diff --git "a/VQA_lora/vqa/checkpoint-6500/trainer_state.json" "b/VQA_lora/vqa/checkpoint-6500/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/VQA_lora/vqa/checkpoint-6500/trainer_state.json"
@@ -0,0 +1,39016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.701627486437613,
+  "global_step": 6500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.615384615384617e-08,
+      "loss": 10.3394,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9230769230769234e-07,
+      "loss": 10.3294,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.884615384615385e-07,
+      "loss": 10.2876,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.846153846153847e-07,
+      "loss": 10.2936,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.807692307692308e-07,
+      "loss": 10.2744,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.76923076923077e-07,
+      "loss": 10.224,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.730769230769231e-07,
+      "loss": 10.24,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 10.1923,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.653846153846154e-07,
+      "loss": 10.1606,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.615384615384617e-07,
+      "loss": 10.0549,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0576923076923078e-06,
+      "loss": 10.0548,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.153846153846154e-06,
+      "loss": 9.9685,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.25e-06,
+      "loss": 9.8973,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3461538461538462e-06,
+      "loss": 9.8162,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4423076923076922e-06,
+      "loss": 9.753,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 9.6169,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6346153846153848e-06,
+      "loss": 9.5226,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7307692307692308e-06,
+      "loss": 9.4535,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.826923076923077e-06,
+      "loss": 9.2957,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9230769230769234e-06,
+      "loss": 9.1875,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0192307692307692e-06,
+      "loss": 9.083,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1153846153846155e-06,
+      "loss": 8.9262,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.211538461538462e-06,
+      "loss": 8.7582,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 8.6682,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.403846153846154e-06,
+      "loss": 8.4646,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5e-06,
+      "loss": 8.3287,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5961538461538465e-06,
+      "loss": 8.1129,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6923076923076923e-06,
+      "loss": 7.9287,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7884615384615386e-06,
+      "loss": 7.7177,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8846153846153845e-06,
+      "loss": 7.5651,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.980769230769231e-06,
+      "loss": 7.341,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 7.137,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1730769230769233e-06,
+      "loss": 6.9006,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2692307692307696e-06,
+      "loss": 6.6547,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.365384615384616e-06,
+      "loss": 6.4136,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4615384615384617e-06,
+      "loss": 6.1879,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.557692307692308e-06,
+      "loss": 5.9355,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.653846153846154e-06,
+      "loss": 5.6785,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 5.4539,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 5.1883,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.942307692307692e-06,
+      "loss": 4.9794,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.0384615384615385e-06,
+      "loss": 4.7848,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.134615384615385e-06,
+      "loss": 4.6471,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.230769230769231e-06,
+      "loss": 4.3799,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.326923076923077e-06,
+      "loss": 4.2038,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.423076923076924e-06,
+      "loss": 4.0222,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.51923076923077e-06,
+      "loss": 3.8671,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 3.7002,
+      "step": 48
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.711538461538462e-06,
+      "loss": 3.5262,
+      "step": 49
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.807692307692308e-06,
+      "loss": 3.412,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.903846153846154e-06,
+      "loss": 3.2542,
+      "step": 51
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5e-06,
+      "loss": 3.0786,
+      "step": 52
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.096153846153846e-06,
+      "loss": 2.9713,
+      "step": 53
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.192307692307693e-06,
+      "loss": 3.0104,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.288461538461539e-06,
+      "loss": 2.8988,
+      "step": 55
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.384615384615385e-06,
+      "loss": 2.7661,
+      "step": 56
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.480769230769232e-06,
+      "loss": 2.6857,
+      "step": 57
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.576923076923077e-06,
+      "loss": 2.507,
+      "step": 58
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.6730769230769235e-06,
+      "loss": 2.4025,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.769230769230769e-06,
+      "loss": 2.388,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.865384615384616e-06,
+      "loss": 2.2386,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.961538461538462e-06,
+      "loss": 2.1366,
+      "step": 62
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.057692307692308e-06,
+      "loss": 2.0364,
+      "step": 63
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 1.9973,
+      "step": 64
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.25e-06,
+      "loss": 1.8432,
+      "step": 65
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.3461538461538466e-06,
+      "loss": 1.8442,
+      "step": 66
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.442307692307693e-06,
+      "loss": 1.6175,
+      "step": 67
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.538461538461539e-06,
+      "loss": 1.6448,
+      "step": 68
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.6346153846153846e-06,
+      "loss": 1.57,
+      "step": 69
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.730769230769232e-06,
+      "loss": 1.4747,
+      "step": 70
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.826923076923078e-06,
+      "loss": 1.4357,
+      "step": 71
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 1.3034,
+      "step": 72
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.01923076923077e-06,
+      "loss": 1.281,
+      "step": 73
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.115384615384616e-06,
+      "loss": 1.2369,
+      "step": 74
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.211538461538462e-06,
+      "loss": 1.134,
+      "step": 75
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.307692307692308e-06,
+      "loss": 1.066,
+      "step": 76
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.403846153846155e-06,
+      "loss": 1.0677,
+      "step": 77
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 1.0208,
+      "step": 78
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.5961538461538465e-06,
+      "loss": 0.9828,
+      "step": 79
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 0.9954,
+      "step": 80
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.78846153846154e-06,
+      "loss": 0.9651,
+      "step": 81
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.884615384615384e-06,
+      "loss": 0.9687,
+      "step": 82
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.980769230769232e-06,
+      "loss": 0.9537,
+      "step": 83
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.076923076923077e-06,
+      "loss": 0.9434,
+      "step": 84
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.173076923076923e-06,
+      "loss": 0.9292,
+      "step": 85
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.26923076923077e-06,
+      "loss": 0.9182,
+      "step": 86
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.365384615384616e-06,
+      "loss": 0.9262,
+      "step": 87
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 0.9258,
+      "step": 88
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.557692307692308e-06,
+      "loss": 0.9312,
+      "step": 89
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.653846153846155e-06,
+      "loss": 0.9263,
+      "step": 90
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.750000000000001e-06,
+      "loss": 0.9246,
+      "step": 91
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.846153846153847e-06,
+      "loss": 0.9225,
+      "step": 92
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.942307692307693e-06,
+      "loss": 0.9287,
+      "step": 93
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.03846153846154e-06,
+      "loss": 0.9259,
+      "step": 94
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.134615384615384e-06,
+      "loss": 0.9237,
+      "step": 95
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 0.9028,
+      "step": 96
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.326923076923079e-06,
+      "loss": 0.9113,
+      "step": 97
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.423076923076923e-06,
+      "loss": 0.9251,
+      "step": 98
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.51923076923077e-06,
+      "loss": 0.9501,
+      "step": 99
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.615384615384616e-06,
+      "loss": 0.9349,
+      "step": 100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.711538461538462e-06,
+      "loss": 0.8954,
+      "step": 101
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.807692307692308e-06,
+      "loss": 0.9089,
+      "step": 102
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.903846153846155e-06,
+      "loss": 0.9086,
+      "step": 103
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1e-05,
+      "loss": 0.9099,
+      "step": 104
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0096153846153847e-05,
+      "loss": 0.9033,
+      "step": 105
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0192307692307692e-05,
+      "loss": 0.9012,
+      "step": 106
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0288461538461538e-05,
+      "loss": 0.9096,
+      "step": 107
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0384615384615386e-05,
+      "loss": 0.8801,
+      "step": 108
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0480769230769232e-05,
+      "loss": 0.8938,
+      "step": 109
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0576923076923078e-05,
+      "loss": 0.9299,
+      "step": 110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0673076923076923e-05,
+      "loss": 0.9113,
+      "step": 111
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 0.915,
+      "step": 112
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0865384615384616e-05,
+      "loss": 0.8925,
+      "step": 113
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0961538461538464e-05,
+      "loss": 0.898,
+      "step": 114
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.105769230769231e-05,
+      "loss": 0.8924,
+      "step": 115
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1153846153846154e-05,
+      "loss": 0.8869,
+      "step": 116
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.125e-05,
+      "loss": 0.89,
+      "step": 117
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1346153846153847e-05,
+      "loss": 0.8889,
+      "step": 118
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1442307692307693e-05,
+      "loss": 0.8836,
+      "step": 119
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1538461538461538e-05,
+      "loss": 0.8743,
+      "step": 120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1634615384615386e-05,
+      "loss": 0.8779,
+      "step": 121
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1730769230769232e-05,
+      "loss": 0.8769,
+      "step": 122
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1826923076923078e-05,
+      "loss": 0.8993,
+      "step": 123
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1923076923076925e-05,
+      "loss": 0.9147,
+      "step": 124
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.201923076923077e-05,
+      "loss": 0.8547,
+      "step": 125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2115384615384615e-05,
+      "loss": 0.8773,
+      "step": 126
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2211538461538463e-05,
+      "loss": 0.8768,
+      "step": 127
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 0.8583,
+      "step": 128
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2403846153846156e-05,
+      "loss": 0.8727,
+      "step": 129
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.25e-05,
+      "loss": 0.8598,
+      "step": 130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2596153846153847e-05,
+      "loss": 0.855,
+      "step": 131
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2692307692307693e-05,
+      "loss": 0.8395,
+      "step": 132
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2788461538461538e-05,
+      "loss": 0.8703,
+      "step": 133
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2884615384615386e-05,
+      "loss": 0.8747,
+      "step": 134
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2980769230769232e-05,
+      "loss": 0.8515,
+      "step": 135
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3076923076923078e-05,
+      "loss": 0.8789,
+      "step": 136
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3173076923076925e-05,
+      "loss": 0.841,
+      "step": 137
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3269230769230769e-05,
+      "loss": 0.8273,
+      "step": 138
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3365384615384615e-05,
+      "loss": 0.8359,
+      "step": 139
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3461538461538463e-05,
+      "loss": 0.848,
+      "step": 140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.355769230769231e-05,
+      "loss": 0.8299,
+      "step": 141
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3653846153846156e-05,
+      "loss": 0.8235,
+      "step": 142
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.375e-05,
+      "loss": 0.8194,
+      "step": 143
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 0.8095,
+      "step": 144
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3942307692307693e-05,
+      "loss": 0.7942,
+      "step": 145
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.403846153846154e-05,
+      "loss": 0.8247,
+      "step": 146
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4134615384615387e-05,
+      "loss": 0.8238,
+      "step": 147
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4230769230769232e-05,
+      "loss": 0.8113,
+      "step": 148
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4326923076923078e-05,
+      "loss": 0.7945,
+      "step": 149
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4423076923076924e-05,
+      "loss": 0.7767,
+      "step": 150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.451923076923077e-05,
+      "loss": 0.7734,
+      "step": 151
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4615384615384615e-05,
+      "loss": 0.7668,
+      "step": 152
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4711538461538463e-05,
+      "loss": 0.7695,
+      "step": 153
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.480769230769231e-05,
+      "loss": 0.766,
+      "step": 154
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4903846153846156e-05,
+      "loss": 0.7395,
+      "step": 155
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.7674,
+      "step": 156
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5096153846153847e-05,
+      "loss": 0.7365,
+      "step": 157
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5192307692307693e-05,
+      "loss": 0.7404,
+      "step": 158
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.528846153846154e-05,
+      "loss": 0.7336,
+      "step": 159
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 0.7088,
+      "step": 160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5480769230769232e-05,
+      "loss": 0.6949,
+      "step": 161
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.557692307692308e-05,
+      "loss": 0.6934,
+      "step": 162
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5673076923076924e-05,
+      "loss": 0.6857,
+      "step": 163
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.576923076923077e-05,
+      "loss": 0.6888,
+      "step": 164
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5865384615384617e-05,
+      "loss": 0.672,
+      "step": 165
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5961538461538465e-05,
+      "loss": 0.6663,
+      "step": 166
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.605769230769231e-05,
+      "loss": 0.665,
+      "step": 167
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6153846153846154e-05,
+      "loss": 0.6607,
+      "step": 168
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6250000000000002e-05,
+      "loss": 0.6522,
+      "step": 169
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6346153846153847e-05,
+      "loss": 0.6278,
+      "step": 170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6442307692307695e-05,
+      "loss": 0.6374,
+      "step": 171
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.653846153846154e-05,
+      "loss": 0.6131,
+      "step": 172
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6634615384615387e-05,
+      "loss": 0.6025,
+      "step": 173
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.673076923076923e-05,
+      "loss": 0.5916,
+      "step": 174
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.682692307692308e-05,
+      "loss": 0.5753,
+      "step": 175
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 0.6018,
+      "step": 176
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.701923076923077e-05,
+      "loss": 0.5575,
+      "step": 177
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7115384615384617e-05,
+      "loss": 0.5214,
+      "step": 178
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7211538461538465e-05,
+      "loss": 0.517,
+      "step": 179
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.730769230769231e-05,
+      "loss": 0.5161,
+      "step": 180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7403846153846157e-05,
+      "loss": 0.5195,
+      "step": 181
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 0.5139,
+      "step": 182
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7596153846153846e-05,
+      "loss": 0.5009,
+      "step": 183
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7692307692307694e-05,
+      "loss": 0.486,
+      "step": 184
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.778846153846154e-05,
+      "loss": 0.502,
+      "step": 185
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7884615384615387e-05,
+      "loss": 0.5109,
+      "step": 186
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.798076923076923e-05,
+      "loss": 0.5011,
+      "step": 187
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.807692307692308e-05,
+      "loss": 0.5013,
+      "step": 188
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8173076923076924e-05,
+      "loss": 0.471,
+      "step": 189
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.826923076923077e-05,
+      "loss": 0.4812,
+      "step": 190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8365384615384617e-05,
+      "loss": 0.4746,
+      "step": 191
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 0.4711,
+      "step": 192
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.855769230769231e-05,
+      "loss": 0.472,
+      "step": 193
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8653846153846157e-05,
+      "loss": 0.4598,
+      "step": 194
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 0.4725,
+      "step": 195
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8846153846153846e-05,
+      "loss": 0.4623,
+      "step": 196
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8942307692307694e-05,
+      "loss": 0.4627,
+      "step": 197
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.903846153846154e-05,
+      "loss": 0.4633,
+      "step": 198
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9134615384615387e-05,
+      "loss": 0.4559,
+      "step": 199
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 0.4726,
+      "step": 200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.932692307692308e-05,
+      "loss": 0.4904,
+      "step": 201
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9423076923076924e-05,
+      "loss": 0.4452,
+      "step": 202
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9519230769230772e-05,
+      "loss": 0.4531,
+      "step": 203
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9615384615384617e-05,
+      "loss": 0.4743,
+      "step": 204
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9711538461538465e-05,
+      "loss": 0.4692,
+      "step": 205
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.980769230769231e-05,
+      "loss": 0.4649,
+      "step": 206
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9903846153846157e-05,
+      "loss": 0.4509,
+      "step": 207
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.4945,
+      "step": 208
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999999890134612e-05,
+      "loss": 0.5003,
+      "step": 209
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999999560538471e-05,
+      "loss": 0.5066,
+      "step": 210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999990112116507e-05,
+      "loss": 0.4825,
+      "step": 211
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999982421542713e-05,
+      "loss": 0.462,
+      "step": 212
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999997253366501e-05,
+      "loss": 0.4585,
+      "step": 213
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999960448485584e-05,
+      "loss": 0.4708,
+      "step": 214
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999946166007077e-05,
+      "loss": 0.4545,
+      "step": 215
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999929686232637e-05,
+      "loss": 0.4493,
+      "step": 216
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999991100916588e-05,
+      "loss": 0.4595,
+      "step": 217
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999989013481092e-05,
+      "loss": 0.4553,
+      "step": 218
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999867063172328e-05,
+      "loss": 0.4594,
+      "step": 219
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999841794255185e-05,
+      "loss": 0.4434,
+      "step": 220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999981432806504e-05,
+      "loss": 0.4429,
+      "step": 221
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999784664607923e-05,
+      "loss": 0.4748,
+      "step": 222
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999975280389036e-05,
+      "loss": 0.4904,
+      "step": 223
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999971874591935e-05,
+      "loss": 0.4705,
+      "step": 224
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999682490702377e-05,
+      "loss": 0.441,
+      "step": 225
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999644038247405e-05,
+      "loss": 0.4563,
+      "step": 226
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999960338856288e-05,
+      "loss": 0.4647,
+      "step": 227
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999956054165774e-05,
+      "loss": 0.4878,
+      "step": 228
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99995154975414e-05,
+      "loss": 0.5034,
+      "step": 229
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999946825622375e-05,
+      "loss": 0.4934,
+      "step": 230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999418817715178e-05,
+      "loss": 0.4692,
+      "step": 231
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999367182026547e-05,
+      "loss": 0.4534,
+      "step": 232
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99993133491692e-05,
+      "loss": 0.4656,
+      "step": 233
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999925731915496e-05,
+      "loss": 0.5152,
+      "step": 234
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999919909199615e-05,
+      "loss": 0.4769,
+      "step": 235
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999138667705563e-05,
+      "loss": 0.4748,
+      "step": 236
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999076046296467e-05,
+      "loss": 0.4875,
+      "step": 237
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999901122778263e-05,
+      "loss": 0.4378,
+      "step": 238
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999894421217829e-05,
+      "loss": 0.4514,
+      "step": 239
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998874999498174e-05,
+      "loss": 0.466,
+      "step": 240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999880358975749e-05,
+      "loss": 0.4674,
+      "step": 241
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999872998297193e-05,
+      "loss": 0.46,
+      "step": 242
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999865417915767e-05,
+      "loss": 0.458,
+      "step": 243
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999857617833136e-05,
+      "loss": 0.4707,
+      "step": 244
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998495980510145e-05,
+      "loss": 0.4408,
+      "step": 245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998413585711642e-05,
+      "loss": 0.4355,
+      "step": 246
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998328993953957e-05,
+      "loss": 0.4719,
+      "step": 247
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999824220525568e-05,
+      "loss": 0.4451,
+      "step": 248
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999815321963588e-05,
+      "loss": 0.4574,
+      "step": 249
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999806203711411e-05,
+      "loss": 0.4586,
+      "step": 250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997968657710404e-05,
+      "loss": 0.4824,
+      "step": 251
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997873081445284e-05,
+      "loss": 0.4756,
+      "step": 252
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997775308339747e-05,
+      "loss": 0.4714,
+      "step": 253
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999767533841528e-05,
+      "loss": 0.4629,
+      "step": 254
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997573171693845e-05,
+      "loss": 0.4676,
+      "step": 255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999746880819789e-05,
+      "loss": 0.461,
+      "step": 256
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999736224795036e-05,
+      "loss": 0.4567,
+      "step": 257
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997253490974656e-05,
+      "loss": 0.4491,
+      "step": 258
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999714253729468e-05,
+      "loss": 0.4687,
+      "step": 259
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9997029386934812e-05,
+      "loss": 0.4674,
+      "step": 260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996914039919914e-05,
+      "loss": 0.4661,
+      "step": 261
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999679649627533e-05,
+      "loss": 0.4587,
+      "step": 262
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996676756026893e-05,
+      "loss": 0.4651,
+      "step": 263
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996554819200907e-05,
+      "loss": 0.4686,
+      "step": 264
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999643068582417e-05,
+      "loss": 0.4663,
+      "step": 265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996304355923954e-05,
+      "loss": 0.4491,
+      "step": 266
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999617582952802e-05,
+      "loss": 0.4481,
+      "step": 267
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999604510666461e-05,
+      "loss": 0.4852,
+      "step": 268
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.999591218736245e-05,
+      "loss": 0.4948,
+      "step": 269
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995777071650737e-05,
+      "loss": 0.4897,
+      "step": 270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995639759559173e-05,
+      "loss": 0.4757,
+      "step": 271
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999550025111792e-05,
+      "loss": 0.4604,
+      "step": 272
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995358546357634e-05,
+      "loss": 0.4566,
+      "step": 273
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995214645309455e-05,
+      "loss": 0.4664,
+      "step": 274
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995068548005e-05,
+      "loss": 0.4807,
+      "step": 275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994920254476374e-05,
+      "loss": 0.4941,
+      "step": 276
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994769764756157e-05,
+      "loss": 0.4621,
+      "step": 277
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994617078877422e-05,
+      "loss": 0.4645,
+      "step": 278
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999446219687371e-05,
+      "loss": 0.4511,
+      "step": 279
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994305118779064e-05,
+      "loss": 0.4849,
+      "step": 280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999414584462799e-05,
+      "loss": 0.4676,
+      "step": 281
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993984374455493e-05,
+      "loss": 0.4572,
+      "step": 282
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9993820708297048e-05,
+      "loss": 0.4439,
+      "step": 283
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993654846188617e-05,
+      "loss": 0.4535,
+      "step": 284
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993486788166647e-05,
+      "loss": 0.4421,
+      "step": 285
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993316534268065e-05,
+      "loss": 0.4497,
+      "step": 286
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993144084530284e-05,
+      "loss": 0.461,
+      "step": 287
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999296943899119e-05,
+      "loss": 0.4631,
+      "step": 288
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992792597689162e-05,
+      "loss": 0.4751,
+      "step": 289
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992613560663062e-05,
+      "loss": 0.4504,
+      "step": 290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992432327952224e-05,
+      "loss": 0.4568,
+      "step": 291
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992248899596467e-05,
+      "loss": 0.4588,
+      "step": 292
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99920632756361e-05,
+      "loss": 0.4765,
+      "step": 293
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991875456111917e-05,
+      "loss": 0.4612,
+      "step": 294
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999168544106518e-05,
+      "loss": 0.4532,
+      "step": 295
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999149323053764e-05,
+      "loss": 0.4565,
+      "step": 296
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999129882457153e-05,
+      "loss": 0.5026,
+      "step": 297
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9991102223209578e-05,
+      "loss": 0.4801,
+      "step": 298
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990903426494977e-05,
+      "loss": 0.47,
+      "step": 299
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990702434471405e-05,
+      "loss": 0.462,
+      "step": 300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999049924718303e-05,
+      "loss": 0.4556,
+      "step": 301
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990293864674494e-05,
+      "loss": 0.4766,
+      "step": 302
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9990086286990936e-05,
+      "loss": 0.4536,
+      "step": 303
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989876514177958e-05,
+      "loss": 0.4766,
+      "step": 304
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998966454628166e-05,
+      "loss": 0.4519,
+      "step": 305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989450383348608e-05,
+      "loss": 0.4442,
+      "step": 306
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998923402542587e-05,
+      "loss": 0.4546,
+      "step": 307
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989015472560985e-05,
+      "loss": 0.4695,
+      "step": 308
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998879472480197e-05,
+      "loss": 0.4509,
+      "step": 309
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998857178219734e-05,
+      "loss": 0.462,
+      "step": 310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998834664479607e-05,
+      "loss": 0.4678,
+      "step": 311
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9988119312647638e-05,
+      "loss": 0.4657,
+      "step": 312
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987889785801995e-05,
+      "loss": 0.4535,
+      "step": 313
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998765806430957e-05,
+      "loss": 0.4683,
+      "step": 314
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998742414822129e-05,
+      "loss": 0.4449,
+      "step": 315
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987188037588542e-05,
+      "loss": 0.4462,
+      "step": 316
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998694973246321e-05,
+      "loss": 0.466,
+      "step": 317
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998670923289766e-05,
+      "loss": 0.4333,
+      "step": 318
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986466538944736e-05,
+      "loss": 0.4768,
+      "step": 319
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986221650657767e-05,
+      "loss": 0.4766,
+      "step": 320
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985974568090558e-05,
+      "loss": 0.4543,
+      "step": 321
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985725291297405e-05,
+      "loss": 0.4576,
+      "step": 322
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998547382033308e-05,
+      "loss": 0.4603,
+      "step": 323
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998522015525284e-05,
+      "loss": 0.455,
+      "step": 324
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984964296112416e-05,
+      "loss": 0.4777,
+      "step": 325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984706242968038e-05,
+      "loss": 0.4615,
+      "step": 326
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984445995876404e-05,
+      "loss": 0.4595,
+      "step": 327
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984183554894697e-05,
+      "loss": 0.4553,
+      "step": 328
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983918920080587e-05,
+      "loss": 0.4686,
+      "step": 329
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983652091492218e-05,
+      "loss": 0.4608,
+      "step": 330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983383069188225e-05,
+      "loss": 0.4705,
+      "step": 331
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983111853227718e-05,
+      "loss": 0.4527,
+      "step": 332
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982838443670293e-05,
+      "loss": 0.4533,
+      "step": 333
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982562840576022e-05,
+      "loss": 0.451,
+      "step": 334
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982285044005468e-05,
+      "loss": 0.4596,
+      "step": 335
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998200505401967e-05,
+      "loss": 0.4391,
+      "step": 336
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.998172287068015e-05,
+      "loss": 0.4574,
+      "step": 337
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981438494048916e-05,
+      "loss": 0.4726,
+      "step": 338
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998115192418845e-05,
+      "loss": 0.4664,
+      "step": 339
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998086316116172e-05,
+      "loss": 0.4665,
+      "step": 340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980572205032175e-05,
+      "loss": 0.4734,
+      "step": 341
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980279055863752e-05,
+      "loss": 0.4602,
+      "step": 342
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979983713720866e-05,
+      "loss": 0.4557,
+      "step": 343
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979686178668406e-05,
+      "loss": 0.481,
+      "step": 344
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979386450771753e-05,
+      "loss": 0.4714,
+      "step": 345
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9979084530096765e-05,
+      "loss": 0.4603,
+      "step": 346
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978780416709786e-05,
+      "loss": 0.4477,
+      "step": 347
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978474110677635e-05,
+      "loss": 0.457,
+      "step": 348
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9978165612067623e-05,
+      "loss": 0.4587,
+      "step": 349
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977854920947534e-05,
+      "loss": 0.4649,
+      "step": 350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997754203738563e-05,
+      "loss": 0.4618,
+      "step": 351
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9977226961450673e-05,
+      "loss": 0.4895,
+      "step": 352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976909693211886e-05,
+      "loss": 0.4706,
+      "step": 353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976590232738982e-05,
+      "loss": 0.4598,
+      "step": 354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9976268580102165e-05,
+      "loss": 0.458,
+      "step": 355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975944735372104e-05,
+      "loss": 0.4565,
+      "step": 356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997561869861996e-05,
+      "loss": 0.4678,
+      "step": 357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975290469917373e-05,
+      "loss": 0.4822,
+      "step": 358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974960049336466e-05,
+      "loss": 0.4533,
+      "step": 359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997462743694984e-05,
+      "loss": 0.4619,
+      "step": 360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9974292632830584e-05,
+      "loss": 0.447,
+      "step": 361
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973955637052262e-05,
+      "loss": 0.4576,
+      "step": 362
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973616449688923e-05,
+      "loss": 0.4559,
+      "step": 363
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973275070815098e-05,
+      "loss": 0.465,
+      "step": 364
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972931500505797e-05,
+      "loss": 0.4759,
+      "step": 365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972585738836514e-05,
+      "loss": 0.4677,
+      "step": 366
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9972237785883223e-05,
+      "loss": 0.4757,
+      "step": 367
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.997188764172238e-05,
+      "loss": 0.4481,
+      "step": 368
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971535306430922e-05,
+      "loss": 0.4417,
+      "step": 369
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971180780086267e-05,
+      "loss": 0.4725,
+      "step": 370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970824062766318e-05,
+      "loss": 0.4787,
+      "step": 371
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970465154549458e-05,
+      "loss": 0.4474,
+      "step": 372
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970104055514544e-05,
+      "loss": 0.473,
+      "step": 373
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969740765740924e-05,
+      "loss": 0.4677,
+      "step": 374
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996937528530843e-05,
+      "loss": 0.4552,
+      "step": 375
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969007614297354e-05,
+      "loss": 0.4499,
+      "step": 376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.99686377527885e-05,
+      "loss": 0.4643,
+      "step": 377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996826570086313e-05,
+      "loss": 0.4638,
+      "step": 378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9967891458603e-05,
+      "loss": 0.4635,
+      "step": 379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.996751502609034e-05,
+      "loss": 0.4561,
+      "step": 380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996713640340786e-05,
+      "loss": 0.4579,
+      "step": 381
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996675559063876e-05,
+      "loss": 0.4611,
+      "step": 382
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966372587866716e-05,
+      "loss": 0.4648,
+      "step": 383
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965987395175882e-05,
+      "loss": 0.4646,
+      "step": 384
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965600012650902e-05,
+      "loss": 0.4749,
+      "step": 385
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965210440376892e-05,
+      "loss": 0.4543,
+      "step": 386
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964818678439453e-05,
+      "loss": 0.4582,
+      "step": 387
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964424726924672e-05,
+      "loss": 0.4518,
+      "step": 388
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964028585919104e-05,
+      "loss": 0.4482,
+      "step": 389
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.99636302555098e-05,
+      "loss": 0.4583,
+      "step": 390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9963229735784285e-05,
+      "loss": 0.4749,
+      "step": 391
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.996282702683056e-05,
+      "loss": 0.4595,
+      "step": 392
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962422128737116e-05,
+      "loss": 0.4682,
+      "step": 393
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9962015041592927e-05,
+      "loss": 0.4573,
+      "step": 394
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961605765487435e-05,
+      "loss": 0.467,
+      "step": 395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961194300510574e-05,
+      "loss": 0.4548,
+      "step": 396
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960780646752755e-05,
+      "loss": 0.4553,
+      "step": 397
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9960364804304863e-05,
+      "loss": 0.4608,
+      "step": 398
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959946773258286e-05,
+      "loss": 0.456,
+      "step": 399
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995952655370487e-05,
+      "loss": 0.4469,
+      "step": 400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9959104145736946e-05,
+      "loss": 0.4484,
+      "step": 401
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995867954944734e-05,
+      "loss": 0.4468,
+      "step": 402
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995825276492934e-05,
+      "loss": 0.4478,
+      "step": 403
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957823792276733e-05,
+      "loss": 0.4493,
+      "step": 404
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9957392631583766e-05,
+      "loss": 0.4538,
+      "step": 405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956959282945186e-05,
+      "loss": 0.4397,
+      "step": 406
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9956523746456212e-05,
+      "loss": 0.4564,
+      "step": 407
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9956086022212547e-05,
+      "loss": 0.4489,
+      "step": 408
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955646110310366e-05,
+      "loss": 0.4459,
+      "step": 409
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955204010846334e-05,
+      "loss": 0.4534,
+      "step": 410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954759723917595e-05,
+      "loss": 0.4581,
+      "step": 411
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9954313249621774e-05,
+      "loss": 0.4567,
+      "step": 412
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995386458805697e-05,
+      "loss": 0.4525,
+      "step": 413
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9953413739321778e-05,
+      "loss": 0.4492,
+      "step": 414
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995296070351525e-05,
+      "loss": 0.455,
+      "step": 415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995250548073694e-05,
+      "loss": 0.4605,
+      "step": 416
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995204807108687e-05,
+      "loss": 0.4577,
+      "step": 417
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951588474665552e-05,
+      "loss": 0.4598,
+      "step": 418
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9951126691573976e-05,
+      "loss": 0.4465,
+      "step": 419
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.99506627219136e-05,
+      "loss": 0.4638,
+      "step": 420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9950196565786374e-05,
+      "loss": 0.4599,
+      "step": 421
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9949728223294734e-05,
+      "loss": 0.4694,
+      "step": 422
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9949257694541587e-05,
+      "loss": 0.4649,
+      "step": 423
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994878497963032e-05,
+      "loss": 0.4616,
+      "step": 424
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9948310078664804e-05,
+      "loss": 0.4673,
+      "step": 425
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994783299174939e-05,
+      "loss": 0.4708,
+      "step": 426
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9947353718988907e-05,
+      "loss": 0.4581,
+      "step": 427
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946872260488673e-05,
+      "loss": 0.4628,
+      "step": 428
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946388616354468e-05,
+      "loss": 0.4741,
+      "step": 429
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994590278669257e-05,
+      "loss": 0.4593,
+      "step": 430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994541477160973e-05,
+      "loss": 0.4579,
+      "step": 431
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994492457121318e-05,
+      "loss": 0.4586,
+      "step": 432
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994443218561063e-05,
+      "loss": 0.4522,
+      "step": 433
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994393761491028e-05,
+      "loss": 0.4618,
+      "step": 434
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994344085922079e-05,
+      "loss": 0.4694,
+      "step": 435
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994294191865132e-05,
+      "loss": 0.4549,
+      "step": 436
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9942440793311502e-05,
+      "loss": 0.4535,
+      "step": 437
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941937483311454e-05,
+      "loss": 0.459,
+      "step": 438
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941431988761756e-05,
+      "loss": 0.4705,
+      "step": 439
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994092430977349e-05,
+      "loss": 0.4484,
+      "step": 440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940414446458207e-05,
+      "loss": 0.4657,
+      "step": 441
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9939902398927938e-05,
+      "loss": 0.4623,
+      "step": 442
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.99393881672952e-05,
+      "loss": 0.4672,
+      "step": 443
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938871751672976e-05,
+      "loss": 0.4587,
+      "step": 444
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938353152174748e-05,
+      "loss": 0.4395,
+      "step": 445
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9937832368914465e-05,
+      "loss": 0.4698,
+      "step": 446
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993730940200656e-05,
+      "loss": 0.4459,
+      "step": 447
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993678425156594e-05,
+      "loss": 0.4641,
+      "step": 448
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9936256917708005e-05,
+      "loss": 0.4653,
+      "step": 449
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993572740054862e-05,
+      "loss": 0.4537,
+      "step": 450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9935195700204138e-05,
+      "loss": 0.4485,
+      "step": 451
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993466181679139e-05,
+      "loss": 0.4582,
+      "step": 452
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993412575042769e-05,
+      "loss": 0.447,
+      "step": 453
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993358750123082e-05,
+      "loss": 0.445,
+      "step": 454
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9933047069319058e-05,
+      "loss": 0.4727,
+      "step": 455
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993250445481115e-05,
+      "loss": 0.4645,
+      "step": 456
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993195965782633e-05,
+      "loss": 0.4455,
+      "step": 457
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9931412678484294e-05,
+      "loss": 0.4569,
+      "step": 458
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9930863516905242e-05,
+      "loss": 0.4425,
+      "step": 459
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993031217320984e-05,
+      "loss": 0.4632,
+      "step": 460
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.992975864751923e-05,
+      "loss": 0.4522,
+      "step": 461
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929202939955044e-05,
+      "loss": 0.4578,
+      "step": 462
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9928645050639388e-05,
+      "loss": 0.4531,
+      "step": 463
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9928084979694844e-05,
+      "loss": 0.4477,
+      "step": 464
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992752272724448e-05,
+      "loss": 0.4567,
+      "step": 465
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9926958293411833e-05,
+      "loss": 0.4545,
+      "step": 466
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9926391678320937e-05,
+      "loss": 0.4605,
+      "step": 467
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9925822882096288e-05,
+      "loss": 0.4529,
+      "step": 468
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992525190486287e-05,
+      "loss": 0.4547,
+      "step": 469
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924678746746146e-05,
+      "loss": 0.4588,
+      "step": 470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924103407872052e-05,
+      "loss": 0.4557,
+      "step": 471
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992352588836701e-05,
+      "loss": 0.4654,
+      "step": 472
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9922946188357923e-05,
+      "loss": 0.4584,
+      "step": 473
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9922364307972163e-05,
+      "loss": 0.4675,
+      "step": 474
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9921780247337586e-05,
+      "loss": 0.4679,
+      "step": 475
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9921194006582532e-05,
+      "loss": 0.4566,
+      "step": 476
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9920605585835817e-05,
+      "loss": 0.4486,
+      "step": 477
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9920014985226736e-05,
+      "loss": 0.4574,
+      "step": 478
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919422204885054e-05,
+      "loss": 0.4577,
+      "step": 479
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991882724494103e-05,
+      "loss": 0.4505,
+      "step": 480
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9918230105525395e-05,
+      "loss": 0.4617,
+      "step": 481
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991763078676936e-05,
+      "loss": 0.4411,
+      "step": 482
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9917029288804606e-05,
+      "loss": 0.4615,
+      "step": 483
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991642561176331e-05,
+      "loss": 0.4453,
+      "step": 484
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915819755778112e-05,
+      "loss": 0.4479,
+      "step": 485
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9915211720982138e-05,
+      "loss": 0.4506,
+      "step": 486
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9914601507508996e-05,
+      "loss": 0.4653,
+      "step": 487
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9913989115492765e-05,
+      "loss": 0.458,
+      "step": 488
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.991337454506801e-05,
+      "loss": 0.4549,
+      "step": 489
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9912757796369768e-05,
+      "loss": 0.4576,
+      "step": 490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9912138869533553e-05,
+      "loss": 0.4593,
+      "step": 491
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9911517764695375e-05,
+      "loss": 0.4471,
+      "step": 492
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9910894481991698e-05,
+      "loss": 0.4431,
+      "step": 493
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9910269021559482e-05,
+      "loss": 0.4425,
+      "step": 494
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9909641383536158e-05,
+      "loss": 0.4449,
+      "step": 495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990901156805964e-05,
+      "loss": 0.463,
+      "step": 496
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990837957526831e-05,
+      "loss": 0.4535,
+      "step": 497
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990774540530105e-05,
+      "loss": 0.4483,
+      "step": 498
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9907109058297198e-05,
+      "loss": 0.4444,
+      "step": 499
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9906470534396574e-05,
+      "loss": 0.4466,
+      "step": 500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990582983373949e-05,
+      "loss": 0.4485,
+      "step": 501
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905186956466726e-05,
+      "loss": 0.4532,
+      "step": 502
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.990454190271954e-05,
+      "loss": 0.456,
+      "step": 503
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9903894672639673e-05,
+      "loss": 0.4534,
+      "step": 504
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9903245266369337e-05,
+      "loss": 0.4428,
+      "step": 505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9902593684051227e-05,
+      "loss": 0.4623,
+      "step": 506
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901939925828523e-05,
+      "loss": 0.4521,
+      "step": 507
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9901283991844863e-05,
+      "loss": 0.4576,
+      "step": 508
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990062588224439e-05,
+      "loss": 0.4732,
+      "step": 509
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9899965597171703e-05,
+      "loss": 0.4611,
+      "step": 510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9899303136771883e-05,
+      "loss": 0.4442,
+      "step": 511
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9898638501190503e-05,
+      "loss": 0.4496,
+      "step": 512
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9897971690573597e-05,
+      "loss": 0.4492,
+      "step": 513
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9897302705067686e-05,
+      "loss": 0.4621,
+      "step": 514
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9896631544819767e-05,
+      "loss": 0.4562,
+      "step": 515
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9895958209977312e-05,
+      "loss": 0.4704,
+      "step": 516
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.989528270068828e-05,
+      "loss": 0.4578,
+      "step": 517
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9894605017101095e-05,
+      "loss": 0.4477,
+      "step": 518
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9893925159364663e-05,
+      "loss": 0.4516,
+      "step": 519
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9893243127628375e-05,
+      "loss": 0.4712,
+      "step": 520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9892558922042094e-05,
+      "loss": 0.4584,
+      "step": 521
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989187254275616e-05,
+      "loss": 0.4436,
+      "step": 522
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989118398992139e-05,
+      "loss": 0.4637,
+      "step": 523
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9890493263689078e-05,
+      "loss": 0.4803,
+      "step": 524
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988980036421101e-05,
+      "loss": 0.455,
+      "step": 525
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988910529163942e-05,
+      "loss": 0.4618,
+      "step": 526
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988840804612705e-05,
+      "loss": 0.4647,
+      "step": 527
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.98877086278271e-05,
+      "loss": 0.4599,
+      "step": 528
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9887007036893252e-05,
+      "loss": 0.4595,
+      "step": 529
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9886303273479676e-05,
+      "loss": 0.4627,
+      "step": 530
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9885597337741005e-05,
+      "loss": 0.4449,
+      "step": 531
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9884889229832354e-05,
+      "loss": 0.458,
+      "step": 532
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9884178949909315e-05,
+      "loss": 0.4673,
+      "step": 533
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9883466498127964e-05,
+      "loss": 0.4645,
+      "step": 534
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9882751874644844e-05,
+      "loss": 0.4724,
+      "step": 535
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9882035079616977e-05,
+      "loss": 0.4563,
+      "step": 536
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9881316113201873e-05,
+      "loss": 0.4448,
+      "step": 537
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9880594975557506e-05,
+      "loss": 0.4507,
+      "step": 538
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987987166684233e-05,
+      "loss": 0.4699,
+      "step": 539
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9879146187215284e-05,
+      "loss": 0.4779,
+      "step": 540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9878418536835772e-05,
+      "loss": 0.4679,
+      "step": 541
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987768871586369e-05,
+      "loss": 0.4672,
+      "step": 542
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.987695672445939e-05,
+      "loss": 0.4637,
+      "step": 543
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9876222562783725e-05,
+      "loss": 0.4511,
+      "step": 544
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9875486230998005e-05,
+      "loss": 0.4519,
+      "step": 545
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9874747729264027e-05,
+      "loss": 0.4601,
+      "step": 546
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987400705774406e-05,
+      "loss": 0.4731,
+      "step": 547
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987326421660086e-05,
+      "loss": 0.4608,
+      "step": 548
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9872519205997646e-05,
+      "loss": 0.4545,
+      "step": 549
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987177202609812e-05,
+      "loss": 0.4637,
+      "step": 550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9871022677066466e-05,
+      "loss": 0.4602,
+      "step": 551
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987027115906733e-05,
+      "loss": 0.4438,
+      "step": 552
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9869517472265845e-05,
+      "loss": 0.4697,
+      "step": 553
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986876161682763e-05,
+      "loss": 0.4796,
+      "step": 554
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9868003592918755e-05,
+      "loss": 0.476,
+      "step": 555
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986724340070579e-05,
+      "loss": 0.4711,
+      "step": 556
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866481040355772e-05,
+      "loss": 0.4548,
+      "step": 557
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9865716512036217e-05,
+      "loss": 0.4543,
+      "step": 558
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9864949815915106e-05,
+      "loss": 0.4421,
+      "step": 559
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986418095216091e-05,
+      "loss": 0.4526,
+      "step": 560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9863409920942584e-05,
+      "loss": 0.4783,
+      "step": 561
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986263672242953e-05,
+      "loss": 0.4693,
+      "step": 562
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986186135679165e-05,
+      "loss": 0.4494,
+      "step": 563
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986108382419932e-05,
+      "loss": 0.4564,
+      "step": 564
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.986030412482338e-05,
+      "loss": 0.4622,
+      "step": 565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9859522258835165e-05,
+      "loss": 0.4629,
+      "step": 566
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9858738226406463e-05,
+      "loss": 0.4694,
+      "step": 567
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985795202770956e-05,
+      "loss": 0.4498,
+      "step": 568
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.98571636629172e-05,
+      "loss": 0.4657,
+      "step": 569
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985637313220261e-05,
+      "loss": 0.4434,
+      "step": 570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9855580435739507e-05,
+      "loss": 0.4493,
+      "step": 571
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9854785573702053e-05,
+      "loss": 0.4433,
+      "step": 572
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985398854626492e-05,
+      "loss": 0.4536,
+      "step": 573
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985318935360323e-05,
+      "loss": 0.476,
+      "step": 574
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9852387995892592e-05,
+      "loss": 0.4802,
+      "step": 575
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985158447330909e-05,
+      "loss": 0.4526,
+      "step": 576
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.985077878602928e-05,
+      "loss": 0.4522,
+      "step": 577
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.98499709342302e-05,
+      "loss": 0.4596,
+      "step": 578
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984916091808936e-05,
+      "loss": 0.4704,
+      "step": 579
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984834873778474e-05,
+      "loss": 0.471,
+      "step": 580
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9847534393494806e-05,
+      "loss": 0.4968,
+      "step": 581
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9846717885398496e-05,
+      "loss": 0.4927,
+      "step": 582
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9845899213675218e-05,
+      "loss": 0.4633,
+      "step": 583
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.984507837850486e-05,
+      "loss": 0.4516,
+      "step": 584
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9844255380067782e-05,
+      "loss": 0.4386,
+      "step": 585
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9843430218544827e-05,
+      "loss": 0.4827,
+      "step": 586
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9842602894117307e-05,
+      "loss": 0.4894,
+      "step": 587
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.984177340696701e-05,
+      "loss": 0.4952,
+      "step": 588
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9840941757276204e-05,
+      "loss": 0.478,
+      "step": 589
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9840107945227623e-05,
+      "loss": 0.4463,
+      "step": 590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983927197100448e-05,
+      "loss": 0.4472,
+      "step": 591
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9838433834790468e-05,
+      "loss": 0.4479,
+      "step": 592
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983759353676975e-05,
+      "loss": 0.4717,
+      "step": 593
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9836751077126967e-05,
+      "loss": 0.4649,
+      "step": 594
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983590645604723e-05,
+      "loss": 0.481,
+      "step": 595
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983505967371613e-05,
+      "loss": 0.4618,
+      "step": 596
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983421073031973e-05,
+      "loss": 0.4389,
+      "step": 597
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983335962604457e-05,
+      "loss": 0.4423,
+      "step": 598
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9832506361077667e-05,
+      "loss": 0.4588,
+      "step": 599
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9831650935606503e-05,
+      "loss": 0.4618,
+      "step": 600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9830793349819045e-05,
+      "loss": 0.4567,
+      "step": 601
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9829933603903732e-05,
+      "loss": 0.4568,
+      "step": 602
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9829071698049474e-05,
+      "loss": 0.4415,
+      "step": 603
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982820763244566e-05,
+      "loss": 0.4392,
+      "step": 604
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9827341407282157e-05,
+      "loss": 0.4503,
+      "step": 605
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982647302274929e-05,
+      "loss": 0.4608,
+      "step": 606
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9825602479037875e-05,
+      "loss": 0.466,
+      "step": 607
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.98247297763392e-05,
+      "loss": 0.4525,
+      "step": 608
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9823854914845022e-05,
+      "loss": 0.4434,
+      "step": 609
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9822977894747574e-05,
+      "loss": 0.4595,
+      "step": 610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982209871623957e-05,
+      "loss": 0.4483,
+      "step": 611
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9821217379514185e-05,
+      "loss": 0.4816,
+      "step": 612
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.982033388476508e-05,
+      "loss": 0.4568,
+      "step": 613
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.981944823218638e-05,
+      "loss": 0.443,
+      "step": 614
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9818560421972703e-05,
+      "loss": 0.4439,
+      "step": 615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9817670454319118e-05,
+      "loss": 0.4538,
+      "step": 616
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981677832942118e-05,
+      "loss": 0.4597,
+      "step": 617
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9815884047474917e-05,
+      "loss": 0.4454,
+      "step": 618
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981498760867683e-05,
+      "loss": 0.4551,
+      "step": 619
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9814089013223892e-05,
+      "loss": 0.4357,
+      "step": 620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981318826131356e-05,
+      "loss": 0.4299,
+      "step": 621
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981228535314375e-05,
+      "loss": 0.4368,
+      "step": 622
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9811380288912855e-05,
+      "loss": 0.4807,
+      "step": 623
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981047306881976e-05,
+      "loss": 0.4651,
+      "step": 624
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980956369306379e-05,
+      "loss": 0.4623,
+      "step": 625
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980865216184478e-05,
+      "loss": 0.4558,
+      "step": 626
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980773847536301e-05,
+      "loss": 0.4396,
+      "step": 627
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9806822633819252e-05,
+      "loss": 0.456,
+      "step": 628
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980590463741474e-05,
+      "loss": 0.4491,
+      "step": 629
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9804984486351187e-05,
+      "loss": 0.4681,
+      "step": 630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9804062180830782e-05,
+      "loss": 0.4477,
+      "step": 631
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980313772105618e-05,
+      "loss": 0.447,
+      "step": 632
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9802211107230517e-05,
+      "loss": 0.4402,
+      "step": 633
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980128233955739e-05,
+      "loss": 0.4829,
+      "step": 634
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.980035141824089e-05,
+      "loss": 0.451,
+      "step": 635
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979941834348556e-05,
+      "loss": 0.442,
+      "step": 636
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9798483115496426e-05,
+      "loss": 0.4282,
+      "step": 637
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9797545734478994e-05,
+      "loss": 0.4357,
+      "step": 638
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9796606200639224e-05,
+      "loss": 0.4509,
+      "step": 639
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9795664514183574e-05,
+      "loss": 0.4706,
+      "step": 640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979472067531895e-05,
+      "loss": 0.4469,
+      "step": 641
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9793774684252743e-05,
+      "loss": 0.4742,
+      "step": 642
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9792826541192825e-05,
+      "loss": 0.4501,
+      "step": 643
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9791876246347526e-05,
+      "loss": 0.4452,
+      "step": 644
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9790923799925657e-05,
+      "loss": 0.4367,
+      "step": 645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9789969202136495e-05,
+      "loss": 0.4577,
+      "step": 646
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.97890124531898e-05,
+      "loss": 0.4563,
+      "step": 647
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.97880535532958e-05,
+      "loss": 0.4437,
+      "step": 648
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978709250266519e-05,
+      "loss": 0.4539,
+      "step": 649
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9786129301509146e-05,
+      "loss": 0.4865,
+      "step": 650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978516395003931e-05,
+      "loss": 0.4443,
+      "step": 651
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9784196448467803e-05,
+      "loss": 0.4608,
+      "step": 652
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9783226797007214e-05,
+      "loss": 0.4405,
+      "step": 653
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.97822549958706e-05,
+      "loss": 0.4688,
+      "step": 654
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9781281045271507e-05,
+      "loss": 0.4614,
+      "step": 655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978030494542393e-05,
+      "loss": 0.4357,
+      "step": 656
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9779326696542355e-05,
+      "loss": 0.4585,
+      "step": 657
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9778346298841727e-05,
+      "loss": 0.4426,
+      "step": 658
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977736375253748e-05,
+      "loss": 0.4495,
+      "step": 659
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9776379057845498e-05,
+      "loss": 0.4258,
+      "step": 660
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9775392214982158e-05,
+      "loss": 0.4374,
+      "step": 661
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9774403224164295e-05,
+      "loss": 0.494,
+      "step": 662
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9773412085609222e-05,
+      "loss": 0.4852,
+      "step": 663
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9772418799534724e-05,
+      "loss": 0.4775,
+      "step": 664
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9771423366159053e-05,
+      "loss": 0.4505,
+      "step": 665
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9770425785700936e-05,
+      "loss": 0.4526,
+      "step": 666
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976942605837958e-05,
+      "loss": 0.438,
+      "step": 667
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.976842418441465e-05,
+      "loss": 0.4589,
+      "step": 668
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9767420164026285e-05,
+      "loss": 0.4512,
+      "step": 669
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9766413997435102e-05,
+      "loss": 0.4436,
+      "step": 670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.976540568486219e-05,
+      "loss": 0.4542,
+      "step": 671
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9764395226529105e-05,
+      "loss": 0.4545,
+      "step": 672
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9763382622657875e-05,
+      "loss": 0.445,
+      "step": 673
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9762367873471e-05,
+      "loss": 0.4391,
+      "step": 674
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9761350979191453e-05,
+      "loss": 0.4523,
+      "step": 675
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9760331940042673e-05,
+      "loss": 0.4424,
+      "step": 676
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975931075624858e-05,
+      "loss": 0.4614,
+      "step": 677
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9758287428033553e-05,
+      "loss": 0.4438,
+      "step": 678
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9757261955622454e-05,
+      "loss": 0.4482,
+      "step": 679
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975623433924061e-05,
+      "loss": 0.4446,
+      "step": 680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975520457911382e-05,
+      "loss": 0.4485,
+      "step": 681
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.975417267546835e-05,
+      "loss": 0.4535,
+      "step": 682
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9753138628530942e-05,
+      "loss": 0.449,
+      "step": 683
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9752102438528814e-05,
+      "loss": 0.4584,
+      "step": 684
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9751064105689643e-05,
+      "loss": 0.4548,
+      "step": 685
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9750023630241585e-05,
+      "loss": 0.4467,
+      "step": 686
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9748981012413263e-05,
+      "loss": 0.4558,
+      "step": 687
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974793625243378e-05,
+      "loss": 0.4562,
+      "step": 688
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9746889350532685e-05,
+      "loss": 0.4574,
+      "step": 689
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.974584030694003e-05,
+      "loss": 0.438,
+      "step": 690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9744789121886314e-05,
+      "loss": 0.4592,
+      "step": 691
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9743735795602517e-05,
+      "loss": 0.446,
+      "step": 692
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9742680328320093e-05,
+      "loss": 0.4374,
+      "step": 693
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9741622720270954e-05,
+      "loss": 0.4483,
+      "step": 694
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9740562971687487e-05,
+      "loss": 0.4576,
+      "step": 695
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9739501082802555e-05,
+      "loss": 0.4811,
+      "step": 696
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.973843705384949e-05,
+      "loss": 0.4469,
+      "step": 697
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.973737088506209e-05,
+      "loss": 0.4485,
+      "step": 698
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973630257667462e-05,
+      "loss": 0.4521,
+      "step": 699
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9735232128921828e-05,
+      "loss": 0.4549,
+      "step": 700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9734159542038924e-05,
+      "loss": 0.4651,
+      "step": 701
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.973308481626158e-05,
+      "loss": 0.4595,
+      "step": 702
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9732007951825952e-05,
+      "loss": 0.4616,
+      "step": 703
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9730928948968664e-05,
+      "loss": 0.4363,
+      "step": 704
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.97298478079268e-05,
+      "loss": 0.4521,
+      "step": 705
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972876452893792e-05,
+      "loss": 0.4476,
+      "step": 706
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9727679112240058e-05,
+      "loss": 0.4779,
+      "step": 707
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972659155807171e-05,
+      "loss": 0.4839,
+      "step": 708
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972550186667185e-05,
+      "loss": 0.4379,
+      "step": 709
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972441003827991e-05,
+      "loss": 0.4537,
+      "step": 710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9723316073135798e-05,
+      "loss": 0.4425,
+      "step": 711
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.97222199714799e-05,
+      "loss": 0.4478,
+      "step": 712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9721121733553057e-05,
+      "loss": 0.4403,
+      "step": 713
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9720021359596587e-05,
+      "loss": 0.4485,
+      "step": 714
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9718918849852277e-05,
+      "loss": 0.4543,
+      "step": 715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971781420456238e-05,
+      "loss": 0.4574,
+      "step": 716
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9716707423969622e-05,
+      "loss": 0.4498,
+      "step": 717
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.97155985083172e-05,
+      "loss": 0.4449,
+      "step": 718
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971448745784877e-05,
+      "loss": 0.4579,
+      "step": 719
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971337427280847e-05,
+      "loss": 0.4569,
+      "step": 720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9712258953440898e-05,
+      "loss": 0.4597,
+      "step": 721
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9711141499991125e-05,
+      "loss": 0.443,
+      "step": 722
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971002191270469e-05,
+      "loss": 0.4578,
+      "step": 723
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.97089001918276e-05,
+      "loss": 0.4637,
+      "step": 724
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.970777633760633e-05,
+      "loss": 0.4342,
+      "step": 725
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9706650350287825e-05,
+      "loss": 0.4396,
+      "step": 726
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970552223011951e-05,
+      "loss": 0.4317,
+      "step": 727
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9704391977349253e-05,
+      "loss": 0.4649,
+      "step": 728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9703259592225412e-05,
+      "loss": 0.4656,
+      "step": 729
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9702125074996806e-05,
+      "loss": 0.4526,
+      "step": 730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9700988425912727e-05,
+      "loss": 0.4447,
+      "step": 731
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9699849645222927e-05,
+      "loss": 0.4336,
+      "step": 732
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9698708733177628e-05,
+      "loss": 0.4354,
+      "step": 733
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9697565690027534e-05,
+      "loss": 0.4409,
+      "step": 734
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9696420516023796e-05,
+      "loss": 0.4542,
+      "step": 735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9695273211418055e-05,
+      "loss": 0.4402,
+      "step": 736
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.96941237764624e-05,
+      "loss": 0.4414,
+      "step": 737
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9692972211409398e-05,
+      "loss": 0.4446,
+      "step": 738
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9691818516512088e-05,
+      "loss": 0.4643,
+      "step": 739
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9690662692023967e-05,
+      "loss": 0.4496,
+      "step": 740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968950473819901e-05,
+      "loss": 0.4277,
+      "step": 741
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9688344655291657e-05,
+      "loss": 0.447,
+      "step": 742
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9687182443556807e-05,
+      "loss": 0.4515,
+      "step": 743
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9686018103249833e-05,
+      "loss": 0.4784,
+      "step": 744
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9684851634626582e-05,
+      "loss": 0.4659,
+      "step": 745
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9683683037943365e-05,
+      "loss": 0.4576,
+      "step": 746
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968251231345695e-05,
+      "loss": 0.4677,
+      "step": 747
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9681339461424587e-05,
+      "loss": 0.4487,
+      "step": 748
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9680164482103987e-05,
+      "loss": 0.4667,
+      "step": 749
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967898737575333e-05,
+      "loss": 0.4512,
+      "step": 750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9677808142631263e-05,
+      "loss": 0.4462,
+      "step": 751
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9676626782996892e-05,
+      "loss": 0.4357,
+      "step": 752
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9675443297109812e-05,
+      "loss": 0.4546,
+      "step": 753
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967425768523006e-05,
+      "loss": 0.4441,
+      "step": 754
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9673069947618152e-05,
+      "loss": 0.455,
+      "step": 755
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9671880084535076e-05,
+      "loss": 0.4628,
+      "step": 756
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9670688096242277e-05,
+      "loss": 0.4456,
+      "step": 757
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9669493983001674e-05,
+      "loss": 0.436,
+      "step": 758
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966829774507565e-05,
+      "loss": 0.4368,
+      "step": 759
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9667099382727057e-05,
+      "loss": 0.4281,
+      "step": 760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.966589889621921e-05,
+      "loss": 0.4517,
+      "step": 761
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9664696285815893e-05,
+      "loss": 0.4708,
+      "step": 762
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9663491551781356e-05,
+      "loss": 0.4536,
+      "step": 763
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9662284694380313e-05,
+      "loss": 0.4786,
+      "step": 764
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9661075713877954e-05,
+      "loss": 0.4666,
+      "step": 765
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9659864610539928e-05,
+      "loss": 0.4573,
+      "step": 766
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9658651384632353e-05,
+      "loss": 0.4492,
+      "step": 767
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9657436036421808e-05,
+      "loss": 0.4596,
+      "step": 768
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965621856617534e-05,
+      "loss": 0.4418,
+      "step": 769
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965499897416047e-05,
+      "loss": 0.4397,
+      "step": 770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.965377726064518e-05,
+      "loss": 0.455,
+      "step": 771
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9652553425897914e-05,
+      "loss": 0.4365,
+      "step": 772
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9651327470187593e-05,
+      "loss": 0.4603,
+      "step": 773
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9650099393783585e-05,
+      "loss": 0.4505,
+      "step": 774
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9648869196955753e-05,
+      "loss": 0.4765,
+      "step": 775
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9647636879974394e-05,
+      "loss": 0.4497,
+      "step": 776
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9646402443110295e-05,
+      "loss": 0.4428,
+      "step": 777
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9645165886634692e-05,
+      "loss": 0.4309,
+      "step": 778
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9643927210819305e-05,
+      "loss": 0.4418,
+      "step": 779
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9642686415936304e-05,
+      "loss": 0.4631,
+      "step": 780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9641443502258326e-05,
+      "loss": 0.4688,
+      "step": 781
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.964019847005848e-05,
+      "loss": 0.4496,
+      "step": 782
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963895131961034e-05,
+      "loss": 0.4584,
+      "step": 783
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9637702051187945e-05,
+      "loss": 0.4516,
+      "step": 784
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963645066506579e-05,
+      "loss": 0.4577,
+      "step": 785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9635197161518852e-05,
+      "loss": 0.4397,
+      "step": 786
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963394154082256e-05,
+      "loss": 0.451,
+      "step": 787
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963268380325281e-05,
+      "loss": 0.4623,
+      "step": 788
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963142394908597e-05,
+      "loss": 0.4539,
+      "step": 789
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9630161978598865e-05,
+      "loss": 0.4586,
+      "step": 790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962889789206879e-05,
+      "loss": 0.4855,
+      "step": 791
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9627631689773508e-05,
+      "loss": 0.4836,
+      "step": 792
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9626363371991237e-05,
+      "loss": 0.474,
+      "step": 793
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9625092939000665e-05,
+      "loss": 0.471,
+      "step": 794
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.962382039108095e-05,
+      "loss": 0.4438,
+      "step": 795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9622545728511707e-05,
+      "loss": 0.4632,
+      "step": 796
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9621268951573022e-05,
+      "loss": 0.4635,
+      "step": 797
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961999006054544e-05,
+      "loss": 0.4356,
+      "step": 798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9618709055709967e-05,
+      "loss": 0.4598,
+      "step": 799
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961742593734809e-05,
+      "loss": 0.4305,
+      "step": 800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961614070574174e-05,
+      "loss": 0.4542,
+      "step": 801
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9614853361173326e-05,
+      "loss": 0.4472,
+      "step": 802
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9613563903925718e-05,
+      "loss": 0.4516,
+      "step": 803
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961227233428225e-05,
+      "loss": 0.4817,
+      "step": 804
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9610978652526714e-05,
+      "loss": 0.4514,
+      "step": 805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9609682858943383e-05,
+      "loss": 0.4554,
+      "step": 806
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960838495381697e-05,
+      "loss": 0.4499,
+      "step": 807
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.960708493743267e-05,
+      "loss": 0.4479,
+      "step": 808
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960578281007614e-05,
+      "loss": 0.434,
+      "step": 809
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960447857203349e-05,
+      "loss": 0.4477,
+      "step": 810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960317222359131e-05,
+      "loss": 0.4768,
+      "step": 811
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9601863765036643e-05,
+      "loss": 0.464,
+      "step": 812
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960055319665699e-05,
+      "loss": 0.4743,
+      "step": 813
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9599240518740327e-05,
+      "loss": 0.469,
+      "step": 814
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9597925731575095e-05,
+      "loss": 0.4526,
+      "step": 815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9596608835450185e-05,
+      "loss": 0.4483,
+      "step": 816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9595289830654967e-05,
+      "loss": 0.4494,
+      "step": 817
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9593968717479266e-05,
+      "loss": 0.4692,
+      "step": 818
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9592645496213367e-05,
+      "loss": 0.4596,
+      "step": 819
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9591320167148025e-05,
+      "loss": 0.4624,
+      "step": 820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.958999273057445e-05,
+      "loss": 0.4701,
+      "step": 821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9588663186784334e-05,
+      "loss": 0.4486,
+      "step": 822
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9587331536069808e-05,
+      "loss": 0.4555,
+      "step": 823
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9585997778723474e-05,
+      "loss": 0.4705,
+      "step": 824
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958466191503841e-05,
+      "loss": 0.4617,
+      "step": 825
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9583323945308138e-05,
+      "loss": 0.4521,
+      "step": 826
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9581983869826658e-05,
+      "loss": 0.4435,
+      "step": 827
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.958064168888842e-05,
+      "loss": 0.4488,
+      "step": 828
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9579297402788343e-05,
+      "loss": 0.4439,
+      "step": 829
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957795101182181e-05,
+      "loss": 0.4606,
+      "step": 830
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9576602516284664e-05,
+      "loss": 0.447,
+      "step": 831
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957525191647321e-05,
+      "loss": 0.4597,
+      "step": 832
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9573899212684223e-05,
+      "loss": 0.4271,
+      "step": 833
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957254440521492e-05,
+      "loss": 0.4451,
+      "step": 834
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9571187494363005e-05,
+      "loss": 0.4248,
+      "step": 835
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9569828480426632e-05,
+      "loss": 0.4412,
+      "step": 836
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956846736370441e-05,
+      "loss": 0.4582,
+      "step": 837
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9567104144495434e-05,
+      "loss": 0.4616,
+      "step": 838
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956573882309923e-05,
+      "loss": 0.4447,
+      "step": 839
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9564371399815807e-05,
+      "loss": 0.4478,
+      "step": 840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956300187494563e-05,
+      "loss": 0.4321,
+      "step": 841
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9561630248789624e-05,
+      "loss": 0.4592,
+      "step": 842
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956025652164918e-05,
+      "loss": 0.4437,
+      "step": 843
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9558880693826146e-05,
+      "loss": 0.4388,
+      "step": 844
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955750276562284e-05,
+      "loss": 0.4539,
+      "step": 845
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9556122737342026e-05,
+      "loss": 0.4416,
+      "step": 846
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955474060928694e-05,
+      "loss": 0.4688,
+      "step": 847
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9553356381761284e-05,
+      "loss": 0.4753,
+      "step": 848
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9551970055069213e-05,
+      "loss": 0.4729,
+      "step": 849
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9550581629515344e-05,
+      "loss": 0.4302,
+      "step": 850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9549191105404754e-05,
+      "loss": 0.4523,
+      "step": 851
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9547798483042992e-05,
+      "loss": 0.4365,
+      "step": 852
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9546403762736053e-05,
+      "loss": 0.4466,
+      "step": 853
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9545006944790405e-05,
+      "loss": 0.4493,
+      "step": 854
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9543608029512968e-05,
+      "loss": 0.4516,
+      "step": 855
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9542207017211127e-05,
+      "loss": 0.451,
+      "step": 856
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.954080390819273e-05,
+      "loss": 0.437,
+      "step": 857
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9539398702766085e-05,
+      "loss": 0.4467,
+      "step": 858
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953799140123995e-05,
+      "loss": 0.4482,
+      "step": 859
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953658200392356e-05,
+      "loss": 0.4539,
+      "step": 860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9535170511126603e-05,
+      "loss": 0.4282,
+      "step": 861
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9533756923159226e-05,
+      "loss": 0.4399,
+      "step": 862
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9532341240332034e-05,
+      "loss": 0.4758,
+      "step": 863
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9530923462956106e-05,
+      "loss": 0.4317,
+      "step": 864
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952950359134296e-05,
+      "loss": 0.4514,
+      "step": 865
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952808162580459e-05,
+      "loss": 0.4513,
+      "step": 866
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9526657566653447e-05,
+      "loss": 0.4405,
+      "step": 867
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.952523141420244e-05,
+      "loss": 0.4655,
+      "step": 868
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9523803168764934e-05,
+      "loss": 0.4385,
+      "step": 869
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9522372830654767e-05,
+      "loss": 0.4521,
+      "step": 870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9520940400186222e-05,
+      "loss": 0.4513,
+      "step": 871
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951950587767405e-05,
+      "loss": 0.4533,
+      "step": 872
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951806926343346e-05,
+      "loss": 0.4421,
+      "step": 873
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9516630557780124e-05,
+      "loss": 0.442,
+      "step": 874
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951518976103016e-05,
+      "loss": 0.4575,
+      "step": 875
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9513746873500167e-05,
+      "loss": 0.4406,
+      "step": 876
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.951230189550718e-05,
+      "loss": 0.4345,
+      "step": 877
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9510854827368716e-05,
+      "loss": 0.4663,
+      "step": 878
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9509405669402732e-05,
+      "loss": 0.4417,
+      "step": 879
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950795442192766e-05,
+      "loss": 0.4464,
+      "step": 880
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950650108526238e-05,
+      "loss": 0.4349,
+      "step": 881
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.950504565972623e-05,
+      "loss": 0.4506,
+      "step": 882
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9503588145639022e-05,
+      "loss": 0.4596,
+      "step": 883
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9502128543321008e-05,
+      "loss": 0.454,
+      "step": 884
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9500666853092914e-05,
+      "loss": 0.4484,
+      "step": 885
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9499203075275912e-05,
+      "loss": 0.4659,
+      "step": 886
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9497737210191642e-05,
+      "loss": 0.4506,
+      "step": 887
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9496269258162202e-05,
+      "loss": 0.4562,
+      "step": 888
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9494799219510142e-05,
+      "loss": 0.4566,
+      "step": 889
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949332709455848e-05,
+      "loss": 0.444,
+      "step": 890
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949185288363068e-05,
+      "loss": 0.4367,
+      "step": 891
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9490376587050682e-05,
+      "loss": 0.4424,
+      "step": 892
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9488898205142862e-05,
+      "loss": 0.4525,
+      "step": 893
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948741773823207e-05,
+      "loss": 0.4569,
+      "step": 894
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9485935186643617e-05,
+      "loss": 0.4284,
+      "step": 895
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9484450550703254e-05,
+      "loss": 0.4399,
+      "step": 896
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948296383073721e-05,
+      "loss": 0.4551,
+      "step": 897
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.948147502707216e-05,
+      "loss": 0.4288,
+      "step": 898
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947998414003524e-05,
+      "loss": 0.4411,
+      "step": 899
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9478491169954042e-05,
+      "loss": 0.4591,
+      "step": 900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9476996117156623e-05,
+      "loss": 0.4452,
+      "step": 901
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9475498981971487e-05,
+      "loss": 0.4476,
+      "step": 902
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.94739997647276e-05,
+      "loss": 0.4504,
+      "step": 903
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947249846575439e-05,
+      "loss": 0.4423,
+      "step": 904
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9470995085381738e-05,
+      "loss": 0.4486,
+      "step": 905
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946948962393998e-05,
+      "loss": 0.4495,
+      "step": 906
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9467982081759914e-05,
+      "loss": 0.4435,
+      "step": 907
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946647245917279e-05,
+      "loss": 0.4548,
+      "step": 908
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946496075651033e-05,
+      "loss": 0.4488,
+      "step": 909
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9463446974104688e-05,
+      "loss": 0.4412,
+      "step": 910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9461931112288494e-05,
+      "loss": 0.4457,
+      "step": 911
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.946041317139483e-05,
+      "loss": 0.4524,
+      "step": 912
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9458893151757237e-05,
+      "loss": 0.46,
+      "step": 913
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.94573710537097e-05,
+      "loss": 0.4554,
+      "step": 914
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9455846877586686e-05,
+      "loss": 0.449,
+      "step": 915
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.945432062372309e-05,
+      "loss": 0.4545,
+      "step": 916
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9452792292454282e-05,
+      "loss": 0.4506,
+      "step": 917
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9451261884116088e-05,
+      "loss": 0.4398,
+      "step": 918
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9449729399044777e-05,
+      "loss": 0.4445,
+      "step": 919
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944819483757709e-05,
+      "loss": 0.4371,
+      "step": 920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9446658200050215e-05,
+      "loss": 0.456,
+      "step": 921
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9445119486801798e-05,
+      "loss": 0.4326,
+      "step": 922
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9443578698169944e-05,
+      "loss": 0.4929,
+      "step": 923
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9442035834493204e-05,
+      "loss": 0.435,
+      "step": 924
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9440490896110603e-05,
+      "loss": 0.4342,
+      "step": 925
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9438943883361606e-05,
+      "loss": 0.4494,
+      "step": 926
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9437394796586143e-05,
+      "loss": 0.4352,
+      "step": 927
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9435843636124588e-05,
+      "loss": 0.4399,
+      "step": 928
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9434290402317787e-05,
+      "loss": 0.4602,
+      "step": 929
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.943273509550703e-05,
+      "loss": 0.4227,
+      "step": 930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9431177716034066e-05,
+      "loss": 0.4349,
+      "step": 931
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.94296182642411e-05,
+      "loss": 0.4507,
+      "step": 932
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.942805674047079e-05,
+      "loss": 0.4464,
+      "step": 933
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942649314506625e-05,
+      "loss": 0.4446,
+      "step": 934
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9424927478371054e-05,
+      "loss": 0.4498,
+      "step": 935
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9423359740729224e-05,
+      "loss": 0.4362,
+      "step": 936
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.942178993248524e-05,
+      "loss": 0.4516,
+      "step": 937
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9420218053984042e-05,
+      "loss": 0.4508,
+      "step": 938
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9418644105571013e-05,
+      "loss": 0.445,
+      "step": 939
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9417068087592002e-05,
+      "loss": 0.4517,
+      "step": 940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9415490000393308e-05,
+      "loss": 0.4558,
+      "step": 941
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941390984432168e-05,
+      "loss": 0.4474,
+      "step": 942
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9412327619724342e-05,
+      "loss": 0.4469,
+      "step": 943
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941074332694894e-05,
+      "loss": 0.4472,
+      "step": 944
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.94091569663436e-05,
+      "loss": 0.4573,
+      "step": 945
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9407568538256898e-05,
+      "loss": 0.4243,
+      "step": 946
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9405978043037852e-05,
+      "loss": 0.4423,
+      "step": 947
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.940438548103595e-05,
+      "loss": 0.4394,
+      "step": 948
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9402790852601124e-05,
+      "loss": 0.4435,
+      "step": 949
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.940119415808376e-05,
+      "loss": 0.449,
+      "step": 950
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9399595397834705e-05,
+      "loss": 0.4404,
+      "step": 951
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9397994572205254e-05,
+      "loss": 0.4395,
+      "step": 952
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9396391681547156e-05,
+      "loss": 0.437,
+      "step": 953
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9394786726212623e-05,
+      "loss": 0.4339,
+      "step": 954
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9393179706554303e-05,
+      "loss": 0.4488,
+      "step": 955
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9391570622925316e-05,
+      "loss": 0.4453,
+      "step": 956
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9389959475679223e-05,
+      "loss": 0.447,
+      "step": 957
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9388346265170042e-05,
+      "loss": 0.4516,
+      "step": 958
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9386730991752244e-05,
+      "loss": 0.4524,
+      "step": 959
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9385113655780762e-05,
+      "loss": 0.4515,
+      "step": 960
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9383494257610965e-05,
+      "loss": 0.4386,
+      "step": 961
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.938187279759869e-05,
+      "loss": 0.4511,
+      "step": 962
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9380249276100217e-05,
+      "loss": 0.4383,
+      "step": 963
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937862369347229e-05,
+      "loss": 0.4354,
+      "step": 964
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9376996050072095e-05,
+      "loss": 0.4438,
+      "step": 965
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937536634625728e-05,
+      "loss": 0.4328,
+      "step": 966
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9373734582385936e-05,
+      "loss": 0.4569,
+      "step": 967
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937210075881661e-05,
+      "loss": 0.4853,
+      "step": 968
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9370464875908312e-05,
+      "loss": 0.4855,
+      "step": 969
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936882693402049e-05,
+      "loss": 0.4526,
+      "step": 970
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9367186933513048e-05,
+      "loss": 0.4554,
+      "step": 971
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936554487474635e-05,
+      "loss": 0.452,
+      "step": 972
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9363900758081206e-05,
+      "loss": 0.471,
+      "step": 973
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9362254583878872e-05,
+      "loss": 0.4753,
+      "step": 974
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.936060635250107e-05,
+      "loss": 0.4792,
+      "step": 975
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9358956064309972e-05,
+      "loss": 0.4566,
+      "step": 976
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9357303719668184e-05,
+      "loss": 0.4546,
+      "step": 977
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935564931893879e-05,
+      "loss": 0.4545,
+      "step": 978
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.93539928624853e-05,
+      "loss": 0.4391,
+      "step": 979
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.93523343506717e-05,
+      "loss": 0.4433,
+      "step": 980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.935067378386241e-05,
+      "loss": 0.4506,
+      "step": 981
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9349011162422305e-05,
+      "loss": 0.4653,
+      "step": 982
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934734648671672e-05,
+      "loss": 0.4326,
+      "step": 983
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9345679757111434e-05,
+      "loss": 0.4454,
+      "step": 984
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934401097397268e-05,
+      "loss": 0.4669,
+      "step": 985
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9342340137667136e-05,
+      "loss": 0.469,
+      "step": 986
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9340667248561943e-05,
+      "loss": 0.441,
+      "step": 987
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9338992307024677e-05,
+      "loss": 0.4704,
+      "step": 988
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9337315313423386e-05,
+      "loss": 0.4476,
+      "step": 989
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9335636268126548e-05,
+      "loss": 0.4531,
+      "step": 990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9333955171503103e-05,
+      "loss": 0.4324,
+      "step": 991
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933227202392244e-05,
+      "loss": 0.4499,
+      "step": 992
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.93305868257544e-05,
+      "loss": 0.4492,
+      "step": 993
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9328899577369273e-05,
+      "loss": 0.4475,
+      "step": 994
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9327210279137794e-05,
+      "loss": 0.4487,
+      "step": 995
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9325518931431164e-05,
+      "loss": 0.425,
+      "step": 996
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932382553462101e-05,
+      "loss": 0.4464,
+      "step": 997
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9322130089079438e-05,
+      "loss": 0.4399,
+      "step": 998
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9320432595178983e-05,
+      "loss": 0.4562,
+      "step": 999
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9318733053292632e-05,
+      "loss": 0.4639,
+      "step": 1000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9317031463793835e-05,
+      "loss": 0.4385,
+      "step": 1001
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9315327827056477e-05,
+      "loss": 0.4412,
+      "step": 1002
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9313622143454904e-05,
+      "loss": 0.4623,
+      "step": 1003
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.931191441336391e-05,
+      "loss": 0.433,
+      "step": 1004
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9310204637158723e-05,
+      "loss": 0.457,
+      "step": 1005
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9308492815215044e-05,
+      "loss": 0.4456,
+      "step": 1006
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9306778947909014e-05,
+      "loss": 0.4728,
+      "step": 1007
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930506303561722e-05,
+      "loss": 0.4314,
+      "step": 1008
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9303345078716695e-05,
+      "loss": 0.4386,
+      "step": 1009
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9301625077584935e-05,
+      "loss": 0.4476,
+      "step": 1010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9299903032599872e-05,
+      "loss": 0.4592,
+      "step": 1011
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9298178944139895e-05,
+      "loss": 0.4749,
+      "step": 1012
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9296452812583842e-05,
+      "loss": 0.4565,
+      "step": 1013
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929472463831099e-05,
+      "loss": 0.462,
+      "step": 1014
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9292994421701075e-05,
+      "loss": 0.442,
+      "step": 1015
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9291262163134282e-05,
+      "loss": 0.4287,
+      "step": 1016
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928952786299124e-05,
+      "loss": 0.4441,
+      "step": 1017
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9287791521653027e-05,
+      "loss": 0.4552,
+      "step": 1018
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9286053139501173e-05,
+      "loss": 0.4682,
+      "step": 1019
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9284312716917652e-05,
+      "loss": 0.4633,
+      "step": 1020
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9282570254284883e-05,
+      "loss": 0.4295,
+      "step": 1021
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.928082575198575e-05,
+      "loss": 0.4367,
+      "step": 1022
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9279079210403562e-05,
+      "loss": 0.4325,
+      "step": 1023
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.92773306299221e-05,
+      "loss": 0.4382,
+      "step": 1024
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927558001092557e-05,
+      "loss": 0.4443,
+      "step": 1025
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9273827353798647e-05,
+      "loss": 0.4604,
+      "step": 1026
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9272072658926437e-05,
+      "loss": 0.441,
+      "step": 1027
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.92703159266945e-05,
+      "loss": 0.4614,
+      "step": 1028
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.926855715748885e-05,
+      "loss": 0.4373,
+      "step": 1029
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9266796351695936e-05,
+      "loss": 0.4628,
+      "step": 1030
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9265033509702662e-05,
+      "loss": 0.4264,
+      "step": 1031
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9263268631896382e-05,
+      "loss": 0.4403,
+      "step": 1032
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9261501718664895e-05,
+      "loss": 0.471,
+      "step": 1033
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925973277039644e-05,
+      "loss": 0.4489,
+      "step": 1034
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925796178747972e-05,
+      "loss": 0.4446,
+      "step": 1035
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9256188770303862e-05,
+      "loss": 0.4278,
+      "step": 1036
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.925441371925846e-05,
+      "loss": 0.4444,
+      "step": 1037
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9252636634733542e-05,
+      "loss": 0.4267,
+      "step": 1038
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9250857517119596e-05,
+      "loss": 0.4666,
+      "step": 1039
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9249076366807543e-05,
+      "loss": 0.455,
+      "step": 1040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.924729318418876e-05,
+      "loss": 0.4621,
+      "step": 1041
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.924550796965506e-05,
+      "loss": 0.4467,
+      "step": 1042
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.924372072359872e-05,
+      "loss": 0.4412,
+      "step": 1043
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9241931446412446e-05,
+      "loss": 0.4345,
+      "step": 1044
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.92401401384894e-05,
+      "loss": 0.4405,
+      "step": 1045
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9238346800223183e-05,
+      "loss": 0.4561,
+      "step": 1046
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9236551432007854e-05,
+      "loss": 0.4531,
+      "step": 1047
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9234754034237907e-05,
+      "loss": 0.458,
+      "step": 1048
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9232954607308282e-05,
+      "loss": 0.4623,
+      "step": 1049
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9231153151614376e-05,
+      "loss": 0.4505,
+      "step": 1050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922934966755202e-05,
+      "loss": 0.4394,
+      "step": 1051
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922754415551749e-05,
+      "loss": 0.4483,
+      "step": 1052
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922573661590752e-05,
+      "loss": 0.4492,
+      "step": 1053
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922392704911928e-05,
+      "loss": 0.429,
+      "step": 1054
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922211545555039e-05,
+      "loss": 0.4578,
+      "step": 1055
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9220301835598906e-05,
+      "loss": 0.4533,
+      "step": 1056
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9218486189663342e-05,
+      "loss": 0.4486,
+      "step": 1057
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.921666851814265e-05,
+      "loss": 0.4381,
+      "step": 1058
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9214848821436224e-05,
+      "loss": 0.4487,
+      "step": 1059
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9213027099943916e-05,
+      "loss": 0.438,
+      "step": 1060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9211203354066004e-05,
+      "loss": 0.4311,
+      "step": 1061
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920937758420323e-05,
+      "loss": 0.4386,
+      "step": 1062
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9207549790756772e-05,
+      "loss": 0.4346,
+      "step": 1063
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9205719974128245e-05,
+      "loss": 0.4411,
+      "step": 1064
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920388813471972e-05,
+      "loss": 0.4427,
+      "step": 1065
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920205427293371e-05,
+      "loss": 0.4412,
+      "step": 1066
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920021838917317e-05,
+      "loss": 0.4383,
+      "step": 1067
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9198380483841497e-05,
+      "loss": 0.4582,
+      "step": 1068
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9196540557342537e-05,
+      "loss": 0.4284,
+      "step": 1069
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919469861008058e-05,
+      "loss": 0.4524,
+      "step": 1070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.919285464246036e-05,
+      "loss": 0.4353,
+      "step": 1071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9191008654887052e-05,
+      "loss": 0.4706,
+      "step": 1072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918916064776627e-05,
+      "loss": 0.4761,
+      "step": 1073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918731062150409e-05,
+      "loss": 0.4434,
+      "step": 1074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918545857650701e-05,
+      "loss": 0.4594,
+      "step": 1075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9183604513181984e-05,
+      "loss": 0.4298,
+      "step": 1076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918174843193641e-05,
+      "loss": 0.472,
+      "step": 1077
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917989033317812e-05,
+      "loss": 0.4653,
+      "step": 1078
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.91780302173154e-05,
+      "loss": 0.4512,
+      "step": 1079
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9176168084756973e-05,
+      "loss": 0.467,
+      "step": 1080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917430393591201e-05,
+      "loss": 0.4548,
+      "step": 1081
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.917243777119012e-05,
+      "loss": 0.4443,
+      "step": 1082
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9170569591001356e-05,
+      "loss": 0.4585,
+      "step": 1083
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.916869939575621e-05,
+      "loss": 0.455,
+      "step": 1084
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.916682718586563e-05,
+      "loss": 0.4446,
+      "step": 1085
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9164952961740997e-05,
+      "loss": 0.4695,
+      "step": 1086
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.916307672379413e-05,
+      "loss": 0.4506,
+      "step": 1087
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9161198472437302e-05,
+      "loss": 0.4389,
+      "step": 1088
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9159318208083216e-05,
+      "loss": 0.4569,
+      "step": 1089
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915743593114503e-05,
+      "loss": 0.4318,
+      "step": 1090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915555164203634e-05,
+      "loss": 0.4497,
+      "step": 1091
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915366534117117e-05,
+      "loss": 0.4435,
+      "step": 1092
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9151777028964013e-05,
+      "loss": 0.4649,
+      "step": 1093
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9149886705829783e-05,
+      "loss": 0.4378,
+      "step": 1094
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914799437218384e-05,
+      "loss": 0.4456,
+      "step": 1095
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914610002844199e-05,
+      "loss": 0.4424,
+      "step": 1096
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914420367502048e-05,
+      "loss": 0.4505,
+      "step": 1097
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9142305312335998e-05,
+      "loss": 0.4538,
+      "step": 1098
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914040494080567e-05,
+      "loss": 0.4536,
+      "step": 1099
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9138502560847066e-05,
+      "loss": 0.4412,
+      "step": 1100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9136598172878198e-05,
+      "loss": 0.4461,
+      "step": 1101
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913469177731752e-05,
+      "loss": 0.4401,
+      "step": 1102
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9132783374583927e-05,
+      "loss": 0.4445,
+      "step": 1103
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9130872965096747e-05,
+      "loss": 0.4747,
+      "step": 1104
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9128960549275764e-05,
+      "loss": 0.4465,
+      "step": 1105
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9127046127541187e-05,
+      "loss": 0.4413,
+      "step": 1106
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912512970031368e-05,
+      "loss": 0.4389,
+      "step": 1107
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912321126801434e-05,
+      "loss": 0.4464,
+      "step": 1108
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.91212908310647e-05,
+      "loss": 0.4376,
+      "step": 1109
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9119368389886746e-05,
+      "loss": 0.4467,
+      "step": 1110
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9117443944902893e-05,
+      "loss": 0.4454,
+      "step": 1111
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9115517496536006e-05,
+      "loss": 0.4553,
+      "step": 1112
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9113589045209377e-05,
+      "loss": 0.4559,
+      "step": 1113
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9111658591346752e-05,
+      "loss": 0.4405,
+      "step": 1114
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910972613537231e-05,
+      "loss": 0.4394,
+      "step": 1115
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9107791677710667e-05,
+      "loss": 0.4632,
+      "step": 1116
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9105855218786892e-05,
+      "loss": 0.4708,
+      "step": 1117
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9103916759026474e-05,
+      "loss": 0.4508,
+      "step": 1118
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9101976298855364e-05,
+      "loss": 0.4426,
+      "step": 1119
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910003383869993e-05,
+      "loss": 0.4402,
+      "step": 1120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9098089378986995e-05,
+      "loss": 0.4509,
+      "step": 1121
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9096142920143817e-05,
+      "loss": 0.4498,
+      "step": 1122
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909419446259809e-05,
+      "loss": 0.4481,
+      "step": 1123
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9092244006777955e-05,
+      "loss": 0.4638,
+      "step": 1124
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909029155311198e-05,
+      "loss": 0.4565,
+      "step": 1125
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9088337102029187e-05,
+      "loss": 0.446,
+      "step": 1126
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9086380653959027e-05,
+      "loss": 0.4235,
+      "step": 1127
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.908442220933139e-05,
+      "loss": 0.4307,
+      "step": 1128
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9082461768576604e-05,
+      "loss": 0.4508,
+      "step": 1129
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9080499332125442e-05,
+      "loss": 0.4593,
+      "step": 1130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907853490040911e-05,
+      "loss": 0.4336,
+      "step": 1131
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.907656847385926e-05,
+      "loss": 0.4658,
+      "step": 1132
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9074600052907967e-05,
+      "loss": 0.4587,
+      "step": 1133
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9072629637987757e-05,
+      "loss": 0.444,
+      "step": 1134
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9070657229531596e-05,
+      "loss": 0.4654,
+      "step": 1135
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9068682827972873e-05,
+      "loss": 0.4413,
+      "step": 1136
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9066706433745435e-05,
+      "loss": 0.4632,
+      "step": 1137
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906472804728355e-05,
+      "loss": 0.4406,
+      "step": 1138
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9062747669021934e-05,
+      "loss": 0.4457,
+      "step": 1139
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9060765299395735e-05,
+      "loss": 0.4423,
+      "step": 1140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905878093884054e-05,
+      "loss": 0.4467,
+      "step": 1141
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905679458779238e-05,
+      "loss": 0.4512,
+      "step": 1142
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9054806246687706e-05,
+      "loss": 0.4546,
+      "step": 1143
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9052815915963427e-05,
+      "loss": 0.4449,
+      "step": 1144
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.905082359605688e-05,
+      "loss": 0.4459,
+      "step": 1145
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9048829287405833e-05,
+      "loss": 0.4442,
+      "step": 1146
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9046832990448503e-05,
+      "loss": 0.4768,
+      "step": 1147
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.904483470562353e-05,
+      "loss": 0.4312,
+      "step": 1148
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.904283443337001e-05,
+      "loss": 0.4516,
+      "step": 1149
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9040832174127456e-05,
+      "loss": 0.4468,
+      "step": 1150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9038827928335827e-05,
+      "loss": 0.4452,
+      "step": 1151
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9036821696435522e-05,
+      "loss": 0.4308,
+      "step": 1152
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9034813478867368e-05,
+      "loss": 0.4474,
+      "step": 1153
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.903280327607263e-05,
+      "loss": 0.4493,
+      "step": 1154
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9030791088493018e-05,
+      "loss": 0.4557,
+      "step": 1155
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9028776916570663e-05,
+      "loss": 0.4426,
+      "step": 1156
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.902676076074815e-05,
+      "loss": 0.4455,
+      "step": 1157
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9024742621468483e-05,
+      "loss": 0.4453,
+      "step": 1158
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9022722499175113e-05,
+      "loss": 0.4509,
+      "step": 1159
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9020700394311922e-05,
+      "loss": 0.4155,
+      "step": 1160
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9018676307323227e-05,
+      "loss": 0.4153,
+      "step": 1161
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9016650238653786e-05,
+      "loss": 0.4206,
+      "step": 1162
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901462218874879e-05,
+      "loss": 0.4592,
+      "step": 1163
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9012592158053857e-05,
+      "loss": 0.4775,
+      "step": 1164
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901056014701505e-05,
+      "loss": 0.4639,
+      "step": 1165
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.900852615607887e-05,
+      "loss": 0.4218,
+      "step": 1166
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9006490185692235e-05,
+      "loss": 0.4461,
+      "step": 1167
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9004452236302527e-05,
+      "loss": 0.4539,
+      "step": 1168
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9002412308357532e-05,
+      "loss": 0.4503,
+      "step": 1169
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.900037040230549e-05,
+      "loss": 0.4659,
+      "step": 1170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899832651859507e-05,
+      "loss": 0.4663,
+      "step": 1171
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8996280657675383e-05,
+      "loss": 0.4504,
+      "step": 1172
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8994232819995953e-05,
+      "loss": 0.4531,
+      "step": 1173
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.899218300600677e-05,
+      "loss": 0.4536,
+      "step": 1174
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8990131216158225e-05,
+      "loss": 0.4306,
+      "step": 1175
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898807745090117e-05,
+      "loss": 0.4565,
+      "step": 1176
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898602171068688e-05,
+      "loss": 0.4501,
+      "step": 1177
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.898396399596706e-05,
+      "loss": 0.4494,
+      "step": 1178
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8981904307193855e-05,
+      "loss": 0.4828,
+      "step": 1179
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8979842644819842e-05,
+      "loss": 0.4569,
+      "step": 1180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8977779009298036e-05,
+      "loss": 0.4525,
+      "step": 1181
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8975713401081872e-05,
+      "loss": 0.4337,
+      "step": 1182
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8973645820625232e-05,
+      "loss": 0.4549,
+      "step": 1183
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.897157626838243e-05,
+      "loss": 0.4362,
+      "step": 1184
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896950474480821e-05,
+      "loss": 0.4445,
+      "step": 1185
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8967431250357747e-05,
+      "loss": 0.4437,
+      "step": 1186
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8965355785486652e-05,
+      "loss": 0.4412,
+      "step": 1187
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8963278350650967e-05,
+      "loss": 0.4535,
+      "step": 1188
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8961198946307172e-05,
+      "loss": 0.4423,
+      "step": 1189
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8959117572912173e-05,
+      "loss": 0.4363,
+      "step": 1190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8957034230923314e-05,
+      "loss": 0.4506,
+      "step": 1191
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895494892079837e-05,
+      "loss": 0.4384,
+      "step": 1192
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.895286164299554e-05,
+      "loss": 0.4268,
+      "step": 1193
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8950772397973475e-05,
+      "loss": 0.4494,
+      "step": 1194
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8948681186191244e-05,
+      "loss": 0.4462,
+      "step": 1195
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.894658800810834e-05,
+      "loss": 0.429,
+      "step": 1196
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.894449286418471e-05,
+      "loss": 0.4543,
+      "step": 1197
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8942395754880716e-05,
+      "loss": 0.4337,
+      "step": 1198
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8940296680657158e-05,
+      "loss": 0.4403,
+      "step": 1199
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8938195641975274e-05,
+      "loss": 0.4412,
+      "step": 1200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8936092639296716e-05,
+      "loss": 0.4346,
+      "step": 1201
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893398767308358e-05,
+      "loss": 0.4202,
+      "step": 1202
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8931880743798404e-05,
+      "loss": 0.454,
+      "step": 1203
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8929771851904132e-05,
+      "loss": 0.4301,
+      "step": 1204
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8927660997864157e-05,
+      "loss": 0.4282,
+      "step": 1205
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.89255481821423e-05,
+      "loss": 0.446,
+      "step": 1206
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8923433405202808e-05,
+      "loss": 0.4437,
+      "step": 1207
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8921316667510367e-05,
+      "loss": 0.4389,
+      "step": 1208
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8919197969530088e-05,
+      "loss": 0.4421,
+      "step": 1209
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8917077311727514e-05,
+      "loss": 0.4585,
+      "step": 1210
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8914954694568618e-05,
+      "loss": 0.4426,
+      "step": 1211
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8912830118519803e-05,
+      "loss": 0.4415,
+      "step": 1212
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8910703584047906e-05,
+      "loss": 0.445,
+      "step": 1213
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8908575091620188e-05,
+      "loss": 0.4345,
+      "step": 1214
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8906444641704354e-05,
+      "loss": 0.4403,
+      "step": 1215
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890431223476852e-05,
+      "loss": 0.4374,
+      "step": 1216
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8902177871281245e-05,
+      "loss": 0.4507,
+      "step": 1217
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8900041551711513e-05,
+      "loss": 0.4379,
+      "step": 1218
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889790327652874e-05,
+      "loss": 0.4565,
+      "step": 1219
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8895763046202772e-05,
+      "loss": 0.454,
+      "step": 1220
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889362086120388e-05,
+      "loss": 0.4452,
+      "step": 1221
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8891476722002774e-05,
+      "loss": 0.4462,
+      "step": 1222
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.888933062907058e-05,
+      "loss": 0.4737,
+      "step": 1223
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.888718258287887e-05,
+      "loss": 0.4528,
+      "step": 1224
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8885032583899625e-05,
+      "loss": 0.4394,
+      "step": 1225
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8882880632605273e-05,
+      "loss": 0.4425,
+      "step": 1226
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8880726729468662e-05,
+      "loss": 0.4205,
+      "step": 1227
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8878570874963073e-05,
+      "loss": 0.4302,
+      "step": 1228
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8876413069562208e-05,
+      "loss": 0.4263,
+      "step": 1229
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.887425331374021e-05,
+      "loss": 0.4718,
+      "step": 1230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8872091607971637e-05,
+      "loss": 0.4383,
+      "step": 1231
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8869927952731486e-05,
+      "loss": 0.4693,
+      "step": 1232
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886776234849518e-05,
+      "loss": 0.4334,
+      "step": 1233
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8865594795738568e-05,
+      "loss": 0.4342,
+      "step": 1234
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.886342529493793e-05,
+      "loss": 0.4541,
+      "step": 1235
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8861253846569966e-05,
+      "loss": 0.4477,
+      "step": 1236
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8859080451111813e-05,
+      "loss": 0.4259,
+      "step": 1237
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8856905109041034e-05,
+      "loss": 0.439,
+      "step": 1238
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8854727820835622e-05,
+      "loss": 0.4366,
+      "step": 1239
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8852548586973986e-05,
+      "loss": 0.4502,
+      "step": 1240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.885036740793498e-05,
+      "loss": 0.4265,
+      "step": 1241
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8848184284197863e-05,
+      "loss": 0.4423,
+      "step": 1242
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8845999216242347e-05,
+      "loss": 0.43,
+      "step": 1243
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8843812204548553e-05,
+      "loss": 0.4156,
+      "step": 1244
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.884162324959704e-05,
+      "loss": 0.446,
+      "step": 1245
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883943235186878e-05,
+      "loss": 0.454,
+      "step": 1246
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883723951184519e-05,
+      "loss": 0.4293,
+      "step": 1247
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.88350447300081e-05,
+      "loss": 0.4268,
+      "step": 1248
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.883284800683977e-05,
+      "loss": 0.4615,
+      "step": 1249
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8830649342822887e-05,
+      "loss": 0.4417,
+      "step": 1250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8828448738440568e-05,
+      "loss": 0.4405,
+      "step": 1251
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8826246194176356e-05,
+      "loss": 0.4456,
+      "step": 1252
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8824041710514208e-05,
+      "loss": 0.447,
+      "step": 1253
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.882183528793853e-05,
+      "loss": 0.4318,
+      "step": 1254
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8819626926934127e-05,
+      "loss": 0.4409,
+      "step": 1255
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8817416627986254e-05,
+      "loss": 0.4441,
+      "step": 1256
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.881520439158058e-05,
+      "loss": 0.4426,
+      "step": 1257
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.88129902182032e-05,
+      "loss": 0.4361,
+      "step": 1258
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8810774108340635e-05,
+      "loss": 0.4266,
+      "step": 1259
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8808556062479835e-05,
+      "loss": 0.4183,
+      "step": 1260
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880633608110817e-05,
+      "loss": 0.4251,
+      "step": 1261
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880411416471344e-05,
+      "loss": 0.452,
+      "step": 1262
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8801890313783867e-05,
+      "loss": 0.441,
+      "step": 1263
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8799664528808104e-05,
+      "loss": 0.4591,
+      "step": 1264
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.879743681027522e-05,
+      "loss": 0.4463,
+      "step": 1265
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8795207158674714e-05,
+      "loss": 0.4236,
+      "step": 1266
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.879297557449651e-05,
+      "loss": 0.4295,
+      "step": 1267
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8790742058230958e-05,
+      "loss": 0.4404,
+      "step": 1268
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8788506610368823e-05,
+      "loss": 0.4284,
+      "step": 1269
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.878626923140131e-05,
+      "loss": 0.4629,
+      "step": 1270
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8784029921820032e-05,
+      "loss": 0.4556,
+      "step": 1271
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8781788682117043e-05,
+      "loss": 0.4183,
+      "step": 1272
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8779545512784807e-05,
+      "loss": 0.4237,
+      "step": 1273
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8777300414316216e-05,
+      "loss": 0.4515,
+      "step": 1274
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8775053387204592e-05,
+      "loss": 0.4331,
+      "step": 1275
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.877280443194367e-05,
+      "loss": 0.4396,
+      "step": 1276
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8770553549027625e-05,
+      "loss": 0.4508,
+      "step": 1277
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.876830073895103e-05,
+      "loss": 0.4552,
+      "step": 1278
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876604600220891e-05,
+      "loss": 0.4444,
+      "step": 1279
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8763789339296694e-05,
+      "loss": 0.4339,
+      "step": 1280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8761530750710243e-05,
+      "loss": 0.4511,
+      "step": 1281
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8759270236945835e-05,
+      "loss": 0.4478,
+      "step": 1282
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8757007798500178e-05,
+      "loss": 0.4433,
+      "step": 1283
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8754743435870394e-05,
+      "loss": 0.443,
+      "step": 1284
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8752477149554042e-05,
+      "loss": 0.4563,
+      "step": 1285
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8750208940049085e-05,
+      "loss": 0.4303,
+      "step": 1286
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8747938807853927e-05,
+      "loss": 0.4228,
+      "step": 1287
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874566675346738e-05,
+      "loss": 0.4517,
+      "step": 1288
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8743392777388688e-05,
+      "loss": 0.45,
+      "step": 1289
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8741116880117508e-05,
+      "loss": 0.4838,
+      "step": 1290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8738839062153934e-05,
+      "loss": 0.4118,
+      "step": 1291
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8736559323998465e-05,
+      "loss": 0.4491,
+      "step": 1292
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.873427766615203e-05,
+      "loss": 0.4473,
+      "step": 1293
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8731994089115984e-05,
+      "loss": 0.4498,
+      "step": 1294
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8729708593392096e-05,
+      "loss": 0.445,
+      "step": 1295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872742117948256e-05,
+      "loss": 0.4596,
+      "step": 1296
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.872513184788999e-05,
+      "loss": 0.4533,
+      "step": 1297
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8722840599117424e-05,
+      "loss": 0.4518,
+      "step": 1298
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8720547433668323e-05,
+      "loss": 0.443,
+      "step": 1299
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.871825235204656e-05,
+      "loss": 0.4728,
+      "step": 1300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8715955354756438e-05,
+      "loss": 0.4614,
+      "step": 1301
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8713656442302678e-05,
+      "loss": 0.444,
+      "step": 1302
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8711355615190423e-05,
+      "loss": 0.4347,
+      "step": 1303
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8709052873925236e-05,
+      "loss": 0.4572,
+      "step": 1304
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8706748219013094e-05,
+      "loss": 0.4547,
+      "step": 1305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.870444165096041e-05,
+      "loss": 0.4631,
+      "step": 1306
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8702133170274e-05,
+      "loss": 0.4234,
+      "step": 1307
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8699822777461114e-05,
+      "loss": 0.4283,
+      "step": 1308
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869751047302941e-05,
+      "loss": 0.4444,
+      "step": 1309
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8695196257486983e-05,
+      "loss": 0.4242,
+      "step": 1310
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8692880131342325e-05,
+      "loss": 0.4437,
+      "step": 1311
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8690562095104368e-05,
+      "loss": 0.4481,
+      "step": 1312
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8688242149282452e-05,
+      "loss": 0.4532,
+      "step": 1313
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8685920294386343e-05,
+      "loss": 0.421,
+      "step": 1314
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8683596530926222e-05,
+      "loss": 0.4529,
+      "step": 1315
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.868127085941269e-05,
+      "loss": 0.4427,
+      "step": 1316
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8678943280356778e-05,
+      "loss": 0.4401,
+      "step": 1317
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867661379426992e-05,
+      "loss": 0.4363,
+      "step": 1318
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867428240166397e-05,
+      "loss": 0.4432,
+      "step": 1319
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867194910305121e-05,
+      "loss": 0.4361,
+      "step": 1320
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866961389894435e-05,
+      "loss": 0.433,
+      "step": 1321
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8667276789856486e-05,
+      "loss": 0.4352,
+      "step": 1322
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8664937776301172e-05,
+      "loss": 0.4405,
+      "step": 1323
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866259685879235e-05,
+      "loss": 0.4383,
+      "step": 1324
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.4221,
+      "step": 1325
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8657909313972085e-05,
+      "loss": 0.4349,
+      "step": 1326
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8655562687690648e-05,
+      "loss": 0.4373,
+      "step": 1327
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8653214159515695e-05,
+      "loss": 0.4409,
+      "step": 1328
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.865086372996328e-05,
+      "loss": 0.4303,
+      "step": 1329
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.864851139954986e-05,
+      "loss": 0.45,
+      "step": 1330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8646157168792315e-05,
+      "loss": 0.4328,
+      "step": 1331
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.864380103820794e-05,
+      "loss": 0.441,
+      "step": 1332
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8641443008314453e-05,
+      "loss": 0.4528,
+      "step": 1333
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8639083079629983e-05,
+      "loss": 0.4542,
+      "step": 1334
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.863672125267308e-05,
+      "loss": 0.4464,
+      "step": 1335
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.863435752796271e-05,
+      "loss": 0.4298,
+      "step": 1336
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8631991906018254e-05,
+      "loss": 0.4339,
+      "step": 1337
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.862962438735951e-05,
+      "loss": 0.4573,
+      "step": 1338
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8627254972506707e-05,
+      "loss": 0.4358,
+      "step": 1339
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8624883661980465e-05,
+      "loss": 0.4348,
+      "step": 1340
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8622510456301843e-05,
+      "loss": 0.4439,
+      "step": 1341
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.86201353559923e-05,
+      "loss": 0.4327,
+      "step": 1342
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861775836157372e-05,
+      "loss": 0.4493,
+      "step": 1343
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861537947356841e-05,
+      "loss": 0.4203,
+      "step": 1344
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861299869249907e-05,
+      "loss": 0.4385,
+      "step": 1345
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8610616018888846e-05,
+      "loss": 0.4424,
+      "step": 1346
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8608231453261276e-05,
+      "loss": 0.4741,
+      "step": 1347
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8605844996140322e-05,
+      "loss": 0.4467,
+      "step": 1348
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8603456648050368e-05,
+      "loss": 0.4378,
+      "step": 1349
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8601066409516206e-05,
+      "loss": 0.4683,
+      "step": 1350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859867428106304e-05,
+      "loss": 0.4563,
+      "step": 1351
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8596280263216497e-05,
+      "loss": 0.4357,
+      "step": 1352
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8593884356502616e-05,
+      "loss": 0.4462,
+      "step": 1353
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8591486561447856e-05,
+      "loss": 0.4545,
+      "step": 1354
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.858908687857908e-05,
+      "loss": 0.4628,
+      "step": 1355
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8586685308423577e-05,
+      "loss": 0.4586,
+      "step": 1356
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.858428185150904e-05,
+      "loss": 0.4671,
+      "step": 1357
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8581876508363588e-05,
+      "loss": 0.4295,
+      "step": 1358
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.857946927951575e-05,
+      "loss": 0.4552,
+      "step": 1359
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8577060165494462e-05,
+      "loss": 0.4461,
+      "step": 1360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8574649166829082e-05,
+      "loss": 0.4618,
+      "step": 1361
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.857223628404939e-05,
+      "loss": 0.4355,
+      "step": 1362
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8569821517685555e-05,
+      "loss": 0.4364,
+      "step": 1363
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8567404868268187e-05,
+      "loss": 0.4341,
+      "step": 1364
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8564986336328294e-05,
+      "loss": 0.4363,
+      "step": 1365
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8562565922397302e-05,
+      "loss": 0.4327,
+      "step": 1366
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8560143627007052e-05,
+      "loss": 0.4417,
+      "step": 1367
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8557719450689798e-05,
+      "loss": 0.4391,
+      "step": 1368
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.85552933939782e-05,
+      "loss": 0.4376,
+      "step": 1369
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8552865457405348e-05,
+      "loss": 0.4323,
+      "step": 1370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.855043564150472e-05,
+      "loss": 0.4397,
+      "step": 1371
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8548003946810234e-05,
+      "loss": 0.4368,
+      "step": 1372
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8545570373856205e-05,
+      "loss": 0.4456,
+      "step": 1373
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8543134923177355e-05,
+      "loss": 0.4669,
+      "step": 1374
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8540697595308842e-05,
+      "loss": 0.4252,
+      "step": 1375
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.853825839078621e-05,
+      "loss": 0.4402,
+      "step": 1376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8535817310145433e-05,
+      "loss": 0.4431,
+      "step": 1377
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.853337435392289e-05,
+      "loss": 0.4493,
+      "step": 1378
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8530929522655373e-05,
+      "loss": 0.4169,
+      "step": 1379
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8528482816880094e-05,
+      "loss": 0.4388,
+      "step": 1380
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8526034237134658e-05,
+      "loss": 0.4401,
+      "step": 1381
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.85235837839571e-05,
+      "loss": 0.4245,
+      "step": 1382
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8521131457885858e-05,
+      "loss": 0.4334,
+      "step": 1383
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8518677259459785e-05,
+      "loss": 0.4345,
+      "step": 1384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8516221189218143e-05,
+      "loss": 0.4562,
+      "step": 1385
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8513763247700607e-05,
+      "loss": 0.4263,
+      "step": 1386
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.851130343544726e-05,
+      "loss": 0.4152,
+      "step": 1387
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8508841752998602e-05,
+      "loss": 0.4353,
+      "step": 1388
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8506378200895537e-05,
+      "loss": 0.431,
+      "step": 1389
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8503912779679388e-05,
+      "loss": 0.4676,
+      "step": 1390
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8501445489891875e-05,
+      "loss": 0.4353,
+      "step": 1391
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8498976332075148e-05,
+      "loss": 0.4286,
+      "step": 1392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8496505306771754e-05,
+      "loss": 0.4438,
+      "step": 1393
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8494032414524647e-05,
+      "loss": 0.4641,
+      "step": 1394
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.84915576558772e-05,
+      "loss": 0.4475,
+      "step": 1395
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.84890810313732e-05,
+      "loss": 0.4367,
+      "step": 1396
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848660254155683e-05,
+      "loss": 0.459,
+      "step": 1397
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8484122186972696e-05,
+      "loss": 0.4245,
+      "step": 1398
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8481639968165807e-05,
+      "loss": 0.4249,
+      "step": 1399
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8479155885681585e-05,
+      "loss": 0.4465,
+      "step": 1400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847666994006585e-05,
+      "loss": 0.4698,
+      "step": 1401
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847418213186485e-05,
+      "loss": 0.4568,
+      "step": 1402
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8471692461625227e-05,
+      "loss": 0.4286,
+      "step": 1403
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8469200929894043e-05,
+      "loss": 0.44,
+      "step": 1404
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.846670753721876e-05,
+      "loss": 0.4298,
+      "step": 1405
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8464212284147263e-05,
+      "loss": 0.4399,
+      "step": 1406
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8461715171227823e-05,
+      "loss": 0.4493,
+      "step": 1407
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845921619900914e-05,
+      "loss": 0.4349,
+      "step": 1408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8456715368040308e-05,
+      "loss": 0.4477,
+      "step": 1409
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8454212678870845e-05,
+      "loss": 0.4514,
+      "step": 1410
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.845170813205067e-05,
+      "loss": 0.4293,
+      "step": 1411
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.84492017281301e-05,
+      "loss": 0.4438,
+      "step": 1412
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8446693467659872e-05,
+      "loss": 0.4363,
+      "step": 1413
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844418335119113e-05,
+      "loss": 0.441,
+      "step": 1414
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.844167137927543e-05,
+      "loss": 0.447,
+      "step": 1415
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8439157552464718e-05,
+      "loss": 0.4464,
+      "step": 1416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.843664187131136e-05,
+      "loss": 0.4302,
+      "step": 1417
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8434124336368136e-05,
+      "loss": 0.4371,
+      "step": 1418
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8431604948188226e-05,
+      "loss": 0.4544,
+      "step": 1419
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842908370732521e-05,
+      "loss": 0.4432,
+      "step": 1420
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8426560614333086e-05,
+      "loss": 0.425,
+      "step": 1421
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842403566976626e-05,
+      "loss": 0.4215,
+      "step": 1422
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8421508874179528e-05,
+      "loss": 0.4365,
+      "step": 1423
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8418980228128114e-05,
+      "loss": 0.4339,
+      "step": 1424
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.841644973216764e-05,
+      "loss": 0.4722,
+      "step": 1425
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8413917386854126e-05,
+      "loss": 0.4706,
+      "step": 1426
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8411383192744016e-05,
+      "loss": 0.4519,
+      "step": 1427
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8408847150394143e-05,
+      "loss": 0.4399,
+      "step": 1428
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8406309260361758e-05,
+      "loss": 0.4439,
+      "step": 1429
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.840376952320451e-05,
+      "loss": 0.4402,
+      "step": 1430
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.840122793948046e-05,
+      "loss": 0.4454,
+      "step": 1431
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8398684509748073e-05,
+      "loss": 0.4393,
+      "step": 1432
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8396139234566217e-05,
+      "loss": 0.437,
+      "step": 1433
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8393592114494166e-05,
+      "loss": 0.4498,
+      "step": 1434
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8391043150091604e-05,
+      "loss": 0.4347,
+      "step": 1435
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8388492341918612e-05,
+      "loss": 0.4354,
+      "step": 1436
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8385939690535685e-05,
+      "loss": 0.4694,
+      "step": 1437
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.838338519650372e-05,
+      "loss": 0.4153,
+      "step": 1438
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8380828860384015e-05,
+      "loss": 0.4493,
+      "step": 1439
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8378270682738277e-05,
+      "loss": 0.4296,
+      "step": 1440
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8375710664128615e-05,
+      "loss": 0.4291,
+      "step": 1441
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837314880511755e-05,
+      "loss": 0.437,
+      "step": 1442
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.837058510626799e-05,
+      "loss": 0.4625,
+      "step": 1443
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.836801956814327e-05,
+      "loss": 0.4474,
+      "step": 1444
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.836545219130711e-05,
+      "loss": 0.4517,
+      "step": 1445
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8362882976323644e-05,
+      "loss": 0.4391,
+      "step": 1446
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8360311923757406e-05,
+      "loss": 0.4383,
+      "step": 1447
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835773903417334e-05,
+      "loss": 0.4459,
+      "step": 1448
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8355164308136788e-05,
+      "loss": 0.466,
+      "step": 1449
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8352587746213494e-05,
+      "loss": 0.4412,
+      "step": 1450
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8350009348969605e-05,
+      "loss": 0.4473,
+      "step": 1451
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834742911697168e-05,
+      "loss": 0.4409,
+      "step": 1452
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8344847050786675e-05,
+      "loss": 0.4117,
+      "step": 1453
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.834226315098195e-05,
+      "loss": 0.4474,
+      "step": 1454
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8339677418125262e-05,
+      "loss": 0.4448,
+      "step": 1455
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8337089852784776e-05,
+      "loss": 0.4449,
+      "step": 1456
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.833450045552907e-05,
+      "loss": 0.4464,
+      "step": 1457
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8331909226927104e-05,
+      "loss": 0.4498,
+      "step": 1458
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.832931616754825e-05,
+      "loss": 0.452,
+      "step": 1459
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8326721277962294e-05,
+      "loss": 0.4512,
+      "step": 1460
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8324124558739404e-05,
+      "loss": 0.4353,
+      "step": 1461
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8321526010450156e-05,
+      "loss": 0.4393,
+      "step": 1462
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.831892563366554e-05,
+      "loss": 0.4249,
+      "step": 1463
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8316323428956938e-05,
+      "loss": 0.44,
+      "step": 1464
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8313719396896125e-05,
+      "loss": 0.4435,
+      "step": 1465
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8311113538055298e-05,
+      "loss": 0.4564,
+      "step": 1466
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8308505853007036e-05,
+      "loss": 0.4459,
+      "step": 1467
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8305896342324337e-05,
+      "loss": 0.4658,
+      "step": 1468
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.830328500658058e-05,
+      "loss": 0.4393,
+      "step": 1469
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8300671846349564e-05,
+      "loss": 0.4458,
+      "step": 1470
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8298056862205476e-05,
+      "loss": 0.4553,
+      "step": 1471
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.829544005472291e-05,
+      "loss": 0.4443,
+      "step": 1472
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8292821424476863e-05,
+      "loss": 0.4533,
+      "step": 1473
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.829020097204272e-05,
+      "loss": 0.4595,
+      "step": 1474
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8287578697996285e-05,
+      "loss": 0.439,
+      "step": 1475
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8284954602913745e-05,
+      "loss": 0.4363,
+      "step": 1476
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8282328687371698e-05,
+      "loss": 0.4541,
+      "step": 1477
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8279700951947136e-05,
+      "loss": 0.4233,
+      "step": 1478
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8277071397217455e-05,
+      "loss": 0.4491,
+      "step": 1479
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8274440023760452e-05,
+      "loss": 0.4377,
+      "step": 1480
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8271806832154312e-05,
+      "loss": 0.4471,
+      "step": 1481
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8269171822977637e-05,
+      "loss": 0.429,
+      "step": 1482
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8266534996809414e-05,
+      "loss": 0.4404,
+      "step": 1483
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8263896354229042e-05,
+      "loss": 0.4416,
+      "step": 1484
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8261255895816307e-05,
+      "loss": 0.44,
+      "step": 1485
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8258613622151396e-05,
+      "loss": 0.4416,
+      "step": 1486
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.82559695338149e-05,
+      "loss": 0.441,
+      "step": 1487
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8253323631387808e-05,
+      "loss": 0.4361,
+      "step": 1488
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.825067591545151e-05,
+      "loss": 0.4266,
+      "step": 1489
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8248026386587782e-05,
+      "loss": 0.4471,
+      "step": 1490
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8245375045378812e-05,
+      "loss": 0.4336,
+      "step": 1491
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8242721892407186e-05,
+      "loss": 0.4801,
+      "step": 1492
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8240066928255874e-05,
+      "loss": 0.4226,
+      "step": 1493
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.823741015350826e-05,
+      "loss": 0.4399,
+      "step": 1494
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8234751568748112e-05,
+      "loss": 0.4264,
+      "step": 1495
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.823209117455961e-05,
+      "loss": 0.4403,
+      "step": 1496
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.822942897152732e-05,
+      "loss": 0.4385,
+      "step": 1497
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8226764960236215e-05,
+      "loss": 0.4389,
+      "step": 1498
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.822409914127166e-05,
+      "loss": 0.453,
+      "step": 1499
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.822143151521941e-05,
+      "loss": 0.4551,
+      "step": 1500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.821876208266563e-05,
+      "loss": 0.4477,
+      "step": 1501
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8216090844196877e-05,
+      "loss": 0.4401,
+      "step": 1502
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8213417800400105e-05,
+      "loss": 0.4292,
+      "step": 1503
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8210742951862656e-05,
+      "loss": 0.4281,
+      "step": 1504
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8208066299172287e-05,
+      "loss": 0.4608,
+      "step": 1505
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8205387842917138e-05,
+      "loss": 0.5,
+      "step": 1506
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8202707583685742e-05,
+      "loss": 0.4232,
+      "step": 1507
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8200025522067043e-05,
+      "loss": 0.4527,
+      "step": 1508
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.819734165865037e-05,
+      "loss": 0.4559,
+      "step": 1509
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8194655994025446e-05,
+      "loss": 0.4567,
+      "step": 1510
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8191968528782398e-05,
+      "loss": 0.4501,
+      "step": 1511
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8189279263511745e-05,
+      "loss": 0.4517,
+      "step": 1512
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.81865881988044e-05,
+      "loss": 0.4574,
+      "step": 1513
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8183895335251673e-05,
+      "loss": 0.4292,
+      "step": 1514
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.818120067344527e-05,
+      "loss": 0.4314,
+      "step": 1515
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.817850421397729e-05,
+      "loss": 0.4504,
+      "step": 1516
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.817580595744023e-05,
+      "loss": 0.441,
+      "step": 1517
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8173105904426976e-05,
+      "loss": 0.4232,
+      "step": 1518
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8170404055530818e-05,
+      "loss": 0.4434,
+      "step": 1519
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816770041134543e-05,
+      "loss": 0.4406,
+      "step": 1520
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816499497246489e-05,
+      "loss": 0.4494,
+      "step": 1521
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8162287739483665e-05,
+      "loss": 0.4498,
+      "step": 1522
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8159578712996615e-05,
+      "loss": 0.4485,
+      "step": 1523
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8156867893599002e-05,
+      "loss": 0.457,
+      "step": 1524
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.815415528188647e-05,
+      "loss": 0.4342,
+      "step": 1525
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8151440878455067e-05,
+      "loss": 0.4393,
+      "step": 1526
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8148724683901232e-05,
+      "loss": 0.463,
+      "step": 1527
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.814600669882179e-05,
+      "loss": 0.4419,
+      "step": 1528
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8143286923813975e-05,
+      "loss": 0.4629,
+      "step": 1529
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.81405653594754e-05,
+      "loss": 0.482,
+      "step": 1530
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.813784200640408e-05,
+      "loss": 0.451,
+      "step": 1531
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.813511686519841e-05,
+      "loss": 0.4393,
+      "step": 1532
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8132389936457203e-05,
+      "loss": 0.4296,
+      "step": 1533
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8129661220779636e-05,
+      "loss": 0.4352,
+      "step": 1534
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8126930718765298e-05,
+      "loss": 0.4379,
+      "step": 1535
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8124198431014158e-05,
+      "loss": 0.4455,
+      "step": 1536
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812146435812659e-05,
+      "loss": 0.4489,
+      "step": 1537
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8118728500703357e-05,
+      "loss": 0.4402,
+      "step": 1538
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8115990859345605e-05,
+      "loss": 0.4529,
+      "step": 1539
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8113251434654876e-05,
+      "loss": 0.4231,
+      "step": 1540
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8110510227233115e-05,
+      "loss": 0.4543,
+      "step": 1541
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.810776723768264e-05,
+      "loss": 0.451,
+      "step": 1542
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8105022466606175e-05,
+      "loss": 0.4438,
+      "step": 1543
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.810227591460683e-05,
+      "loss": 0.4394,
+      "step": 1544
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8099527582288106e-05,
+      "loss": 0.4438,
+      "step": 1545
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8096777470253898e-05,
+      "loss": 0.443,
+      "step": 1546
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.809402557910849e-05,
+      "loss": 0.441,
+      "step": 1547
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8091271909456556e-05,
+      "loss": 0.4407,
+      "step": 1548
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.808851646190316e-05,
+      "loss": 0.4473,
+      "step": 1549
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8085759237053765e-05,
+      "loss": 0.4391,
+      "step": 1550
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8083000235514214e-05,
+      "loss": 0.4204,
+      "step": 1551
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.808023945789074e-05,
+      "loss": 0.4537,
+      "step": 1552
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8077476904789985e-05,
+      "loss": 0.4141,
+      "step": 1553
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8074712576818952e-05,
+      "loss": 0.4471,
+      "step": 1554
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.807194647458505e-05,
+      "loss": 0.4208,
+      "step": 1555
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.806917859869609e-05,
+      "loss": 0.4479,
+      "step": 1556
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8066408949760245e-05,
+      "loss": 0.4427,
+      "step": 1557
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.80636375283861e-05,
+      "loss": 0.4337,
+      "step": 1558
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.806086433518262e-05,
+      "loss": 0.4694,
+      "step": 1559
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.805808937075916e-05,
+      "loss": 0.4377,
+      "step": 1560
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8055312635725467e-05,
+      "loss": 0.4371,
+      "step": 1561
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.805253413069167e-05,
+      "loss": 0.4442,
+      "step": 1562
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.80497538562683e-05,
+      "loss": 0.4257,
+      "step": 1563
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8046971813066265e-05,
+      "loss": 0.4214,
+      "step": 1564
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8044188001696857e-05,
+      "loss": 0.4497,
+      "step": 1565
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.804140242277178e-05,
+      "loss": 0.4448,
+      "step": 1566
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8038615076903102e-05,
+      "loss": 0.4557,
+      "step": 1567
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8035825964703296e-05,
+      "loss": 0.4699,
+      "step": 1568
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8033035086785208e-05,
+      "loss": 0.4652,
+      "step": 1569
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.803024244376208e-05,
+      "loss": 0.4504,
+      "step": 1570
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.802744803624755e-05,
+      "loss": 0.4521,
+      "step": 1571
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8024651864855625e-05,
+      "loss": 0.4576,
+      "step": 1572
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8021853930200716e-05,
+      "loss": 0.4479,
+      "step": 1573
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8019054232897614e-05,
+      "loss": 0.4581,
+      "step": 1574
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8016252773561497e-05,
+      "loss": 0.4526,
+      "step": 1575
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.801344955280794e-05,
+      "loss": 0.4566,
+      "step": 1576
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8010644571252884e-05,
+      "loss": 0.4536,
+      "step": 1577
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.800783782951268e-05,
+      "loss": 0.4407,
+      "step": 1578
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8005029328204048e-05,
+      "loss": 0.4417,
+      "step": 1579
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8002219067944108e-05,
+      "loss": 0.4486,
+      "step": 1580
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7999407049350357e-05,
+      "loss": 0.4532,
+      "step": 1581
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7996593273040685e-05,
+      "loss": 0.4395,
+      "step": 1582
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7993777739633365e-05,
+      "loss": 0.4373,
+      "step": 1583
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7990960449747056e-05,
+      "loss": 0.4447,
+      "step": 1584
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7988141404000798e-05,
+      "loss": 0.4431,
+      "step": 1585
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7985320603014028e-05,
+      "loss": 0.4237,
+      "step": 1586
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.798249804740656e-05,
+      "loss": 0.4223,
+      "step": 1587
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7979673737798605e-05,
+      "loss": 0.4432,
+      "step": 1588
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7976847674810738e-05,
+      "loss": 0.4452,
+      "step": 1589
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.797401985906394e-05,
+      "loss": 0.4668,
+      "step": 1590
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.797119029117956e-05,
+      "loss": 0.4505,
+      "step": 1591
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7968358971779353e-05,
+      "loss": 0.4476,
+      "step": 1592
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.796552590148544e-05,
+      "loss": 0.4501,
+      "step": 1593
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.796269108092034e-05,
+      "loss": 0.4502,
+      "step": 1594
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7959854510706943e-05,
+      "loss": 0.4351,
+      "step": 1595
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7957016191468533e-05,
+      "loss": 0.4508,
+      "step": 1596
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7954176123828776e-05,
+      "loss": 0.4401,
+      "step": 1597
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7951334308411725e-05,
+      "loss": 0.4386,
+      "step": 1598
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.794849074584181e-05,
+      "loss": 0.4294,
+      "step": 1599
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.794564543674385e-05,
+      "loss": 0.4346,
+      "step": 1600
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7942798381743053e-05,
+      "loss": 0.4473,
+      "step": 1601
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7939949581464996e-05,
+      "loss": 0.4362,
+      "step": 1602
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7937099036535657e-05,
+      "loss": 0.4646,
+      "step": 1603
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.793424674758138e-05,
+      "loss": 0.4591,
+      "step": 1604
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7931392715228905e-05,
+      "loss": 0.4387,
+      "step": 1605
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7928536940105352e-05,
+      "loss": 0.4387,
+      "step": 1606
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7925679422838216e-05,
+      "loss": 0.4396,
+      "step": 1607
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7922820164055393e-05,
+      "loss": 0.4378,
+      "step": 1608
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7919959164385137e-05,
+      "loss": 0.4195,
+      "step": 1609
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7917096424456105e-05,
+      "loss": 0.4487,
+      "step": 1610
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7914231944897335e-05,
+      "loss": 0.4396,
+      "step": 1611
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7911365726338227e-05,
+      "loss": 0.4498,
+      "step": 1612
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7908497769408586e-05,
+      "loss": 0.4403,
+      "step": 1613
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.790562807473859e-05,
+      "loss": 0.4165,
+      "step": 1614
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.79027566429588e-05,
+      "loss": 0.427,
+      "step": 1615
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7899883474700155e-05,
+      "loss": 0.4353,
+      "step": 1616
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.789700857059398e-05,
+      "loss": 0.4331,
+      "step": 1617
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.789413193127198e-05,
+      "loss": 0.4435,
+      "step": 1618
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.789125355736624e-05,
+      "loss": 0.4329,
+      "step": 1619
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.788837344950923e-05,
+      "loss": 0.4494,
+      "step": 1620
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7885491608333796e-05,
+      "loss": 0.4315,
+      "step": 1621
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.788260803447317e-05,
+      "loss": 0.4259,
+      "step": 1622
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7879722728560954e-05,
+      "loss": 0.4325,
+      "step": 1623
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.787683569123115e-05,
+      "loss": 0.453,
+      "step": 1624
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7873946923118123e-05,
+      "loss": 0.4363,
+      "step": 1625
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.787105642485662e-05,
+      "loss": 0.4431,
+      "step": 1626
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7868164197081783e-05,
+      "loss": 0.4331,
+      "step": 1627
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7865270240429114e-05,
+      "loss": 0.4357,
+      "step": 1628
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7862374555534506e-05,
+      "loss": 0.4535,
+      "step": 1629
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7859477143034233e-05,
+      "loss": 0.4352,
+      "step": 1630
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7856578003564946e-05,
+      "loss": 0.4338,
+      "step": 1631
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7853677137763673e-05,
+      "loss": 0.4308,
+      "step": 1632
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7850774546267825e-05,
+      "loss": 0.4342,
+      "step": 1633
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7847870229715193e-05,
+      "loss": 0.4298,
+      "step": 1634
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7844964188743932e-05,
+      "loss": 0.4349,
+      "step": 1635
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7842056423992607e-05,
+      "loss": 0.444,
+      "step": 1636
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7839146936100135e-05,
+      "loss": 0.4419,
+      "step": 1637
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7836235725705814e-05,
+      "loss": 0.463,
+      "step": 1638
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7833322793449336e-05,
+      "loss": 0.4604,
+      "step": 1639
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7830408139970758e-05,
+      "loss": 0.4544,
+      "step": 1640
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7827491765910515e-05,
+      "loss": 0.4103,
+      "step": 1641
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7824573671909432e-05,
+      "loss": 0.4501,
+      "step": 1642
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.78216538586087e-05,
+      "loss": 0.4413,
+      "step": 1643
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.781873232664989e-05,
+      "loss": 0.4538,
+      "step": 1644
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7815809076674958e-05,
+      "loss": 0.4298,
+      "step": 1645
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7812884109326227e-05,
+      "loss": 0.4239,
+      "step": 1646
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7809957425246406e-05,
+      "loss": 0.4422,
+      "step": 1647
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7807029025078572e-05,
+      "loss": 0.4507,
+      "step": 1648
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7804098909466193e-05,
+      "loss": 0.4761,
+      "step": 1649
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7801167079053095e-05,
+      "loss": 0.4266,
+      "step": 1650
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.77982335344835e-05,
+      "loss": 0.4376,
+      "step": 1651
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7795298276402e-05,
+      "loss": 0.4291,
+      "step": 1652
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.779236130545355e-05,
+      "loss": 0.4449,
+      "step": 1653
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.77894226222835e-05,
+      "loss": 0.4293,
+      "step": 1654
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7786482227537574e-05,
+      "loss": 0.4231,
+      "step": 1655
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7783540121861857e-05,
+      "loss": 0.4255,
+      "step": 1656
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7780596305902828e-05,
+      "loss": 0.4731,
+      "step": 1657
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7777650780307328e-05,
+      "loss": 0.4313,
+      "step": 1658
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7774703545722585e-05,
+      "loss": 0.4382,
+      "step": 1659
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7771754602796197e-05,
+      "loss": 0.4572,
+      "step": 1660
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7768803952176133e-05,
+      "loss": 0.4156,
+      "step": 1661
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7765851594510743e-05,
+      "loss": 0.444,
+      "step": 1662
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7762897530448753e-05,
+      "loss": 0.4366,
+      "step": 1663
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7759941760639264e-05,
+      "loss": 0.4596,
+      "step": 1664
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.775698428573174e-05,
+      "loss": 0.4525,
+      "step": 1665
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.775402510637604e-05,
+      "loss": 0.4341,
+      "step": 1666
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7751064223222382e-05,
+      "loss": 0.4198,
+      "step": 1667
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7748101636921366e-05,
+      "loss": 0.4387,
+      "step": 1668
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7745137348123957e-05,
+      "loss": 0.4306,
+      "step": 1669
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7742171357481506e-05,
+      "loss": 0.4418,
+      "step": 1670
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.773920366564573e-05,
+      "loss": 0.422,
+      "step": 1671
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7736234273268724e-05,
+      "loss": 0.4395,
+      "step": 1672
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7733263181002956e-05,
+      "loss": 0.4653,
+      "step": 1673
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.773029038950126e-05,
+      "loss": 0.4841,
+      "step": 1674
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7727315899416856e-05,
+      "loss": 0.4244,
+      "step": 1675
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.772433971140333e-05,
+      "loss": 0.442,
+      "step": 1676
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.772136182611464e-05,
+      "loss": 0.4249,
+      "step": 1677
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7718382244205122e-05,
+      "loss": 0.4533,
+      "step": 1678
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7715400966329484e-05,
+      "loss": 0.448,
+      "step": 1679
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7712417993142793e-05,
+      "loss": 0.4361,
+      "step": 1680
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.770943332530051e-05,
+      "loss": 0.4527,
+      "step": 1681
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.770644696345846e-05,
+      "loss": 0.4654,
+      "step": 1682
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7703458908272832e-05,
+      "loss": 0.4576,
+      "step": 1683
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7700469160400196e-05,
+      "loss": 0.4451,
+      "step": 1684
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7697477720497492e-05,
+      "loss": 0.4419,
+      "step": 1685
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.769448458922203e-05,
+      "loss": 0.4276,
+      "step": 1686
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.76914897672315e-05,
+      "loss": 0.4506,
+      "step": 1687
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.768849325518395e-05,
+      "loss": 0.4183,
+      "step": 1688
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.76854950537378e-05,
+      "loss": 0.4441,
+      "step": 1689
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.768249516355186e-05,
+      "loss": 0.4444,
+      "step": 1690
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7679493585285292e-05,
+      "loss": 0.4393,
+      "step": 1691
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7676490319597635e-05,
+      "loss": 0.4546,
+      "step": 1692
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7673485367148798e-05,
+      "loss": 0.4567,
+      "step": 1693
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7670478728599065e-05,
+      "loss": 0.4398,
+      "step": 1694
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7667470404609083e-05,
+      "loss": 0.4516,
+      "step": 1695
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7664460395839876e-05,
+      "loss": 0.4329,
+      "step": 1696
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7661448702952834e-05,
+      "loss": 0.437,
+      "step": 1697
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.765843532660972e-05,
+      "loss": 0.4595,
+      "step": 1698
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7655420267472668e-05,
+      "loss": 0.4391,
+      "step": 1699
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7652403526204174e-05,
+      "loss": 0.4718,
+      "step": 1700
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.764938510346711e-05,
+      "loss": 0.4592,
+      "step": 1701
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7646364999924718e-05,
+      "loss": 0.4445,
+      "step": 1702
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.764334321624061e-05,
+      "loss": 0.438,
+      "step": 1703
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7640319753078755e-05,
+      "loss": 0.4266,
+      "step": 1704
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7637294611103516e-05,
+      "loss": 0.4515,
+      "step": 1705
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7634267790979598e-05,
+      "loss": 0.4259,
+      "step": 1706
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.763123929337209e-05,
+      "loss": 0.444,
+      "step": 1707
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7628209118946448e-05,
+      "loss": 0.4397,
+      "step": 1708
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.762517726836849e-05,
+      "loss": 0.4381,
+      "step": 1709
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7622143742304414e-05,
+      "loss": 0.434,
+      "step": 1710
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7619108541420774e-05,
+      "loss": 0.4284,
+      "step": 1711
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7616071666384495e-05,
+      "loss": 0.433,
+      "step": 1712
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.761303311786288e-05,
+      "loss": 0.4332,
+      "step": 1713
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7609992896523582e-05,
+      "loss": 0.4354,
+      "step": 1714
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760695100303464e-05,
+      "loss": 0.444,
+      "step": 1715
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7603907438064446e-05,
+      "loss": 0.435,
+      "step": 1716
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7600862202281763e-05,
+      "loss": 0.4327,
+      "step": 1717
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7597815296355727e-05,
+      "loss": 0.4231,
+      "step": 1718
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759476672095584e-05,
+      "loss": 0.4478,
+      "step": 1719
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7591716476751963e-05,
+      "loss": 0.4459,
+      "step": 1720
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7588664564414327e-05,
+      "loss": 0.4524,
+      "step": 1721
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7585610984613538e-05,
+      "loss": 0.4648,
+      "step": 1722
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7582555738020556e-05,
+      "loss": 0.4337,
+      "step": 1723
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.757949882530671e-05,
+      "loss": 0.4418,
+      "step": 1724
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.75764402471437e-05,
+      "loss": 0.4683,
+      "step": 1725
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.75733800042036e-05,
+      "loss": 0.4523,
+      "step": 1726
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7570318097158825e-05,
+      "loss": 0.4302,
+      "step": 1727
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.756725452668218e-05,
+      "loss": 0.4371,
+      "step": 1728
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.756418929344682e-05,
+      "loss": 0.4477,
+      "step": 1729
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7561122398126272e-05,
+      "loss": 0.4316,
+      "step": 1730
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.755805384139443e-05,
+      "loss": 0.4308,
+      "step": 1731
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7554983623925547e-05,
+      "loss": 0.4454,
+      "step": 1732
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.755191174639425e-05,
+      "loss": 0.4301,
+      "step": 1733
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7548838209475515e-05,
+      "loss": 0.4579,
+      "step": 1734
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7545763013844702e-05,
+      "loss": 0.4283,
+      "step": 1735
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.754268616017752e-05,
+      "loss": 0.4298,
+      "step": 1736
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7539607649150055e-05,
+      "loss": 0.4489,
+      "step": 1737
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.753652748143875e-05,
+      "loss": 0.4302,
+      "step": 1738
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7533445657720403e-05,
+      "loss": 0.4388,
+      "step": 1739
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7530362178672195e-05,
+      "loss": 0.4237,
+      "step": 1740
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7527277044971654e-05,
+      "loss": 0.4389,
+      "step": 1741
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7524190257296686e-05,
+      "loss": 0.4693,
+      "step": 1742
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.752110181632555e-05,
+      "loss": 0.473,
+      "step": 1743
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7518011722736874e-05,
+      "loss": 0.4522,
+      "step": 1744
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7514919977209642e-05,
+      "loss": 0.4334,
+      "step": 1745
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7511826580423207e-05,
+      "loss": 0.4533,
+      "step": 1746
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7508731533057287e-05,
+      "loss": 0.4286,
+      "step": 1747
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.750563483579196e-05,
+      "loss": 0.4378,
+      "step": 1748
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7502536489307656e-05,
+      "loss": 0.4686,
+      "step": 1749
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7499436494285188e-05,
+      "loss": 0.4464,
+      "step": 1750
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.749633485140571e-05,
+      "loss": 0.4554,
+      "step": 1751
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7493231561350762e-05,
+      "loss": 0.4301,
+      "step": 1752
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.749012662480222e-05,
+      "loss": 0.453,
+      "step": 1753
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.748702004244234e-05,
+      "loss": 0.4414,
+      "step": 1754
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7483911814953733e-05,
+      "loss": 0.4366,
+      "step": 1755
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.748080194301937e-05,
+      "loss": 0.4494,
+      "step": 1756
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.747769042732259e-05,
+      "loss": 0.4331,
+      "step": 1757
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7474577268547084e-05,
+      "loss": 0.451,
+      "step": 1758
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7471462467376913e-05,
+      "loss": 0.4262,
+      "step": 1759
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7468346024496487e-05,
+      "loss": 0.436,
+      "step": 1760
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7465227940590596e-05,
+      "loss": 0.437,
+      "step": 1761
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.746210821634437e-05,
+      "loss": 0.4511,
+      "step": 1762
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.745898685244331e-05,
+      "loss": 0.4422,
+      "step": 1763
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7455863849573283e-05,
+      "loss": 0.4334,
+      "step": 1764
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7452739208420497e-05,
+      "loss": 0.4276,
+      "step": 1765
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7449612929671537e-05,
+      "loss": 0.4471,
+      "step": 1766
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.744648501401335e-05,
+      "loss": 0.446,
+      "step": 1767
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7443355462133224e-05,
+      "loss": 0.4152,
+      "step": 1768
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.744022427471882e-05,
+      "loss": 0.4347,
+      "step": 1769
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7437091452458165e-05,
+      "loss": 0.4395,
+      "step": 1770
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7433956996039628e-05,
+      "loss": 0.4229,
+      "step": 1771
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7430820906151948e-05,
+      "loss": 0.4342,
+      "step": 1772
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.742768318348422e-05,
+      "loss": 0.4586,
+      "step": 1773
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.74245438287259e-05,
+      "loss": 0.4421,
+      "step": 1774
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7421402842566798e-05,
+      "loss": 0.4169,
+      "step": 1775
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7418260225697085e-05,
+      "loss": 0.451,
+      "step": 1776
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7415115978807296e-05,
+      "loss": 0.4329,
+      "step": 1777
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.741197010258831e-05,
+      "loss": 0.4434,
+      "step": 1778
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7408822597731383e-05,
+      "loss": 0.4358,
+      "step": 1779
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.740567346492811e-05,
+      "loss": 0.4368,
+      "step": 1780
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.740252270487046e-05,
+      "loss": 0.4345,
+      "step": 1781
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7399370318250745e-05,
+      "loss": 0.4594,
+      "step": 1782
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.739621630576165e-05,
+      "loss": 0.4376,
+      "step": 1783
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7393060668096195e-05,
+      "loss": 0.4411,
+      "step": 1784
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7389903405947783e-05,
+      "loss": 0.4276,
+      "step": 1785
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7386744520010156e-05,
+      "loss": 0.4335,
+      "step": 1786
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7383584010977422e-05,
+      "loss": 0.453,
+      "step": 1787
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7380421879544038e-05,
+      "loss": 0.4344,
+      "step": 1788
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7377258126404823e-05,
+      "loss": 0.4428,
+      "step": 1789
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7374092752254953e-05,
+      "loss": 0.4455,
+      "step": 1790
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7370925757789957e-05,
+      "loss": 0.4396,
+      "step": 1791
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.736775714370572e-05,
+      "loss": 0.4618,
+      "step": 1792
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7364586910698486e-05,
+      "loss": 0.4638,
+      "step": 1793
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.736141505946485e-05,
+      "loss": 0.4244,
+      "step": 1794
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7358241590701768e-05,
+      "loss": 0.4254,
+      "step": 1795
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.735506650510655e-05,
+      "loss": 0.4482,
+      "step": 1796
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7351889803376854e-05,
+      "loss": 0.454,
+      "step": 1797
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7348711486210704e-05,
+      "loss": 0.4722,
+      "step": 1798
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7345531554306472e-05,
+      "loss": 0.4302,
+      "step": 1799
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.734235000836289e-05,
+      "loss": 0.4531,
+      "step": 1800
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7339166849079035e-05,
+      "loss": 0.4336,
+      "step": 1801
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7335982077154353e-05,
+      "loss": 0.4473,
+      "step": 1802
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7332795693288635e-05,
+      "loss": 0.4249,
+      "step": 1803
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.732960769818202e-05,
+      "loss": 0.4307,
+      "step": 1804
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.732641809253502e-05,
+      "loss": 0.4428,
+      "step": 1805
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.732322687704848e-05,
+      "loss": 0.4149,
+      "step": 1806
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.732003405242361e-05,
+      "loss": 0.4639,
+      "step": 1807
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7316839619361977e-05,
+      "loss": 0.4549,
+      "step": 1808
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.731364357856549e-05,
+      "loss": 0.4575,
+      "step": 1809
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7310445930736427e-05,
+      "loss": 0.4408,
+      "step": 1810
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7307246676577397e-05,
+      "loss": 0.4391,
+      "step": 1811
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7304045816791383e-05,
+      "loss": 0.4629,
+      "step": 1812
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.730084335208171e-05,
+      "loss": 0.4598,
+      "step": 1813
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7297639283152057e-05,
+      "loss": 0.4366,
+      "step": 1814
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7294433610706457e-05,
+      "loss": 0.4604,
+      "step": 1815
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.72912263354493e-05,
+      "loss": 0.442,
+      "step": 1816
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.728801745808532e-05,
+      "loss": 0.4196,
+      "step": 1817
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7284806979319596e-05,
+      "loss": 0.4356,
+      "step": 1818
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7281594899857585e-05,
+      "loss": 0.4477,
+      "step": 1819
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.727838122040507e-05,
+      "loss": 0.4539,
+      "step": 1820
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.72751659416682e-05,
+      "loss": 0.4529,
+      "step": 1821
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7271949064353464e-05,
+      "loss": 0.4486,
+      "step": 1822
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7268730589167717e-05,
+      "loss": 0.422,
+      "step": 1823
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7265510516818152e-05,
+      "loss": 0.4331,
+      "step": 1824
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7262288848012324e-05,
+      "loss": 0.4544,
+      "step": 1825
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.725906558345812e-05,
+      "loss": 0.4478,
+      "step": 1826
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7255840723863807e-05,
+      "loss": 0.4321,
+      "step": 1827
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7252614269937972e-05,
+      "loss": 0.4508,
+      "step": 1828
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7249386222389577e-05,
+      "loss": 0.4214,
+      "step": 1829
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7246156581927912e-05,
+      "loss": 0.4306,
+      "step": 1830
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.724292534926264e-05,
+      "loss": 0.4198,
+      "step": 1831
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7239692525103752e-05,
+      "loss": 0.4298,
+      "step": 1832
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.723645811016161e-05,
+      "loss": 0.4464,
+      "step": 1833
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7233222105146902e-05,
+      "loss": 0.4311,
+      "step": 1834
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7229984510770688e-05,
+      "loss": 0.4701,
+      "step": 1835
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7226745327744364e-05,
+      "loss": 0.436,
+      "step": 1836
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.722350455677968e-05,
+      "loss": 0.4395,
+      "step": 1837
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7220262198588726e-05,
+      "loss": 0.4364,
+      "step": 1838
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7217018253883954e-05,
+      "loss": 0.4177,
+      "step": 1839
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7213772723378156e-05,
+      "loss": 0.4389,
+      "step": 1840
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.721052560778448e-05,
+      "loss": 0.427,
+      "step": 1841
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.720727690781641e-05,
+      "loss": 0.4282,
+      "step": 1842
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7204026624187785e-05,
+      "loss": 0.4482,
+      "step": 1843
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7200774757612802e-05,
+      "loss": 0.424,
+      "step": 1844
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.719752130880599e-05,
+      "loss": 0.4308,
+      "step": 1845
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7194266278482227e-05,
+      "loss": 0.4261,
+      "step": 1846
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7191009667356754e-05,
+      "loss": 0.4113,
+      "step": 1847
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.718775147614514e-05,
+      "loss": 0.4101,
+      "step": 1848
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7184491705563317e-05,
+      "loss": 0.474,
+      "step": 1849
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.718123035632755e-05,
+      "loss": 0.4742,
+      "step": 1850
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7177967429154457e-05,
+      "loss": 0.4441,
+      "step": 1851
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7174702924761013e-05,
+      "loss": 0.4379,
+      "step": 1852
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.717143684386452e-05,
+      "loss": 0.4431,
+      "step": 1853
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7168169187182644e-05,
+      "loss": 0.4275,
+      "step": 1854
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7164899955433384e-05,
+      "loss": 0.4379,
+      "step": 1855
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7161629149335095e-05,
+      "loss": 0.4286,
+      "step": 1856
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.715835676960647e-05,
+      "loss": 0.4487,
+      "step": 1857
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7155082816966556e-05,
+      "loss": 0.4194,
+      "step": 1858
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7151807292134738e-05,
+      "loss": 0.4346,
+      "step": 1859
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.714853019583075e-05,
+      "loss": 0.4292,
+      "step": 1860
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.714525152877467e-05,
+      "loss": 0.4393,
+      "step": 1861
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7141971291686926e-05,
+      "loss": 0.421,
+      "step": 1862
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.713868948528828e-05,
+      "loss": 0.4287,
+      "step": 1863
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7135406110299856e-05,
+      "loss": 0.4617,
+      "step": 1864
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7132121167443105e-05,
+      "loss": 0.4374,
+      "step": 1865
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.712883465743983e-05,
+      "loss": 0.4074,
+      "step": 1866
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7125546581012184e-05,
+      "loss": 0.4581,
+      "step": 1867
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7122256938882652e-05,
+      "loss": 0.4245,
+      "step": 1868
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.711896573177407e-05,
+      "loss": 0.4382,
+      "step": 1869
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7115672960409625e-05,
+      "loss": 0.4523,
+      "step": 1870
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.711237862551283e-05,
+      "loss": 0.4401,
+      "step": 1871
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7109082727807564e-05,
+      "loss": 0.4311,
+      "step": 1872
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7105785268018023e-05,
+      "loss": 0.4323,
+      "step": 1873
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.710248624686877e-05,
+      "loss": 0.4296,
+      "step": 1874
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.70991856650847e-05,
+      "loss": 0.4438,
+      "step": 1875
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.709588352339105e-05,
+      "loss": 0.4448,
+      "step": 1876
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7092579822513406e-05,
+      "loss": 0.472,
+      "step": 1877
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7089274563177687e-05,
+      "loss": 0.4432,
+      "step": 1878
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7085967746110166e-05,
+      "loss": 0.4416,
+      "step": 1879
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.708265937203745e-05,
+      "loss": 0.4354,
+      "step": 1880
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7079349441686487e-05,
+      "loss": 0.4539,
+      "step": 1881
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7076037955784576e-05,
+      "loss": 0.4399,
+      "step": 1882
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7072724915059355e-05,
+      "loss": 0.4509,
+      "step": 1883
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.706941032023879e-05,
+      "loss": 0.4477,
+      "step": 1884
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7066094172051212e-05,
+      "loss": 0.4503,
+      "step": 1885
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.706277647122527e-05,
+      "loss": 0.4694,
+      "step": 1886
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.705945721848997e-05,
+      "loss": 0.432,
+      "step": 1887
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7056136414574654e-05,
+      "loss": 0.4723,
+      "step": 1888
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7052814060209005e-05,
+      "loss": 0.452,
+      "step": 1889
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.704949015612305e-05,
+      "loss": 0.4652,
+      "step": 1890
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7046164703047146e-05,
+      "loss": 0.4339,
+      "step": 1891
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7042837701711997e-05,
+      "loss": 0.436,
+      "step": 1892
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7039509152848657e-05,
+      "loss": 0.4414,
+      "step": 1893
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.70361790571885e-05,
+      "loss": 0.4401,
+      "step": 1894
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.703284741546326e-05,
+      "loss": 0.4431,
+      "step": 1895
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7029514228404994e-05,
+      "loss": 0.441,
+      "step": 1896
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7026179496746106e-05,
+      "loss": 0.4325,
+      "step": 1897
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7022843221219344e-05,
+      "loss": 0.4467,
+      "step": 1898
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7019505402557787e-05,
+      "loss": 0.4856,
+      "step": 1899
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.701616604149486e-05,
+      "loss": 0.4448,
+      "step": 1900
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7012825138764318e-05,
+      "loss": 0.4333,
+      "step": 1901
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7009482695100262e-05,
+      "loss": 0.4466,
+      "step": 1902
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7006138711237134e-05,
+      "loss": 0.4349,
+      "step": 1903
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7002793187909702e-05,
+      "loss": 0.4222,
+      "step": 1904
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6999446125853088e-05,
+      "loss": 0.4467,
+      "step": 1905
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6996097525802742e-05,
+      "loss": 0.4198,
+      "step": 1906
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6992747388494456e-05,
+      "loss": 0.443,
+      "step": 1907
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6989395714664354e-05,
+      "loss": 0.4663,
+      "step": 1908
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.698604250504891e-05,
+      "loss": 0.4552,
+      "step": 1909
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6982687760384913e-05,
+      "loss": 0.4475,
+      "step": 1910
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6979331481409515e-05,
+      "loss": 0.4383,
+      "step": 1911
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.697597366886019e-05,
+      "loss": 0.4554,
+      "step": 1912
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.697261432347476e-05,
+      "loss": 0.4475,
+      "step": 1913
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6969253445991366e-05,
+      "loss": 0.4547,
+      "step": 1914
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6965891037148502e-05,
+      "loss": 0.4435,
+      "step": 1915
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6962527097684992e-05,
+      "loss": 0.4492,
+      "step": 1916
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6959161628339993e-05,
+      "loss": 0.4522,
+      "step": 1917
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.695579462985301e-05,
+      "loss": 0.4396,
+      "step": 1918
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.695242610296387e-05,
+      "loss": 0.425,
+      "step": 1919
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6949056048412744e-05,
+      "loss": 0.4379,
+      "step": 1920
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.694568446694014e-05,
+      "loss": 0.4387,
+      "step": 1921
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6942311359286886e-05,
+      "loss": 0.4531,
+      "step": 1922
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6938936726194173e-05,
+      "loss": 0.4344,
+      "step": 1923
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6935560568403504e-05,
+      "loss": 0.4432,
+      "step": 1924
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6932182886656726e-05,
+      "loss": 0.4498,
+      "step": 1925
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6928803681696024e-05,
+      "loss": 0.4475,
+      "step": 1926
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6925422954263902e-05,
+      "loss": 0.4242,
+      "step": 1927
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.692204070510322e-05,
+      "loss": 0.432,
+      "step": 1928
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6918656934957157e-05,
+      "loss": 0.4226,
+      "step": 1929
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6915271644569235e-05,
+      "loss": 0.4509,
+      "step": 1930
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6911884834683305e-05,
+      "loss": 0.4352,
+      "step": 1931
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.690849650604355e-05,
+      "loss": 0.4397,
+      "step": 1932
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6905106659394497e-05,
+      "loss": 0.4416,
+      "step": 1933
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6901715295480995e-05,
+      "loss": 0.4541,
+      "step": 1934
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6898322415048232e-05,
+      "loss": 0.4501,
+      "step": 1935
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6894928018841726e-05,
+      "loss": 0.4304,
+      "step": 1936
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6891532107607337e-05,
+      "loss": 0.4422,
+      "step": 1937
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6888134682091246e-05,
+      "loss": 0.4485,
+      "step": 1938
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.688473574303997e-05,
+      "loss": 0.4333,
+      "step": 1939
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6881335291200365e-05,
+      "loss": 0.4203,
+      "step": 1940
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6877933327319613e-05,
+      "loss": 0.4243,
+      "step": 1941
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.687452985214523e-05,
+      "loss": 0.445,
+      "step": 1942
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6871124866425063e-05,
+      "loss": 0.4475,
+      "step": 1943
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.68677183709073e-05,
+      "loss": 0.4517,
+      "step": 1944
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.686431036634044e-05,
+      "loss": 0.4485,
+      "step": 1945
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6860900853473335e-05,
+      "loss": 0.45,
+      "step": 1946
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6857489833055158e-05,
+      "loss": 0.436,
+      "step": 1947
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6854077305835414e-05,
+      "loss": 0.4392,
+      "step": 1948
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6850663272563944e-05,
+      "loss": 0.4222,
+      "step": 1949
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6847247733990914e-05,
+      "loss": 0.4397,
+      "step": 1950
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.684383069086682e-05,
+      "loss": 0.4327,
+      "step": 1951
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6840412143942494e-05,
+      "loss": 0.4432,
+      "step": 1952
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.68369920939691e-05,
+      "loss": 0.4436,
+      "step": 1953
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.683357054169812e-05,
+      "loss": 0.4308,
+      "step": 1954
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6830147487881384e-05,
+      "loss": 0.4317,
+      "step": 1955
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.682672293327103e-05,
+      "loss": 0.4278,
+      "step": 1956
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.682329687861955e-05,
+      "loss": 0.4423,
+      "step": 1957
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6819869324679746e-05,
+      "loss": 0.421,
+      "step": 1958
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.681644027220476e-05,
+      "loss": 0.435,
+      "step": 1959
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6813009721948063e-05,
+      "loss": 0.4302,
+      "step": 1960
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6809577674663446e-05,
+      "loss": 0.4337,
+      "step": 1961
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.680614413110504e-05,
+      "loss": 0.4473,
+      "step": 1962
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.68027090920273e-05,
+      "loss": 0.4285,
+      "step": 1963
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.679927255818501e-05,
+      "loss": 0.4374,
+      "step": 1964
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.679583453033328e-05,
+      "loss": 0.4445,
+      "step": 1965
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6792395009227548e-05,
+      "loss": 0.4078,
+      "step": 1966
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6788953995623593e-05,
+      "loss": 0.4366,
+      "step": 1967
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.67855114902775e-05,
+      "loss": 0.4223,
+      "step": 1968
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.67820674939457e-05,
+      "loss": 0.4193,
+      "step": 1969
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6778622007384947e-05,
+      "loss": 0.4134,
+      "step": 1970
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.677517503135231e-05,
+      "loss": 0.4135,
+      "step": 1971
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6771726566605206e-05,
+      "loss": 0.4404,
+      "step": 1972
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6768276613901367e-05,
+      "loss": 0.446,
+      "step": 1973
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6764825173998847e-05,
+      "loss": 0.4278,
+      "step": 1974
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6761372247656044e-05,
+      "loss": 0.4286,
+      "step": 1975
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6757917835631664e-05,
+      "loss": 0.4372,
+      "step": 1976
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.675446193868475e-05,
+      "loss": 0.4497,
+      "step": 1977
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6751004557574674e-05,
+      "loss": 0.4278,
+      "step": 1978
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6747545693061123e-05,
+      "loss": 0.4269,
+      "step": 1979
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6744085345904112e-05,
+      "loss": 0.4381,
+      "step": 1980
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6740623516863997e-05,
+      "loss": 0.4291,
+      "step": 1981
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.673716020670144e-05,
+      "loss": 0.4518,
+      "step": 1982
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.673369541617744e-05,
+      "loss": 0.448,
+      "step": 1983
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6730229146053316e-05,
+      "loss": 0.4396,
+      "step": 1984
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.672676139709072e-05,
+      "loss": 0.4189,
+      "step": 1985
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6723292170051616e-05,
+      "loss": 0.4509,
+      "step": 1986
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6719821465698304e-05,
+      "loss": 0.4246,
+      "step": 1987
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6716349284793404e-05,
+      "loss": 0.4209,
+      "step": 1988
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.671287562809986e-05,
+      "loss": 0.4179,
+      "step": 1989
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6709400496380938e-05,
+      "loss": 0.4344,
+      "step": 1990
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.670592389040024e-05,
+      "loss": 0.4405,
+      "step": 1991
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6702445810921677e-05,
+      "loss": 0.4399,
+      "step": 1992
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.669896625870949e-05,
+      "loss": 0.4198,
+      "step": 1993
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6695485234528246e-05,
+      "loss": 0.4357,
+      "step": 1994
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6692002739142835e-05,
+      "loss": 0.455,
+      "step": 1995
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6688518773318462e-05,
+      "loss": 0.4263,
+      "step": 1996
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6685033337820667e-05,
+      "loss": 0.4513,
+      "step": 1997
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.668154643341531e-05,
+      "loss": 0.4661,
+      "step": 1998
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.667805806086856e-05,
+      "loss": 0.4236,
+      "step": 1999
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.667456822094693e-05,
+      "loss": 0.436,
+      "step": 2000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6671076914417242e-05,
+      "loss": 0.4305,
+      "step": 2001
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.666758414204664e-05,
+      "loss": 0.4198,
+      "step": 2002
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6664089904602604e-05,
+      "loss": 0.4133,
+      "step": 2003
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6660594202852917e-05,
+      "loss": 0.4517,
+      "step": 2004
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.665709703756569e-05,
+      "loss": 0.4349,
+      "step": 2005
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6653598409509365e-05,
+      "loss": 0.44,
+      "step": 2006
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6650098319452692e-05,
+      "loss": 0.4477,
+      "step": 2007
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6646596768164754e-05,
+      "loss": 0.437,
+      "step": 2008
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6643093756414948e-05,
+      "loss": 0.4405,
+      "step": 2009
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6639589284972992e-05,
+      "loss": 0.4414,
+      "step": 2010
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6636083354608925e-05,
+      "loss": 0.428,
+      "step": 2011
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6632575966093114e-05,
+      "loss": 0.4525,
+      "step": 2012
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.662906712019623e-05,
+      "loss": 0.4212,
+      "step": 2013
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6625556817689285e-05,
+      "loss": 0.4122,
+      "step": 2014
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6622045059343593e-05,
+      "loss": 0.4444,
+      "step": 2015
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.66185318459308e-05,
+      "loss": 0.4331,
+      "step": 2016
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.661501717822286e-05,
+      "loss": 0.4318,
+      "step": 2017
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6611501056992064e-05,
+      "loss": 0.4514,
+      "step": 2018
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.660798348301101e-05,
+      "loss": 0.467,
+      "step": 2019
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6604464457052607e-05,
+      "loss": 0.4404,
+      "step": 2020
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6600943979890106e-05,
+      "loss": 0.4505,
+      "step": 2021
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6597422052297056e-05,
+      "loss": 0.4474,
+      "step": 2022
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6593898675047334e-05,
+      "loss": 0.4431,
+      "step": 2023
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.659037384891514e-05,
+      "loss": 0.4151,
+      "step": 2024
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.658684757467498e-05,
+      "loss": 0.4423,
+      "step": 2025
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.658331985310169e-05,
+      "loss": 0.4362,
+      "step": 2026
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6579790684970417e-05,
+      "loss": 0.4261,
+      "step": 2027
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6576260071056625e-05,
+      "loss": 0.421,
+      "step": 2028
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.65727280121361e-05,
+      "loss": 0.4441,
+      "step": 2029
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6569194508984948e-05,
+      "loss": 0.4282,
+      "step": 2030
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6565659562379584e-05,
+      "loss": 0.4611,
+      "step": 2031
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.656212317309675e-05,
+      "loss": 0.4401,
+      "step": 2032
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6558585341913492e-05,
+      "loss": 0.4386,
+      "step": 2033
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6555046069607188e-05,
+      "loss": 0.4404,
+      "step": 2034
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6551505356955516e-05,
+      "loss": 0.4363,
+      "step": 2035
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6547963204736486e-05,
+      "loss": 0.415,
+      "step": 2036
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.654441961372842e-05,
+      "loss": 0.4157,
+      "step": 2037
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.654087458470995e-05,
+      "loss": 0.4335,
+      "step": 2038
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.653732811846003e-05,
+      "loss": 0.4305,
+      "step": 2039
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.653378021575792e-05,
+      "loss": 0.4429,
+      "step": 2040
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6530230877383214e-05,
+      "loss": 0.4591,
+      "step": 2041
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6526680104115803e-05,
+      "loss": 0.4366,
+      "step": 2042
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6523127896735907e-05,
+      "loss": 0.4566,
+      "step": 2043
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.651957425602406e-05,
+      "loss": 0.4396,
+      "step": 2044
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6516019182761088e-05,
+      "loss": 0.4361,
+      "step": 2045
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.651246267772817e-05,
+      "loss": 0.4341,
+      "step": 2046
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6508904741706767e-05,
+      "loss": 0.445,
+      "step": 2047
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.650534537547867e-05,
+      "loss": 0.4521,
+      "step": 2048
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6501784579825984e-05,
+      "loss": 0.4196,
+      "step": 2049
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6498222355531123e-05,
+      "loss": 0.4338,
+      "step": 2050
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6494658703376822e-05,
+      "loss": 0.4317,
+      "step": 2051
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6491093624146117e-05,
+      "loss": 0.418,
+      "step": 2052
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6487527118622368e-05,
+      "loss": 0.4215,
+      "step": 2053
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.648395918758925e-05,
+      "loss": 0.4311,
+      "step": 2054
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6480389831830744e-05,
+      "loss": 0.4562,
+      "step": 2055
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.647681905213115e-05,
+      "loss": 0.4888,
+      "step": 2056
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6473246849275077e-05,
+      "loss": 0.4372,
+      "step": 2057
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6469673224047443e-05,
+      "loss": 0.4511,
+      "step": 2058
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6466098177233495e-05,
+      "loss": 0.4312,
+      "step": 2059
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.646252170961877e-05,
+      "loss": 0.4516,
+      "step": 2060
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6458943821989128e-05,
+      "loss": 0.445,
+      "step": 2061
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.645536451513075e-05,
+      "loss": 0.4215,
+      "step": 2062
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6451783789830113e-05,
+      "loss": 0.4622,
+      "step": 2063
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6448201646874012e-05,
+      "loss": 0.4322,
+      "step": 2064
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6444618087049557e-05,
+      "loss": 0.4258,
+      "step": 2065
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6441033111144168e-05,
+      "loss": 0.4388,
+      "step": 2066
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.643744671994557e-05,
+      "loss": 0.4327,
+      "step": 2067
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6433858914241807e-05,
+      "loss": 0.4361,
+      "step": 2068
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6430269694821228e-05,
+      "loss": 0.4329,
+      "step": 2069
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6426679062472497e-05,
+      "loss": 0.4464,
+      "step": 2070
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6423087017984586e-05,
+      "loss": 0.4482,
+      "step": 2071
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6419493562146773e-05,
+      "loss": 0.449,
+      "step": 2072
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.641589869574866e-05,
+      "loss": 0.4549,
+      "step": 2073
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.641230241958014e-05,
+      "loss": 0.4152,
+      "step": 2074
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6408704734431432e-05,
+      "loss": 0.4405,
+      "step": 2075
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6405105641093054e-05,
+      "loss": 0.4123,
+      "step": 2076
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6401505140355845e-05,
+      "loss": 0.4418,
+      "step": 2077
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6397903233010936e-05,
+      "loss": 0.4135,
+      "step": 2078
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6394299919849785e-05,
+      "loss": 0.4278,
+      "step": 2079
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6390695201664147e-05,
+      "loss": 0.4515,
+      "step": 2080
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6387089079246093e-05,
+      "loss": 0.4476,
+      "step": 2081
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.638348155338799e-05,
+      "loss": 0.4512,
+      "step": 2082
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6379872624882533e-05,
+      "loss": 0.4503,
+      "step": 2083
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6376262294522707e-05,
+      "loss": 0.4566,
+      "step": 2084
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6372650563101818e-05,
+      "loss": 0.4409,
+      "step": 2085
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6369037431413473e-05,
+      "loss": 0.4399,
+      "step": 2086
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6365422900251588e-05,
+      "loss": 0.4459,
+      "step": 2087
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6361806970410384e-05,
+      "loss": 0.4395,
+      "step": 2088
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6358189642684394e-05,
+      "loss": 0.444,
+      "step": 2089
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6354570917868458e-05,
+      "loss": 0.4446,
+      "step": 2090
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.635095079675772e-05,
+      "loss": 0.4286,
+      "step": 2091
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6347329280147634e-05,
+      "loss": 0.4224,
+      "step": 2092
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6343706368833955e-05,
+      "loss": 0.4393,
+      "step": 2093
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.634008206361275e-05,
+      "loss": 0.4384,
+      "step": 2094
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.633645636528039e-05,
+      "loss": 0.4477,
+      "step": 2095
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6332829274633553e-05,
+      "loss": 0.4463,
+      "step": 2096
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.632920079246922e-05,
+      "loss": 0.4388,
+      "step": 2097
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6325570919584687e-05,
+      "loss": 0.4359,
+      "step": 2098
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.632193965677754e-05,
+      "loss": 0.4384,
+      "step": 2099
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6318307004845684e-05,
+      "loss": 0.4445,
+      "step": 2100
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6314672964587322e-05,
+      "loss": 0.4304,
+      "step": 2101
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6311037536800972e-05,
+      "loss": 0.4316,
+      "step": 2102
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.630740072228544e-05,
+      "loss": 0.4192,
+      "step": 2103
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.630376252183985e-05,
+      "loss": 0.4536,
+      "step": 2104
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6300122936263623e-05,
+      "loss": 0.4429,
+      "step": 2105
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6296481966356495e-05,
+      "loss": 0.4324,
+      "step": 2106
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6292839612918493e-05,
+      "loss": 0.433,
+      "step": 2107
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.628919587674996e-05,
+      "loss": 0.4318,
+      "step": 2108
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6285550758651533e-05,
+      "loss": 0.43,
+      "step": 2109
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6281904259424154e-05,
+      "loss": 0.4317,
+      "step": 2110
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6278256379869075e-05,
+      "loss": 0.4349,
+      "step": 2111
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6274607120787848e-05,
+      "loss": 0.4258,
+      "step": 2112
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6270956482982324e-05,
+      "loss": 0.4501,
+      "step": 2113
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6267304467254663e-05,
+      "loss": 0.4285,
+      "step": 2114
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.626365107440733e-05,
+      "loss": 0.4572,
+      "step": 2115
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6259996305243077e-05,
+      "loss": 0.4088,
+      "step": 2116
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.625634016056498e-05,
+      "loss": 0.4354,
+      "step": 2117
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6252682641176397e-05,
+      "loss": 0.4296,
+      "step": 2118
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6249023747881005e-05,
+      "loss": 0.4175,
+      "step": 2119
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.624536348148277e-05,
+      "loss": 0.4477,
+      "step": 2120
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6241701842785968e-05,
+      "loss": 0.4323,
+      "step": 2121
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6238038832595172e-05,
+      "loss": 0.4405,
+      "step": 2122
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6234374451715265e-05,
+      "loss": 0.4416,
+      "step": 2123
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6230708700951418e-05,
+      "loss": 0.4133,
+      "step": 2124
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6227041581109103e-05,
+      "loss": 0.4393,
+      "step": 2125
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.622337309299411e-05,
+      "loss": 0.443,
+      "step": 2126
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6219703237412518e-05,
+      "loss": 0.4488,
+      "step": 2127
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6216032015170705e-05,
+      "loss": 0.446,
+      "step": 2128
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6212359427075346e-05,
+      "loss": 0.4415,
+      "step": 2129
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6208685473933425e-05,
+      "loss": 0.4425,
+      "step": 2130
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.620501015655223e-05,
+      "loss": 0.4415,
+      "step": 2131
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6201333475739328e-05,
+      "loss": 0.4191,
+      "step": 2132
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6197655432302614e-05,
+      "loss": 0.483,
+      "step": 2133
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6193976027050253e-05,
+      "loss": 0.4577,
+      "step": 2134
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6190295260790734e-05,
+      "loss": 0.4536,
+      "step": 2135
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.618661313433283e-05,
+      "loss": 0.4329,
+      "step": 2136
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6182929648485615e-05,
+      "loss": 0.4226,
+      "step": 2137
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6179244804058473e-05,
+      "loss": 0.4422,
+      "step": 2138
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6175558601861066e-05,
+      "loss": 0.4337,
+      "step": 2139
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6171871042703377e-05,
+      "loss": 0.4425,
+      "step": 2140
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.616818212739567e-05,
+      "loss": 0.448,
+      "step": 2141
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.616449185674851e-05,
+      "loss": 0.4445,
+      "step": 2142
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6160800231572773e-05,
+      "loss": 0.4316,
+      "step": 2143
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6157107252679612e-05,
+      "loss": 0.47,
+      "step": 2144
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.61534129208805e-05,
+      "loss": 0.4648,
+      "step": 2145
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6149717236987186e-05,
+      "loss": 0.4432,
+      "step": 2146
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.614602020181173e-05,
+      "loss": 0.4446,
+      "step": 2147
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.614232181616648e-05,
+      "loss": 0.4275,
+      "step": 2148
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.613862208086409e-05,
+      "loss": 0.4411,
+      "step": 2149
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.61349209967175e-05,
+      "loss": 0.4473,
+      "step": 2150
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.613121856453996e-05,
+      "loss": 0.469,
+      "step": 2151
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6127514785145003e-05,
+      "loss": 0.4464,
+      "step": 2152
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.612380965934646e-05,
+      "loss": 0.4504,
+      "step": 2153
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.612010318795847e-05,
+      "loss": 0.4397,
+      "step": 2154
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6116395371795456e-05,
+      "loss": 0.4337,
+      "step": 2155
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.611268621167213e-05,
+      "loss": 0.4484,
+      "step": 2156
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6108975708403526e-05,
+      "loss": 0.4542,
+      "step": 2157
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.610526386280494e-05,
+      "loss": 0.4547,
+      "step": 2158
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6101550675691987e-05,
+      "loss": 0.4223,
+      "step": 2159
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6097836147880564e-05,
+      "loss": 0.4483,
+      "step": 2160
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.609412028018687e-05,
+      "loss": 0.4351,
+      "step": 2161
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6090403073427395e-05,
+      "loss": 0.4284,
+      "step": 2162
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.608668452841893e-05,
+      "loss": 0.4265,
+      "step": 2163
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6082964645978537e-05,
+      "loss": 0.404,
+      "step": 2164
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6079243426923604e-05,
+      "loss": 0.4418,
+      "step": 2165
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.607552087207179e-05,
+      "loss": 0.4436,
+      "step": 2166
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.607179698224106e-05,
+      "loss": 0.4347,
+      "step": 2167
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.606807175824966e-05,
+      "loss": 0.4225,
+      "step": 2168
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.606434520091615e-05,
+      "loss": 0.4461,
+      "step": 2169
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.606061731105935e-05,
+      "loss": 0.4415,
+      "step": 2170
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.605688808949841e-05,
+      "loss": 0.4372,
+      "step": 2171
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.605315753705274e-05,
+      "loss": 0.4268,
+      "step": 2172
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6049425654542068e-05,
+      "loss": 0.4526,
+      "step": 2173
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6045692442786398e-05,
+      "loss": 0.4261,
+      "step": 2174
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6041957902606036e-05,
+      "loss": 0.4385,
+      "step": 2175
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.603822203482157e-05,
+      "loss": 0.437,
+      "step": 2176
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.603448484025389e-05,
+      "loss": 0.4228,
+      "step": 2177
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6030746319724168e-05,
+      "loss": 0.4235,
+      "step": 2178
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6027006474053872e-05,
+      "loss": 0.4391,
+      "step": 2179
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6023265304064767e-05,
+      "loss": 0.4191,
+      "step": 2180
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6019522810578904e-05,
+      "loss": 0.4433,
+      "step": 2181
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.601577899441861e-05,
+      "loss": 0.4413,
+      "step": 2182
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.601203385640653e-05,
+      "loss": 0.4329,
+      "step": 2183
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6008287397365587e-05,
+      "loss": 0.4408,
+      "step": 2184
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6004539618118985e-05,
+      "loss": 0.4276,
+      "step": 2185
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6000790519490233e-05,
+      "loss": 0.4254,
+      "step": 2186
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5997040102303117e-05,
+      "loss": 0.4341,
+      "step": 2187
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5993288367381725e-05,
+      "loss": 0.4277,
+      "step": 2188
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5989535315550428e-05,
+      "loss": 0.4318,
+      "step": 2189
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.598578094763388e-05,
+      "loss": 0.4195,
+      "step": 2190
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5982025264457038e-05,
+      "loss": 0.4231,
+      "step": 2191
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.597826826684514e-05,
+      "loss": 0.4256,
+      "step": 2192
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5974509955623715e-05,
+      "loss": 0.4234,
+      "step": 2193
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5970750331618582e-05,
+      "loss": 0.4089,
+      "step": 2194
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5966989395655837e-05,
+      "loss": 0.4475,
+      "step": 2195
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5963227148561878e-05,
+      "loss": 0.4086,
+      "step": 2196
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.595946359116339e-05,
+      "loss": 0.4126,
+      "step": 2197
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.595569872428734e-05,
+      "loss": 0.4249,
+      "step": 2198
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.595193254876098e-05,
+      "loss": 0.4321,
+      "step": 2199
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5948165065411867e-05,
+      "loss": 0.4421,
+      "step": 2200
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5944396275067822e-05,
+      "loss": 0.4376,
+      "step": 2201
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5940626178556967e-05,
+      "loss": 0.4189,
+      "step": 2202
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.593685477670771e-05,
+      "loss": 0.4548,
+      "step": 2203
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.593308207034874e-05,
+      "loss": 0.4465,
+      "step": 2204
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5929308060309043e-05,
+      "loss": 0.4374,
+      "step": 2205
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.592553274741788e-05,
+      "loss": 0.462,
+      "step": 2206
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.592175613250481e-05,
+      "loss": 0.4309,
+      "step": 2207
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.591797821639966e-05,
+      "loss": 0.4352,
+      "step": 2208
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5914198999932564e-05,
+      "loss": 0.4216,
+      "step": 2209
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5910418483933933e-05,
+      "loss": 0.4259,
+      "step": 2210
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5906636669234453e-05,
+      "loss": 0.434,
+      "step": 2211
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5902853556665114e-05,
+      "loss": 0.4432,
+      "step": 2212
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.589906914705718e-05,
+      "loss": 0.4393,
+      "step": 2213
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.58952834412422e-05,
+      "loss": 0.4215,
+      "step": 2214
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5891496440052015e-05,
+      "loss": 0.4447,
+      "step": 2215
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.588770814431874e-05,
+      "loss": 0.4258,
+      "step": 2216
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5883918554874785e-05,
+      "loss": 0.4251,
+      "step": 2217
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5880127672552835e-05,
+      "loss": 0.4311,
+      "step": 2218
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5876335498185865e-05,
+      "loss": 0.4372,
+      "step": 2219
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5872542032607136e-05,
+      "loss": 0.421,
+      "step": 2220
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.586874727665018e-05,
+      "loss": 0.4377,
+      "step": 2221
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5864951231148834e-05,
+      "loss": 0.4486,
+      "step": 2222
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.58611538969372e-05,
+      "loss": 0.4491,
+      "step": 2223
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5857355274849667e-05,
+      "loss": 0.4328,
+      "step": 2224
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5853555365720912e-05,
+      "loss": 0.4364,
+      "step": 2225
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.584975417038589e-05,
+      "loss": 0.4295,
+      "step": 2226
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5845951689679843e-05,
+      "loss": 0.4268,
+      "step": 2227
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.584214792443829e-05,
+      "loss": 0.4304,
+      "step": 2228
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.583834287549704e-05,
+      "loss": 0.4168,
+      "step": 2229
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5834536543692177e-05,
+      "loss": 0.4146,
+      "step": 2230
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5830728929860068e-05,
+      "loss": 0.4434,
+      "step": 2231
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.582692003483736e-05,
+      "loss": 0.424,
+      "step": 2232
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5823109859460993e-05,
+      "loss": 0.4243,
+      "step": 2233
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5819298404568174e-05,
+      "loss": 0.4427,
+      "step": 2234
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5815485670996394e-05,
+      "loss": 0.4268,
+      "step": 2235
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5811671659583435e-05,
+      "loss": 0.4397,
+      "step": 2236
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5807856371167352e-05,
+      "loss": 0.4373,
+      "step": 2237
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5804039806586477e-05,
+      "loss": 0.4347,
+      "step": 2238
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5800221966679427e-05,
+      "loss": 0.4555,
+      "step": 2239
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5796402852285103e-05,
+      "loss": 0.4386,
+      "step": 2240
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5792582464242678e-05,
+      "loss": 0.4656,
+      "step": 2241
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.578876080339161e-05,
+      "loss": 0.4859,
+      "step": 2242
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5784937870571634e-05,
+      "loss": 0.4421,
+      "step": 2243
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5781113666622766e-05,
+      "loss": 0.4536,
+      "step": 2244
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.577728819238531e-05,
+      "loss": 0.4353,
+      "step": 2245
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.577346144869983e-05,
+      "loss": 0.4425,
+      "step": 2246
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5769633436407183e-05,
+      "loss": 0.4224,
+      "step": 2247
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5765804156348497e-05,
+      "loss": 0.4286,
+      "step": 2248
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.576197360936519e-05,
+      "loss": 0.4277,
+      "step": 2249
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.575814179629895e-05,
+      "loss": 0.4209,
+      "step": 2250
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.575430871799174e-05,
+      "loss": 0.4105,
+      "step": 2251
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5750474375285808e-05,
+      "loss": 0.4335,
+      "step": 2252
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5746638769023676e-05,
+      "loss": 0.4212,
+      "step": 2253
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5742801900048147e-05,
+      "loss": 0.4394,
+      "step": 2254
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5738963769202293e-05,
+      "loss": 0.4384,
+      "step": 2255
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5735124377329475e-05,
+      "loss": 0.434,
+      "step": 2256
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5731283725273323e-05,
+      "loss": 0.4332,
+      "step": 2257
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5727441813877753e-05,
+      "loss": 0.409,
+      "step": 2258
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.572359864398694e-05,
+      "loss": 0.4489,
+      "step": 2259
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5719754216445357e-05,
+      "loss": 0.4172,
+      "step": 2260
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5715908532097737e-05,
+      "loss": 0.455,
+      "step": 2261
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5712061591789098e-05,
+      "loss": 0.4487,
+      "step": 2262
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5708213396364726e-05,
+      "loss": 0.4218,
+      "step": 2263
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5704363946670196e-05,
+      "loss": 0.4259,
+      "step": 2264
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5700513243551346e-05,
+      "loss": 0.4381,
+      "step": 2265
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5696661287854293e-05,
+      "loss": 0.429,
+      "step": 2266
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5692808080425435e-05,
+      "loss": 0.4457,
+      "step": 2267
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5688953622111432e-05,
+      "loss": 0.4769,
+      "step": 2268
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5685097913759237e-05,
+      "loss": 0.439,
+      "step": 2269
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5681240956216064e-05,
+      "loss": 0.4244,
+      "step": 2270
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.56773827503294e-05,
+      "loss": 0.4262,
+      "step": 2271
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5673523296947018e-05,
+      "loss": 0.4377,
+      "step": 2272
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.566966259691696e-05,
+      "loss": 0.4506,
+      "step": 2273
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.566580065108753e-05,
+      "loss": 0.4199,
+      "step": 2274
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5661937460307327e-05,
+      "loss": 0.4321,
+      "step": 2275
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.565807302542521e-05,
+      "loss": 0.4283,
+      "step": 2276
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.565420734729031e-05,
+      "loss": 0.4477,
+      "step": 2277
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5650340426752042e-05,
+      "loss": 0.4232,
+      "step": 2278
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5646472264660083e-05,
+      "loss": 0.4286,
+      "step": 2279
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.564260286186439e-05,
+      "loss": 0.4185,
+      "step": 2280
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.563873221921519e-05,
+      "loss": 0.3953,
+      "step": 2281
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.563486033756298e-05,
+      "loss": 0.4292,
+      "step": 2282
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5630987217758535e-05,
+      "loss": 0.4183,
+      "step": 2283
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5627112860652892e-05,
+      "loss": 0.4306,
+      "step": 2284
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5623237267097374e-05,
+      "loss": 0.4515,
+      "step": 2285
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.561936043794356e-05,
+      "loss": 0.4239,
+      "step": 2286
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.561548237404332e-05,
+      "loss": 0.4326,
+      "step": 2287
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5611603076248777e-05,
+      "loss": 0.4193,
+      "step": 2288
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.560772254541233e-05,
+      "loss": 0.4245,
+      "step": 2289
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5603840782386655e-05,
+      "loss": 0.4522,
+      "step": 2290
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.559995778802469e-05,
+      "loss": 0.4393,
+      "step": 2291
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5596073563179658e-05,
+      "loss": 0.443,
+      "step": 2292
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5592188108705034e-05,
+      "loss": 0.4443,
+      "step": 2293
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5588301425454574e-05,
+      "loss": 0.4371,
+      "step": 2294
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.55844135142823e-05,
+      "loss": 0.4406,
+      "step": 2295
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5580524376042513e-05,
+      "loss": 0.4204,
+      "step": 2296
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5576634011589768e-05,
+      "loss": 0.4308,
+      "step": 2297
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.55727424217789e-05,
+      "loss": 0.4163,
+      "step": 2298
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5568849607465016e-05,
+      "loss": 0.4408,
+      "step": 2299
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5564955569503488e-05,
+      "loss": 0.4354,
+      "step": 2300
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5561060308749943e-05,
+      "loss": 0.4148,
+      "step": 2301
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.55571638260603e-05,
+      "loss": 0.4315,
+      "step": 2302
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5553266122290737e-05,
+      "loss": 0.4462,
+      "step": 2303
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5549367198297694e-05,
+      "loss": 0.4037,
+      "step": 2304
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.554546705493789e-05,
+      "loss": 0.4468,
+      "step": 2305
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5541565693068305e-05,
+      "loss": 0.4099,
+      "step": 2306
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5537663113546187e-05,
+      "loss": 0.415,
+      "step": 2307
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5533759317229053e-05,
+      "loss": 0.4174,
+      "step": 2308
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5529854304974686e-05,
+      "loss": 0.4312,
+      "step": 2309
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.552594807764114e-05,
+      "loss": 0.4007,
+      "step": 2310
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.552204063608673e-05,
+      "loss": 0.4058,
+      "step": 2311
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5518131981170045e-05,
+      "loss": 0.4562,
+      "step": 2312
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5514222113749933e-05,
+      "loss": 0.4532,
+      "step": 2313
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5510311034685518e-05,
+      "loss": 0.444,
+      "step": 2314
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5506398744836175e-05,
+      "loss": 0.4285,
+      "step": 2315
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.550248524506156e-05,
+      "loss": 0.4234,
+      "step": 2316
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5498570536221596e-05,
+      "loss": 0.4517,
+      "step": 2317
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5494654619176456e-05,
+      "loss": 0.4283,
+      "step": 2318
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.549073749478659e-05,
+      "loss": 0.4386,
+      "step": 2319
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5486819163912707e-05,
+      "loss": 0.4468,
+      "step": 2320
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5482899627415794e-05,
+      "loss": 0.4344,
+      "step": 2321
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5478978886157086e-05,
+      "loss": 0.435,
+      "step": 2322
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5475056940998093e-05,
+      "loss": 0.4227,
+      "step": 2323
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5471133792800587e-05,
+      "loss": 0.4181,
+      "step": 2324
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5467209442426604e-05,
+      "loss": 0.4112,
+      "step": 2325
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5463283890738446e-05,
+      "loss": 0.4402,
+      "step": 2326
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5459357138598675e-05,
+      "loss": 0.4342,
+      "step": 2327
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5455429186870122e-05,
+      "loss": 0.4447,
+      "step": 2328
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5451500036415877e-05,
+      "loss": 0.4361,
+      "step": 2329
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5447569688099296e-05,
+      "loss": 0.4359,
+      "step": 2330
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5443638142784e-05,
+      "loss": 0.4261,
+      "step": 2331
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5439705401333864e-05,
+      "loss": 0.4391,
+      "step": 2332
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.543577146461304e-05,
+      "loss": 0.4508,
+      "step": 2333
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5431836333485926e-05,
+      "loss": 0.4202,
+      "step": 2334
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.54279000088172e-05,
+      "loss": 0.4193,
+      "step": 2335
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5423962491471795e-05,
+      "loss": 0.437,
+      "step": 2336
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5420023782314894e-05,
+      "loss": 0.441,
+      "step": 2337
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.541608388221196e-05,
+      "loss": 0.4307,
+      "step": 2338
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5412142792028707e-05,
+      "loss": 0.4496,
+      "step": 2339
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5408200512631122e-05,
+      "loss": 0.4492,
+      "step": 2340
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5404257044885435e-05,
+      "loss": 0.4402,
+      "step": 2341
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5400312389658153e-05,
+      "loss": 0.4118,
+      "step": 2342
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5396366547816032e-05,
+      "loss": 0.4334,
+      "step": 2343
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5392419520226102e-05,
+      "loss": 0.4314,
+      "step": 2344
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5388471307755643e-05,
+      "loss": 0.432,
+      "step": 2345
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.53845219112722e-05,
+      "loss": 0.4427,
+      "step": 2346
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5380571331643577e-05,
+      "loss": 0.4336,
+      "step": 2347
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.537661956973784e-05,
+      "loss": 0.4163,
+      "step": 2348
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5372666626423304e-05,
+      "loss": 0.4359,
+      "step": 2349
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5368712502568558e-05,
+      "loss": 0.4345,
+      "step": 2350
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.536475719904245e-05,
+      "loss": 0.4364,
+      "step": 2351
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5360800716714073e-05,
+      "loss": 0.4125,
+      "step": 2352
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5356843056452797e-05,
+      "loss": 0.4316,
+      "step": 2353
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5352884219128233e-05,
+      "loss": 0.4345,
+      "step": 2354
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5348924205610265e-05,
+      "loss": 0.448,
+      "step": 2355
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5344963016769025e-05,
+      "loss": 0.4222,
+      "step": 2356
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5341000653474912e-05,
+      "loss": 0.4595,
+      "step": 2357
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.533703711659858e-05,
+      "loss": 0.4258,
+      "step": 2358
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5333072407010936e-05,
+      "loss": 0.4293,
+      "step": 2359
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5329106525583153e-05,
+      "loss": 0.4308,
+      "step": 2360
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5325139473186652e-05,
+      "loss": 0.4339,
+      "step": 2361
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.532117125069312e-05,
+      "loss": 0.4406,
+      "step": 2362
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5317201858974505e-05,
+      "loss": 0.4186,
+      "step": 2363
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.531323129890299e-05,
+      "loss": 0.4492,
+      "step": 2364
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5309259571351035e-05,
+      "loss": 0.4176,
+      "step": 2365
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5305286677191355e-05,
+      "loss": 0.4331,
+      "step": 2366
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5301312617296912e-05,
+      "loss": 0.4234,
+      "step": 2367
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.529733739254093e-05,
+      "loss": 0.4343,
+      "step": 2368
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5293361003796895e-05,
+      "loss": 0.4148,
+      "step": 2369
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.528938345193853e-05,
+      "loss": 0.4436,
+      "step": 2370
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5285404737839838e-05,
+      "loss": 0.4498,
+      "step": 2371
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5281424862375053e-05,
+      "loss": 0.4221,
+      "step": 2372
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5277443826418684e-05,
+      "loss": 0.4239,
+      "step": 2373
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5273461630845482e-05,
+      "loss": 0.4448,
+      "step": 2374
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5269478276530463e-05,
+      "loss": 0.42,
+      "step": 2375
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5265493764348897e-05,
+      "loss": 0.4399,
+      "step": 2376
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5261508095176293e-05,
+      "loss": 0.43,
+      "step": 2377
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5257521269888429e-05,
+      "loss": 0.4355,
+      "step": 2378
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5253533289361335e-05,
+      "loss": 0.4457,
+      "step": 2379
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.524954415447129e-05,
+      "loss": 0.4308,
+      "step": 2380
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5245553866094832e-05,
+      "loss": 0.4248,
+      "step": 2381
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5241562425108754e-05,
+      "loss": 0.4424,
+      "step": 2382
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.523756983239009e-05,
+      "loss": 0.4283,
+      "step": 2383
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.523357608881614e-05,
+      "loss": 0.4269,
+      "step": 2384
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5229581195264456e-05,
+      "loss": 0.4388,
+      "step": 2385
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5225585152612832e-05,
+      "loss": 0.4114,
+      "step": 2386
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5221587961739324e-05,
+      "loss": 0.4222,
+      "step": 2387
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5217589623522241e-05,
+      "loss": 0.4314,
+      "step": 2388
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5213590138840139e-05,
+      "loss": 0.4258,
+      "step": 2389
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5209589508571825e-05,
+      "loss": 0.4412,
+      "step": 2390
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5205587733596364e-05,
+      "loss": 0.4371,
+      "step": 2391
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.520158481479307e-05,
+      "loss": 0.4131,
+      "step": 2392
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5197580753041503e-05,
+      "loss": 0.4446,
+      "step": 2393
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5193575549221484e-05,
+      "loss": 0.4234,
+      "step": 2394
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5189569204213075e-05,
+      "loss": 0.4183,
+      "step": 2395
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5185561718896594e-05,
+      "loss": 0.4265,
+      "step": 2396
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5181553094152612e-05,
+      "loss": 0.4387,
+      "step": 2397
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5177543330861944e-05,
+      "loss": 0.4306,
+      "step": 2398
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5173532429905659e-05,
+      "loss": 0.4179,
+      "step": 2399
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5169520392165078e-05,
+      "loss": 0.417,
+      "step": 2400
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.516550721852177e-05,
+      "loss": 0.4492,
+      "step": 2401
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5161492909857542e-05,
+      "loss": 0.4496,
+      "step": 2402
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5157477467054473e-05,
+      "loss": 0.4734,
+      "step": 2403
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5153460890994876e-05,
+      "loss": 0.4357,
+      "step": 2404
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5149443182561315e-05,
+      "loss": 0.4198,
+      "step": 2405
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5145424342636607e-05,
+      "loss": 0.4437,
+      "step": 2406
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5141404372103811e-05,
+      "loss": 0.4339,
+      "step": 2407
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.513738327184624e-05,
+      "loss": 0.4328,
+      "step": 2408
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5133361042747454e-05,
+      "loss": 0.4399,
+      "step": 2409
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.512933768569126e-05,
+      "loss": 0.4291,
+      "step": 2410
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5125313201561711e-05,
+      "loss": 0.4404,
+      "step": 2411
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5121287591243117e-05,
+      "loss": 0.4449,
+      "step": 2412
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5117260855620022e-05,
+      "loss": 0.4145,
+      "step": 2413
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5113232995577226e-05,
+      "loss": 0.4293,
+      "step": 2414
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5109204011999772e-05,
+      "loss": 0.4147,
+      "step": 2415
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5105173905772956e-05,
+      "loss": 0.4632,
+      "step": 2416
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.510114267778231e-05,
+      "loss": 0.4269,
+      "step": 2417
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5097110328913624e-05,
+      "loss": 0.4194,
+      "step": 2418
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5093076860052928e-05,
+      "loss": 0.4119,
+      "step": 2419
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.50890422720865e-05,
+      "loss": 0.4137,
+      "step": 2420
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5085006565900859e-05,
+      "loss": 0.4324,
+      "step": 2421
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5080969742382779e-05,
+      "loss": 0.4396,
+      "step": 2422
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.507693180241927e-05,
+      "loss": 0.4433,
+      "step": 2423
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5072892746897599e-05,
+      "loss": 0.4378,
+      "step": 2424
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.506885257670526e-05,
+      "loss": 0.4421,
+      "step": 2425
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5064811292730011e-05,
+      "loss": 0.4222,
+      "step": 2426
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5060768895859844e-05,
+      "loss": 0.4255,
+      "step": 2427
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5056725386982995e-05,
+      "loss": 0.459,
+      "step": 2428
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5052680766987952e-05,
+      "loss": 0.433,
+      "step": 2429
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5048635036763442e-05,
+      "loss": 0.4394,
+      "step": 2430
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5044588197198437e-05,
+      "loss": 0.4461,
+      "step": 2431
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5040540249182147e-05,
+      "loss": 0.4058,
+      "step": 2432
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5036491193604033e-05,
+      "loss": 0.4145,
+      "step": 2433
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.50324410313538e-05,
+      "loss": 0.4204,
+      "step": 2434
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.502838976332139e-05,
+      "loss": 0.4268,
+      "step": 2435
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5024337390396996e-05,
+      "loss": 0.4726,
+      "step": 2436
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5020283913471045e-05,
+      "loss": 0.4539,
+      "step": 2437
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5016229333434211e-05,
+      "loss": 0.4543,
+      "step": 2438
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5012173651177411e-05,
+      "loss": 0.4519,
+      "step": 2439
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5008116867591802e-05,
+      "loss": 0.444,
+      "step": 2440
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5004058983568786e-05,
+      "loss": 0.4214,
+      "step": 2441
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.4292,
+      "step": 2442
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4995939917777335e-05,
+      "loss": 0.4427,
+      "step": 2443
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4991878737792913e-05,
+      "loss": 0.4385,
+      "step": 2444
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4987816460939097e-05,
+      "loss": 0.4434,
+      "step": 2445
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4983753088108501e-05,
+      "loss": 0.437,
+      "step": 2446
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4979688620193969e-05,
+      "loss": 0.4082,
+      "step": 2447
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4975623058088587e-05,
+      "loss": 0.4393,
+      "step": 2448
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4971556402685688e-05,
+      "loss": 0.4286,
+      "step": 2449
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4967488654878838e-05,
+      "loss": 0.4273,
+      "step": 2450
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.496341981556185e-05,
+      "loss": 0.4536,
+      "step": 2451
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4959349885628771e-05,
+      "loss": 0.4356,
+      "step": 2452
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4955278865973895e-05,
+      "loss": 0.4402,
+      "step": 2453
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4951206757491746e-05,
+      "loss": 0.4434,
+      "step": 2454
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4947133561077087e-05,
+      "loss": 0.4296,
+      "step": 2455
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4943059277624929e-05,
+      "loss": 0.4332,
+      "step": 2456
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.493898390803052e-05,
+      "loss": 0.438,
+      "step": 2457
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4934907453189343e-05,
+      "loss": 0.4348,
+      "step": 2458
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4930829913997118e-05,
+      "loss": 0.4518,
+      "step": 2459
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4926751291349808e-05,
+      "loss": 0.436,
+      "step": 2460
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4922671586143609e-05,
+      "loss": 0.4306,
+      "step": 2461
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4918590799274964e-05,
+      "loss": 0.4429,
+      "step": 2462
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4914508931640539e-05,
+      "loss": 0.4478,
+      "step": 2463
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4910425984137251e-05,
+      "loss": 0.431,
+      "step": 2464
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4906341957662252e-05,
+      "loss": 0.4428,
+      "step": 2465
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4902256853112922e-05,
+      "loss": 0.3992,
+      "step": 2466
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4898170671386888e-05,
+      "loss": 0.4309,
+      "step": 2467
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4894083413382008e-05,
+      "loss": 0.4251,
+      "step": 2468
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4889995079996383e-05,
+      "loss": 0.4645,
+      "step": 2469
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4885905672128338e-05,
+      "loss": 0.4409,
+      "step": 2470
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4881815190676447e-05,
+      "loss": 0.4396,
+      "step": 2471
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4877723636539515e-05,
+      "loss": 0.4281,
+      "step": 2472
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4873631010616578e-05,
+      "loss": 0.4442,
+      "step": 2473
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4869537313806916e-05,
+      "loss": 0.426,
+      "step": 2474
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4865442547010038e-05,
+      "loss": 0.4498,
+      "step": 2475
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4861346711125695e-05,
+      "loss": 0.4461,
+      "step": 2476
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.485724980705386e-05,
+      "loss": 0.3999,
+      "step": 2477
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4853151835694757e-05,
+      "loss": 0.4304,
+      "step": 2478
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4849052797948828e-05,
+      "loss": 0.4236,
+      "step": 2479
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4844952694716765e-05,
+      "loss": 0.4524,
+      "step": 2480
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4840851526899484e-05,
+      "loss": 0.4193,
+      "step": 2481
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4836749295398138e-05,
+      "loss": 0.4333,
+      "step": 2482
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4832646001114114e-05,
+      "loss": 0.428,
+      "step": 2483
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4828541644949032e-05,
+      "loss": 0.4527,
+      "step": 2484
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4824436227804747e-05,
+      "loss": 0.451,
+      "step": 2485
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4820329750583342e-05,
+      "loss": 0.4216,
+      "step": 2486
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4816222214187137e-05,
+      "loss": 0.4377,
+      "step": 2487
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4812113619518684e-05,
+      "loss": 0.4474,
+      "step": 2488
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4808003967480773e-05,
+      "loss": 0.4315,
+      "step": 2489
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4803893258976412e-05,
+      "loss": 0.4282,
+      "step": 2490
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4799781494908859e-05,
+      "loss": 0.4229,
+      "step": 2491
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.479566867618159e-05,
+      "loss": 0.4461,
+      "step": 2492
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4791554803698316e-05,
+      "loss": 0.4307,
+      "step": 2493
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4787439878362988e-05,
+      "loss": 0.4568,
+      "step": 2494
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4783323901079778e-05,
+      "loss": 0.4721,
+      "step": 2495
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4779206872753091e-05,
+      "loss": 0.4122,
+      "step": 2496
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4775088794287566e-05,
+      "loss": 0.444,
+      "step": 2497
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4770969666588075e-05,
+      "loss": 0.4446,
+      "step": 2498
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4766849490559714e-05,
+      "loss": 0.4095,
+      "step": 2499
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4762728267107813e-05,
+      "loss": 0.4303,
+      "step": 2500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4758605997137934e-05,
+      "loss": 0.4201,
+      "step": 2501
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4754482681555859e-05,
+      "loss": 0.4419,
+      "step": 2502
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4750358321267615e-05,
+      "loss": 0.4348,
+      "step": 2503
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4746232917179448e-05,
+      "loss": 0.4353,
+      "step": 2504
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4742106470197837e-05,
+      "loss": 0.4556,
+      "step": 2505
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4737978981229488e-05,
+      "loss": 0.4285,
+      "step": 2506
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4733850451181342e-05,
+      "loss": 0.4415,
+      "step": 2507
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4729720880960561e-05,
+      "loss": 0.4318,
+      "step": 2508
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4725590271474535e-05,
+      "loss": 0.438,
+      "step": 2509
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4721458623630889e-05,
+      "loss": 0.4263,
+      "step": 2510
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4717325938337475e-05,
+      "loss": 0.4428,
+      "step": 2511
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4713192216502368e-05,
+      "loss": 0.4266,
+      "step": 2512
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4709057459033878e-05,
+      "loss": 0.4366,
+      "step": 2513
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4704921666840535e-05,
+      "loss": 0.4358,
+      "step": 2514
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4700784840831102e-05,
+      "loss": 0.4119,
+      "step": 2515
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4696646981914562e-05,
+      "loss": 0.4123,
+      "step": 2516
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4692508091000135e-05,
+      "loss": 0.4296,
+      "step": 2517
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4688368168997263e-05,
+      "loss": 0.4359,
+      "step": 2518
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4684227216815614e-05,
+      "loss": 0.4512,
+      "step": 2519
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4680085235365078e-05,
+      "loss": 0.4437,
+      "step": 2520
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4675942225555781e-05,
+      "loss": 0.4133,
+      "step": 2521
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4671798188298067e-05,
+      "loss": 0.4225,
+      "step": 2522
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4667653124502512e-05,
+      "loss": 0.448,
+      "step": 2523
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4663507035079909e-05,
+      "loss": 0.4204,
+      "step": 2524
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4659359920941287e-05,
+      "loss": 0.4121,
+      "step": 2525
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4655211782997888e-05,
+      "loss": 0.4165,
+      "step": 2526
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4651062622161192e-05,
+      "loss": 0.4352,
+      "step": 2527
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4646912439342891e-05,
+      "loss": 0.422,
+      "step": 2528
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4642761235454915e-05,
+      "loss": 0.4276,
+      "step": 2529
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4638609011409405e-05,
+      "loss": 0.4344,
+      "step": 2530
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4634455768118739e-05,
+      "loss": 0.4468,
+      "step": 2531
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4630301506495506e-05,
+      "loss": 0.4623,
+      "step": 2532
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4626146227452527e-05,
+      "loss": 0.4276,
+      "step": 2533
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4621989931902845e-05,
+      "loss": 0.414,
+      "step": 2534
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4617832620759727e-05,
+      "loss": 0.4433,
+      "step": 2535
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4613674294936661e-05,
+      "loss": 0.4118,
+      "step": 2536
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4609514955347361e-05,
+      "loss": 0.4159,
+      "step": 2537
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4605354602905761e-05,
+      "loss": 0.4292,
+      "step": 2538
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4601193238526017e-05,
+      "loss": 0.4277,
+      "step": 2539
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.459703086312251e-05,
+      "loss": 0.4259,
+      "step": 2540
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.459286747760984e-05,
+      "loss": 0.4307,
+      "step": 2541
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4588703082902838e-05,
+      "loss": 0.4129,
+      "step": 2542
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4584537679916539e-05,
+      "loss": 0.4436,
+      "step": 2543
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4580371269566219e-05,
+      "loss": 0.4315,
+      "step": 2544
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.457620385276736e-05,
+      "loss": 0.4339,
+      "step": 2545
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4572035430435679e-05,
+      "loss": 0.4342,
+      "step": 2546
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4567866003487101e-05,
+      "loss": 0.4162,
+      "step": 2547
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.456369557283778e-05,
+      "loss": 0.4253,
+      "step": 2548
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4559524139404087e-05,
+      "loss": 0.426,
+      "step": 2549
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4555351704102611e-05,
+      "loss": 0.4472,
+      "step": 2550
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.455117826785017e-05,
+      "loss": 0.4497,
+      "step": 2551
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4547003831563796e-05,
+      "loss": 0.4179,
+      "step": 2552
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4542828396160739e-05,
+      "loss": 0.4316,
+      "step": 2553
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4538651962558473e-05,
+      "loss": 0.4321,
+      "step": 2554
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4534474531674683e-05,
+      "loss": 0.4285,
+      "step": 2555
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4530296104427283e-05,
+      "loss": 0.4125,
+      "step": 2556
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4526116681734403e-05,
+      "loss": 0.4514,
+      "step": 2557
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.452193626451439e-05,
+      "loss": 0.446,
+      "step": 2558
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4517754853685811e-05,
+      "loss": 0.4465,
+      "step": 2559
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4513572450167452e-05,
+      "loss": 0.4483,
+      "step": 2560
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4509389054878311e-05,
+      "loss": 0.4287,
+      "step": 2561
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4505204668737615e-05,
+      "loss": 0.4239,
+      "step": 2562
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4501019292664795e-05,
+      "loss": 0.4404,
+      "step": 2563
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4496832927579512e-05,
+      "loss": 0.4213,
+      "step": 2564
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4492645574401638e-05,
+      "loss": 0.4375,
+      "step": 2565
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4488457234051266e-05,
+      "loss": 0.4604,
+      "step": 2566
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4484267907448696e-05,
+      "loss": 0.4189,
+      "step": 2567
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.448007759551446e-05,
+      "loss": 0.436,
+      "step": 2568
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4475886299169296e-05,
+      "loss": 0.448,
+      "step": 2569
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4471694019334156e-05,
+      "loss": 0.4266,
+      "step": 2570
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4467500756930219e-05,
+      "loss": 0.4355,
+      "step": 2571
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4463306512878874e-05,
+      "loss": 0.4435,
+      "step": 2572
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4459111288101722e-05,
+      "loss": 0.4423,
+      "step": 2573
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.445491508352058e-05,
+      "loss": 0.4576,
+      "step": 2574
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4450717900057491e-05,
+      "loss": 0.4207,
+      "step": 2575
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4446519738634704e-05,
+      "loss": 0.4257,
+      "step": 2576
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.444232060017468e-05,
+      "loss": 0.4156,
+      "step": 2577
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4438120485600102e-05,
+      "loss": 0.4294,
+      "step": 2578
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4433919395833863e-05,
+      "loss": 0.4185,
+      "step": 2579
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4429717331799068e-05,
+      "loss": 0.4316,
+      "step": 2580
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.442551429441905e-05,
+      "loss": 0.4103,
+      "step": 2581
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4421310284617335e-05,
+      "loss": 0.4156,
+      "step": 2582
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4417105303317682e-05,
+      "loss": 0.4484,
+      "step": 2583
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4412899351444049e-05,
+      "loss": 0.4093,
+      "step": 2584
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4408692429920618e-05,
+      "loss": 0.4331,
+      "step": 2585
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4404484539671772e-05,
+      "loss": 0.4417,
+      "step": 2586
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.440027568162212e-05,
+      "loss": 0.4203,
+      "step": 2587
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4396065856696475e-05,
+      "loss": 0.4442,
+      "step": 2588
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4391855065819865e-05,
+      "loss": 0.4358,
+      "step": 2589
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4387643309917534e-05,
+      "loss": 0.4535,
+      "step": 2590
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4383430589914928e-05,
+      "loss": 0.4274,
+      "step": 2591
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4379216906737718e-05,
+      "loss": 0.4397,
+      "step": 2592
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4375002261311774e-05,
+      "loss": 0.4159,
+      "step": 2593
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4370786654563187e-05,
+      "loss": 0.4212,
+      "step": 2594
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4366570087418254e-05,
+      "loss": 0.4213,
+      "step": 2595
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.436235256080349e-05,
+      "loss": 0.423,
+      "step": 2596
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4358134075645604e-05,
+      "loss": 0.4282,
+      "step": 2597
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4353914632871536e-05,
+      "loss": 0.4381,
+      "step": 2598
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4349694233408427e-05,
+      "loss": 0.4488,
+      "step": 2599
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4345472878183624e-05,
+      "loss": 0.4264,
+      "step": 2600
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4341250568124689e-05,
+      "loss": 0.4316,
+      "step": 2601
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4337027304159401e-05,
+      "loss": 0.434,
+      "step": 2602
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4332803087215733e-05,
+      "loss": 0.4273,
+      "step": 2603
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4328577918221878e-05,
+      "loss": 0.4439,
+      "step": 2604
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4324351798106236e-05,
+      "loss": 0.438,
+      "step": 2605
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4320124727797415e-05,
+      "loss": 0.431,
+      "step": 2606
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4315896708224236e-05,
+      "loss": 0.4401,
+      "step": 2607
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.431166774031572e-05,
+      "loss": 0.4536,
+      "step": 2608
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4307437825001103e-05,
+      "loss": 0.4545,
+      "step": 2609
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4303206963209828e-05,
+      "loss": 0.4343,
+      "step": 2610
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4298975155871547e-05,
+      "loss": 0.4128,
+      "step": 2611
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4294742403916113e-05,
+      "loss": 0.4399,
+      "step": 2612
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4290508708273599e-05,
+      "loss": 0.4401,
+      "step": 2613
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4286274069874275e-05,
+      "loss": 0.4403,
+      "step": 2614
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4282038489648624e-05,
+      "loss": 0.4469,
+      "step": 2615
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4277801968527327e-05,
+      "loss": 0.4297,
+      "step": 2616
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4273564507441282e-05,
+      "loss": 0.4243,
+      "step": 2617
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.426932610732159e-05,
+      "loss": 0.4302,
+      "step": 2618
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4265086769099558e-05,
+      "loss": 0.4198,
+      "step": 2619
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4260846493706697e-05,
+      "loss": 0.4562,
+      "step": 2620
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4256605282074726e-05,
+      "loss": 0.4435,
+      "step": 2621
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4252363135135577e-05,
+      "loss": 0.4493,
+      "step": 2622
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4248120053821371e-05,
+      "loss": 0.4195,
+      "step": 2623
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4243876039064447e-05,
+      "loss": 0.4316,
+      "step": 2624
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4239631091797346e-05,
+      "loss": 0.4293,
+      "step": 2625
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.423538521295281e-05,
+      "loss": 0.4311,
+      "step": 2626
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4231138403463796e-05,
+      "loss": 0.4359,
+      "step": 2627
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4226890664263453e-05,
+      "loss": 0.4249,
+      "step": 2628
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4222641996285142e-05,
+      "loss": 0.445,
+      "step": 2629
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4218392400462428e-05,
+      "loss": 0.427,
+      "step": 2630
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.421414187772908e-05,
+      "loss": 0.4531,
+      "step": 2631
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4209890429019059e-05,
+      "loss": 0.432,
+      "step": 2632
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4205638055266546e-05,
+      "loss": 0.4465,
+      "step": 2633
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4201384757405918e-05,
+      "loss": 0.4283,
+      "step": 2634
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4197130536371754e-05,
+      "loss": 0.4353,
+      "step": 2635
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4192875393098842e-05,
+      "loss": 0.4303,
+      "step": 2636
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4188619328522163e-05,
+      "loss": 0.4276,
+      "step": 2637
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4184362343576904e-05,
+      "loss": 0.4225,
+      "step": 2638
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4180104439198461e-05,
+      "loss": 0.4439,
+      "step": 2639
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4175845616322422e-05,
+      "loss": 0.4326,
+      "step": 2640
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.417158587588458e-05,
+      "loss": 0.4181,
+      "step": 2641
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.416732521882094e-05,
+      "loss": 0.4446,
+      "step": 2642
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4163063646067691e-05,
+      "loss": 0.4179,
+      "step": 2643
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4158801158561236e-05,
+      "loss": 0.4256,
+      "step": 2644
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.415453775723817e-05,
+      "loss": 0.4264,
+      "step": 2645
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4150273443035301e-05,
+      "loss": 0.4372,
+      "step": 2646
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4146008216889623e-05,
+      "loss": 0.4195,
+      "step": 2647
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4141742079738339e-05,
+      "loss": 0.4377,
+      "step": 2648
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4137475032518855e-05,
+      "loss": 0.4505,
+      "step": 2649
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4133207076168768e-05,
+      "loss": 0.4241,
+      "step": 2650
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.412893821162588e-05,
+      "loss": 0.4249,
+      "step": 2651
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4124668439828193e-05,
+      "loss": 0.4079,
+      "step": 2652
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4120397761713908e-05,
+      "loss": 0.4374,
+      "step": 2653
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4116126178221424e-05,
+      "loss": 0.4275,
+      "step": 2654
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4111853690289337e-05,
+      "loss": 0.4136,
+      "step": 2655
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4107580298856447e-05,
+      "loss": 0.4221,
+      "step": 2656
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4103306004861746e-05,
+      "loss": 0.4494,
+      "step": 2657
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4099030809244431e-05,
+      "loss": 0.436,
+      "step": 2658
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4094754712943895e-05,
+      "loss": 0.4206,
+      "step": 2659
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4090477716899727e-05,
+      "loss": 0.461,
+      "step": 2660
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4086199822051714e-05,
+      "loss": 0.4503,
+      "step": 2661
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4081921029339838e-05,
+      "loss": 0.4516,
+      "step": 2662
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4077641339704287e-05,
+      "loss": 0.4467,
+      "step": 2663
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4073360754085439e-05,
+      "loss": 0.4193,
+      "step": 2664
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4069079273423867e-05,
+      "loss": 0.4245,
+      "step": 2665
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4064796898660346e-05,
+      "loss": 0.4363,
+      "step": 2666
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.406051363073585e-05,
+      "loss": 0.4084,
+      "step": 2667
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.405622947059154e-05,
+      "loss": 0.4282,
+      "step": 2668
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.405194441916878e-05,
+      "loss": 0.4139,
+      "step": 2669
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4047658477409122e-05,
+      "loss": 0.439,
+      "step": 2670
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4043371646254327e-05,
+      "loss": 0.431,
+      "step": 2671
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4039083926646339e-05,
+      "loss": 0.4512,
+      "step": 2672
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4034795319527305e-05,
+      "loss": 0.4387,
+      "step": 2673
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4030505825839562e-05,
+      "loss": 0.4495,
+      "step": 2674
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4026215446525643e-05,
+      "loss": 0.4253,
+      "step": 2675
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.402192418252828e-05,
+      "loss": 0.4489,
+      "step": 2676
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4017632034790391e-05,
+      "loss": 0.4308,
+      "step": 2677
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4013339004255097e-05,
+      "loss": 0.444,
+      "step": 2678
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4009045091865708e-05,
+      "loss": 0.4151,
+      "step": 2679
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4004750298565726e-05,
+      "loss": 0.4223,
+      "step": 2680
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4000454625298851e-05,
+      "loss": 0.4098,
+      "step": 2681
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3996158073008977e-05,
+      "loss": 0.4373,
+      "step": 2682
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3991860642640185e-05,
+      "loss": 0.4409,
+      "step": 2683
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3987562335136755e-05,
+      "loss": 0.4508,
+      "step": 2684
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3983263151443158e-05,
+      "loss": 0.4567,
+      "step": 2685
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3978963092504055e-05,
+      "loss": 0.4339,
+      "step": 2686
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3974662159264298e-05,
+      "loss": 0.4338,
+      "step": 2687
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3970360352668944e-05,
+      "loss": 0.423,
+      "step": 2688
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3966057673663222e-05,
+      "loss": 0.4395,
+      "step": 2689
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3961754123192572e-05,
+      "loss": 0.4261,
+      "step": 2690
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.395744970220261e-05,
+      "loss": 0.4227,
+      "step": 2691
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3953144411639155e-05,
+      "loss": 0.4348,
+      "step": 2692
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3948838252448207e-05,
+      "loss": 0.4478,
+      "step": 2693
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3944531225575961e-05,
+      "loss": 0.4312,
+      "step": 2694
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3940223331968807e-05,
+      "loss": 0.4208,
+      "step": 2695
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3935914572573322e-05,
+      "loss": 0.4364,
+      "step": 2696
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3931604948336273e-05,
+      "loss": 0.4223,
+      "step": 2697
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3927294460204613e-05,
+      "loss": 0.4366,
+      "step": 2698
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3922983109125495e-05,
+      "loss": 0.4487,
+      "step": 2699
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3918670896046249e-05,
+      "loss": 0.4353,
+      "step": 2700
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3914357821914406e-05,
+      "loss": 0.4392,
+      "step": 2701
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.391004388767768e-05,
+      "loss": 0.4416,
+      "step": 2702
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3905729094283973e-05,
+      "loss": 0.4284,
+      "step": 2703
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.390141344268138e-05,
+      "loss": 0.4337,
+      "step": 2704
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.389709693381818e-05,
+      "loss": 0.4357,
+      "step": 2705
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3892779568642846e-05,
+      "loss": 0.4257,
+      "step": 2706
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3888461348104037e-05,
+      "loss": 0.4128,
+      "step": 2707
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3884142273150593e-05,
+      "loss": 0.4214,
+      "step": 2708
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3879822344731553e-05,
+      "loss": 0.4374,
+      "step": 2709
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3875501563796137e-05,
+      "loss": 0.42,
+      "step": 2710
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3871179931293751e-05,
+      "loss": 0.4488,
+      "step": 2711
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3866857448173994e-05,
+      "loss": 0.4416,
+      "step": 2712
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3862534115386648e-05,
+      "loss": 0.4283,
+      "step": 2713
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3858209933881682e-05,
+      "loss": 0.4526,
+      "step": 2714
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3853884904609252e-05,
+      "loss": 0.4132,
+      "step": 2715
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3849559028519697e-05,
+      "loss": 0.4184,
+      "step": 2716
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3845232306563548e-05,
+      "loss": 0.4531,
+      "step": 2717
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3840904739691519e-05,
+      "loss": 0.4247,
+      "step": 2718
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3836576328854507e-05,
+      "loss": 0.4289,
+      "step": 2719
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3832247075003603e-05,
+      "loss": 0.4219,
+      "step": 2720
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3827916979090072e-05,
+      "loss": 0.4323,
+      "step": 2721
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3823586042065373e-05,
+      "loss": 0.4327,
+      "step": 2722
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.381925426488114e-05,
+      "loss": 0.4208,
+      "step": 2723
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3814921648489203e-05,
+      "loss": 0.4266,
+      "step": 2724
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3810588193841569e-05,
+      "loss": 0.4438,
+      "step": 2725
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3806253901890436e-05,
+      "loss": 0.4208,
+      "step": 2726
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3801918773588174e-05,
+      "loss": 0.452,
+      "step": 2727
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3797582809887348e-05,
+      "loss": 0.4146,
+      "step": 2728
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3793246011740706e-05,
+      "loss": 0.4372,
+      "step": 2729
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.378890838010117e-05,
+      "loss": 0.4296,
+      "step": 2730
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3784569915921853e-05,
+      "loss": 0.4352,
+      "step": 2731
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.378023062015605e-05,
+      "loss": 0.4198,
+      "step": 2732
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3775890493757239e-05,
+      "loss": 0.4646,
+      "step": 2733
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3771549537679074e-05,
+      "loss": 0.4351,
+      "step": 2734
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3767207752875403e-05,
+      "loss": 0.4367,
+      "step": 2735
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3762865140300248e-05,
+      "loss": 0.4275,
+      "step": 2736
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3758521700907812e-05,
+      "loss": 0.447,
+      "step": 2737
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3754177435652487e-05,
+      "loss": 0.4355,
+      "step": 2738
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3749832345488838e-05,
+      "loss": 0.4353,
+      "step": 2739
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3745486431371616e-05,
+      "loss": 0.4385,
+      "step": 2740
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.374113969425575e-05,
+      "loss": 0.4437,
+      "step": 2741
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3736792135096355e-05,
+      "loss": 0.4455,
+      "step": 2742
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.373244375484872e-05,
+      "loss": 0.4458,
+      "step": 2743
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3728094554468328e-05,
+      "loss": 0.417,
+      "step": 2744
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3723744534910821e-05,
+      "loss": 0.4379,
+      "step": 2745
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3719393697132034e-05,
+      "loss": 0.4186,
+      "step": 2746
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3715042042087983e-05,
+      "loss": 0.4369,
+      "step": 2747
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.371068957073486e-05,
+      "loss": 0.4451,
+      "step": 2748
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3706336284029034e-05,
+      "loss": 0.4263,
+      "step": 2749
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.370198218292706e-05,
+      "loss": 0.43,
+      "step": 2750
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3697627268385667e-05,
+      "loss": 0.4427,
+      "step": 2751
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3693271541361766e-05,
+      "loss": 0.4273,
+      "step": 2752
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3688915002812437e-05,
+      "loss": 0.4341,
+      "step": 2753
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.368455765369495e-05,
+      "loss": 0.4563,
+      "step": 2754
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3680199494966747e-05,
+      "loss": 0.4462,
+      "step": 2755
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3675840527585457e-05,
+      "loss": 0.421,
+      "step": 2756
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3671480752508873e-05,
+      "loss": 0.4533,
+      "step": 2757
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3667120170694967e-05,
+      "loss": 0.4585,
+      "step": 2758
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3662758783101901e-05,
+      "loss": 0.4404,
+      "step": 2759
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3658396590688006e-05,
+      "loss": 0.4123,
+      "step": 2760
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3654033594411785e-05,
+      "loss": 0.4384,
+      "step": 2761
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3649669795231928e-05,
+      "loss": 0.4211,
+      "step": 2762
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.364530519410729e-05,
+      "loss": 0.442,
+      "step": 2763
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.364093979199691e-05,
+      "loss": 0.4311,
+      "step": 2764
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3636573589860003e-05,
+      "loss": 0.4271,
+      "step": 2765
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3632206588655958e-05,
+      "loss": 0.4165,
+      "step": 2766
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3627838789344341e-05,
+      "loss": 0.4295,
+      "step": 2767
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3623470192884888e-05,
+      "loss": 0.4186,
+      "step": 2768
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3619100800237517e-05,
+      "loss": 0.4225,
+      "step": 2769
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3614730612362314e-05,
+      "loss": 0.4267,
+      "step": 2770
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3610359630219549e-05,
+      "loss": 0.4274,
+      "step": 2771
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3605987854769657e-05,
+      "loss": 0.4257,
+      "step": 2772
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3601615286973256e-05,
+      "loss": 0.425,
+      "step": 2773
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3597241927791131e-05,
+      "loss": 0.4391,
+      "step": 2774
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3592867778184244e-05,
+      "loss": 0.4199,
+      "step": 2775
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3588492839113732e-05,
+      "loss": 0.4214,
+      "step": 2776
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.35841171115409e-05,
+      "loss": 0.4222,
+      "step": 2777
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.357974059642723e-05,
+      "loss": 0.432,
+      "step": 2778
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.357536329473438e-05,
+      "loss": 0.4237,
+      "step": 2779
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3570985207424181e-05,
+      "loss": 0.4317,
+      "step": 2780
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3566606335458628e-05,
+      "loss": 0.4091,
+      "step": 2781
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3562226679799893e-05,
+      "loss": 0.4437,
+      "step": 2782
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3557846241410326e-05,
+      "loss": 0.4067,
+      "step": 2783
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.355346502125244e-05,
+      "loss": 0.4163,
+      "step": 2784
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3549083020288925e-05,
+      "loss": 0.4097,
+      "step": 2785
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3544700239482644e-05,
+      "loss": 0.4387,
+      "step": 2786
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3540316679796628e-05,
+      "loss": 0.4358,
+      "step": 2787
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3535932342194077e-05,
+      "loss": 0.4428,
+      "step": 2788
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3531547227638364e-05,
+      "loss": 0.4142,
+      "step": 2789
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3527161337093043e-05,
+      "loss": 0.4359,
+      "step": 2790
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.352277467152182e-05,
+      "loss": 0.4501,
+      "step": 2791
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3518387231888582e-05,
+      "loss": 0.426,
+      "step": 2792
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3513999019157385e-05,
+      "loss": 0.4205,
+      "step": 2793
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3509610034292455e-05,
+      "loss": 0.4279,
+      "step": 2794
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3505220278258186e-05,
+      "loss": 0.4528,
+      "step": 2795
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3500829752019145e-05,
+      "loss": 0.4393,
+      "step": 2796
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3496438456540066e-05,
+      "loss": 0.4264,
+      "step": 2797
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3492046392785847e-05,
+      "loss": 0.4391,
+      "step": 2798
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3487653561721565e-05,
+      "loss": 0.4691,
+      "step": 2799
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3483259964312456e-05,
+      "loss": 0.4214,
+      "step": 2800
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3478865601523932e-05,
+      "loss": 0.4445,
+      "step": 2801
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3474470474321566e-05,
+      "loss": 0.442,
+      "step": 2802
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3470074583671107e-05,
+      "loss": 0.4287,
+      "step": 2803
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3465677930538465e-05,
+      "loss": 0.4291,
+      "step": 2804
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3461280515889719e-05,
+      "loss": 0.437,
+      "step": 2805
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3456882340691118e-05,
+      "loss": 0.4276,
+      "step": 2806
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3452483405909077e-05,
+      "loss": 0.4098,
+      "step": 2807
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3448083712510174e-05,
+      "loss": 0.4185,
+      "step": 2808
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3443683261461163e-05,
+      "loss": 0.4261,
+      "step": 2809
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3439282053728953e-05,
+      "loss": 0.402,
+      "step": 2810
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3434880090280625e-05,
+      "loss": 0.4263,
+      "step": 2811
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3430477372083429e-05,
+      "loss": 0.4279,
+      "step": 2812
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3426073900104777e-05,
+      "loss": 0.4097,
+      "step": 2813
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3421669675312246e-05,
+      "loss": 0.4398,
+      "step": 2814
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3417264698673583e-05,
+      "loss": 0.4407,
+      "step": 2815
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.341285897115669e-05,
+      "loss": 0.4239,
+      "step": 2816
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3408452493729646e-05,
+      "loss": 0.4608,
+      "step": 2817
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.340404526736069e-05,
+      "loss": 0.408,
+      "step": 2818
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3399637293018223e-05,
+      "loss": 0.4313,
+      "step": 2819
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3395228571670815e-05,
+      "loss": 0.4531,
+      "step": 2820
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3390819104287199e-05,
+      "loss": 0.4281,
+      "step": 2821
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3386408891836267e-05,
+      "loss": 0.4312,
+      "step": 2822
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3381997935287077e-05,
+      "loss": 0.4104,
+      "step": 2823
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3377586235608856e-05,
+      "loss": 0.4382,
+      "step": 2824
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3373173793770987e-05,
+      "loss": 0.4208,
+      "step": 2825
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3368760610743024e-05,
+      "loss": 0.4404,
+      "step": 2826
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3364346687494675e-05,
+      "loss": 0.4282,
+      "step": 2827
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3359932024995817e-05,
+      "loss": 0.4064,
+      "step": 2828
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3355516624216485e-05,
+      "loss": 0.4252,
+      "step": 2829
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.335110048612688e-05,
+      "loss": 0.4528,
+      "step": 2830
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3346683611697363e-05,
+      "loss": 0.4103,
+      "step": 2831
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3342266001898456e-05,
+      "loss": 0.4339,
+      "step": 2832
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.333784765770085e-05,
+      "loss": 0.4241,
+      "step": 2833
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3333428580075381e-05,
+      "loss": 0.4501,
+      "step": 2834
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3329008769993064e-05,
+      "loss": 0.4358,
+      "step": 2835
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3324588228425065e-05,
+      "loss": 0.4469,
+      "step": 2836
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3320166956342712e-05,
+      "loss": 0.4313,
+      "step": 2837
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3315744954717495e-05,
+      "loss": 0.4428,
+      "step": 2838
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3311322224521068e-05,
+      "loss": 0.4443,
+      "step": 2839
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3306898766725236e-05,
+      "loss": 0.442,
+      "step": 2840
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3302474582301967e-05,
+      "loss": 0.428,
+      "step": 2841
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3298049672223397e-05,
+      "loss": 0.4397,
+      "step": 2842
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3293624037461809e-05,
+      "loss": 0.4268,
+      "step": 2843
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3289197678989659e-05,
+      "loss": 0.4235,
+      "step": 2844
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3284770597779547e-05,
+      "loss": 0.4343,
+      "step": 2845
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3280342794804237e-05,
+      "loss": 0.4543,
+      "step": 2846
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3275914271036658e-05,
+      "loss": 0.4014,
+      "step": 2847
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3271485027449896e-05,
+      "loss": 0.4637,
+      "step": 2848
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3267055065017184e-05,
+      "loss": 0.4044,
+      "step": 2849
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.326262438471193e-05,
+      "loss": 0.4412,
+      "step": 2850
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3258192987507687e-05,
+      "loss": 0.4488,
+      "step": 2851
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3253760874378168e-05,
+      "loss": 0.4298,
+      "step": 2852
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3249328046297246e-05,
+      "loss": 0.4268,
+      "step": 2853
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3244894504238945e-05,
+      "loss": 0.4296,
+      "step": 2854
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3240460249177457e-05,
+      "loss": 0.4265,
+      "step": 2855
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3236025282087123e-05,
+      "loss": 0.4353,
+      "step": 2856
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3231589603942445e-05,
+      "loss": 0.4294,
+      "step": 2857
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.322715321571807e-05,
+      "loss": 0.4326,
+      "step": 2858
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3222716118388813e-05,
+      "loss": 0.3969,
+      "step": 2859
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3218278312929643e-05,
+      "loss": 0.4404,
+      "step": 2860
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3213839800315677e-05,
+      "loss": 0.4208,
+      "step": 2861
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3209400581522199e-05,
+      "loss": 0.4474,
+      "step": 2862
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3204960657524638e-05,
+      "loss": 0.4203,
+      "step": 2863
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3200520029298583e-05,
+      "loss": 0.419,
+      "step": 2864
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3196078697819776e-05,
+      "loss": 0.4281,
+      "step": 2865
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3191636664064116e-05,
+      "loss": 0.4263,
+      "step": 2866
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3187193929007653e-05,
+      "loss": 0.4344,
+      "step": 2867
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3182750493626595e-05,
+      "loss": 0.4271,
+      "step": 2868
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.31783063588973e-05,
+      "loss": 0.4211,
+      "step": 2869
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3173861525796277e-05,
+      "loss": 0.4447,
+      "step": 2870
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3169415995300199e-05,
+      "loss": 0.437,
+      "step": 2871
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3164969768385883e-05,
+      "loss": 0.4035,
+      "step": 2872
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3160522846030301e-05,
+      "loss": 0.4053,
+      "step": 2873
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.315607522921058e-05,
+      "loss": 0.4405,
+      "step": 2874
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3151626918904004e-05,
+      "loss": 0.4091,
+      "step": 2875
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3147177916087992e-05,
+      "loss": 0.396,
+      "step": 2876
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3142728221740132e-05,
+      "loss": 0.4034,
+      "step": 2877
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.313827783683816e-05,
+      "loss": 0.4193,
+      "step": 2878
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3133826762359965e-05,
+      "loss": 0.4293,
+      "step": 2879
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3129374999283581e-05,
+      "loss": 0.4146,
+      "step": 2880
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3124922548587197e-05,
+      "loss": 0.4106,
+      "step": 2881
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3120469411249157e-05,
+      "loss": 0.4294,
+      "step": 2882
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.311601558824795e-05,
+      "loss": 0.4075,
+      "step": 2883
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3111561080562218e-05,
+      "loss": 0.4405,
+      "step": 2884
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3107105889170753e-05,
+      "loss": 0.4366,
+      "step": 2885
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3102650015052501e-05,
+      "loss": 0.4357,
+      "step": 2886
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3098193459186547e-05,
+      "loss": 0.4024,
+      "step": 2887
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.309373622255214e-05,
+      "loss": 0.4078,
+      "step": 2888
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3089278306128673e-05,
+      "loss": 0.4216,
+      "step": 2889
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3084819710895685e-05,
+      "loss": 0.4232,
+      "step": 2890
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3080360437832863e-05,
+      "loss": 0.4171,
+      "step": 2891
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3075900487920054e-05,
+      "loss": 0.4262,
+      "step": 2892
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.307143986213724e-05,
+      "loss": 0.4174,
+      "step": 2893
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3066978561464561e-05,
+      "loss": 0.4129,
+      "step": 2894
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3062516586882301e-05,
+      "loss": 0.4383,
+      "step": 2895
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3058053939370891e-05,
+      "loss": 0.425,
+      "step": 2896
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3053590619910921e-05,
+      "loss": 0.4263,
+      "step": 2897
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3049126629483109e-05,
+      "loss": 0.4449,
+      "step": 2898
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3044661969068335e-05,
+      "loss": 0.4099,
+      "step": 2899
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3040196639647621e-05,
+      "loss": 0.4104,
+      "step": 2900
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.303573064220214e-05,
+      "loss": 0.4108,
+      "step": 2901
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3031263977713206e-05,
+      "loss": 0.4363,
+      "step": 2902
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3026796647162287e-05,
+      "loss": 0.4126,
+      "step": 2903
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3022328651530991e-05,
+      "loss": 0.4329,
+      "step": 2904
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3017859991801072e-05,
+      "loss": 0.4403,
+      "step": 2905
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3013390668954435e-05,
+      "loss": 0.4431,
+      "step": 2906
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3008920683973125e-05,
+      "loss": 0.4344,
+      "step": 2907
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3004450037839336e-05,
+      "loss": 0.4358,
+      "step": 2908
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.2999978731535408e-05,
+      "loss": 0.4457,
+      "step": 2909
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.2995506766043827e-05,
+      "loss": 0.451,
+      "step": 2910
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2991034142347213e-05,
+      "loss": 0.4185,
+      "step": 2911
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2986560861428346e-05,
+      "loss": 0.4296,
+      "step": 2912
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2982086924270145e-05,
+      "loss": 0.4378,
+      "step": 2913
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2977612331855664e-05,
+      "loss": 0.4419,
+      "step": 2914
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2973137085168117e-05,
+      "loss": 0.4602,
+      "step": 2915
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.296866118519085e-05,
+      "loss": 0.4276,
+      "step": 2916
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2964184632907353e-05,
+      "loss": 0.4267,
+      "step": 2917
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2959707429301265e-05,
+      "loss": 0.4206,
+      "step": 2918
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2955229575356366e-05,
+      "loss": 0.4279,
+      "step": 2919
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.295075107205658e-05,
+      "loss": 0.4088,
+      "step": 2920
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2946271920385967e-05,
+      "loss": 0.4249,
+      "step": 2921
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2941792121328739e-05,
+      "loss": 0.4339,
+      "step": 2922
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2937311675869241e-05,
+      "loss": 0.4398,
+      "step": 2923
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.293283058499197e-05,
+      "loss": 0.4296,
+      "step": 2924
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2928348849681556e-05,
+      "loss": 0.4041,
+      "step": 2925
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2923866470922775e-05,
+      "loss": 0.41,
+      "step": 2926
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2919383449700544e-05,
+      "loss": 0.42,
+      "step": 2927
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2914899786999924e-05,
+      "loss": 0.432,
+      "step": 2928
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2910415483806109e-05,
+      "loss": 0.4191,
+      "step": 2929
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2905930541104438e-05,
+      "loss": 0.4273,
+      "step": 2930
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2901444959880394e-05,
+      "loss": 0.4348,
+      "step": 2931
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2896958741119594e-05,
+      "loss": 0.4164,
+      "step": 2932
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2892471885807803e-05,
+      "loss": 0.4455,
+      "step": 2933
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.288798439493092e-05,
+      "loss": 0.4449,
+      "step": 2934
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2883496269474979e-05,
+      "loss": 0.3985,
+      "step": 2935
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2879007510426166e-05,
+      "loss": 0.4246,
+      "step": 2936
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2874518118770798e-05,
+      "loss": 0.4415,
+      "step": 2937
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.287002809549533e-05,
+      "loss": 0.4736,
+      "step": 2938
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.286553744158636e-05,
+      "loss": 0.4398,
+      "step": 2939
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2861046158030625e-05,
+      "loss": 0.4427,
+      "step": 2940
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2856554245814996e-05,
+      "loss": 0.4454,
+      "step": 2941
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2852061705926483e-05,
+      "loss": 0.4057,
+      "step": 2942
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2847568539352237e-05,
+      "loss": 0.4285,
+      "step": 2943
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2843074747079544e-05,
+      "loss": 0.4352,
+      "step": 2944
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.283858033009583e-05,
+      "loss": 0.441,
+      "step": 2945
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2834085289388658e-05,
+      "loss": 0.4168,
+      "step": 2946
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2829589625945722e-05,
+      "loss": 0.44,
+      "step": 2947
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2825093340754859e-05,
+      "loss": 0.4419,
+      "step": 2948
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2820596434804044e-05,
+      "loss": 0.4315,
+      "step": 2949
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2816098909081384e-05,
+      "loss": 0.4476,
+      "step": 2950
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2811600764575128e-05,
+      "loss": 0.4245,
+      "step": 2951
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.280710200227365e-05,
+      "loss": 0.4325,
+      "step": 2952
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.280260262316547e-05,
+      "loss": 0.4366,
+      "step": 2953
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2798102628239237e-05,
+      "loss": 0.4323,
+      "step": 2954
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2793602018483742e-05,
+      "loss": 0.4384,
+      "step": 2955
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2789100794887906e-05,
+      "loss": 0.4561,
+      "step": 2956
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2784598958440786e-05,
+      "loss": 0.4218,
+      "step": 2957
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2780096510131576e-05,
+      "loss": 0.4412,
+      "step": 2958
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2775593450949603e-05,
+      "loss": 0.4513,
+      "step": 2959
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2771089781884325e-05,
+      "loss": 0.4325,
+      "step": 2960
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2766585503925335e-05,
+      "loss": 0.4313,
+      "step": 2961
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2762080618062365e-05,
+      "loss": 0.4169,
+      "step": 2962
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2757575125285278e-05,
+      "loss": 0.4247,
+      "step": 2963
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2753069026584065e-05,
+      "loss": 0.4296,
+      "step": 2964
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2748562322948856e-05,
+      "loss": 0.4265,
+      "step": 2965
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2744055015369917e-05,
+      "loss": 0.4217,
+      "step": 2966
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2739547104837638e-05,
+      "loss": 0.4268,
+      "step": 2967
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2735038592342544e-05,
+      "loss": 0.4118,
+      "step": 2968
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.27305294788753e-05,
+      "loss": 0.4374,
+      "step": 2969
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.272601976542669e-05,
+      "loss": 0.427,
+      "step": 2970
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2721509452987641e-05,
+      "loss": 0.4216,
+      "step": 2971
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2716998542549206e-05,
+      "loss": 0.4389,
+      "step": 2972
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2712487035102572e-05,
+      "loss": 0.4389,
+      "step": 2973
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2707974931639058e-05,
+      "loss": 0.4403,
+      "step": 2974
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.270346223315011e-05,
+      "loss": 0.4475,
+      "step": 2975
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2698948940627302e-05,
+      "loss": 0.4311,
+      "step": 2976
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2694435055062352e-05,
+      "loss": 0.4418,
+      "step": 2977
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.268992057744709e-05,
+      "loss": 0.4322,
+      "step": 2978
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2685405508773493e-05,
+      "loss": 0.457,
+      "step": 2979
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.268088985003366e-05,
+      "loss": 0.4254,
+      "step": 2980
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2676373602219818e-05,
+      "loss": 0.4198,
+      "step": 2981
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2671856766324329e-05,
+      "loss": 0.4224,
+      "step": 2982
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2667339343339674e-05,
+      "loss": 0.4081,
+      "step": 2983
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2662821334258472e-05,
+      "loss": 0.4426,
+      "step": 2984
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2658302740073473e-05,
+      "loss": 0.4339,
+      "step": 2985
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.265378356177755e-05,
+      "loss": 0.4327,
+      "step": 2986
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2649263800363705e-05,
+      "loss": 0.4171,
+      "step": 2987
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2644743456825065e-05,
+      "loss": 0.4069,
+      "step": 2988
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2640222532154893e-05,
+      "loss": 0.406,
+      "step": 2989
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2635701027346573e-05,
+      "loss": 0.4283,
+      "step": 2990
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2631178943393623e-05,
+      "loss": 0.4191,
+      "step": 2991
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2626656281289678e-05,
+      "loss": 0.4423,
+      "step": 2992
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.262213304202851e-05,
+      "loss": 0.4392,
+      "step": 2993
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2617609226604011e-05,
+      "loss": 0.4344,
+      "step": 2994
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2613084836010203e-05,
+      "loss": 0.4206,
+      "step": 2995
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2608559871241236e-05,
+      "loss": 0.441,
+      "step": 2996
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2604034333291386e-05,
+      "loss": 0.4185,
+      "step": 2997
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2599508223155049e-05,
+      "loss": 0.4339,
+      "step": 2998
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.259498154182675e-05,
+      "loss": 0.429,
+      "step": 2999
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2590454290301142e-05,
+      "loss": 0.4168,
+      "step": 3000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2585926469573e-05,
+      "loss": 0.4177,
+      "step": 3001
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2581398080637229e-05,
+      "loss": 0.4482,
+      "step": 3002
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.257686912448885e-05,
+      "loss": 0.4348,
+      "step": 3003
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2572339602123023e-05,
+      "loss": 0.4322,
+      "step": 3004
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2567809514535017e-05,
+      "loss": 0.4316,
+      "step": 3005
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.256327886272023e-05,
+      "loss": 0.4246,
+      "step": 3006
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2558747647674188e-05,
+      "loss": 0.4225,
+      "step": 3007
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2554215870392539e-05,
+      "loss": 0.4267,
+      "step": 3008
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.254968353187105e-05,
+      "loss": 0.4069,
+      "step": 3009
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2545150633105623e-05,
+      "loss": 0.4337,
+      "step": 3010
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2540617175092274e-05,
+      "loss": 0.4375,
+      "step": 3011
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2536083158827135e-05,
+      "loss": 0.417,
+      "step": 3012
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2531548585306477e-05,
+      "loss": 0.4576,
+      "step": 3013
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.252701345552668e-05,
+      "loss": 0.4321,
+      "step": 3014
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2522477770484257e-05,
+      "loss": 0.4261,
+      "step": 3015
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2517941531175833e-05,
+      "loss": 0.435,
+      "step": 3016
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2513404738598163e-05,
+      "loss": 0.4487,
+      "step": 3017
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2508867393748117e-05,
+      "loss": 0.426,
+      "step": 3018
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2504329497622689e-05,
+      "loss": 0.4351,
+      "step": 3019
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2499791051218999e-05,
+      "loss": 0.4542,
+      "step": 3020
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2495252055534278e-05,
+      "loss": 0.4401,
+      "step": 3021
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2490712511565884e-05,
+      "loss": 0.4173,
+      "step": 3022
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.24861724203113e-05,
+      "loss": 0.4494,
+      "step": 3023
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2481631782768116e-05,
+      "loss": 0.4124,
+      "step": 3024
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2477090599934054e-05,
+      "loss": 0.4412,
+      "step": 3025
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2472548872806951e-05,
+      "loss": 0.4324,
+      "step": 3026
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2468006602384768e-05,
+      "loss": 0.4304,
+      "step": 3027
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2463463789665573e-05,
+      "loss": 0.4321,
+      "step": 3028
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.245892043564757e-05,
+      "loss": 0.4282,
+      "step": 3029
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.245437654132907e-05,
+      "loss": 0.4233,
+      "step": 3030
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2449832107708504e-05,
+      "loss": 0.4084,
+      "step": 3031
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2445287135784427e-05,
+      "loss": 0.4165,
+      "step": 3032
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.244074162655551e-05,
+      "loss": 0.4231,
+      "step": 3033
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.243619558102054e-05,
+      "loss": 0.4225,
+      "step": 3034
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2431649000178425e-05,
+      "loss": 0.45,
+      "step": 3035
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2427101885028188e-05,
+      "loss": 0.4496,
+      "step": 3036
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2422554236568966e-05,
+      "loss": 0.3935,
+      "step": 3037
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.241800605580002e-05,
+      "loss": 0.4009,
+      "step": 3038
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2413457343720727e-05,
+      "loss": 0.4475,
+      "step": 3039
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.240890810133058e-05,
+      "loss": 0.4188,
+      "step": 3040
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2404358329629185e-05,
+      "loss": 0.4289,
+      "step": 3041
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2399808029616264e-05,
+      "loss": 0.4291,
+      "step": 3042
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2395257202291663e-05,
+      "loss": 0.4047,
+      "step": 3043
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2390705848655337e-05,
+      "loss": 0.4104,
+      "step": 3044
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.238615396970736e-05,
+      "loss": 0.4416,
+      "step": 3045
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2381601566447918e-05,
+      "loss": 0.4307,
+      "step": 3046
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2377048639877311e-05,
+      "loss": 0.4257,
+      "step": 3047
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2372495190995964e-05,
+      "loss": 0.4232,
+      "step": 3048
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2367941220804404e-05,
+      "loss": 0.4334,
+      "step": 3049
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2363386730303284e-05,
+      "loss": 0.4278,
+      "step": 3050
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2358831720493361e-05,
+      "loss": 0.4347,
+      "step": 3051
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2354276192375513e-05,
+      "loss": 0.4404,
+      "step": 3052
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.234972014695073e-05,
+      "loss": 0.4355,
+      "step": 3053
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2345163585220112e-05,
+      "loss": 0.4307,
+      "step": 3054
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.234060650818488e-05,
+      "loss": 0.4466,
+      "step": 3055
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2336048916846361e-05,
+      "loss": 0.4662,
+      "step": 3056
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2331490812206e-05,
+      "loss": 0.4493,
+      "step": 3057
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2326932195265355e-05,
+      "loss": 0.4294,
+      "step": 3058
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2322373067026092e-05,
+      "loss": 0.4421,
+      "step": 3059
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2317813428489988e-05,
+      "loss": 0.4485,
+      "step": 3060
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2313253280658942e-05,
+      "loss": 0.4443,
+      "step": 3061
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2308692624534955e-05,
+      "loss": 0.4409,
+      "step": 3062
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2304131461120145e-05,
+      "loss": 0.4158,
+      "step": 3063
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.229956979141674e-05,
+      "loss": 0.4346,
+      "step": 3064
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2295007616427081e-05,
+      "loss": 0.4243,
+      "step": 3065
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2290444937153616e-05,
+      "loss": 0.4235,
+      "step": 3066
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2285881754598905e-05,
+      "loss": 0.4252,
+      "step": 3067
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2281318069765623e-05,
+      "loss": 0.4265,
+      "step": 3068
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2276753883656547e-05,
+      "loss": 0.4363,
+      "step": 3069
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2272189197274576e-05,
+      "loss": 0.4494,
+      "step": 3070
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2267624011622705e-05,
+      "loss": 0.4359,
+      "step": 3071
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2263058327704052e-05,
+      "loss": 0.4397,
+      "step": 3072
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2258492146521832e-05,
+      "loss": 0.432,
+      "step": 3073
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2253925469079384e-05,
+      "loss": 0.427,
+      "step": 3074
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2249358296380139e-05,
+      "loss": 0.4386,
+      "step": 3075
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.224479062942765e-05,
+      "loss": 0.4041,
+      "step": 3076
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2240222469225572e-05,
+      "loss": 0.4168,
+      "step": 3077
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2235653816777671e-05,
+      "loss": 0.4297,
+      "step": 3078
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2231084673087822e-05,
+      "loss": 0.4328,
+      "step": 3079
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2226515039160003e-05,
+      "loss": 0.4333,
+      "step": 3080
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.222194491599831e-05,
+      "loss": 0.4475,
+      "step": 3081
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2217374304606936e-05,
+      "loss": 0.4301,
+      "step": 3082
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2212803205990179e-05,
+      "loss": 0.4254,
+      "step": 3083
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2208231621152455e-05,
+      "loss": 0.4154,
+      "step": 3084
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2203659551098286e-05,
+      "loss": 0.4116,
+      "step": 3085
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.219908699683229e-05,
+      "loss": 0.4355,
+      "step": 3086
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2194513959359201e-05,
+      "loss": 0.4038,
+      "step": 3087
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2189940439683859e-05,
+      "loss": 0.4365,
+      "step": 3088
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2185366438811204e-05,
+      "loss": 0.4273,
+      "step": 3089
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.218079195774628e-05,
+      "loss": 0.4614,
+      "step": 3090
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2176216997494246e-05,
+      "loss": 0.4295,
+      "step": 3091
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2171641559060364e-05,
+      "loss": 0.467,
+      "step": 3092
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2167065643449993e-05,
+      "loss": 0.4203,
+      "step": 3093
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.216248925166861e-05,
+      "loss": 0.4214,
+      "step": 3094
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2157912384721782e-05,
+      "loss": 0.4449,
+      "step": 3095
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2153335043615189e-05,
+      "loss": 0.4238,
+      "step": 3096
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2148757229354618e-05,
+      "loss": 0.4245,
+      "step": 3097
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2144178942945948e-05,
+      "loss": 0.4283,
+      "step": 3098
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2139600185395178e-05,
+      "loss": 0.4369,
+      "step": 3099
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2135020957708396e-05,
+      "loss": 0.4344,
+      "step": 3100
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2130441260891801e-05,
+      "loss": 0.4309,
+      "step": 3101
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2125861095951694e-05,
+      "loss": 0.4313,
+      "step": 3102
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2121280463894476e-05,
+      "loss": 0.4257,
+      "step": 3103
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2116699365726659e-05,
+      "loss": 0.4327,
+      "step": 3104
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2112117802454842e-05,
+      "loss": 0.4454,
+      "step": 3105
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2107535775085743e-05,
+      "loss": 0.424,
+      "step": 3106
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2102953284626171e-05,
+      "loss": 0.4216,
+      "step": 3107
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.209837033208304e-05,
+      "loss": 0.4286,
+      "step": 3108
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2093786918463365e-05,
+      "loss": 0.4583,
+      "step": 3109
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2089203044774268e-05,
+      "loss": 0.4089,
+      "step": 3110
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2084618712022963e-05,
+      "loss": 0.4351,
+      "step": 3111
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.208003392121677e-05,
+      "loss": 0.4166,
+      "step": 3112
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2075448673363108e-05,
+      "loss": 0.4349,
+      "step": 3113
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2070862969469497e-05,
+      "loss": 0.4193,
+      "step": 3114
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2066276810543555e-05,
+      "loss": 0.4402,
+      "step": 3115
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2061690197593008e-05,
+      "loss": 0.4276,
+      "step": 3116
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.205710313162567e-05,
+      "loss": 0.4447,
+      "step": 3117
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.205251561364947e-05,
+      "loss": 0.44,
+      "step": 3118
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2047927644672416e-05,
+      "loss": 0.4156,
+      "step": 3119
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2043339225702632e-05,
+      "loss": 0.44,
+      "step": 3120
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2038750357748332e-05,
+      "loss": 0.4371,
+      "step": 3121
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2034161041817833e-05,
+      "loss": 0.4099,
+      "step": 3122
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.202957127891955e-05,
+      "loss": 0.4381,
+      "step": 3123
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2024981070061994e-05,
+      "loss": 0.4248,
+      "step": 3124
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2020390416253773e-05,
+      "loss": 0.4199,
+      "step": 3125
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2015799318503599e-05,
+      "loss": 0.4271,
+      "step": 3126
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2011207777820274e-05,
+      "loss": 0.4273,
+      "step": 3127
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2006615795212704e-05,
+      "loss": 0.4157,
+      "step": 3128
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2002023371689884e-05,
+      "loss": 0.4379,
+      "step": 3129
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1997430508260917e-05,
+      "loss": 0.4183,
+      "step": 3130
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1992837205934993e-05,
+      "loss": 0.43,
+      "step": 3131
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.19882434657214e-05,
+      "loss": 0.4265,
+      "step": 3132
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1983649288629528e-05,
+      "loss": 0.4307,
+      "step": 3133
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1979054675668859e-05,
+      "loss": 0.4226,
+      "step": 3134
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1974459627848969e-05,
+      "loss": 0.4308,
+      "step": 3135
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1969864146179533e-05,
+      "loss": 0.4295,
+      "step": 3136
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1965268231670316e-05,
+      "loss": 0.4291,
+      "step": 3137
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1960671885331182e-05,
+      "loss": 0.4328,
+      "step": 3138
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1956075108172096e-05,
+      "loss": 0.4237,
+      "step": 3139
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1951477901203106e-05,
+      "loss": 0.451,
+      "step": 3140
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1946880265434363e-05,
+      "loss": 0.4149,
+      "step": 3141
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1942282201876108e-05,
+      "loss": 0.4538,
+      "step": 3142
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1937683711538676e-05,
+      "loss": 0.4401,
+      "step": 3143
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1933084795432497e-05,
+      "loss": 0.405,
+      "step": 3144
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1928485454568093e-05,
+      "loss": 0.4362,
+      "step": 3145
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1923885689956085e-05,
+      "loss": 0.4272,
+      "step": 3146
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.191928550260718e-05,
+      "loss": 0.4597,
+      "step": 3147
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1914684893532182e-05,
+      "loss": 0.4205,
+      "step": 3148
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1910083863741983e-05,
+      "loss": 0.4298,
+      "step": 3149
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1905482414247577e-05,
+      "loss": 0.4296,
+      "step": 3150
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1900880546060039e-05,
+      "loss": 0.4318,
+      "step": 3151
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1896278260190542e-05,
+      "loss": 0.4218,
+      "step": 3152
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1891675557650351e-05,
+      "loss": 0.416,
+      "step": 3153
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.188707243945082e-05,
+      "loss": 0.4216,
+      "step": 3154
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1882468906603396e-05,
+      "loss": 0.4273,
+      "step": 3155
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1877864960119618e-05,
+      "loss": 0.4049,
+      "step": 3156
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1873260601011119e-05,
+      "loss": 0.4279,
+      "step": 3157
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1868655830289609e-05,
+      "loss": 0.4465,
+      "step": 3158
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1864050648966904e-05,
+      "loss": 0.4313,
+      "step": 3159
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.18594450580549e-05,
+      "loss": 0.4378,
+      "step": 3160
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.185483905856559e-05,
+      "loss": 0.4308,
+      "step": 3161
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1850232651511053e-05,
+      "loss": 0.4272,
+      "step": 3162
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.184562583790346e-05,
+      "loss": 0.4481,
+      "step": 3163
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1841018618755067e-05,
+      "loss": 0.4509,
+      "step": 3164
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1836410995078226e-05,
+      "loss": 0.4143,
+      "step": 3165
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1831802967885373e-05,
+      "loss": 0.4087,
+      "step": 3166
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1827194538189027e-05,
+      "loss": 0.4433,
+      "step": 3167
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1822585707001808e-05,
+      "loss": 0.4281,
+      "step": 3168
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1817976475336415e-05,
+      "loss": 0.4308,
+      "step": 3169
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.181336684420564e-05,
+      "loss": 0.4368,
+      "step": 3170
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1808756814622362e-05,
+      "loss": 0.439,
+      "step": 3171
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1804146387599544e-05,
+      "loss": 0.4215,
+      "step": 3172
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.179953556415024e-05,
+      "loss": 0.4536,
+      "step": 3173
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1794924345287589e-05,
+      "loss": 0.4305,
+      "step": 3174
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1790312732024815e-05,
+      "loss": 0.4213,
+      "step": 3175
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1785700725375236e-05,
+      "loss": 0.4341,
+      "step": 3176
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.178108832635225e-05,
+      "loss": 0.4468,
+      "step": 3177
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1776475535969343e-05,
+      "loss": 0.4365,
+      "step": 3178
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1771862355240086e-05,
+      "loss": 0.4238,
+      "step": 3179
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.176724878517814e-05,
+      "loss": 0.4311,
+      "step": 3180
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1762634826797248e-05,
+      "loss": 0.4094,
+      "step": 3181
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1758020481111232e-05,
+      "loss": 0.4487,
+      "step": 3182
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1753405749134015e-05,
+      "loss": 0.4392,
+      "step": 3183
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1748790631879588e-05,
+      "loss": 0.412,
+      "step": 3184
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1744175130362038e-05,
+      "loss": 0.4347,
+      "step": 3185
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1739559245595532e-05,
+      "loss": 0.4085,
+      "step": 3186
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1734942978594323e-05,
+      "loss": 0.409,
+      "step": 3187
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1730326330372748e-05,
+      "loss": 0.4493,
+      "step": 3188
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1725709301945224e-05,
+      "loss": 0.4156,
+      "step": 3189
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1721091894326253e-05,
+      "loss": 0.435,
+      "step": 3190
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1716474108530424e-05,
+      "loss": 0.45,
+      "step": 3191
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1711855945572407e-05,
+      "loss": 0.4179,
+      "step": 3192
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1707237406466954e-05,
+      "loss": 0.4243,
+      "step": 3193
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1702618492228899e-05,
+      "loss": 0.432,
+      "step": 3194
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1697999203873162e-05,
+      "loss": 0.4239,
+      "step": 3195
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1693379542414744e-05,
+      "loss": 0.4374,
+      "step": 3196
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1688759508868721e-05,
+      "loss": 0.437,
+      "step": 3197
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1684139104250258e-05,
+      "loss": 0.4258,
+      "step": 3198
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1679518329574604e-05,
+      "loss": 0.43,
+      "step": 3199
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1674897185857083e-05,
+      "loss": 0.4351,
+      "step": 3200
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1670275674113105e-05,
+      "loss": 0.3949,
+      "step": 3201
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1665653795358155e-05,
+      "loss": 0.4355,
+      "step": 3202
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.16610315506078e-05,
+      "loss": 0.42,
+      "step": 3203
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1656408940877697e-05,
+      "loss": 0.4147,
+      "step": 3204
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.165178596718357e-05,
+      "loss": 0.4332,
+      "step": 3205
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1647162630541226e-05,
+      "loss": 0.4337,
+      "step": 3206
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1642538931966565e-05,
+      "loss": 0.4356,
+      "step": 3207
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1637914872475544e-05,
+      "loss": 0.4242,
+      "step": 3208
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1633290453084215e-05,
+      "loss": 0.4424,
+      "step": 3209
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.162866567480871e-05,
+      "loss": 0.43,
+      "step": 3210
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1624040538665233e-05,
+      "loss": 0.4477,
+      "step": 3211
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1619415045670064e-05,
+      "loss": 0.4507,
+      "step": 3212
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1614789196839573e-05,
+      "loss": 0.4395,
+      "step": 3213
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1610162993190195e-05,
+      "loss": 0.401,
+      "step": 3214
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1605536435738452e-05,
+      "loss": 0.4213,
+      "step": 3215
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1600909525500944e-05,
+      "loss": 0.4113,
+      "step": 3216
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1596282263494339e-05,
+      "loss": 0.4306,
+      "step": 3217
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1591654650735395e-05,
+      "loss": 0.4194,
+      "step": 3218
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1587026688240938e-05,
+      "loss": 0.4684,
+      "step": 3219
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1582398377027877e-05,
+      "loss": 0.447,
+      "step": 3220
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1577769718113189e-05,
+      "loss": 0.4167,
+      "step": 3221
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1573140712513938e-05,
+      "loss": 0.4218,
+      "step": 3222
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1568511361247254e-05,
+      "loss": 0.433,
+      "step": 3223
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1563881665330352e-05,
+      "loss": 0.4246,
+      "step": 3224
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1559251625780517e-05,
+      "loss": 0.4251,
+      "step": 3225
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1554621243615112e-05,
+      "loss": 0.4302,
+      "step": 3226
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1549990519851574e-05,
+      "loss": 0.4293,
+      "step": 3227
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1545359455507416e-05,
+      "loss": 0.418,
+      "step": 3228
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1540728051600223e-05,
+      "loss": 0.4231,
+      "step": 3229
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1536096309147661e-05,
+      "loss": 0.4165,
+      "step": 3230
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1531464229167464e-05,
+      "loss": 0.4361,
+      "step": 3231
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1526831812677441e-05,
+      "loss": 0.4269,
+      "step": 3232
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1522199060695478e-05,
+      "loss": 0.4188,
+      "step": 3233
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1517565974239538e-05,
+      "loss": 0.4335,
+      "step": 3234
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1512932554327643e-05,
+      "loss": 0.4206,
+      "step": 3235
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1508298801977907e-05,
+      "loss": 0.4443,
+      "step": 3236
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.15036647182085e-05,
+      "loss": 0.4335,
+      "step": 3237
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1499030304037679e-05,
+      "loss": 0.3959,
+      "step": 3238
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1494395560483764e-05,
+      "loss": 0.4262,
+      "step": 3239
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1489760488565153e-05,
+      "loss": 0.4318,
+      "step": 3240
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1485125089300315e-05,
+      "loss": 0.4212,
+      "step": 3241
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1480489363707785e-05,
+      "loss": 0.411,
+      "step": 3242
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1475853312806182e-05,
+      "loss": 0.4204,
+      "step": 3243
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.147121693761418e-05,
+      "loss": 0.4295,
+      "step": 3244
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.146658023915054e-05,
+      "loss": 0.4087,
+      "step": 3245
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1461943218434083e-05,
+      "loss": 0.4312,
+      "step": 3246
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1457305876483708e-05,
+      "loss": 0.4168,
+      "step": 3247
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1452668214318386e-05,
+      "loss": 0.4141,
+      "step": 3248
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1448030232957144e-05,
+      "loss": 0.4404,
+      "step": 3249
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1443391933419097e-05,
+      "loss": 0.4343,
+      "step": 3250
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1438753316723416e-05,
+      "loss": 0.3963,
+      "step": 3251
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1434114383889355e-05,
+      "loss": 0.4322,
+      "step": 3252
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1429475135936226e-05,
+      "loss": 0.4215,
+      "step": 3253
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1424835573883418e-05,
+      "loss": 0.4032,
+      "step": 3254
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1420195698750378e-05,
+      "loss": 0.4238,
+      "step": 3255
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1415555511556638e-05,
+      "loss": 0.4323,
+      "step": 3256
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1410915013321786e-05,
+      "loss": 0.4444,
+      "step": 3257
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1406274205065482e-05,
+      "loss": 0.4163,
+      "step": 3258
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1401633087807453e-05,
+      "loss": 0.4498,
+      "step": 3259
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1396991662567501e-05,
+      "loss": 0.4299,
+      "step": 3260
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1392349930365484e-05,
+      "loss": 0.408,
+      "step": 3261
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1387707892221334e-05,
+      "loss": 0.4068,
+      "step": 3262
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1383065549155052e-05,
+      "loss": 0.4245,
+      "step": 3263
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1378422902186703e-05,
+      "loss": 0.4333,
+      "step": 3264
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1373779952336424e-05,
+      "loss": 0.4273,
+      "step": 3265
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1369136700624406e-05,
+      "loss": 0.4247,
+      "step": 3266
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1364493148070916e-05,
+      "loss": 0.4223,
+      "step": 3267
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1359849295696288e-05,
+      "loss": 0.4386,
+      "step": 3268
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1355205144520917e-05,
+      "loss": 0.4226,
+      "step": 3269
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1350560695565269e-05,
+      "loss": 0.4351,
+      "step": 3270
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.134591594984987e-05,
+      "loss": 0.4319,
+      "step": 3271
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1341270908395318e-05,
+      "loss": 0.4339,
+      "step": 3272
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1336625572222267e-05,
+      "loss": 0.4157,
+      "step": 3273
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1331979942351437e-05,
+      "loss": 0.4342,
+      "step": 3274
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1327334019803622e-05,
+      "loss": 0.4452,
+      "step": 3275
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1322687805599672e-05,
+      "loss": 0.4441,
+      "step": 3276
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1318041300760503e-05,
+      "loss": 0.4221,
+      "step": 3277
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1313394506307098e-05,
+      "loss": 0.4333,
+      "step": 3278
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1308747423260496e-05,
+      "loss": 0.4308,
+      "step": 3279
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1304100052641806e-05,
+      "loss": 0.4406,
+      "step": 3280
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1299452395472202e-05,
+      "loss": 0.4138,
+      "step": 3281
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.129480445277291e-05,
+      "loss": 0.4304,
+      "step": 3282
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1290156225565232e-05,
+      "loss": 0.4069,
+      "step": 3283
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1285507714870525e-05,
+      "loss": 0.4397,
+      "step": 3284
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1280858921710209e-05,
+      "loss": 0.4448,
+      "step": 3285
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1276209847105769e-05,
+      "loss": 0.4264,
+      "step": 3286
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1271560492078745e-05,
+      "loss": 0.449,
+      "step": 3287
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.126691085765075e-05,
+      "loss": 0.4466,
+      "step": 3288
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1262260944843446e-05,
+      "loss": 0.4375,
+      "step": 3289
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1257610754678568e-05,
+      "loss": 0.4309,
+      "step": 3290
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1252960288177901e-05,
+      "loss": 0.4398,
+      "step": 3291
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1248309546363293e-05,
+      "loss": 0.4252,
+      "step": 3292
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1243658530256663e-05,
+      "loss": 0.421,
+      "step": 3293
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1239007240879978e-05,
+      "loss": 0.4331,
+      "step": 3294
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.123435567925527e-05,
+      "loss": 0.4558,
+      "step": 3295
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1229703846404632e-05,
+      "loss": 0.4176,
+      "step": 3296
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.122505174335021e-05,
+      "loss": 0.4328,
+      "step": 3297
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1220399371114219e-05,
+      "loss": 0.4403,
+      "step": 3298
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1215746730718924e-05,
+      "loss": 0.443,
+      "step": 3299
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1211093823186657e-05,
+      "loss": 0.4367,
+      "step": 3300
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1206440649539802e-05,
+      "loss": 0.4196,
+      "step": 3301
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1201787210800811e-05,
+      "loss": 0.4094,
+      "step": 3302
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1197133507992181e-05,
+      "loss": 0.4217,
+      "step": 3303
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1192479542136476e-05,
+      "loss": 0.4224,
+      "step": 3304
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1187825314256313e-05,
+      "loss": 0.4339,
+      "step": 3305
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1183170825374372e-05,
+      "loss": 0.4148,
+      "step": 3306
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1178516076513391e-05,
+      "loss": 0.4162,
+      "step": 3307
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1173861068696152e-05,
+      "loss": 0.4172,
+      "step": 3308
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1169205802945511e-05,
+      "loss": 0.4245,
+      "step": 3309
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1164550280284369e-05,
+      "loss": 0.4277,
+      "step": 3310
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1159894501735695e-05,
+      "loss": 0.4221,
+      "step": 3311
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1155238468322497e-05,
+      "loss": 0.4182,
+      "step": 3312
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1150582181067855e-05,
+      "loss": 0.4239,
+      "step": 3313
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1145925640994896e-05,
+      "loss": 0.4164,
+      "step": 3314
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1141268849126808e-05,
+      "loss": 0.4149,
+      "step": 3315
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1136611806486826e-05,
+      "loss": 0.4241,
+      "step": 3316
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.113195451409825e-05,
+      "loss": 0.4342,
+      "step": 3317
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1127296972984434e-05,
+      "loss": 0.4206,
+      "step": 3318
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1122639184168777e-05,
+      "loss": 0.4232,
+      "step": 3319
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1117981148674737e-05,
+      "loss": 0.4074,
+      "step": 3320
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1113322867525833e-05,
+      "loss": 0.4266,
+      "step": 3321
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1108664341745631e-05,
+      "loss": 0.4234,
+      "step": 3322
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1104005572357753e-05,
+      "loss": 0.4435,
+      "step": 3323
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1099346560385871e-05,
+      "loss": 0.425,
+      "step": 3324
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.109468730685372e-05,
+      "loss": 0.416,
+      "step": 3325
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1090027812785072e-05,
+      "loss": 0.4375,
+      "step": 3326
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.108536807920377e-05,
+      "loss": 0.4411,
+      "step": 3327
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1080708107133695e-05,
+      "loss": 0.4493,
+      "step": 3328
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1076047897598789e-05,
+      "loss": 0.4275,
+      "step": 3329
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1071387451623044e-05,
+      "loss": 0.4126,
+      "step": 3330
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1066726770230503e-05,
+      "loss": 0.4186,
+      "step": 3331
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.106206585444526e-05,
+      "loss": 0.4328,
+      "step": 3332
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.105740470529146e-05,
+      "loss": 0.4161,
+      "step": 3333
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1052743323793305e-05,
+      "loss": 0.4242,
+      "step": 3334
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1048081710975041e-05,
+      "loss": 0.4102,
+      "step": 3335
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1043419867860968e-05,
+      "loss": 0.422,
+      "step": 3336
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1038757795475438e-05,
+      "loss": 0.4565,
+      "step": 3337
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1034095494842851e-05,
+      "loss": 0.4483,
+      "step": 3338
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1029432966987658e-05,
+      "loss": 0.4291,
+      "step": 3339
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1024770212934358e-05,
+      "loss": 0.4571,
+      "step": 3340
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1020107233707509e-05,
+      "loss": 0.4052,
+      "step": 3341
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1015444030331701e-05,
+      "loss": 0.4183,
+      "step": 3342
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.101078060383159e-05,
+      "loss": 0.4355,
+      "step": 3343
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1006116955231869e-05,
+      "loss": 0.4197,
+      "step": 3344
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.100145308555729e-05,
+      "loss": 0.4148,
+      "step": 3345
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0996788995832648e-05,
+      "loss": 0.4365,
+      "step": 3346
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0992124687082784e-05,
+      "loss": 0.4232,
+      "step": 3347
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0987460160332593e-05,
+      "loss": 0.433,
+      "step": 3348
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0982795416607013e-05,
+      "loss": 0.408,
+      "step": 3349
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0978130456931035e-05,
+      "loss": 0.4223,
+      "step": 3350
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0973465282329692e-05,
+      "loss": 0.4334,
+      "step": 3351
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0968799893828066e-05,
+      "loss": 0.4446,
+      "step": 3352
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0964134292451285e-05,
+      "loss": 0.4279,
+      "step": 3353
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.095946847922453e-05,
+      "loss": 0.4104,
+      "step": 3354
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0954802455173022e-05,
+      "loss": 0.4481,
+      "step": 3355
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0950136221322025e-05,
+      "loss": 0.4272,
+      "step": 3356
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0945469778696862e-05,
+      "loss": 0.4306,
+      "step": 3357
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.094080312832289e-05,
+      "loss": 0.4359,
+      "step": 3358
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0936136271225515e-05,
+      "loss": 0.4232,
+      "step": 3359
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0931469208430187e-05,
+      "loss": 0.4389,
+      "step": 3360
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0926801940962412e-05,
+      "loss": 0.4339,
+      "step": 3361
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0922134469847725e-05,
+      "loss": 0.4351,
+      "step": 3362
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0917466796111716e-05,
+      "loss": 0.4249,
+      "step": 3363
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0912798920780014e-05,
+      "loss": 0.4348,
+      "step": 3364
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0908130844878296e-05,
+      "loss": 0.4096,
+      "step": 3365
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0903462569432282e-05,
+      "loss": 0.4271,
+      "step": 3366
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.089879409546774e-05,
+      "loss": 0.4096,
+      "step": 3367
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0894125424010468e-05,
+      "loss": 0.4232,
+      "step": 3368
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0889456556086325e-05,
+      "loss": 0.3966,
+      "step": 3369
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.08847874927212e-05,
+      "loss": 0.4391,
+      "step": 3370
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0880118234941034e-05,
+      "loss": 0.4152,
+      "step": 3371
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0875448783771808e-05,
+      "loss": 0.4346,
+      "step": 3372
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0870779140239539e-05,
+      "loss": 0.4455,
+      "step": 3373
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.086610930537029e-05,
+      "loss": 0.4269,
+      "step": 3374
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0861439280190173e-05,
+      "loss": 0.4467,
+      "step": 3375
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0856769065725335e-05,
+      "loss": 0.4449,
+      "step": 3376
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0852098663001965e-05,
+      "loss": 0.4052,
+      "step": 3377
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.084742807304629e-05,
+      "loss": 0.43,
+      "step": 3378
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0842757296884592e-05,
+      "loss": 0.4139,
+      "step": 3379
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0838086335543178e-05,
+      "loss": 0.4303,
+      "step": 3380
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0833415190048403e-05,
+      "loss": 0.4341,
+      "step": 3381
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0828743861426659e-05,
+      "loss": 0.3904,
+      "step": 3382
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.082407235070438e-05,
+      "loss": 0.4525,
+      "step": 3383
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0819400658908048e-05,
+      "loss": 0.4062,
+      "step": 3384
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0814728787064173e-05,
+      "loss": 0.4156,
+      "step": 3385
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0810056736199306e-05,
+      "loss": 0.4126,
+      "step": 3386
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0805384507340046e-05,
+      "loss": 0.4371,
+      "step": 3387
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0800712101513022e-05,
+      "loss": 0.4315,
+      "step": 3388
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0796039519744905e-05,
+      "loss": 0.4399,
+      "step": 3389
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0791366763062409e-05,
+      "loss": 0.418,
+      "step": 3390
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0786693832492279e-05,
+      "loss": 0.432,
+      "step": 3391
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.07820207290613e-05,
+      "loss": 0.4299,
+      "step": 3392
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0777347453796298e-05,
+      "loss": 0.4262,
+      "step": 3393
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0772674007724139e-05,
+      "loss": 0.4381,
+      "step": 3394
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0768000391871722e-05,
+      "loss": 0.4242,
+      "step": 3395
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0763326607265982e-05,
+      "loss": 0.4529,
+      "step": 3396
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0758652654933892e-05,
+      "loss": 0.4433,
+      "step": 3397
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0753978535902466e-05,
+      "loss": 0.4505,
+      "step": 3398
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0749304251198747e-05,
+      "loss": 0.4289,
+      "step": 3399
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0744629801849826e-05,
+      "loss": 0.4201,
+      "step": 3400
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.073995518888282e-05,
+      "loss": 0.4248,
+      "step": 3401
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.073528041332489e-05,
+      "loss": 0.4172,
+      "step": 3402
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.073060547620322e-05,
+      "loss": 0.4112,
+      "step": 3403
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0725930378545042e-05,
+      "loss": 0.4197,
+      "step": 3404
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0721255121377616e-05,
+      "loss": 0.433,
+      "step": 3405
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0716579705728245e-05,
+      "loss": 0.4402,
+      "step": 3406
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0711904132624257e-05,
+      "loss": 0.4213,
+      "step": 3407
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0707228403093023e-05,
+      "loss": 0.4395,
+      "step": 3408
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0702552518161941e-05,
+      "loss": 0.4201,
+      "step": 3409
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0697876478858451e-05,
+      "loss": 0.4421,
+      "step": 3410
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.069320028621002e-05,
+      "loss": 0.4284,
+      "step": 3411
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.068852394124415e-05,
+      "loss": 0.4221,
+      "step": 3412
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.068384744498838e-05,
+      "loss": 0.4348,
+      "step": 3413
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0679170798470282e-05,
+      "loss": 0.4016,
+      "step": 3414
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0674494002717456e-05,
+      "loss": 0.4371,
+      "step": 3415
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0669817058757536e-05,
+      "loss": 0.4202,
+      "step": 3416
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0665139967618196e-05,
+      "loss": 0.4434,
+      "step": 3417
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0660462730327137e-05,
+      "loss": 0.4495,
+      "step": 3418
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0655785347912086e-05,
+      "loss": 0.425,
+      "step": 3419
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0651107821400815e-05,
+      "loss": 0.4562,
+      "step": 3420
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0646430151821112e-05,
+      "loss": 0.4386,
+      "step": 3421
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0641752340200813e-05,
+      "loss": 0.4067,
+      "step": 3422
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0637074387567771e-05,
+      "loss": 0.4414,
+      "step": 3423
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.063239629494988e-05,
+      "loss": 0.4527,
+      "step": 3424
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0627718063375064e-05,
+      "loss": 0.4252,
+      "step": 3425
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0623039693871269e-05,
+      "loss": 0.4303,
+      "step": 3426
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0618361187466477e-05,
+      "loss": 0.4414,
+      "step": 3427
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.06136825451887e-05,
+      "loss": 0.4202,
+      "step": 3428
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0609003768065983e-05,
+      "loss": 0.4197,
+      "step": 3429
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0604324857126392e-05,
+      "loss": 0.4266,
+      "step": 3430
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0599645813398032e-05,
+      "loss": 0.4247,
+      "step": 3431
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0594966637909035e-05,
+      "loss": 0.417,
+      "step": 3432
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0590287331687554e-05,
+      "loss": 0.4536,
+      "step": 3433
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.058560789576178e-05,
+      "loss": 0.4507,
+      "step": 3434
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0580928331159927e-05,
+      "loss": 0.4133,
+      "step": 3435
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0576248638910242e-05,
+      "loss": 0.4406,
+      "step": 3436
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0571568820040995e-05,
+      "loss": 0.4204,
+      "step": 3437
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.056688887558049e-05,
+      "loss": 0.4465,
+      "step": 3438
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.056220880655705e-05,
+      "loss": 0.4152,
+      "step": 3439
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0557528613999033e-05,
+      "loss": 0.4169,
+      "step": 3440
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0552848298934822e-05,
+      "loss": 0.4019,
+      "step": 3441
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0548167862392824e-05,
+      "loss": 0.46,
+      "step": 3442
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0543487305401474e-05,
+      "loss": 0.4123,
+      "step": 3443
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.053880662898924e-05,
+      "loss": 0.4298,
+      "step": 3444
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0534125834184602e-05,
+      "loss": 0.441,
+      "step": 3445
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0529444922016083e-05,
+      "loss": 0.4264,
+      "step": 3446
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.052476389351222e-05,
+      "loss": 0.4077,
+      "step": 3447
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.052008274970158e-05,
+      "loss": 0.4393,
+      "step": 3448
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0515401491612753e-05,
+      "loss": 0.4409,
+      "step": 3449
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0510720120274356e-05,
+      "loss": 0.4149,
+      "step": 3450
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.050603863671503e-05,
+      "loss": 0.4194,
+      "step": 3451
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.050135704196344e-05,
+      "loss": 0.4278,
+      "step": 3452
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.049667533704828e-05,
+      "loss": 0.4518,
+      "step": 3453
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.049199352299826e-05,
+      "loss": 0.4296,
+      "step": 3454
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0487311600842125e-05,
+      "loss": 0.4267,
+      "step": 3455
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.048262957160863e-05,
+      "loss": 0.451,
+      "step": 3456
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.047794743632657e-05,
+      "loss": 0.4222,
+      "step": 3457
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0473265196024742e-05,
+      "loss": 0.4391,
+      "step": 3458
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0468582851731987e-05,
+      "loss": 0.4204,
+      "step": 3459
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0463900404477158e-05,
+      "loss": 0.4095,
+      "step": 3460
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0459217855289135e-05,
+      "loss": 0.4143,
+      "step": 3461
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0454535205196817e-05,
+      "loss": 0.4236,
+      "step": 3462
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0449852455229121e-05,
+      "loss": 0.4293,
+      "step": 3463
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0445169606415e-05,
+      "loss": 0.4347,
+      "step": 3464
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0440486659783413e-05,
+      "loss": 0.4259,
+      "step": 3465
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.043580361636335e-05,
+      "loss": 0.4261,
+      "step": 3466
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0431120477183821e-05,
+      "loss": 0.4165,
+      "step": 3467
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0426437243273855e-05,
+      "loss": 0.4159,
+      "step": 3468
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0421753915662503e-05,
+      "loss": 0.4285,
+      "step": 3469
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0417070495378834e-05,
+      "loss": 0.4119,
+      "step": 3470
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0412386983451942e-05,
+      "loss": 0.4344,
+      "step": 3471
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0407703380910938e-05,
+      "loss": 0.4295,
+      "step": 3472
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0403019688784952e-05,
+      "loss": 0.435,
+      "step": 3473
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0398335908103141e-05,
+      "loss": 0.4442,
+      "step": 3474
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0393652039894668e-05,
+      "loss": 0.4177,
+      "step": 3475
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0388968085188727e-05,
+      "loss": 0.4296,
+      "step": 3476
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0384284045014525e-05,
+      "loss": 0.4142,
+      "step": 3477
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0379599920401293e-05,
+      "loss": 0.436,
+      "step": 3478
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0374915712378277e-05,
+      "loss": 0.4488,
+      "step": 3479
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0370231421974744e-05,
+      "loss": 0.4169,
+      "step": 3480
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0365547050219966e-05,
+      "loss": 0.4499,
+      "step": 3481
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0360862598143253e-05,
+      "loss": 0.4271,
+      "step": 3482
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.035617806677392e-05,
+      "loss": 0.4198,
+      "step": 3483
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0351493457141306e-05,
+      "loss": 0.4307,
+      "step": 3484
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0346808770274763e-05,
+      "loss": 0.4372,
+      "step": 3485
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0342124007203657e-05,
+      "loss": 0.4077,
+      "step": 3486
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.033743916895738e-05,
+      "loss": 0.435,
+      "step": 3487
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0332754256565329e-05,
+      "loss": 0.4327,
+      "step": 3488
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0328069271056929e-05,
+      "loss": 0.4114,
+      "step": 3489
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0323384213461611e-05,
+      "loss": 0.4075,
+      "step": 3490
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0318699084808832e-05,
+      "loss": 0.4342,
+      "step": 3491
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0314013886128053e-05,
+      "loss": 0.426,
+      "step": 3492
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.030932861844876e-05,
+      "loss": 0.4224,
+      "step": 3493
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0304643282800451e-05,
+      "loss": 0.4288,
+      "step": 3494
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0299957880212634e-05,
+      "loss": 0.4109,
+      "step": 3495
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0295272411714841e-05,
+      "loss": 0.4209,
+      "step": 3496
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0290586878336611e-05,
+      "loss": 0.4274,
+      "step": 3497
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.02859012811075e-05,
+      "loss": 0.4223,
+      "step": 3498
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.028121562105708e-05,
+      "loss": 0.4507,
+      "step": 3499
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.027652989921493e-05,
+      "loss": 0.4108,
+      "step": 3500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0271844116610652e-05,
+      "loss": 0.4534,
+      "step": 3501
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0267158274273858e-05,
+      "loss": 0.4566,
+      "step": 3502
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0262472373234169e-05,
+      "loss": 0.4367,
+      "step": 3503
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0257786414521218e-05,
+      "loss": 0.4376,
+      "step": 3504
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.025310039916466e-05,
+      "loss": 0.4438,
+      "step": 3505
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0248414328194154e-05,
+      "loss": 0.4403,
+      "step": 3506
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0243728202639373e-05,
+      "loss": 0.4417,
+      "step": 3507
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0239042023530007e-05,
+      "loss": 0.4173,
+      "step": 3508
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0234355791895754e-05,
+      "loss": 0.426,
+      "step": 3509
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.022966950876632e-05,
+      "loss": 0.4283,
+      "step": 3510
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0224983175171424e-05,
+      "loss": 0.4355,
+      "step": 3511
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0220296792140801e-05,
+      "loss": 0.4373,
+      "step": 3512
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0215610360704191e-05,
+      "loss": 0.4354,
+      "step": 3513
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.021092388189135e-05,
+      "loss": 0.4272,
+      "step": 3514
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0206237356732042e-05,
+      "loss": 0.4288,
+      "step": 3515
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.020155078625604e-05,
+      "loss": 0.4185,
+      "step": 3516
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0196864171493127e-05,
+      "loss": 0.4219,
+      "step": 3517
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0192177513473096e-05,
+      "loss": 0.4372,
+      "step": 3518
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.018749081322575e-05,
+      "loss": 0.4507,
+      "step": 3519
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.01828040717809e-05,
+      "loss": 0.4452,
+      "step": 3520
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0178117290168375e-05,
+      "loss": 0.4233,
+      "step": 3521
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0173430469417996e-05,
+      "loss": 0.436,
+      "step": 3522
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0168743610559606e-05,
+      "loss": 0.4287,
+      "step": 3523
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0164056714623049e-05,
+      "loss": 0.4405,
+      "step": 3524
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0159369782638189e-05,
+      "loss": 0.4017,
+      "step": 3525
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0154682815634878e-05,
+      "loss": 0.4192,
+      "step": 3526
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0149995814642994e-05,
+      "loss": 0.4136,
+      "step": 3527
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0145308780692411e-05,
+      "loss": 0.4307,
+      "step": 3528
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0140621714813017e-05,
+      "loss": 0.4334,
+      "step": 3529
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0135934618034704e-05,
+      "loss": 0.4278,
+      "step": 3530
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0131247491387373e-05,
+      "loss": 0.4257,
+      "step": 3531
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.012656033590093e-05,
+      "loss": 0.4104,
+      "step": 3532
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0121873152605286e-05,
+      "loss": 0.413,
+      "step": 3533
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0117185942530358e-05,
+      "loss": 0.4386,
+      "step": 3534
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0112498706706071e-05,
+      "loss": 0.4147,
+      "step": 3535
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0107811446162357e-05,
+      "loss": 0.4193,
+      "step": 3536
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.010312416192915e-05,
+      "loss": 0.4216,
+      "step": 3537
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.009843685503639e-05,
+      "loss": 0.4388,
+      "step": 3538
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0093749526514027e-05,
+      "loss": 0.4208,
+      "step": 3539
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0089062177392003e-05,
+      "loss": 0.4215,
+      "step": 3540
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0084374808700283e-05,
+      "loss": 0.4247,
+      "step": 3541
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0079687421468816e-05,
+      "loss": 0.4223,
+      "step": 3542
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0075000016727572e-05,
+      "loss": 0.4109,
+      "step": 3543
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0070312595506516e-05,
+      "loss": 0.4429,
+      "step": 3544
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0065625158835619e-05,
+      "loss": 0.4501,
+      "step": 3545
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0060937707744854e-05,
+      "loss": 0.4191,
+      "step": 3546
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0056250243264199e-05,
+      "loss": 0.4034,
+      "step": 3547
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0051562766423637e-05,
+      "loss": 0.4278,
+      "step": 3548
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0046875278253146e-05,
+      "loss": 0.4253,
+      "step": 3549
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0042187779782713e-05,
+      "loss": 0.4495,
+      "step": 3550
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0037500272042329e-05,
+      "loss": 0.4336,
+      "step": 3551
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0032812756061977e-05,
+      "loss": 0.4425,
+      "step": 3552
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0028125232871654e-05,
+      "loss": 0.4363,
+      "step": 3553
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0023437703501353e-05,
+      "loss": 0.4109,
+      "step": 3554
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0018750168981068e-05,
+      "loss": 0.4156,
+      "step": 3555
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0014062630340793e-05,
+      "loss": 0.4218,
+      "step": 3556
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0009375088610527e-05,
+      "loss": 0.4405,
+      "step": 3557
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0004687544820264e-05,
+      "loss": 0.4363,
+      "step": 3558
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1e-05,
+      "loss": 0.4314,
+      "step": 3559
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.995312455179743e-06,
+      "loss": 0.4093,
+      "step": 3560
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.990624911389477e-06,
+      "loss": 0.4261,
+      "step": 3561
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.98593736965921e-06,
+      "loss": 0.4251,
+      "step": 3562
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.981249831018933e-06,
+      "loss": 0.4214,
+      "step": 3563
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.97656229649865e-06,
+      "loss": 0.4405,
+      "step": 3564
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.97187476712835e-06,
+      "loss": 0.4216,
+      "step": 3565
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.967187243938024e-06,
+      "loss": 0.4396,
+      "step": 3566
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.962499727957678e-06,
+      "loss": 0.4426,
+      "step": 3567
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.957812220217288e-06,
+      "loss": 0.4541,
+      "step": 3568
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.95312472174686e-06,
+      "loss": 0.4268,
+      "step": 3569
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.948437233576366e-06,
+      "loss": 0.415,
+      "step": 3570
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.943749756735805e-06,
+      "loss": 0.4218,
+      "step": 3571
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.939062292255151e-06,
+      "loss": 0.4311,
+      "step": 3572
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.934374841164384e-06,
+      "loss": 0.411,
+      "step": 3573
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.929687404493488e-06,
+      "loss": 0.4322,
+      "step": 3574
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.924999983272428e-06,
+      "loss": 0.4239,
+      "step": 3575
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.920312578531186e-06,
+      "loss": 0.4131,
+      "step": 3576
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.915625191299722e-06,
+      "loss": 0.4248,
+      "step": 3577
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.910937822607998e-06,
+      "loss": 0.4114,
+      "step": 3578
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.906250473485976e-06,
+      "loss": 0.4388,
+      "step": 3579
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.901563144963611e-06,
+      "loss": 0.4469,
+      "step": 3580
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.896875838070852e-06,
+      "loss": 0.4329,
+      "step": 3581
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.892188553837645e-06,
+      "loss": 0.4338,
+      "step": 3582
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.88750129329393e-06,
+      "loss": 0.4421,
+      "step": 3583
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.882814057469647e-06,
+      "loss": 0.4137,
+      "step": 3584
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.878126847394717e-06,
+      "loss": 0.411,
+      "step": 3585
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.873439664099073e-06,
+      "loss": 0.4242,
+      "step": 3586
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.868752508612627e-06,
+      "loss": 0.4395,
+      "step": 3587
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.864065381965298e-06,
+      "loss": 0.4086,
+      "step": 3588
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.859378285186988e-06,
+      "loss": 0.4399,
+      "step": 3589
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.854691219307594e-06,
+      "loss": 0.4128,
+      "step": 3590
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.85000418535701e-06,
+      "loss": 0.4528,
+      "step": 3591
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.845317184365124e-06,
+      "loss": 0.4248,
+      "step": 3592
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.840630217361814e-06,
+      "loss": 0.4333,
+      "step": 3593
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.835943285376951e-06,
+      "loss": 0.4259,
+      "step": 3594
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.831256389440397e-06,
+      "loss": 0.436,
+      "step": 3595
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.82656953058201e-06,
+      "loss": 0.4336,
+      "step": 3596
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.821882709831627e-06,
+      "loss": 0.4222,
+      "step": 3597
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.817195928219101e-06,
+      "loss": 0.4326,
+      "step": 3598
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.81250918677425e-06,
+      "loss": 0.4311,
+      "step": 3599
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.807822486526907e-06,
+      "loss": 0.4138,
+      "step": 3600
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.803135828506878e-06,
+      "loss": 0.4301,
+      "step": 3601
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.798449213743963e-06,
+      "loss": 0.4267,
+      "step": 3602
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.793762643267963e-06,
+      "loss": 0.4323,
+      "step": 3603
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.78907611810865e-06,
+      "loss": 0.4285,
+      "step": 3604
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.784389639295812e-06,
+      "loss": 0.427,
+      "step": 3605
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.7797032078592e-06,
+      "loss": 0.4249,
+      "step": 3606
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.775016824828579e-06,
+      "loss": 0.4192,
+      "step": 3607
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.770330491233685e-06,
+      "loss": 0.435,
+      "step": 3608
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.76564420810425e-06,
+      "loss": 0.4145,
+      "step": 3609
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.760957976469996e-06,
+      "loss": 0.4433,
+      "step": 3610
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.756271797360627e-06,
+      "loss": 0.427,
+      "step": 3611
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.75158567180585e-06,
+      "loss": 0.4301,
+      "step": 3612
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.746899600835346e-06,
+      "loss": 0.4149,
+      "step": 3613
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.742213585478784e-06,
+      "loss": 0.4104,
+      "step": 3614
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.737527626765836e-06,
+      "loss": 0.4517,
+      "step": 3615
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.732841725726145e-06,
+      "loss": 0.4197,
+      "step": 3616
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.728155883389351e-06,
+      "loss": 0.4261,
+      "step": 3617
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.723470100785071e-06,
+      "loss": 0.4247,
+      "step": 3618
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.718784378942924e-06,
+      "loss": 0.4103,
+      "step": 3619
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.714098718892506e-06,
+      "loss": 0.4288,
+      "step": 3620
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.70941312166339e-06,
+      "loss": 0.4329,
+      "step": 3621
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.704727588285164e-06,
+      "loss": 0.4333,
+      "step": 3622
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.700042119787369e-06,
+      "loss": 0.4112,
+      "step": 3623
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.695356717199552e-06,
+      "loss": 0.4378,
+      "step": 3624
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.690671381551244e-06,
+      "loss": 0.4427,
+      "step": 3625
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.68598611387195e-06,
+      "loss": 0.4418,
+      "step": 3626
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.681300915191173e-06,
+      "loss": 0.437,
+      "step": 3627
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.676615786538389e-06,
+      "loss": 0.4388,
+      "step": 3628
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.671930728943075e-06,
+      "loss": 0.4153,
+      "step": 3629
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.667245743434671e-06,
+      "loss": 0.4287,
+      "step": 3630
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.662560831042624e-06,
+      "loss": 0.4349,
+      "step": 3631
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.657875992796346e-06,
+      "loss": 0.4336,
+      "step": 3632
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.65319122972524e-06,
+      "loss": 0.4566,
+      "step": 3633
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.648506542858697e-06,
+      "loss": 0.4294,
+      "step": 3634
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.64382193322608e-06,
+      "loss": 0.4195,
+      "step": 3635
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.63913740185675e-06,
+      "loss": 0.4155,
+      "step": 3636
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.634452949780039e-06,
+      "loss": 0.4267,
+      "step": 3637
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.629768578025261e-06,
+      "loss": 0.4339,
+      "step": 3638
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.625084287621725e-06,
+      "loss": 0.415,
+      "step": 3639
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.620400079598707e-06,
+      "loss": 0.427,
+      "step": 3640
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.615715954985478e-06,
+      "loss": 0.429,
+      "step": 3641
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.611031914811273e-06,
+      "loss": 0.4354,
+      "step": 3642
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.606347960105335e-06,
+      "loss": 0.4302,
+      "step": 3643
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.601664091896866e-06,
+      "loss": 0.4123,
+      "step": 3644
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.596980311215048e-06,
+      "loss": 0.4051,
+      "step": 3645
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.592296619089065e-06,
+      "loss": 0.4342,
+      "step": 3646
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.58761301654806e-06,
+      "loss": 0.4068,
+      "step": 3647
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.58292950462117e-06,
+      "loss": 0.425,
+      "step": 3648
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.578246084337502e-06,
+      "loss": 0.4145,
+      "step": 3649
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.573562756726147e-06,
+      "loss": 0.4392,
+      "step": 3650
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.568879522816182e-06,
+      "loss": 0.4247,
+      "step": 3651
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.564196383636651e-06,
+      "loss": 0.4404,
+      "step": 3652
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.55951334021659e-06,
+      "loss": 0.4344,
+      "step": 3653
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.554830393585005e-06,
+      "loss": 0.438,
+      "step": 3654
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.550147544770882e-06,
+      "loss": 0.4306,
+      "step": 3655
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.54546479480319e-06,
+      "loss": 0.4359,
+      "step": 3656
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.540782144710865e-06,
+      "loss": 0.4268,
+      "step": 3657
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.536099595522843e-06,
+      "loss": 0.4347,
+      "step": 3658
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.531417148268013e-06,
+      "loss": 0.4296,
+      "step": 3659
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.52673480397526e-06,
+      "loss": 0.4145,
+      "step": 3660
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.522052563673437e-06,
+      "loss": 0.4244,
+      "step": 3661
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.517370428391371e-06,
+      "loss": 0.4473,
+      "step": 3662
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.512688399157878e-06,
+      "loss": 0.4231,
+      "step": 3663
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.50800647700174e-06,
+      "loss": 0.4413,
+      "step": 3664
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.503324662951724e-06,
+      "loss": 0.425,
+      "step": 3665
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.498642958036565e-06,
+      "loss": 0.4349,
+      "step": 3666
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.493961363284973e-06,
+      "loss": 0.4239,
+      "step": 3667
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.489279879725648e-06,
+      "loss": 0.4209,
+      "step": 3668
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.484598508387249e-06,
+      "loss": 0.4324,
+      "step": 3669
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.479917250298423e-06,
+      "loss": 0.4047,
+      "step": 3670
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.475236106487782e-06,
+      "loss": 0.4354,
+      "step": 3671
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.47055507798392e-06,
+      "loss": 0.3916,
+      "step": 3672
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.465874165815403e-06,
+      "loss": 0.4283,
+      "step": 3673
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.461193371010764e-06,
+      "loss": 0.415,
+      "step": 3674
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.45651269459853e-06,
+      "loss": 0.4201,
+      "step": 3675
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.451832137607178e-06,
+      "loss": 0.4211,
+      "step": 3676
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.44715170106518e-06,
+      "loss": 0.4085,
+      "step": 3677
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.44247138600097e-06,
+      "loss": 0.4244,
+      "step": 3678
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.437791193442952e-06,
+      "loss": 0.4195,
+      "step": 3679
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.433111124419514e-06,
+      "loss": 0.4209,
+      "step": 3680
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.428431179959005e-06,
+      "loss": 0.4299,
+      "step": 3681
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.42375136108976e-06,
+      "loss": 0.4208,
+      "step": 3682
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.419071668840071e-06,
+      "loss": 0.4186,
+      "step": 3683
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.414392104238221e-06,
+      "loss": 0.4108,
+      "step": 3684
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.40971266831245e-06,
+      "loss": 0.4161,
+      "step": 3685
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.405033362090968e-06,
+      "loss": 0.4324,
+      "step": 3686
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.40035418660197e-06,
+      "loss": 0.4345,
+      "step": 3687
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.395675142873608e-06,
+      "loss": 0.4314,
+      "step": 3688
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.39099623193402e-06,
+      "loss": 0.407,
+      "step": 3689
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.386317454811305e-06,
+      "loss": 0.4303,
+      "step": 3690
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.381638812533526e-06,
+      "loss": 0.432,
+      "step": 3691
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.376960306128735e-06,
+      "loss": 0.423,
+      "step": 3692
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.372281936624938e-06,
+      "loss": 0.4188,
+      "step": 3693
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.367603705050122e-06,
+      "loss": 0.4015,
+      "step": 3694
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.362925612432229e-06,
+      "loss": 0.4449,
+      "step": 3695
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.35824765979919e-06,
+      "loss": 0.4121,
+      "step": 3696
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.353569848178891e-06,
+      "loss": 0.4295,
+      "step": 3697
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.348892178599188e-06,
+      "loss": 0.4388,
+      "step": 3698
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.344214652087916e-06,
+      "loss": 0.4302,
+      "step": 3699
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.339537269672867e-06,
+      "loss": 0.4338,
+      "step": 3700
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.334860032381807e-06,
+      "loss": 0.4322,
+      "step": 3701
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.330182941242469e-06,
+      "loss": 0.4232,
+      "step": 3702
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.325505997282548e-06,
+      "loss": 0.4307,
+      "step": 3703
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.320829201529723e-06,
+      "loss": 0.4066,
+      "step": 3704
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.316152555011621e-06,
+      "loss": 0.4122,
+      "step": 3705
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.311476058755853e-06,
+      "loss": 0.4488,
+      "step": 3706
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.306799713789983e-06,
+      "loss": 0.4378,
+      "step": 3707
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.302123521141552e-06,
+      "loss": 0.4495,
+      "step": 3708
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.297447481838062e-06,
+      "loss": 0.4497,
+      "step": 3709
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.292771596906978e-06,
+      "loss": 0.4217,
+      "step": 3710
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.288095867375746e-06,
+      "loss": 0.4367,
+      "step": 3711
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.283420294271756e-06,
+      "loss": 0.4344,
+      "step": 3712
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.278744878622386e-06,
+      "loss": 0.4278,
+      "step": 3713
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.274069621454963e-06,
+      "loss": 0.4278,
+      "step": 3714
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.269394523796782e-06,
+      "loss": 0.3996,
+      "step": 3715
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.264719586675114e-06,
+      "loss": 0.4434,
+      "step": 3716
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.26004481111718e-06,
+      "loss": 0.4275,
+      "step": 3717
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.255370198150177e-06,
+      "loss": 0.4088,
+      "step": 3718
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.250695748801253e-06,
+      "loss": 0.4328,
+      "step": 3719
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.24602146409754e-06,
+      "loss": 0.4122,
+      "step": 3720
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.241347345066114e-06,
+      "loss": 0.4155,
+      "step": 3721
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.236673392734021e-06,
+      "loss": 0.4395,
+      "step": 3722
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.231999608128282e-06,
+      "loss": 0.4167,
+      "step": 3723
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.227325992275863e-06,
+      "loss": 0.4196,
+      "step": 3724
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.222652546203705e-06,
+      "loss": 0.4303,
+      "step": 3725
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.217979270938706e-06,
+      "loss": 0.445,
+      "step": 3726
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.213306167507725e-06,
+      "loss": 0.4159,
+      "step": 3727
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.208633236937594e-06,
+      "loss": 0.43,
+      "step": 3728
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.203960480255095e-06,
+      "loss": 0.4256,
+      "step": 3729
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.19928789848698e-06,
+      "loss": 0.4313,
+      "step": 3730
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.194615492659959e-06,
+      "loss": 0.4006,
+      "step": 3731
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.189943263800698e-06,
+      "loss": 0.4184,
+      "step": 3732
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.185271212935833e-06,
+      "loss": 0.4295,
+      "step": 3733
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.180599341091954e-06,
+      "loss": 0.4238,
+      "step": 3734
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.175927649295622e-06,
+      "loss": 0.4314,
+      "step": 3735
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.171256138573343e-06,
+      "loss": 0.4256,
+      "step": 3736
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.166584809951602e-06,
+      "loss": 0.4213,
+      "step": 3737
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.161913664456827e-06,
+      "loss": 0.4227,
+      "step": 3738
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.15724270311541e-06,
+      "loss": 0.4364,
+      "step": 3739
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.152571926953711e-06,
+      "loss": 0.4149,
+      "step": 3740
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.147901336998037e-06,
+      "loss": 0.4395,
+      "step": 3741
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.143230934274667e-06,
+      "loss": 0.4335,
+      "step": 3742
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.13856071980983e-06,
+      "loss": 0.4349,
+      "step": 3743
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.13389069462971e-06,
+      "loss": 0.4348,
+      "step": 3744
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.129220859760466e-06,
+      "loss": 0.4208,
+      "step": 3745
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.124551216228196e-06,
+      "loss": 0.4217,
+      "step": 3746
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.119881765058967e-06,
+      "loss": 0.4109,
+      "step": 3747
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.1152125072788e-06,
+      "loss": 0.4221,
+      "step": 3748
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.110543443913677e-06,
+      "loss": 0.4307,
+      "step": 3749
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.105874575989535e-06,
+      "loss": 0.4119,
+      "step": 3750
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.101205904532265e-06,
+      "loss": 0.4395,
+      "step": 3751
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.09653743056772e-06,
+      "loss": 0.4364,
+      "step": 3752
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.091869155121706e-06,
+      "loss": 0.41,
+      "step": 3753
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.087201079219989e-06,
+      "loss": 0.4415,
+      "step": 3754
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.082533203888288e-06,
+      "loss": 0.4336,
+      "step": 3755
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.077865530152278e-06,
+      "loss": 0.4398,
+      "step": 3756
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.073198059037593e-06,
+      "loss": 0.4175,
+      "step": 3757
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.068530791569811e-06,
+      "loss": 0.4445,
+      "step": 3758
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.063863728774488e-06,
+      "loss": 0.421,
+      "step": 3759
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.059196871677112e-06,
+      "loss": 0.4253,
+      "step": 3760
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.05453022130314e-06,
+      "loss": 0.4081,
+      "step": 3761
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.049863778677978e-06,
+      "loss": 0.4268,
+      "step": 3762
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.045197544826982e-06,
+      "loss": 0.4206,
+      "step": 3763
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.040531520775474e-06,
+      "loss": 0.419,
+      "step": 3764
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.035865707548715e-06,
+      "loss": 0.4145,
+      "step": 3765
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.031200106171939e-06,
+      "loss": 0.4322,
+      "step": 3766
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.026534717670313e-06,
+      "loss": 0.4244,
+      "step": 3767
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.021869543068966e-06,
+      "loss": 0.4273,
+      "step": 3768
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.017204583392989e-06,
+      "loss": 0.4206,
+      "step": 3769
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.012539839667408e-06,
+      "loss": 0.4229,
+      "step": 3770
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.00787531291722e-06,
+      "loss": 0.4375,
+      "step": 3771
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.003211004167354e-06,
+      "loss": 0.4122,
+      "step": 3772
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.998546914442712e-06,
+      "loss": 0.4277,
+      "step": 3773
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.993883044768135e-06,
+      "loss": 0.4284,
+      "step": 3774
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.989219396168413e-06,
+      "loss": 0.4286,
+      "step": 3775
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.984555969668302e-06,
+      "loss": 0.406,
+      "step": 3776
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.979892766292494e-06,
+      "loss": 0.4191,
+      "step": 3777
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.975229787065643e-06,
+      "loss": 0.418,
+      "step": 3778
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.970567033012347e-06,
+      "loss": 0.4076,
+      "step": 3779
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.96590450515715e-06,
+      "loss": 0.4209,
+      "step": 3780
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.961242204524567e-06,
+      "loss": 0.4357,
+      "step": 3781
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.956580132139034e-06,
+      "loss": 0.4168,
+      "step": 3782
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.951918289024964e-06,
+      "loss": 0.4375,
+      "step": 3783
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.9472566762067e-06,
+      "loss": 0.4024,
+      "step": 3784
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.942595294708545e-06,
+      "loss": 0.4189,
+      "step": 3785
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.937934145554747e-06,
+      "loss": 0.4522,
+      "step": 3786
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.9332732297695e-06,
+      "loss": 0.409,
+      "step": 3787
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.928612548376959e-06,
+      "loss": 0.4118,
+      "step": 3788
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.923952102401211e-06,
+      "loss": 0.4186,
+      "step": 3789
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.919291892866306e-06,
+      "loss": 0.4022,
+      "step": 3790
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.914631920796235e-06,
+      "loss": 0.4007,
+      "step": 3791
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.90997218721493e-06,
+      "loss": 0.4232,
+      "step": 3792
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.905312693146284e-06,
+      "loss": 0.4249,
+      "step": 3793
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.90065343961413e-06,
+      "loss": 0.4149,
+      "step": 3794
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.895994427642252e-06,
+      "loss": 0.436,
+      "step": 3795
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.89133565825437e-06,
+      "loss": 0.4041,
+      "step": 3796
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.88667713247417e-06,
+      "loss": 0.4322,
+      "step": 3797
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.882018851325266e-06,
+      "loss": 0.4277,
+      "step": 3798
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.877360815831226e-06,
+      "loss": 0.4181,
+      "step": 3799
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.872703027015571e-06,
+      "loss": 0.4236,
+      "step": 3800
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.86804548590175e-06,
+      "loss": 0.409,
+      "step": 3801
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.863388193513177e-06,
+      "loss": 0.4029,
+      "step": 3802
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.858731150873199e-06,
+      "loss": 0.4361,
+      "step": 3803
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.854074359005105e-06,
+      "loss": 0.4433,
+      "step": 3804
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.849417818932148e-06,
+      "loss": 0.4081,
+      "step": 3805
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.844761531677504e-06,
+      "loss": 0.4128,
+      "step": 3806
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.840105498264308e-06,
+      "loss": 0.4217,
+      "step": 3807
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.83544971971563e-06,
+      "loss": 0.4259,
+      "step": 3808
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.830794197054492e-06,
+      "loss": 0.4188,
+      "step": 3809
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.826138931303853e-06,
+      "loss": 0.4341,
+      "step": 3810
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.821483923486612e-06,
+      "loss": 0.4199,
+      "step": 3811
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.81682917462563e-06,
+      "loss": 0.4026,
+      "step": 3812
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.812174685743687e-06,
+      "loss": 0.4197,
+      "step": 3813
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.807520457863527e-06,
+      "loss": 0.4123,
+      "step": 3814
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.802866492007822e-06,
+      "loss": 0.3937,
+      "step": 3815
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.798212789199192e-06,
+      "loss": 0.4371,
+      "step": 3816
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.7935593504602e-06,
+      "loss": 0.4467,
+      "step": 3817
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.788906176813345e-06,
+      "loss": 0.4323,
+      "step": 3818
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.784253269281079e-06,
+      "loss": 0.4238,
+      "step": 3819
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.779600628885786e-06,
+      "loss": 0.4274,
+      "step": 3820
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.774948256649793e-06,
+      "loss": 0.4273,
+      "step": 3821
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.770296153595373e-06,
+      "loss": 0.4264,
+      "step": 3822
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.765644320744731e-06,
+      "loss": 0.4122,
+      "step": 3823
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.760992759120026e-06,
+      "loss": 0.4153,
+      "step": 3824
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.756341469743337e-06,
+      "loss": 0.4458,
+      "step": 3825
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.751690453636708e-06,
+      "loss": 0.4379,
+      "step": 3826
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.747039711822106e-06,
+      "loss": 0.4173,
+      "step": 3827
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.742389245321434e-06,
+      "loss": 0.4434,
+      "step": 3828
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.737739055156556e-06,
+      "loss": 0.4301,
+      "step": 3829
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.733089142349253e-06,
+      "loss": 0.4207,
+      "step": 3830
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.728439507921258e-06,
+      "loss": 0.4439,
+      "step": 3831
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.723790152894236e-06,
+      "loss": 0.4313,
+      "step": 3832
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.719141078289795e-06,
+      "loss": 0.4215,
+      "step": 3833
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.71449228512948e-06,
+      "loss": 0.4197,
+      "step": 3834
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.70984377443477e-06,
+      "loss": 0.4526,
+      "step": 3835
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.705195547227093e-06,
+      "loss": 0.4152,
+      "step": 3836
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.700547604527802e-06,
+      "loss": 0.4363,
+      "step": 3837
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.695899947358196e-06,
+      "loss": 0.4088,
+      "step": 3838
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.691252576739509e-06,
+      "loss": 0.4264,
+      "step": 3839
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.686605493692905e-06,
+      "loss": 0.435,
+      "step": 3840
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.6819586992395e-06,
+      "loss": 0.4327,
+      "step": 3841
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.67731219440033e-06,
+      "loss": 0.4256,
+      "step": 3842
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.672665980196381e-06,
+      "loss": 0.4297,
+      "step": 3843
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.668020057648568e-06,
+      "loss": 0.4242,
+      "step": 3844
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.663374427777737e-06,
+      "loss": 0.4374,
+      "step": 3845
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.658729091604685e-06,
+      "loss": 0.4295,
+      "step": 3846
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.65408405015013e-06,
+      "loss": 0.4295,
+      "step": 3847
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.649439304434733e-06,
+      "loss": 0.4097,
+      "step": 3848
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.644794855479081e-06,
+      "loss": 0.4306,
+      "step": 3849
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.640150704303715e-06,
+      "loss": 0.4029,
+      "step": 3850
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.63550685192909e-06,
+      "loss": 0.435,
+      "step": 3851
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.630863299375597e-06,
+      "loss": 0.4457,
+      "step": 3852
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.626220047663581e-06,
+      "loss": 0.4277,
+      "step": 3853
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.621577097813297e-06,
+      "loss": 0.4282,
+      "step": 3854
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.61693445084495e-06,
+      "loss": 0.4518,
+      "step": 3855
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.612292107778671e-06,
+      "loss": 0.4051,
+      "step": 3856
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.60765006963452e-06,
+      "loss": 0.4168,
+      "step": 3857
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.603008337432504e-06,
+      "loss": 0.425,
+      "step": 3858
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.598366912192548e-06,
+      "loss": 0.4081,
+      "step": 3859
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.593725794934523e-06,
+      "loss": 0.4169,
+      "step": 3860
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.589084986678218e-06,
+      "loss": 0.4497,
+      "step": 3861
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.584444488443366e-06,
+      "loss": 0.4295,
+      "step": 3862
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.579804301249626e-06,
+      "loss": 0.4368,
+      "step": 3863
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.575164426116586e-06,
+      "loss": 0.4448,
+      "step": 3864
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.570524864063777e-06,
+      "loss": 0.3962,
+      "step": 3865
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.565885616110645e-06,
+      "loss": 0.4263,
+      "step": 3866
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.561246683276585e-06,
+      "loss": 0.4323,
+      "step": 3867
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.556608066580909e-06,
+      "loss": 0.4231,
+      "step": 3868
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.55196976704286e-06,
+      "loss": 0.4348,
+      "step": 3869
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.547331785681621e-06,
+      "loss": 0.4276,
+      "step": 3870
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.542694123516291e-06,
+      "loss": 0.4053,
+      "step": 3871
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.53805678156592e-06,
+      "loss": 0.4448,
+      "step": 3872
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.533419760849461e-06,
+      "loss": 0.4221,
+      "step": 3873
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.528783062385822e-06,
+      "loss": 0.4296,
+      "step": 3874
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.524146687193823e-06,
+      "loss": 0.4204,
+      "step": 3875
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.519510636292217e-06,
+      "loss": 0.4128,
+      "step": 3876
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.514874910699687e-06,
+      "loss": 0.4279,
+      "step": 3877
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.510239511434847e-06,
+      "loss": 0.397,
+      "step": 3878
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.505604439516238e-06,
+      "loss": 0.4177,
+      "step": 3879
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.500969695962326e-06,
+      "loss": 0.4261,
+      "step": 3880
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.496335281791501e-06,
+      "loss": 0.4136,
+      "step": 3881
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.491701198022098e-06,
+      "loss": 0.4269,
+      "step": 3882
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.487067445672357e-06,
+      "loss": 0.4202,
+      "step": 3883
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.482434025760466e-06,
+      "loss": 0.4234,
+      "step": 3884
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.47780093930452e-06,
+      "loss": 0.4191,
+      "step": 3885
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.47316818732256e-06,
+      "loss": 0.4111,
+      "step": 3886
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.468535770832541e-06,
+      "loss": 0.4127,
+      "step": 3887
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.46390369085234e-06,
+      "loss": 0.4242,
+      "step": 3888
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.459271948399778e-06,
+      "loss": 0.414,
+      "step": 3889
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.454640544492585e-06,
+      "loss": 0.4367,
+      "step": 3890
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.450009480148427e-06,
+      "loss": 0.4426,
+      "step": 3891
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.445378756384891e-06,
+      "loss": 0.4541,
+      "step": 3892
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.440748374219485e-06,
+      "loss": 0.4095,
+      "step": 3893
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.436118334669653e-06,
+      "loss": 0.4205,
+      "step": 3894
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.431488638752747e-06,
+      "loss": 0.4141,
+      "step": 3895
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.426859287486067e-06,
+      "loss": 0.4458,
+      "step": 3896
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.422230281886811e-06,
+      "loss": 0.4227,
+      "step": 3897
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.417601622972126e-06,
+      "loss": 0.4478,
+      "step": 3898
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.412973311759064e-06,
+      "loss": 0.4292,
+      "step": 3899
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.408345349264607e-06,
+      "loss": 0.4431,
+      "step": 3900
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.403717736505665e-06,
+      "loss": 0.424,
+      "step": 3901
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.399090474499058e-06,
+      "loss": 0.4302,
+      "step": 3902
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.39446356426155e-06,
+      "loss": 0.4483,
+      "step": 3903
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.38983700680981e-06,
+      "loss": 0.432,
+      "step": 3904
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.38521080316043e-06,
+      "loss": 0.4332,
+      "step": 3905
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.380584954329938e-06,
+      "loss": 0.4277,
+      "step": 3906
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.375959461334769e-06,
+      "loss": 0.4471,
+      "step": 3907
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.371334325191292e-06,
+      "loss": 0.4161,
+      "step": 3908
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.366709546915788e-06,
+      "loss": 0.4271,
+      "step": 3909
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.362085127524459e-06,
+      "loss": 0.4206,
+      "step": 3910
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.357461068033442e-06,
+      "loss": 0.4314,
+      "step": 3911
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.352837369458774e-06,
+      "loss": 0.4246,
+      "step": 3912
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.348214032816434e-06,
+      "loss": 0.4142,
+      "step": 3913
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.343591059122306e-06,
+      "loss": 0.429,
+      "step": 3914
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.338968449392202e-06,
+      "loss": 0.4146,
+      "step": 3915
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.33434620464185e-06,
+      "loss": 0.4294,
+      "step": 3916
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.329724325886897e-06,
+      "loss": 0.4321,
+      "step": 3917
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.32510281414292e-06,
+      "loss": 0.4251,
+      "step": 3918
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.320481670425396e-06,
+      "loss": 0.4113,
+      "step": 3919
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.315860895749744e-06,
+      "loss": 0.437,
+      "step": 3920
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.311240491131286e-06,
+      "loss": 0.4134,
+      "step": 3921
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.30662045758526e-06,
+      "loss": 0.4302,
+      "step": 3922
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.30200079612684e-06,
+      "loss": 0.4275,
+      "step": 3923
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.297381507771101e-06,
+      "loss": 0.4234,
+      "step": 3924
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.29276259353305e-06,
+      "loss": 0.4297,
+      "step": 3925
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.288144054427594e-06,
+      "loss": 0.4258,
+      "step": 3926
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.283525891469578e-06,
+      "loss": 0.4294,
+      "step": 3927
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.27890810567375e-06,
+      "loss": 0.4499,
+      "step": 3928
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.274290698054779e-06,
+      "loss": 0.4577,
+      "step": 3929
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.269673669627256e-06,
+      "loss": 0.4162,
+      "step": 3930
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.265057021405679e-06,
+      "loss": 0.4221,
+      "step": 3931
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.260440754404471e-06,
+      "loss": 0.4323,
+      "step": 3932
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.255824869637967e-06,
+      "loss": 0.3878,
+      "step": 3933
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.251209368120413e-06,
+      "loss": 0.4237,
+      "step": 3934
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.24659425086599e-06,
+      "loss": 0.4243,
+      "step": 3935
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.24197951888877e-06,
+      "loss": 0.427,
+      "step": 3936
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.237365173202757e-06,
+      "loss": 0.3968,
+      "step": 3937
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.23275121482186e-06,
+      "loss": 0.4257,
+      "step": 3938
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.228137644759915e-06,
+      "loss": 0.41,
+      "step": 3939
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.223524464030662e-06,
+      "loss": 0.4372,
+      "step": 3940
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.218911673647751e-06,
+      "loss": 0.4047,
+      "step": 3941
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.214299274624767e-06,
+      "loss": 0.4052,
+      "step": 3942
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.209687267975187e-06,
+      "loss": 0.4383,
+      "step": 3943
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.205075654712416e-06,
+      "loss": 0.4294,
+      "step": 3944
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.200464435849765e-06,
+      "loss": 0.4493,
+      "step": 3945
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.195853612400459e-06,
+      "loss": 0.4323,
+      "step": 3946
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.191243185377643e-06,
+      "loss": 0.4073,
+      "step": 3947
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.18663315579436e-06,
+      "loss": 0.4391,
+      "step": 3948
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.182023524663587e-06,
+      "loss": 0.4054,
+      "step": 3949
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.177414292998194e-06,
+      "loss": 0.4246,
+      "step": 3950
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.172805461810975e-06,
+      "loss": 0.4667,
+      "step": 3951
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.168197032114633e-06,
+      "loss": 0.4303,
+      "step": 3952
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.163589004921776e-06,
+      "loss": 0.4191,
+      "step": 3953
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.158981381244934e-06,
+      "loss": 0.4326,
+      "step": 3954
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.154374162096541e-06,
+      "loss": 0.4065,
+      "step": 3955
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.149767348488949e-06,
+      "loss": 0.4394,
+      "step": 3956
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.145160941434415e-06,
+      "loss": 0.4577,
+      "step": 3957
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.140554941945104e-06,
+      "loss": 0.4391,
+      "step": 3958
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.135949351033101e-06,
+      "loss": 0.4182,
+      "step": 3959
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.131344169710396e-06,
+      "loss": 0.4126,
+      "step": 3960
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.126739398988886e-06,
+      "loss": 0.4252,
+      "step": 3961
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.122135039880382e-06,
+      "loss": 0.4327,
+      "step": 3962
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.117531093396605e-06,
+      "loss": 0.4177,
+      "step": 3963
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.112927560549185e-06,
+      "loss": 0.4137,
+      "step": 3964
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.10832444234965e-06,
+      "loss": 0.4531,
+      "step": 3965
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.103721739809461e-06,
+      "loss": 0.4481,
+      "step": 3966
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.099119453939961e-06,
+      "loss": 0.4111,
+      "step": 3967
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.094517585752424e-06,
+      "loss": 0.444,
+      "step": 3968
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.089916136258018e-06,
+      "loss": 0.417,
+      "step": 3969
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.085315106467821e-06,
+      "loss": 0.4202,
+      "step": 3970
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.080714497392824e-06,
+      "loss": 0.4314,
+      "step": 3971
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.076114310043917e-06,
+      "loss": 0.4239,
+      "step": 3972
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.071514545431908e-06,
+      "loss": 0.4178,
+      "step": 3973
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.066915204567503e-06,
+      "loss": 0.4229,
+      "step": 3974
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.062316288461327e-06,
+      "loss": 0.416,
+      "step": 3975
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.057717798123895e-06,
+      "loss": 0.4088,
+      "step": 3976
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.053119734565638e-06,
+      "loss": 0.4226,
+      "step": 3977
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.048522098796896e-06,
+      "loss": 0.4436,
+      "step": 3978
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.043924891827905e-06,
+      "loss": 0.4405,
+      "step": 3979
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.03932811466882e-06,
+      "loss": 0.411,
+      "step": 3980
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.03473176832969e-06,
+      "loss": 0.4038,
+      "step": 3981
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.030135853820472e-06,
+      "loss": 0.4066,
+      "step": 3982
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.025540372151034e-06,
+      "loss": 0.4472,
+      "step": 3983
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.020945324331143e-06,
+      "loss": 0.4133,
+      "step": 3984
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.016350711370475e-06,
+      "loss": 0.4427,
+      "step": 3985
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.0117565342786e-06,
+      "loss": 0.4554,
+      "step": 3986
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.00716279406501e-06,
+      "loss": 0.4302,
+      "step": 3987
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.002569491739088e-06,
+      "loss": 0.4211,
+      "step": 3988
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.997976628310117e-06,
+      "loss": 0.4297,
+      "step": 3989
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.9933842047873e-06,
+      "loss": 0.4311,
+      "step": 3990
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.988792222179729e-06,
+      "loss": 0.4595,
+      "step": 3991
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.984200681496406e-06,
+      "loss": 0.4136,
+      "step": 3992
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.979609583746232e-06,
+      "loss": 0.4158,
+      "step": 3993
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.97501892993801e-06,
+      "loss": 0.4269,
+      "step": 3994
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.970428721080455e-06,
+      "loss": 0.4326,
+      "step": 3995
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.965838958182167e-06,
+      "loss": 0.4196,
+      "step": 3996
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.96124964225167e-06,
+      "loss": 0.4216,
+      "step": 3997
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.956660774297373e-06,
+      "loss": 0.4312,
+      "step": 3998
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.952072355327587e-06,
+      "loss": 0.4194,
+      "step": 3999
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.947484386350534e-06,
+      "loss": 0.4304,
+      "step": 4000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.942896868374327e-06,
+      "loss": 0.425,
+      "step": 4001
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.938309802406996e-06,
+      "loss": 0.4327,
+      "step": 4002
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.933723189456445e-06,
+      "loss": 0.4114,
+      "step": 4003
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.929137030530508e-06,
+      "loss": 0.4399,
+      "step": 4004
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.924551326636899e-06,
+      "loss": 0.4208,
+      "step": 4005
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.919966078783235e-06,
+      "loss": 0.4325,
+      "step": 4006
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.91538128797704e-06,
+      "loss": 0.4281,
+      "step": 4007
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.910796955225734e-06,
+      "loss": 0.4181,
+      "step": 4008
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.906213081536636e-06,
+      "loss": 0.4479,
+      "step": 4009
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.901629667916967e-06,
+      "loss": 0.4166,
+      "step": 4010
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.897046715373832e-06,
+      "loss": 0.4409,
+      "step": 4011
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.892464224914262e-06,
+      "loss": 0.4218,
+      "step": 4012
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.88788219754516e-06,
+      "loss": 0.4079,
+      "step": 4013
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.883300634273346e-06,
+      "loss": 0.427,
+      "step": 4014
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.878719536105524e-06,
+      "loss": 0.4249,
+      "step": 4015
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.874138904048309e-06,
+      "loss": 0.4337,
+      "step": 4016
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.869558739108204e-06,
+      "loss": 0.4093,
+      "step": 4017
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.864979042291607e-06,
+      "loss": 0.4133,
+      "step": 4018
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.860399814604827e-06,
+      "loss": 0.4324,
+      "step": 4019
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.855821057054052e-06,
+      "loss": 0.4596,
+      "step": 4020
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.851242770645387e-06,
+      "loss": 0.4187,
+      "step": 4021
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.846664956384815e-06,
+      "loss": 0.421,
+      "step": 4022
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.842087615278223e-06,
+      "loss": 0.4319,
+      "step": 4023
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.837510748331396e-06,
+      "loss": 0.3987,
+      "step": 4024
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.832934356550007e-06,
+      "loss": 0.4109,
+      "step": 4025
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.82835844093964e-06,
+      "loss": 0.4249,
+      "step": 4026
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.823783002505754e-06,
+      "loss": 0.4293,
+      "step": 4027
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.819208042253723e-06,
+      "loss": 0.4286,
+      "step": 4028
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.814633561188803e-06,
+      "loss": 0.4299,
+      "step": 4029
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.810059560316145e-06,
+      "loss": 0.431,
+      "step": 4030
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.805486040640802e-06,
+      "loss": 0.4209,
+      "step": 4031
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.80091300316771e-06,
+      "loss": 0.4199,
+      "step": 4032
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.796340448901719e-06,
+      "loss": 0.4184,
+      "step": 4033
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.791768378847549e-06,
+      "loss": 0.4372,
+      "step": 4034
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.787196794009824e-06,
+      "loss": 0.4307,
+      "step": 4035
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.782625695393071e-06,
+      "loss": 0.4334,
+      "step": 4036
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.778055084001693e-06,
+      "loss": 0.4403,
+      "step": 4037
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.77348496084e-06,
+      "loss": 0.4096,
+      "step": 4038
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.76891532691218e-06,
+      "loss": 0.4092,
+      "step": 4039
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.764346183222332e-06,
+      "loss": 0.4447,
+      "step": 4040
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.759777530774433e-06,
+      "loss": 0.4118,
+      "step": 4041
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.755209370572353e-06,
+      "loss": 0.4323,
+      "step": 4042
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.750641703619865e-06,
+      "loss": 0.4159,
+      "step": 4043
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.74607453092062e-06,
+      "loss": 0.4303,
+      "step": 4044
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.74150785347817e-06,
+      "loss": 0.414,
+      "step": 4045
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.736941672295952e-06,
+      "loss": 0.4027,
+      "step": 4046
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.732375988377297e-06,
+      "loss": 0.4251,
+      "step": 4047
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.727810802725428e-06,
+      "loss": 0.4524,
+      "step": 4048
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.723246116343453e-06,
+      "loss": 0.4262,
+      "step": 4049
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.71868193023438e-06,
+      "loss": 0.4111,
+      "step": 4050
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.714118245401097e-06,
+      "loss": 0.4042,
+      "step": 4051
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.709555062846385e-06,
+      "loss": 0.4341,
+      "step": 4052
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.704992383572922e-06,
+      "loss": 0.4366,
+      "step": 4053
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.700430208583261e-06,
+      "loss": 0.4349,
+      "step": 4054
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.695868538879858e-06,
+      "loss": 0.4188,
+      "step": 4055
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.691307375465046e-06,
+      "loss": 0.4044,
+      "step": 4056
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.686746719341061e-06,
+      "loss": 0.4159,
+      "step": 4057
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.682186571510017e-06,
+      "loss": 0.4348,
+      "step": 4058
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.677626932973912e-06,
+      "loss": 0.3961,
+      "step": 4059
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.673067804734649e-06,
+      "loss": 0.4153,
+      "step": 4060
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.668509187794e-06,
+      "loss": 0.4027,
+      "step": 4061
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.663951083153642e-06,
+      "loss": 0.4312,
+      "step": 4062
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.659393491815121e-06,
+      "loss": 0.4458,
+      "step": 4063
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.65483641477989e-06,
+      "loss": 0.4095,
+      "step": 4064
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.650279853049274e-06,
+      "loss": 0.442,
+      "step": 4065
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.645723807624489e-06,
+      "loss": 0.4334,
+      "step": 4066
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.64116827950664e-06,
+      "loss": 0.4278,
+      "step": 4067
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.636613269696718e-06,
+      "loss": 0.4132,
+      "step": 4068
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.632058779195598e-06,
+      "loss": 0.4128,
+      "step": 4069
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.6275048090040406e-06,
+      "loss": 0.425,
+      "step": 4070
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.6229513601226905e-06,
+      "loss": 0.4014,
+      "step": 4071
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.618398433552088e-06,
+      "loss": 0.4177,
+      "step": 4072
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.6138460302926424e-06,
+      "loss": 0.4182,
+      "step": 4073
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.609294151344665e-06,
+      "loss": 0.4355,
+      "step": 4074
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.60474279770834e-06,
+      "loss": 0.4199,
+      "step": 4075
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.600191970383739e-06,
+      "loss": 0.4311,
+      "step": 4076
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.595641670370821e-06,
+      "loss": 0.4468,
+      "step": 4077
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.591091898669423e-06,
+      "loss": 0.4495,
+      "step": 4078
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.586542656279274e-06,
+      "loss": 0.4139,
+      "step": 4079
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.581993944199979e-06,
+      "loss": 0.4159,
+      "step": 4080
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.577445763431036e-06,
+      "loss": 0.4065,
+      "step": 4081
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.5728981149718175e-06,
+      "loss": 0.4388,
+      "step": 4082
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.568350999821576e-06,
+      "loss": 0.449,
+      "step": 4083
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.563804418979462e-06,
+      "loss": 0.4324,
+      "step": 4084
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.55925837344449e-06,
+      "loss": 0.432,
+      "step": 4085
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.554712864215576e-06,
+      "loss": 0.4398,
+      "step": 4086
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5501678922915015e-06,
+      "loss": 0.4212,
+      "step": 4087
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.545623458670934e-06,
+      "loss": 0.4166,
+      "step": 4088
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.541079564352434e-06,
+      "loss": 0.4241,
+      "step": 4089
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5365362103344285e-06,
+      "loss": 0.3977,
+      "step": 4090
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.531993397615236e-06,
+      "loss": 0.4392,
+      "step": 4091
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.527451127193049e-06,
+      "loss": 0.4188,
+      "step": 4092
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.522909400065947e-06,
+      "loss": 0.4484,
+      "step": 4093
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.518368217231888e-06,
+      "loss": 0.4132,
+      "step": 4094
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.513827579688703e-06,
+      "loss": 0.4035,
+      "step": 4095
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.5092874884341185e-06,
+      "loss": 0.4387,
+      "step": 4096
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.504747944465723e-06,
+      "loss": 0.4097,
+      "step": 4097
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.500208948781005e-06,
+      "loss": 0.4152,
+      "step": 4098
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.495670502377314e-06,
+      "loss": 0.4133,
+      "step": 4099
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.491132606251887e-06,
+      "loss": 0.4212,
+      "step": 4100
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4865952614018435e-06,
+      "loss": 0.4137,
+      "step": 4101
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4820584688241694e-06,
+      "loss": 0.4323,
+      "step": 4102
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4775222295157465e-06,
+      "loss": 0.4055,
+      "step": 4103
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.472986544473319e-06,
+      "loss": 0.4229,
+      "step": 4104
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.468451414693526e-06,
+      "loss": 0.4284,
+      "step": 4105
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4639168411728695e-06,
+      "loss": 0.4411,
+      "step": 4106
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.45938282490773e-06,
+      "loss": 0.4518,
+      "step": 4107
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.454849366894379e-06,
+      "loss": 0.4136,
+      "step": 4108
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4503164681289486e-06,
+      "loss": 0.447,
+      "step": 4109
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.445784129607465e-06,
+      "loss": 0.4178,
+      "step": 4110
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.441252352325818e-06,
+      "loss": 0.421,
+      "step": 4111
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.436721137279774e-06,
+      "loss": 0.4274,
+      "step": 4112
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.432190485464989e-06,
+      "loss": 0.4281,
+      "step": 4113
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.427660397876979e-06,
+      "loss": 0.4227,
+      "step": 4114
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.423130875511151e-06,
+      "loss": 0.4203,
+      "step": 4115
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.418601919362773e-06,
+      "loss": 0.4178,
+      "step": 4116
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.414073530427001e-06,
+      "loss": 0.4178,
+      "step": 4117
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.409545709698863e-06,
+      "loss": 0.4238,
+      "step": 4118
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.405018458173252e-06,
+      "loss": 0.4313,
+      "step": 4119
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.4004917768449555e-06,
+      "loss": 0.4225,
+      "step": 4120
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.3959656667086165e-06,
+      "loss": 0.4508,
+      "step": 4121
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.391440128758765e-06,
+      "loss": 0.4314,
+      "step": 4122
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.386915163989801e-06,
+      "loss": 0.4241,
+      "step": 4123
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.382390773395992e-06,
+      "loss": 0.4331,
+      "step": 4124
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.3778669579714956e-06,
+      "loss": 0.423,
+      "step": 4125
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.373343718710324e-06,
+      "loss": 0.417,
+      "step": 4126
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.368821056606381e-06,
+      "loss": 0.4286,
+      "step": 4127
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3642989726534275e-06,
+      "loss": 0.4235,
+      "step": 4128
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.359777467845111e-06,
+      "loss": 0.4414,
+      "step": 4129
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.355256543174938e-06,
+      "loss": 0.4185,
+      "step": 4130
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.350736199636299e-06,
+      "loss": 0.4408,
+      "step": 4131
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3462164382224535e-06,
+      "loss": 0.4165,
+      "step": 4132
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.341697259926527e-06,
+      "loss": 0.4096,
+      "step": 4133
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.337178665741529e-06,
+      "loss": 0.4382,
+      "step": 4134
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.332660656660332e-06,
+      "loss": 0.4269,
+      "step": 4135
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3281432336756754e-06,
+      "loss": 0.4335,
+      "step": 4136
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.323626397780185e-06,
+      "loss": 0.4353,
+      "step": 4137
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.319110149966342e-06,
+      "loss": 0.4488,
+      "step": 4138
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.31459449122651e-06,
+      "loss": 0.4018,
+      "step": 4139
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.310079422552909e-06,
+      "loss": 0.4191,
+      "step": 4140
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.3055649449376535e-06,
+      "loss": 0.4316,
+      "step": 4141
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.301051059372701e-06,
+      "loss": 0.4154,
+      "step": 4142
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.296537766849894e-06,
+      "loss": 0.4553,
+      "step": 4143
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.292025068360944e-06,
+      "loss": 0.4286,
+      "step": 4144
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.287512964897427e-06,
+      "loss": 0.438,
+      "step": 4145
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.283001457450796e-06,
+      "loss": 0.4416,
+      "step": 4146
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2784905470123645e-06,
+      "loss": 0.4229,
+      "step": 4147
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.273980234573313e-06,
+      "loss": 0.4351,
+      "step": 4148
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.269470521124707e-06,
+      "loss": 0.4245,
+      "step": 4149
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.264961407657456e-06,
+      "loss": 0.4376,
+      "step": 4150
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.260452895162366e-06,
+      "loss": 0.4378,
+      "step": 4151
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.255944984630086e-06,
+      "loss": 0.4253,
+      "step": 4152
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.251437677051146e-06,
+      "loss": 0.4257,
+      "step": 4153
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.24693097341594e-06,
+      "loss": 0.44,
+      "step": 4154
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.242424874714726e-06,
+      "loss": 0.4262,
+      "step": 4155
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.237919381937639e-06,
+      "loss": 0.4336,
+      "step": 4156
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.233414496074667e-06,
+      "loss": 0.4285,
+      "step": 4157
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.228910218115679e-06,
+      "loss": 0.4324,
+      "step": 4158
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.2244065490504e-06,
+      "loss": 0.4359,
+      "step": 4159
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.219903489868425e-06,
+      "loss": 0.434,
+      "step": 4160
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.215401041559216e-06,
+      "loss": 0.4172,
+      "step": 4161
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.210899205112095e-06,
+      "loss": 0.423,
+      "step": 4162
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.206397981516261e-06,
+      "loss": 0.4368,
+      "step": 4163
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.201897371760768e-06,
+      "loss": 0.4225,
+      "step": 4164
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.197397376834534e-06,
+      "loss": 0.4343,
+      "step": 4165
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.192897997726353e-06,
+      "loss": 0.4046,
+      "step": 4166
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.188399235424875e-06,
+      "loss": 0.406,
+      "step": 4167
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.183901090918617e-06,
+      "loss": 0.4083,
+      "step": 4168
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.179403565195955e-06,
+      "loss": 0.4405,
+      "step": 4169
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.174906659245143e-06,
+      "loss": 0.3938,
+      "step": 4170
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.170410374054284e-06,
+      "loss": 0.4131,
+      "step": 4171
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.165914710611346e-06,
+      "loss": 0.4161,
+      "step": 4172
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.161419669904173e-06,
+      "loss": 0.4327,
+      "step": 4173
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.156925252920458e-06,
+      "loss": 0.4174,
+      "step": 4174
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.152431460647766e-06,
+      "loss": 0.4323,
+      "step": 4175
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.147938294073521e-06,
+      "loss": 0.3933,
+      "step": 4176
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.143445754185008e-06,
+      "loss": 0.4088,
+      "step": 4177
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.138953841969378e-06,
+      "loss": 0.4172,
+      "step": 4178
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.134462558413639e-06,
+      "loss": 0.412,
+      "step": 4179
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1299719045046725e-06,
+      "loss": 0.4001,
+      "step": 4180
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1254818812292025e-06,
+      "loss": 0.4157,
+      "step": 4181
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.120992489573835e-06,
+      "loss": 0.4136,
+      "step": 4182
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.116503730525024e-06,
+      "loss": 0.428,
+      "step": 4183
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.1120156050690844e-06,
+      "loss": 0.437,
+      "step": 4184
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.1075281141922005e-06,
+      "loss": 0.4216,
+      "step": 4185
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.103041258880405e-06,
+      "loss": 0.4103,
+      "step": 4186
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.098555040119609e-06,
+      "loss": 0.422,
+      "step": 4187
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.094069458895566e-06,
+      "loss": 0.4066,
+      "step": 4188
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.0895845161938935e-06,
+      "loss": 0.4009,
+      "step": 4189
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.0851002130000804e-06,
+      "loss": 0.4003,
+      "step": 4190
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.080616550299457e-06,
+      "loss": 0.441,
+      "step": 4191
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.076133529077228e-06,
+      "loss": 0.4196,
+      "step": 4192
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.071651150318446e-06,
+      "loss": 0.4324,
+      "step": 4193
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.067169415008033e-06,
+      "loss": 0.4565,
+      "step": 4194
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.062688324130764e-06,
+      "loss": 0.4381,
+      "step": 4195
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.058207878671264e-06,
+      "loss": 0.441,
+      "step": 4196
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.053728079614037e-06,
+      "loss": 0.4313,
+      "step": 4197
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.049248927943424e-06,
+      "loss": 0.4428,
+      "step": 4198
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.044770424643637e-06,
+      "loss": 0.4104,
+      "step": 4199
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.0402925706987405e-06,
+      "loss": 0.4119,
+      "step": 4200
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.03581536709265e-06,
+      "loss": 0.4427,
+      "step": 4201
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.031338814809156e-06,
+      "loss": 0.4293,
+      "step": 4202
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.026862914831884e-06,
+      "loss": 0.4151,
+      "step": 4203
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.022387668144337e-06,
+      "loss": 0.4305,
+      "step": 4204
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.017913075729859e-06,
+      "loss": 0.4166,
+      "step": 4205
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.013439138571656e-06,
+      "loss": 0.4292,
+      "step": 4206
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.008965857652792e-06,
+      "loss": 0.4556,
+      "step": 4207
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.004493233956178e-06,
+      "loss": 0.4012,
+      "step": 4208
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.000021268464595e-06,
+      "loss": 0.4168,
+      "step": 4209
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.995549962160665e-06,
+      "loss": 0.4159,
+      "step": 4210
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.991079316026879e-06,
+      "loss": 0.4236,
+      "step": 4211
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.986609331045571e-06,
+      "loss": 0.4235,
+      "step": 4212
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.982140008198932e-06,
+      "loss": 0.4136,
+      "step": 4213
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.977671348469013e-06,
+      "loss": 0.4273,
+      "step": 4214
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.973203352837714e-06,
+      "loss": 0.4314,
+      "step": 4215
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.968736022286796e-06,
+      "loss": 0.4364,
+      "step": 4216
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.964269357797862e-06,
+      "loss": 0.4333,
+      "step": 4217
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.959803360352382e-06,
+      "loss": 0.3909,
+      "step": 4218
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.955338030931671e-06,
+      "loss": 0.4042,
+      "step": 4219
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.950873370516893e-06,
+      "loss": 0.4573,
+      "step": 4220
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.946409380089083e-06,
+      "loss": 0.4341,
+      "step": 4221
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.941946060629108e-06,
+      "loss": 0.428,
+      "step": 4222
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.937483413117703e-06,
+      "loss": 0.4023,
+      "step": 4223
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.933021438535443e-06,
+      "loss": 0.4201,
+      "step": 4224
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.928560137862762e-06,
+      "loss": 0.412,
+      "step": 4225
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.92409951207995e-06,
+      "loss": 0.431,
+      "step": 4226
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.919639562167137e-06,
+      "loss": 0.4036,
+      "step": 4227
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.915180289104318e-06,
+      "loss": 0.3996,
+      "step": 4228
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.91072169387133e-06,
+      "loss": 0.4057,
+      "step": 4229
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.9062637774478615e-06,
+      "loss": 0.414,
+      "step": 4230
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.901806540813458e-06,
+      "loss": 0.4539,
+      "step": 4231
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.897349984947504e-06,
+      "loss": 0.4313,
+      "step": 4232
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.892894110829251e-06,
+      "loss": 0.4179,
+      "step": 4233
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.888438919437783e-06,
+      "loss": 0.4275,
+      "step": 4234
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.883984411752052e-06,
+      "loss": 0.4352,
+      "step": 4235
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.879530588750847e-06,
+      "loss": 0.3927,
+      "step": 4236
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.875077451412804e-06,
+      "loss": 0.4265,
+      "step": 4237
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.870625000716423e-06,
+      "loss": 0.432,
+      "step": 4238
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.866173237640036e-06,
+      "loss": 0.4404,
+      "step": 4239
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.861722163161842e-06,
+      "loss": 0.4195,
+      "step": 4240
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.857271778259868e-06,
+      "loss": 0.426,
+      "step": 4241
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.852822083912012e-06,
+      "loss": 0.4256,
+      "step": 4242
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.848373081096002e-06,
+      "loss": 0.4563,
+      "step": 4243
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.843924770789421e-06,
+      "loss": 0.4288,
+      "step": 4244
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.839477153969703e-06,
+      "loss": 0.4242,
+      "step": 4245
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.835030231614119e-06,
+      "loss": 0.4273,
+      "step": 4246
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.830584004699805e-06,
+      "loss": 0.4311,
+      "step": 4247
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.826138474203728e-06,
+      "loss": 0.428,
+      "step": 4248
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.821693641102705e-06,
+      "loss": 0.4221,
+      "step": 4249
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.817249506373409e-06,
+      "loss": 0.43,
+      "step": 4250
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.812806070992349e-06,
+      "loss": 0.4217,
+      "step": 4251
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.808363335935888e-06,
+      "loss": 0.4256,
+      "step": 4252
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.803921302180227e-06,
+      "loss": 0.4057,
+      "step": 4253
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.79947997070142e-06,
+      "loss": 0.4228,
+      "step": 4254
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.795039342475367e-06,
+      "loss": 0.4112,
+      "step": 4255
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.790599418477804e-06,
+      "loss": 0.4095,
+      "step": 4256
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.786160199684326e-06,
+      "loss": 0.4309,
+      "step": 4257
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.781721687070361e-06,
+      "loss": 0.4426,
+      "step": 4258
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.777283881611191e-06,
+      "loss": 0.4281,
+      "step": 4259
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.772846784281933e-06,
+      "loss": 0.4157,
+      "step": 4260
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.768410396057559e-06,
+      "loss": 0.4301,
+      "step": 4261
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.7639747179128775e-06,
+      "loss": 0.4226,
+      "step": 4262
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.759539750822542e-06,
+      "loss": 0.4076,
+      "step": 4263
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.755105495761057e-06,
+      "loss": 0.4204,
+      "step": 4264
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.750671953702761e-06,
+      "loss": 0.4102,
+      "step": 4265
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.746239125621834e-06,
+      "loss": 0.4254,
+      "step": 4266
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.741807012492316e-06,
+      "loss": 0.4078,
+      "step": 4267
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.737375615288071e-06,
+      "loss": 0.397,
+      "step": 4268
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.732944934982817e-06,
+      "loss": 0.4173,
+      "step": 4269
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.728514972550106e-06,
+      "loss": 0.439,
+      "step": 4270
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.724085728963343e-06,
+      "loss": 0.4045,
+      "step": 4271
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.719657205195769e-06,
+      "loss": 0.4366,
+      "step": 4272
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7152294022204575e-06,
+      "loss": 0.4273,
+      "step": 4273
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.710802321010345e-06,
+      "loss": 0.4444,
+      "step": 4274
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.706375962538192e-06,
+      "loss": 0.4064,
+      "step": 4275
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.7019503277766075e-06,
+      "loss": 0.4174,
+      "step": 4276
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.697525417698037e-06,
+      "loss": 0.4102,
+      "step": 4277
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.693101233274768e-06,
+      "loss": 0.4292,
+      "step": 4278
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6886777754789365e-06,
+      "loss": 0.4404,
+      "step": 4279
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.684255045282505e-06,
+      "loss": 0.4138,
+      "step": 4280
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.679833043657291e-06,
+      "loss": 0.427,
+      "step": 4281
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.675411771574938e-06,
+      "loss": 0.445,
+      "step": 4282
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6709912300069405e-06,
+      "loss": 0.4334,
+      "step": 4283
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.666571419924623e-06,
+      "loss": 0.4063,
+      "step": 4284
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.662152342299154e-06,
+      "loss": 0.4162,
+      "step": 4285
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.657733998101546e-06,
+      "loss": 0.4291,
+      "step": 4286
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.653316388302637e-06,
+      "loss": 0.4239,
+      "step": 4287
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.648899513873123e-06,
+      "loss": 0.3936,
+      "step": 4288
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.644483375783519e-06,
+      "loss": 0.4227,
+      "step": 4289
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.640067975004187e-06,
+      "loss": 0.4194,
+      "step": 4290
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.635653312505329e-06,
+      "loss": 0.3986,
+      "step": 4291
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6312393892569786e-06,
+      "loss": 0.415,
+      "step": 4292
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.626826206229016e-06,
+      "loss": 0.4276,
+      "step": 4293
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.6224137643911444e-06,
+      "loss": 0.4342,
+      "step": 4294
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.6180020647129255e-06,
+      "loss": 0.4171,
+      "step": 4295
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.613591108163739e-06,
+      "loss": 0.4216,
+      "step": 4296
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.609180895712805e-06,
+      "loss": 0.417,
+      "step": 4297
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.604771428329188e-06,
+      "loss": 0.4317,
+      "step": 4298
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.600362706981778e-06,
+      "loss": 0.4186,
+      "step": 4299
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.595954732639314e-06,
+      "loss": 0.4064,
+      "step": 4300
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.591547506270358e-06,
+      "loss": 0.4194,
+      "step": 4301
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.587141028843314e-06,
+      "loss": 0.4128,
+      "step": 4302
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.5827353013264225e-06,
+      "loss": 0.403,
+      "step": 4303
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.5783303246877565e-06,
+      "loss": 0.4226,
+      "step": 4304
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.573926099895225e-06,
+      "loss": 0.4189,
+      "step": 4305
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.569522627916572e-06,
+      "loss": 0.4233,
+      "step": 4306
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.565119909719378e-06,
+      "loss": 0.4442,
+      "step": 4307
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.560717946271053e-06,
+      "loss": 0.4039,
+      "step": 4308
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.556316738538841e-06,
+      "loss": 0.4435,
+      "step": 4309
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.551916287489829e-06,
+      "loss": 0.4336,
+      "step": 4310
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.547516594090925e-06,
+      "loss": 0.4015,
+      "step": 4311
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.543117659308884e-06,
+      "loss": 0.4025,
+      "step": 4312
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.538719484110285e-06,
+      "loss": 0.4038,
+      "step": 4313
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.534322069461539e-06,
+      "loss": 0.4368,
+      "step": 4314
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.529925416328898e-06,
+      "loss": 0.408,
+      "step": 4315
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.5255295256784335e-06,
+      "loss": 0.41,
+      "step": 4316
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.521134398476071e-06,
+      "loss": 0.4328,
+      "step": 4317
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.516740035687543e-06,
+      "loss": 0.433,
+      "step": 4318
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.512346438278436e-06,
+      "loss": 0.4117,
+      "step": 4319
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.507953607214155e-06,
+      "loss": 0.4027,
+      "step": 4320
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.503561543459936e-06,
+      "loss": 0.4136,
+      "step": 4321
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.499170247980857e-06,
+      "loss": 0.421,
+      "step": 4322
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.494779721741814e-06,
+      "loss": 0.4247,
+      "step": 4323
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.490389965707548e-06,
+      "loss": 0.4576,
+      "step": 4324
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.486000980842619e-06,
+      "loss": 0.4308,
+      "step": 4325
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.481612768111421e-06,
+      "loss": 0.43,
+      "step": 4326
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.477225328478185e-06,
+      "loss": 0.4379,
+      "step": 4327
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4728386629069605e-06,
+      "loss": 0.4336,
+      "step": 4328
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.4684527723616375e-06,
+      "loss": 0.4113,
+      "step": 4329
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.464067657805926e-06,
+      "loss": 0.4283,
+      "step": 4330
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.459683320203376e-06,
+      "loss": 0.4199,
+      "step": 4331
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.45529976051736e-06,
+      "loss": 0.4428,
+      "step": 4332
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.450916979711075e-06,
+      "loss": 0.404,
+      "step": 4333
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.446534978747563e-06,
+      "loss": 0.4056,
+      "step": 4334
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.442153758589677e-06,
+      "loss": 0.428,
+      "step": 4335
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.43777332020011e-06,
+      "loss": 0.4421,
+      "step": 4336
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4333936645413786e-06,
+      "loss": 0.4224,
+      "step": 4337
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.429014792575821e-06,
+      "loss": 0.3955,
+      "step": 4338
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.424636705265621e-06,
+      "loss": 0.4107,
+      "step": 4339
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.420259403572769e-06,
+      "loss": 0.421,
+      "step": 4340
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.415882888459104e-06,
+      "loss": 0.3787,
+      "step": 4341
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.411507160886272e-06,
+      "loss": 0.4177,
+      "step": 4342
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.407132221815759e-06,
+      "loss": 0.4131,
+      "step": 4343
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.40275807220887e-06,
+      "loss": 0.4072,
+      "step": 4344
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.398384713026744e-06,
+      "loss": 0.4108,
+      "step": 4345
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.394012145230344e-06,
+      "loss": 0.4252,
+      "step": 4346
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.389640369780452e-06,
+      "loss": 0.4223,
+      "step": 4347
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.385269387637687e-06,
+      "loss": 0.3905,
+      "step": 4348
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.380899199762488e-06,
+      "loss": 0.42,
+      "step": 4349
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.376529807115113e-06,
+      "loss": 0.4161,
+      "step": 4350
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.372161210655662e-06,
+      "loss": 0.4078,
+      "step": 4351
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.367793411344043e-06,
+      "loss": 0.4397,
+      "step": 4352
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.363426410139999e-06,
+      "loss": 0.4269,
+      "step": 4353
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.359060208003095e-06,
+      "loss": 0.4225,
+      "step": 4354
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.3546948058927134e-06,
+      "loss": 0.3873,
+      "step": 4355
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.350330204768078e-06,
+      "loss": 0.3811,
+      "step": 4356
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.345966405588215e-06,
+      "loss": 0.4348,
+      "step": 4357
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.341603409311996e-06,
+      "loss": 0.4321,
+      "step": 4358
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.337241216898099e-06,
+      "loss": 0.4155,
+      "step": 4359
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.332879829305036e-06,
+      "loss": 0.4029,
+      "step": 4360
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.328519247491134e-06,
+      "loss": 0.4238,
+      "step": 4361
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.324159472414545e-06,
+      "loss": 0.4256,
+      "step": 4362
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.319800505033253e-06,
+      "loss": 0.4281,
+      "step": 4363
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.315442346305052e-06,
+      "loss": 0.4301,
+      "step": 4364
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.311084997187566e-06,
+      "loss": 0.4301,
+      "step": 4365
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.306728458638241e-06,
+      "loss": 0.415,
+      "step": 4366
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.302372731614335e-06,
+      "loss": 0.4331,
+      "step": 4367
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.298017817072943e-06,
+      "loss": 0.4292,
+      "step": 4368
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.293663715970967e-06,
+      "loss": 0.4269,
+      "step": 4369
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.289310429265144e-06,
+      "loss": 0.4307,
+      "step": 4370
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.284957957912018e-06,
+      "loss": 0.4069,
+      "step": 4371
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.280606302867969e-06,
+      "loss": 0.4514,
+      "step": 4372
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.276255465089185e-06,
+      "loss": 0.4414,
+      "step": 4373
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.271905445531676e-06,
+      "loss": 0.414,
+      "step": 4374
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.2675562451512804e-06,
+      "loss": 0.4179,
+      "step": 4375
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.263207864903646e-06,
+      "loss": 0.4128,
+      "step": 4376
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.258860305744253e-06,
+      "loss": 0.4078,
+      "step": 4377
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.254513568628391e-06,
+      "loss": 0.4625,
+      "step": 4378
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.250167654511165e-06,
+      "loss": 0.4209,
+      "step": 4379
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.245822564347517e-06,
+      "loss": 0.44,
+      "step": 4380
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.24147829909219e-06,
+      "loss": 0.4374,
+      "step": 4381
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.237134859699756e-06,
+      "loss": 0.4297,
+      "step": 4382
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.232792247124597e-06,
+      "loss": 0.3969,
+      "step": 4383
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.228450462320927e-06,
+      "loss": 0.4198,
+      "step": 4384
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.224109506242767e-06,
+      "loss": 0.4067,
+      "step": 4385
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.219769379843952e-06,
+      "loss": 0.4389,
+      "step": 4386
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.215430084078152e-06,
+      "loss": 0.4191,
+      "step": 4387
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.211091619898834e-06,
+      "loss": 0.4248,
+      "step": 4388
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.206753988259296e-06,
+      "loss": 0.3921,
+      "step": 4389
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.202417190112653e-06,
+      "loss": 0.4135,
+      "step": 4390
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.198081226411828e-06,
+      "loss": 0.4335,
+      "step": 4391
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.193746098109569e-06,
+      "loss": 0.4285,
+      "step": 4392
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.189411806158431e-06,
+      "loss": 0.4204,
+      "step": 4393
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.185078351510799e-06,
+      "loss": 0.4159,
+      "step": 4394
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.18074573511886e-06,
+      "loss": 0.4094,
+      "step": 4395
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.17641395793463e-06,
+      "loss": 0.4074,
+      "step": 4396
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.172083020909931e-06,
+      "loss": 0.4294,
+      "step": 4397
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.167752924996398e-06,
+      "loss": 0.4372,
+      "step": 4398
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.163423671145494e-06,
+      "loss": 0.424,
+      "step": 4399
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.159095260308482e-06,
+      "loss": 0.4432,
+      "step": 4400
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.154767693436454e-06,
+      "loss": 0.4207,
+      "step": 4401
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.150440971480307e-06,
+      "loss": 0.4251,
+      "step": 4402
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.14611509539075e-06,
+      "loss": 0.4357,
+      "step": 4403
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.141790066118321e-06,
+      "loss": 0.4056,
+      "step": 4404
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.137465884613354e-06,
+      "loss": 0.4194,
+      "step": 4405
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1331425518260075e-06,
+      "loss": 0.4163,
+      "step": 4406
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.128820068706249e-06,
+      "loss": 0.415,
+      "step": 4407
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.124498436203865e-06,
+      "loss": 0.4594,
+      "step": 4408
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.120177655268451e-06,
+      "loss": 0.4266,
+      "step": 4409
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.1158577268494075e-06,
+      "loss": 0.4264,
+      "step": 4410
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.111538651895967e-06,
+      "loss": 0.4283,
+      "step": 4411
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.107220431357155e-06,
+      "loss": 0.4325,
+      "step": 4412
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.102903066181822e-06,
+      "loss": 0.431,
+      "step": 4413
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.0985865573186265e-06,
+      "loss": 0.4471,
+      "step": 4414
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.09427090571603e-06,
+      "loss": 0.4277,
+      "step": 4415
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.089956112322324e-06,
+      "loss": 0.4153,
+      "step": 4416
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.085642178085595e-06,
+      "loss": 0.4151,
+      "step": 4417
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.081329103953752e-06,
+      "loss": 0.4442,
+      "step": 4418
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0770168908745095e-06,
+      "loss": 0.4268,
+      "step": 4419
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0727055397953895e-06,
+      "loss": 0.4101,
+      "step": 4420
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.068395051663731e-06,
+      "loss": 0.4124,
+      "step": 4421
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.064085427426679e-06,
+      "loss": 0.4242,
+      "step": 4422
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.059776668031195e-06,
+      "loss": 0.423,
+      "step": 4423
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.055468774424039e-06,
+      "loss": 0.4192,
+      "step": 4424
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.051161747551797e-06,
+      "loss": 0.4156,
+      "step": 4425
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0468555883608495e-06,
+      "loss": 0.433,
+      "step": 4426
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.042550297797392e-06,
+      "loss": 0.3986,
+      "step": 4427
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.03824587680743e-06,
+      "loss": 0.4146,
+      "step": 4428
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.033942326336778e-06,
+      "loss": 0.4148,
+      "step": 4429
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0296396473310605e-06,
+      "loss": 0.4191,
+      "step": 4430
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0253378407357055e-06,
+      "loss": 0.4433,
+      "step": 4431
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.021036907495949e-06,
+      "loss": 0.4046,
+      "step": 4432
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.016736848556846e-06,
+      "loss": 0.4309,
+      "step": 4433
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.0124376648632465e-06,
+      "loss": 0.4324,
+      "step": 4434
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.008139357359817e-06,
+      "loss": 0.384,
+      "step": 4435
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.003841926991025e-06,
+      "loss": 0.4406,
+      "step": 4436
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.999545374701151e-06,
+      "loss": 0.4227,
+      "step": 4437
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.9952497014342784e-06,
+      "loss": 0.4307,
+      "step": 4438
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.990954908134295e-06,
+      "loss": 0.4133,
+      "step": 4439
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.986660995744906e-06,
+      "loss": 0.4049,
+      "step": 4440
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.982367965209608e-06,
+      "loss": 0.4332,
+      "step": 4441
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.978075817471722e-06,
+      "loss": 0.3866,
+      "step": 4442
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.97378455347436e-06,
+      "loss": 0.4014,
+      "step": 4443
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.96949417416044e-06,
+      "loss": 0.3944,
+      "step": 4444
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9652046804726996e-06,
+      "loss": 0.4034,
+      "step": 4445
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.960916073353662e-06,
+      "loss": 0.4212,
+      "step": 4446
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.956628353745676e-06,
+      "loss": 0.4373,
+      "step": 4447
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.952341522590878e-06,
+      "loss": 0.388,
+      "step": 4448
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.948055580831225e-06,
+      "loss": 0.4051,
+      "step": 4449
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.943770529408464e-06,
+      "loss": 0.4044,
+      "step": 4450
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.939486369264151e-06,
+      "loss": 0.4093,
+      "step": 4451
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.935203101339656e-06,
+      "loss": 0.4453,
+      "step": 4452
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.930920726576135e-06,
+      "loss": 0.4341,
+      "step": 4453
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.926639245914565e-06,
+      "loss": 0.4333,
+      "step": 4454
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.922358660295718e-06,
+      "loss": 0.4045,
+      "step": 4455
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.918078970660164e-06,
+      "loss": 0.3799,
+      "step": 4456
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.913800177948291e-06,
+      "loss": 0.3911,
+      "step": 4457
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.909522283100276e-06,
+      "loss": 0.4034,
+      "step": 4458
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.905245287056109e-06,
+      "loss": 0.437,
+      "step": 4459
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.90096919075557e-06,
+      "loss": 0.4277,
+      "step": 4460
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.896693995138256e-06,
+      "loss": 0.4133,
+      "step": 4461
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.892419701143559e-06,
+      "loss": 0.4409,
+      "step": 4462
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.888146309710665e-06,
+      "loss": 0.397,
+      "step": 4463
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.88387382177858e-06,
+      "loss": 0.4427,
+      "step": 4464
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.879602238286094e-06,
+      "loss": 0.4301,
+      "step": 4465
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.87533156017181e-06,
+      "loss": 0.4329,
+      "step": 4466
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.871061788374123e-06,
+      "loss": 0.4215,
+      "step": 4467
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.866792923831235e-06,
+      "loss": 0.4199,
+      "step": 4468
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.862524967481149e-06,
+      "loss": 0.4077,
+      "step": 4469
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.8582579202616605e-06,
+      "loss": 0.4162,
+      "step": 4470
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.85399178311038e-06,
+      "loss": 0.4236,
+      "step": 4471
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.849726556964701e-06,
+      "loss": 0.4415,
+      "step": 4472
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.84546224276183e-06,
+      "loss": 0.4474,
+      "step": 4473
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.841198841438768e-06,
+      "loss": 0.4253,
+      "step": 4474
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8369363539323105e-06,
+      "loss": 0.4251,
+      "step": 4475
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8326747811790644e-06,
+      "loss": 0.4311,
+      "step": 4476
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.828414124115419e-06,
+      "loss": 0.4279,
+      "step": 4477
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.824154383677582e-06,
+      "loss": 0.4121,
+      "step": 4478
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.819895560801545e-06,
+      "loss": 0.4008,
+      "step": 4479
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.815637656423096e-06,
+      "loss": 0.4016,
+      "step": 4480
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.811380671477841e-06,
+      "loss": 0.4255,
+      "step": 4481
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.807124606901158e-06,
+      "loss": 0.417,
+      "step": 4482
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.802869463628244e-06,
+      "loss": 0.4283,
+      "step": 4483
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.798615242594083e-06,
+      "loss": 0.4073,
+      "step": 4484
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.794361944733458e-06,
+      "loss": 0.4196,
+      "step": 4485
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.790109570980948e-06,
+      "loss": 0.4379,
+      "step": 4486
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7858581222709255e-06,
+      "loss": 0.4171,
+      "step": 4487
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.781607599537575e-06,
+      "loss": 0.3943,
+      "step": 4488
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.777358003714858e-06,
+      "loss": 0.4085,
+      "step": 4489
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.77310933573655e-06,
+      "loss": 0.4296,
+      "step": 4490
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.768861596536209e-06,
+      "loss": 0.399,
+      "step": 4491
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.764614787047191e-06,
+      "loss": 0.4288,
+      "step": 4492
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.7603689082026605e-06,
+      "loss": 0.4017,
+      "step": 4493
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.756123960935556e-06,
+      "loss": 0.4094,
+      "step": 4494
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.751879946178632e-06,
+      "loss": 0.3953,
+      "step": 4495
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.747636864864423e-06,
+      "loss": 0.4022,
+      "step": 4496
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.743394717925273e-06,
+      "loss": 0.4098,
+      "step": 4497
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.739153506293304e-06,
+      "loss": 0.4097,
+      "step": 4498
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.734913230900443e-06,
+      "loss": 0.4251,
+      "step": 4499
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.730673892678411e-06,
+      "loss": 0.4166,
+      "step": 4500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.726435492558721e-06,
+      "loss": 0.4104,
+      "step": 4501
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.722198031472678e-06,
+      "loss": 0.4382,
+      "step": 4502
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.717961510351383e-06,
+      "loss": 0.3999,
+      "step": 4503
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.713725930125727e-06,
+      "loss": 0.4482,
+      "step": 4504
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.709491291726404e-06,
+      "loss": 0.4292,
+      "step": 4505
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.705257596083886e-06,
+      "loss": 0.4432,
+      "step": 4506
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.701024844128458e-06,
+      "loss": 0.4362,
+      "step": 4507
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.696793036790177e-06,
+      "loss": 0.4267,
+      "step": 4508
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.692562174998899e-06,
+      "loss": 0.4401,
+      "step": 4509
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.6883322596842845e-06,
+      "loss": 0.4142,
+      "step": 4510
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.684103291775767e-06,
+      "loss": 0.4426,
+      "step": 4511
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.679875272202586e-06,
+      "loss": 0.4045,
+      "step": 4512
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.675648201893764e-06,
+      "loss": 0.4198,
+      "step": 4513
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.671422081778124e-06,
+      "loss": 0.4178,
+      "step": 4514
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.667196912784271e-06,
+      "loss": 0.42,
+      "step": 4515
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.662972695840605e-06,
+      "loss": 0.4136,
+      "step": 4516
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6587494318753145e-06,
+      "loss": 0.4392,
+      "step": 4517
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.65452712181638e-06,
+      "loss": 0.4106,
+      "step": 4518
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6503057665915795e-06,
+      "loss": 0.4137,
+      "step": 4519
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.646085367128469e-06,
+      "loss": 0.4303,
+      "step": 4520
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.641865924354398e-06,
+      "loss": 0.4483,
+      "step": 4521
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6376474391965166e-06,
+      "loss": 0.4472,
+      "step": 4522
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.633429912581745e-06,
+      "loss": 0.4297,
+      "step": 4523
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.629213345436815e-06,
+      "loss": 0.4214,
+      "step": 4524
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.624997738688225e-06,
+      "loss": 0.4162,
+      "step": 4525
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.620783093262283e-06,
+      "loss": 0.4267,
+      "step": 4526
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.616569410085073e-06,
+      "loss": 0.4033,
+      "step": 4527
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.612356690082468e-06,
+      "loss": 0.4375,
+      "step": 4528
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.608144934180135e-06,
+      "loss": 0.458,
+      "step": 4529
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.603934143303529e-06,
+      "loss": 0.4431,
+      "step": 4530
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5997243183778845e-06,
+      "loss": 0.4396,
+      "step": 4531
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.59551546032823e-06,
+      "loss": 0.423,
+      "step": 4532
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.591307570079388e-06,
+      "loss": 0.4108,
+      "step": 4533
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.587100648555954e-06,
+      "loss": 0.4167,
+      "step": 4534
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5828946966823195e-06,
+      "loss": 0.4092,
+      "step": 4535
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.578689715382667e-06,
+      "loss": 0.4479,
+      "step": 4536
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.574485705580952e-06,
+      "loss": 0.4129,
+      "step": 4537
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.570282668200932e-06,
+      "loss": 0.4386,
+      "step": 4538
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.566080604166144e-06,
+      "loss": 0.3947,
+      "step": 4539
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.561879514399901e-06,
+      "loss": 0.4294,
+      "step": 4540
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.557679399825324e-06,
+      "loss": 0.4118,
+      "step": 4541
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.553480261365297e-06,
+      "loss": 0.3969,
+      "step": 4542
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5492820999425104e-06,
+      "loss": 0.3934,
+      "step": 4543
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.545084916479422e-06,
+      "loss": 0.4172,
+      "step": 4544
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.54088871189828e-06,
+      "loss": 0.434,
+      "step": 4545
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.53669348712113e-06,
+      "loss": 0.4118,
+      "step": 4546
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5324992430697835e-06,
+      "loss": 0.4264,
+      "step": 4547
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.528305980665848e-06,
+      "loss": 0.424,
+      "step": 4548
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5241137008307076e-06,
+      "loss": 0.4453,
+      "step": 4549
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.519922404485544e-06,
+      "loss": 0.4121,
+      "step": 4550
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.515732092551308e-06,
+      "loss": 0.448,
+      "step": 4551
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.511542765948738e-06,
+      "loss": 0.4494,
+      "step": 4552
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.507354425598365e-06,
+      "loss": 0.4287,
+      "step": 4553
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.503167072420489e-06,
+      "loss": 0.4376,
+      "step": 4554
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.498980707335207e-06,
+      "loss": 0.4255,
+      "step": 4555
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.49479533126239e-06,
+      "loss": 0.422,
+      "step": 4556
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.490610945121688e-06,
+      "loss": 0.4343,
+      "step": 4557
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.48642754983255e-06,
+      "loss": 0.4139,
+      "step": 4558
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.482245146314186e-06,
+      "loss": 0.4524,
+      "step": 4559
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.47806373548561e-06,
+      "loss": 0.4087,
+      "step": 4560
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.473883318265598e-06,
+      "loss": 0.4147,
+      "step": 4561
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.469703895572721e-06,
+      "loss": 0.4318,
+      "step": 4562
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.465525468325323e-06,
+      "loss": 0.4221,
+      "step": 4563
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.461348037441533e-06,
+      "loss": 0.4205,
+      "step": 4564
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.457171603839266e-06,
+      "loss": 0.4204,
+      "step": 4565
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.452996168436206e-06,
+      "loss": 0.4416,
+      "step": 4566
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4488217321498315e-06,
+      "loss": 0.3973,
+      "step": 4567
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.444648295897392e-06,
+      "loss": 0.4152,
+      "step": 4568
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.440475860595917e-06,
+      "loss": 0.432,
+      "step": 4569
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.436304427162225e-06,
+      "loss": 0.4275,
+      "step": 4570
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.432133996512901e-06,
+      "loss": 0.3941,
+      "step": 4571
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.427964569564324e-06,
+      "loss": 0.4275,
+      "step": 4572
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.423796147232639e-06,
+      "loss": 0.4321,
+      "step": 4573
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.419628730433784e-06,
+      "loss": 0.427,
+      "step": 4574
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.415462320083465e-06,
+      "loss": 0.4464,
+      "step": 4575
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.411296917097168e-06,
+      "loss": 0.4104,
+      "step": 4576
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.407132522390161e-06,
+      "loss": 0.4197,
+      "step": 4577
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.4029691368774936e-06,
+      "loss": 0.4455,
+      "step": 4578
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.398806761473988e-06,
+      "loss": 0.3972,
+      "step": 4579
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.3946453970942445e-06,
+      "loss": 0.4275,
+      "step": 4580
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.390485044652642e-06,
+      "loss": 0.44,
+      "step": 4581
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.386325705063342e-06,
+      "loss": 0.4191,
+      "step": 4582
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.382167379240275e-06,
+      "loss": 0.4004,
+      "step": 4583
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.378010068097158e-06,
+      "loss": 0.4313,
+      "step": 4584
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.373853772547475e-06,
+      "loss": 0.4282,
+      "step": 4585
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.369698493504497e-06,
+      "loss": 0.4205,
+      "step": 4586
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3655442318812655e-06,
+      "loss": 0.4197,
+      "step": 4587
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3613909885905945e-06,
+      "loss": 0.4331,
+      "step": 4588
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3572387645450874e-06,
+      "loss": 0.4191,
+      "step": 4589
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.353087560657107e-06,
+      "loss": 0.4168,
+      "step": 4590
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.34893737783881e-06,
+      "loss": 0.4169,
+      "step": 4591
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.344788217002115e-06,
+      "loss": 0.424,
+      "step": 4592
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.340640079058718e-06,
+      "loss": 0.4306,
+      "step": 4593
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.3364929649200946e-06,
+      "loss": 0.3932,
+      "step": 4594
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.332346875497491e-06,
+      "loss": 0.4088,
+      "step": 4595
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.328201811701935e-06,
+      "loss": 0.4179,
+      "step": 4596
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.324057774444224e-06,
+      "loss": 0.4176,
+      "step": 4597
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.3199147646349235e-06,
+      "loss": 0.4359,
+      "step": 4598
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.315772783184391e-06,
+      "loss": 0.4358,
+      "step": 4599
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.311631831002737e-06,
+      "loss": 0.4251,
+      "step": 4600
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.307491908999867e-06,
+      "loss": 0.4491,
+      "step": 4601
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.303353018085437e-06,
+      "loss": 0.3876,
+      "step": 4602
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.299215159168902e-06,
+      "loss": 0.4298,
+      "step": 4603
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.295078333159469e-06,
+      "loss": 0.3979,
+      "step": 4604
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.290942540966123e-06,
+      "loss": 0.4055,
+      "step": 4605
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.286807783497633e-06,
+      "loss": 0.424,
+      "step": 4606
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.282674061662528e-06,
+      "loss": 0.4224,
+      "step": 4607
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.2785413763691155e-06,
+      "loss": 0.4305,
+      "step": 4608
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.274409728525472e-06,
+      "loss": 0.4075,
+      "step": 4609
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.270279119039443e-06,
+      "loss": 0.4521,
+      "step": 4610
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2661495488186625e-06,
+      "loss": 0.3855,
+      "step": 4611
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2620210187705116e-06,
+      "loss": 0.4132,
+      "step": 4612
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2578935298021646e-06,
+      "loss": 0.4097,
+      "step": 4613
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.253767082820552e-06,
+      "loss": 0.412,
+      "step": 4614
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.2496416787323865e-06,
+      "loss": 0.4438,
+      "step": 4615
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.245517318444145e-06,
+      "loss": 0.4423,
+      "step": 4616
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.24139400286207e-06,
+      "loss": 0.3998,
+      "step": 4617
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.237271732892189e-06,
+      "loss": 0.3862,
+      "step": 4618
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.233150509440285e-06,
+      "loss": 0.4299,
+      "step": 4619
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.229030333411927e-06,
+      "loss": 0.4256,
+      "step": 4620
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.224911205712435e-06,
+      "loss": 0.4219,
+      "step": 4621
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.22079312724691e-06,
+      "loss": 0.4263,
+      "step": 4622
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.216676098920226e-06,
+      "loss": 0.4228,
+      "step": 4623
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.212560121637016e-06,
+      "loss": 0.4039,
+      "step": 4624
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.208445196301688e-06,
+      "loss": 0.4136,
+      "step": 4625
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.204331323818413e-06,
+      "loss": 0.4075,
+      "step": 4626
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.200218505091146e-06,
+      "loss": 0.3922,
+      "step": 4627
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.196106741023592e-06,
+      "loss": 0.4204,
+      "step": 4628
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.191996032519231e-06,
+      "loss": 0.4263,
+      "step": 4629
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.187886380481318e-06,
+      "loss": 0.432,
+      "step": 4630
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.183777785812865e-06,
+      "loss": 0.3994,
+      "step": 4631
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.179670249416662e-06,
+      "loss": 0.4094,
+      "step": 4632
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.175563772195257e-06,
+      "loss": 0.4417,
+      "step": 4633
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.171458355050967e-06,
+      "loss": 0.438,
+      "step": 4634
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.167353998885886e-06,
+      "loss": 0.4027,
+      "step": 4635
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.163250704601861e-06,
+      "loss": 0.4258,
+      "step": 4636
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.159148473100516e-06,
+      "loss": 0.4364,
+      "step": 4637
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1550473052832365e-06,
+      "loss": 0.4231,
+      "step": 4638
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.150947202051175e-06,
+      "loss": 0.4102,
+      "step": 4639
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1468481643052506e-06,
+      "loss": 0.4305,
+      "step": 4640
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1427501929461424e-06,
+      "loss": 0.4338,
+      "step": 4641
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.138653288874311e-06,
+      "loss": 0.4282,
+      "step": 4642
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.134557452989963e-06,
+      "loss": 0.4257,
+      "step": 4643
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.130462686193087e-06,
+      "loss": 0.4196,
+      "step": 4644
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.126368989383427e-06,
+      "loss": 0.422,
+      "step": 4645
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.122276363460488e-06,
+      "loss": 0.4095,
+      "step": 4646
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.118184809323556e-06,
+      "loss": 0.4128,
+      "step": 4647
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.114094327871664e-06,
+      "loss": 0.4476,
+      "step": 4648
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1100049200036205e-06,
+      "loss": 0.4172,
+      "step": 4649
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1059165866179905e-06,
+      "loss": 0.4216,
+      "step": 4650
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.101829328613113e-06,
+      "loss": 0.4444,
+      "step": 4651
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.09774314688708e-06,
+      "loss": 0.4486,
+      "step": 4652
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.093658042337753e-06,
+      "loss": 0.3988,
+      "step": 4653
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.089574015862753e-06,
+      "loss": 0.4111,
+      "step": 4654
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.085491068359463e-06,
+      "loss": 0.428,
+      "step": 4655
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.081409200725043e-06,
+      "loss": 0.4097,
+      "step": 4656
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.077328413856395e-06,
+      "loss": 0.3945,
+      "step": 4657
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.073248708650196e-06,
+      "loss": 0.4118,
+      "step": 4658
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.069170086002886e-06,
+      "loss": 0.4428,
+      "step": 4659
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.0650925468106595e-06,
+      "loss": 0.3946,
+      "step": 4660
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.061016091969482e-06,
+      "loss": 0.4235,
+      "step": 4661
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.05694072237507e-06,
+      "loss": 0.4152,
+      "step": 4662
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.052866438922917e-06,
+      "loss": 0.4251,
+      "step": 4663
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.048793242508261e-06,
+      "loss": 0.4114,
+      "step": 4664
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.044721134026106e-06,
+      "loss": 0.4141,
+      "step": 4665
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.040650114371228e-06,
+      "loss": 0.4288,
+      "step": 4666
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.036580184438149e-06,
+      "loss": 0.4093,
+      "step": 4667
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.032511345121163e-06,
+      "loss": 0.4192,
+      "step": 4668
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.028443597314317e-06,
+      "loss": 0.4221,
+      "step": 4669
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.024376941911417e-06,
+      "loss": 0.4579,
+      "step": 4670
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.020311379806037e-06,
+      "loss": 0.4309,
+      "step": 4671
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.016246911891502e-06,
+      "loss": 0.4347,
+      "step": 4672
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.012183539060904e-06,
+      "loss": 0.4352,
+      "step": 4673
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.008121262207088e-06,
+      "loss": 0.4351,
+      "step": 4674
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.004060082222666e-06,
+      "loss": 0.4463,
+      "step": 4675
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.4446,
+      "step": 4676
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.995941016431217e-06,
+      "loss": 0.4187,
+      "step": 4677
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.991883132408201e-06,
+      "loss": 0.4203,
+      "step": 4678
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.9878263488225895e-06,
+      "loss": 0.41,
+      "step": 4679
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.983770666565791e-06,
+      "loss": 0.4269,
+      "step": 4680
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.979716086528958e-06,
+      "loss": 0.4235,
+      "step": 4681
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9756626096030035e-06,
+      "loss": 0.442,
+      "step": 4682
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.971610236678609e-06,
+      "loss": 0.4089,
+      "step": 4683
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.967558968646202e-06,
+      "loss": 0.4304,
+      "step": 4684
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.96350880639597e-06,
+      "loss": 0.39,
+      "step": 4685
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.95945975081786e-06,
+      "loss": 0.423,
+      "step": 4686
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.955411802801567e-06,
+      "loss": 0.4237,
+      "step": 4687
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.951364963236561e-06,
+      "loss": 0.4125,
+      "step": 4688
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.947319233012048e-06,
+      "loss": 0.4031,
+      "step": 4689
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.9432746130170075e-06,
+      "loss": 0.4231,
+      "step": 4690
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.939231104140157e-06,
+      "loss": 0.4275,
+      "step": 4691
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.935188707269992e-06,
+      "loss": 0.4132,
+      "step": 4692
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.931147423294743e-06,
+      "loss": 0.4291,
+      "step": 4693
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.927107253102404e-06,
+      "loss": 0.4007,
+      "step": 4694
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.923068197580732e-06,
+      "loss": 0.3871,
+      "step": 4695
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.919030257617221e-06,
+      "loss": 0.4167,
+      "step": 4696
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.914993434099143e-06,
+      "loss": 0.424,
+      "step": 4697
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.910957727913504e-06,
+      "loss": 0.4352,
+      "step": 4698
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.906923139947075e-06,
+      "loss": 0.4048,
+      "step": 4699
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.902889671086381e-06,
+      "loss": 0.4148,
+      "step": 4700
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.898857322217692e-06,
+      "loss": 0.434,
+      "step": 4701
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.89482609422705e-06,
+      "loss": 0.3925,
+      "step": 4702
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.89079598800023e-06,
+      "loss": 0.3973,
+      "step": 4703
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.8867670044227776e-06,
+      "loss": 0.4452,
+      "step": 4704
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.882739144379984e-06,
+      "loss": 0.4231,
+      "step": 4705
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.878712408756885e-06,
+      "loss": 0.4311,
+      "step": 4706
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.87468679843829e-06,
+      "loss": 0.4293,
+      "step": 4707
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.870662314308742e-06,
+      "loss": 0.4149,
+      "step": 4708
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.866638957252549e-06,
+      "loss": 0.4196,
+      "step": 4709
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.862616728153764e-06,
+      "loss": 0.4026,
+      "step": 4710
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.858595627896191e-06,
+      "loss": 0.4193,
+      "step": 4711
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.854575657363396e-06,
+      "loss": 0.4097,
+      "step": 4712
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8505568174386844e-06,
+      "loss": 0.4281,
+      "step": 4713
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.846539109005125e-06,
+      "loss": 0.4307,
+      "step": 4714
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.842522532945529e-06,
+      "loss": 0.4278,
+      "step": 4715
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.838507090142461e-06,
+      "loss": 0.4262,
+      "step": 4716
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.834492781478239e-06,
+      "loss": 0.3914,
+      "step": 4717
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.830479607834925e-06,
+      "loss": 0.4085,
+      "step": 4718
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.826467570094344e-06,
+      "loss": 0.4218,
+      "step": 4719
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.822456669138058e-06,
+      "loss": 0.425,
+      "step": 4720
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.818446905847392e-06,
+      "loss": 0.4388,
+      "step": 4721
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.81443828110341e-06,
+      "loss": 0.4307,
+      "step": 4722
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.810430795786928e-06,
+      "loss": 0.4251,
+      "step": 4723
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.8064244507785205e-06,
+      "loss": 0.3987,
+      "step": 4724
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.802419246958497e-06,
+      "loss": 0.4067,
+      "step": 4725
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.798415185206933e-06,
+      "loss": 0.436,
+      "step": 4726
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.794412266403635e-06,
+      "loss": 0.4142,
+      "step": 4727
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.790410491428176e-06,
+      "loss": 0.4261,
+      "step": 4728
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.7864098611598645e-06,
+      "loss": 0.4061,
+      "step": 4729
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.782410376477762e-06,
+      "loss": 0.4137,
+      "step": 4730
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.778412038260679e-06,
+      "loss": 0.4233,
+      "step": 4731
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.774414847387171e-06,
+      "loss": 0.4269,
+      "step": 4732
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.770418804735548e-06,
+      "loss": 0.4242,
+      "step": 4733
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.766423911183863e-06,
+      "loss": 0.4418,
+      "step": 4734
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.762430167609911e-06,
+      "loss": 0.4451,
+      "step": 4735
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.75843757489125e-06,
+      "loss": 0.4357,
+      "step": 4736
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7544461339051675e-06,
+      "loss": 0.4006,
+      "step": 4737
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.750455845528712e-06,
+      "loss": 0.441,
+      "step": 4738
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.746466710638665e-06,
+      "loss": 0.4228,
+      "step": 4739
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7424787301115725e-06,
+      "loss": 0.4168,
+      "step": 4740
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.73849190482371e-06,
+      "loss": 0.4129,
+      "step": 4741
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.734506235651104e-06,
+      "loss": 0.4259,
+      "step": 4742
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.730521723469534e-06,
+      "loss": 0.4088,
+      "step": 4743
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7265383691545184e-06,
+      "loss": 0.4211,
+      "step": 4744
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.722556173581321e-06,
+      "loss": 0.4376,
+      "step": 4745
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.71857513762495e-06,
+      "loss": 0.4276,
+      "step": 4746
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.7145952621601685e-06,
+      "loss": 0.4516,
+      "step": 4747
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.710616548061474e-06,
+      "loss": 0.3967,
+      "step": 4748
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.706638996203109e-06,
+      "loss": 0.4164,
+      "step": 4749
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.702662607459072e-06,
+      "loss": 0.4254,
+      "step": 4750
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.698687382703089e-06,
+      "loss": 0.4066,
+      "step": 4751
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.694713322808648e-06,
+      "loss": 0.4262,
+      "step": 4752
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.690740428648967e-06,
+      "loss": 0.4207,
+      "step": 4753
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.686768701097012e-06,
+      "loss": 0.4163,
+      "step": 4754
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.682798141025499e-06,
+      "loss": 0.4165,
+      "step": 4755
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.678828749306877e-06,
+      "loss": 0.4176,
+      "step": 4756
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.674860526813347e-06,
+      "loss": 0.4296,
+      "step": 4757
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.67089347441685e-06,
+      "loss": 0.3927,
+      "step": 4758
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.666927592989064e-06,
+      "loss": 0.4212,
+      "step": 4759
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.662962883401422e-06,
+      "loss": 0.4094,
+      "step": 4760
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.658999346525089e-06,
+      "loss": 0.4236,
+      "step": 4761
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6550369832309794e-06,
+      "loss": 0.3879,
+      "step": 4762
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.651075794389742e-06,
+      "loss": 0.4172,
+      "step": 4763
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6471157808717706e-06,
+      "loss": 0.4259,
+      "step": 4764
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.643156943547208e-06,
+      "loss": 0.414,
+      "step": 4765
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.639199283285927e-06,
+      "loss": 0.4178,
+      "step": 4766
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.635242800957553e-06,
+      "loss": 0.4643,
+      "step": 4767
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.63128749743144e-06,
+      "loss": 0.4289,
+      "step": 4768
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.627333373576699e-06,
+      "loss": 0.4158,
+      "step": 4769
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.623380430262168e-06,
+      "loss": 0.4269,
+      "step": 4770
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.619428668356425e-06,
+      "loss": 0.4178,
+      "step": 4771
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6154780887278014e-06,
+      "loss": 0.3986,
+      "step": 4772
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.611528692244357e-06,
+      "loss": 0.4382,
+      "step": 4773
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6075804797739e-06,
+      "loss": 0.4143,
+      "step": 4774
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.6036334521839705e-06,
+      "loss": 0.4221,
+      "step": 4775
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.599687610341853e-06,
+      "loss": 0.385,
+      "step": 4776
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.595742955114571e-06,
+      "loss": 0.4423,
+      "step": 4777
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.591799487368882e-06,
+      "loss": 0.4247,
+      "step": 4778
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.587857207971294e-06,
+      "loss": 0.4146,
+      "step": 4779
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.583916117788041e-06,
+      "loss": 0.4351,
+      "step": 4780
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.579976217685108e-06,
+      "loss": 0.4291,
+      "step": 4781
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.576037508528211e-06,
+      "loss": 0.4101,
+      "step": 4782
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.572099991182799e-06,
+      "loss": 0.4478,
+      "step": 4783
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.5681636665140735e-06,
+      "loss": 0.4364,
+      "step": 4784
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.564228535386961e-06,
+      "loss": 0.439,
+      "step": 4785
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.560294598666136e-06,
+      "loss": 0.4207,
+      "step": 4786
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.556361857216004e-06,
+      "loss": 0.4323,
+      "step": 4787
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.552430311900703e-06,
+      "loss": 0.4247,
+      "step": 4788
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.548499963584125e-06,
+      "loss": 0.4176,
+      "step": 4789
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.54457081312988e-06,
+      "loss": 0.417,
+      "step": 4790
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.540642861401326e-06,
+      "loss": 0.4158,
+      "step": 4791
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.536716109261556e-06,
+      "loss": 0.4174,
+      "step": 4792
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5327905575733985e-06,
+      "loss": 0.4205,
+      "step": 4793
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5288662071994185e-06,
+      "loss": 0.4149,
+      "step": 4794
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.524943059001909e-06,
+      "loss": 0.4157,
+      "step": 4795
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.521021113842918e-06,
+      "loss": 0.4068,
+      "step": 4796
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5171003725842075e-06,
+      "loss": 0.3952,
+      "step": 4797
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.513180836087294e-06,
+      "loss": 0.4105,
+      "step": 4798
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.509262505213415e-06,
+      "loss": 0.413,
+      "step": 4799
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.505345380823546e-06,
+      "loss": 0.4017,
+      "step": 4800
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.501429463778406e-06,
+      "loss": 0.4101,
+      "step": 4801
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.497514754938437e-06,
+      "loss": 0.4004,
+      "step": 4802
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.493601255163826e-06,
+      "loss": 0.4287,
+      "step": 4803
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.489688965314484e-06,
+      "loss": 0.4008,
+      "step": 4804
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.485777886250068e-06,
+      "loss": 0.4281,
+      "step": 4805
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.481868018829959e-06,
+      "loss": 0.4071,
+      "step": 4806
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4779593639132745e-06,
+      "loss": 0.4172,
+      "step": 4807
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.474051922358867e-06,
+      "loss": 0.4209,
+      "step": 4808
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.470145695025318e-06,
+      "loss": 0.4292,
+      "step": 4809
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.466240682770951e-06,
+      "loss": 0.4143,
+      "step": 4810
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.462336886453819e-06,
+      "loss": 0.4328,
+      "step": 4811
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.458434306931697e-06,
+      "loss": 0.4491,
+      "step": 4812
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.454532945062111e-06,
+      "loss": 0.4498,
+      "step": 4813
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.450632801702305e-06,
+      "loss": 0.415,
+      "step": 4814
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.446733877709265e-06,
+      "loss": 0.4219,
+      "step": 4815
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.442836173939698e-06,
+      "loss": 0.4438,
+      "step": 4816
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4389396912500594e-06,
+      "loss": 0.4151,
+      "step": 4817
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.435044430496519e-06,
+      "loss": 0.438,
+      "step": 4818
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.431150392534983e-06,
+      "loss": 0.4174,
+      "step": 4819
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4272575782211e-06,
+      "loss": 0.4361,
+      "step": 4820
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.423365988410236e-06,
+      "loss": 0.4033,
+      "step": 4821
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.4194756239574945e-06,
+      "loss": 0.4005,
+      "step": 4822
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.415586485717706e-06,
+      "loss": 0.4206,
+      "step": 4823
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.411698574545432e-06,
+      "loss": 0.4146,
+      "step": 4824
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.407811891294973e-06,
+      "loss": 0.4299,
+      "step": 4825
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.403926436820346e-06,
+      "loss": 0.4157,
+      "step": 4826
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.400042211975313e-06,
+      "loss": 0.4146,
+      "step": 4827
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.3961592176133485e-06,
+      "loss": 0.4378,
+      "step": 4828
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.392277454587673e-06,
+      "loss": 0.4478,
+      "step": 4829
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.388396923751228e-06,
+      "loss": 0.4198,
+      "step": 4830
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.3845176259566805e-06,
+      "loss": 0.4113,
+      "step": 4831
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.380639562056439e-06,
+      "loss": 0.4283,
+      "step": 4832
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.376762732902626e-06,
+      "loss": 0.4038,
+      "step": 4833
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.372887139347108e-06,
+      "loss": 0.4227,
+      "step": 4834
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.369012782241469e-06,
+      "loss": 0.4044,
+      "step": 4835
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.365139662437019e-06,
+      "loss": 0.4549,
+      "step": 4836
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.36126778078481e-06,
+      "loss": 0.4156,
+      "step": 4837
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.35739713813561e-06,
+      "loss": 0.4484,
+      "step": 4838
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.35352773533992e-06,
+      "loss": 0.4187,
+      "step": 4839
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.349659573247959e-06,
+      "loss": 0.4198,
+      "step": 4840
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.3457926527096925e-06,
+      "loss": 0.4152,
+      "step": 4841
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.341926974574797e-06,
+      "loss": 0.425,
+      "step": 4842
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.338062539692677e-06,
+      "loss": 0.4127,
+      "step": 4843
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.334199348912474e-06,
+      "loss": 0.4232,
+      "step": 4844
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.330337403083045e-06,
+      "loss": 0.3926,
+      "step": 4845
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.326476703052984e-06,
+      "loss": 0.4252,
+      "step": 4846
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.322617249670603e-06,
+      "loss": 0.4279,
+      "step": 4847
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.318759043783939e-06,
+      "loss": 0.4271,
+      "step": 4848
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.314902086240764e-06,
+      "loss": 0.4122,
+      "step": 4849
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.3110463778885645e-06,
+      "loss": 0.3981,
+      "step": 4850
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.307191919574566e-06,
+      "loss": 0.4262,
+      "step": 4851
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.303338712145708e-06,
+      "loss": 0.4313,
+      "step": 4852
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.299486756448657e-06,
+      "loss": 0.4205,
+      "step": 4853
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.295636053329808e-06,
+      "loss": 0.407,
+      "step": 4854
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.2917866036352755e-06,
+      "loss": 0.4057,
+      "step": 4855
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.2879384082109074e-06,
+      "loss": 0.4154,
+      "step": 4856
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.284091467902265e-06,
+      "loss": 0.4245,
+      "step": 4857
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.280245783554646e-06,
+      "loss": 0.4077,
+      "step": 4858
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.276401356013063e-06,
+      "loss": 0.418,
+      "step": 4859
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.27255818612225e-06,
+      "loss": 0.4112,
+      "step": 4860
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.268716274726679e-06,
+      "loss": 0.4171,
+      "step": 4861
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.264875622670526e-06,
+      "loss": 0.4078,
+      "step": 4862
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.26103623079771e-06,
+      "loss": 0.4159,
+      "step": 4863
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.257198099951859e-06,
+      "loss": 0.4208,
+      "step": 4864
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.253361230976325e-06,
+      "loss": 0.4087,
+      "step": 4865
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.249525624714195e-06,
+      "loss": 0.4065,
+      "step": 4866
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.245691282008263e-06,
+      "loss": 0.419,
+      "step": 4867
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.2418582037010545e-06,
+      "loss": 0.406,
+      "step": 4868
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.23802639063481e-06,
+      "loss": 0.4245,
+      "step": 4869
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.234195843651505e-06,
+      "loss": 0.4197,
+      "step": 4870
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.230366563592824e-06,
+      "loss": 0.4144,
+      "step": 4871
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.226538551300175e-06,
+      "loss": 0.4271,
+      "step": 4872
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.222711807614696e-06,
+      "loss": 0.4179,
+      "step": 4873
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2188863333772336e-06,
+      "loss": 0.4275,
+      "step": 4874
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.21506212942837e-06,
+      "loss": 0.4183,
+      "step": 4875
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.211239196608397e-06,
+      "loss": 0.4087,
+      "step": 4876
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.2074175357573266e-06,
+      "loss": 0.3848,
+      "step": 4877
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.203597147714903e-06,
+      "loss": 0.4034,
+      "step": 4878
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.199778033320575e-06,
+      "loss": 0.4317,
+      "step": 4879
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.195960193413527e-06,
+      "loss": 0.4085,
+      "step": 4880
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.192143628832649e-06,
+      "loss": 0.4049,
+      "step": 4881
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.188328340416564e-06,
+      "loss": 0.4148,
+      "step": 4882
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.184514329003606e-06,
+      "loss": 0.4166,
+      "step": 4883
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.1807015954318305e-06,
+      "loss": 0.4145,
+      "step": 4884
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.176890140539013e-06,
+      "loss": 0.4229,
+      "step": 4885
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.1730799651626405e-06,
+      "loss": 0.4495,
+      "step": 4886
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.169271070139938e-06,
+      "loss": 0.4264,
+      "step": 4887
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1654634563078285e-06,
+      "loss": 0.3962,
+      "step": 4888
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.161657124502962e-06,
+      "loss": 0.4368,
+      "step": 4889
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.157852075561713e-06,
+      "loss": 0.4342,
+      "step": 4890
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.154048310320158e-06,
+      "loss": 0.4285,
+      "step": 4891
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.150245829614112e-06,
+      "loss": 0.4305,
+      "step": 4892
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.146444634279089e-06,
+      "loss": 0.4495,
+      "step": 4893
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.142644725150334e-06,
+      "loss": 0.4265,
+      "step": 4894
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.138846103062803e-06,
+      "loss": 0.4251,
+      "step": 4895
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.135048768851166e-06,
+      "loss": 0.4209,
+      "step": 4896
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.131252723349819e-06,
+      "loss": 0.4123,
+      "step": 4897
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.127457967392869e-06,
+      "loss": 0.4289,
+      "step": 4898
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.123664501814138e-06,
+      "loss": 0.4155,
+      "step": 4899
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.119872327447172e-06,
+      "loss": 0.4435,
+      "step": 4900
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.116081445125219e-06,
+      "loss": 0.4286,
+      "step": 4901
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.112291855681264e-06,
+      "loss": 0.4185,
+      "step": 4902
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.108503559947989e-06,
+      "loss": 0.4256,
+      "step": 4903
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.104716558757802e-06,
+      "loss": 0.4151,
+      "step": 4904
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.100930852942822e-06,
+      "loss": 0.4544,
+      "step": 4905
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.097146443334888e-06,
+      "loss": 0.3996,
+      "step": 4906
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.09336333076555e-06,
+      "loss": 0.4387,
+      "step": 4907
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.089581516066071e-06,
+      "loss": 0.4497,
+      "step": 4908
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.0858010000674375e-06,
+      "loss": 0.4177,
+      "step": 4909
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.08202178360034e-06,
+      "loss": 0.4309,
+      "step": 4910
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.078243867495194e-06,
+      "loss": 0.4422,
+      "step": 4911
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.074467252582122e-06,
+      "loss": 0.4175,
+      "step": 4912
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.070691939690961e-06,
+      "loss": 0.4104,
+      "step": 4913
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.066917929651261e-06,
+      "loss": 0.4434,
+      "step": 4914
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0631452232922945e-06,
+      "loss": 0.422,
+      "step": 4915
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0593738214430375e-06,
+      "loss": 0.4187,
+      "step": 4916
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.055603724932181e-06,
+      "loss": 0.4161,
+      "step": 4917
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.051834934588137e-06,
+      "loss": 0.4229,
+      "step": 4918
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.048067451239022e-06,
+      "loss": 0.4154,
+      "step": 4919
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.044301275712663e-06,
+      "loss": 0.3994,
+      "step": 4920
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.040536408836613e-06,
+      "loss": 0.4176,
+      "step": 4921
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.036772851438122e-06,
+      "loss": 0.4098,
+      "step": 4922
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0330106043441665e-06,
+      "loss": 0.4043,
+      "step": 4923
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.0292496683814245e-06,
+      "loss": 0.4145,
+      "step": 4924
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.025490044376285e-06,
+      "loss": 0.4164,
+      "step": 4925
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.021731733154861e-06,
+      "loss": 0.4103,
+      "step": 4926
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.017974735542961e-06,
+      "loss": 0.399,
+      "step": 4927
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.014219052366122e-06,
+      "loss": 0.4074,
+      "step": 4928
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.010464684449577e-06,
+      "loss": 0.4063,
+      "step": 4929
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.00671163261828e-06,
+      "loss": 0.4009,
+      "step": 4930
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.002959897696887e-06,
+      "loss": 0.4048,
+      "step": 4931
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.999209480509772e-06,
+      "loss": 0.4161,
+      "step": 4932
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.995460381881019e-06,
+      "loss": 0.4193,
+      "step": 4933
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.991712602634414e-06,
+      "loss": 0.4239,
+      "step": 4934
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.9879661435934705e-06,
+      "loss": 0.4116,
+      "step": 4935
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.984221005581392e-06,
+      "loss": 0.4466,
+      "step": 4936
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.980477189421101e-06,
+      "loss": 0.4023,
+      "step": 4937
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.976734695935235e-06,
+      "loss": 0.4175,
+      "step": 4938
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.972993525946127e-06,
+      "loss": 0.4217,
+      "step": 4939
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.969253680275835e-06,
+      "loss": 0.41,
+      "step": 4940
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.965515159746115e-06,
+      "loss": 0.4153,
+      "step": 4941
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.961777965178431e-06,
+      "loss": 0.4214,
+      "step": 4942
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.958042097393967e-06,
+      "loss": 0.388,
+      "step": 4943
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9543075572136045e-06,
+      "loss": 0.4136,
+      "step": 4944
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.950574345457937e-06,
+      "loss": 0.4307,
+      "step": 4945
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9468424629472625e-06,
+      "loss": 0.4387,
+      "step": 4946
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.943111910501597e-06,
+      "loss": 0.4351,
+      "step": 4947
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.939382688940654e-06,
+      "loss": 0.4198,
+      "step": 4948
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.935654799083856e-06,
+      "loss": 0.4179,
+      "step": 4949
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9319282417503414e-06,
+      "loss": 0.3997,
+      "step": 4950
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.928203017758942e-06,
+      "loss": 0.4244,
+      "step": 4951
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.924479127928212e-06,
+      "loss": 0.4217,
+      "step": 4952
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9207565730764e-06,
+      "loss": 0.4414,
+      "step": 4953
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.917035354021465e-06,
+      "loss": 0.4114,
+      "step": 4954
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9133154715810775e-06,
+      "loss": 0.4187,
+      "step": 4955
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.909596926572604e-06,
+      "loss": 0.4304,
+      "step": 4956
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.90587971981313e-06,
+      "loss": 0.4215,
+      "step": 4957
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.902163852119439e-06,
+      "loss": 0.3959,
+      "step": 4958
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.898449324308018e-06,
+      "loss": 0.4129,
+      "step": 4959
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.894736137195061e-06,
+      "loss": 0.439,
+      "step": 4960
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.891024291596478e-06,
+      "loss": 0.4342,
+      "step": 4961
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.88731378832787e-06,
+      "loss": 0.4185,
+      "step": 4962
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.883604628204547e-06,
+      "loss": 0.4045,
+      "step": 4963
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8798968120415315e-06,
+      "loss": 0.419,
+      "step": 4964
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.876190340653541e-06,
+      "loss": 0.4157,
+      "step": 4965
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.872485214855001e-06,
+      "loss": 0.3923,
+      "step": 4966
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.868781435460044e-06,
+      "loss": 0.4153,
+      "step": 4967
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.865079003282499e-06,
+      "loss": 0.4406,
+      "step": 4968
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.861377919135913e-06,
+      "loss": 0.4239,
+      "step": 4969
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.85767818383352e-06,
+      "loss": 0.4399,
+      "step": 4970
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.853979798188272e-06,
+      "loss": 0.4297,
+      "step": 4971
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.850282763012816e-06,
+      "loss": 0.4524,
+      "step": 4972
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8465870791195e-06,
+      "loss": 0.4088,
+      "step": 4973
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.842892747320386e-06,
+      "loss": 0.4187,
+      "step": 4974
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.839199768427231e-06,
+      "loss": 0.3967,
+      "step": 4975
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.835508143251493e-06,
+      "loss": 0.4375,
+      "step": 4976
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.831817872604338e-06,
+      "loss": 0.4121,
+      "step": 4977
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.828128957296628e-06,
+      "loss": 0.4042,
+      "step": 4978
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.824441398138939e-06,
+      "loss": 0.411,
+      "step": 4979
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8207551959415314e-06,
+      "loss": 0.4294,
+      "step": 4980
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.817070351514387e-06,
+      "loss": 0.4304,
+      "step": 4981
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8133868656671726e-06,
+      "loss": 0.3946,
+      "step": 4982
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.809704739209269e-06,
+      "loss": 0.4429,
+      "step": 4983
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.8060239729497494e-06,
+      "loss": 0.4217,
+      "step": 4984
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.802344567697388e-06,
+      "loss": 0.4147,
+      "step": 4985
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7986665242606713e-06,
+      "loss": 0.4431,
+      "step": 4986
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7949898434477706e-06,
+      "loss": 0.4187,
+      "step": 4987
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.791314526066574e-06,
+      "loss": 0.3965,
+      "step": 4988
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.787640572924657e-06,
+      "loss": 0.4123,
+      "step": 4989
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7839679848293007e-06,
+      "loss": 0.4335,
+      "step": 4990
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7802967625874863e-06,
+      "loss": 0.4338,
+      "step": 4991
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7766269070058893e-06,
+      "loss": 0.4183,
+      "step": 4992
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7729584188908985e-06,
+      "loss": 0.4193,
+      "step": 4993
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.769291299048586e-06,
+      "loss": 0.4346,
+      "step": 4994
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.765625548284738e-06,
+      "loss": 0.4103,
+      "step": 4995
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7619611674048297e-06,
+      "loss": 0.4027,
+      "step": 4996
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7582981572140343e-06,
+      "loss": 0.4189,
+      "step": 4997
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.754636518517234e-06,
+      "loss": 0.426,
+      "step": 4998
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7509762521189984e-06,
+      "loss": 0.4093,
+      "step": 4999
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7473173588236055e-06,
+      "loss": 0.4242,
+      "step": 5000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7436598394350255e-06,
+      "loss": 0.4351,
+      "step": 5001
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.740003694756923e-06,
+      "loss": 0.43,
+      "step": 5002
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7363489255926732e-06,
+      "loss": 0.4276,
+      "step": 5003
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7326955327453383e-06,
+      "loss": 0.3958,
+      "step": 5004
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7290435170176763e-06,
+      "loss": 0.414,
+      "step": 5005
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.725392879212155e-06,
+      "loss": 0.4095,
+      "step": 5006
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7217436201309286e-06,
+      "loss": 0.4253,
+      "step": 5007
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.718095740575852e-06,
+      "loss": 0.4345,
+      "step": 5008
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7144492413484725e-06,
+      "loss": 0.4139,
+      "step": 5009
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7108041232500445e-06,
+      "loss": 0.3735,
+      "step": 5010
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7071603870815077e-06,
+      "loss": 0.4232,
+      "step": 5011
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.703518033643508e-06,
+      "loss": 0.4454,
+      "step": 5012
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6998770637363802e-06,
+      "loss": 0.4436,
+      "step": 5013
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.696237478160154e-06,
+      "loss": 0.4096,
+      "step": 5014
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.692599277714565e-06,
+      "loss": 0.4405,
+      "step": 5015
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6889624631990308e-06,
+      "loss": 0.4017,
+      "step": 5016
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.685327035412678e-06,
+      "loss": 0.4107,
+      "step": 5017
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.681692995154316e-06,
+      "loss": 0.3983,
+      "step": 5018
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6780603432224616e-06,
+      "loss": 0.4002,
+      "step": 5019
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.674429080415317e-06,
+      "loss": 0.444,
+      "step": 5020
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.670799207530782e-06,
+      "loss": 0.4188,
+      "step": 5021
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6671707253664512e-06,
+      "loss": 0.423,
+      "step": 5022
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.663543634719612e-06,
+      "loss": 0.4069,
+      "step": 5023
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6599179363872526e-06,
+      "loss": 0.4145,
+      "step": 5024
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.656293631166049e-06,
+      "loss": 0.4153,
+      "step": 5025
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.652670719852368e-06,
+      "loss": 0.4215,
+      "step": 5026
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6490492032422806e-06,
+      "loss": 0.4079,
+      "step": 5027
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6454290821315418e-06,
+      "loss": 0.4351,
+      "step": 5028
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6418103573156072e-06,
+      "loss": 0.4275,
+      "step": 5029
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6381930295896207e-06,
+      "loss": 0.43,
+      "step": 5030
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6345770997484154e-06,
+      "loss": 0.4523,
+      "step": 5031
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6309625685865303e-06,
+      "loss": 0.425,
+      "step": 5032
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.627349436898182e-06,
+      "loss": 0.4273,
+      "step": 5033
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6237377054772938e-06,
+      "loss": 0.4177,
+      "step": 5034
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6201273751174705e-06,
+      "loss": 0.4298,
+      "step": 5035
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.616518446612013e-06,
+      "loss": 0.4221,
+      "step": 5036
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6129109207539135e-06,
+      "loss": 0.4074,
+      "step": 5037
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6093047983358542e-06,
+      "loss": 0.4282,
+      "step": 5038
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6057000801502174e-06,
+      "loss": 0.4029,
+      "step": 5039
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.602096766989064e-06,
+      "loss": 0.4117,
+      "step": 5040
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5984948596441584e-06,
+      "loss": 0.4199,
+      "step": 5041
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5948943589069484e-06,
+      "loss": 0.4174,
+      "step": 5042
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5912952655685708e-06,
+      "loss": 0.419,
+      "step": 5043
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5876975804198646e-06,
+      "loss": 0.4369,
+      "step": 5044
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.584101304251344e-06,
+      "loss": 0.4162,
+      "step": 5045
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5805064378532296e-06,
+      "loss": 0.4016,
+      "step": 5046
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5769129820154167e-06,
+      "loss": 0.4048,
+      "step": 5047
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5733209375275056e-06,
+      "loss": 0.4323,
+      "step": 5048
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.569730305178775e-06,
+      "loss": 0.4093,
+      "step": 5049
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5661410857581935e-06,
+      "loss": 0.4231,
+      "step": 5050
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5625532800544303e-06,
+      "loss": 0.4017,
+      "step": 5051
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5589668888558348e-06,
+      "loss": 0.4241,
+      "step": 5052
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5553819129504454e-06,
+      "loss": 0.4208,
+      "step": 5053
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5517983531259926e-06,
+      "loss": 0.4042,
+      "step": 5054
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.54821621016989e-06,
+      "loss": 0.4073,
+      "step": 5055
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5446354848692543e-06,
+      "loss": 0.4163,
+      "step": 5056
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.541056178010872e-06,
+      "loss": 0.4313,
+      "step": 5057
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5374782903812345e-06,
+      "loss": 0.4177,
+      "step": 5058
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.533901822766508e-06,
+      "loss": 0.4409,
+      "step": 5059
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.530326775952556e-06,
+      "loss": 0.4045,
+      "step": 5060
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5267531507249263e-06,
+      "loss": 0.4118,
+      "step": 5061
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5231809478688504e-06,
+      "loss": 0.4146,
+      "step": 5062
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5196101681692573e-06,
+      "loss": 0.4134,
+      "step": 5063
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5160408124107505e-06,
+      "loss": 0.4317,
+      "step": 5064
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.512472881377633e-06,
+      "loss": 0.4053,
+      "step": 5065
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.508906375853888e-06,
+      "loss": 0.4255,
+      "step": 5066
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5053412966231846e-06,
+      "loss": 0.4084,
+      "step": 5067
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5017776444688813e-06,
+      "loss": 0.4364,
+      "step": 5068
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.498215420174018e-06,
+      "loss": 0.4302,
+      "step": 5069
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.494654624521333e-06,
+      "loss": 0.4209,
+      "step": 5070
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4910952582932355e-06,
+      "loss": 0.3938,
+      "step": 5071
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4875373222718346e-06,
+      "loss": 0.4135,
+      "step": 5072
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4839808172389143e-06,
+      "loss": 0.4345,
+      "step": 5073
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4804257439759446e-06,
+      "loss": 0.4645,
+      "step": 5074
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.476872103264093e-06,
+      "loss": 0.4291,
+      "step": 5075
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4733198958841962e-06,
+      "loss": 0.4064,
+      "step": 5076
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4697691226167884e-06,
+      "loss": 0.4146,
+      "step": 5077
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.466219784242083e-06,
+      "loss": 0.4435,
+      "step": 5078
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4626718815399753e-06,
+      "loss": 0.4258,
+      "step": 5079
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4591254152900533e-06,
+      "loss": 0.417,
+      "step": 5080
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.455580386271583e-06,
+      "loss": 0.4402,
+      "step": 5081
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4520367952635157e-06,
+      "loss": 0.4231,
+      "step": 5082
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4484946430444856e-06,
+      "loss": 0.4236,
+      "step": 5083
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.444953930392817e-06,
+      "loss": 0.4326,
+      "step": 5084
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4414146580865115e-06,
+      "loss": 0.4338,
+      "step": 5085
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4378768269032524e-06,
+      "loss": 0.4101,
+      "step": 5086
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.434340437620417e-06,
+      "loss": 0.4452,
+      "step": 5087
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.430805491015052e-06,
+      "loss": 0.4151,
+      "step": 5088
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4272719878639015e-06,
+      "loss": 0.4507,
+      "step": 5089
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4237399289433803e-06,
+      "loss": 0.425,
+      "step": 5090
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4202093150295868e-06,
+      "loss": 0.4164,
+      "step": 5091
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4166801468983136e-06,
+      "loss": 0.4216,
+      "step": 5092
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4131524253250204e-06,
+      "loss": 0.4272,
+      "step": 5093
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4096261510848626e-06,
+      "loss": 0.4132,
+      "step": 5094
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.406101324952665e-06,
+      "loss": 0.4229,
+      "step": 5095
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.4025779477029463e-06,
+      "loss": 0.4141,
+      "step": 5096
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3990560201098978e-06,
+      "loss": 0.4196,
+      "step": 5097
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3955355429473957e-06,
+      "loss": 0.4448,
+      "step": 5098
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3920165169889974e-06,
+      "loss": 0.4411,
+      "step": 5099
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3884989430079375e-06,
+      "loss": 0.4355,
+      "step": 5100
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3849828217771418e-06,
+      "loss": 0.4281,
+      "step": 5101
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3814681540692064e-06,
+      "loss": 0.4025,
+      "step": 5102
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3779549406564104e-06,
+      "loss": 0.4231,
+      "step": 5103
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3744431823107205e-06,
+      "loss": 0.3948,
+      "step": 5104
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.370932879803771e-06,
+      "loss": 0.439,
+      "step": 5105
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.367424033906891e-06,
+      "loss": 0.4299,
+      "step": 5106
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3639166453910775e-06,
+      "loss": 0.4353,
+      "step": 5107
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3604107150270102e-06,
+      "loss": 0.4163,
+      "step": 5108
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.356906243585054e-06,
+      "loss": 0.4257,
+      "step": 5109
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.353403231835245e-06,
+      "loss": 0.4057,
+      "step": 5110
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3499016805473074e-06,
+      "loss": 0.4099,
+      "step": 5111
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.346401590490638e-06,
+      "loss": 0.4057,
+      "step": 5112
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3429029624343133e-06,
+      "loss": 0.4426,
+      "step": 5113
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3394057971470893e-06,
+      "loss": 0.4103,
+      "step": 5114
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.335910095397399e-06,
+      "loss": 0.4214,
+      "step": 5115
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3324158579533604e-06,
+      "loss": 0.4382,
+      "step": 5116
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.32892308558276e-06,
+      "loss": 0.4171,
+      "step": 5117
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.325431779053073e-06,
+      "loss": 0.4261,
+      "step": 5118
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3219419391314444e-06,
+      "loss": 0.4243,
+      "step": 5119
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.318453566584695e-06,
+      "loss": 0.4094,
+      "step": 5120
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.314966662179335e-06,
+      "loss": 0.4201,
+      "step": 5121
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3114812266815386e-06,
+      "loss": 0.3983,
+      "step": 5122
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3079972608571687e-06,
+      "loss": 0.4375,
+      "step": 5123
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3045147654717537e-06,
+      "loss": 0.4168,
+      "step": 5124
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.3010337412905113e-06,
+      "loss": 0.4343,
+      "step": 5125
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.297554189078327e-06,
+      "loss": 0.4092,
+      "step": 5126
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.294076109599764e-06,
+      "loss": 0.4167,
+      "step": 5127
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.290599503619062e-06,
+      "loss": 0.4208,
+      "step": 5128
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2871243719001443e-06,
+      "loss": 0.4025,
+      "step": 5129
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.283650715206601e-06,
+      "loss": 0.4097,
+      "step": 5130
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.280178534301701e-06,
+      "loss": 0.4184,
+      "step": 5131
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2767078299483858e-06,
+      "loss": 0.4166,
+      "step": 5132
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2732386029092834e-06,
+      "loss": 0.4077,
+      "step": 5133
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.269770853946683e-06,
+      "loss": 0.4108,
+      "step": 5134
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2663045838225615e-06,
+      "loss": 0.4171,
+      "step": 5135
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2628397932985602e-06,
+      "loss": 0.416,
+      "step": 5136
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.259376483136005e-06,
+      "loss": 0.4086,
+      "step": 5137
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2559146540958884e-06,
+      "loss": 0.4346,
+      "step": 5138
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2524543069388804e-06,
+      "loss": 0.4266,
+      "step": 5139
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2489954424253278e-06,
+      "loss": 0.3935,
+      "step": 5140
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2455380613152474e-06,
+      "loss": 0.4324,
+      "step": 5141
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.242082164368335e-06,
+      "loss": 0.4232,
+      "step": 5142
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.238627752343958e-06,
+      "loss": 0.426,
+      "step": 5143
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.235174826001154e-06,
+      "loss": 0.3995,
+      "step": 5144
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2317233860986385e-06,
+      "loss": 0.4209,
+      "step": 5145
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.228273433394795e-06,
+      "loss": 0.4377,
+      "step": 5146
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.224824968647693e-06,
+      "loss": 0.4198,
+      "step": 5147
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.221377992615057e-06,
+      "loss": 0.4347,
+      "step": 5148
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.217932506054302e-06,
+      "loss": 0.4251,
+      "step": 5149
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.214488509722503e-06,
+      "loss": 0.3995,
+      "step": 5150
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2110460043764103e-06,
+      "loss": 0.4362,
+      "step": 5151
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2076049907724526e-06,
+      "loss": 0.3971,
+      "step": 5152
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2041654696667222e-06,
+      "loss": 0.4123,
+      "step": 5153
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2007274418149935e-06,
+      "loss": 0.4265,
+      "step": 5154
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1972909079727034e-06,
+      "loss": 0.4054,
+      "step": 5155
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1938558688949606e-06,
+      "loss": 0.4253,
+      "step": 5156
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1904223253365562e-06,
+      "loss": 0.4223,
+      "step": 5157
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.186990278051941e-06,
+      "loss": 0.4238,
+      "step": 5158
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1835597277952437e-06,
+      "loss": 0.4327,
+      "step": 5159
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1801306753202566e-06,
+      "loss": 0.4373,
+      "step": 5160
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1767031213804544e-06,
+      "loss": 0.4228,
+      "step": 5161
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1732770667289737e-06,
+      "loss": 0.4202,
+      "step": 5162
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1698525121186206e-06,
+      "loss": 0.387,
+      "step": 5163
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1664294583018828e-06,
+      "loss": 0.4162,
+      "step": 5164
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.163007906030903e-06,
+      "loss": 0.4279,
+      "step": 5165
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1595878560575065e-06,
+      "loss": 0.3887,
+      "step": 5166
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1561693091331835e-06,
+      "loss": 0.4574,
+      "step": 5167
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.152752266009089e-06,
+      "loss": 0.3973,
+      "step": 5168
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1493367274360585e-06,
+      "loss": 0.4342,
+      "step": 5169
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.145922694164585e-06,
+      "loss": 0.3925,
+      "step": 5170
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1425101669448443e-06,
+      "loss": 0.4303,
+      "step": 5171
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1390991465266684e-06,
+      "loss": 0.4164,
+      "step": 5172
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.135689633659562e-06,
+      "loss": 0.4223,
+      "step": 5173
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.132281629092705e-06,
+      "loss": 0.4319,
+      "step": 5174
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.128875133574939e-06,
+      "loss": 0.4116,
+      "step": 5175
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1254701478547746e-06,
+      "loss": 0.4339,
+      "step": 5176
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1220666726803893e-06,
+      "loss": 0.4134,
+      "step": 5177
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.118664708799638e-06,
+      "loss": 0.4585,
+      "step": 5178
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1152642569600343e-06,
+      "loss": 0.402,
+      "step": 5179
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1118653179087576e-06,
+      "loss": 0.4058,
+      "step": 5180
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.108467892392666e-06,
+      "loss": 0.4393,
+      "step": 5181
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1050719811582732e-06,
+      "loss": 0.4123,
+      "step": 5182
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.10167758495177e-06,
+      "loss": 0.4127,
+      "step": 5183
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0982847045190055e-06,
+      "loss": 0.407,
+      "step": 5184
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0948933406055047e-06,
+      "loss": 0.4274,
+      "step": 5185
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0915034939564516e-06,
+      "loss": 0.4401,
+      "step": 5186
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0881151653166976e-06,
+      "loss": 0.4073,
+      "step": 5187
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.084728355430767e-06,
+      "loss": 0.4164,
+      "step": 5188
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0813430650428466e-06,
+      "loss": 0.4261,
+      "step": 5189
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.077959294896785e-06,
+      "loss": 0.4217,
+      "step": 5190
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0745770457361036e-06,
+      "loss": 0.4324,
+      "step": 5191
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.071196318303982e-06,
+      "loss": 0.425,
+      "step": 5192
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0678171133432764e-06,
+      "loss": 0.4392,
+      "step": 5193
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0644394315964965e-06,
+      "loss": 0.3999,
+      "step": 5194
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.061063273805829e-06,
+      "loss": 0.3993,
+      "step": 5195
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0576886407131156e-06,
+      "loss": 0.4086,
+      "step": 5196
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.054315533059865e-06,
+      "loss": 0.4075,
+      "step": 5197
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0509439515872595e-06,
+      "loss": 0.4087,
+      "step": 5198
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0475738970361314e-06,
+      "loss": 0.429,
+      "step": 5199
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.044205370146992e-06,
+      "loss": 0.4186,
+      "step": 5200
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0408383716600052e-06,
+      "loss": 0.4221,
+      "step": 5201
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0374729023150106e-06,
+      "loss": 0.4171,
+      "step": 5202
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0341089628515006e-06,
+      "loss": 0.4117,
+      "step": 5203
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0307465540086378e-06,
+      "loss": 0.4043,
+      "step": 5204
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0273856765252453e-06,
+      "loss": 0.4265,
+      "step": 5205
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.02402633113981e-06,
+      "loss": 0.4206,
+      "step": 5206
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0206685185904884e-06,
+      "loss": 0.4038,
+      "step": 5207
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0173122396150924e-06,
+      "loss": 0.4224,
+      "step": 5208
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0139574949510973e-06,
+      "loss": 0.4376,
+      "step": 5209
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0106042853356497e-06,
+      "loss": 0.4119,
+      "step": 5210
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0072526115055454e-06,
+      "loss": 0.4314,
+      "step": 5211
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.0039024741972588e-06,
+      "loss": 0.4147,
+      "step": 5212
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.000553874146911e-06,
+      "loss": 0.421,
+      "step": 5213
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9972068120902975e-06,
+      "loss": 0.4169,
+      "step": 5214
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9938612887628692e-06,
+      "loss": 0.4358,
+      "step": 5215
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9905173048997384e-06,
+      "loss": 0.3989,
+      "step": 5216
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9871748612356845e-06,
+      "loss": 0.3884,
+      "step": 5217
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.983833958505141e-06,
+      "loss": 0.4283,
+      "step": 5218
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.980494597442213e-06,
+      "loss": 0.3991,
+      "step": 5219
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9771567787806575e-06,
+      "loss": 0.4384,
+      "step": 5220
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9738205032538958e-06,
+      "loss": 0.4441,
+      "step": 5221
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9704857715950118e-06,
+      "loss": 0.4045,
+      "step": 5222
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9671525845367433e-06,
+      "loss": 0.4169,
+      "step": 5223
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.963820942811502e-06,
+      "loss": 0.4116,
+      "step": 5224
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.960490847151345e-06,
+      "loss": 0.4119,
+      "step": 5225
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9571622982880044e-06,
+      "loss": 0.4238,
+      "step": 5226
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9538352969528593e-06,
+      "loss": 0.4175,
+      "step": 5227
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.950509843876953e-06,
+      "loss": 0.4352,
+      "step": 5228
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9471859397909965e-06,
+      "loss": 0.4131,
+      "step": 5229
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9438635854253463e-06,
+      "loss": 0.4285,
+      "step": 5230
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9405427815100317e-06,
+      "loss": 0.4353,
+      "step": 5231
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9372235287747343e-06,
+      "loss": 0.4228,
+      "step": 5232
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.933905827948792e-06,
+      "loss": 0.453,
+      "step": 5233
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.930589679761211e-06,
+      "loss": 0.428,
+      "step": 5234
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9272750849406506e-06,
+      "loss": 0.4222,
+      "step": 5235
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.923962044215427e-06,
+      "loss": 0.4207,
+      "step": 5236
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9206505583135134e-06,
+      "loss": 0.4306,
+      "step": 5237
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.917340627962555e-06,
+      "loss": 0.4186,
+      "step": 5238
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.914032253889839e-06,
+      "loss": 0.4273,
+      "step": 5239
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9107254368223147e-06,
+      "loss": 0.4497,
+      "step": 5240
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9074201774865985e-06,
+      "loss": 0.3963,
+      "step": 5241
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.90411647660895e-06,
+      "loss": 0.4414,
+      "step": 5242
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9008143349153016e-06,
+      "loss": 0.4218,
+      "step": 5243
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8975137531312325e-06,
+      "loss": 0.4161,
+      "step": 5244
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.894214731981978e-06,
+      "loss": 0.4083,
+      "step": 5245
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8909172721924405e-06,
+      "loss": 0.3986,
+      "step": 5246
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8876213744871684e-06,
+      "loss": 0.4137,
+      "step": 5247
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8843270395903777e-06,
+      "loss": 0.4254,
+      "step": 5248
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.881034268225932e-06,
+      "loss": 0.4131,
+      "step": 5249
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.877743061117353e-06,
+      "loss": 0.4117,
+      "step": 5250
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8744534189878225e-06,
+      "loss": 0.4152,
+      "step": 5251
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8711653425601717e-06,
+      "loss": 0.3966,
+      "step": 5252
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8678788325568996e-06,
+      "loss": 0.4232,
+      "step": 5253
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.864593889700146e-06,
+      "loss": 0.4396,
+      "step": 5254
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8613105147117205e-06,
+      "loss": 0.3963,
+      "step": 5255
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8580287083130786e-06,
+      "loss": 0.3963,
+      "step": 5256
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8547484712253317e-06,
+      "loss": 0.4235,
+      "step": 5257
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8514698041692536e-06,
+      "loss": 0.4164,
+      "step": 5258
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8481927078652637e-06,
+      "loss": 0.4123,
+      "step": 5259
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8449171830334466e-06,
+      "loss": 0.4085,
+      "step": 5260
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8416432303935293e-06,
+      "loss": 0.4072,
+      "step": 5261
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.838370850664908e-06,
+      "loss": 0.4261,
+      "step": 5262
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.835100044566619e-06,
+      "loss": 0.402,
+      "step": 5263
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8318308128173577e-06,
+      "loss": 0.4342,
+      "step": 5264
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.828563156135481e-06,
+      "loss": 0.4388,
+      "step": 5265
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8252970752389907e-06,
+      "loss": 0.4124,
+      "step": 5266
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8220325708455456e-06,
+      "loss": 0.3925,
+      "step": 5267
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8187696436724566e-06,
+      "loss": 0.4284,
+      "step": 5268
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.815508294436687e-06,
+      "loss": 0.4047,
+      "step": 5269
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8122485238548624e-06,
+      "loss": 0.421,
+      "step": 5270
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8089903326432468e-06,
+      "loss": 0.4214,
+      "step": 5271
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8057337215177725e-06,
+      "loss": 0.4301,
+      "step": 5272
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8024786911940118e-06,
+      "loss": 0.42,
+      "step": 5273
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.799225242387199e-06,
+      "loss": 0.4147,
+      "step": 5274
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7959733758122153e-06,
+      "loss": 0.4174,
+      "step": 5275
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7927230921835926e-06,
+      "loss": 0.4536,
+      "step": 5276
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.789474392215524e-06,
+      "loss": 0.4411,
+      "step": 5277
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7862272766218433e-06,
+      "loss": 0.4156,
+      "step": 5278
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.782981746116048e-06,
+      "loss": 0.4089,
+      "step": 5279
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7797378014112773e-06,
+      "loss": 0.4172,
+      "step": 5280
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7764954432203252e-06,
+      "loss": 0.4522,
+      "step": 5281
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7732546722556397e-06,
+      "loss": 0.4392,
+      "step": 5282
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.770015489229312e-06,
+      "loss": 0.4461,
+      "step": 5283
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.766777894853099e-06,
+      "loss": 0.3982,
+      "step": 5284
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.763541889838396e-06,
+      "loss": 0.4133,
+      "step": 5285
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.760307474896249e-06,
+      "loss": 0.4017,
+      "step": 5286
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7570746507373646e-06,
+      "loss": 0.4367,
+      "step": 5287
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7538434180720884e-06,
+      "loss": 0.4228,
+      "step": 5288
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.750613777610428e-06,
+      "loss": 0.4255,
+      "step": 5289
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7473857300620276e-06,
+      "loss": 0.4245,
+      "step": 5290
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7441592761361957e-06,
+      "loss": 0.4418,
+      "step": 5291
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7409344165418806e-06,
+      "loss": 0.4306,
+      "step": 5292
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7377111519876798e-06,
+      "loss": 0.4028,
+      "step": 5293
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.734489483181849e-06,
+      "loss": 0.4041,
+      "step": 5294
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.731269410832286e-06,
+      "loss": 0.4307,
+      "step": 5295
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.728050935646539e-06,
+      "loss": 0.4134,
+      "step": 5296
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7248340583318046e-06,
+      "loss": 0.441,
+      "step": 5297
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7216187795949334e-06,
+      "loss": 0.4259,
+      "step": 5298
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7184051001424205e-06,
+      "loss": 0.3967,
+      "step": 5299
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7151930206804046e-06,
+      "loss": 0.4296,
+      "step": 5300
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.7119825419146873e-06,
+      "loss": 0.3934,
+      "step": 5301
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7087736645507014e-06,
+      "loss": 0.4323,
+      "step": 5302
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7055663892935426e-06,
+      "loss": 0.4382,
+      "step": 5303
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.702360716847946e-06,
+      "loss": 0.4238,
+      "step": 5304
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6991566479182916e-06,
+      "loss": 0.4102,
+      "step": 5305
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6959541832086187e-06,
+      "loss": 0.4188,
+      "step": 5306
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6927533234226034e-06,
+      "loss": 0.4074,
+      "step": 5307
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6895540692635757e-06,
+      "loss": 0.4245,
+      "step": 5308
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6863564214345093e-06,
+      "loss": 0.4009,
+      "step": 5309
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6831603806380236e-06,
+      "loss": 0.4327,
+      "step": 5310
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6799659475763907e-06,
+      "loss": 0.4288,
+      "step": 5311
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.676773122951524e-06,
+      "loss": 0.4065,
+      "step": 5312
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.673581907464985e-06,
+      "loss": 0.4269,
+      "step": 5313
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.67039230181798e-06,
+      "loss": 0.3941,
+      "step": 5314
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6672043067113694e-06,
+      "loss": 0.415,
+      "step": 5315
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6640179228456496e-06,
+      "loss": 0.4172,
+      "step": 5316
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.660833150920965e-06,
+      "loss": 0.439,
+      "step": 5317
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6576499916371144e-06,
+      "loss": 0.4281,
+      "step": 5318
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.654468445693529e-06,
+      "loss": 0.4161,
+      "step": 5319
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6512885137892984e-06,
+      "loss": 0.3847,
+      "step": 5320
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6481101966231505e-06,
+      "loss": 0.4222,
+      "step": 5321
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6449334948934525e-06,
+      "loss": 0.4082,
+      "step": 5322
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6417584092982327e-06,
+      "loss": 0.4067,
+      "step": 5323
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6385849405351495e-06,
+      "loss": 0.4209,
+      "step": 5324
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.635413089301515e-06,
+      "loss": 0.4068,
+      "step": 5325
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6322428562942813e-06,
+      "loss": 0.3999,
+      "step": 5326
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.629074242210046e-06,
+      "loss": 0.4217,
+      "step": 5327
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6259072477450496e-06,
+      "loss": 0.4073,
+      "step": 5328
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6227418735951784e-06,
+      "loss": 0.4484,
+      "step": 5329
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6195781204559656e-06,
+      "loss": 0.4131,
+      "step": 5330
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6164159890225806e-06,
+      "loss": 0.4093,
+      "step": 5331
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6132554799898457e-06,
+      "loss": 0.4409,
+      "step": 5332
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6100965940522204e-06,
+      "loss": 0.4324,
+      "step": 5333
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.606939331903806e-06,
+      "loss": 0.411,
+      "step": 5334
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6037836942383567e-06,
+      "loss": 0.4315,
+      "step": 5335
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.600629681749255e-06,
+      "loss": 0.4311,
+      "step": 5336
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5974772951295423e-06,
+      "loss": 0.4272,
+      "step": 5337
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.594326535071888e-06,
+      "loss": 0.4168,
+      "step": 5338
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5911774022686185e-06,
+      "loss": 0.4391,
+      "step": 5339
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5880298974116914e-06,
+      "loss": 0.4454,
+      "step": 5340
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5848840211927097e-06,
+      "loss": 0.4104,
+      "step": 5341
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.581739774302917e-06,
+      "loss": 0.4051,
+      "step": 5342
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.578597157433206e-06,
+      "loss": 0.4157,
+      "step": 5343
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5754561712741055e-06,
+      "loss": 0.4427,
+      "step": 5344
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5723168165157854e-06,
+      "loss": 0.4187,
+      "step": 5345
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5691790938480542e-06,
+      "loss": 0.4058,
+      "step": 5346
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5660430039603756e-06,
+      "loss": 0.4213,
+      "step": 5347
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.562908547541836e-06,
+      "loss": 0.4345,
+      "step": 5348
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5597757252811793e-06,
+      "loss": 0.4215,
+      "step": 5349
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.556644537866777e-06,
+      "loss": 0.4157,
+      "step": 5350
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.553514985986654e-06,
+      "loss": 0.4165,
+      "step": 5351
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5503870703284638e-06,
+      "loss": 0.4528,
+      "step": 5352
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5472607915795043e-06,
+      "loss": 0.415,
+      "step": 5353
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5441361504267216e-06,
+      "loss": 0.4238,
+      "step": 5354
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.541013147556688e-06,
+      "loss": 0.4036,
+      "step": 5355
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5378917836556317e-06,
+      "loss": 0.4285,
+      "step": 5356
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5347720594094073e-06,
+      "loss": 0.4276,
+      "step": 5357
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.531653975503515e-06,
+      "loss": 0.4311,
+      "step": 5358
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5285375326230933e-06,
+      "loss": 0.4125,
+      "step": 5359
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.525422731452919e-06,
+      "loss": 0.4318,
+      "step": 5360
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5223095726774137e-06,
+      "loss": 0.4132,
+      "step": 5361
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.519198056980631e-06,
+      "loss": 0.4274,
+      "step": 5362
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5160881850462693e-06,
+      "loss": 0.4106,
+      "step": 5363
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.512979957557663e-06,
+      "loss": 0.4236,
+      "step": 5364
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.509873375197781e-06,
+      "loss": 0.4104,
+      "step": 5365
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.506768438649241e-06,
+      "loss": 0.4221,
+      "step": 5366
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5036651485942877e-06,
+      "loss": 0.4463,
+      "step": 5367
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.500563505714815e-06,
+      "loss": 0.4128,
+      "step": 5368
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.4974635106923463e-06,
+      "loss": 0.3924,
+      "step": 5369
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.4943651642080436e-06,
+      "loss": 0.4045,
+      "step": 5370
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.4912684669427133e-06,
+      "loss": 0.4086,
+      "step": 5371
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.488173419576794e-06,
+      "loss": 0.4226,
+      "step": 5372
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4850800227903617e-06,
+      "loss": 0.4212,
+      "step": 5373
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.481988277263132e-06,
+      "loss": 0.4247,
+      "step": 5374
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.478898183674452e-06,
+      "loss": 0.4332,
+      "step": 5375
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4758097427033178e-06,
+      "loss": 0.4213,
+      "step": 5376
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.472722955028348e-06,
+      "loss": 0.4224,
+      "step": 5377
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4696378213278106e-06,
+      "loss": 0.4306,
+      "step": 5378
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4665543422796e-06,
+      "loss": 0.4068,
+      "step": 5379
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4634725185612553e-06,
+      "loss": 0.4094,
+      "step": 5380
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4603923508499473e-06,
+      "loss": 0.4274,
+      "step": 5381
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4573138398224784e-06,
+      "loss": 0.4107,
+      "step": 5382
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.454236986155301e-06,
+      "loss": 0.4308,
+      "step": 5383
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4511617905244845e-06,
+      "loss": 0.4172,
+      "step": 5384
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.448088253605754e-06,
+      "loss": 0.42,
+      "step": 5385
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.445016376074455e-06,
+      "loss": 0.4004,
+      "step": 5386
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4419461586055703e-06,
+      "loss": 0.4249,
+      "step": 5387
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4388776018737293e-06,
+      "loss": 0.4212,
+      "step": 5388
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4358107065531834e-06,
+      "loss": 0.4195,
+      "step": 5389
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4327454733178247e-06,
+      "loss": 0.4138,
+      "step": 5390
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4296819028411757e-06,
+      "loss": 0.4194,
+      "step": 5391
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4266199957964032e-06,
+      "loss": 0.4039,
+      "step": 5392
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4235597528563004e-06,
+      "loss": 0.4167,
+      "step": 5393
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.420501174693293e-06,
+      "loss": 0.4202,
+      "step": 5394
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4174442619794503e-06,
+      "loss": 0.4355,
+      "step": 5395
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4143890153864645e-06,
+      "loss": 0.4106,
+      "step": 5396
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4113354355856746e-06,
+      "loss": 0.4463,
+      "step": 5397
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4082835232480404e-06,
+      "loss": 0.43,
+      "step": 5398
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4052332790441613e-06,
+      "loss": 0.4043,
+      "step": 5399
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4021847036442737e-06,
+      "loss": 0.4375,
+      "step": 5400
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.399137797718237e-06,
+      "loss": 0.4256,
+      "step": 5401
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.396092561935558e-06,
+      "loss": 0.4076,
+      "step": 5402
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3930489969653635e-06,
+      "loss": 0.4443,
+      "step": 5403
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3900071034764205e-06,
+      "loss": 0.407,
+      "step": 5404
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.386966882137125e-06,
+      "loss": 0.3917,
+      "step": 5405
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.383928333615505e-06,
+      "loss": 0.4273,
+      "step": 5406
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3808914585792296e-06,
+      "loss": 0.4122,
+      "step": 5407
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.377856257695588e-06,
+      "loss": 0.387,
+      "step": 5408
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.374822731631511e-06,
+      "loss": 0.4357,
+      "step": 5409
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.371790881053556e-06,
+      "loss": 0.4288,
+      "step": 5410
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3687607066279115e-06,
+      "loss": 0.3965,
+      "step": 5411
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.365732209020406e-06,
+      "loss": 0.433,
+      "step": 5412
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.362705388896487e-06,
+      "loss": 0.4039,
+      "step": 5413
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3596802469212455e-06,
+      "loss": 0.4113,
+      "step": 5414
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3566567837593924e-06,
+      "loss": 0.422,
+      "step": 5415
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.353635000075284e-06,
+      "loss": 0.4177,
+      "step": 5416
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3506148965328924e-06,
+      "loss": 0.4265,
+      "step": 5417
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3475964737958302e-06,
+      "loss": 0.4135,
+      "step": 5418
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3445797325273366e-06,
+      "loss": 0.4117,
+      "step": 5419
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.341564673390281e-06,
+      "loss": 0.4086,
+      "step": 5420
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.338551297047168e-06,
+      "loss": 0.423,
+      "step": 5421
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3355396041601274e-06,
+      "loss": 0.4202,
+      "step": 5422
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.332529595390919e-06,
+      "loss": 0.4049,
+      "step": 5423
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3295212714009385e-06,
+      "loss": 0.4321,
+      "step": 5424
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3265146328512035e-06,
+      "loss": 0.409,
+      "step": 5425
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3235096804023684e-06,
+      "loss": 0.4156,
+      "step": 5426
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3205064147147084e-06,
+      "loss": 0.411,
+      "step": 5427
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.317504836448141e-06,
+      "loss": 0.4158,
+      "step": 5428
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3145049462622003e-06,
+      "loss": 0.4273,
+      "step": 5429
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3115067448160533e-06,
+      "loss": 0.4175,
+      "step": 5430
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3085102327685014e-06,
+      "loss": 0.4399,
+      "step": 5431
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3055154107779663e-06,
+      "loss": 0.4321,
+      "step": 5432
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3025222795025082e-06,
+      "loss": 0.4138,
+      "step": 5433
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.299530839599805e-06,
+      "loss": 0.4304,
+      "step": 5434
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2965410917271715e-06,
+      "loss": 0.4006,
+      "step": 5435
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2935530365415436e-06,
+      "loss": 0.4135,
+      "step": 5436
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2905666746994893e-06,
+      "loss": 0.4301,
+      "step": 5437
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2875820068572084e-06,
+      "loss": 0.4122,
+      "step": 5438
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2845990336705192e-06,
+      "loss": 0.4172,
+      "step": 5439
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.281617755794878e-06,
+      "loss": 0.4268,
+      "step": 5440
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.278638173885361e-06,
+      "loss": 0.4278,
+      "step": 5441
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2756602885966704e-06,
+      "loss": 0.4382,
+      "step": 5442
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2726841005831447e-06,
+      "loss": 0.4518,
+      "step": 5443
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2697096104987406e-06,
+      "loss": 0.4233,
+      "step": 5444
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2667368189970473e-06,
+      "loss": 0.4143,
+      "step": 5445
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2637657267312785e-06,
+      "loss": 0.4287,
+      "step": 5446
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2607963343542703e-06,
+      "loss": 0.3996,
+      "step": 5447
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2578286425184957e-06,
+      "loss": 0.4385,
+      "step": 5448
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2548626518760455e-06,
+      "loss": 0.4296,
+      "step": 5449
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2518983630786395e-06,
+      "loss": 0.4186,
+      "step": 5450
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.248935776777621e-06,
+      "loss": 0.4034,
+      "step": 5451
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.245974893623961e-06,
+      "loss": 0.395,
+      "step": 5452
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2430157142682617e-06,
+      "loss": 0.4132,
+      "step": 5453
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2400582393607397e-06,
+      "loss": 0.4011,
+      "step": 5454
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2371024695512487e-06,
+      "loss": 0.4118,
+      "step": 5455
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.234148405489257e-06,
+      "loss": 0.4143,
+      "step": 5456
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2311960478238692e-06,
+      "loss": 0.4002,
+      "step": 5457
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2282453972038067e-06,
+      "loss": 0.4236,
+      "step": 5458
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.225296454277415e-06,
+      "loss": 0.4276,
+      "step": 5459
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.222349219692672e-06,
+      "loss": 0.4422,
+      "step": 5460
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2194036940971732e-06,
+      "loss": 0.3968,
+      "step": 5461
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.216459878138144e-06,
+      "loss": 0.4321,
+      "step": 5462
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.213517772462429e-06,
+      "loss": 0.427,
+      "step": 5463
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2105773777165006e-06,
+      "loss": 0.3947,
+      "step": 5464
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2076386945464513e-06,
+      "loss": 0.43,
+      "step": 5465
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.204701723598005e-06,
+      "loss": 0.4127,
+      "step": 5466
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2017664655165004e-06,
+      "loss": 0.4187,
+      "step": 5467
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.198832920946904e-06,
+      "loss": 0.4334,
+      "step": 5468
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1959010905338107e-06,
+      "loss": 0.3991,
+      "step": 5469
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.19297097492143e-06,
+      "loss": 0.4345,
+      "step": 5470
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1900425747535957e-06,
+      "loss": 0.4257,
+      "step": 5471
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1871158906737745e-06,
+      "loss": 0.4255,
+      "step": 5472
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1841909233250425e-06,
+      "loss": 0.4027,
+      "step": 5473
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1812676733501093e-06,
+      "loss": 0.4384,
+      "step": 5474
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1783461413913033e-06,
+      "loss": 0.426,
+      "step": 5475
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1754263280905683e-06,
+      "loss": 0.4214,
+      "step": 5476
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1725082340894865e-06,
+      "loss": 0.4335,
+      "step": 5477
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1695918600292443e-06,
+      "loss": 0.4183,
+      "step": 5478
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.166677206550666e-06,
+      "loss": 0.44,
+      "step": 5479
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.163764274294188e-06,
+      "loss": 0.4164,
+      "step": 5480
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1608530638998705e-06,
+      "loss": 0.4149,
+      "step": 5481
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1579435760073965e-06,
+      "loss": 0.4285,
+      "step": 5482
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1550358112560666e-06,
+      "loss": 0.4242,
+      "step": 5483
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.152129770284813e-06,
+      "loss": 0.4409,
+      "step": 5484
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1492254537321757e-06,
+      "loss": 0.4072,
+      "step": 5485
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1463228622363275e-06,
+      "loss": 0.4275,
+      "step": 5486
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1434219964350554e-06,
+      "loss": 0.421,
+      "step": 5487
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1405228569657666e-06,
+      "loss": 0.4265,
+      "step": 5488
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1376254444654955e-06,
+      "loss": 0.4224,
+      "step": 5489
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.134729759570888e-06,
+      "loss": 0.4145,
+      "step": 5490
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1318358029182207e-06,
+      "loss": 0.4335,
+      "step": 5491
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1289435751433784e-06,
+      "loss": 0.4135,
+      "step": 5492
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.126053076881879e-06,
+      "loss": 0.4131,
+      "step": 5493
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1231643087688514e-06,
+      "loss": 0.4352,
+      "step": 5494
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.120277271439046e-06,
+      "loss": 0.4383,
+      "step": 5495
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.117391965526835e-06,
+      "loss": 0.4413,
+      "step": 5496
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1145083916662057e-06,
+      "loss": 0.4108,
+      "step": 5497
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.111626550490773e-06,
+      "loss": 0.433,
+      "step": 5498
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1087464426337623e-06,
+      "loss": 0.3929,
+      "step": 5499
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.105868068728022e-06,
+      "loss": 0.4161,
+      "step": 5500
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.102991429406023e-06,
+      "loss": 0.421,
+      "step": 5501
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.100116525299847e-06,
+      "loss": 0.4181,
+      "step": 5502
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0972433570412023e-06,
+      "loss": 0.4242,
+      "step": 5503
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.094371925261409e-06,
+      "loss": 0.4419,
+      "step": 5504
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0915022305914145e-06,
+      "loss": 0.416,
+      "step": 5505
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0886342736617757e-06,
+      "loss": 0.4222,
+      "step": 5506
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0857680551026683e-06,
+      "loss": 0.4252,
+      "step": 5507
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.082903575543894e-06,
+      "loss": 0.4126,
+      "step": 5508
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0800408356148638e-06,
+      "loss": 0.4516,
+      "step": 5509
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0771798359446095e-06,
+      "loss": 0.4291,
+      "step": 5510
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0743205771617835e-06,
+      "loss": 0.4314,
+      "step": 5511
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.071463059894652e-06,
+      "loss": 0.3949,
+      "step": 5512
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.068607284771098e-06,
+      "loss": 0.4184,
+      "step": 5513
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0657532524186206e-06,
+      "loss": 0.398,
+      "step": 5514
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.062900963464346e-06,
+      "loss": 0.4169,
+      "step": 5515
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0600504185350024e-06,
+      "loss": 0.4145,
+      "step": 5516
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.057201618256949e-06,
+      "loss": 0.4209,
+      "step": 5517
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.054354563256151e-06,
+      "loss": 0.4015,
+      "step": 5518
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0515092541581915e-06,
+      "loss": 0.3954,
+      "step": 5519
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0486656915882785e-06,
+      "loss": 0.4223,
+      "step": 5520
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0458238761712245e-06,
+      "loss": 0.4105,
+      "step": 5521
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.042983808531469e-06,
+      "loss": 0.4268,
+      "step": 5522
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0401454892930606e-06,
+      "loss": 0.425,
+      "step": 5523
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.037308919079661e-06,
+      "loss": 0.411,
+      "step": 5524
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0344740985145594e-06,
+      "loss": 0.4525,
+      "step": 5525
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0316410282206478e-06,
+      "loss": 0.4178,
+      "step": 5526
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0288097088204416e-06,
+      "loss": 0.3891,
+      "step": 5527
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0259801409360647e-06,
+      "loss": 0.4222,
+      "step": 5528
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.023152325189266e-06,
+      "loss": 0.4201,
+      "step": 5529
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0203262622013996e-06,
+      "loss": 0.4111,
+      "step": 5530
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.017501952593438e-06,
+      "loss": 0.4209,
+      "step": 5531
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0146793969859736e-06,
+      "loss": 0.4127,
+      "step": 5532
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0118585959992033e-06,
+      "loss": 0.4266,
+      "step": 5533
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.009039550252948e-06,
+      "loss": 0.4275,
+      "step": 5534
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.006222260366638e-06,
+      "loss": 0.4085,
+      "step": 5535
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.003406726959315e-06,
+      "loss": 0.4367,
+      "step": 5536
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.0005929506496446e-06,
+      "loss": 0.4188,
+      "step": 5537
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9977809320558928e-06,
+      "loss": 0.4295,
+      "step": 5538
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.994970671795954e-06,
+      "loss": 0.4326,
+      "step": 5539
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.992162170487324e-06,
+      "loss": 0.4049,
+      "step": 5540
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9893554287471195e-06,
+      "loss": 0.4166,
+      "step": 5541
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.986550447192066e-06,
+      "loss": 0.4111,
+      "step": 5542
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9837472264385028e-06,
+      "loss": 0.422,
+      "step": 5543
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.980945767102389e-06,
+      "loss": 0.4016,
+      "step": 5544
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.978146069799286e-06,
+      "loss": 0.4173,
+      "step": 5545
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9753481351443782e-06,
+      "loss": 0.4166,
+      "step": 5546
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.972551963752456e-06,
+      "loss": 0.4193,
+      "step": 5547
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9697575562379213e-06,
+      "loss": 0.3835,
+      "step": 5548
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.966964913214796e-06,
+      "loss": 0.4412,
+      "step": 5549
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9641740352967064e-06,
+      "loss": 0.398,
+      "step": 5550
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9613849230968985e-06,
+      "loss": 0.4388,
+      "step": 5551
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.958597577228223e-06,
+      "loss": 0.4032,
+      "step": 5552
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.955811998303142e-06,
+      "loss": 0.4049,
+      "step": 5553
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.953028186933741e-06,
+      "loss": 0.4161,
+      "step": 5554
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9502461437317045e-06,
+      "loss": 0.411,
+      "step": 5555
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9474658693083303e-06,
+      "loss": 0.4315,
+      "step": 5556
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9446873642745366e-06,
+      "loss": 0.4206,
+      "step": 5557
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.941910629240843e-06,
+      "loss": 0.4318,
+      "step": 5558
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9391356648173842e-06,
+      "loss": 0.4163,
+      "step": 5559
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9363624716139017e-06,
+      "loss": 0.4074,
+      "step": 5560
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.933591050239758e-06,
+      "loss": 0.4328,
+      "step": 5561
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9308214013039127e-06,
+      "loss": 0.393,
+      "step": 5562
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.92805352541495e-06,
+      "loss": 0.3966,
+      "step": 5563
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9252874231810535e-06,
+      "loss": 0.4337,
+      "step": 5564
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9225230952100195e-06,
+      "loss": 0.4347,
+      "step": 5565
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.91976054210926e-06,
+      "loss": 0.4448,
+      "step": 5566
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9169997644857874e-06,
+      "loss": 0.4616,
+      "step": 5567
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9142407629462367e-06,
+      "loss": 0.4179,
+      "step": 5568
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9114835380968388e-06,
+      "loss": 0.4236,
+      "step": 5569
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.908728090543446e-06,
+      "loss": 0.4048,
+      "step": 5570
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9059744208915132e-06,
+      "loss": 0.4332,
+      "step": 5571
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9032225297461049e-06,
+      "loss": 0.3918,
+      "step": 5572
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9004724177118982e-06,
+      "loss": 0.4132,
+      "step": 5573
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8977240853931733e-06,
+      "loss": 0.4231,
+      "step": 5574
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8949775333938292e-06,
+      "loss": 0.4267,
+      "step": 5575
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8922327623173654e-06,
+      "loss": 0.4161,
+      "step": 5576
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8894897727668893e-06,
+      "loss": 0.4288,
+      "step": 5577
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8867485653451257e-06,
+      "loss": 0.4284,
+      "step": 5578
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.884009140654397e-06,
+      "loss": 0.4129,
+      "step": 5579
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8812714992966453e-06,
+      "loss": 0.4042,
+      "step": 5580
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.878535641873407e-06,
+      "loss": 0.4259,
+      "step": 5581
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8758015689858422e-06,
+      "loss": 0.4294,
+      "step": 5582
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8730692812347062e-06,
+      "loss": 0.4325,
+      "step": 5583
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8703387792203652e-06,
+      "loss": 0.4293,
+      "step": 5584
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8676100635428006e-06,
+      "loss": 0.4102,
+      "step": 5585
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8648831348015906e-06,
+      "loss": 0.424,
+      "step": 5586
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8621579935959255e-06,
+      "loss": 0.435,
+      "step": 5587
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8594346405246044e-06,
+      "loss": 0.4148,
+      "step": 5588
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8567130761860274e-06,
+      "loss": 0.4126,
+      "step": 5589
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8539933011782129e-06,
+      "loss": 0.4187,
+      "step": 5590
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8512753160987717e-06,
+      "loss": 0.4242,
+      "step": 5591
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8485591215449361e-06,
+      "loss": 0.4166,
+      "step": 5592
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.845844718113532e-06,
+      "loss": 0.4163,
+      "step": 5593
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.843132106401002e-06,
+      "loss": 0.4146,
+      "step": 5594
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8404212870033877e-06,
+      "loss": 0.4548,
+      "step": 5595
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8377122605163377e-06,
+      "loss": 0.4153,
+      "step": 5596
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8350050275351128e-06,
+      "loss": 0.4113,
+      "step": 5597
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8322995886545703e-06,
+      "loss": 0.4099,
+      "step": 5598
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8295959444691846e-06,
+      "loss": 0.3945,
+      "step": 5599
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8268940955730264e-06,
+      "loss": 0.3786,
+      "step": 5600
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8241940425597715e-06,
+      "loss": 0.3973,
+      "step": 5601
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.821495786022711e-06,
+      "loss": 0.4208,
+      "step": 5602
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8187993265547322e-06,
+      "loss": 0.4013,
+      "step": 5603
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8161046647483306e-06,
+      "loss": 0.4214,
+      "step": 5604
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8134118011956015e-06,
+      "loss": 0.4392,
+      "step": 5605
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8107207364882585e-06,
+      "loss": 0.4121,
+      "step": 5606
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8080314712176051e-06,
+      "loss": 0.409,
+      "step": 5607
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8053440059745565e-06,
+      "loss": 0.4165,
+      "step": 5608
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8026583413496345e-06,
+      "loss": 0.4093,
+      "step": 5609
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7999744779329576e-06,
+      "loss": 0.421,
+      "step": 5610
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7972924163142591e-06,
+      "loss": 0.4054,
+      "step": 5611
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7946121570828668e-06,
+      "loss": 0.4115,
+      "step": 5612
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7919337008277138e-06,
+      "loss": 0.4303,
+      "step": 5613
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.789257048137345e-06,
+      "loss": 0.4316,
+      "step": 5614
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7865821995998978e-06,
+      "loss": 0.4351,
+      "step": 5615
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.783909155803124e-06,
+      "loss": 0.4385,
+      "step": 5616
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.781237917334372e-06,
+      "loss": 0.4256,
+      "step": 5617
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.778568484780594e-06,
+      "loss": 0.415,
+      "step": 5618
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7759008587283456e-06,
+      "loss": 0.4042,
+      "step": 5619
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7732350397637855e-06,
+      "loss": 0.4247,
+      "step": 5620
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7705710284726808e-06,
+      "loss": 0.4234,
+      "step": 5621
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7679088254403919e-06,
+      "loss": 0.4241,
+      "step": 5622
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7652484312518914e-06,
+      "loss": 0.42,
+      "step": 5623
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7625898464917468e-06,
+      "loss": 0.4075,
+      "step": 5624
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7599330717441287e-06,
+      "loss": 0.4119,
+      "step": 5625
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.757278107592818e-06,
+      "loss": 0.4225,
+      "step": 5626
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7546249546211869e-06,
+      "loss": 0.3956,
+      "step": 5627
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.751973613412219e-06,
+      "loss": 0.4193,
+      "step": 5628
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7493240845484937e-06,
+      "loss": 0.4225,
+      "step": 5629
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7466763686121913e-06,
+      "loss": 0.383,
+      "step": 5630
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7440304661851027e-06,
+      "loss": 0.4002,
+      "step": 5631
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7413863778486084e-06,
+      "loss": 0.4226,
+      "step": 5632
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7387441041836995e-06,
+      "loss": 0.42,
+      "step": 5633
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.73610364577096e-06,
+      "loss": 0.4313,
+      "step": 5634
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7334650031905864e-06,
+      "loss": 0.4059,
+      "step": 5635
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7308281770223668e-06,
+      "loss": 0.4175,
+      "step": 5636
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7281931678456899e-06,
+      "loss": 0.4113,
+      "step": 5637
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7255599762395535e-06,
+      "loss": 0.4055,
+      "step": 5638
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7229286027825454e-06,
+      "loss": 0.3934,
+      "step": 5639
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7202990480528659e-06,
+      "loss": 0.432,
+      "step": 5640
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7176713126283062e-06,
+      "loss": 0.4117,
+      "step": 5641
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7150453970862569e-06,
+      "loss": 0.4258,
+      "step": 5642
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7124213020037173e-06,
+      "loss": 0.4047,
+      "step": 5643
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7097990279572797e-06,
+      "loss": 0.4423,
+      "step": 5644
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7071785755231397e-06,
+      "loss": 0.4076,
+      "step": 5645
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7045599452770889e-06,
+      "loss": 0.4288,
+      "step": 5646
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.701943137794525e-06,
+      "loss": 0.4024,
+      "step": 5647
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.699328153650439e-06,
+      "loss": 0.4285,
+      "step": 5648
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6967149934194226e-06,
+      "loss": 0.4195,
+      "step": 5649
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6941036576756687e-06,
+      "loss": 0.4199,
+      "step": 5650
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6914941469929646e-06,
+      "loss": 0.4201,
+      "step": 5651
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6888864619447066e-06,
+      "loss": 0.4158,
+      "step": 5652
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6862806031038781e-06,
+      "loss": 0.4406,
+      "step": 5653
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6836765710430669e-06,
+      "loss": 0.4264,
+      "step": 5654
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.681074366334462e-06,
+      "loss": 0.4394,
+      "step": 5655
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6784739895498448e-06,
+      "loss": 0.422,
+      "step": 5656
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6758754412606016e-06,
+      "loss": 0.4152,
+      "step": 5657
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6732787220377078e-06,
+      "loss": 0.3992,
+      "step": 5658
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6706838324517493e-06,
+      "loss": 0.4456,
+      "step": 5659
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6680907730729002e-06,
+      "loss": 0.42,
+      "step": 5660
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6654995444709321e-06,
+      "loss": 0.414,
+      "step": 5661
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.662910147215223e-06,
+      "loss": 0.4289,
+      "step": 5662
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6603225818747416e-06,
+      "loss": 0.4163,
+      "step": 5663
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6577368490180545e-06,
+      "loss": 0.4307,
+      "step": 5664
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6551529492133277e-06,
+      "loss": 0.4151,
+      "step": 5665
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6525708830283206e-06,
+      "loss": 0.3865,
+      "step": 5666
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6499906510303976e-06,
+      "loss": 0.4099,
+      "step": 5667
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.64741225378651e-06,
+      "loss": 0.4065,
+      "step": 5668
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6448356918632158e-06,
+      "loss": 0.4083,
+      "step": 5669
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6422609658266598e-06,
+      "loss": 0.4051,
+      "step": 5670
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6396880762425948e-06,
+      "loss": 0.3982,
+      "step": 5671
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.63711702367636e-06,
+      "loss": 0.4341,
+      "step": 5672
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6345478086928935e-06,
+      "loss": 0.4161,
+      "step": 5673
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6319804318567345e-06,
+      "loss": 0.4268,
+      "step": 5674
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6294148937320098e-06,
+      "loss": 0.4101,
+      "step": 5675
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.626851194882454e-06,
+      "loss": 0.4475,
+      "step": 5676
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6242893358713862e-06,
+      "loss": 0.3917,
+      "step": 5677
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6217293172617254e-06,
+      "loss": 0.4331,
+      "step": 5678
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.619171139615986e-06,
+      "loss": 0.4094,
+      "step": 5679
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6166148034962825e-06,
+      "loss": 0.4122,
+      "step": 5680
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.614060309464317e-06,
+      "loss": 0.4314,
+      "step": 5681
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6115076580813893e-06,
+      "loss": 0.3991,
+      "step": 5682
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6089568499083997e-06,
+      "loss": 0.4278,
+      "step": 5683
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6064078855058374e-06,
+      "loss": 0.4219,
+      "step": 5684
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6038607654337845e-06,
+      "loss": 0.3966,
+      "step": 5685
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.601315490251929e-06,
+      "loss": 0.4138,
+      "step": 5686
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.598772060519539e-06,
+      "loss": 0.3941,
+      "step": 5687
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.5962304767954906e-06,
+      "loss": 0.42,
+      "step": 5688
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.593690739638244e-06,
+      "loss": 0.4318,
+      "step": 5689
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5911528496058571e-06,
+      "loss": 0.4111,
+      "step": 5690
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.588616807255986e-06,
+      "loss": 0.4127,
+      "step": 5691
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.586082613145873e-06,
+      "loss": 0.4258,
+      "step": 5692
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5835502678323622e-06,
+      "loss": 0.4226,
+      "step": 5693
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5810197718718879e-06,
+      "loss": 0.4033,
+      "step": 5694
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.578491125820475e-06,
+      "loss": 0.4186,
+      "step": 5695
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5759643302337468e-06,
+      "loss": 0.4185,
+      "step": 5696
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.573439385666915e-06,
+      "loss": 0.45,
+      "step": 5697
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5709162926747933e-06,
+      "loss": 0.4054,
+      "step": 5698
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5683950518117763e-06,
+      "loss": 0.4079,
+      "step": 5699
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.565875663631864e-06,
+      "loss": 0.3936,
+      "step": 5700
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5633581286886413e-06,
+      "loss": 0.4266,
+      "step": 5701
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5608424475352868e-06,
+      "loss": 0.4141,
+      "step": 5702
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.558328620724575e-06,
+      "loss": 0.4135,
+      "step": 5703
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5558166488088689e-06,
+      "loss": 0.4389,
+      "step": 5704
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5533065323401286e-06,
+      "loss": 0.419,
+      "step": 5705
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5507982718699012e-06,
+      "loss": 0.4174,
+      "step": 5706
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.548291867949333e-06,
+      "loss": 0.4574,
+      "step": 5707
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.545787321129154e-06,
+      "loss": 0.4101,
+      "step": 5708
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5432846319596927e-06,
+      "loss": 0.4308,
+      "step": 5709
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5407838009908649e-06,
+      "loss": 0.4128,
+      "step": 5710
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5382848287721796e-06,
+      "loss": 0.3989,
+      "step": 5711
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5357877158527413e-06,
+      "loss": 0.4204,
+      "step": 5712
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5332924627812407e-06,
+      "loss": 0.4033,
+      "step": 5713
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5307990701059584e-06,
+      "loss": 0.4061,
+      "step": 5714
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.528307538374776e-06,
+      "loss": 0.4101,
+      "step": 5715
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5258178681351533e-06,
+      "loss": 0.4249,
+      "step": 5716
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5233300599341539e-06,
+      "loss": 0.4257,
+      "step": 5717
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5208441143184215e-06,
+      "loss": 0.4106,
+      "step": 5718
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.518360031834194e-06,
+      "loss": 0.415,
+      "step": 5719
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5158778130273044e-06,
+      "loss": 0.4007,
+      "step": 5720
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5133974584431688e-06,
+      "loss": 0.3887,
+      "step": 5721
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5109189686268023e-06,
+      "loss": 0.4267,
+      "step": 5722
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5084423441228013e-06,
+      "loss": 0.4099,
+      "step": 5723
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5059675854753563e-06,
+      "loss": 0.4449,
+      "step": 5724
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5034946932282512e-06,
+      "loss": 0.4149,
+      "step": 5725
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5010236679248546e-06,
+      "loss": 0.4189,
+      "step": 5726
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4985545101081266e-06,
+      "loss": 0.431,
+      "step": 5727
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4960872203206155e-06,
+      "loss": 0.4139,
+      "step": 5728
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4936217991044654e-06,
+      "loss": 0.4197,
+      "step": 5729
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.491158247001402e-06,
+      "loss": 0.4209,
+      "step": 5730
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4886965645527418e-06,
+      "loss": 0.4198,
+      "step": 5731
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4862367522993958e-06,
+      "loss": 0.409,
+      "step": 5732
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4837788107818585e-06,
+      "loss": 0.4338,
+      "step": 5733
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4813227405402165e-06,
+      "loss": 0.391,
+      "step": 5734
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4788685421141425e-06,
+      "loss": 0.4003,
+      "step": 5735
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.476416216042902e-06,
+      "loss": 0.4607,
+      "step": 5736
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4739657628653448e-06,
+      "loss": 0.4082,
+      "step": 5737
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4715171831199083e-06,
+      "loss": 0.4132,
+      "step": 5738
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4690704773446264e-06,
+      "loss": 0.4321,
+      "step": 5739
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4666256460771123e-06,
+      "loss": 0.4234,
+      "step": 5740
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4641826898545708e-06,
+      "loss": 0.4195,
+      "step": 5741
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.461741609213795e-06,
+      "loss": 0.3922,
+      "step": 5742
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.459302404691163e-06,
+      "loss": 0.411,
+      "step": 5743
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4568650768226478e-06,
+      "loss": 0.417,
+      "step": 5744
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4544296261438007e-06,
+      "loss": 0.4044,
+      "step": 5745
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4519960531897693e-06,
+      "loss": 0.4121,
+      "step": 5746
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.44956435849528e-06,
+      "loss": 0.4071,
+      "step": 5747
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.447134542594657e-06,
+      "loss": 0.41,
+      "step": 5748
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4447066060218017e-06,
+      "loss": 0.4158,
+      "step": 5749
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4422805493102044e-06,
+      "loss": 0.4045,
+      "step": 5750
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4398563729929493e-06,
+      "loss": 0.4229,
+      "step": 5751
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4374340776026985e-06,
+      "loss": 0.4104,
+      "step": 5752
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4350136636717083e-06,
+      "loss": 0.4106,
+      "step": 5753
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4325951317318164e-06,
+      "loss": 0.4122,
+      "step": 5754
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4301784823144482e-06,
+      "loss": 0.4007,
+      "step": 5755
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4277637159506176e-06,
+      "loss": 0.3942,
+      "step": 5756
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4253508331709187e-06,
+      "loss": 0.3745,
+      "step": 5757
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4229398345055423e-06,
+      "loss": 0.4067,
+      "step": 5758
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4205307204842532e-06,
+      "loss": 0.4316,
+      "step": 5759
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.418123491636414e-06,
+      "loss": 0.4067,
+      "step": 5760
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4157181484909632e-06,
+      "loss": 0.4247,
+      "step": 5761
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4133146915764272e-06,
+      "loss": 0.4118,
+      "step": 5762
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4109131214209226e-06,
+      "loss": 0.4147,
+      "step": 5763
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.408513438552146e-06,
+      "loss": 0.4138,
+      "step": 5764
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4061156434973854e-06,
+      "loss": 0.3945,
+      "step": 5765
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4037197367835066e-06,
+      "loss": 0.4126,
+      "step": 5766
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4013257189369634e-06,
+      "loss": 0.4277,
+      "step": 5767
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3989335904837987e-06,
+      "loss": 0.4173,
+      "step": 5768
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3965433519496318e-06,
+      "loss": 0.4118,
+      "step": 5769
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3941550038596773e-06,
+      "loss": 0.4219,
+      "step": 5770
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3917685467387264e-06,
+      "loss": 0.4222,
+      "step": 5771
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3893839811111576e-06,
+      "loss": 0.4276,
+      "step": 5772
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3870013075009325e-06,
+      "loss": 0.4285,
+      "step": 5773
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3846205264315948e-06,
+      "loss": 0.4334,
+      "step": 5774
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3822416384262815e-06,
+      "loss": 0.4362,
+      "step": 5775
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3798646440077025e-06,
+      "loss": 0.4141,
+      "step": 5776
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3774895436981606e-06,
+      "loss": 0.4241,
+      "step": 5777
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3751163380195365e-06,
+      "loss": 0.4193,
+      "step": 5778
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3727450274932942e-06,
+      "loss": 0.4294,
+      "step": 5779
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3703756126404887e-06,
+      "loss": 0.4171,
+      "step": 5780
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3680080939817486e-06,
+      "loss": 0.4095,
+      "step": 5781
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.365642472037294e-06,
+      "loss": 0.386,
+      "step": 5782
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3632787473269215e-06,
+      "loss": 0.4017,
+      "step": 5783
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.360916920370019e-06,
+      "loss": 0.4175,
+      "step": 5784
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.358556991685549e-06,
+      "loss": 0.4106,
+      "step": 5785
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3561989617920624e-06,
+      "loss": 0.4213,
+      "step": 5786
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.353842831207689e-06,
+      "loss": 0.4138,
+      "step": 5787
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3514886004501415e-06,
+      "loss": 0.4111,
+      "step": 5788
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3491362700367216e-06,
+      "loss": 0.4132,
+      "step": 5789
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.346785840484307e-06,
+      "loss": 0.4255,
+      "step": 5790
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3444373123093557e-06,
+      "loss": 0.4116,
+      "step": 5791
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3420906860279181e-06,
+      "loss": 0.4101,
+      "step": 5792
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.4325,
+      "step": 5793
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.337403141207656e-06,
+      "loss": 0.4172,
+      "step": 5794
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.335062223698832e-06,
+      "loss": 0.4104,
+      "step": 5795
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3327232101435129e-06,
+      "loss": 0.4294,
+      "step": 5796
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.330386101055654e-06,
+      "loss": 0.4027,
+      "step": 5797
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.328050896948787e-06,
+      "loss": 0.4209,
+      "step": 5798
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3257175983360325e-06,
+      "loss": 0.4113,
+      "step": 5799
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3233862057300862e-06,
+      "loss": 0.413,
+      "step": 5800
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3210567196432256e-06,
+      "loss": 0.4184,
+      "step": 5801
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3187291405873104e-06,
+      "loss": 0.4222,
+      "step": 5802
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3164034690737804e-06,
+      "loss": 0.4033,
+      "step": 5803
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3140797056136612e-06,
+      "loss": 0.4316,
+      "step": 5804
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3117578507175511e-06,
+      "loss": 0.3941,
+      "step": 5805
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3094379048956363e-06,
+      "loss": 0.4149,
+      "step": 5806
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3071198686576792e-06,
+      "loss": 0.4082,
+      "step": 5807
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3048037425130212e-06,
+      "loss": 0.4178,
+      "step": 5808
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3024895269705906e-06,
+      "loss": 0.4307,
+      "step": 5809
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3001772225388864e-06,
+      "loss": 0.403,
+      "step": 5810
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.297866829726e-06,
+      "loss": 0.4393,
+      "step": 5811
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2955583490395896e-06,
+      "loss": 0.4201,
+      "step": 5812
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.293251780986905e-06,
+      "loss": 0.4215,
+      "step": 5813
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2909471260747664e-06,
+      "loss": 0.4316,
+      "step": 5814
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2886443848095764e-06,
+      "loss": 0.4115,
+      "step": 5815
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2863435576973216e-06,
+      "loss": 0.4167,
+      "step": 5816
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.284044645243564e-06,
+      "loss": 0.4139,
+      "step": 5817
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.281747647953444e-06,
+      "loss": 0.4158,
+      "step": 5818
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2794525663316825e-06,
+      "loss": 0.4141,
+      "step": 5819
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2771594008825782e-06,
+      "loss": 0.4419,
+      "step": 5820
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2748681521100136e-06,
+      "loss": 0.4247,
+      "step": 5821
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2725788205174439e-06,
+      "loss": 0.4028,
+      "step": 5822
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2702914066079076e-06,
+      "loss": 0.4264,
+      "step": 5823
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2680059108840182e-06,
+      "loss": 0.4269,
+      "step": 5824
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2657223338479706e-06,
+      "loss": 0.3989,
+      "step": 5825
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2634406760015382e-06,
+      "loss": 0.3948,
+      "step": 5826
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2611609378460677e-06,
+      "loss": 0.4113,
+      "step": 5827
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2588831198824914e-06,
+      "loss": 0.406,
+      "step": 5828
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2566072226113125e-06,
+      "loss": 0.4228,
+      "step": 5829
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2543332465326208e-06,
+      "loss": 0.4151,
+      "step": 5830
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2520611921460745e-06,
+      "loss": 0.4298,
+      "step": 5831
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2497910599509156e-06,
+      "loss": 0.4134,
+      "step": 5832
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2475228504459624e-06,
+      "loss": 0.4232,
+      "step": 5833
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2452565641296066e-06,
+      "loss": 0.4217,
+      "step": 5834
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2429922014998252e-06,
+      "loss": 0.428,
+      "step": 5835
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.240729763054166e-06,
+      "loss": 0.4529,
+      "step": 5836
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2384692492897598e-06,
+      "loss": 0.4092,
+      "step": 5837
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2362106607033087e-06,
+      "loss": 0.4053,
+      "step": 5838
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.233953997791091e-06,
+      "loss": 0.4025,
+      "step": 5839
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.231699261048971e-06,
+      "loss": 0.404,
+      "step": 5840
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2294464509723792e-06,
+      "loss": 0.421,
+      "step": 5841
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2271955680563297e-06,
+      "loss": 0.4166,
+      "step": 5842
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2249466127954112e-06,
+      "loss": 0.412,
+      "step": 5843
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2226995856837841e-06,
+      "loss": 0.4161,
+      "step": 5844
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2204544872151957e-06,
+      "loss": 0.4284,
+      "step": 5845
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2182113178829602e-06,
+      "loss": 0.4221,
+      "step": 5846
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.21597007817997e-06,
+      "loss": 0.4282,
+      "step": 5847
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2137307685986932e-06,
+      "loss": 0.4457,
+      "step": 5848
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2114933896311787e-06,
+      "loss": 0.4316,
+      "step": 5849
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2092579417690464e-06,
+      "loss": 0.3892,
+      "step": 5850
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2070244255034912e-06,
+      "loss": 0.398,
+      "step": 5851
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2047928413252874e-06,
+      "loss": 0.4122,
+      "step": 5852
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2025631897247814e-06,
+      "loss": 0.4132,
+      "step": 5853
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.200335471191898e-06,
+      "loss": 0.4337,
+      "step": 5854
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1981096862161334e-06,
+      "loss": 0.3966,
+      "step": 5855
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.195885835286561e-06,
+      "loss": 0.4093,
+      "step": 5856
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.193663918891833e-06,
+      "loss": 0.4473,
+      "step": 5857
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.191443937520167e-06,
+      "loss": 0.3991,
+      "step": 5858
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1892258916593669e-06,
+      "loss": 0.3921,
+      "step": 5859
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1870097817968008e-06,
+      "loss": 0.4027,
+      "step": 5860
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1847956084194213e-06,
+      "loss": 0.4176,
+      "step": 5861
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1825833720137469e-06,
+      "loss": 0.4203,
+      "step": 5862
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1803730730658746e-06,
+      "loss": 0.4033,
+      "step": 5863
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1781647120614759e-06,
+      "loss": 0.4042,
+      "step": 5864
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1759582894857923e-06,
+      "loss": 0.4076,
+      "step": 5865
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1737538058236486e-06,
+      "loss": 0.4291,
+      "step": 5866
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1715512615594338e-06,
+      "loss": 0.3943,
+      "step": 5867
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.169350657177114e-06,
+      "loss": 0.4101,
+      "step": 5868
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1671519931602337e-06,
+      "loss": 0.4057,
+      "step": 5869
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.164955269991902e-06,
+      "loss": 0.418,
+      "step": 5870
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1627604881548105e-06,
+      "loss": 0.4535,
+      "step": 5871
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1605676481312178e-06,
+      "loss": 0.4204,
+      "step": 5872
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1583767504029608e-06,
+      "loss": 0.4024,
+      "step": 5873
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1561877954514467e-06,
+      "loss": 0.4101,
+      "step": 5874
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1540007837576528e-06,
+      "loss": 0.4164,
+      "step": 5875
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1518157158021382e-06,
+      "loss": 0.4368,
+      "step": 5876
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1496325920650263e-06,
+      "loss": 0.4309,
+      "step": 5877
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1474514130260173e-06,
+      "loss": 0.4006,
+      "step": 5878
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.145272179164384e-06,
+      "loss": 0.4132,
+      "step": 5879
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1430948909589667e-06,
+      "loss": 0.4288,
+      "step": 5880
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1409195488881898e-06,
+      "loss": 0.4175,
+      "step": 5881
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.138746153430037e-06,
+      "loss": 0.4079,
+      "step": 5882
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1365747050620745e-06,
+      "loss": 0.4255,
+      "step": 5883
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1344052042614339e-06,
+      "loss": 0.4341,
+      "step": 5884
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1322376515048207e-06,
+      "loss": 0.409,
+      "step": 5885
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1300720472685155e-06,
+      "loss": 0.4349,
+      "step": 5886
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1279083920283652e-06,
+      "loss": 0.4129,
+      "step": 5887
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1257466862597943e-06,
+      "loss": 0.4284,
+      "step": 5888
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1235869304377934e-06,
+      "loss": 0.4159,
+      "step": 5889
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1214291250369303e-06,
+      "loss": 0.4331,
+      "step": 5890
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1192732705313402e-06,
+      "loss": 0.3988,
+      "step": 5891
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.11711936739473e-06,
+      "loss": 0.4243,
+      "step": 5892
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1149674161003765e-06,
+      "loss": 0.434,
+      "step": 5893
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1128174171211337e-06,
+      "loss": 0.4274,
+      "step": 5894
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1106693709294203e-06,
+      "loss": 0.4387,
+      "step": 5895
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1085232779972287e-06,
+      "loss": 0.4187,
+      "step": 5896
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1063791387961197e-06,
+      "loss": 0.4219,
+      "step": 5897
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.1042369537972309e-06,
+      "loss": 0.3988,
+      "step": 5898
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.102096723471261e-06,
+      "loss": 0.4092,
+      "step": 5899
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0999584482884896e-06,
+      "loss": 0.4349,
+      "step": 5900
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0978221287187563e-06,
+      "loss": 0.4154,
+      "step": 5901
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0956877652314824e-06,
+      "loss": 0.4047,
+      "step": 5902
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0935553582956482e-06,
+      "loss": 0.3824,
+      "step": 5903
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.09142490837981e-06,
+      "loss": 0.4213,
+      "step": 5904
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.089296415952097e-06,
+      "loss": 0.4271,
+      "step": 5905
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0871698814801979e-06,
+      "loss": 0.4083,
+      "step": 5906
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0850453054313847e-06,
+      "loss": 0.4131,
+      "step": 5907
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0829226882724875e-06,
+      "loss": 0.419,
+      "step": 5908
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.080802030469913e-06,
+      "loss": 0.435,
+      "step": 5909
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.078683332489634e-06,
+      "loss": 0.4247,
+      "step": 5910
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0765665947971916e-06,
+      "loss": 0.4047,
+      "step": 5911
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0744518178577025e-06,
+      "loss": 0.4144,
+      "step": 5912
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0723390021358437e-06,
+      "loss": 0.4206,
+      "step": 5913
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.07022814809587e-06,
+      "loss": 0.4199,
+      "step": 5914
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0681192562015995e-06,
+      "loss": 0.3886,
+      "step": 5915
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.066012326916419e-06,
+      "loss": 0.4308,
+      "step": 5916
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.063907360703288e-06,
+      "loss": 0.4074,
+      "step": 5917
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0618043580247294e-06,
+      "loss": 0.4139,
+      "step": 5918
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.059703319342843e-06,
+      "loss": 0.4075,
+      "step": 5919
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.057604245119287e-06,
+      "loss": 0.4162,
+      "step": 5920
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0555071358152923e-06,
+      "loss": 0.4277,
+      "step": 5921
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.053411991891662e-06,
+      "loss": 0.3854,
+      "step": 5922
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0513188138087616e-06,
+      "loss": 0.413,
+      "step": 5923
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0492276020265268e-06,
+      "loss": 0.4352,
+      "step": 5924
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.047138357004459e-06,
+      "loss": 0.4344,
+      "step": 5925
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0450510792016343e-06,
+      "loss": 0.4201,
+      "step": 5926
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0429657690766891e-06,
+      "loss": 0.418,
+      "step": 5927
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0408824270878282e-06,
+      "loss": 0.4343,
+      "step": 5928
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0388010536928306e-06,
+      "loss": 0.3991,
+      "step": 5929
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.036721649349034e-06,
+      "loss": 0.4227,
+      "step": 5930
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0346442145133506e-06,
+      "loss": 0.4284,
+      "step": 5931
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0325687496422554e-06,
+      "loss": 0.4316,
+      "step": 5932
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0304952551917913e-06,
+      "loss": 0.3901,
+      "step": 5933
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0284237316175704e-06,
+      "loss": 0.4219,
+      "step": 5934
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0263541793747678e-06,
+      "loss": 0.3971,
+      "step": 5935
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0242865989181305e-06,
+      "loss": 0.4234,
+      "step": 5936
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0222209907019687e-06,
+      "loss": 0.4032,
+      "step": 5937
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.020157355180158e-06,
+      "loss": 0.4167,
+      "step": 5938
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.018095692806147e-06,
+      "loss": 0.4267,
+      "step": 5939
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.016036004032943e-06,
+      "loss": 0.4116,
+      "step": 5940
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0139782893131234e-06,
+      "loss": 0.3879,
+      "step": 5941
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0119225490988305e-06,
+      "loss": 0.4043,
+      "step": 5942
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0098687838417765e-06,
+      "loss": 0.4123,
+      "step": 5943
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.007816993993236e-06,
+      "loss": 0.4223,
+      "step": 5944
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.005767180004048e-06,
+      "loss": 0.4064,
+      "step": 5945
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0037193423246227e-06,
+      "loss": 0.4125,
+      "step": 5946
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.00167348140493e-06,
+      "loss": 0.4267,
+      "step": 5947
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.996295976945124e-07,
+      "loss": 0.4081,
+      "step": 5948
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.97587691642471e-07,
+      "loss": 0.4239,
+      "step": 5949
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.955477636974763e-07,
+      "loss": 0.421,
+      "step": 5950
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.935098143077649e-07,
+      "loss": 0.4048,
+      "step": 5951
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.914738439211335e-07,
+      "loss": 0.4251,
+      "step": 5952
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.894398529849514e-07,
+      "loss": 0.4096,
+      "step": 5953
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.874078419461464e-07,
+      "loss": 0.4124,
+      "step": 5954
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.853778112512147e-07,
+      "loss": 0.4308,
+      "step": 5955
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.833497613462162e-07,
+      "loss": 0.4206,
+      "step": 5956
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.813236926767734e-07,
+      "loss": 0.4368,
+      "step": 5957
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.792996056880811e-07,
+      "loss": 0.4094,
+      "step": 5958
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.772775008248892e-07,
+      "loss": 0.3925,
+      "step": 5959
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.7525737853152e-07,
+      "loss": 0.4454,
+      "step": 5960
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.73239239251853e-07,
+      "loss": 0.4098,
+      "step": 5961
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.712230834293378e-07,
+      "loss": 0.3977,
+      "step": 5962
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.69208911506987e-07,
+      "loss": 0.3872,
+      "step": 5963
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.671967239273716e-07,
+      "loss": 0.42,
+      "step": 5964
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.65186521132635e-07,
+      "loss": 0.4138,
+      "step": 5965
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.631783035644792e-07,
+      "loss": 0.3971,
+      "step": 5966
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.611720716641725e-07,
+      "loss": 0.406,
+      "step": 5967
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.591678258725457e-07,
+      "loss": 0.4049,
+      "step": 5968
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.571655666299917e-07,
+      "loss": 0.4039,
+      "step": 5969
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.55165294376471e-07,
+      "loss": 0.404,
+      "step": 5970
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.531670095515e-07,
+      "loss": 0.3894,
+      "step": 5971
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.511707125941694e-07,
+      "loss": 0.4191,
+      "step": 5972
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.49176403943124e-07,
+      "loss": 0.417,
+      "step": 5973
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.471840840365742e-07,
+      "loss": 0.3917,
+      "step": 5974
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.451937533122968e-07,
+      "loss": 0.4022,
+      "step": 5975
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.432054122076239e-07,
+      "loss": 0.419,
+      "step": 5976
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.412190611594607e-07,
+      "loss": 0.4081,
+      "step": 5977
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.392347006042657e-07,
+      "loss": 0.401,
+      "step": 5978
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.372523309780679e-07,
+      "loss": 0.4157,
+      "step": 5979
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.352719527164511e-07,
+      "loss": 0.4173,
+      "step": 5980
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.332935662545661e-07,
+      "loss": 0.3927,
+      "step": 5981
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.313171720271274e-07,
+      "loss": 0.4224,
+      "step": 5982
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.293427704684065e-07,
+      "loss": 0.4082,
+      "step": 5983
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.273703620122442e-07,
+      "loss": 0.4011,
+      "step": 5984
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.25399947092036e-07,
+      "loss": 0.4113,
+      "step": 5985
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.234315261407445e-07,
+      "loss": 0.4568,
+      "step": 5986
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.214650995908925e-07,
+      "loss": 0.4216,
+      "step": 5987
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.1950066787456e-07,
+      "loss": 0.4141,
+      "step": 5988
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.175382314233994e-07,
+      "loss": 0.4321,
+      "step": 5989
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.155777906686136e-07,
+      "loss": 0.4139,
+      "step": 5990
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.136193460409759e-07,
+      "loss": 0.4298,
+      "step": 5991
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.116628979708142e-07,
+      "loss": 0.4234,
+      "step": 5992
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.097084468880202e-07,
+      "loss": 0.4291,
+      "step": 5993
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.077559932220481e-07,
+      "loss": 0.4071,
+      "step": 5994
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.058055374019104e-07,
+      "loss": 0.4278,
+      "step": 5995
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.038570798561863e-07,
+      "loss": 0.4058,
+      "step": 5996
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.01910621013008e-07,
+      "loss": 0.4116,
+      "step": 5997
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.99966161300072e-07,
+      "loss": 0.4511,
+      "step": 5998
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.980237011446392e-07,
+      "loss": 0.4218,
+      "step": 5999
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.960832409735265e-07,
+      "loss": 0.4257,
+      "step": 6000
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.941447812131121e-07,
+      "loss": 0.4095,
+      "step": 6001
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.922083222893329e-07,
+      "loss": 0.4108,
+      "step": 6002
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.902738646276942e-07,
+      "loss": 0.3961,
+      "step": 6003
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.883414086532527e-07,
+      "loss": 0.4243,
+      "step": 6004
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.864109547906263e-07,
+      "loss": 0.4239,
+      "step": 6005
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.844825034639993e-07,
+      "loss": 0.4158,
+      "step": 6006
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.82556055097108e-07,
+      "loss": 0.4243,
+      "step": 6007
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.806316101132561e-07,
+      "loss": 0.4274,
+      "step": 6008
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.787091689353022e-07,
+      "loss": 0.3869,
+      "step": 6009
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.767887319856627e-07,
+      "loss": 0.3998,
+      "step": 6010
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.748702996863212e-07,
+      "loss": 0.4236,
+      "step": 6011
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.729538724588127e-07,
+      "loss": 0.4161,
+      "step": 6012
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.710394507242393e-07,
+      "loss": 0.4166,
+      "step": 6013
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.691270349032543e-07,
+      "loss": 0.4034,
+      "step": 6014
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.672166254160774e-07,
+      "loss": 0.4188,
+      "step": 6015
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.653082226824805e-07,
+      "loss": 0.4017,
+      "step": 6016
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.634018271218025e-07,
+      "loss": 0.4075,
+      "step": 6017
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.614974391529363e-07,
+      "loss": 0.4234,
+      "step": 6018
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.595950591943313e-07,
+      "loss": 0.4205,
+      "step": 6019
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.576946876640036e-07,
+      "loss": 0.4115,
+      "step": 6020
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.557963249795198e-07,
+      "loss": 0.4171,
+      "step": 6021
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.538999715580098e-07,
+      "loss": 0.421,
+      "step": 6022
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.520056278161615e-07,
+      "loss": 0.4288,
+      "step": 6023
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.501132941702183e-07,
+      "loss": 0.4102,
+      "step": 6024
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.48222971035988e-07,
+      "loss": 0.4039,
+      "step": 6025
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.46334658828829e-07,
+      "loss": 0.4062,
+      "step": 6026
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.444483579636642e-07,
+      "loss": 0.3958,
+      "step": 6027
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.425640688549719e-07,
+      "loss": 0.424,
+      "step": 6028
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.406817919167843e-07,
+      "loss": 0.4324,
+      "step": 6029
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.388015275627015e-07,
+      "loss": 0.4213,
+      "step": 6030
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.369232762058721e-07,
+      "loss": 0.4193,
+      "step": 6031
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.35047038259006e-07,
+      "loss": 0.3975,
+      "step": 6032
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.331728141343709e-07,
+      "loss": 0.4014,
+      "step": 6033
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.313006042437899e-07,
+      "loss": 0.3952,
+      "step": 6034
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.294304089986483e-07,
+      "loss": 0.4341,
+      "step": 6035
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.275622288098817e-07,
+      "loss": 0.4075,
+      "step": 6036
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.256960640879907e-07,
+      "loss": 0.4272,
+      "step": 6037
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.238319152430263e-07,
+      "loss": 0.4368,
+      "step": 6038
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.219697826846007e-07,
+      "loss": 0.4171,
+      "step": 6039
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.201096668218822e-07,
+      "loss": 0.4226,
+      "step": 6040
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.182515680635927e-07,
+      "loss": 0.3971,
+      "step": 6041
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.163954868180179e-07,
+      "loss": 0.4328,
+      "step": 6042
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.145414234929916e-07,
+      "loss": 0.4037,
+      "step": 6043
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.126893784959122e-07,
+      "loss": 0.4108,
+      "step": 6044
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.108393522337299e-07,
+      "loss": 0.4321,
+      "step": 6045
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.089913451129527e-07,
+      "loss": 0.4033,
+      "step": 6046
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.071453575396426e-07,
+      "loss": 0.426,
+      "step": 6047
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.053013899194206e-07,
+      "loss": 0.4262,
+      "step": 6048
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.034594426574649e-07,
+      "loss": 0.3995,
+      "step": 6049
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 8.016195161585061e-07,
+      "loss": 0.3858,
+      "step": 6050
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.997816108268341e-07,
+      "loss": 0.4206,
+      "step": 6051
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.979457270662927e-07,
+      "loss": 0.4157,
+      "step": 6052
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.961118652802812e-07,
+      "loss": 0.4164,
+      "step": 6053
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.942800258717565e-07,
+      "loss": 0.4132,
+      "step": 6054
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.924502092432296e-07,
+      "loss": 0.4001,
+      "step": 6055
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.906224157967691e-07,
+      "loss": 0.4149,
+      "step": 6056
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.887966459339958e-07,
+      "loss": 0.4216,
+      "step": 6057
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.869729000560866e-07,
+      "loss": 0.4105,
+      "step": 6058
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.851511785637766e-07,
+      "loss": 0.4205,
+      "step": 6059
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.833314818573545e-07,
+      "loss": 0.4158,
+      "step": 6060
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.815138103366604e-07,
+      "loss": 0.4443,
+      "step": 6061
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.79698164401097e-07,
+      "loss": 0.4037,
+      "step": 6062
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.778845444496141e-07,
+      "loss": 0.4075,
+      "step": 6063
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.760729508807229e-07,
+      "loss": 0.4207,
+      "step": 6064
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.742633840924818e-07,
+      "loss": 0.4073,
+      "step": 6065
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.724558444825126e-07,
+      "loss": 0.4155,
+      "step": 6066
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.706503324479841e-07,
+      "loss": 0.4193,
+      "step": 6067
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.688468483856271e-07,
+      "loss": 0.4294,
+      "step": 6068
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.670453926917187e-07,
+      "loss": 0.4216,
+      "step": 6069
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.652459657620948e-07,
+      "loss": 0.3991,
+      "step": 6070
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.634485679921466e-07,
+      "loss": 0.388,
+      "step": 6071
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.616531997768151e-07,
+      "loss": 0.4273,
+      "step": 6072
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.598598615106012e-07,
+      "loss": 0.4285,
+      "step": 6073
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.580685535875554e-07,
+      "loss": 0.3947,
+      "step": 6074
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.562792764012805e-07,
+      "loss": 0.4079,
+      "step": 6075
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.544920303449387e-07,
+      "loss": 0.4216,
+      "step": 6076
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.527068158112439e-07,
+      "loss": 0.4346,
+      "step": 6077
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.509236331924596e-07,
+      "loss": 0.4008,
+      "step": 6078
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.49142482880405e-07,
+      "loss": 0.4378,
+      "step": 6079
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.473633652664591e-07,
+      "loss": 0.4003,
+      "step": 6080
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.455862807415448e-07,
+      "loss": 0.4188,
+      "step": 6081
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.438112296961409e-07,
+      "loss": 0.407,
+      "step": 6082
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.420382125202852e-07,
+      "loss": 0.411,
+      "step": 6083
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.402672296035595e-07,
+      "loss": 0.4384,
+      "step": 6084
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.384982813351071e-07,
+      "loss": 0.4052,
+      "step": 6085
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.367313681036192e-07,
+      "loss": 0.4059,
+      "step": 6086
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.349664902973386e-07,
+      "loss": 0.4027,
+      "step": 6087
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.332036483040672e-07,
+      "loss": 0.453,
+      "step": 6088
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.31442842511153e-07,
+      "loss": 0.4151,
+      "step": 6089
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.296840733054999e-07,
+      "loss": 0.4175,
+      "step": 6090
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.279273410735655e-07,
+      "loss": 0.4174,
+      "step": 6091
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.26172646201354e-07,
+      "loss": 0.4111,
+      "step": 6092
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.244199890744297e-07,
+      "loss": 0.4438,
+      "step": 6093
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.22669370077902e-07,
+      "loss": 0.383,
+      "step": 6094
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.209207895964376e-07,
+      "loss": 0.4065,
+      "step": 6095
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.191742480142527e-07,
+      "loss": 0.414,
+      "step": 6096
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.174297457151181e-07,
+      "loss": 0.4166,
+      "step": 6097
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.156872830823535e-07,
+      "loss": 0.4112,
+      "step": 6098
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.139468604988287e-07,
+      "loss": 0.4238,
+      "step": 6099
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.12208478346974e-07,
+      "loss": 0.4119,
+      "step": 6100
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.104721370087598e-07,
+      "loss": 0.4129,
+      "step": 6101
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.087378368657171e-07,
+      "loss": 0.4227,
+      "step": 6102
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.070055782989238e-07,
+      "loss": 0.4015,
+      "step": 6103
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.052753616890107e-07,
+      "loss": 0.4147,
+      "step": 6104
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.035471874161615e-07,
+      "loss": 0.4291,
+      "step": 6105
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.018210558601058e-07,
+      "loss": 0.4294,
+      "step": 6106
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 7.00096967400129e-07,
+      "loss": 0.3775,
+      "step": 6107
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.983749224150671e-07,
+      "loss": 0.3815,
+      "step": 6108
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.966549212833073e-07,
+      "loss": 0.417,
+      "step": 6109
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.949369643827853e-07,
+      "loss": 0.432,
+      "step": 6110
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.932210520909877e-07,
+      "loss": 0.4221,
+      "step": 6111
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.915071847849564e-07,
+      "loss": 0.4343,
+      "step": 6112
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.89795362841278e-07,
+      "loss": 0.4045,
+      "step": 6113
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.880855866360958e-07,
+      "loss": 0.4307,
+      "step": 6114
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.863778565450962e-07,
+      "loss": 0.4233,
+      "step": 6115
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.846721729435235e-07,
+      "loss": 0.4219,
+      "step": 6116
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.829685362061678e-07,
+      "loss": 0.3939,
+      "step": 6117
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.812669467073684e-07,
+      "loss": 0.4327,
+      "step": 6118
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.79567404821021e-07,
+      "loss": 0.4042,
+      "step": 6119
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.778699109205622e-07,
+      "loss": 0.4165,
+      "step": 6120
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.761744653789892e-07,
+      "loss": 0.3969,
+      "step": 6121
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.744810685688397e-07,
+      "loss": 0.4331,
+      "step": 6122
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.727897208622069e-07,
+      "loss": 0.4311,
+      "step": 6123
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.711004226307316e-07,
+      "loss": 0.4307,
+      "step": 6124
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.694131742456012e-07,
+      "loss": 0.4163,
+      "step": 6125
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.677279760775623e-07,
+      "loss": 0.4086,
+      "step": 6126
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.66044828496899e-07,
+      "loss": 0.405,
+      "step": 6127
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.643637318734553e-07,
+      "loss": 0.4101,
+      "step": 6128
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.626846865766178e-07,
+      "loss": 0.4432,
+      "step": 6129
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.610076929753228e-07,
+      "loss": 0.4109,
+      "step": 6130
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.593327514380599e-07,
+      "loss": 0.4112,
+      "step": 6131
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.576598623328634e-07,
+      "loss": 0.4069,
+      "step": 6132
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.559890260273216e-07,
+      "loss": 0.4078,
+      "step": 6133
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.543202428885664e-07,
+      "loss": 0.4001,
+      "step": 6134
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.526535132832801e-07,
+      "loss": 0.3765,
+      "step": 6135
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.509888375776962e-07,
+      "loss": 0.3996,
+      "step": 6136
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.493262161375946e-07,
+      "loss": 0.4044,
+      "step": 6137
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.47665649328304e-07,
+      "loss": 0.423,
+      "step": 6138
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.460071375147025e-07,
+      "loss": 0.4248,
+      "step": 6139
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.443506810612144e-07,
+      "loss": 0.4212,
+      "step": 6140
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.426962803318171e-07,
+      "loss": 0.4129,
+      "step": 6141
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.410439356900311e-07,
+      "loss": 0.416,
+      "step": 6142
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.393936474989293e-07,
+      "loss": 0.4206,
+      "step": 6143
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.377454161211271e-07,
+      "loss": 0.4117,
+      "step": 6144
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.360992419187972e-07,
+      "loss": 0.4307,
+      "step": 6145
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.344551252536524e-07,
+      "loss": 0.4043,
+      "step": 6146
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.328130664869525e-07,
+      "loss": 0.4348,
+      "step": 6147
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.311730659795123e-07,
+      "loss": 0.3953,
+      "step": 6148
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.295351240916892e-07,
+      "loss": 0.4258,
+      "step": 6149
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.278992411833895e-07,
+      "loss": 0.4137,
+      "step": 6150
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.26265417614067e-07,
+      "loss": 0.3898,
+      "step": 6151
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.246336537427222e-07,
+      "loss": 0.424,
+      "step": 6152
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.23003949927905e-07,
+      "loss": 0.4032,
+      "step": 6153
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.213763065277112e-07,
+      "loss": 0.4044,
+      "step": 6154
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.197507238997846e-07,
+      "loss": 0.4289,
+      "step": 6155
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.181272024013129e-07,
+      "loss": 0.4314,
+      "step": 6156
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.165057423890385e-07,
+      "loss": 0.4044,
+      "step": 6157
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.148863442192421e-07,
+      "loss": 0.4161,
+      "step": 6158
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.132690082477565e-07,
+      "loss": 0.4328,
+      "step": 6159
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.116537348299612e-07,
+      "loss": 0.3995,
+      "step": 6160
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.100405243207797e-07,
+      "loss": 0.409,
+      "step": 6161
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.084293770746863e-07,
+      "loss": 0.4447,
+      "step": 6162
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.068202934456979e-07,
+      "loss": 0.4339,
+      "step": 6163
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.052132737873795e-07,
+      "loss": 0.4007,
+      "step": 6164
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.036083184528441e-07,
+      "loss": 0.407,
+      "step": 6165
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.020054277947473e-07,
+      "loss": 0.4371,
+      "step": 6166
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.004046021652976e-07,
+      "loss": 0.4208,
+      "step": 6167
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.988058419162435e-07,
+      "loss": 0.4242,
+      "step": 6168
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.972091473988805e-07,
+      "loss": 0.4191,
+      "step": 6169
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.956145189640538e-07,
+      "loss": 0.4011,
+      "step": 6170
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.940219569621497e-07,
+      "loss": 0.4265,
+      "step": 6171
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.924314617431059e-07,
+      "loss": 0.4182,
+      "step": 6172
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.908430336564008e-07,
+      "loss": 0.3954,
+      "step": 6173
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.892566730510629e-07,
+      "loss": 0.3973,
+      "step": 6174
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.876723802756635e-07,
+      "loss": 0.4135,
+      "step": 6175
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.860901556783194e-07,
+      "loss": 0.4292,
+      "step": 6176
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.845099996066972e-07,
+      "loss": 0.4126,
+      "step": 6177
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.829319124080013e-07,
+      "loss": 0.4266,
+      "step": 6178
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.81355894428991e-07,
+      "loss": 0.4112,
+      "step": 6179
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.797819460159615e-07,
+      "loss": 0.3964,
+      "step": 6180
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.782100675147606e-07,
+      "loss": 0.4213,
+      "step": 6181
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.766402592707787e-07,
+      "loss": 0.441,
+      "step": 6182
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.750725216289488e-07,
+      "loss": 0.418,
+      "step": 6183
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.73506854933753e-07,
+      "loss": 0.3969,
+      "step": 6184
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.719432595292129e-07,
+      "loss": 0.4443,
+      "step": 6185
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.703817357589036e-07,
+      "loss": 0.4504,
+      "step": 6186
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.688222839659374e-07,
+      "loss": 0.4121,
+      "step": 6187
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.672649044929723e-07,
+      "loss": 0.4354,
+      "step": 6188
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.657095976822158e-07,
+      "loss": 0.4073,
+      "step": 6189
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.641563638754132e-07,
+      "loss": 0.4282,
+      "step": 6190
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.626052034138618e-07,
+      "loss": 0.4459,
+      "step": 6191
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.610561166383943e-07,
+      "loss": 0.4255,
+      "step": 6192
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.595091038893985e-07,
+      "loss": 0.4349,
+      "step": 6193
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.579641655067969e-07,
+      "loss": 0.3865,
+      "step": 6194
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.56421301830059e-07,
+      "loss": 0.3988,
+      "step": 6195
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.548805131982038e-07,
+      "loss": 0.4138,
+      "step": 6196
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.533417999497858e-07,
+      "loss": 0.4297,
+      "step": 6197
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.518051624229104e-07,
+      "loss": 0.4123,
+      "step": 6198
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.502706009552228e-07,
+      "loss": 0.4149,
+      "step": 6199
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.487381158839145e-07,
+      "loss": 0.4096,
+      "step": 6200
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.472077075457183e-07,
+      "loss": 0.4334,
+      "step": 6201
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.456793762769108e-07,
+      "loss": 0.4047,
+      "step": 6202
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.441531224133167e-07,
+      "loss": 0.4219,
+      "step": 6203
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.426289462902989e-07,
+      "loss": 0.3822,
+      "step": 6204
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.411068482427672e-07,
+      "loss": 0.4242,
+      "step": 6205
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.39586828605172e-07,
+      "loss": 0.4073,
+      "step": 6206
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.38068887711507e-07,
+      "loss": 0.4301,
+      "step": 6207
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.365530258953156e-07,
+      "loss": 0.4335,
+      "step": 6208
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.350392434896734e-07,
+      "loss": 0.4307,
+      "step": 6209
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.335275408272089e-07,
+      "loss": 0.4311,
+      "step": 6210
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.320179182400887e-07,
+      "loss": 0.3851,
+      "step": 6211
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.305103760600217e-07,
+      "loss": 0.429,
+      "step": 6212
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.290049146182652e-07,
+      "loss": 0.4076,
+      "step": 6213
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.275015342456114e-07,
+      "loss": 0.4099,
+      "step": 6214
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.260002352724014e-07,
+      "loss": 0.4297,
+      "step": 6215
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.245010180285148e-07,
+      "loss": 0.4327,
+      "step": 6216
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.230038828433792e-07,
+      "loss": 0.4357,
+      "step": 6217
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.215088300459581e-07,
+      "loss": 0.4338,
+      "step": 6218
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.200158599647609e-07,
+      "loss": 0.4008,
+      "step": 6219
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.185249729278408e-07,
+      "loss": 0.4492,
+      "step": 6220
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.170361692627901e-07,
+      "loss": 0.4086,
+      "step": 6221
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.155494492967461e-07,
+      "loss": 0.4317,
+      "step": 6222
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.140648133563864e-07,
+      "loss": 0.4003,
+      "step": 6223
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.125822617679299e-07,
+      "loss": 0.4316,
+      "step": 6224
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.111017948571406e-07,
+      "loss": 0.407,
+      "step": 6225
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.096234129493216e-07,
+      "loss": 0.4313,
+      "step": 6226
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.081471163693197e-07,
+      "loss": 0.4161,
+      "step": 6227
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.06672905441522e-07,
+      "loss": 0.4049,
+      "step": 6228
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.052007804898595e-07,
+      "loss": 0.452,
+      "step": 6229
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.037307418378001e-07,
+      "loss": 0.3844,
+      "step": 6230
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.022627898083598e-07,
+      "loss": 0.4307,
+      "step": 6231
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.007969247240918e-07,
+      "loss": 0.3987,
+      "step": 6232
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.993331469070895e-07,
+      "loss": 0.4296,
+      "step": 6233
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.978714566789944e-07,
+      "loss": 0.4211,
+      "step": 6234
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.96411854360982e-07,
+      "loss": 0.3816,
+      "step": 6235
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.949543402737711e-07,
+      "loss": 0.4062,
+      "step": 6236
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.934989147376246e-07,
+      "loss": 0.4031,
+      "step": 6237
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.920455780723421e-07,
+      "loss": 0.395,
+      "step": 6238
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.905943305972682e-07,
+      "loss": 0.3981,
+      "step": 6239
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.891451726312868e-07,
+      "loss": 0.3974,
+      "step": 6240
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.876981044928208e-07,
+      "loss": 0.4243,
+      "step": 6241
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.862531264998371e-07,
+      "loss": 0.4323,
+      "step": 6242
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.848102389698395e-07,
+      "loss": 0.4354,
+      "step": 6243
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.833694422198786e-07,
+      "loss": 0.4081,
+      "step": 6244
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.819307365665394e-07,
+      "loss": 0.4145,
+      "step": 6245
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.804941223259507e-07,
+      "loss": 0.4397,
+      "step": 6246
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.790595998137803e-07,
+      "loss": 0.4038,
+      "step": 6247
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.77627169345235e-07,
+      "loss": 0.4296,
+      "step": 6248
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7619683123506735e-07,
+      "loss": 0.4044,
+      "step": 6249
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.747685857975637e-07,
+      "loss": 0.4185,
+      "step": 6250
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.733424333465564e-07,
+      "loss": 0.4325,
+      "step": 6251
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7191837419541366e-07,
+      "loss": 0.4136,
+      "step": 6252
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7049640865704296e-07,
+      "loss": 0.4197,
+      "step": 6253
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.6907653704389875e-07,
+      "loss": 0.4163,
+      "step": 6254
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.6765875966796493e-07,
+      "loss": 0.4243,
+      "step": 6255
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.662430768407755e-07,
+      "loss": 0.3977,
+      "step": 6256
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6482948887339617e-07,
+      "loss": 0.4364,
+      "step": 6257
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.634179960764396e-07,
+      "loss": 0.4211,
+      "step": 6258
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.620085987600509e-07,
+      "loss": 0.4336,
+      "step": 6259
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6060129723391913e-07,
+      "loss": 0.4444,
+      "step": 6260
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.591960918072713e-07,
+      "loss": 0.4146,
+      "step": 6261
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5779298278887274e-07,
+      "loss": 0.4336,
+      "step": 6262
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5639197048703343e-07,
+      "loss": 0.4244,
+      "step": 6263
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5499305520959713e-07,
+      "loss": 0.423,
+      "step": 6264
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.5359623726394684e-07,
+      "loss": 0.4097,
+      "step": 6265
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.522015169570093e-07,
+      "loss": 0.4001,
+      "step": 6266
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.50808894595246e-07,
+      "loss": 0.4205,
+      "step": 6267
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.494183704846589e-07,
+      "loss": 0.4134,
+      "step": 6268
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.4802994493078897e-07,
+      "loss": 0.3898,
+      "step": 6269
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4664361823871684e-07,
+      "loss": 0.4104,
+      "step": 6270
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.452593907130609e-07,
+      "loss": 0.447,
+      "step": 6271
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4387726265797683e-07,
+      "loss": 0.4038,
+      "step": 6272
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.424972343771627e-07,
+      "loss": 0.4216,
+      "step": 6273
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.411193061738539e-07,
+      "loss": 0.4012,
+      "step": 6274
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.397434783508203e-07,
+      "loss": 0.4356,
+      "step": 6275
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3836975121037686e-07,
+      "loss": 0.4297,
+      "step": 6276
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.36998125054372e-07,
+      "loss": 0.4197,
+      "step": 6277
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3562860018419583e-07,
+      "loss": 0.4117,
+      "step": 6278
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3426117690077206e-07,
+      "loss": 0.434,
+      "step": 6279
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3289585550456923e-07,
+      "loss": 0.4145,
+      "step": 6280
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.315326362955874e-07,
+      "loss": 0.4118,
+      "step": 6281
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.301715195733702e-07,
+      "loss": 0.4243,
+      "step": 6282
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.2881250563699627e-07,
+      "loss": 0.4309,
+      "step": 6283
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2745559478508005e-07,
+      "loss": 0.4183,
+      "step": 6284
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2610078731578186e-07,
+      "loss": 0.4164,
+      "step": 6285
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.247480835267892e-07,
+      "loss": 0.4221,
+      "step": 6286
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.233974837153376e-07,
+      "loss": 0.4351,
+      "step": 6287
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.2204898817819194e-07,
+      "loss": 0.4179,
+      "step": 6288
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.207025972116585e-07,
+      "loss": 0.3988,
+      "step": 6289
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1935831111158286e-07,
+      "loss": 0.4231,
+      "step": 6290
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.180161301733454e-07,
+      "loss": 0.4007,
+      "step": 6291
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.166760546918636e-07,
+      "loss": 0.4129,
+      "step": 6292
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1533808496159177e-07,
+      "loss": 0.3916,
+      "step": 6293
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.14002221276526e-07,
+      "loss": 0.4169,
+      "step": 6294
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.12668463930197e-07,
+      "loss": 0.4167,
+      "step": 6295
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.113368132156681e-07,
+      "loss": 0.4068,
+      "step": 6296
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1000726942554967e-07,
+      "loss": 0.4342,
+      "step": 6297
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.086798328519781e-07,
+      "loss": 0.4182,
+      "step": 6298
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0735450378663556e-07,
+      "loss": 0.4058,
+      "step": 6299
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.060312825207369e-07,
+      "loss": 0.4215,
+      "step": 6300
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0471016934503284e-07,
+      "loss": 0.4134,
+      "step": 6301
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.0339116454981454e-07,
+      "loss": 0.3914,
+      "step": 6302
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.020742684249068e-07,
+      "loss": 0.4062,
+      "step": 6303
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.007594812596738e-07,
+      "loss": 0.4244,
+      "step": 6304
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9944680334301435e-07,
+      "loss": 0.4066,
+      "step": 6305
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9813623496336216e-07,
+      "loss": 0.4048,
+      "step": 6306
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.968277764086914e-07,
+      "loss": 0.4162,
+      "step": 6307
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9552142796650983e-07,
+      "loss": 0.4185,
+      "step": 6308
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.942171899238623e-07,
+      "loss": 0.421,
+      "step": 6309
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.929150625673306e-07,
+      "loss": 0.4128,
+      "step": 6310
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.916150461830326e-07,
+      "loss": 0.4107,
+      "step": 6311
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.903171410566209e-07,
+      "loss": 0.4064,
+      "step": 6312
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8902134747328515e-07,
+      "loss": 0.3901,
+      "step": 6313
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.87727665717752e-07,
+      "loss": 0.415,
+      "step": 6314
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.864360960742819e-07,
+      "loss": 0.4397,
+      "step": 6315
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8514663882667447e-07,
+      "loss": 0.4267,
+      "step": 6316
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8385929425826196e-07,
+      "loss": 0.3907,
+      "step": 6317
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8257406265191254e-07,
+      "loss": 0.4071,
+      "step": 6318
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8129094429003364e-07,
+      "loss": 0.439,
+      "step": 6319
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.80009939454562e-07,
+      "loss": 0.4269,
+      "step": 6320
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.7873104842697903e-07,
+      "loss": 0.421,
+      "step": 6321
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.774542714882923e-07,
+      "loss": 0.4153,
+      "step": 6322
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.761796089190506e-07,
+      "loss": 0.3938,
+      "step": 6323
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.749070609993366e-07,
+      "loss": 0.4101,
+      "step": 6324
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7363662800876663e-07,
+      "loss": 0.3893,
+      "step": 6325
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.7236831022649524e-07,
+      "loss": 0.4284,
+      "step": 6326
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.711021079312105e-07,
+      "loss": 0.4283,
+      "step": 6327
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6983802140113765e-07,
+      "loss": 0.4116,
+      "step": 6328
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.685760509140346e-07,
+      "loss": 0.4083,
+      "step": 6329
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6731619674719275e-07,
+      "loss": 0.4185,
+      "step": 6330
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.66058459177443e-07,
+      "loss": 0.4274,
+      "step": 6331
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.648028384811486e-07,
+      "loss": 0.4305,
+      "step": 6332
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6354933493420896e-07,
+      "loss": 0.3975,
+      "step": 6333
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.622979488120559e-07,
+      "loss": 0.4283,
+      "step": 6334
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.6104868038965845e-07,
+      "loss": 0.4385,
+      "step": 6335
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.598015299415203e-07,
+      "loss": 0.4138,
+      "step": 6336
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.5855649774167666e-07,
+      "loss": 0.4333,
+      "step": 6337
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.573135840636999e-07,
+      "loss": 0.4079,
+      "step": 6338
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.560727891806959e-07,
+      "loss": 0.4351,
+      "step": 6339
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5483411336530773e-07,
+      "loss": 0.4194,
+      "step": 6340
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5359755688970876e-07,
+      "loss": 0.4219,
+      "step": 6341
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5236312002560835e-07,
+      "loss": 0.4207,
+      "step": 6342
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5113080304425175e-07,
+      "loss": 0.4343,
+      "step": 6343
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4990060621641454e-07,
+      "loss": 0.4054,
+      "step": 6344
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.486725298124116e-07,
+      "loss": 0.4093,
+      "step": 6345
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4744657410208713e-07,
+      "loss": 0.4059,
+      "step": 6346
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4622273935482234e-07,
+      "loss": 0.4438,
+      "step": 6347
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.45001025839532e-07,
+      "loss": 0.4343,
+      "step": 6348
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4378143382466147e-07,
+      "loss": 0.4165,
+      "step": 6349
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4256396357819633e-07,
+      "loss": 0.4214,
+      "step": 6350
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.413486153676504e-07,
+      "loss": 0.4156,
+      "step": 6351
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.401353894600723e-07,
+      "loss": 0.4058,
+      "step": 6352
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3892428612204543e-07,
+      "loss": 0.4126,
+      "step": 6353
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.377153056196869e-07,
+      "loss": 0.4225,
+      "step": 6354
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.365084482186487e-07,
+      "loss": 0.4076,
+      "step": 6355
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.353037141841098e-07,
+      "loss": 0.4016,
+      "step": 6356
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3410110378079284e-07,
+      "loss": 0.3925,
+      "step": 6357
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3290061727294433e-07,
+      "loss": 0.3958,
+      "step": 6358
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.317022549243498e-07,
+      "loss": 0.3901,
+      "step": 6359
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.305060169983276e-07,
+      "loss": 0.4187,
+      "step": 6360
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2931190375772417e-07,
+      "loss": 0.3947,
+      "step": 6361
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.281199154649273e-07,
+      "loss": 0.4289,
+      "step": 6362
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.269300523818497e-07,
+      "loss": 0.4493,
+      "step": 6363
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2574231476994457e-07,
+      "loss": 0.4143,
+      "step": 6364
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.245567028901919e-07,
+      "loss": 0.4076,
+      "step": 6365
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2337321700310786e-07,
+      "loss": 0.4255,
+      "step": 6366
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2219185736874104e-07,
+      "loss": 0.418,
+      "step": 6367
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.210126242466716e-07,
+      "loss": 0.4215,
+      "step": 6368
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.198355178960144e-07,
+      "loss": 0.4162,
+      "step": 6369
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.186605385754138e-07,
+      "loss": 0.401,
+      "step": 6370
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.174876865430521e-07,
+      "loss": 0.4351,
+      "step": 6371
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.163169620566398e-07,
+      "loss": 0.4107,
+      "step": 6372
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.151483653734189e-07,
+      "loss": 0.395,
+      "step": 6373
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1398189675016844e-07,
+      "loss": 0.409,
+      "step": 6374
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.128175564431968e-07,
+      "loss": 0.4087,
+      "step": 6375
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.11655344708347e-07,
+      "loss": 0.4393,
+      "step": 6376
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1049526180099046e-07,
+      "loss": 0.4016,
+      "step": 6377
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.09337307976032e-07,
+      "loss": 0.4341,
+      "step": 6378
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.0818148348791377e-07,
+      "loss": 0.433,
+      "step": 6379
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.0702778859060255e-07,
+      "loss": 0.4196,
+      "step": 6380
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.058762235376034e-07,
+      "loss": 0.4031,
+      "step": 6381
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0472678858194827e-07,
+      "loss": 0.4062,
+      "step": 6382
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.035794839762041e-07,
+      "loss": 0.4323,
+      "step": 6383
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0243430997246916e-07,
+      "loss": 0.4142,
+      "step": 6384
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.0129126682237217e-07,
+      "loss": 0.3913,
+      "step": 6385
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.001503547770779e-07,
+      "loss": 0.4087,
+      "step": 6386
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9901157408727565e-07,
+      "loss": 0.4178,
+      "step": 6387
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9787492500319536e-07,
+      "loss": 0.4105,
+      "step": 6388
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9674040777459055e-07,
+      "loss": 0.4122,
+      "step": 6389
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9560802265074964e-07,
+      "loss": 0.4036,
+      "step": 6390
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.944777698804946e-07,
+      "loss": 0.4211,
+      "step": 6391
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.933496497121746e-07,
+      "loss": 0.4253,
+      "step": 6392
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.922236623936736e-07,
+      "loss": 0.4257,
+      "step": 6393
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.910998081724037e-07,
+      "loss": 0.3968,
+      "step": 6394
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8997808729531396e-07,
+      "loss": 0.41,
+      "step": 6395
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.888585000088784e-07,
+      "loss": 0.4279,
+      "step": 6396
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8774104655910463e-07,
+      "loss": 0.4251,
+      "step": 6397
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.866257271915329e-07,
+      "loss": 0.4178,
+      "step": 6398
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.855125421512317e-07,
+      "loss": 0.4119,
+      "step": 6399
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8440149168280304e-07,
+      "loss": 0.4125,
+      "step": 6400
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8329257603037954e-07,
+      "loss": 0.4356,
+      "step": 6401
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.821857954376217e-07,
+      "loss": 0.4087,
+      "step": 6402
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.810811501477262e-07,
+      "loss": 0.4157,
+      "step": 6403
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.7997864040341436e-07,
+      "loss": 0.4057,
+      "step": 6404
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.788782664469447e-07,
+      "loss": 0.3893,
+      "step": 6405
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.7778002852010267e-07,
+      "loss": 0.3984,
+      "step": 6406
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.7668392686420295e-07,
+      "loss": 0.4226,
+      "step": 6407
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.755899617200941e-07,
+      "loss": 0.4337,
+      "step": 6408
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.744981333281538e-07,
+      "loss": 0.4254,
+      "step": 6409
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.7340844192829117e-07,
+      "loss": 0.4253,
+      "step": 6410
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.723208877599426e-07,
+      "loss": 0.4431,
+      "step": 6411
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.712354710620801e-07,
+      "loss": 0.4291,
+      "step": 6412
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.701521920732031e-07,
+      "loss": 0.4009,
+      "step": 6413
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.690710510313388e-07,
+      "loss": 0.4291,
+      "step": 6414
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6799204817404835e-07,
+      "loss": 0.413,
+      "step": 6415
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6691518373842094e-07,
+      "loss": 0.4196,
+      "step": 6416
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.658404579610796e-07,
+      "loss": 0.411,
+      "step": 6417
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6476787107817205e-07,
+      "loss": 0.4056,
+      "step": 6418
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6369742332537975e-07,
+      "loss": 0.4052,
+      "step": 6419
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.626291149379123e-07,
+      "loss": 0.4252,
+      "step": 6420
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6156294615050983e-07,
+      "loss": 0.421,
+      "step": 6421
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.604989171974448e-07,
+      "loss": 0.4257,
+      "step": 6422
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.594370283125136e-07,
+      "loss": 0.4136,
+      "step": 6423
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5837727972904845e-07,
+      "loss": 0.4086,
+      "step": 6424
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.573196716799087e-07,
+      "loss": 0.3936,
+      "step": 6425
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.562642043974806e-07,
+      "loss": 0.4016,
+      "step": 6426
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.552108781136864e-07,
+      "loss": 0.4306,
+      "step": 6427
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5415969305997326e-07,
+      "loss": 0.4294,
+      "step": 6428
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5311064946731744e-07,
+      "loss": 0.4255,
+      "step": 6429
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5206374756622684e-07,
+      "loss": 0.4151,
+      "step": 6430
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5101898758673746e-07,
+      "loss": 0.4113,
+      "step": 6431
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.4997636975841676e-07,
+      "loss": 0.4189,
+      "step": 6432
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.4893589431035815e-07,
+      "loss": 0.4172,
+      "step": 6433
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.478975614711887e-07,
+      "loss": 0.4205,
+      "step": 6434
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.468613714690593e-07,
+      "loss": 0.4402,
+      "step": 6435
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.458273245316545e-07,
+      "loss": 0.3937,
+      "step": 6436
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.447954208861858e-07,
+      "loss": 0.4328,
+      "step": 6437
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.43765660759393e-07,
+      "loss": 0.396,
+      "step": 6438
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4273804437754843e-07,
+      "loss": 0.4158,
+      "step": 6439
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4171257196644815e-07,
+      "loss": 0.4088,
+      "step": 6440
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4068924375142413e-07,
+      "loss": 0.419,
+      "step": 6441
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.396680599573287e-07,
+      "loss": 0.4162,
+      "step": 6442
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3864902080855015e-07,
+      "loss": 0.4042,
+      "step": 6443
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.376321265290016e-07,
+      "loss": 0.4069,
+      "step": 6444
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3661737734212653e-07,
+      "loss": 0.4281,
+      "step": 6445
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.356047734708966e-07,
+      "loss": 0.4154,
+      "step": 6446
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.345943151378105e-07,
+      "loss": 0.4232,
+      "step": 6447
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.335860025648984e-07,
+      "loss": 0.4079,
+      "step": 6448
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3257983597371858e-07,
+      "loss": 0.4033,
+      "step": 6449
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3157581558535424e-07,
+      "loss": 0.4132,
+      "step": 6450
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.3057394162042223e-07,
+      "loss": 0.4118,
+      "step": 6451
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2957421429906314e-07,
+      "loss": 0.4312,
+      "step": 6452
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2857663384095007e-07,
+      "loss": 0.4086,
+      "step": 6453
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2758120046527998e-07,
+      "loss": 0.4296,
+      "step": 6454
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.26587914390779e-07,
+      "loss": 0.4348,
+      "step": 6455
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2559677583570694e-07,
+      "loss": 0.413,
+      "step": 6456
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2460778501784297e-07,
+      "loss": 0.4435,
+      "step": 6457
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2362094215450325e-07,
+      "loss": 0.413,
+      "step": 6458
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2263624746252433e-07,
+      "loss": 0.4083,
+      "step": 6459
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2165370115827423e-07,
+      "loss": 0.3997,
+      "step": 6460
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2067330345765025e-07,
+      "loss": 0.4156,
+      "step": 6461
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.1969505457607343e-07,
+      "loss": 0.4192,
+      "step": 6462
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.1871895472849736e-07,
+      "loss": 0.3746,
+      "step": 6463
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1774500412939935e-07,
+      "loss": 0.4018,
+      "step": 6464
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1677320299278826e-07,
+      "loss": 0.4043,
+      "step": 6465
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1580355153219877e-07,
+      "loss": 0.4146,
+      "step": 6466
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1483604996069053e-07,
+      "loss": 0.435,
+      "step": 6467
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1387069849085563e-07,
+      "loss": 0.4222,
+      "step": 6468
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.129074973348111e-07,
+      "loss": 0.4541,
+      "step": 6469
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1194644670420206e-07,
+      "loss": 0.4115,
+      "step": 6470
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1098754681019962e-07,
+      "loss": 0.43,
+      "step": 6471
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1003079786350632e-07,
+      "loss": 0.4287,
+      "step": 6472
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0907620007434736e-07,
+      "loss": 0.4177,
+      "step": 6473
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0812375365247718e-07,
+      "loss": 0.4024,
+      "step": 6474
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.071734588071772e-07,
+      "loss": 0.4289,
+      "step": 6475
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0622531574725714e-07,
+      "loss": 0.4155,
+      "step": 6476
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.052793246810547e-07,
+      "loss": 0.4125,
+      "step": 6477
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0433548581643036e-07,
+      "loss": 0.4041,
+      "step": 6478
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0339379936077596e-07,
+      "loss": 0.417,
+      "step": 6479
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0245426552100933e-07,
+      "loss": 0.4118,
+      "step": 6480
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.015168845035742e-07,
+      "loss": 0.3969,
+      "step": 6481
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0058165651444362e-07,
+      "loss": 0.4331,
+      "step": 6482
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9964858175911427e-07,
+      "loss": 0.4049,
+      "step": 6483
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.987176604426111e-07,
+      "loss": 0.4187,
+      "step": 6484
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.977888927694882e-07,
+      "loss": 0.4378,
+      "step": 6485
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9686227894382127e-07,
+      "loss": 0.434,
+      "step": 6486
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.959378191692196e-07,
+      "loss": 0.4102,
+      "step": 6487
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.950155136488141e-07,
+      "loss": 0.4132,
+      "step": 6488
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9409536258526153e-07,
+      "loss": 0.4268,
+      "step": 6489
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9317736618075012e-07,
+      "loss": 0.3968,
+      "step": 6490
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9226152463699188e-07,
+      "loss": 0.425,
+      "step": 6491
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9134783815522363e-07,
+      "loss": 0.4038,
+      "step": 6492
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.904363069362103e-07,
+      "loss": 0.4306,
+      "step": 6493
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.89526931180245e-07,
+      "loss": 0.399,
+      "step": 6494
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8861971108714462e-07,
+      "loss": 0.4074,
+      "step": 6495
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8771464685625406e-07,
+      "loss": 0.4048,
+      "step": 6496
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.868117386864432e-07,
+      "loss": 0.4023,
+      "step": 6497
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8591098677610776e-07,
+      "loss": 0.3754,
+      "step": 6498
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8501239132317272e-07,
+      "loss": 0.4272,
+      "step": 6499
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8411595252508575e-07,
+      "loss": 0.397,
+      "step": 6500
+    }
+  ],
+  "max_steps": 6910,
+  "num_train_epochs": 5,
+  "total_flos": 0.0,
+  "trial_name": null,
+  "trial_params": null
+}