457
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.613024744364912e-05,
+      "loss": 1.267,
+      "step": 458
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.611090903665067e-05,
+      "loss": 1.4629,
+      "step": 459
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.609152650440501e-05,
+      "loss": 1.4228,
+      "step": 460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6072099887424804e-05,
+      "loss": 1.4829,
+      "step": 461
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.605262922631483e-05,
+      "loss": 1.4424,
+      "step": 462
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6033114561771966e-05,
+      "loss": 1.4557,
+      "step": 463
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.601355593458505e-05,
+      "loss": 1.4372,
+      "step": 464
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5993953385634794e-05,
+      "loss": 1.4484,
+      "step": 465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5974306955893745e-05,
+      "loss": 1.3254,
+      "step": 466
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.595461668642612e-05,
+      "loss": 1.5284,
+      "step": 467
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.593488261838783e-05,
+      "loss": 1.4588,
+      "step": 468
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5915104793026284e-05,
+      "loss": 1.353,
+      "step": 469
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5895283251680364e-05,
+      "loss": 1.4668,
+      "step": 470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.587541803578034e-05,
+      "loss": 1.4828,
+      "step": 471
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.585550918684774e-05,
+      "loss": 1.3344,
+      "step": 472
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5835556746495325e-05,
+      "loss": 1.535,
+      "step": 473
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5815560756426944e-05,
+      "loss": 1.4413,
+      "step": 474
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5795521258437494e-05,
+      "loss": 1.4273,
+      "step": 475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5775438294412775e-05,
+      "loss": 1.4547,
+      "step": 476
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.5755311906329484e-05,
+      "loss": 1.4024,
+      "step": 477
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.573514213625505e-05,
+      "loss": 1.4502,
+      "step": 478
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.571492902634758e-05,
+      "loss": 1.4351,
+      "step": 479
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.569467261885578e-05,
+      "loss": 1.4262,
+      "step": 480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.567437295611886e-05,
+      "loss": 1.4453,
+      "step": 481
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5654030080566415e-05,
+      "loss": 1.4741,
+      "step": 482
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.563364403471838e-05,
+      "loss": 1.4301,
+      "step": 483
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5613214861184913e-05,
+      "loss": 1.3098,
+      "step": 484
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5592742602666344e-05,
+      "loss": 1.4449,
+      "step": 485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.557222730195302e-05,
+      "loss": 1.4261,
+      "step": 486
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5551669001925274e-05,
+      "loss": 1.4389,
+      "step": 487
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.55310677455533e-05,
+      "loss": 1.4479,
+      "step": 488
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.551042357589711e-05,
+      "loss": 1.511,
+      "step": 489
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.548973653610637e-05,
+      "loss": 1.4668,
+      "step": 490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5469006669420375e-05,
+      "loss": 1.3722,
+      "step": 491
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5448234019167945e-05,
+      "loss": 1.4466,
+      "step": 492
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5427418628767296e-05,
+      "loss": 1.5311,
+      "step": 493
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5406560541726e-05,
+      "loss": 1.4608,
+      "step": 494
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.538565980164085e-05,
+      "loss": 1.4373,
+      "step": 495
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.536471645219783e-05,
+      "loss": 1.4464,
+      "step": 496
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5343730537171944e-05,
+      "loss": 1.349,
+      "step": 497
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.532270210042719e-05,
+      "loss": 1.4824,
+      "step": 498
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5301631185916425e-05,
+      "loss": 1.469,
+      "step": 499
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.528051783768131e-05,
+      "loss": 1.4006,
+      "step": 500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5259362099852186e-05,
+      "loss": 1.4785,
+      "step": 501
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.5238164016648004e-05,
+      "loss": 1.4989,
+      "step": 502
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.521692363237622e-05,
+      "loss": 1.3894,
+      "step": 503
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.519564099143269e-05,
+      "loss": 1.4322,
+      "step": 504
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.517431613830163e-05,
+      "loss": 1.4175,
+      "step": 505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.5152949117555454e-05,
+      "loss": 1.4458,
+      "step": 506
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.513153997385473e-05,
+      "loss": 1.4912,
+      "step": 507
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.511008875194805e-05,
+      "loss": 1.4013,
+      "step": 508
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.5088595496672e-05,
+      "loss": 1.3933,
+      "step": 509
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.5067060252950965e-05,
+      "loss": 1.4505,
+      "step": 510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.504548306579713e-05,
+      "loss": 1.399,
+      "step": 511
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.502386398031034e-05,
+      "loss": 1.5295,
+      "step": 512
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.500220304167802e-05,
+      "loss": 1.2613,
+      "step": 513
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.498050029517506e-05,
+      "loss": 1.4026,
+      "step": 514
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.495875578616375e-05,
+      "loss": 1.5012,
+      "step": 515
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4936969560093654e-05,
+      "loss": 1.4519,
+      "step": 516
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4915141662501567e-05,
+      "loss": 1.3909,
+      "step": 517
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.489327213901133e-05,
+      "loss": 1.4882,
+      "step": 518
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4871361035333836e-05,
+      "loss": 1.3921,
+      "step": 519
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.484940839726686e-05,
+      "loss": 1.4689,
+      "step": 520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4827414270695015e-05,
+      "loss": 1.4391,
+      "step": 521
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4805378701589594e-05,
+      "loss": 1.3979,
+      "step": 522
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.4783301736008556e-05,
+      "loss": 1.471,
+      "step": 523
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.476118342009635e-05,
+      "loss": 1.4851,
+      "step": 524
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.473902380008389e-05,
+      "loss": 1.3799,
+      "step": 525
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.471682292228837e-05,
+      "loss": 1.4714,
+      "step": 526
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.469458083311327e-05,
+      "loss": 1.4557,
+      "step": 527
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.467229757904818e-05,
+      "loss": 1.4001,
+      "step": 528
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.464997320666874e-05,
+      "loss": 1.26,
+      "step": 529
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.462760776263655e-05,
+      "loss": 1.4969,
+      "step": 530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.460520129369902e-05,
+      "loss": 1.4471,
+      "step": 531
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.458275384668933e-05,
+      "loss": 1.4758,
+      "step": 532
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.456026546852632e-05,
+      "loss": 1.3975,
+      "step": 533
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.453773620621436e-05,
+      "loss": 1.4744,
+      "step": 534
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.451516610684331e-05,
+      "loss": 1.4896,
+      "step": 535
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4492555217588336e-05,
+      "loss": 1.4138,
+      "step": 536
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4469903585709914e-05,
+      "loss": 1.4513,
+      "step": 537
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4447211258553636e-05,
+      "loss": 1.4685,
+      "step": 538
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.442447828355017e-05,
+      "loss": 1.4093,
+      "step": 539
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.440170470821515e-05,
+      "loss": 1.4914,
+      "step": 540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.437889058014908e-05,
+      "loss": 1.36,
+      "step": 541
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.43560359470372e-05,
+      "loss": 1.4271,
+      "step": 542
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.433314085664943e-05,
+      "loss": 1.4765,
+      "step": 543
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.431020535684024e-05,
+      "loss": 1.4263,
+      "step": 544
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.428722949554857e-05,
+      "loss": 1.3977,
+      "step": 545
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.426421332079774e-05,
+      "loss": 1.4839,
+      "step": 546
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.42411568806953e-05,
+      "loss": 1.4571,
+      "step": 547
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.4218060223432966e-05,
+      "loss": 1.3693,
+      "step": 548
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.419492339728654e-05,
+      "loss": 1.3074,
+      "step": 549
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4171746450615756e-05,
+      "loss": 1.376,
+      "step": 550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.4148529431864226e-05,
+      "loss": 1.4505,
+      "step": 551
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.41252723895593e-05,
+      "loss": 1.4934,
+      "step": 552
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.410197537231201e-05,
+      "loss": 1.3711,
+      "step": 553
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.407863842881691e-05,
+      "loss": 1.4547,
+      "step": 554
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.405526160785203e-05,
+      "loss": 1.4188,
+      "step": 555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.403184495827873e-05,
+      "loss": 1.4228,
+      "step": 556
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.400838852904166e-05,
+      "loss": 1.4312,
+      "step": 557
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.398489236916856e-05,
+      "loss": 1.4635,
+      "step": 558
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3961356527770266e-05,
+      "loss": 1.4209,
+      "step": 559
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3937781054040505e-05,
+      "loss": 1.437,
+      "step": 560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.391416599725588e-05,
+      "loss": 1.3334,
+      "step": 561
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3890511406775714e-05,
+      "loss": 1.3946,
+      "step": 562
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.386681733204196e-05,
+      "loss": 1.4985,
+      "step": 563
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3843083822579113e-05,
+      "loss": 1.4498,
+      "step": 564
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.381931092799407e-05,
+      "loss": 1.4114,
+      "step": 565
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.379549869797607e-05,
+      "loss": 1.4428,
+      "step": 566
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.377164718229657e-05,
+      "loss": 1.4207,
+      "step": 567
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.374775643080912e-05,
+      "loss": 1.4166,
+      "step": 568
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.37238264934493e-05,
+      "loss": 1.4455,
+      "step": 569
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.369985742023457e-05,
+      "loss": 1.361,
+      "step": 570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.3675849261264236e-05,
+      "loss": 1.4541,
+      "step": 571
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.365180206671925e-05,
+      "loss": 1.4685,
+      "step": 572
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3627715886862175e-05,
+      "loss": 1.3909,
+      "step": 573
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3603590772037067e-05,
+      "loss": 1.4428,
+      "step": 574
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.357942677266935e-05,
+      "loss": 1.4489,
+      "step": 575
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3555223939265744e-05,
+      "loss": 1.4009,
+      "step": 576
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3530982322414104e-05,
+      "loss": 1.469,
+      "step": 577
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.350670197278338e-05,
+      "loss": 1.3071,
+      "step": 578
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3482382941123454e-05,
+      "loss": 1.3181,
+      "step": 579
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3458025278265093e-05,
+      "loss": 1.4803,
+      "step": 580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.343362903511978e-05,
+      "loss": 1.4027,
+      "step": 581
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3409194262679636e-05,
+      "loss": 1.4653,
+      "step": 582
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.338472101201734e-05,
+      "loss": 1.4747,
+      "step": 583
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.336020933428598e-05,
+      "loss": 1.3326,
+      "step": 584
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3335659280718964e-05,
+      "loss": 1.3837,
+      "step": 585
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.331107090262991e-05,
+      "loss": 1.4854,
+      "step": 586
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.328644425141254e-05,
+      "loss": 1.4696,
+      "step": 587
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.326177937854059e-05,
+      "loss": 1.3836,
+      "step": 588
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.323707633556766e-05,
+      "loss": 1.463,
+      "step": 589
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.321233517412715e-05,
+      "loss": 1.3059,
+      "step": 590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.3187555945932124e-05,
+      "loss": 1.4653,
+      "step": 591
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.316273870277521e-05,
+      "loss": 1.4485,
+      "step": 592
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.313788349652851e-05,
+      "loss": 1.3743,
+      "step": 593
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.311299037914346e-05,
+      "loss": 1.509,
+      "step": 594
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.3088059402650745e-05,
+      "loss": 1.3882,
+      "step": 595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.306309061916016e-05,
+      "loss": 1.4503,
+      "step": 596
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.303808408086056e-05,
+      "loss": 1.2727,
+      "step": 597
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.301303984001967e-05,
+      "loss": 1.4776,
+      "step": 598
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2987957948984074e-05,
+      "loss": 1.4035,
+      "step": 599
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2962838460179e-05,
+      "loss": 1.4625,
+      "step": 600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.293768142610828e-05,
+      "loss": 1.3844,
+      "step": 601
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2912486899354235e-05,
+      "loss": 1.4551,
+      "step": 602
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.288725493257753e-05,
+      "loss": 1.461,
+      "step": 603
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2861985578517096e-05,
+      "loss": 1.3434,
+      "step": 604
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.283667888999001e-05,
+      "loss": 1.4484,
+      "step": 605
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.281133491989139e-05,
+      "loss": 1.4764,
+      "step": 606
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2785953721194275e-05,
+      "loss": 1.383,
+      "step": 607
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.27605353469495e-05,
+      "loss": 1.4767,
+      "step": 608
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.273507985028563e-05,
+      "loss": 1.4156,
+      "step": 609
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.27095872844088e-05,
+      "loss": 1.3994,
+      "step": 610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2684057702602645e-05,
+      "loss": 1.4275,
+      "step": 611
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2658491158228154e-05,
+      "loss": 1.4952,
+      "step": 612
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2632887704723586e-05,
+      "loss": 1.3618,
+      "step": 613
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2607247395604335e-05,
+      "loss": 1.4669,
+      "step": 614
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.258157028446283e-05,
+      "loss": 1.5097,
+      "step": 615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.2555856424968434e-05,
+      "loss": 1.396,
+      "step": 616
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.253010587086731e-05,
+      "loss": 1.4474,
+      "step": 617
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2504318675982316e-05,
+      "loss": 1.4696,
+      "step": 618
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2478494894212906e-05,
+      "loss": 1.334,
+      "step": 619
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2452634579535e-05,
+      "loss": 1.3524,
+      "step": 620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.242673778600088e-05,
+      "loss": 1.2579,
+      "step": 621
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.240080456773907e-05,
+      "loss": 1.4445,
+      "step": 622
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2374834978954235e-05,
+      "loss": 1.4406,
+      "step": 623
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2348829073927046e-05,
+      "loss": 1.3832,
+      "step": 624
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.23227869070141e-05,
+      "loss": 1.5153,
+      "step": 625
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2296708532647785e-05,
+      "loss": 1.372,
+      "step": 626
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.227059400533614e-05,
+      "loss": 1.3698,
+      "step": 627
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2244443379662816e-05,
+      "loss": 1.4936,
+      "step": 628
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2218256710286886e-05,
+      "loss": 1.4159,
+      "step": 629
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.219203405194274e-05,
+      "loss": 1.4565,
+      "step": 630
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.216577545944005e-05,
+      "loss": 1.403,
+      "step": 631
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.213948098766355e-05,
+      "loss": 1.399,
+      "step": 632
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.211315069157297e-05,
+      "loss": 1.3659,
+      "step": 633
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.2086784626202935e-05,
+      "loss": 1.4724,
+      "step": 634
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.206038284666284e-05,
+      "loss": 1.5025,
+      "step": 635
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.203394540813669e-05,
+      "loss": 1.353,
+      "step": 636
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.200747236588307e-05,
+      "loss": 1.4558,
+      "step": 637
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.198096377523495e-05,
+      "loss": 1.381,
+      "step": 638
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1954419691599614e-05,
+      "loss": 1.4377,
+      "step": 639
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.192784017045854e-05,
+      "loss": 1.4245,
+      "step": 640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.190122526736726e-05,
+      "loss": 1.429,
+      "step": 641
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.187457503795527e-05,
+      "loss": 1.3758,
+      "step": 642
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.18478895379259e-05,
+      "loss": 1.4329,
+      "step": 643
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1821168823056193e-05,
+      "loss": 1.28,
+      "step": 644
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1794412949196824e-05,
+      "loss": 1.4309,
+      "step": 645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.176762197227194e-05,
+      "loss": 1.4749,
+      "step": 646
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.174079594827904e-05,
+      "loss": 1.3876,
+      "step": 647
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.17139349332889e-05,
+      "loss": 1.4235,
+      "step": 648
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.168703898344544e-05,
+      "loss": 1.411,
+      "step": 649
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.166010815496557e-05,
+      "loss": 1.358,
+      "step": 650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.163314250413913e-05,
+      "loss": 1.5126,
+      "step": 651
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1606142087328734e-05,
+      "loss": 1.3405,
+      "step": 652
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.157910696096966e-05,
+      "loss": 1.4388,
+      "step": 653
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.155203718156974e-05,
+      "loss": 1.4179,
+      "step": 654
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.152493280570922e-05,
+      "loss": 1.4358,
+      "step": 655
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.149779389004069e-05,
+      "loss": 1.3814,
+      "step": 656
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1470620491288905e-05,
+      "loss": 1.4002,
+      "step": 657
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.1443412666250695e-05,
+      "loss": 1.4703,
+      "step": 658
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.141617047179488e-05,
+      "loss": 1.3858,
+      "step": 659
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.138889396486209e-05,
+      "loss": 1.4679,
+      "step": 660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.136158320246466e-05,
+      "loss": 1.3345,
+      "step": 661
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1334238241686566e-05,
+      "loss": 1.4114,
+      "step": 662
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1306859139683225e-05,
+      "loss": 1.428,
+      "step": 663
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.127944595368144e-05,
+      "loss": 1.3722,
+      "step": 664
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1251998740979244e-05,
+      "loss": 1.4686,
+      "step": 665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1224517558945784e-05,
+      "loss": 1.4322,
+      "step": 666
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.119700246502124e-05,
+      "loss": 1.3586,
+      "step": 667
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.116945351671664e-05,
+      "loss": 1.4157,
+      "step": 668
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1141870771613775e-05,
+      "loss": 1.3584,
+      "step": 669
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1114254287365104e-05,
+      "loss": 1.4431,
+      "step": 670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.108660412169357e-05,
+      "loss": 1.4637,
+      "step": 671
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.105892033239255e-05,
+      "loss": 1.4077,
+      "step": 672
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.1031202977325676e-05,
+      "loss": 1.4703,
+      "step": 673
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.100345211442675e-05,
+      "loss": 1.2957,
+      "step": 674
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.09756678016996e-05,
+      "loss": 1.3775,
+      "step": 675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.094785009721799e-05,
+      "loss": 1.3959,
+      "step": 676
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.091999905912545e-05,
+      "loss": 1.3636,
+      "step": 677
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.089211474563521e-05,
+      "loss": 1.4609,
+      "step": 678
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.086419721503003e-05,
+      "loss": 1.4376,
+      "step": 679
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0836246525662114e-05,
+      "loss": 1.4557,
+      "step": 680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.080826273595296e-05,
+      "loss": 1.3841,
+      "step": 681
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.078024590439327e-05,
+      "loss": 1.3889,
+      "step": 682
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.075219608954278e-05,
+      "loss": 1.4088,
+      "step": 683
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.07241133500302e-05,
+      "loss": 1.4128,
+      "step": 684
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.069599774455303e-05,
+      "loss": 1.4473,
+      "step": 685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0667849331877474e-05,
+      "loss": 1.4626,
+      "step": 686
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.063966817083833e-05,
+      "loss": 1.4214,
+      "step": 687
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0611454320338804e-05,
+      "loss": 1.462,
+      "step": 688
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.058320783935047e-05,
+      "loss": 1.4427,
+      "step": 689
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.055492878691307e-05,
+      "loss": 1.3883,
+      "step": 690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.052661722213446e-05,
+      "loss": 1.4059,
+      "step": 691
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.049827320419041e-05,
+      "loss": 1.387,
+      "step": 692
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.046989679232456e-05,
+      "loss": 1.3087,
+      "step": 693
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0441488045848236e-05,
+      "loss": 1.4243,
+      "step": 694
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0413047024140374e-05,
+      "loss": 1.3771,
+      "step": 695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.038457378664734e-05,
+      "loss": 1.4079,
+      "step": 696
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.035606839288285e-05,
+      "loss": 1.3335,
+      "step": 697
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.032753090242783e-05,
+      "loss": 1.4679,
+      "step": 698
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0298961374930306e-05,
+      "loss": 1.365,
+      "step": 699
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.027035987010525e-05,
+      "loss": 1.3951,
+      "step": 700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0241726447734476e-05,
+      "loss": 1.4256,
+      "step": 701
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.021306116766651e-05,
+      "loss": 1.4162,
+      "step": 702
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0184364089816476e-05,
+      "loss": 1.4734,
+      "step": 703
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.015563527416595e-05,
+      "loss": 1.3969,
+      "step": 704
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.0126874780762844e-05,
+      "loss": 1.4269,
+      "step": 705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.0098082669721296e-05,
+      "loss": 1.3978,
+      "step": 706
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.0069259001221503e-05,
+      "loss": 1.4828,
+      "step": 707
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.004040383550965e-05,
+      "loss": 1.4061,
+      "step": 708
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.001151723289774e-05,
+      "loss": 1.3531,
+      "step": 709
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.998259925376351e-05,
+      "loss": 1.4347,
+      "step": 710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9953649958550226e-05,
+      "loss": 1.4747,
+      "step": 711
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.992466940776666e-05,
+      "loss": 1.3505,
+      "step": 712
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.98956576619869e-05,
+      "loss": 1.4707,
+      "step": 713
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.986661478185021e-05,
+      "loss": 1.2867,
+      "step": 714
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9837540828060983e-05,
+      "loss": 1.3927,
+      "step": 715
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.98084358613885e-05,
+      "loss": 1.4612,
+      "step": 716
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.977929994266693e-05,
+      "loss": 1.347,
+      "step": 717
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.975013313279506e-05,
+      "loss": 1.3872,
+      "step": 718
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.972093549273631e-05,
+      "loss": 1.4249,
+      "step": 719
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9691707083518516e-05,
+      "loss": 1.3525,
+      "step": 720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.966244796623382e-05,
+      "loss": 1.4188,
+      "step": 721
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.963315820203855e-05,
+      "loss": 1.4205,
+      "step": 722
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.960383785215311e-05,
+      "loss": 1.4064,
+      "step": 723
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.95744869778618e-05,
+      "loss": 1.3952,
+      "step": 724
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.954510564051276e-05,
+      "loss": 1.4259,
+      "step": 725
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.951569390151778e-05,
+      "loss": 1.375,
+      "step": 726
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.948625182235218e-05,
+      "loss": 1.4414,
+      "step": 727
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9456779464554726e-05,
+      "loss": 1.3874,
+      "step": 728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.942727688972746e-05,
+      "loss": 1.4083,
+      "step": 729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.939774415953558e-05,
+      "loss": 1.4699,
+      "step": 730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.936818133570731e-05,
+      "loss": 1.4225,
+      "step": 731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9338588480033775e-05,
+      "loss": 1.3567,
+      "step": 732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9308965654368897e-05,
+      "loss": 1.4681,
+      "step": 733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9279312920629195e-05,
+      "loss": 1.4069,
+      "step": 734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.924963034079374e-05,
+      "loss": 1.3752,
+      "step": 735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.921991797690396e-05,
+      "loss": 1.4895,
+      "step": 736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9190175891063564e-05,
+      "loss": 1.3903,
+      "step": 737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9160404145438345e-05,
+      "loss": 1.3153,
+      "step": 738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9130602802256136e-05,
+      "loss": 1.4188,
+      "step": 739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.910077192380659e-05,
+      "loss": 1.3863,
+      "step": 740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9070911572441124e-05,
+      "loss": 1.3678,
+      "step": 741
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.904102181057275e-05,
+      "loss": 1.4289,
+      "step": 742
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.901110270067595e-05,
+      "loss": 1.4288,
+      "step": 743
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.898115430528655e-05,
+      "loss": 1.3499,
+      "step": 744
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.895117668700159e-05,
+      "loss": 1.4454,
+      "step": 745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8921169908479175e-05,
+      "loss": 1.2764,
+      "step": 746
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8891134032438384e-05,
+      "loss": 1.4371,
+      "step": 747
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.88610691216591e-05,
+      "loss": 1.4375,
+      "step": 748
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8830975238981896e-05,
+      "loss": 1.3296,
+      "step": 749
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.88008524473079e-05,
+      "loss": 1.4418,
+      "step": 750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8770700809598656e-05,
+      "loss": 1.427,
+      "step": 751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.874052038887603e-05,
+      "loss": 1.3432,
+      "step": 752
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.871031124822202e-05,
+      "loss": 1.4696,
+      "step": 753
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.868007345077866e-05,
+      "loss": 1.4402,
+      "step": 754
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.864980705974789e-05,
+      "loss": 1.3383,
+      "step": 755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.86195121383914e-05,
+      "loss": 1.4491,
+      "step": 756
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.858918875003053e-05,
+      "loss": 1.4089,
+      "step": 757
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.855883695804612e-05,
+      "loss": 1.3733,
+      "step": 758
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8528456825878355e-05,
+      "loss": 1.4156,
+      "step": 759
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8498048417026684e-05,
+      "loss": 1.4105,
+      "step": 760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.846761179504965e-05,
+      "loss": 1.4397,
+      "step": 761
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.843714702356475e-05,
+      "loss": 1.4299,
+      "step": 762
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.840665416624834e-05,
+      "loss": 1.3897,
+      "step": 763
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.837613328683547e-05,
+      "loss": 1.3003,
+      "step": 764
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8345584449119776e-05,
+      "loss": 1.4038,
+      "step": 765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8315007716953285e-05,
+      "loss": 1.384,
+      "step": 766
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8284403154246384e-05,
+      "loss": 1.4369,
+      "step": 767
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8253770824967604e-05,
+      "loss": 1.328,
+      "step": 768
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.822311079314351e-05,
+      "loss": 1.4736,
+      "step": 769
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.819242312285857e-05,
+      "loss": 1.4346,
+      "step": 770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8161707878255046e-05,
+      "loss": 1.4083,
+      "step": 771
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8130965123532805e-05,
+      "loss": 1.4,
+      "step": 772
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8100194922949205e-05,
+      "loss": 1.4532,
+      "step": 773
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.806939734081902e-05,
+      "loss": 1.3833,
+      "step": 774
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.803857244151421e-05,
+      "loss": 1.3394,
+      "step": 775
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.800772028946385e-05,
+      "loss": 1.4109,
+      "step": 776
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.797684094915397e-05,
+      "loss": 1.3723,
+      "step": 777
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.794593448512745e-05,
+      "loss": 1.4423,
+      "step": 778
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7915000961983835e-05,
+      "loss": 1.4741,
+      "step": 779
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.788404044437924e-05,
+      "loss": 1.3549,
+      "step": 780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7853052997026206e-05,
+      "loss": 1.3994,
+      "step": 781
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.782203868469356e-05,
+      "loss": 1.4222,
+      "step": 782
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7790997572206267e-05,
+      "loss": 1.4094,
+      "step": 783
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.775992972444534e-05,
+      "loss": 1.4241,
+      "step": 784
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7728835206347634e-05,
+      "loss": 1.4769,
+      "step": 785
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7697714082905784e-05,
+      "loss": 1.314,
+      "step": 786
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.766656641916801e-05,
+      "loss": 1.4238,
+      "step": 787
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7635392280238025e-05,
+      "loss": 1.3661,
+      "step": 788
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7604191731274874e-05,
+      "loss": 1.3244,
+      "step": 789
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.757296483749278e-05,
+      "loss": 1.4055,
+      "step": 790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.754171166416106e-05,
+      "loss": 1.3838,
+      "step": 791
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7510432276603955e-05,
+      "loss": 1.5015,
+      "step": 792
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7479126740200486e-05,
+      "loss": 1.2878,
+      "step": 793
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.7447795120384345e-05,
+      "loss": 1.3332,
+      "step": 794
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.741643748264372e-05,
+      "loss": 1.4526,
+      "step": 795
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7385053892521205e-05,
+      "loss": 1.4507,
+      "step": 796
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7353644415613624e-05,
+      "loss": 1.3476,
+      "step": 797
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.732220911757192e-05,
+      "loss": 1.3755,
+      "step": 798
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.729074806410099e-05,
+      "loss": 1.3755,
+      "step": 799
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.725926132095957e-05,
+      "loss": 1.3802,
+      "step": 800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7227748953960104e-05,
+      "loss": 1.4282,
+      "step": 801
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7196211028968586e-05,
+      "loss": 1.396,
+      "step": 802
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.716464761190442e-05,
+      "loss": 1.412,
+      "step": 803
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.713305876874031e-05,
+      "loss": 1.4232,
+      "step": 804
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.7101444565502097e-05,
+      "loss": 1.359,
+      "step": 805
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.706980506826863e-05,
+      "loss": 1.4223,
+      "step": 806
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.70381403431716e-05,
+      "loss": 1.4079,
+      "step": 807
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.700645045639549e-05,
+      "loss": 1.3654,
+      "step": 808
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.697473547417732e-05,
+      "loss": 1.2847,
+      "step": 809
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.694299546280657e-05,
+      "loss": 1.4665,
+      "step": 810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.691123048862507e-05,
+      "loss": 1.3672,
+      "step": 811
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6879440618026796e-05,
+      "loss": 1.4604,
+      "step": 812
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6847625917457765e-05,
+      "loss": 1.4562,
+      "step": 813
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6815786453415885e-05,
+      "loss": 1.3612,
+      "step": 814
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6783922292450856e-05,
+      "loss": 1.4632,
+      "step": 815
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.675203350116397e-05,
+      "loss": 1.3189,
+      "step": 816
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6720120146208006e-05,
+      "loss": 1.385,
+      "step": 817
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.668818229428708e-05,
+      "loss": 1.4457,
+      "step": 818
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.665622001215654e-05,
+      "loss": 1.3545,
+      "step": 819
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.662423336662275e-05,
+      "loss": 1.4419,
+      "step": 820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6592222424543035e-05,
+      "loss": 1.3856,
+      "step": 821
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.656018725282549e-05,
+      "loss": 1.3725,
+      "step": 822
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6528127918428854e-05,
+      "loss": 1.4058,
+      "step": 823
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6496044488362366e-05,
+      "loss": 1.4336,
+      "step": 824
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.646393702968565e-05,
+      "loss": 1.3625,
+      "step": 825
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.643180560950853e-05,
+      "loss": 1.4157,
+      "step": 826
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.63996502949909e-05,
+      "loss": 1.4213,
+      "step": 827
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.636747115334265e-05,
+      "loss": 1.3712,
+      "step": 828
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.633526825182343e-05,
+      "loss": 1.2875,
+      "step": 829
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.630304165774256e-05,
+      "loss": 1.474,
+      "step": 830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6270791438458874e-05,
+      "loss": 1.3597,
+      "step": 831
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6238517661380613e-05,
+      "loss": 1.424,
+      "step": 832
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6206220393965236e-05,
+      "loss": 1.3101,
+      "step": 833
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.61738997037193e-05,
+      "loss": 1.4436,
+      "step": 834
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.614155565819833e-05,
+      "loss": 1.4252,
+      "step": 835
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6109188325006674e-05,
+      "loss": 1.334,
+      "step": 836
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.6076797771797324e-05,
+      "loss": 1.4624,
+      "step": 837
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.604438406627183e-05,
+      "loss": 1.4158,
+      "step": 838
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.601194727618015e-05,
+      "loss": 1.3612,
+      "step": 839
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.597948746932044e-05,
+      "loss": 1.403,
+      "step": 840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.594700471353902e-05,
+      "loss": 1.2942,
+      "step": 841
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.591449907673015e-05,
+      "loss": 1.3707,
+      "step": 842
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.588197062683591e-05,
+      "loss": 1.4495,
+      "step": 843
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5849419431846075e-05,
+      "loss": 1.3781,
+      "step": 844
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5816845559797965e-05,
+      "loss": 1.4135,
+      "step": 845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.578424907877629e-05,
+      "loss": 1.3833,
+      "step": 846
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.575163005691302e-05,
+      "loss": 1.3377,
+      "step": 847
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.571898856238722e-05,
+      "loss": 1.3875,
+      "step": 848
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5686324663424966e-05,
+      "loss": 1.4401,
+      "step": 849
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.565363842829913e-05,
+      "loss": 1.3917,
+      "step": 850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.562092992532928e-05,
+      "loss": 1.4001,
+      "step": 851
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.558819922288154e-05,
+      "loss": 1.3964,
+      "step": 852
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.55554463893684e-05,
+      "loss": 1.3409,
+      "step": 853
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.552267149324866e-05,
+      "loss": 1.3618,
+      "step": 854
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.548987460302717e-05,
+      "loss": 1.429,
+      "step": 855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5457055787254823e-05,
+      "loss": 1.3856,
+      "step": 856
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.542421511452828e-05,
+      "loss": 1.3884,
+      "step": 857
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5391352653489913e-05,
+      "loss": 1.3061,
+      "step": 858
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.535846847282763e-05,
+      "loss": 1.3357,
+      "step": 859
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.532556264127475e-05,
+      "loss": 1.4042,
+      "step": 860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5292635227609825e-05,
+      "loss": 1.3873,
+      "step": 861
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.5259686300656525e-05,
+      "loss": 1.4163,
+      "step": 862
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5226715929283506e-05,
+      "loss": 1.4309,
+      "step": 863
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5193724182404195e-05,
+      "loss": 1.3506,
+      "step": 864
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.516071112897675e-05,
+      "loss": 1.365,
+      "step": 865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.5127676838003854e-05,
+      "loss": 1.412,
+      "step": 866
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.509462137853256e-05,
+      "loss": 1.4255,
+      "step": 867
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.506154481965415e-05,
+      "loss": 1.3934,
+      "step": 868
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.502844723050408e-05,
+      "loss": 1.3839,
+      "step": 869
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4995328680261684e-05,
+      "loss": 1.4043,
+      "step": 870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.496218923815015e-05,
+      "loss": 1.4006,
+      "step": 871
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4929028973436325e-05,
+      "loss": 1.425,
+      "step": 872
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.489584795543057e-05,
+      "loss": 1.3257,
+      "step": 873
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4862646253486644e-05,
+      "loss": 1.4569,
+      "step": 874
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4829423937001506e-05,
+      "loss": 1.3938,
+      "step": 875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4796181075415245e-05,
+      "loss": 1.4113,
+      "step": 876
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4762917738210856e-05,
+      "loss": 1.2503,
+      "step": 877
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.472963399491414e-05,
+      "loss": 1.4566,
+      "step": 878
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4696329915093584e-05,
+      "loss": 1.3404,
+      "step": 879
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.466300556836012e-05,
+      "loss": 1.4268,
+      "step": 880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4629661024367096e-05,
+      "loss": 1.3188,
+      "step": 881
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.459629635281004e-05,
+      "loss": 1.4173,
+      "step": 882
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4562911623426574e-05,
+      "loss": 1.4359,
+      "step": 883
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.4529506905996224e-05,
+      "loss": 1.3663,
+      "step": 884
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.449608227034031e-05,
+      "loss": 1.3972,
+      "step": 885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.446263778632179e-05,
+      "loss": 1.4292,
+      "step": 886
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4429173523845085e-05,
+      "loss": 1.3802,
+      "step": 887
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4395689552855955e-05,
+      "loss": 1.3848,
+      "step": 888
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.436218594334139e-05,
+      "loss": 1.3376,
+      "step": 889
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4328662765329394e-05,
+      "loss": 1.3658,
+      "step": 890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.4295120088888896e-05,
+      "loss": 1.4193,
+      "step": 891
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.426155798412954e-05,
+      "loss": 1.4144,
+      "step": 892
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.422797652120163e-05,
+      "loss": 1.319,
+      "step": 893
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.41943757702959e-05,
+      "loss": 1.3842,
+      "step": 894
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.416075580164341e-05,
+      "loss": 1.4274,
+      "step": 895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.412711668551537e-05,
+      "loss": 1.3317,
+      "step": 896
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.409345849222304e-05,
+      "loss": 1.4376,
+      "step": 897
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.405978129211754e-05,
+      "loss": 1.4138,
+      "step": 898
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.40260851555897e-05,
+      "loss": 1.3146,
+      "step": 899
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3992370153069975e-05,
+      "loss": 1.4309,
+      "step": 900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.395863635502821e-05,
+      "loss": 1.2027,
+      "step": 901
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.392488383197355e-05,
+      "loss": 1.4375,
+      "step": 902
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3891112654454285e-05,
+      "loss": 1.4186,
+      "step": 903
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3857322893057694e-05,
+      "loss": 1.2755,
+      "step": 904
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.382351461840988e-05,
+      "loss": 1.4821,
+      "step": 905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.378968790117569e-05,
+      "loss": 1.3333,
+      "step": 906
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.375584281205845e-05,
+      "loss": 1.3744,
+      "step": 907
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.372197942179994e-05,
+      "loss": 1.4433,
+      "step": 908
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.368809780118017e-05,
+      "loss": 1.4146,
+      "step": 909
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.365419802101727e-05,
+      "loss": 1.3906,
+      "step": 910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.362028015216731e-05,
+      "loss": 1.3949,
+      "step": 911
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.358634426552417e-05,
+      "loss": 1.3252,
+      "step": 912
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.355239043201942e-05,
+      "loss": 1.3578,
+      "step": 913
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.351841872262207e-05,
+      "loss": 1.3912,
+      "step": 914
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.348442920833859e-05,
+      "loss": 1.4337,
+      "step": 915
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3450421960212566e-05,
+      "loss": 1.388,
+      "step": 916
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.341639704932474e-05,
+      "loss": 1.4097,
+      "step": 917
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.338235454679272e-05,
+      "loss": 1.4359,
+      "step": 918
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3348294523770884e-05,
+      "loss": 1.3518,
+      "step": 919
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.331421705145026e-05,
+      "loss": 1.4589,
+      "step": 920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.328012220105831e-05,
+      "loss": 1.3119,
+      "step": 921
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.324601004385885e-05,
+      "loss": 1.3888,
+      "step": 922
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.321188065115187e-05,
+      "loss": 1.4045,
+      "step": 923
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.317773409427335e-05,
+      "loss": 1.236,
+      "step": 924
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.3143570444595183e-05,
+      "loss": 1.4315,
+      "step": 925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.310938977352498e-05,
+      "loss": 1.3747,
+      "step": 926
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.307519215250591e-05,
+      "loss": 1.4064,
+      "step": 927
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.304097765301661e-05,
+      "loss": 1.4308,
+      "step": 928
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.300674634657094e-05,
+      "loss": 1.3077,
+      "step": 929
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2972498304717946e-05,
+      "loss": 1.3421,
+      "step": 930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2938233599041615e-05,
+      "loss": 1.457,
+      "step": 931
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2903952301160786e-05,
+      "loss": 1.345,
+      "step": 932
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.286965448272896e-05,
+      "loss": 1.3976,
+      "step": 933
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.283534021543418e-05,
+      "loss": 1.4372,
+      "step": 934
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.280100957099887e-05,
+      "loss": 1.4092,
+      "step": 935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2766662621179684e-05,
+      "loss": 1.3442,
+      "step": 936
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.273229943776736e-05,
+      "loss": 1.4824,
+      "step": 937
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.269792009258655e-05,
+      "loss": 1.3581,
+      "step": 938
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.266352465749571e-05,
+      "loss": 1.2968,
+      "step": 939
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2629113204386914e-05,
+      "loss": 1.416,
+      "step": 940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.259468580518572e-05,
+      "loss": 1.4005,
+      "step": 941
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.256024253185101e-05,
+      "loss": 1.3608,
+      "step": 942
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2525783456374846e-05,
+      "loss": 1.4474,
+      "step": 943
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.249130865078233e-05,
+      "loss": 1.3662,
+      "step": 944
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2456818187131445e-05,
+      "loss": 1.3948,
+      "step": 945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.242231213751286e-05,
+      "loss": 1.4082,
+      "step": 946
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.238779057404989e-05,
+      "loss": 1.3646,
+      "step": 947
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.235325356889822e-05,
+      "loss": 1.4138,
+      "step": 948
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.2318701194245826e-05,
+      "loss": 1.4251,
+      "step": 949
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.228413352231283e-05,
+      "loss": 1.33,
+      "step": 950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.22495506253513e-05,
+      "loss": 1.3957,
+      "step": 951
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.221495257564515e-05,
+      "loss": 1.3297,
+      "step": 952
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.218033944550993e-05,
+      "loss": 1.4207,
+      "step": 953
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.2145711307292755e-05,
+      "loss": 1.2876,
+      "step": 954
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.2111068233372076e-05,
+      "loss": 1.3449,
+      "step": 955
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.2076410296157586e-05,
+      "loss": 1.401,
+      "step": 956
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.2041737568090025e-05,
+      "loss": 1.4021,
+      "step": 957
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.200705012164107e-05,
+      "loss": 1.4055,
+      "step": 958
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1972348029313124e-05,
+      "loss": 1.373,
+      "step": 959
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.193763136363925e-05,
+      "loss": 1.406,
+      "step": 960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.190290019718293e-05,
+      "loss": 1.3307,
+      "step": 961
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1868154602537987e-05,
+      "loss": 1.3969,
+      "step": 962
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.183339465232838e-05,
+      "loss": 1.4304,
+      "step": 963
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.179862041920808e-05,
+      "loss": 1.3642,
+      "step": 964
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.176383197586093e-05,
+      "loss": 1.3882,
+      "step": 965
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.172902939500044e-05,
+      "loss": 1.3981,
+      "step": 966
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.169421274936968e-05,
+      "loss": 1.3506,
+      "step": 967
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.165938211174114e-05,
+      "loss": 1.3932,
+      "step": 968
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.162453755491655e-05,
+      "loss": 1.4338,
+      "step": 969
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.158967915172669e-05,
+      "loss": 1.2828,
+      "step": 970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.155480697503135e-05,
+      "loss": 1.3645,
+      "step": 971
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1519921097719054e-05,
+      "loss": 1.3341,
+      "step": 972
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.148502159270698e-05,
+      "loss": 1.2727,
+      "step": 973
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1450108532940805e-05,
+      "loss": 1.388,
+      "step": 974
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.141518199139451e-05,
+      "loss": 1.3373,
+      "step": 975
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.13802420410703e-05,
+      "loss": 1.4545,
+      "step": 976
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1345288754998345e-05,
+      "loss": 1.3572,
+      "step": 977
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.131032220623673e-05,
+      "loss": 1.3478,
+      "step": 978
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.127534246787126e-05,
+      "loss": 1.3802,
+      "step": 979
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.124034961301529e-05,
+      "loss": 1.2753,
+      "step": 980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.12053437148096e-05,
+      "loss": 1.4224,
+      "step": 981
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.117032484642225e-05,
+      "loss": 1.3601,
+      "step": 982
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1135293081048375e-05,
+      "loss": 1.4221,
+      "step": 983
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.110024849191009e-05,
+      "loss": 1.3353,
+      "step": 984
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1065191152256304e-05,
+      "loss": 1.4153,
+      "step": 985
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.10301211353626e-05,
+      "loss": 1.3303,
+      "step": 986
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0995038514531026e-05,
+      "loss": 1.4223,
+      "step": 987
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.095994336308999e-05,
+      "loss": 1.3686,
+      "step": 988
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0924835754394094e-05,
+      "loss": 1.3902,
+      "step": 989
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0889715761823965e-05,
+      "loss": 1.3238,
+      "step": 990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.085458345878613e-05,
+      "loss": 1.4147,
+      "step": 991
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.081943891871283e-05,
+      "loss": 1.3214,
+      "step": 992
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.078428221506191e-05,
+      "loss": 1.4259,
+      "step": 993
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.07491134213166e-05,
+      "loss": 1.2653,
+      "step": 994
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.071393261098544e-05,
+      "loss": 1.3424,
+      "step": 995
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0678739857602057e-05,
+      "loss": 1.4086,
+      "step": 996
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.064353523472505e-05,
+      "loss": 1.3391,
+      "step": 997
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0608318815937845e-05,
+      "loss": 1.4063,
+      "step": 998
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.057309067484849e-05,
+      "loss": 1.4483,
+      "step": 999
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0537850885089576e-05,
+      "loss": 1.345,
+      "step": 1000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0502599520317986e-05,
+      "loss": 1.3887,
+      "step": 1001
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0467336654214862e-05,
+      "loss": 1.3164,
+      "step": 1002
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.043206236048533e-05,
+      "loss": 1.3456,
+      "step": 1003
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0396776712858427e-05,
+      "loss": 1.4358,
+      "step": 1004
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0361479785086927e-05,
+      "loss": 1.3893,
+      "step": 1005
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0326171650947173e-05,
+      "loss": 1.3529,
+      "step": 1006
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.029085238423891e-05,
+      "loss": 1.3916,
+      "step": 1007
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0255522058785203e-05,
+      "loss": 1.3904,
+      "step": 1008
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.022018074843218e-05,
+      "loss": 1.2912,
+      "step": 1009
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0184828527048965e-05,
+      "loss": 1.4424,
+      "step": 1010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.014946546852746e-05,
+      "loss": 1.3673,
+      "step": 1011
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0114091646782245e-05,
+      "loss": 1.3371,
+      "step": 1012
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.007870713575038e-05,
+      "loss": 1.4538,
+      "step": 1013
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0043312009391255e-05,
+      "loss": 1.369,
+      "step": 1014
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.000790634168649e-05,
+      "loss": 1.3345,
+      "step": 1015
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.99724902066397e-05,
+      "loss": 1.4316,
+      "step": 1016
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9937063678276385e-05,
+      "loss": 1.3012,
+      "step": 1017
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9901626830643774e-05,
+      "loss": 1.3952,
+      "step": 1018
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9866179737810686e-05,
+      "loss": 1.2376,
+      "step": 1019
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9830722473867313e-05,
+      "loss": 1.3712,
+      "step": 1020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9795255112925124e-05,
+      "loss": 1.3111,
+      "step": 1021
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.975977772911671e-05,
+      "loss": 1.4496,
+      "step": 1022
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.972429039659559e-05,
+      "loss": 1.3617,
+      "step": 1023
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.968879318953608e-05,
+      "loss": 1.3589,
+      "step": 1024
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9653286182133146e-05,
+      "loss": 1.3296,
+      "step": 1025
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9617769448602233e-05,
+      "loss": 1.3178,
+      "step": 1026
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9582243063179104e-05,
+      "loss": 1.3714,
+      "step": 1027
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.954670710011971e-05,
+      "loss": 1.4173,
+      "step": 1028
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9511161633700018e-05,
+      "loss": 1.3362,
+      "step": 1029
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.947560673821586e-05,
+      "loss": 1.3601,
+      "step": 1030
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.944004248798276e-05,
+      "loss": 1.3893,
+      "step": 1031
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9404468957335836e-05,
+      "loss": 1.3488,
+      "step": 1032
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.936888622062955e-05,
+      "loss": 1.4437,
+      "step": 1033
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9333294352237637e-05,
+      "loss": 1.3774,
+      "step": 1034
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9297693426552924e-05,
+      "loss": 1.3973,
+      "step": 1035
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9262083517987155e-05,
+      "loss": 1.4011,
+      "step": 1036
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.922646470097085e-05,
+      "loss": 1.3111,
+      "step": 1037
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.919083704995315e-05,
+      "loss": 1.359,
+      "step": 1038
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.915520063940168e-05,
+      "loss": 1.4065,
+      "step": 1039
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.911955554380234e-05,
+      "loss": 1.3751,
+      "step": 1040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9083901837659207e-05,
+      "loss": 1.4127,
+      "step": 1041
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9048239595494348e-05,
+      "loss": 1.4339,
+      "step": 1042
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9012568891847675e-05,
+      "loss": 1.323,
+      "step": 1043
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.897688980127678e-05,
+      "loss": 1.2053,
+      "step": 1044
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8941202398356792e-05,
+      "loss": 1.3854,
+      "step": 1045
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8905506757680207e-05,
+      "loss": 1.322,
+      "step": 1046
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.886980295385674e-05,
+      "loss": 1.4148,
+      "step": 1047
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.883409106151318e-05,
+      "loss": 1.3247,
+      "step": 1048
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8798371155293214e-05,
+      "loss": 1.4272,
+      "step": 1049
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8762643309857272e-05,
+      "loss": 1.3256,
+      "step": 1050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8726907599882384e-05,
+      "loss": 1.3384,
+      "step": 1051
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8691164100062034e-05,
+      "loss": 1.3241,
+      "step": 1052
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8655412885105953e-05,
+      "loss": 1.3916,
+      "step": 1053
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.861965402974004e-05,
+      "loss": 1.4277,
+      "step": 1054
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8583887608706113e-05,
+      "loss": 1.3428,
+      "step": 1055
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8548113696761856e-05,
+      "loss": 1.3777,
+      "step": 1056
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8512332368680583e-05,
+      "loss": 1.2774,
+      "step": 1057
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8476543699251107e-05,
+      "loss": 1.4081,
+      "step": 1058
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8440747763277596e-05,
+      "loss": 1.413,
+      "step": 1059
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8404944635579395e-05,
+      "loss": 1.3292,
+      "step": 1060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8369134390990902e-05,
+      "loss": 1.4091,
+      "step": 1061
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8333317104361352e-05,
+      "loss": 1.3887,
+      "step": 1062
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8297492850554747e-05,
+      "loss": 1.3161,
+      "step": 1063
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8261661704449615e-05,
+      "loss": 1.422,
+      "step": 1064
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.82258237409389e-05,
+      "loss": 1.4032,
+      "step": 1065
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8189979034929804e-05,
+      "loss": 1.2859,
+      "step": 1066
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8154127661343622e-05,
+      "loss": 1.4188,
+      "step": 1067
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8118269695115555e-05,
+      "loss": 1.3059,
+      "step": 1068
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.808240521119464e-05,
+      "loss": 1.2396,
+      "step": 1069
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8046534284543486e-05,
+      "loss": 1.4072,
+      "step": 1070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8010656990138183e-05,
+      "loss": 1.2896,
+      "step": 1071
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.797477340296815e-05,
+      "loss": 1.4018,
+      "step": 1072
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7938883598035935e-05,
+      "loss": 1.3127,
+      "step": 1073
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7902987650357092e-05,
+      "loss": 1.2875,
+      "step": 1074
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7867085634960016e-05,
+      "loss": 1.4377,
+      "step": 1075
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7831177626885773e-05,
+      "loss": 1.3539,
+      "step": 1076
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7795263701187972e-05,
+      "loss": 1.367,
+      "step": 1077
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7759343932932586e-05,
+      "loss": 1.2916,
+      "step": 1078
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.772341839719778e-05,
+      "loss": 1.336,
+      "step": 1079
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.768748716907381e-05,
+      "loss": 1.4192,
+      "step": 1080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7651550323662796e-05,
+      "loss": 1.3697,
+      "step": 1081
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.761560793607862e-05,
+      "loss": 1.4532,
+      "step": 1082
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7579660081446734e-05,
+      "loss": 1.3176,
+      "step": 1083
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7543706834904044e-05,
+      "loss": 1.3827,
+      "step": 1084
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.750774827159867e-05,
+      "loss": 1.327,
+      "step": 1085
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.7471784466689926e-05,
+      "loss": 1.3939,
+      "step": 1086
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7435815495348e-05,
+      "loss": 1.3796,
+      "step": 1087
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.739984143275393e-05,
+      "loss": 1.3144,
+      "step": 1088
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7363862354099383e-05,
+      "loss": 1.3237,
+      "step": 1089
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.732787833458651e-05,
+      "loss": 1.3724,
+      "step": 1090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7291889449427786e-05,
+      "loss": 1.3385,
+      "step": 1091
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7255895773845845e-05,
+      "loss": 1.4407,
+      "step": 1092
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7219897383073373e-05,
+      "loss": 1.4266,
+      "step": 1093
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.718389435235286e-05,
+      "loss": 1.2877,
+      "step": 1094
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7147886756936543e-05,
+      "loss": 1.4044,
+      "step": 1095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.711187467208614e-05,
+      "loss": 1.3622,
+      "step": 1096
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7075858173072834e-05,
+      "loss": 1.3283,
+      "step": 1097
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7039837335176954e-05,
+      "loss": 1.4122,
+      "step": 1098
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7003812233687947e-05,
+      "loss": 1.3597,
+      "step": 1099
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.696778294390416e-05,
+      "loss": 1.3662,
+      "step": 1100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.693174954113269e-05,
+      "loss": 1.2736,
+      "step": 1101
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6895712100689226e-05,
+      "loss": 1.3594,
+      "step": 1102
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6859670697897914e-05,
+      "loss": 1.4173,
+      "step": 1103
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6823625408091167e-05,
+      "loss": 1.4225,
+      "step": 1104
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.678757630660953e-05,
+      "loss": 1.3829,
+      "step": 1105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6751523468801504e-05,
+      "loss": 1.3767,
+      "step": 1106
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.671546697002341e-05,
+      "loss": 1.3871,
+      "step": 1107
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6679406885639225e-05,
+      "loss": 1.3191,
+      "step": 1108
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.66433432910204e-05,
+      "loss": 1.2095,
+      "step": 1109
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6607276261545756e-05,
+      "loss": 1.433,
+      "step": 1110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6571205872601258e-05,
+      "loss": 1.3569,
+      "step": 1111
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.653513219957991e-05,
+      "loss": 1.4373,
+      "step": 1112
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6499055317881587e-05,
+      "loss": 1.3963,
+      "step": 1113
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6462975302912862e-05,
+      "loss": 1.343,
+      "step": 1114
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6426892230086865e-05,
+      "loss": 1.354,
+      "step": 1115
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6390806174823085e-05,
+      "loss": 1.2979,
+      "step": 1116
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.63547172125473e-05,
+      "loss": 1.4333,
+      "step": 1117
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.631862541869131e-05,
+      "loss": 1.4255,
+      "step": 1118
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6282530868692874e-05,
+      "loss": 1.2693,
+      "step": 1119
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.624643363799549e-05,
+      "loss": 1.4078,
+      "step": 1120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6210333802048263e-05,
+      "loss": 1.2894,
+      "step": 1121
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6174231436305752e-05,
+      "loss": 1.3784,
+      "step": 1122
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6138126616227786e-05,
+      "loss": 1.3753,
+      "step": 1123
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6102019417279344e-05,
+      "loss": 1.3752,
+      "step": 1124
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6065909914930358e-05,
+      "loss": 1.3576,
+      "step": 1125
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.6029798184655602e-05,
+      "loss": 1.3548,
+      "step": 1126
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5993684301934467e-05,
+      "loss": 1.3816,
+      "step": 1127
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.595756834225089e-05,
+      "loss": 1.3238,
+      "step": 1128
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.59214503810931e-05,
+      "loss": 1.4034,
+      "step": 1129
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5885330493953548e-05,
+      "loss": 1.3816,
+      "step": 1130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.584920875632869e-05,
+      "loss": 1.3233,
+      "step": 1131
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5813085243718864e-05,
+      "loss": 1.3831,
+      "step": 1132
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5776960031628093e-05,
+      "loss": 1.3987,
+      "step": 1133
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5740833195563996e-05,
+      "loss": 1.2786,
+      "step": 1134
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.570470481103754e-05,
+      "loss": 1.3908,
+      "step": 1135
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5668574953562952e-05,
+      "loss": 1.3483,
+      "step": 1136
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.563244369865754e-05,
+      "loss": 1.4082,
+      "step": 1137
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5596311121841516e-05,
+      "loss": 1.2528,
+      "step": 1138
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5560177298637878e-05,
+      "loss": 1.3461,
+      "step": 1139
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.55240423045722e-05,
+      "loss": 1.4026,
+      "step": 1140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5487906215172545e-05,
+      "loss": 1.3262,
+      "step": 1141
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.545176910596922e-05,
+      "loss": 1.4057,
+      "step": 1142
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5415631052494693e-05,
+      "loss": 1.4369,
+      "step": 1143
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5379492130283396e-05,
+      "loss": 1.3217,
+      "step": 1144
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5343352414871578e-05,
+      "loss": 1.3553,
+      "step": 1145
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5307211981797147e-05,
+      "loss": 1.4282,
+      "step": 1146
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5271070906599503e-05,
+      "loss": 1.3617,
+      "step": 1147
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5234929264819397e-05,
+      "loss": 1.3557,
+      "step": 1148
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.519878713199877e-05,
+      "loss": 1.4082,
+      "step": 1149
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5162644583680574e-05,
+      "loss": 1.2683,
+      "step": 1150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5126501695408632e-05,
+      "loss": 1.3638,
+      "step": 1151
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.5090358542727506e-05,
+      "loss": 1.4131,
+      "step": 1152
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.505421520118225e-05,
+      "loss": 1.4144,
+      "step": 1153
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.501807174631839e-05,
+      "loss": 1.3538,
+      "step": 1154
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4981928253681616e-05,
+      "loss": 1.396,
+      "step": 1155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4945784798817752e-05,
+      "loss": 1.3196,
+      "step": 1156
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4909641457272507e-05,
+      "loss": 1.1821,
+      "step": 1157
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4873498304591363e-05,
+      "loss": 1.3992,
+      "step": 1158
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.483735541631943e-05,
+      "loss": 1.3062,
+      "step": 1159
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.480121286800124e-05,
+      "loss": 1.3968,
+      "step": 1160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4765070735180606e-05,
+      "loss": 1.3799,
+      "step": 1161
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4728929093400506e-05,
+      "loss": 1.4042,
+      "step": 1162
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4692788018202866e-05,
+      "loss": 1.3845,
+      "step": 1163
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.465664758512843e-05,
+      "loss": 1.3407,
+      "step": 1164
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4620507869716606e-05,
+      "loss": 1.3535,
+      "step": 1165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4584368947505306e-05,
+      "loss": 1.3714,
+      "step": 1166
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4548230894030786e-05,
+      "loss": 1.3726,
+      "step": 1167
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.451209378482746e-05,
+      "loss": 1.3774,
+      "step": 1168
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4475957695427802e-05,
+      "loss": 1.3929,
+      "step": 1169
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4439822701362135e-05,
+      "loss": 1.3422,
+      "step": 1170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.440368887815849e-05,
+      "loss": 1.3507,
+      "step": 1171
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4367556301342464e-05,
+      "loss": 1.3709,
+      "step": 1172
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4331425046437054e-05,
+      "loss": 1.3807,
+      "step": 1173
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4295295188962463e-05,
+      "loss": 1.3312,
+      "step": 1174
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4259166804436006e-05,
+      "loss": 1.3865,
+      "step": 1175
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.422303996837191e-05,
+      "loss": 1.2939,
+      "step": 1176
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4186914756281145e-05,
+      "loss": 1.424,
+      "step": 1177
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4150791243671322e-05,
+      "loss": 1.4015,
+      "step": 1178
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.411466950604645e-05,
+      "loss": 1.2859,
+      "step": 1179
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4078549618906904e-05,
+      "loss": 1.3922,
+      "step": 1180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4042431657749117e-05,
+      "loss": 1.1814,
+      "step": 1181
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4006315698065536e-05,
+      "loss": 1.3456,
+      "step": 1182
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3970201815344407e-05,
+      "loss": 1.3974,
+      "step": 1183
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3934090085069644e-05,
+      "loss": 1.3254,
+      "step": 1184
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.389798058272067e-05,
+      "loss": 1.4154,
+      "step": 1185
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.386187338377222e-05,
+      "loss": 1.3154,
+      "step": 1186
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3825768563694254e-05,
+      "loss": 1.3672,
+      "step": 1187
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.378966619795174e-05,
+      "loss": 1.3961,
+      "step": 1188
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.375356636200452e-05,
+      "loss": 1.3918,
+      "step": 1189
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3717469131307132e-05,
+      "loss": 1.383,
+      "step": 1190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3681374581308702e-05,
+      "loss": 1.3538,
+      "step": 1191
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3645282787452703e-05,
+      "loss": 1.2766,
+      "step": 1192
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3609193825176917e-05,
+      "loss": 1.3027,
+      "step": 1193
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3573107769913144e-05,
+      "loss": 1.4077,
+      "step": 1194
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3537024697087144e-05,
+      "loss": 1.4151,
+      "step": 1195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3500944682118415e-05,
+      "loss": 1.3047,
+      "step": 1196
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3464867800420093e-05,
+      "loss": 1.435,
+      "step": 1197
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3428794127398758e-05,
+      "loss": 1.3262,
+      "step": 1198
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.339272373845425e-05,
+      "loss": 1.334,
+      "step": 1199
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3356656708979603e-05,
+      "loss": 1.3859,
+      "step": 1200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.332059311436078e-05,
+      "loss": 1.2627,
+      "step": 1201
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3284533029976595e-05,
+      "loss": 1.432,
+      "step": 1202
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3248476531198505e-05,
+      "loss": 1.3113,
+      "step": 1203
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3212423693390482e-05,
+      "loss": 1.3495,
+      "step": 1204
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.317637459190884e-05,
+      "loss": 1.2281,
+      "step": 1205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.314032930210209e-05,
+      "loss": 1.3643,
+      "step": 1206
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3104287899310773e-05,
+      "loss": 1.3057,
+      "step": 1207
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.306825045886732e-05,
+      "loss": 1.3807,
+      "step": 1208
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3032217056095844e-05,
+      "loss": 1.349,
+      "step": 1209
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.299618776631206e-05,
+      "loss": 1.3286,
+      "step": 1210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2960162664823055e-05,
+      "loss": 1.4117,
+      "step": 1211
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.292414182692718e-05,
+      "loss": 1.2967,
+      "step": 1212
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2888125327913857e-05,
+      "loss": 1.3994,
+      "step": 1213
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2852113243063466e-05,
+      "loss": 1.3314,
+      "step": 1214
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.281610564764714e-05,
+      "loss": 1.3285,
+      "step": 1215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2780102616926633e-05,
+      "loss": 1.3458,
+      "step": 1216
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2744104226154157e-05,
+      "loss": 1.4015,
+      "step": 1217
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2708110550572223e-05,
+      "loss": 1.3001,
+      "step": 1218
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2672121665413495e-05,
+      "loss": 1.3001,
+      "step": 1219
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2636137645900613e-05,
+      "loss": 1.4037,
+      "step": 1220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2600158567246075e-05,
+      "loss": 1.2976,
+      "step": 1221
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2564184504652004e-05,
+      "loss": 1.2908,
+      "step": 1222
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2528215533310084e-05,
+      "loss": 1.3443,
+      "step": 1223
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2492251728401332e-05,
+      "loss": 1.3666,
+      "step": 1224
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.245629316509597e-05,
+      "loss": 1.4203,
+      "step": 1225
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.242033991855326e-05,
+      "loss": 1.3778,
+      "step": 1226
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2384392063921386e-05,
+      "loss": 1.3394,
+      "step": 1227
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2348449676337206e-05,
+      "loss": 1.3599,
+      "step": 1228
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2312512830926195e-05,
+      "loss": 1.403,
+      "step": 1229
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2276581602802226e-05,
+      "loss": 1.2402,
+      "step": 1230
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2240656067067427e-05,
+      "loss": 1.3854,
+      "step": 1231
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.220473629881203e-05,
+      "loss": 1.3029,
+      "step": 1232
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2168822373114226e-05,
+      "loss": 1.3865,
+      "step": 1233
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2132914365039993e-05,
+      "loss": 1.2099,
+      "step": 1234
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.209701234964291e-05,
+      "loss": 1.3301,
+      "step": 1235
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.206111640196407e-05,
+      "loss": 1.3683,
+      "step": 1236
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2025226597031856e-05,
+      "loss": 1.4423,
+      "step": 1237
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.198934300986182e-05,
+      "loss": 1.3212,
+      "step": 1238
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1953465715456526e-05,
+      "loss": 1.3733,
+      "step": 1239
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1917594788805363e-05,
+      "loss": 1.2897,
+      "step": 1240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.188173030488444e-05,
+      "loss": 1.3067,
+      "step": 1241
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1845872338656383e-05,
+      "loss": 1.2695,
+      "step": 1242
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1810020965070198e-05,
+      "loss": 1.3448,
+      "step": 1243
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1774176259061103e-05,
+      "loss": 1.3392,
+      "step": 1244
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1738338295550397e-05,
+      "loss": 1.3668,
+      "step": 1245
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1702507149445262e-05,
+      "loss": 1.4056,
+      "step": 1246
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.166668289563865e-05,
+      "loss": 1.33,
+      "step": 1247
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1630865609009104e-05,
+      "loss": 1.3281,
+      "step": 1248
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1595055364420607e-05,
+      "loss": 1.3862,
+      "step": 1249
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.155925223672241e-05,
+      "loss": 1.3388,
+      "step": 1250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.15234563007489e-05,
+      "loss": 1.39,
+      "step": 1251
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1487667631319426e-05,
+      "loss": 1.2717,
+      "step": 1252
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1451886303238146e-05,
+      "loss": 1.262,
+      "step": 1253
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.141611239129389e-05,
+      "loss": 1.3375,
+      "step": 1254
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1380345970259967e-05,
+      "loss": 1.3384,
+      "step": 1255
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1344587114894053e-05,
+      "loss": 1.398,
+      "step": 1256
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1308835899937972e-05,
+      "loss": 1.3315,
+      "step": 1257
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1273092400117618e-05,
+      "loss": 1.2832,
+      "step": 1258
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1237356690142734e-05,
+      "loss": 1.4077,
+      "step": 1259
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1201628844706785e-05,
+      "loss": 1.2908,
+      "step": 1260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1165908938486815e-05,
+      "loss": 1.3575,
+      "step": 1261
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1130197046143263e-05,
+      "loss": 1.3827,
+      "step": 1262
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1094493242319802e-05,
+      "loss": 1.3982,
+      "step": 1263
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1058797601643214e-05,
+      "loss": 1.282,
+      "step": 1264
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1023110198723233e-05,
+      "loss": 1.4156,
+      "step": 1265
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0987431108152334e-05,
+      "loss": 1.3475,
+      "step": 1266
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0951760404505655e-05,
+      "loss": 1.3961,
+      "step": 1267
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0916098162340796e-05,
+      "loss": 1.3719,
+      "step": 1268
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.088044445619767e-05,
+      "loss": 1.3919,
+      "step": 1269
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0844799360598326e-05,
+      "loss": 1.308,
+      "step": 1270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0809162950046855e-05,
+      "loss": 1.334,
+      "step": 1271
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.077353529902916e-05,
+      "loss": 1.2895,
+      "step": 1272
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0737916482012854e-05,
+      "loss": 1.3973,
+      "step": 1273
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0702306573447075e-05,
+      "loss": 1.2694,
+      "step": 1274
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0666705647762365e-05,
+      "loss": 1.3066,
+      "step": 1275
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0631113779370456e-05,
+      "loss": 1.4231,
+      "step": 1276
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.059553104266417e-05,
+      "loss": 1.3803,
+      "step": 1277
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0559957512017243e-05,
+      "loss": 1.3102,
+      "step": 1278
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.052439326178415e-05,
+      "loss": 1.4231,
+      "step": 1279
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0488838366299994e-05,
+      "loss": 1.3747,
+      "step": 1280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0453292899880293e-05,
+      "loss": 1.2515,
+      "step": 1281
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0417756936820902e-05,
+      "loss": 1.3222,
+      "step": 1282
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0382230551397772e-05,
+      "loss": 1.3198,
+      "step": 1283
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0346713817866857e-05,
+      "loss": 1.3658,
+      "step": 1284
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0311206810463922e-05,
+      "loss": 1.3807,
+      "step": 1285
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0275709603404416e-05,
+      "loss": 1.3918,
+      "step": 1286
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0240222270883288e-05,
+      "loss": 1.3668,
+      "step": 1287
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.0204744887074878e-05,
+      "loss": 1.39,
+      "step": 1288
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0169277526132696e-05,
+      "loss": 1.2752,
+      "step": 1289
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0133820262189316e-05,
+      "loss": 1.4209,
+      "step": 1290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0098373169356228e-05,
+      "loss": 1.3859,
+      "step": 1291
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0062936321723624e-05,
+      "loss": 1.2626,
+      "step": 1292
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0027509793360317e-05,
+      "loss": 1.4064,
+      "step": 1293
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.999209365831351e-05,
+      "loss": 1.4056,
+      "step": 1294
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9956687990608748e-05,
+      "loss": 1.3228,
+      "step": 1295
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.992129286424963e-05,
+      "loss": 1.3739,
+      "step": 1296
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9885908353217764e-05,
+      "loss": 1.3886,
+      "step": 1297
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9850534531472546e-05,
+      "loss": 1.312,
+      "step": 1298
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.981517147295104e-05,
+      "loss": 1.1809,
+      "step": 1299
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.977981925156783e-05,
+      "loss": 1.3091,
+      "step": 1300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.97444779412148e-05,
+      "loss": 1.3623,
+      "step": 1301
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.970914761576109e-05,
+      "loss": 1.3906,
+      "step": 1302
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9673828349052836e-05,
+      "loss": 1.2808,
+      "step": 1303
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9638520214913082e-05,
+      "loss": 1.3765,
+      "step": 1304
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9603223287141576e-05,
+      "loss": 1.2705,
+      "step": 1305
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9567937639514684e-05,
+      "loss": 1.2694,
+      "step": 1306
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.953266334578515e-05,
+      "loss": 1.364,
+      "step": 1307
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.949740047968201e-05,
+      "loss": 1.4058,
+      "step": 1308
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.946214911491043e-05,
+      "loss": 1.3357,
+      "step": 1309
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.942690932515151e-05,
+      "loss": 1.3195,
+      "step": 1310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.939168118406216e-05,
+      "loss": 1.344,
+      "step": 1311
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9356464765274954e-05,
+      "loss": 1.3178,
+      "step": 1312
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9321260142397956e-05,
+      "loss": 1.3945,
+      "step": 1313
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9286067389014573e-05,
+      "loss": 1.3325,
+      "step": 1314
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9250886578683403e-05,
+      "loss": 1.3494,
+      "step": 1315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9215717784938096e-05,
+      "loss": 1.3732,
+      "step": 1316
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9180561081287174e-05,
+      "loss": 1.3084,
+      "step": 1317
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.914541654121388e-05,
+      "loss": 1.2786,
+      "step": 1318
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9110284238176048e-05,
+      "loss": 1.4191,
+      "step": 1319
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9075164245605915e-05,
+      "loss": 1.3176,
+      "step": 1320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9040056636910013e-05,
+      "loss": 1.3034,
+      "step": 1321
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9004961485468973e-05,
+      "loss": 1.4154,
+      "step": 1322
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8969878864637405e-05,
+      "loss": 1.3007,
+      "step": 1323
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8934808847743695e-05,
+      "loss": 1.3569,
+      "step": 1324
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.889975150808992e-05,
+      "loss": 1.212,
+      "step": 1325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.886470691895164e-05,
+      "loss": 1.3188,
+      "step": 1326
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8829675153577765e-05,
+      "loss": 1.3348,
+      "step": 1327
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.87946562851904e-05,
+      "loss": 1.3189,
+      "step": 1328
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8759650386984714e-05,
+      "loss": 1.3999,
+      "step": 1329
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8724657532128747e-05,
+      "loss": 1.2851,
+      "step": 1330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.868967779376327e-05,
+      "loss": 1.2984,
+      "step": 1331
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8654711245001667e-05,
+      "loss": 1.3786,
+      "step": 1332
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8619757958929706e-05,
+      "loss": 1.3462,
+      "step": 1333
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8584818008605487e-05,
+      "loss": 1.3941,
+      "step": 1334
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8549891467059194e-05,
+      "loss": 1.3458,
+      "step": 1335
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8514978407293022e-05,
+      "loss": 1.391,
+      "step": 1336
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8480078902280952e-05,
+      "loss": 1.3266,
+      "step": 1337
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8445193024968656e-05,
+      "loss": 1.3436,
+      "step": 1338
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8410320848273315e-05,
+      "loss": 1.3332,
+      "step": 1339
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8375462445083464e-05,
+      "loss": 1.2883,
+      "step": 1340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8340617888258866e-05,
+      "loss": 1.3749,
+      "step": 1341
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8305787250630318e-05,
+      "loss": 1.3702,
+      "step": 1342
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8270970604999567e-05,
+      "loss": 1.2932,
+      "step": 1343
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8236168024139073e-05,
+      "loss": 1.4109,
+      "step": 1344
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8201379580791922e-05,
+      "loss": 1.3905,
+      "step": 1345
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8166605347671626e-05,
+      "loss": 1.1925,
+      "step": 1346
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.813184539746202e-05,
+      "loss": 1.3909,
+      "step": 1347
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.809709980281708e-05,
+      "loss": 1.2844,
+      "step": 1348
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8062368636360757e-05,
+      "loss": 1.2215,
+      "step": 1349
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.802765197068688e-05,
+      "loss": 1.4082,
+      "step": 1350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7992949878358935e-05,
+      "loss": 1.3297,
+      "step": 1351
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7958262431909977e-05,
+      "loss": 1.329,
+      "step": 1352
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.792358970384242e-05,
+      "loss": 1.3024,
+      "step": 1353
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.788893176662793e-05,
+      "loss": 1.32,
+      "step": 1354
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7854288692707247e-05,
+      "loss": 1.3563,
+      "step": 1355
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7819660554490077e-05,
+      "loss": 1.3818,
+      "step": 1356
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7785047424354862e-05,
+      "loss": 1.2941,
+      "step": 1357
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7750449374648706e-05,
+      "loss": 1.3519,
+      "step": 1358
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7715866477687176e-05,
+      "loss": 1.3992,
+      "step": 1359
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7681298805754176e-05,
+      "loss": 1.2821,
+      "step": 1360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7646746431101797e-05,
+      "loss": 1.356,
+      "step": 1361
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7612209425950116e-05,
+      "loss": 1.372,
+      "step": 1362
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7577687862487142e-05,
+      "loss": 1.3283,
+      "step": 1363
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7543181812868564e-05,
+      "loss": 1.376,
+      "step": 1364
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7508691349217677e-05,
+      "loss": 1.2977,
+      "step": 1365
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.747421654362516e-05,
+      "loss": 1.3208,
+      "step": 1366
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7439757468149006e-05,
+      "loss": 1.4122,
+      "step": 1367
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7405314194814294e-05,
+      "loss": 1.3289,
+      "step": 1368
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.737088679561309e-05,
+      "loss": 1.381,
+      "step": 1369
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7336475342504292e-05,
+      "loss": 1.2189,
+      "step": 1370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7302079907413455e-05,
+      "loss": 1.2732,
+      "step": 1371
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7267700562232648e-05,
+      "loss": 1.4345,
+      "step": 1372
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7233337378820318e-05,
+      "loss": 1.4056,
+      "step": 1373
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.719899042900114e-05,
+      "loss": 1.3056,
+      "step": 1374
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.716465978456583e-05,
+      "loss": 1.385,
+      "step": 1375
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7130345517271045e-05,
+      "loss": 1.382,
+      "step": 1376
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.709604769883922e-05,
+      "loss": 1.2411,
+      "step": 1377
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.706176640095839e-05,
+      "loss": 1.3934,
+      "step": 1378
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7027501695282057e-05,
+      "loss": 1.4144,
+      "step": 1379
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6993253653429063e-05,
+      "loss": 1.2853,
+      "step": 1380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.69590223469834e-05,
+      "loss": 1.2687,
+      "step": 1381
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6924807847494086e-05,
+      "loss": 1.4178,
+      "step": 1382
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6890610226475022e-05,
+      "loss": 1.3054,
+      "step": 1383
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6856429555404822e-05,
+      "loss": 1.3706,
+      "step": 1384
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6822265905726656e-05,
+      "loss": 1.2983,
+      "step": 1385
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.678811934884814e-05,
+      "loss": 1.4074,
+      "step": 1386
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6753989956141154e-05,
+      "loss": 1.3402,
+      "step": 1387
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6719877798941696e-05,
+      "loss": 1.3187,
+      "step": 1388
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6685782948549744e-05,
+      "loss": 1.2494,
+      "step": 1389
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6651705476229118e-05,
+      "loss": 1.3614,
+      "step": 1390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6617645453207287e-05,
+      "loss": 1.293,
+      "step": 1391
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.658360295067526e-05,
+      "loss": 1.3761,
+      "step": 1392
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6549578039787436e-05,
+      "loss": 1.3701,
+      "step": 1393
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6515570791661424e-05,
+      "loss": 1.3163,
+      "step": 1394
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.648158127737793e-05,
+      "loss": 1.3446,
+      "step": 1395
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6447609567980585e-05,
+      "loss": 1.2063,
+      "step": 1396
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6413655734475823e-05,
+      "loss": 1.3369,
+      "step": 1397
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.637971984783269e-05,
+      "loss": 1.3778,
+      "step": 1398
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6345801978982735e-05,
+      "loss": 1.2789,
+      "step": 1399
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6311902198819833e-05,
+      "loss": 1.3627,
+      "step": 1400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.627802057820007e-05,
+      "loss": 1.2851,
+      "step": 1401
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6244157187941563e-05,
+      "loss": 1.3851,
+      "step": 1402
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.621031209882432e-05,
+      "loss": 1.2293,
+      "step": 1403
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6176485381590117e-05,
+      "loss": 1.3569,
+      "step": 1404
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6142677106942308e-05,
+      "loss": 1.3275,
+      "step": 1405
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6108887345545717e-05,
+      "loss": 1.3467,
+      "step": 1406
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6075116168026456e-05,
+      "loss": 1.324,
+      "step": 1407
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6041363644971802e-05,
+      "loss": 1.2914,
+      "step": 1408
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6007629846930034e-05,
+      "loss": 1.3477,
+      "step": 1409
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5973914844410297e-05,
+      "loss": 1.3487,
+      "step": 1410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.594021870788247e-05,
+      "loss": 1.3322,
+      "step": 1411
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5906541507776966e-05,
+      "loss": 1.3454,
+      "step": 1412
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5872883314484636e-05,
+      "loss": 1.3566,
+      "step": 1413
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.58392441983566e-05,
+      "loss": 1.2743,
+      "step": 1414
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.580562422970411e-05,
+      "loss": 1.3856,
+      "step": 1415
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5772023478798374e-05,
+      "loss": 1.2632,
+      "step": 1416
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.573844201587046e-05,
+      "loss": 1.3842,
+      "step": 1417
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.570487991111111e-05,
+      "loss": 1.2745,
+      "step": 1418
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5671337234670608e-05,
+      "loss": 1.2471,
+      "step": 1419
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5637814056658616e-05,
+      "loss": 1.3895,
+      "step": 1420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.560431044714405e-05,
+      "loss": 1.3288,
+      "step": 1421
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.557082647615493e-05,
+      "loss": 1.3155,
+      "step": 1422
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5537362213678214e-05,
+      "loss": 1.3589,
+      "step": 1423
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5503917729659682e-05,
+      "loss": 1.2916,
+      "step": 1424
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.547049309400378e-05,
+      "loss": 1.2926,
+      "step": 1425
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5437088376573432e-05,
+      "loss": 1.3665,
+      "step": 1426
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5403703647189965e-05,
+      "loss": 1.3701,
+      "step": 1427
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5370338975632913e-05,
+      "loss": 1.2709,
+      "step": 1428
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5336994431639885e-05,
+      "loss": 1.3962,
+      "step": 1429
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.530367008490642e-05,
+      "loss": 1.3405,
+      "step": 1430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5270366005085852e-05,
+      "loss": 1.3386,
+      "step": 1431
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5237082261789149e-05,
+      "loss": 1.346,
+      "step": 1432
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5203818924584759e-05,
+      "loss": 1.3829,
+      "step": 1433
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5170576062998498e-05,
+      "loss": 1.3003,
+      "step": 1434
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5137353746513367e-05,
+      "loss": 1.3868,
+      "step": 1435
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5104152044569436e-05,
+      "loss": 1.3129,
+      "step": 1436
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.507097102656368e-05,
+      "loss": 1.1924,
+      "step": 1437
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5037810761849857e-05,
+      "loss": 1.3211,
+      "step": 1438
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5004671319738323e-05,
+      "loss": 1.3125,
+      "step": 1439
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4971552769495929e-05,
+      "loss": 1.356,
+      "step": 1440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4938455180345854e-05,
+      "loss": 1.3741,
+      "step": 1441
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4905378621467458e-05,
+      "loss": 1.3096,
+      "step": 1442
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.487232316199616e-05,
+      "loss": 1.3591,
+      "step": 1443
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4839288871023244e-05,
+      "loss": 1.2916,
+      "step": 1444
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.480627581759581e-05,
+      "loss": 1.3066,
+      "step": 1445
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4773284070716503e-05,
+      "loss": 1.3598,
+      "step": 1446
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4740313699343477e-05,
+      "loss": 1.2996,
+      "step": 1447
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.470736477239018e-05,
+      "loss": 1.364,
+      "step": 1448
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4674437358725252e-05,
+      "loss": 1.3411,
+      "step": 1449
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4641531527172364e-05,
+      "loss": 1.3191,
+      "step": 1450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.460864734651009e-05,
+      "loss": 1.3058,
+      "step": 1451
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4575784885471722e-05,
+      "loss": 1.3532,
+      "step": 1452
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4542944212745186e-05,
+      "loss": 1.3394,
+      "step": 1453
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.451012539697283e-05,
+      "loss": 1.2813,
+      "step": 1454
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.447732850675135e-05,
+      "loss": 1.398,
+      "step": 1455
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.44445536106316e-05,
+      "loss": 1.2815,
+      "step": 1456
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4411800777118466e-05,
+      "loss": 1.4076,
+      "step": 1457
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.437907007467072e-05,
+      "loss": 1.3357,
+      "step": 1458
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4346361571700873e-05,
+      "loss": 1.3049,
+      "step": 1459
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4313675336575036e-05,
+      "loss": 1.3056,
+      "step": 1460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4281011437612778e-05,
+      "loss": 1.1398,
+      "step": 1461
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4248369943086998e-05,
+      "loss": 1.3875,
+      "step": 1462
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.421575092122372e-05,
+      "loss": 1.3256,
+      "step": 1463
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.418315444020203e-05,
+      "loss": 1.3469,
+      "step": 1464
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.415058056815393e-05,
+      "loss": 1.3521,
+      "step": 1465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4118029373164102e-05,
+      "loss": 1.2816,
+      "step": 1466
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.4085500923269862e-05,
+      "loss": 1.3008,
+      "step": 1467
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4052995286460988e-05,
+      "loss": 1.3621,
+      "step": 1468
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.4020512530679564e-05,
+      "loss": 1.294,
+      "step": 1469
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3988052723819861e-05,
+      "loss": 1.3334,
+      "step": 1470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3955615933728172e-05,
+      "loss": 1.3672,
+      "step": 1471
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3923202228202681e-05,
+      "loss": 1.2487,
+      "step": 1472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3890811674993331e-05,
+      "loss": 1.2584,
+      "step": 1473
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.385844434180168e-05,
+      "loss": 1.3678,
+      "step": 1474
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3826100296280708e-05,
+      "loss": 1.3493,
+      "step": 1475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3793779606034777e-05,
+      "loss": 1.286,
+      "step": 1476
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3761482338619397e-05,
+      "loss": 1.4023,
+      "step": 1477
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3729208561541135e-05,
+      "loss": 1.3678,
+      "step": 1478
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3696958342257455e-05,
+      "loss": 1.2396,
+      "step": 1479
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3664731748176577e-05,
+      "loss": 1.4138,
+      "step": 1480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.363252884665735e-05,
+      "loss": 1.2741,
+      "step": 1481
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3600349705009097e-05,
+      "loss": 1.3517,
+      "step": 1482
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.356819439049149e-05,
+      "loss": 1.3231,
+      "step": 1483
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3536062970314351e-05,
+      "loss": 1.2862,
+      "step": 1484
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3503955511637628e-05,
+      "loss": 1.1775,
+      "step": 1485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3471872081571147e-05,
+      "loss": 1.399,
+      "step": 1486
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3439812747174521e-05,
+      "loss": 1.2998,
+      "step": 1487
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3407777575456975e-05,
+      "loss": 1.3029,
+      "step": 1488
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3375766633377262e-05,
+      "loss": 1.2916,
+      "step": 1489
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3343779987843474e-05,
+      "loss": 1.3157,
+      "step": 1490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3311817705712925e-05,
+      "loss": 1.3586,
+      "step": 1491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3279879853792005e-05,
+      "loss": 1.3966,
+      "step": 1492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3247966498836037e-05,
+      "loss": 1.2999,
+      "step": 1493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3216077707549146e-05,
+      "loss": 1.2931,
+      "step": 1494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3184213546584117e-05,
+      "loss": 1.3398,
+      "step": 1495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3152374082542251e-05,
+      "loss": 1.2885,
+      "step": 1496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3120559381973216e-05,
+      "loss": 1.4222,
+      "step": 1497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3088769511374926e-05,
+      "loss": 1.2686,
+      "step": 1498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3057004537193423e-05,
+      "loss": 1.3371,
+      "step": 1499
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.3025264525822692e-05,
+      "loss": 1.4064,
+      "step": 1500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2993549543604516e-05,
+      "loss": 1.2454,
+      "step": 1501
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.29618596568284e-05,
+      "loss": 1.2946,
+      "step": 1502
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2930194931731382e-05,
+      "loss": 1.3819,
+      "step": 1503
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2898555434497906e-05,
+      "loss": 1.2821,
+      "step": 1504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2866941231259689e-05,
+      "loss": 1.3657,
+      "step": 1505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2835352388095578e-05,
+      "loss": 1.391,
+      "step": 1506
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2803788971031416e-05,
+      "loss": 1.2948,
+      "step": 1507
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2772251046039895e-05,
+      "loss": 1.3204,
+      "step": 1508
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2740738679040439e-05,
+      "loss": 1.406,
+      "step": 1509
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2709251935899024e-05,
+      "loss": 1.1861,
+      "step": 1510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2677790882428093e-05,
+      "loss": 1.3485,
+      "step": 1511
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2646355584386372e-05,
+      "loss": 1.2726,
+      "step": 1512
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2614946107478804e-05,
+      "loss": 1.3955,
+      "step": 1513
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2583562517356288e-05,
+      "loss": 1.2084,
+      "step": 1514
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2552204879615665e-05,
+      "loss": 1.2868,
+      "step": 1515
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2520873259799518e-05,
+      "loss": 1.3357,
+      "step": 1516
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.248956772339605e-05,
+      "loss": 1.4066,
+      "step": 1517
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2458288335838944e-05,
+      "loss": 1.3147,
+      "step": 1518
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2427035162507225e-05,
+      "loss": 1.4025,
+      "step": 1519
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2395808268725134e-05,
+      "loss": 1.2654,
+      "step": 1520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2364607719761972e-05,
+      "loss": 1.2934,
+      "step": 1521
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2333433580831996e-05,
+      "loss": 1.3221,
+      "step": 1522
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2302285917094225e-05,
+      "loss": 1.3774,
+      "step": 1523
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2271164793652374e-05,
+      "loss": 1.3017,
+      "step": 1524
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2240070275554662e-05,
+      "loss": 1.3455,
+      "step": 1525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.220900242779374e-05,
+      "loss": 1.3269,
+      "step": 1526
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2177961315306449e-05,
+      "loss": 1.333,
+      "step": 1527
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2146947002973796e-05,
+      "loss": 1.3251,
+      "step": 1528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2115959555620762e-05,
+      "loss": 1.3792,
+      "step": 1529
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2084999038016168e-05,
+      "loss": 1.2813,
+      "step": 1530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.205406551487256e-05,
+      "loss": 1.3504,
+      "step": 1531
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.2023159050846027e-05,
+      "loss": 1.2517,
+      "step": 1532
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1992279710536152e-05,
+      "loss": 1.3814,
+      "step": 1533
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1961427558485794e-05,
+      "loss": 1.1932,
+      "step": 1534
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.193060265918099e-05,
+      "loss": 1.2594,
+      "step": 1535
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1899805077050802e-05,
+      "loss": 1.4133,
+      "step": 1536
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.186903487646721e-05,
+      "loss": 1.3014,
+      "step": 1537
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1838292121744962e-05,
+      "loss": 1.2769,
+      "step": 1538
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1807576877141433e-05,
+      "loss": 1.3179,
+      "step": 1539
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1776889206856498e-05,
+      "loss": 1.3007,
+      "step": 1540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1746229175032403e-05,
+      "loss": 1.2979,
+      "step": 1541
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.171559684575362e-05,
+      "loss": 1.3497,
+      "step": 1542
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1684992283046719e-05,
+      "loss": 1.3383,
+      "step": 1543
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1654415550880243e-05,
+      "loss": 1.273,
+      "step": 1544
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1623866713164524e-05,
+      "loss": 1.3581,
+      "step": 1545
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1593345833751656e-05,
+      "loss": 1.4088,
+      "step": 1546
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1562852976435249e-05,
+      "loss": 1.2891,
+      "step": 1547
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1532388204950362e-05,
+      "loss": 1.351,
+      "step": 1548
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1501951582973321e-05,
+      "loss": 1.4268,
+      "step": 1549
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1471543174121652e-05,
+      "loss": 1.2843,
+      "step": 1550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.144116304195389e-05,
+      "loss": 1.2504,
+      "step": 1551
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1410811249969475e-05,
+      "loss": 1.2681,
+      "step": 1552
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1380487861608605e-05,
+      "loss": 1.3839,
+      "step": 1553
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.135019294025212e-05,
+      "loss": 1.2264,
+      "step": 1554
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1319926549221343e-05,
+      "loss": 1.2849,
+      "step": 1555
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1289688751777982e-05,
+      "loss": 1.4183,
+      "step": 1556
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1259479611123977e-05,
+      "loss": 1.3037,
+      "step": 1557
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.122929919040135e-05,
+      "loss": 1.3254,
+      "step": 1558
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1199147552692103e-05,
+      "loss": 1.3174,
+      "step": 1559
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1169024761018105e-05,
+      "loss": 1.3631,
+      "step": 1560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1138930878340906e-05,
+      "loss": 1.1825,
+      "step": 1561
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1108865967561621e-05,
+      "loss": 1.3824,
+      "step": 1562
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1078830091520828e-05,
+      "loss": 1.2672,
+      "step": 1563
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1048823312998416e-05,
+      "loss": 1.3456,
+      "step": 1564
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.1018845694713454e-05,
+      "loss": 1.3858,
+      "step": 1565
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0988897299324053e-05,
+      "loss": 1.4148,
+      "step": 1566
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0958978189427253e-05,
+      "loss": 1.2401,
+      "step": 1567
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0929088427558878e-05,
+      "loss": 1.3425,
+      "step": 1568
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0899228076193416e-05,
+      "loss": 1.2837,
+      "step": 1569
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.086939719774388e-05,
+      "loss": 1.3585,
+      "step": 1570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0839595854561666e-05,
+      "loss": 1.3603,
+      "step": 1571
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0809824108936448e-05,
+      "loss": 1.2897,
+      "step": 1572
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0780082023096037e-05,
+      "loss": 1.3257,
+      "step": 1573
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0750369659206266e-05,
+      "loss": 1.341,
+      "step": 1574
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0720687079370809e-05,
+      "loss": 1.3002,
+      "step": 1575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.069103434563111e-05,
+      "loss": 1.348,
+      "step": 1576
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0661411519966224e-05,
+      "loss": 1.3699,
+      "step": 1577
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0631818664292697e-05,
+      "loss": 1.301,
+      "step": 1578
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0602255840464426e-05,
+      "loss": 1.1833,
+      "step": 1579
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0572723110272542e-05,
+      "loss": 1.3221,
+      "step": 1580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0543220535445275e-05,
+      "loss": 1.2233,
+      "step": 1581
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0513748177647822e-05,
+      "loss": 1.3836,
+      "step": 1582
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0484306098482236e-05,
+      "loss": 1.2884,
+      "step": 1583
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0454894359487246e-05,
+      "loss": 1.3846,
+      "step": 1584
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0425513022138203e-05,
+      "loss": 1.2835,
+      "step": 1585
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0396162147846892e-05,
+      "loss": 1.3145,
+      "step": 1586
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0366841797961455e-05,
+      "loss": 1.2544,
+      "step": 1587
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0337552033766187e-05,
+      "loss": 1.3352,
+      "step": 1588
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0308292916481488e-05,
+      "loss": 1.354,
+      "step": 1589
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0279064507263691e-05,
+      "loss": 1.2289,
+      "step": 1590
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.024986686720494e-05,
+      "loss": 1.3833,
+      "step": 1591
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0220700057333087e-05,
+      "loss": 1.2671,
+      "step": 1592
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0191564138611494e-05,
+      "loss": 1.3251,
+      "step": 1593
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.016245917193902e-05,
+      "loss": 1.3791,
+      "step": 1594
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0133385218149787e-05,
+      "loss": 1.2537,
+      "step": 1595
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0104342338013115e-05,
+      "loss": 1.3307,
+      "step": 1596
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0075330592233348e-05,
+      "loss": 1.3673,
+      "step": 1597
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0046350041449783e-05,
+      "loss": 1.2579,
+      "step": 1598
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.0017400746236503e-05,
+      "loss": 1.361,
+      "step": 1599
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.988482767102258e-06,
+      "loss": 1.3384,
+      "step": 1600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.95959616449035e-06,
+      "loss": 1.2603,
+      "step": 1601
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.930740998778499e-06,
+      "loss": 1.3579,
+      "step": 1602
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.90191733027871e-06,
+      "loss": 1.285,
+      "step": 1603
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.873125219237154e-06,
+      "loss": 1.36,
+      "step": 1604
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.844364725834057e-06,
+      "loss": 1.2254,
+      "step": 1605
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.81563591018353e-06,
+      "loss": 1.3089,
+      "step": 1606
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.786938832333487e-06,
+      "loss": 1.3571,
+      "step": 1607
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.758273552265523e-06,
+      "loss": 1.2678,
+      "step": 1608
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.729640129894758e-06,
+      "loss": 1.3299,
+      "step": 1609
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.7010386250697e-06,
+      "loss": 1.3086,
+      "step": 1610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.672469097572176e-06,
+      "loss": 1.3145,
+      "step": 1611
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.64393160711716e-06,
+      "loss": 1.2979,
+      "step": 1612
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.615426213352672e-06,
+      "loss": 1.351,
+      "step": 1613
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.586952975859633e-06,
+      "loss": 1.3565,
+      "step": 1614
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.558511954151766e-06,
+      "loss": 1.3278,
+      "step": 1615
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.530103207675447e-06,
+      "loss": 1.3611,
+      "step": 1616
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.501726795809599e-06,
+      "loss": 1.2681,
+      "step": 1617
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.473382777865559e-06,
+      "loss": 1.3238,
+      "step": 1618
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.445071213086936e-06,
+      "loss": 1.3787,
+      "step": 1619
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.416792160649532e-06,
+      "loss": 1.2834,
+      "step": 1620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.388545679661193e-06,
+      "loss": 1.3439,
+      "step": 1621
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.360331829161678e-06,
+      "loss": 1.358,
+      "step": 1622
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.332150668122528e-06,
+      "loss": 1.3058,
+      "step": 1623
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.304002255446978e-06,
+      "loss": 1.3198,
+      "step": 1624
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.27588664996981e-06,
+      "loss": 1.3621,
+      "step": 1625
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.247803910457226e-06,
+      "loss": 1.2675,
+      "step": 1626
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.219754095606738e-06,
+      "loss": 1.3243,
+      "step": 1627
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.191737264047043e-06,
+      "loss": 1.284,
+      "step": 1628
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.163753474337888e-06,
+      "loss": 1.2134,
+      "step": 1629
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.135802784969971e-06,
+      "loss": 1.3314,
+      "step": 1630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.1078852543648e-06,
+      "loss": 1.3287,
+      "step": 1631
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.080000940874558e-06,
+      "loss": 1.3508,
+      "step": 1632
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.05214990278202e-06,
+      "loss": 1.195,
+      "step": 1633
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.024332198300398e-06,
+      "loss": 1.308,
+      "step": 1634
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.99654788557326e-06,
+      "loss": 1.3225,
+      "step": 1635
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.968797022674333e-06,
+      "loss": 1.2934,
+      "step": 1636
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.941079667607458e-06,
+      "loss": 1.3588,
+      "step": 1637
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.913395878306435e-06,
+      "loss": 1.2525,
+      "step": 1638
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.885745712634906e-06,
+      "loss": 1.3352,
+      "step": 1639
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.858129228386236e-06,
+      "loss": 1.2486,
+      "step": 1640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.830546483283364e-06,
+      "loss": 1.331,
+      "step": 1641
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.802997534978761e-06,
+      "loss": 1.396,
+      "step": 1642
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.775482441054211e-06,
+      "loss": 1.2454,
+      "step": 1643
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.748001259020769e-06,
+      "loss": 1.4158,
+      "step": 1644
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.720554046318572e-06,
+      "loss": 1.2528,
+      "step": 1645
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.693140860316785e-06,
+      "loss": 1.3314,
+      "step": 1646
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.665761758313437e-06,
+      "loss": 1.3015,
+      "step": 1647
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.638416797535346e-06,
+      "loss": 1.3123,
+      "step": 1648
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.611106035137923e-06,
+      "loss": 1.334,
+      "step": 1649
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.583829528205123e-06,
+      "loss": 1.1809,
+      "step": 1650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.556587333749306e-06,
+      "loss": 1.2876,
+      "step": 1651
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.529379508711104e-06,
+      "loss": 1.3547,
+      "step": 1652
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.502206109959323e-06,
+      "loss": 1.3447,
+      "step": 1653
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.475067194290779e-06,
+      "loss": 1.3079,
+      "step": 1654
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.447962818430263e-06,
+      "loss": 1.3453,
+      "step": 1655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.420893039030336e-06,
+      "loss": 1.3521,
+      "step": 1656
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.393857912671269e-06,
+      "loss": 1.2517,
+      "step": 1657
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.36685749586087e-06,
+      "loss": 1.2988,
+      "step": 1658
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.33989184503443e-06,
+      "loss": 1.3595,
+      "step": 1659
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.312961016554566e-06,
+      "loss": 1.28,
+      "step": 1660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.286065066711101e-06,
+      "loss": 1.2538,
+      "step": 1661
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.259204051720965e-06,
+      "loss": 1.3806,
+      "step": 1662
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.232378027728067e-06,
+      "loss": 1.2424,
+      "step": 1663
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.205587050803174e-06,
+      "loss": 1.3846,
+      "step": 1664
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.178831176943806e-06,
+      "loss": 1.2361,
+      "step": 1665
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.152110462074113e-06,
+      "loss": 1.3598,
+      "step": 1666
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.125424962044742e-06,
+      "loss": 1.3428,
+      "step": 1667
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.098774732632742e-06,
+      "loss": 1.2953,
+      "step": 1668
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.072159829541463e-06,
+      "loss": 1.3226,
+      "step": 1669
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.045580308400391e-06,
+      "loss": 1.1837,
+      "step": 1670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.019036224765059e-06,
+      "loss": 1.2684,
+      "step": 1671
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.992527634116939e-06,
+      "loss": 1.3324,
+      "step": 1672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.966054591863314e-06,
+      "loss": 1.3519,
+      "step": 1673
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.939617153337169e-06,
+      "loss": 1.2718,
+      "step": 1674
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.913215373797064e-06,
+      "loss": 1.3104,
+      "step": 1675
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.886849308427034e-06,
+      "loss": 1.2566,
+      "step": 1676
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.860519012336453e-06,
+      "loss": 1.3568,
+      "step": 1677
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.834224540559948e-06,
+      "loss": 1.3267,
+      "step": 1678
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.807965948057264e-06,
+      "loss": 1.2541,
+      "step": 1679
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.781743289713128e-06,
+      "loss": 1.3324,
+      "step": 1680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.755556620337181e-06,
+      "loss": 1.1838,
+      "step": 1681
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.729405994663854e-06,
+      "loss": 1.3486,
+      "step": 1682
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.703291467352224e-06,
+      "loss": 1.3195,
+      "step": 1683
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.677213092985902e-06,
+      "loss": 1.3436,
+      "step": 1684
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.65117092607296e-06,
+      "loss": 1.2898,
+      "step": 1685
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.6251650210457715e-06,
+      "loss": 1.3409,
+      "step": 1686
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.599195432260933e-06,
+      "loss": 1.3063,
+      "step": 1687
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.573262213999122e-06,
+      "loss": 1.2759,
+      "step": 1688
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.547365420464999e-06,
+      "loss": 1.3552,
+      "step": 1689
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.521505105787094e-06,
+      "loss": 1.3218,
+      "step": 1690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.495681324017684e-06,
+      "loss": 1.2966,
+      "step": 1691
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.469894129132701e-06,
+      "loss": 1.3479,
+      "step": 1692
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.4441435750315755e-06,
+      "loss": 1.3699,
+      "step": 1693
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.4184297155371784e-06,
+      "loss": 1.2333,
+      "step": 1694
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.392752604395667e-06,
+      "loss": 1.354,
+      "step": 1695
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.367112295276421e-06,
+      "loss": 1.3488,
+      "step": 1696
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.34150884177185e-06,
+      "loss": 1.3469,
+      "step": 1697
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.31594229739736e-06,
+      "loss": 1.3781,
+      "step": 1698
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.290412715591205e-06,
+      "loss": 1.0942,
+      "step": 1699
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.264920149714377e-06,
+      "loss": 1.2761,
+      "step": 1700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.239464653050512e-06,
+      "loss": 1.3757,
+      "step": 1701
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.21404627880573e-06,
+      "loss": 1.2871,
+      "step": 1702
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.188665080108609e-06,
+      "loss": 1.4018,
+      "step": 1703
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.163321110009988e-06,
+      "loss": 1.3492,
+      "step": 1704
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.138014421482914e-06,
+      "loss": 1.1927,
+      "step": 1705
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.112745067422483e-06,
+      "loss": 1.3097,
+      "step": 1706
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.087513100645776e-06,
+      "loss": 1.3344,
+      "step": 1707
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.062318573891716e-06,
+      "loss": 1.2961,
+      "step": 1708
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.037161539821005e-06,
+      "loss": 1.3112,
+      "step": 1709
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.0120420510159294e-06,
+      "loss": 1.284,
+      "step": 1710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.986960159980327e-06,
+      "loss": 1.3209,
+      "step": 1711
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.9619159191394445e-06,
+      "loss": 1.3195,
+      "step": 1712
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.93690938083984e-06,
+      "loss": 1.3734,
+      "step": 1713
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.911940597349268e-06,
+      "loss": 1.282,
+      "step": 1714
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.887009620856538e-06,
+      "loss": 1.3298,
+      "step": 1715
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.862116503471486e-06,
+      "loss": 1.2613,
+      "step": 1716
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.837261297224787e-06,
+      "loss": 1.1616,
+      "step": 1717
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.812444054067885e-06,
+      "loss": 1.3381,
+      "step": 1718
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.787664825872858e-06,
+      "loss": 1.269,
+      "step": 1719
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.762923664432344e-06,
+      "loss": 1.3077,
+      "step": 1720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.738220621459415e-06,
+      "loss": 1.4048,
+      "step": 1721
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.713555748587461e-06,
+      "loss": 1.2348,
+      "step": 1722
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.6889290973700945e-06,
+      "loss": 1.3323,
+      "step": 1723
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.6643407192810395e-06,
+      "loss": 1.2909,
+      "step": 1724
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.6397906657140215e-06,
+      "loss": 1.3038,
+      "step": 1725
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.615278987982659e-06,
+      "loss": 1.3664,
+      "step": 1726
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.590805737320371e-06,
+      "loss": 1.3239,
+      "step": 1727
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.566370964880234e-06,
+      "loss": 1.316,
+      "step": 1728
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.541974721734906e-06,
+      "loss": 1.3608,
+      "step": 1729
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.517617058876541e-06,
+      "loss": 1.2851,
+      "step": 1730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.493298027216627e-06,
+      "loss": 1.2761,
+      "step": 1731
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.469017677585901e-06,
+      "loss": 1.3307,
+      "step": 1732
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.444776060734262e-06,
+      "loss": 1.3599,
+      "step": 1733
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.420573227330651e-06,
+      "loss": 1.2513,
+      "step": 1734
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.3964092279629375e-06,
+      "loss": 1.3242,
+      "step": 1735
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.37228411313783e-06,
+      "loss": 1.2853,
+      "step": 1736
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.348197933280758e-06,
+      "loss": 1.303,
+      "step": 1737
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.32415073873577e-06,
+      "loss": 1.3767,
+      "step": 1738
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.30014257976543e-06,
+      "loss": 1.2449,
+      "step": 1739
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.276173506550714e-06,
+      "loss": 1.3856,
+      "step": 1740
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.252243569190891e-06,
+      "loss": 1.1303,
+      "step": 1741
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.228352817703431e-06,
+      "loss": 1.2747,
+      "step": 1742
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.2045013020239265e-06,
+      "loss": 1.3475,
+      "step": 1743
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.180689072005935e-06,
+      "loss": 1.3247,
+      "step": 1744
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.156916177420896e-06,
+      "loss": 1.2946,
+      "step": 1745
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.1331826679580424e-06,
+      "loss": 1.2655,
+      "step": 1746
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.109488593224291e-06,
+      "loss": 1.3057,
+      "step": 1747
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.085834002744123e-06,
+      "loss": 1.3488,
+      "step": 1748
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.062218945959497e-06,
+      "loss": 1.3629,
+      "step": 1749
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.0386434722297384e-06,
+      "loss": 1.2806,
+      "step": 1750
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.015107630831435e-06,
+      "loss": 1.3511,
+      "step": 1751
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.991611470958339e-06,
+      "loss": 1.2356,
+      "step": 1752
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.968155041721271e-06,
+      "loss": 1.3031,
+      "step": 1753
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.94473839214798e-06,
+      "loss": 1.3142,
+      "step": 1754
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.921361571183098e-06,
+      "loss": 1.3551,
+      "step": 1755
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.8980246276879896e-06,
+      "loss": 1.2271,
+      "step": 1756
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.8747276104407e-06,
+      "loss": 1.3809,
+      "step": 1757
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.851470568135778e-06,
+      "loss": 1.3323,
+      "step": 1758
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.8282535493842435e-06,
+      "loss": 1.2363,
+      "step": 1759
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.805076602713464e-06,
+      "loss": 1.3427,
+      "step": 1760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.781939776567036e-06,
+      "loss": 1.3756,
+      "step": 1761
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.758843119304713e-06,
+      "loss": 1.2996,
+      "step": 1762
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.735786679202263e-06,
+      "loss": 1.2792,
+      "step": 1763
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.712770504451426e-06,
+      "loss": 1.2623,
+      "step": 1764
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.6897946431597626e-06,
+      "loss": 1.1807,
+      "step": 1765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.666859143350581e-06,
+      "loss": 1.3662,
+      "step": 1766
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.6439640529628055e-06,
+      "loss": 1.266,
+      "step": 1767
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.621109419850926e-06,
+      "loss": 1.3737,
+      "step": 1768
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.59829529178485e-06,
+      "loss": 1.2537,
+      "step": 1769
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.575521716449836e-06,
+      "loss": 1.2794,
+      "step": 1770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.552788741446374e-06,
+      "loss": 1.3501,
+      "step": 1771
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.530096414290092e-06,
+      "loss": 1.2661,
+      "step": 1772
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.507444782411664e-06,
+      "loss": 1.2874,
+      "step": 1773
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.484833893156693e-06,
+      "loss": 1.2907,
+      "step": 1774
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.462263793785644e-06,
+      "loss": 1.3442,
+      "step": 1775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4397345314736835e-06,
+      "loss": 1.3182,
+      "step": 1776
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.417246153310671e-06,
+      "loss": 1.3301,
+      "step": 1777
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.394798706300991e-06,
+      "loss": 1.237,
+      "step": 1778
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.372392237363458e-06,
+      "loss": 1.2702,
+      "step": 1779
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.350026793331259e-06,
+      "loss": 1.3461,
+      "step": 1780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.327702420951824e-06,
+      "loss": 1.2521,
+      "step": 1781
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.305419166886735e-06,
+      "loss": 1.348,
+      "step": 1782
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.283177077711632e-06,
+      "loss": 1.3217,
+      "step": 1783
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.260976199916118e-06,
+      "loss": 1.2681,
+      "step": 1784
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.238816579903647e-06,
+      "loss": 1.3191,
+      "step": 1785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.216698263991446e-06,
+      "loss": 1.3826,
+      "step": 1786
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.194621298410412e-06,
+      "loss": 1.273,
+      "step": 1787
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.172585729304996e-06,
+      "loss": 1.2781,
+      "step": 1788
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.150591602733146e-06,
+      "loss": 1.4024,
+      "step": 1789
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.128638964666166e-06,
+      "loss": 1.199,
+      "step": 1790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.106727860988678e-06,
+      "loss": 1.3178,
+      "step": 1791
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.0848583374984435e-06,
+      "loss": 1.3182,
+      "step": 1792
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.063030439906347e-06,
+      "loss": 1.314,
+      "step": 1793
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.041244213836257e-06,
+      "loss": 1.1558,
+      "step": 1794
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.0194997048249455e-06,
+      "loss": 1.3131,
+      "step": 1795
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.99779695832199e-06,
+      "loss": 1.3148,
+      "step": 1796
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.976136019689662e-06,
+      "loss": 1.3175,
+      "step": 1797
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.954516934202874e-06,
+      "loss": 1.2716,
+      "step": 1798
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.93293974704904e-06,
+      "loss": 1.329,
+      "step": 1799
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.91140450332801e-06,
+      "loss": 1.309,
+      "step": 1800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.889911248051951e-06,
+      "loss": 1.3249,
+      "step": 1801
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.868460026145278e-06,
+      "loss": 1.226,
+      "step": 1802
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.847050882444551e-06,
+      "loss": 1.3683,
+      "step": 1803
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8256838616983745e-06,
+      "loss": 1.2284,
+      "step": 1804
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.804359008567311e-06,
+      "loss": 1.3234,
+      "step": 1805
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.783076367623788e-06,
+      "loss": 1.3417,
+      "step": 1806
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.761835983351998e-06,
+      "loss": 1.2723,
+      "step": 1807
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7406379001478155e-06,
+      "loss": 1.2997,
+      "step": 1808
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.719482162318698e-06,
+      "loss": 1.3662,
+      "step": 1809
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.698368814083575e-06,
+      "loss": 1.2666,
+      "step": 1810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.677297899572813e-06,
+      "loss": 1.3206,
+      "step": 1811
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.656269462828055e-06,
+      "loss": 1.262,
+      "step": 1812
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6352835478021775e-06,
+      "loss": 1.3333,
+      "step": 1813
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.614340198359154e-06,
+      "loss": 1.1837,
+      "step": 1814
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.593439458274012e-06,
+      "loss": 1.2416,
+      "step": 1815
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.572581371232712e-06,
+      "loss": 1.3694,
+      "step": 1816
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.551765980832059e-06,
+      "loss": 1.3083,
+      "step": 1817
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.530993330579625e-06,
+      "loss": 1.2728,
+      "step": 1818
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.510263463893632e-06,
+      "loss": 1.3176,
+      "step": 1819
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.489576424102893e-06,
+      "loss": 1.2613,
+      "step": 1820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.468932254446698e-06,
+      "loss": 1.254,
+      "step": 1821
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.448330998074737e-06,
+      "loss": 1.33,
+      "step": 1822
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.427772698046987e-06,
+      "loss": 1.3399,
+      "step": 1823
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.407257397333655e-06,
+      "loss": 1.2698,
+      "step": 1824
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.386785138815083e-06,
+      "loss": 1.3433,
+      "step": 1825
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.36635596528163e-06,
+      "loss": 1.3321,
+      "step": 1826
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.345969919433593e-06,
+      "loss": 1.2626,
+      "step": 1827
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.325627043881148e-06,
+      "loss": 1.3497,
+      "step": 1828
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.3053273811442194e-06,
+      "loss": 1.3155,
+      "step": 1829
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.285070973652425e-06,
+      "loss": 1.2804,
+      "step": 1830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.264857863744956e-06,
+      "loss": 1.3101,
+      "step": 1831
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.244688093670518e-06,
+      "loss": 1.2719,
+      "step": 1832
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.224561705587226e-06,
+      "loss": 1.3566,
+      "step": 1833
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.204478741562512e-06,
+      "loss": 1.2146,
+      "step": 1834
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.18443924357306e-06,
+      "loss": 1.2647,
+      "step": 1835
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1644432535046796e-06,
+      "loss": 1.3109,
+      "step": 1836
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.144490813152258e-06,
+      "loss": 1.2964,
+      "step": 1837
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.124581964219662e-06,
+      "loss": 1.2603,
+      "step": 1838
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.104716748319637e-06,
+      "loss": 1.368,
+      "step": 1839
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.084895206973721e-06,
+      "loss": 1.2962,
+      "step": 1840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0651173816121716e-06,
+      "loss": 1.1978,
+      "step": 1841
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.045383313573878e-06,
+      "loss": 1.3292,
+      "step": 1842
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.025693044106263e-06,
+      "loss": 1.2192,
+      "step": 1843
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.006046614365208e-06,
+      "loss": 1.3429,
+      "step": 1844
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.986444065414954e-06,
+      "loss": 1.3225,
+      "step": 1845
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.966885438228035e-06,
+      "loss": 1.3895,
+      "step": 1846
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.947370773685172e-06,
+      "loss": 1.2308,
+      "step": 1847
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.927900112575209e-06,
+      "loss": 1.355,
+      "step": 1848
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.9084734955949915e-06,
+      "loss": 1.2648,
+      "step": 1849
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.889090963349329e-06,
+      "loss": 1.2626,
+      "step": 1850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.869752556350875e-06,
+      "loss": 1.346,
+      "step": 1851
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.850458315020075e-06,
+      "loss": 1.255,
+      "step": 1852
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.831208279685025e-06,
+      "loss": 1.338,
+      "step": 1853
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.8120024905814513e-06,
+      "loss": 1.307,
+      "step": 1854
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.792840987852583e-06,
+      "loss": 1.2569,
+      "step": 1855
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.773723811549096e-06,
+      "loss": 1.3734,
+      "step": 1856
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.754651001629014e-06,
+      "loss": 1.3032,
+      "step": 1857
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7356225979576055e-06,
+      "loss": 1.3142,
+      "step": 1858
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.716638640307363e-06,
+      "loss": 1.194,
+      "step": 1859
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6976991683578487e-06,
+      "loss": 1.3463,
+      "step": 1860
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6788042216956557e-06,
+      "loss": 1.2289,
+      "step": 1861
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6599538398143007e-06,
+      "loss": 1.3324,
+      "step": 1862
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6411480621141663e-06,
+      "loss": 1.2851,
+      "step": 1863
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6223869279023937e-06,
+      "loss": 1.2675,
+      "step": 1864
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6036704763928204e-06,
+      "loss": 1.2849,
+      "step": 1865
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5849987467058866e-06,
+      "loss": 1.284,
+      "step": 1866
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.566371777868557e-06,
+      "loss": 1.2441,
+      "step": 1867
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.547789608814234e-06,
+      "loss": 1.3925,
+      "step": 1868
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5292522783826913e-06,
+      "loss": 1.3085,
+      "step": 1869
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5107598253199758e-06,
+      "loss": 1.1736,
+      "step": 1870
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4923122882783237e-06,
+      "loss": 1.2971,
+      "step": 1871
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4739097058161114e-06,
+      "loss": 1.2606,
+      "step": 1872
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4555521163977395e-06,
+      "loss": 1.3186,
+      "step": 1873
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4372395583935705e-06,
+      "loss": 1.3682,
+      "step": 1874
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.418972070079829e-06,
+      "loss": 1.2721,
+      "step": 1875
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.400749689638558e-06,
+      "loss": 1.3332,
+      "step": 1876
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3825724551575068e-06,
+      "loss": 1.3435,
+      "step": 1877
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.364440404630062e-06,
+      "loss": 1.202,
+      "step": 1878
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3463535759551716e-06,
+      "loss": 1.3033,
+      "step": 1879
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.328312006937262e-06,
+      "loss": 1.3476,
+      "step": 1880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.310315735286157e-06,
+      "loss": 1.2832,
+      "step": 1881
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.292364798617004e-06,
+      "loss": 1.3669,
+      "step": 1882
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2744592344501945e-06,
+      "loss": 1.2715,
+      "step": 1883
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.256599080211273e-06,
+      "loss": 1.3452,
+      "step": 1884
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2387843732308798e-06,
+      "loss": 1.1967,
+      "step": 1885
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.221015150744669e-06,
+      "loss": 1.2141,
+      "step": 1886
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2032914498932226e-06,
+      "loss": 1.3193,
+      "step": 1887
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1856133077219517e-06,
+      "loss": 1.3024,
+      "step": 1888
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.167980761181072e-06,
+      "loss": 1.3006,
+      "step": 1889
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1503938471254796e-06,
+      "loss": 1.265,
+      "step": 1890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.132852602314701e-06,
+      "loss": 1.3571,
+      "step": 1891
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.115357063412802e-06,
+      "loss": 1.2562,
+      "step": 1892
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.097907266988312e-06,
+      "loss": 1.2762,
+      "step": 1893
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0805032495141577e-06,
+      "loss": 1.3382,
+      "step": 1894
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.06314504736758e-06,
+      "loss": 1.2711,
+      "step": 1895
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.045832696830056e-06,
+      "loss": 1.311,
+      "step": 1896
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0285662340872184e-06,
+      "loss": 1.2717,
+      "step": 1897
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.011345695228798e-06,
+      "loss": 1.2852,
+      "step": 1898
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9941711162485287e-06,
+      "loss": 1.3487,
+      "step": 1899
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9770425330441025e-06,
+      "loss": 1.3219,
+      "step": 1900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9599599814170346e-06,
+      "loss": 1.2664,
+      "step": 1901
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9429234970726553e-06,
+      "loss": 1.3311,
+      "step": 1902
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.92593311562e-06,
+      "loss": 1.2503,
+      "step": 1903
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9089888725717387e-06,
+      "loss": 1.3357,
+      "step": 1904
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.892090803344105e-06,
+      "loss": 1.3464,
+      "step": 1905
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.875238943256828e-06,
+      "loss": 1.2551,
+      "step": 1906
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8584333275330415e-06,
+      "loss": 1.2946,
+      "step": 1907
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8416739912992318e-06,
+      "loss": 1.3555,
+      "step": 1908
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.824960969585147e-06,
+      "loss": 1.1213,
+      "step": 1909
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.80829429732373e-06,
+      "loss": 1.3251,
+      "step": 1910
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7916740093510447e-06,
+      "loss": 1.2808,
+      "step": 1911
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.775100140406206e-06,
+      "loss": 1.3367,
+      "step": 1912
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.75857272513132e-06,
+      "loss": 1.2452,
+      "step": 1913
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7420917980713694e-06,
+      "loss": 1.2369,
+      "step": 1914
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7256573936741886e-06,
+      "loss": 1.2984,
+      "step": 1915
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7092695462903632e-06,
+      "loss": 1.3825,
+      "step": 1916
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6929282901731746e-06,
+      "loss": 1.3227,
+      "step": 1917
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6766336594785196e-06,
+      "loss": 1.2001,
+      "step": 1918
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6603856882648216e-06,
+      "loss": 1.3585,
+      "step": 1919
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6441844104930088e-06,
+      "loss": 1.21,
+      "step": 1920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.628029860026393e-06,
+      "loss": 1.3494,
+      "step": 1921
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.6119220706306293e-06,
+      "loss": 1.3492,
+      "step": 1922
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.595861075973613e-06,
+      "loss": 1.28,
+      "step": 1923
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.579846909625455e-06,
+      "loss": 1.3344,
+      "step": 1924
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.563879605058375e-06,
+      "loss": 1.2706,
+      "step": 1925
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5479591956466515e-06,
+      "loss": 1.3577,
+      "step": 1926
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.532085714666535e-06,
+      "loss": 1.244,
+      "step": 1927
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5162591952961946e-06,
+      "loss": 1.2636,
+      "step": 1928
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5004796706156436e-06,
+      "loss": 1.3242,
+      "step": 1929
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.484747173606658e-06,
+      "loss": 1.1775,
+      "step": 1930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4690617371527364e-06,
+      "loss": 1.3052,
+      "step": 1931
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4534233940389924e-06,
+      "loss": 1.311,
+      "step": 1932
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4378321769521125e-06,
+      "loss": 1.2941,
+      "step": 1933
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.422288118480298e-06,
+      "loss": 1.2795,
+      "step": 1934
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4067912511131747e-06,
+      "loss": 1.3651,
+      "step": 1935
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3913416072417065e-06,
+      "loss": 1.2908,
+      "step": 1936
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.375939219158185e-06,
+      "loss": 1.2265,
+      "step": 1937
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.360584119056114e-06,
+      "loss": 1.3328,
+      "step": 1938
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3452763390301587e-06,
+      "loss": 1.3315,
+      "step": 1939
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3300159110760804e-06,
+      "loss": 1.2527,
+      "step": 1940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3148028670906653e-06,
+      "loss": 1.2961,
+      "step": 1941
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2996372388716576e-06,
+      "loss": 1.3315,
+      "step": 1942
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2845190581177014e-06,
+      "loss": 1.2195,
+      "step": 1943
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2694483564282636e-06,
+      "loss": 1.3387,
+      "step": 1944
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.254425165303564e-06,
+      "loss": 1.2916,
+      "step": 1945
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2394495161445235e-06,
+      "loss": 1.3136,
+      "step": 1946
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2245214402527054e-06,
+      "loss": 1.3447,
+      "step": 1947
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.209640968830229e-06,
+      "loss": 1.2328,
+      "step": 1948
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.194808132979695e-06,
+      "loss": 1.2365,
+      "step": 1949
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1800229637041566e-06,
+      "loss": 1.2128,
+      "step": 1950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1652854919070374e-06,
+      "loss": 1.2899,
+      "step": 1951
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1505957483920564e-06,
+      "loss": 1.3287,
+      "step": 1952
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.135953763863177e-06,
+      "loss": 1.3537,
+      "step": 1953
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1213595689245386e-06,
+      "loss": 1.2795,
+      "step": 1954
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.10681319408039e-06,
+      "loss": 1.3006,
+      "step": 1955
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.092314669735032e-06,
+      "loss": 1.2573,
+      "step": 1956
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0778640261927498e-06,
+      "loss": 1.3141,
+      "step": 1957
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0634612936577426e-06,
+      "loss": 1.3123,
+      "step": 1958
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0491065022340742e-06,
+      "loss": 1.3686,
+      "step": 1959
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0347996819256033e-06,
+      "loss": 1.3132,
+      "step": 1960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.020540862635928e-06,
+      "loss": 1.1813,
+      "step": 1961
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.0063300741683e-06,
+      "loss": 1.2886,
+      "step": 1962
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9921673462255904e-06,
+      "loss": 1.2799,
+      "step": 1963
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9780527084102135e-06,
+      "loss": 1.3605,
+      "step": 1964
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.963986190224068e-06,
+      "loss": 1.2883,
+      "step": 1965
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.949967821068471e-06,
+      "loss": 1.2997,
+      "step": 1966
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9359976302441075e-06,
+      "loss": 1.2938,
+      "step": 1967
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9220756469509547e-06,
+      "loss": 1.1613,
+      "step": 1968
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9082019002882338e-06,
+      "loss": 1.3247,
+      "step": 1969
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8943764192543427e-06,
+      "loss": 1.3253,
+      "step": 1970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8805992327467853e-06,
+      "loss": 1.284,
+      "step": 1971
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8668703695621347e-06,
+      "loss": 1.3374,
+      "step": 1972
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8531898583959567e-06,
+      "loss": 1.3853,
+      "step": 1973
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.839557727842764e-06,
+      "loss": 1.1632,
+      "step": 1974
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8259740063959224e-06,
+      "loss": 1.3252,
+      "step": 1975
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8124387224476347e-06,
+      "loss": 1.3229,
+      "step": 1976
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7989519042888508e-06,
+      "loss": 1.2577,
+      "step": 1977
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7855135801092243e-06,
+      "loss": 1.3477,
+      "step": 1978
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7721237779970506e-06,
+      "loss": 1.0991,
+      "step": 1979
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7587825259391955e-06,
+      "loss": 1.3297,
+      "step": 1980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7454898518210582e-06,
+      "loss": 1.3626,
+      "step": 1981
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7322457834265004e-06,
+      "loss": 1.2351,
+      "step": 1982
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7190503484377895e-06,
+      "loss": 1.3964,
+      "step": 1983
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7059035744355328e-06,
+      "loss": 1.2968,
+      "step": 1984
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6928054888986356e-06,
+      "loss": 1.1971,
+      "step": 1985
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.67975611920424e-06,
+      "loss": 1.3235,
+      "step": 1986
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6667554926276556e-06,
+      "loss": 1.3404,
+      "step": 1987
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6538036363423159e-06,
+      "loss": 1.2679,
+      "step": 1988
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6409005774197183e-06,
+      "loss": 1.3157,
+      "step": 1989
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.628046342829359e-06,
+      "loss": 1.3093,
+      "step": 1990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.615240959438691e-06,
+      "loss": 1.2669,
+      "step": 1991
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6024844540130562e-06,
+      "loss": 1.297,
+      "step": 1992
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5897768532156298e-06,
+      "loss": 1.3396,
+      "step": 1993
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5771181836073706e-06,
+      "loss": 1.3003,
+      "step": 1994
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5645084716469777e-06,
+      "loss": 1.3553,
+      "step": 1995
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5519477436908015e-06,
+      "loss": 1.2182,
+      "step": 1996
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5394360259928075e-06,
+      "loss": 1.2769,
+      "step": 1997
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.526973344704538e-06,
+      "loss": 1.1877,
+      "step": 1998
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5145597258750282e-06,
+      "loss": 1.3114,
+      "step": 1999
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5021951954507702e-06,
+      "loss": 1.2558,
+      "step": 2000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4898797792756492e-06,
+      "loss": 1.3648,
+      "step": 2001
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4776135030908966e-06,
+      "loss": 1.2036,
+      "step": 2002
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4653963925350345e-06,
+      "loss": 1.3247,
+      "step": 2003
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4532284731438196e-06,
+      "loss": 1.3297,
+      "step": 2004
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4411097703501909e-06,
+      "loss": 1.1878,
+      "step": 2005
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4290403094842115e-06,
+      "loss": 1.3279,
+      "step": 2006
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4170201157730213e-06,
+      "loss": 1.2853,
+      "step": 2007
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4050492143408006e-06,
+      "loss": 1.3019,
+      "step": 2008
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3931276302086877e-06,
+      "loss": 1.3011,
+      "step": 2009
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3812553882947304e-06,
+      "loss": 1.2911,
+      "step": 2010
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.369432513413857e-06,
+      "loss": 1.2833,
+      "step": 2011
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3576590302778052e-06,
+      "loss": 1.2988,
+      "step": 2012
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3459349634950825e-06,
+      "loss": 1.3652,
+      "step": 2013
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3342603375708973e-06,
+      "loss": 1.1814,
+      "step": 2014
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3226351769071278e-06,
+      "loss": 1.3358,
+      "step": 2015
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3110595058022562e-06,
+      "loss": 1.301,
+      "step": 2016
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2995333484513289e-06,
+      "loss": 1.3248,
+      "step": 2017
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.288056728945894e-06,
+      "loss": 1.3167,
+      "step": 2018
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2766296712739573e-06,
+      "loss": 1.2685,
+      "step": 2019
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2652521993199412e-06,
+      "loss": 1.2974,
+      "step": 2020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.253924336864612e-06,
+      "loss": 1.1374,
+      "step": 2021
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2426461075850633e-06,
+      "loss": 1.3033,
+      "step": 2022
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2314175350546298e-06,
+      "loss": 1.3954,
+      "step": 2023
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2202386427428574e-06,
+      "loss": 1.3214,
+      "step": 2024
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.209109454015464e-06,
+      "loss": 1.2127,
+      "step": 2025
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1980299921342697e-06,
+      "loss": 1.2392,
+      "step": 2026
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1870002802571583e-06,
+      "loss": 1.3131,
+      "step": 2027
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1760203414380222e-06,
+      "loss": 1.3222,
+      "step": 2028
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1650901986267365e-06,
+      "loss": 1.3774,
+      "step": 2029
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1542098746690793e-06,
+      "loss": 1.2922,
+      "step": 2030
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1433793923067093e-06,
+      "loss": 1.3064,
+      "step": 2031
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1325987741770961e-06,
+      "loss": 1.273,
+      "step": 2032
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1218680428134954e-06,
+      "loss": 1.3298,
+      "step": 2033
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1111872206448854e-06,
+      "loss": 1.2339,
+      "step": 2034
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1005563299959388e-06,
+      "loss": 1.334,
+      "step": 2035
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0899753930869394e-06,
+      "loss": 1.2688,
+      "step": 2036
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0794444320337827e-06,
+      "loss": 1.3431,
+      "step": 2037
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.068963468847889e-06,
+      "loss": 1.3168,
+      "step": 2038
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.05853252543619e-06,
+      "loss": 1.269,
+      "step": 2039
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0481516236010623e-06,
+      "loss": 1.3604,
+      "step": 2040
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0378207850402749e-06,
+      "loss": 1.3098,
+      "step": 2041
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0275400313469768e-06,
+      "loss": 1.2465,
+      "step": 2042
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0173093840096182e-06,
+      "loss": 1.2716,
+      "step": 2043
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0071288644119269e-06,
+      "loss": 1.334,
+      "step": 2044
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.969984938328486e-07,
+      "loss": 1.115,
+      "step": 2045
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.869182934465094e-07,
+      "loss": 1.379,
+      "step": 2046
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.768882843221749e-07,
+      "loss": 1.2502,
+      "step": 2047
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.669084874242062e-07,
+      "loss": 1.3205,
+      "step": 2048
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.569789236120035e-07,
+      "loss": 1.2866,
+      "step": 2049
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.470996136399818e-07,
+      "loss": 1.2139,
+      "step": 2050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.372705781575064e-07,
+      "loss": 1.3576,
+      "step": 2051
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.274918377088714e-07,
+      "loss": 1.3117,
+      "step": 2052
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.177634127332412e-07,
+      "loss": 1.2988,
+      "step": 2053
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.080853235646086e-07,
+      "loss": 1.347,
+      "step": 2054
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.984575904317616e-07,
+      "loss": 1.2528,
+      "step": 2055
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.888802334582363e-07,
+      "loss": 1.2573,
+      "step": 2056
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.793532726622727e-07,
+      "loss": 1.3313,
+      "step": 2057
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.698767279567726e-07,
+      "loss": 1.3031,
+      "step": 2058
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.604506191492639e-07,
+      "loss": 1.2191,
+      "step": 2059
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.510749659418532e-07,
+      "loss": 1.3251,
+      "step": 2060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.417497879311842e-07,
+      "loss": 1.3441,
+      "step": 2061
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.324751046084017e-07,
+      "loss": 1.2546,
+      "step": 2062
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.232509353591128e-07,
+      "loss": 1.2591,
+      "step": 2063
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.140772994633339e-07,
+      "loss": 1.3124,
+      "step": 2064
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.049542160954632e-07,
+      "loss": 1.281,
+      "step": 2065
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.958817043242389e-07,
+      "loss": 1.3032,
+      "step": 2066
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.868597831126812e-07,
+      "loss": 1.2422,
+      "step": 2067
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.778884713180862e-07,
+      "loss": 1.317,
+      "step": 2068
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.689677876919599e-07,
+      "loss": 1.3529,
+      "step": 2069
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.600977508799901e-07,
+      "loss": 1.2377,
+      "step": 2070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.512783794219963e-07,
+      "loss": 1.2609,
+      "step": 2071
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.425096917519109e-07,
+      "loss": 1.3651,
+      "step": 2072
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.337917061977173e-07,
+      "loss": 1.2413,
+      "step": 2073
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.25124440981434e-07,
+      "loss": 1.1712,
+      "step": 2074
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.165079142190583e-07,
+      "loss": 1.3007,
+      "step": 2075
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.079421439205369e-07,
+      "loss": 1.2991,
+      "step": 2076
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.994271479897314e-07,
+      "loss": 1.3711,
+      "step": 2077
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.909629442243715e-07,
+      "loss": 1.2469,
+      "step": 2078
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.825495503160279e-07,
+      "loss": 1.3655,
+      "step": 2079
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.741869838500642e-07,
+      "loss": 1.2716,
+      "step": 2080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.658752623056124e-07,
+      "loss": 1.3199,
+      "step": 2081
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.576144030555259e-07,
+      "loss": 1.2162,
+      "step": 2082
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.49404423366351e-07,
+      "loss": 1.3345,
+      "step": 2083
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.412453403982832e-07,
+      "loss": 1.2722,
+      "step": 2084
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.331371712051448e-07,
+      "loss": 1.3082,
+      "step": 2085
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.250799327343265e-07,
+      "loss": 1.3744,
+      "step": 2086
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.170736418267765e-07,
+      "loss": 1.2591,
+      "step": 2087
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.0911831521695e-07,
+      "loss": 1.3073,
+      "step": 2088
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.01213969532774e-07,
+      "loss": 1.3868,
+      "step": 2089
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.933606212956244e-07,
+      "loss": 1.2362,
+      "step": 2090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.855582869202875e-07,
+      "loss": 1.2692,
+      "step": 2091
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.778069827149096e-07,
+      "loss": 1.2394,
+      "step": 2092
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.701067248809811e-07,
+      "loss": 1.3485,
+      "step": 2093
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.624575295133022e-07,
+      "loss": 1.1474,
+      "step": 2094
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.548594125999396e-07,
+      "loss": 1.2548,
+      "step": 2095
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.473123900221949e-07,
+      "loss": 1.3234,
+      "step": 2096
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.398164775545833e-07,
+      "loss": 1.3015,
+      "step": 2097
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.32371690864783e-07,
+      "loss": 1.1665,
+      "step": 2098
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.249780455136133e-07,
+      "loss": 1.3719,
+      "step": 2099
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.176355569550012e-07,
+      "loss": 1.2716,
+      "step": 2100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.103442405359482e-07,
+      "loss": 1.2647,
+      "step": 2101
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.031041114964913e-07,
+      "loss": 1.3746,
+      "step": 2102
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.959151849696864e-07,
+      "loss": 1.3134,
+      "step": 2103
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.887774759815667e-07,
+      "loss": 1.2699,
+      "step": 2104
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.816909994511037e-07,
+      "loss": 1.3119,
+      "step": 2105
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.746557701901938e-07,
+      "loss": 1.3203,
+      "step": 2106
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6767180290361314e-07,
+      "loss": 1.2578,
+      "step": 2107
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.607391121889959e-07,
+      "loss": 1.3244,
+      "step": 2108
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5385771253679546e-07,
+      "loss": 1.3316,
+      "step": 2109
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4702761833026184e-07,
+      "loss": 1.2291,
+      "step": 2110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.402488438454089e-07,
+      "loss": 1.2682,
+      "step": 2111
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.3352140325098046e-07,
+      "loss": 1.2974,
+      "step": 2112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.268453106084286e-07,
+      "loss": 1.3129,
+      "step": 2113
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.202205798718717e-07,
+      "loss": 1.3832,
+      "step": 2114
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.136472248880835e-07,
+      "loss": 1.1342,
+      "step": 2115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.071252593964431e-07,
+      "loss": 1.278,
+      "step": 2116
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.006546970289349e-07,
+      "loss": 1.3073,
+      "step": 2117
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.9423555131007925e-07,
+      "loss": 1.2836,
+      "step": 2118
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.878678356569465e-07,
+      "loss": 1.3301,
+      "step": 2119
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.8155156337909836e-07,
+      "loss": 1.2956,
+      "step": 2120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.752867476785743e-07,
+      "loss": 1.2862,
+      "step": 2121
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.6907340164986926e-07,
+      "loss": 1.2602,
+      "step": 2122
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.629115382798809e-07,
+      "loss": 1.1962,
+      "step": 2123
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.568011704479152e-07,
+      "loss": 1.3104,
+      "step": 2124
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.5074231092563913e-07,
+      "loss": 1.3267,
+      "step": 2125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4473497237706156e-07,
+      "loss": 1.3561,
+      "step": 2126
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.387791673584939e-07,
+      "loss": 1.2154,
+      "step": 2127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.3287490831854506e-07,
+      "loss": 1.3851,
+      "step": 2128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2702220759807666e-07,
+      "loss": 1.2311,
+      "step": 2129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.212210774301977e-07,
+      "loss": 1.2545,
+      "step": 2130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.154715299402117e-07,
+      "loss": 1.3724,
+      "step": 2131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.097735771456084e-07,
+      "loss": 1.2009,
+      "step": 2132
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.041272309560472e-07,
+      "loss": 1.3174,
+      "step": 2133
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9853250317330716e-07,
+      "loss": 1.3274,
+      "step": 2134
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9298940549128964e-07,
+      "loss": 1.2824,
+      "step": 2135
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8749794949596566e-07,
+      "loss": 1.3244,
+      "step": 2136
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8205814666537876e-07,
+      "loss": 1.3101,
+      "step": 2137
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.766700083696089e-07,
+      "loss": 1.265,
+      "step": 2138
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.713335458707417e-07,
+      "loss": 1.1097,
+      "step": 2139
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.660487703228548e-07,
+      "loss": 1.3026,
+      "step": 2140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.6081569277199294e-07,
+      "loss": 1.1653,
+      "step": 2141
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.5563432415613987e-07,
+      "loss": 1.3393,
+      "step": 2142
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.505046753052076e-07,
+      "loss": 1.2918,
+      "step": 2143
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.454267569409946e-07,
+      "loss": 1.3117,
+      "step": 2144
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.40400579677183e-07,
+      "loss": 1.267,
+      "step": 2145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3542615401930257e-07,
+      "loss": 1.2776,
+      "step": 2146
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3050349036471686e-07,
+      "loss": 1.2736,
+      "step": 2147
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2563259900259536e-07,
+      "loss": 1.343,
+      "step": 2148
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2081349011389418e-07,
+      "loss": 1.3201,
+      "step": 2149
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.160461737713393e-07,
+      "loss": 1.2338,
+      "step": 2150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.113306599394044e-07,
+      "loss": 1.3181,
+      "step": 2151
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.066669584742803e-07,
+      "loss": 1.2674,
+      "step": 2152
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0205507912386124e-07,
+      "loss": 1.3157,
+      "step": 2153
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9749503152773352e-07,
+      "loss": 1.344,
+      "step": 2154
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9298682521713963e-07,
+      "loss": 1.2495,
+      "step": 2155
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8853046961496423e-07,
+      "loss": 1.3105,
+      "step": 2156
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8412597403572041e-07,
+      "loss": 1.3339,
+      "step": 2157
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7977334768552456e-07,
+      "loss": 1.2266,
+      "step": 2158
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7547259966207708e-07,
+      "loss": 1.3203,
+      "step": 2159
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.712237389546373e-07,
+      "loss": 1.2857,
+      "step": 2160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6702677444402627e-07,
+      "loss": 1.2765,
+      "step": 2161
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6288171490257963e-07,
+      "loss": 1.338,
+      "step": 2162
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.587885689941504e-07,
+      "loss": 1.2513,
+      "step": 2163
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5474734527408108e-07,
+      "loss": 1.3288,
+      "step": 2164
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.507580521891927e-07,
+      "loss": 1.1633,
+      "step": 2165
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4682069807775421e-07,
+      "loss": 1.2423,
+      "step": 2166
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.4293529116947978e-07,
+      "loss": 1.3026,
+      "step": 2167
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.391018395855065e-07,
+      "loss": 1.3516,
+      "step": 2168
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3532035133837217e-07,
+      "loss": 1.2869,
+      "step": 2169
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3159083433200715e-07,
+      "loss": 1.2953,
+      "step": 2170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2791329636170912e-07,
+      "loss": 1.2862,
+      "step": 2171
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2428774511413222e-07,
+      "loss": 1.2661,
+      "step": 2172
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.20714188167273e-07,
+      "loss": 1.1873,
+      "step": 2173
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1719263299044825e-07,
+      "loss": 1.3269,
+      "step": 2174
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1372308694428125e-07,
+      "loss": 1.2835,
+      "step": 2175
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1030555728069325e-07,
+      "loss": 1.3029,
+      "step": 2176
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0694005114287587e-07,
+      "loss": 1.249,
+      "step": 2177
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0362657556528822e-07,
+      "loss": 1.3275,
+      "step": 2178
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0036513747363474e-07,
+      "loss": 1.3128,
+      "step": 2179
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.715574368485692e-08,
+      "loss": 1.3491,
+      "step": 2180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.399840090710544e-08,
+      "loss": 1.2711,
+      "step": 2181
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.089311573974857e-08,
+      "loss": 1.3096,
+      "step": 2182
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.783989467333609e-08,
+      "loss": 1.2453,
+      "step": 2183
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.483874408959647e-08,
+      "loss": 1.3368,
+      "step": 2184
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.188967026143135e-08,
+      "loss": 1.37,
+      "step": 2185
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.899267935287946e-08,
+      "loss": 1.2237,
+      "step": 2186
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.614777741913048e-08,
+      "loss": 1.2644,
+      "step": 2187
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.335497040648898e-08,
+      "loss": 1.3675,
+      "step": 2188
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.061426415237992e-08,
+      "loss": 1.026,
+      "step": 2189
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.792566438532932e-08,
+      "loss": 1.3266,
+      "step": 2190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.52891767249475e-08,
+      "loss": 1.3164,
+      "step": 2191
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.270480668192358e-08,
+      "loss": 1.3313,
+      "step": 2192
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.017255965801161e-08,
+      "loss": 1.3392,
+      "step": 2193
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.769244094602499e-08,
+      "loss": 1.1914,
+      "step": 2194
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.526445572981709e-08,
+      "loss": 1.2619,
+      "step": 2195
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.2888609084270066e-08,
+      "loss": 1.2899,
+      "step": 2196
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.0564905975292174e-08,
+      "loss": 1.3346,
+      "step": 2197
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.829335125980383e-08,
+      "loss": 1.1302,
+      "step": 2198
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.607394968572376e-08,
+      "loss": 1.3394,
+      "step": 2199
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.390670589196622e-08,
+      "loss": 1.2302,
+      "step": 2200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.1791624408421546e-08,
+      "loss": 1.3265,
+      "step": 2201
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.972870965595621e-08,
+      "loss": 1.325,
+      "step": 2202
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7717965946401644e-08,
+      "loss": 1.2608,
+      "step": 2203
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.575939748253765e-08,
+      "loss": 1.3667,
+      "step": 2204
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.385300835809235e-08,
+      "loss": 1.225,
+      "step": 2205
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.199880255773113e-08,
+      "loss": 1.3422,
+      "step": 2206
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.01967839570455e-08,
+      "loss": 1.2843,
+      "step": 2207
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8446956322550344e-08,
+      "loss": 1.3131,
+      "step": 2208
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.674932331167279e-08,
+      "loss": 1.2509,
+      "step": 2209
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5103888472738367e-08,
+      "loss": 1.1701,
+      "step": 2210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.351065524497653e-08,
+      "loss": 1.3128,
+      "step": 2211
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1969626958506795e-08,
+      "loss": 1.2392,
+      "step": 2212
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0480806834330402e-08,
+      "loss": 1.3656,
+      "step": 2213
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9044197984319222e-08,
+      "loss": 1.2622,
+      "step": 2214
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.765980341122686e-08,
+      "loss": 1.3135,
+      "step": 2215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6327626008658115e-08,
+      "loss": 1.3486,
+      "step": 2216
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5047668561085636e-08,
+      "loss": 1.2617,
+      "step": 2217
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3819933743827729e-08,
+      "loss": 1.2846,
+      "step": 2218
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2644424123051113e-08,
+      "loss": 1.3159,
+      "step": 2219
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.152114215576261e-08,
+      "loss": 1.2509,
+      "step": 2220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0450090189803586e-08,
+      "loss": 1.3292,
+      "step": 2221
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.431270463852725e-09,
+      "loss": 1.2105,
+      "step": 2222
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.46468510740106e-09,
+      "loss": 1.2406,
+      "step": 2223
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.550336140776936e-09,
+      "loss": 1.3241,
+      "step": 2224
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.688225475118271e-09,
+      "loss": 1.3325,
+      "step": 2225
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.878354912375316e-09,
+      "loss": 1.2572,
+      "step": 2226
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.120726145307897e-09,
+      "loss": 1.3758,
+      "step": 2227
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.415340757485398e-09,
+      "loss": 1.3424,
+      "step": 2228
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.76220022328122e-09,
+      "loss": 1.1988,
+      "step": 2229
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1613059078644536e-09,
+      "loss": 1.1578,
+      "step": 2230
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6126590671998742e-09,
+      "loss": 1.3002,
+      "step": 2231
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1162608480507217e-09,
+      "loss": 1.3046,
+      "step": 2232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6721122879703732e-09,
+      "loss": 1.3116,
+      "step": 2233
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2802143152995661e-09,
+      "loss": 1.3122,
+      "step": 2234
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.405677491747255e-10,
+      "loss": 1.3079,
+      "step": 2235
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.531732995057605e-10,
+      "loss": 1.2328,
+      "step": 2236
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1803156700104264e-10,
+      "loss": 1.2879,
+      "step": 2237
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.351430431396517e-10,
+      "loss": 1.3624,
+      "step": 2238
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0450811018802854e-10,
+      "loss": 1.2655,
+      "step": 2239
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6127041199974777e-11,
+      "loss": 1.3061,
+      "step": 2240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.1817,
+      "step": 2241
+    },
+    {
+      "epoch": 1.0,
+      "step": 2241,
+      "total_flos": 0.0,
+      "train_loss": 0.25396791189603113,
+      "train_runtime": 8319.8841,
+      "train_samples_per_second": 552.686,
+      "train_steps_per_second": 0.269
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2241,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 200,
+  "total_flos": 0.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}